Estudio de biomarcadores de neuroimagen multimodal implicados en la enfermedad de Parkinson

TESIS DOCTORAL ESTUDIO DE BIOMARCADORES BASADOS EN NEUROIMAGEN MULTIMODAL IMPLICADOS EN LA PROGRESIÓN DE LA ENFERMEDAD DE PARKINSON Autor Miguel Ángel Labrador Espinosa Directores Pablo Mir Rivera Michel J. Grothe UNIVERSIDAD DE SEVILLA DEPARTAMENTO DE MEDICINA | FACULTAD DE MEDICINA SEVILLA, NOVIEMBRE 2023 Tesis doctoral presentada para optar al grado de Doctor con mención internacional por la Universidad de Sevilla Instituto de Biomedicina de Sevilla – IBiS Campus Hospital Universitario Virgen del Rocío Avda. Manuel Siurot, s/n - 41013 Sevilla D. Pablo Mir Rivera, Profesor Titular vinculado del Departamento de Medicina de la Facultad de Medicina de la Universidad de Sevilla, y D. Michel J. Grothe, Doctor vinculado al Instituto de Biomedicina de Sevilla (IBiS), CERTIFICAN: Que la memoria del trabajo titulada “Estudio de biomarcadores basados en neuroimagen multimodal implicados en la progresión de la enfermedad de Parkinson” elaborada por Miguel Ángel Labrador Espinosa, graduado en Ingeniería de la Salud por la Universidad de Sevilla, ha sido realizada bajo su dirección y cumple las condiciones exigidas para optar al grado de Doctor con Mención Internacional por la Universidad de Sevilla. Para que así conste, firman la presente en Sevilla, 14 de noviembre de 2023. Fdo.: D. Pablo Mir Rivera Fdo.: D. Michel J. Grothe AGRADECIMIENTOS Me gustaría expresar mi más sincero agradecimiento a mis directores de tesis, ambos hombres grandes y de excelencia. A Michel por consolidar en mí una vocación y profunda pasión por la ciencia, la cual me motiva a continuar mi carrera científica. Gracias por ser todo un referente y mentor, por impulsarme a nivel internacional, por enseñarme tanto y al detalle, ya sea profesional como personal. Gracias por todos los bonitos momentos que hemos vivido juntos y que viviremos, por ser germano y macareno, por ser un gran y mejor amigo de los que pocos quedan, que para mí lo serás para toda la vida. Sin ti no hubiese llegado donde me encuentro actualmente. A Pablo por creer y apostar por mí desde el principio de mi carrera científica, por crear un maravilloso e insólito entorno donde la ingeniería y la neurología van de la mano y en el que yo como ingeniero he tenido la oportunidad de aprender lo que hoy es mi oficio bajo su tutela. Gracias por seguir liderando todo esto frente a las adversidades de estos años, por abrirme las puertas, por acogerme en la gran familia científica que hoy es tu grupo e impulsarme desde tu casa. Una vez más, sin ti no hubiese llegado donde me encuentro actualmente. Y aquí va la gran familia científica del laboratorio 104, al menos con los que he tenido el placer de trabajar próximamente y, aun así, somos muchos… Tendré cientos de historias para contar que creo que no hace falta ni acentuar aquí. Quiero agradecer a Jesús y a Sandra por ser fantásticos compañeros con los que me he podido sentir integrado en un equipo de neuroimagen. Gracias por nuestras quedadas con Michel en la calle Feria y por todos los buenos ratos que hemos reído juntos, siento que empezamos una nueva familia. Gracias a Sergio, Juanfran, Juan, Pablo, Fran, Miguel y Alejandro por las buenas terrazas en el IBiS, mostos y canciones del Mani que hemos echado. Gracias a Lorena por ser de mis primeras compañeras en el lab junto con Flor y, por supuesto, gracias por tu eficacia como research assistant, no se te escapa ni un trámite. A Elena, Ana, Manuela, Brenda, Cristina, Paloma y a todo el equipo de neurofisio por hacer del lab un espacio divertido donde reír. A Piluca, Tere, Marta, Loli, Rocío, Sergio, Rafa, Cristina y a todo el equipo de genética y wet-lab que son fieles representantes de la alegría del 104 para todo aquel que se asome por la puerta azul. Gracias a Dani, Silvia, Astrid, Laura, Elena y a todo el equipo de neurólogos que hace posible que esto funcione, gracias por vuestra vocación y quedaros tras las consultas, gracias por haber escogido el camino de la excelencia que os brinda Pablo. Gracias a Ismael por haber abierto y allanado el camino para que otra generación de ingenieros también puedan recorrerlo. Gracias a Flori, David y Paula por la exitosa colaboración que he podido tener con radiología y medicina nuclear, acercando la actividad de sus servicios al IBiS. Gracias a Blanca Delgado por ser una persona maravillosa y tan eficiente en su trabajo, siempre con una sonrisa para todo aquel que pregunta en su mesa. En especial quiero agradecer a Mario y Antonio por ser las mejores personas con las que he podido trabajar, y así ha sido que sois de mis mejores amigos. Gracias por todo el apoyo que me habéis dado durante estos años y los bonitos recuerdos que tenemos y tendremos juntos. Aquí tenéis otro apoyo para toda la vida. I would also like to express my gratitude to my international colleagues. Thanks to Eric Westman and the NEO team for the wonderful adventures in Stockholm at the Karolinska Institutet. Thanks to Niels Okkels for our successful collaboration and for the amazing moments and culinary experiences we shared in Pumarejo. Thanks to Michael Schöll and his team for the great moments at the AD/PD congresses and for giving me the opportunity to continue my career as a postdoc. No olvido a los docentes que han marcado bonitos valores durante mi formación académica, gracias a Juan Antonio Álvarez García, a Santiago y, en especial, a Reyes Rodríguez Vicent. Mi mayor agradecimiento va para mis seres queridos. A Ana, gracias por todo tu amor incondicional, paciencia infinita y tu constante apoyo que me has dado a lo largo de estos años. Quién iba a imaginar que me reencontraría con aquella chica tan dulce como enigmática de mi adolescencia, daría un vuelco a todo y convertiría estos años de mayor esfuerzo en los más hermosos de mi vida. Te agradezco de corazón por crear un hogar donde ambos encontramos la felicidad. Gracias. A nuestro peludo de cuatro patas Haku, siempre recostado a mi lado en las largas jornadas de trabajo sin importarle la hora. A Javier, Pepe y Conchi, gracias por las tardes y buenos ratos que hemos echado que hacían despejarme un rato y, sobre todo, gracias por el amor incondicional que le dais a Ana, que la hace tan feliz y, por ende, a mí. Al equipo dominguero, José Luis, Miguel, Emilio y Loli que siempre han sabido sacarme carcajadas entre olivos, cervezas, paellas y dominós, gracias de corazón. A mi familia que ya no está y, en especial, a los que se fueron recientemente, Facundo e Isabel, no había un día que no me preguntarais con interés y orgullo como llevaba esto que estaba haciendo. A mis sobrinos Sandra, Iris, Karel, Laura e Indira, a los que he visto crecer conmigo y me han mostrado siempre todo su cariño. Veo en sus ojos los buenos valores que sus padres y abuelos han transmitido a la siguiente generación. A mis cuñados, por haber sido también grandes apoyos para mí y para mis hermanos. A mi cuñá Yoli, gracias por contagiarme siempre tu alegría y entusiasmo, gracias por tus buenos consejos, gracias por hacer tan feliz a mi hermano, gracias por ser como una hermana para mí. A mi cuñao David, gracias por haberme escuchado en momentos complicados, por nuestros debates con cervezas y risas, por el fútbol y, por supuesto, por hacer feliz a mi hermana. A mi cuñao Antonio, gracias por el apoyo incondicional y felicidad diaria que le das a mi hermana. A mis hermanos, personas maravillosas, a los que les debo todo y encima no dejan de expresar lo orgullosos que están de mí. Orgulloso estoy yo de tener a unos hermanos que siempre han estado ahí, que siempre me han aconsejado desde su experiencia y que me han dado todo su cariño hasta convertirme en la persona que soy. Gracias Sandra, agradezco enormemente tu amor ilimitado desde mi niñez, tus sabios consejos, por tu contagiosa alegría, por tu presencia constante que ha sido un regalo inestimable en mi vida. Gracias por nuestras aventuras que me permitieron desconectar en los momentos más difíciles. ¡Si es que incluso viniste al Karolinska! Gracias Piri, no puedo agradecerte lo suficiente todo el cariño incondicional que me has dado desde mi infancia, por tus sabios consejos y por ser mi guía en los momentos más cruciales. Gracias por mostrarme el arte de la música que hoy día sigo disfrutando. Gracias de corazón por los sólidos principios y valores que me enseñaste. Gracias Chari por cuidar de aquel enano que ya no lo es tanto, siempre llevaré tu amor en mi corazón pase lo que pase. Los ecos del flamenco de aquellos años suenan en mi casa ahora. Finalmente, y por encima de todo, si hoy soy el hombre que he podido llegar a ser se lo debo a mis padres. Sin tener nada, lo habéis creado todo para dárselo a vuestros hijos. Tal que así ha sido que llegué a la universidad y me saqué un grado en ingeniería. ¡Pero es que luego hice hasta máster! Quien iba a pensar que vuestro hijo iba a llegar a ser doctor, cuando ni sabíamos que este nivel existía. Gracias por ser personas ejemplares a seguir que me han enseñado con hechos que el amor, la bondad, la familia, la amistad, la educación y el esfuerzo es un derecho y un deber que te lleva a cualquier destino que quieras alcanzar y, lo que es más importante, te permite encontrar la felicidad durante todo el recorrido. Jamás he sentido que estos valores hayan faltado en nuestra familia pese a todos estos años de adversidades que sé que han sido muy duros. Pese a todo, jamás habéis dejado de darme vuestro cariño, comprenderme y apoyarme con todo lo que habéis tenido a vuestro alcance. Si he podido llegar hasta aquí y sentir que hoy soy plenamente feliz es sin lugar a duda gracias a vosotros. Os quiero muchísimo, muchas gracias. “Si he logrado ver más lejos ha sido porque he subido a hombros de gigantes”, es todo un cliché en ciencia, pero siento firmemente que es mi caso. Durante toda mi vida he tenido la suerte de crecer y rodearme de personas maravillosas que siempre han sacado de mí una mejor versión, ya sea en el ámbito personal, académico como profesional o, al menos yo, me quedo con estas personas que han hecho de mi vida un camino que volvería a recorrer. Tengo mucho más que agradecer y me gustaría que no os quedéis únicamente con las palabras que os dedico en este papel ya que no sería ni el más mínimo reflejo de todo lo que me habéis aportado durante estos años. Prefiero que mantengamos vivo los buenos recuerdos que hemos pasado juntos hasta aquí. ¡Muchas gracias a todos! CONTENIDO ABREVIATURAS .................................................................................................................. 1 INTRODUCCIÓN .................................................................................................................. 4 Antecedentes históricos y epidemiología de la EP ...................................................................... 4 Neuropatología e implicaciones fisiopatológicas ........................................................................ 5 Etiopatogenia ........................................................................................................................................ 5 Sistema dopaminérgico, circuitos de los ganglios basales y su relación con los síntomas de la EP ...... 8 Patología de Lewy y estadios de Braak ............................................................................................... 11 Clínica de la EP .......................................................................................................................... 13 Características principales ................................................................................................................... 13 Diagnóstico y el uso de [I123]FP-CIT SPECT .......................................................................................... 13 Tratamiento principal de la EP ............................................................................................................ 15 Amplio espectro clínico de la progresión de la EP .............................................................................. 17 Discinesias inducidas por levodopa en la EP ....................................................................................... 19 Deterioro cognitivo en la EP ................................................................................................................ 22 Biomarcadores de neuroimagen en la progresión de la EP ........................................................ 24 Biomarcadores de neuroimagen de las DILs en la EP .......................................................................... 27 Biomarcadores de neuroimagen del deterioro cognitivo en la EP ...................................................... 29 HIPÓTESIS / HYPOTHESIS ................................................................................................. 33 OBJETIVOS / OBJECTIVES .................................................................................................. 38 MÉTODOS ........................................................................................................................ 45 Métodos comunes .................................................................................................................... 45 Cuantificación de los niveles de los DAT del estriado usando [123I]FP-CIT SPECT ............................... 45 Cuantificación del volumen de regiones de interés y del grosor cortical usando MRI ....................... 46 Cuantificación de la difusividad media de regiones de interés usando DTI ........................................ 47 Cuantificación del metabolismo cortical usando [18F]FDG PET ........................................................... 48 Material y métodos de cada estudio individual ......................................................................... 48 Estudio 1. “Levodopa-induced dyskinesia in Parkinson’s disease specifically associates with dopaminergic depletion in sensorimotor-related functional subregions of the striatum” .................. 48 Estudio 2. “In vivo cholinergic basal forebrain degeneration and cognition in Parkinson’s disease: Imaging results from the COPPADIS study” ......................................................................................... 50 Estudio 3. “Basal forebrain atrophy, cortical thinning, and amyloid-β status in Parkinson’s diseaserelated cognitive decline” .................................................................................................................... 51 Estudio 4. “Cholinergic basal forebrain atrophy associates with cortical hypometabolism in Parkinson’s disease-related cognitive impairment” ............................................................................ 53 RESULTADOS .................................................................................................................... 57 Resumen de resultados del estudio 1: “Levodopa-induced dyskinesia in Parkinson’s disease specifically associates with dopaminergic depletion in sensorimotor-related functional subregions of the striatum” ....................................................................................................................................... 57 Resumen de resultados del estudio 2: “In vivo cholinergic basal forebrain degeneration and cognition in Parkinson’s disease: Imaging results from the COPPADIS study” .................................... 59 Resumen de resultados del estudio 3. “Basal forebrain atrophy, cortical thinning, and amyloid-β status in Parkinson’s disease-related cognitive decline” ..................................................................... 61 Resumen de resultados del estudio 4. “Cholinergic basal forebrain atrophy associates with cortical hypometabolism in Parkinson’s disease-related cognitive impairment” ............................................ 63 PUBLICACIONES / PUBLICATIONS ..................................................................................... 65 Levodopa-induced dyskinesia in Parkinson’s disease specifically associates with dopaminergic depletion in sensorimotor-related functional subregions of the striatum ...................................... 66 In vivo cholinergic basal forebrain degeneration and cognition in Parkinson’s disease: Imaging results from the COPPADIS study .................................................................................................... 81 Basal forebrain atrophy, cortical thinning, and amyloid-β status in Parkinson’s disease-related cognitive decline .............................................................................................................................. 97 Cholinergic basal forebrain atrophy associates with cortical hypometabolism in Parkinson’s disease-related cognitive impairment ........................................................................................... 128 DISCUSIÓN ..................................................................................................................... 153 Discinesias inducidas por levodopa en la EP y agotamiento dopaminérgico en regiones sensitivomotoras del estriado ..................................................................................................................... 153 Asociaciones in vivo de la degeneración del prosencéfalo basal colinérgico y el deterioro cognitivo en la EP .......................................................................................................................................... 159 Atrofia del prosencéfalo basal colinérgico, adelgazamiento cortical y estado de β-amiloide en la EP ....................................................................................................................................................... 163 Atrofia del prosencéfalo basal colinérgico, hipometabolismo cortical y deterioro cognitivo en la EP ....................................................................................................................................................... 168 Limitaciones y perspectivas de futuro ........................................................................................... 173 CONCLUSIONES / CONCLUSIONS .................................................................................... 177 REFERENCIAS ................................................................................................................. 180 7 biológica plausible de algunos de estos factores, la replicabilidad de estas observaciones no ha sido uniforme. En conjunto, la creciente evidencia en la literatura científica actual apoya que el envejecimiento y la combinación de estos factores genéticos y ambientales desencadena distintos procesos fisiopatológicos a nivel celular y molecular. Entre estos procesos fisiopatológicos destacarían la proteostasis y degradación de la α-sinucleína, la propagación priónica de α-sinucleína como proteína patógena, la disfunción mitocondrial, la alteración de las vías de autofagia-lisosomal, el estrés oxidativo, la homeostasis del calcio, el transporte axonal y la neuroinflamación (figura 1). Finalmente, el desencadenamiento de estos procesos celulares culminaría en la muerte neuronal, la neurodegeneración en los tejidos implicados y la aparición de la sintomatología de la enfermedad (Bloem et al., 2021; Jankovic & Tan, 2020; Minakaki et al., 2020). Figura 1. Patogénesis de la EP. Una variedad de mecanismos celulares en el contexto del estrés oxidativo, junto con factores genéticos/ambientales y de estilo de vida, contribuyen a la neurodegeneración relacionada con la EP. [Figura adaptada de (Jankovic & Tan, 2020)]. Patogénesis de la enfermedad de Parkinson 8 Sistema dopaminérgico, circuitos de los ganglios basales y su relación con los síntomas de la EP El principal determinante neuropatológico de la EP es la degeneración progresiva de las neuronas dopaminérgicas de la SN del mesencéfalo (figura 2). El patrón de la pérdida de las células dopaminérgicas en la SN es una característica distintiva de la EP, ya que la pérdida celular más notable se produce en la región ventrolateral de la SN pars compacta (SNpc). La pérdida inicial de las neuronas en la zona ventrolateral de la SNpc afecta a las proyecciones dopaminérgicas de la vía nigroestriada que conecta con los ganglios basales. En primer lugar, la degeneración dopaminérgica afecta a la parte posterior del estriado, particularmente al putamen. Posteriormente y de manera progresiva, también afectará a las neuronas que proyectan a la zona anterior del estriado, incluyendo el caudado y el núcleo accumbens. La pérdida notable de las neuronas dopaminérgicas en la SNpc incluso en etapas tempranas de la enfermedad sugiere que la degeneración en estas regiones comienza antes de la aparición de los síntomas, lo cual está respaldado por varios estudios clínico-patológicos (Dijkstra et al., 2014; Iacono et al., 2015). Figura 2. Despigmentación de la SN en la EP. (A) Plano transversal del corte correspondiente al mesencéfalo, donde es visible una porción de la SN; (B) Corte transversal del mesencéfalo y porción transversal de la SN; (C) Despigmentación de la SN en un paciente con EP (panel derecho) en comparación con un sujeto sano control (panel izquierdo). Se muestran secciones C A B 9 macroscópicas (recuadro superior) tras tinción inmunohistoquímica con tirosina hidroxilasa. En el corte histológico es evidente la pérdida selectiva de las partes ventrolaterales de la SN con preservación de las regiones mediales y dorsales. [Figura adaptada de (Poewe et al., 2017)]. Esta neurodegeneración de la vía nigroestriada produce una denervación dopaminérgica de los ganglios basales y conlleva un decremento de los niveles de dopamina en estas regiones anatómicas del estriado, lo que afecta a las conexiones del circuito córtico-estriado-talamo- cortical que conecta los ganglios basales y la corteza. Este fenómeno provoca una disfunción del circuito motor y causa la aparición de los síntomas motores cardinales de la enfermedad (figura 3). Figura 3. Esquema de la vía directa e indirecta del circuito córtico-estriado-talamo- cortical. Esquema simplificado del modelo fisiopatológico clásico de los circuitos de los ganglios basales. (A) En estado sano, las áreas motoras corticales proyectan axones glutamatérgicos al putamen localizado en el estriado y que este a su vez envía proyecciones GABAérgicas al GPi y a la SNpr en dos vías: el “circuito directo” (putamen- GPi-tálamo) y el “circuito indirecto” (putamen-GPe-STN-GPi/SNr-tálamo). Las proyecciones dopaminérgicas de la SNpc activan las neuronas putaminales en la vía directa e inhiben las de la vía indirecta. La activación de la vía directa conduce a una reducción de la activación neuronal en GPi/SNpr y la facilitación del movimiento, mientras que la activación de la vía indirecta los suprime. El STN también se activa mediante una proyección excitadora de la corteza (vía “hiperdirecta”), una conexión que no estaba incluida en el modelo original y no Thalamus GPe STN SNpc SNpr Cortex Striatum GPi Thalamus GPe STN SNpc SNpr Cortex Striatum GPi A. Estado sano B. Enfermedad de Parkinson Circuito directo Circuito indirecto Proyección inhibitoria Proyección excitatoria 10 está representada en la figura para simplificar las conexiones. (B) En la EP, el déficit de dopamina conduce a una mayor actividad en el circuito indirecto (con una hiperactividad del STN como característica clave) y a una hipoactividad en el circuito directo. La acción conjunta de ambos fenómenos provoca una mayor inhibición de las proyecciones del GPi y la SNpr hacia el tálamo y causa un decremento de la activación de las regiones motoras corticales y del tronco encefálico. Abreviaturas: SNpr = Sustancia negra pars reticulada; SNpc = Sustancia negra pars compacta; GPi=Globo pálido interno; GPe=Globo pálido externo; STN=Núcleo subtalámico; Thalamus=Tálamo; Striatum=Estriado (compuesto por caudado y putamen). [Figura adaptada de (Picconi et al., 2018; Rodriguez-Oroz et al., 2009)]. Además de la disfunción motora causada por la afectación de la vía nigroestriada, otros circuitos funcionales de los ganglios basales también se ven comprometidos en mayor o menor medida por el déficit de los niveles de dopamina del sistema dopaminérgico (figura 4). Figura 4. Organización funcional de los circuitos de los ganglios basales. Los ganglios basales pueden dividirse en subregiones con (A) función motora, (B) función asociativa/ejecutiva y (C) función límbica. Estas subregiones están distribuidas a través de las diversas estructuras anatómicas y forman circuitos funcionales según sus conexiones topográficas, como lo resaltan las áreas coloreadas en rojo (corteza motora), verde (corteza prefrontal) y azul (corteza del cíngulo anterior). [Figura adaptada de (Rodriguez-Oroz et al., 2009)]. A. Cirtcuito motor B. Cirtcuito asociativo C. Cirtcuito límbico 11 La denervación del caudado y del estriado ventral en etapas más tardías de la enfermedad afecta a las conexiones fronto-estriatales del circuito asociativo y el límbico (De La Fuente- Fernández, 2013). El circuito asociativo conecta la corteza prefrontal dorsolateral con el núcleo caudado dorsal, cuya integridad es esencial para el correcto funcionamiento cognitivo. De hecho, diversos estudios han encontrado asociaciones entre los niveles de dopamina del caudado y el déficit cognitivo en la EP (incluyendo tareas de memoria ejecutiva y de trabajo) (Brooks & Piccini, 2006). El circuito límbico incluye conexiones del núcleo caudado ventral y del núcleo accumbens con la corteza orbitofrontal, relacionándose con las emociones y los estados de ánimo. Existen estudios que muestran la relación entre los déficits de dopamina en el estriado ventral y trastornos psiquiátricos como depresión, ansiedad, alucinaciones visuales y trastornos del control de impulsos (Grachev, 2014; Kiferle et al., 2014; Vriend et al., 2014; Jaakkola et al., 2017). Estas investigaciones respaldan la posible conexión entre la degeneración progresiva del sistema dopaminérgico y la manifestación de los síntomas cognitivos y conductuales durante el transcurso de la enfermedad. Patología de Lewy y estadios de Braak Otra característica neuropatológica principal y distintiva de la EP es el anormal depósito y acumulación de la proteína 𝛼-sinucleína en su estado mal plegada. En este estado, la 𝛼sinucleína se precipita y se agrega para formar inclusiones intracelulares conocidas como cuerpos de Lewy (CL), si estas aparecen en el soma de la neurona, y neuritas de Lewy (NL), si estas aparecen en las prolongaciones (figura 5). Figura 5. Patología de Lewy en la EP. Tinción inmunohistoquímica de α-sinucleína que muestra (A) los cuerpos de Lewy intra-citoplasmáticos redondos, (A y B) depósitos granulares difusos de α-sinucleína, (B), depósitos en procesos de células neuronales y (C) esferoides de α-sinucleína en los axones. [Figura adaptada de (Poewe et al., 2017)]. A B C 12 Estas inclusiones de"𝛼-sinucleína no sólo afectan a las neuronas del sistema dopaminérgico, sino que otras estructuras implicadas en otros sistemas de neurotransmisión también se ven afectadas. Por ejemplo, se han descrito inclusiones de 𝛼-sinucleína en el locus coeruleus, el núcleo Rafe o el prosencéfalo basal, las cuales se han relacionado con la pérdida de neuronas en dichas estructuras, así como un déficit en la actividad del sistema noradrenérgico, serotoninérgico y colinérgico, respectivamente (Jellinger, 1987; Halliday et al., 1990; Seidel et al., 2015). Basándose en la aparición temporal y espacial de estas inclusiones de 𝛼-sinucleína en las neuronas de las distintas estructuras anatómicas del cerebro afectadas durante la progresión de la enfermedad, Braak y colaboradores han sugerido seis etapas clínico-patológicas que explicarían la progresión de la patología de Lewy (figura 6). Según esta secuencia, la patología de la EP empezaría en la periferia, en el sistema nervioso entérico, y luego seguiría una progresión caudo-rostral afectando primero a los núcleos del nervio vago y glosofaríngeo localizados en la parte inferior del tronco del encéfalo, así como al bulbo olfatorio (fase I y II de Braak). Posteriormente, la patología afectaría a las neuronas del mesencéfalo, incluyendo las neuronas dopaminérgicas de la SNpc, y a las neuronas colinérgicas del prosencéfalo basal colinérgico (fase III y IV de Braak). Finalmente, afectaría a regiones cerebrales límbicas y neocorticales a medida que la progresión de la enfermedad avanza (fase V y VI de Braak) (Braak, Tredici, Rüb, Vos, et al., 2003; Halliday & McCann, 2010). Figura 6. Fases de Braak en la EP. Representación de la progresión teórica de la aparición de las agregaciones de 𝛼-sinucleína en las distintas regiones anatómicas afectadas por la patología de Lewy en la EP. [Figura adaptada de (Poewe et al., 2017)]. Severidad de la patología Cuerpos de Lewy corticales Cuerpos de Lewy en el tronco encefálico Fase I y II de Braak Fase III y IV de Braak Fase V y VI de Braak 13 Desde que los CL y las NL fueron descritos e identificados hace más de un siglo, aún se discute si estos depósitos de 𝛼-sinucleína contribuyen directamente a la neurotoxicidad y a la muerte celular de las neuronas, si surgen como resultante de los mecanismos de defensa de la célula contra la toxicidad de la proteína 𝛼-sinucleína en su forma anómala, o si resultan del fracaso de dichos mecanismos de protección (Gispert-Sanchez & Auburger, 2006). Sin embargo, no cabe lugar a duda de que la existencia de los agregados anómalos de 𝛼-sinucleína, la pérdida gradual de las neuronas dopaminérgicas de la sustancia negra y la consiguiente escasez de dopamina en los circuitos de los ganglios basales son los principales determinantes neuropatológicos de la aparición de los síntomas de la EP. Clínica de la EP Características principales La EP se define clínicamente por la manifestación de los síntomas motores cardinales de la enfermedad, que consiste en la presencia de bradicinesia (lentitud y decremento del movimiento) acompañada de temblor en reposo y/o rigidez muscular. La conjunción de estos síntomas se denomina parkinsonismo y es mandatorio para su diagnóstico. Además, pueden darse otros signos motores como la alteración de reflejos posturales, incluyendo la postura flexionada del tronco y las extremidades, junto con la inestabilidad postural (Poewe et al., 2017). Estas características motoras clásicas se presentan de manera heterogénea y suelen clasificarse según el patrón de manifestación clínica: 1) predominio de temblor, 2) predominio rígido-acinético y 3) fenotipo mixto (Postuma et al., 2015). Diagnóstico y el uso de [I123]FP-CIT SPECT En la práctica clínica, los criterios diagnósticos aceptados actualmente son los establecidos por la International Parkinson’s and Movement Disorder Society (MDS) según los cuales el diagnóstico clínico de la EP requiere la presencia de parkinsonismo y dos criterios de apoyo, así como la ausencia de criterios de exclusión sugerentes de otros parkinsonismos neurodegenerativos distintos a la EP (Postuma et al., 2015). Aunque han pasado más de 200 años desde la primera descripción clínica por James Parkinson, el diagnóstico preciso y definitivo de la EP solo puede confirmarse mediante la valoración neuropatológica de la autopsia (Bloem et al., 2021). Si bien estos criterios diagnósticos clínicos están bien definidos 14 y alcanzan una sensibilidad de más del 90%, en etapas iniciales puede resultar difícil establecer un diagnóstico correcto si los síntomas son sutiles, monosintomáticos (p. ej. un temblor aislado) o equívocos. Por esta razón, en la práctica clínica se usan diversos biomarcadores diagnósticos que ayuden a evaluar in vivo el estado neurodegenerativo del paciente. En este sentido, las técnicas de imagen molecular se utilizan ampliamente para el diagnóstico, permitiendo valorar el estado del circuito dopaminérgico en los pacientes. Como se ha comentado previamente, la integridad del sistema dopaminérgico de la vía nigroestriada juega un papel fundamental en la EP y la imagen molecular de esta vía es un biomarcador preciso de la neurodegeneración existente. La disfunción de los circuitos dopaminérgicos se puede evaluar utilizando diferentes objetivos moleculares en la sinapsis dopaminérgica con tomografía por emisión de positrones (PET) y su versión monofónica (SPECT). Hay varias opciones disponibles, pero el enfoque más utilizado es evaluar el estado de los terminales presinápticos con la imagen del transportador de dopamina (DAT). La ventaja inmediata de esta prueba es que es mínimamente invasiva. El déficit de dopamina de la vía dopaminérgica nigroestriada se puede analizar en las regiones anatómicas de los ganglios basales, particularmente el putamen y el caudado, donde las neuronas de la SNpc finalizan sus proyecciones y se conectan a las neuronas postsinápticas utilizando la dopamina como neurotransmisor. Los DAT se encuentran en las terminales nerviosas presinápticas y son responsables de la recaptación de dopamina de la hendidura sináptica, por lo que la imagen de los DAT permite la evaluación in vivo de la integridad de los terminales nerviosos presinápticos dopaminérgicos (figura 7) (Djang et al., 2012). El I123-2-carbometoxi-3-(4-yodofenil)-N-(3-fluoropropil) nortropano ([I123]FP-CIT o I123- ioflupano) es un marcador para la obtención de imágenes moleculares que se une a los DAT. El [I123]FP-CIT es estable 3-6h después de su administración y la mejor imagen se obtiene 3-4h después de la inyección de este radiotrazador, logrando una alta unión específica al estriado en comparación con el resto del cerebro. Estas características hacen que el [I123]FPCIT SPECT (o DAT SCAN en su nombre comercial) sea ideal para el uso clínico. 15 Figura 7. Imagen molecular de los DAT. A. Representación de la vía nigroestriada que proyecta desde las neuronas dopaminérgicas de la SN del mesencéfalo a los ganglios basales; B. Esquema de la sinapsis dopaminérgica indicando el marcaje presináptico del DAT usando el radiotrazador [I123]FP-CIT; C. Exploraciones axiales de [I123]FP-CIT SPECT de un sujeto control (izquierda) y un paciente con EP (derecha). Puede observarse un déficit de los niveles de los DAT en el estriado del paciente con EP en comparación con el sujeto control. Tratamiento principal de la EP Las terapias actualmente disponibles en la práctica clínica para abordar la EP tratan únicamente la sintomatología de la enfermedad, por lo que no existen tratamientos decisivos que modifiquen su progresión, ya sea para ralentizar o detener el proceso neurodegenerativo subyacente (Kalia & Lang, 2015; Jankovic & Tan, 2020; Bloem et al., 2021). Por ello, las terapias actualmente usadas en la EP tienen como principal objetivo suplir el déficit de dopamina para Sección coronal del cerebro Mesencéfalo Sustancia Negra (Sitio de degeneración dopaminérgica) Vía nigroestriada Nucleo caudado Putamen [I123]FP-CIT Neurona dopaminérgica presináptica Neurona postsináptica CONTROL PACIENTE EP AB C 16 atenuar los síntomas. Éstos son muy eficaces durante los estadios iniciales de la enfermedad generalmente, siendo el uso de la levodopa (o L-DOPA) el tratamiento más usado y el nominado como “gold-standard”. Este fármaco se combina con la carbidopa, que es un inhibidor periférico de la dopa-descarboxilasa, evitando así su conversión sistémica a dopamina y los efectos secundarios derivados de ello. Por otro lado, otros de los fármacos más empleados los constituyen los agonistas dopaminérgicos. Éstos actúan uniéndose a los receptores dopaminérgicos postsinápticos y ejerciendo una acción similar a la función de la dopamina. A su vez, estos tratamientos se pueden complementar con el uso de inhibidores enzimáticos como los inhibidores periféricos de la descarboxilasa, inhibidores de la catecol O- metiltransferasa e inhibidores de la monoamino oxidasa. Dichos fármacos aumentan la disponibilidad de dopamina en el cerebro al inhibir las enzimas que metabolizan este neurotransmisor en las neuronas dopaminérgicas y pueden retrasar o reducir la necesidad de L-DOPA o agonistas dopaminérgicos en las primeras etapas de la enfermedad (figura 8). Figura 8. Dianas de fármacos dopaminérgicos en la EP. Los tratamientos terapéuticos presinápticos incluyen la L-DOPA combinada con inhibidores periféricos de la descarboxilasa (AADC) o la catecol-O-metiltransferasa (COMT). Los inhibidores de la monoamino-oxidasa tipo B (MAO-B) aumentan la disponibilidad sináptica de dopamina (tanto endógena como exógena), mientras que los agonistas de la dopamina actúan postsinápticamente. La flecha discontinua que va desde el torrente sanguíneo al tejido cerebral designa el transporte de L-DOPA por la barrera hematoencefálica (BHE). La flecha discontinua que atraviesa el DAT indica la recaptación de dopamina desde la hendidura sináptica. [Figura adaptada de (Poewe et al., 2017)]. 23 mayor de desarrollar demencia que las personas sin EP en edades avanzadas (Perez et al., 2012). La proporción transversal de pacientes con EP y demencia es del 24% al 31% (Aarsland et al., 2005). No obstante, la prevalencia acumulada de pacientes con EP y demencia en pacientes con una edad media entre 54 a 70 años en el momento del diagnóstico es del 17% a los 5 años después del diagnóstico, del 46% a los 10 años después del diagnóstico y del 83% tras 20 años del diagnóstico (Hely et al., 2008; Williams-Gray et al., 2009, 2013). Los mecanismos fisiopatológicos involucrados en la aparición del deterioro cognitivo y la demencia asociada a la EP aún no están claros. Se han propuesto una gran variedad de hipótesis para subyacer los cambios tisulares asociados con el deterioro cognitivo en la EP, los cuales involucran múltiples fenómenos y mecanismos degenerativos (Aarsland et al., 2021). Por un lado, la neurodegeneración progresiva del sistema dopaminérgico en la EP puede sugerir un factor importante en la aparición de deterioro cognitivo dada su implicación en la disfunción de conexiones fronto-estriatales de los circuitos límbicos y ejecutivos (De La Fuente-Fernández, 2013). No obstante, otros sistemas neuronales incluyendo el colinérgico y el noradrenérgico también se ven afectados desde etapas tempranas de la enfermedad y pueden tener relación directa con la aparición de síntomas cognitivos. Según el modelo clínico-patológico de Braak, la patología de Lewy presente en la EP no solo afecta a las neuronas dopaminérgicas de la SNpc en las fases iniciales, sino que las neuronas del prosencéfalo basal colinérgico y del locus coeruleus noradrenérgico también se ven afectadas desde fases tempranas (Braak, Tredici, Rüb, De Vos, et al., 2003; Braak et al., 2005). En este sentido, una de las cuestiones actuales de investigación es cuándo y dónde la α-sinucleína puede tener un efecto significativo en la cognición de estos pacientes. Por ello, diversos trabajos han tenido como objetivo comprender el patrón de deterioro de los diferentes dominios cognitivos afectos en la EP y si estos se relacionan de manera diferencial con la afectación de los distintos sistemas neuronales. Una síntesis de la investigación en este campo condujo a la hipótesis del "síndrome dual" (Kehagia et al., 2013). Esta hipótesis postula que podría haber una distinción entre deficiencias ejecutivas fronto-estriatales que están mediadas por la degeneración dopaminérgica sin presagiar necesariamente la progresión hacia un deterioro cognitivo grave, mientras que otro fenómeno ocurre con deficiencias en memoria y funciones visuoespaciales mediado por la afectación del sistema colinérgico, con daños en la corteza y lóbulo temporal posterior, que derivará a un síndrome de demencia. 24 Más allá de esta hipótesis, existe una clara evidencia de la notable contribución de los cambios neurodegenerativos del sistema colinérgico como factor clave en el deterioro cognitivo y el desarrollo de demencia durante la progresión de la enfermedad (Pasquini et al., 2021; Bohnen et al., 2022). Ya en una de las primeras descripciones neuropatológicas de la EP, Friedrich Lewy informó de la presencia de inclusiones intraneuronales de α-sinucleína en el núcleo basal de Meynert del prosencéfalo basal colinérgico (PBc) (Lewy, 1913). El PBc es la fuente de una densa red de proyecciones corticales del sistema colinérgico central que modulan la función y la fisiología de sus neuronas corticales diana. El complejo del PBc consta de cuatro grupos principales de neuronas colinérgicas que proyectan axones ipsilaterales: el grupo Ch1/Ch2, formado por los núcleos septales mediales y el brazo vertical de la banda diagonal de Broca que envían axones al hipocampo; el grupo Ch3, formado por el núcleo del brazo horizontal de la banda diagonal de Broca que proporciona inervación a la corteza olfatorio; y el grupo Ch4, formado por el núcleo basal de Meynert que proyecta amplia y organizadamente hacia la corteza cerebral y la amígdala (Ballinger et al., 2016; Ovsepian et al., 2015). Desde estos primeros reportes de autopsia por Lewy, otros estudios neuropatológicos posteriores han identificado la patología de Lewy en el PBc, la atrofia de los núcleos colinérgicos que lo constituyen y la consecuente denervación colinérgica cortical como uno de los mayores determinantes de las manifestaciones cognitivas en la EP (Jellinger, 1987; Liu et al., 2015). Biomarcadores de neuroimagen en la progresión de la EP La presencia de los heterogéneos mecanismos neurodegenerativos implicados en la aparición de estos síntomas demanda la búsqueda de biomarcadores que reflejen los procesos fisiopatológicos subyacentes y permitan profundizar en el entendimiento de la etiología de la EP. Estos biomarcadores ayudarían a la identificación precoz de aquellos pacientes susceptibles a estos síntomas específicos, el desarrollo de nuevas terapias personalizadas y la caracterización de dianas terapéuticas que permitan una modificación de la progresión de la enfermedad. 25 En el sentido riguroso, un biomarcador es una característica específica y medible que actúa objetivamente como un indicador de procesos biológicos normales, procesos patológicos o respuestas a intervenciones (‘Biomarkers and Surrogate Endpoints: Preferred Definitions and Conceptual Framework’, 2001). Se pueden identificar distintos tipos de biomarcadores en función de su utilidad, como son los biomarcadores de riesgo de susceptibilidad (los cuales indican la probabilidad de desarrollar una enfermedad en un individuo), biomarcadores diagnósticos (los cuales detectan o confirman una enfermedad e incluso identifican individuos con un subtipo de enfermedad), o pronósticos (los cuales identifican la probabilidad de un evento durante el transcurso y progresión de una enfermedad en los pacientes). Los tipos de biomarcadores según su naturaleza es amplia, incluyendo clínicos, de imagen, bioquímicos, moleculares, y/o genéticos. Recientemente, han surgido distintas iniciativas y colaboraciones internacionales que potencian la investigación de biomarcadores en la EP. Su objetivo es construir grandes biobancos a disposición de la comunidad investigadora para estudiar y caracterizar exhaustivamente a los pacientes, promoviendo así el descubrimiento y la validación de biomarcadores en la enfermedad. Una de las iniciativas internacionales más importantes en la investigación de la EP es PPMI, por sus siglas en ingles “Parkinson’s Progression Markers Initiative”. PPMI es una cohorte internacional, multicéntrica, prospectiva y longitudinal que recoge una multitud de evaluaciones y exploraciones clínicas así cómo datos bioquímicos, genéticos y de neuroimagen con el objetivo de investigar biomarcadores asociados a la EP. A nivel nacional, COPPADIS (de sus siglas en inglés “COhort of Patients with PArkinson zs DIsease in Spain“) representa uno de los proyectos longitudinales más relevantes en el entendimiento de los mecanismos implicados en la progresión de la EP a través del estudio y la integración de biomarcadores de diversa naturaleza, con especial interés en información clínica objetiva a través de escalas estandarizadas, biomarcadores séricos, perfil genético y datos de neuroimagen. Este proyecto ofrece una gran cohorte multicéntrica de pacientes con EP, cuidadores y controles sanos clínicamente bien caracterizados. Entre los distintos tipos disponibles, las exploraciones de neuroimagen son biomarcadores prometedores ya que permiten evaluar in vivo y de manera no invasiva el estado neurodegenerativo, tanto a nivel molecular como estructural, que ocurre en el sistema nervioso de los pacientes. Los nuevos métodos computacionales de neuroimagen permiten 26 procesar la información procedente de la imagen y cuantificar métricas e índices precisos de la integridad de los sistemas neuronales. Por ello, estos marcadores cuantitativos derivados de la neuroimagen se posicionan como claros candidatos para la investigación de biomarcadores que reflejen los procesos neurodegenerativos durante la progresión heterogénea de la EP (Mitchell et al., 2021). En este sentido, como se ha descrito previamente, el [123I]FP-CIT SPECT es uno de los marcadores basados en neuroimagen molecular más utilizados en la práctica clínica durante el diagnóstico de la enfermedad ya que permite reflejar de manera precisa la degeneración dopaminérgica de la vía nigroestriada a través de la cuantificación de los DAT. Teniendo en cuenta su exactitud para reflejar los mecanismos dopaminérgicos alterados, las medidas cuantitativas del [123I]FP-CIT SPECT pueden ser de especial utilidad no sólo en el momento del diagnóstico, sino también para estudiar la relación entre las reducciones progresivas de los DAT (como índice de denervación dopaminérgica) con la aparición de los síntomas durante el transcurso de la enfermedad (Huertas et al., 2015; Huertas et al., 2017). El uso de radiotrazadores específicos para evaluar la integridad de otros sistemas comprometidos en la progresión de la EP no directamente relacionados con el sistema dopaminérgico también es de especial interés. Por ejemplo, el [18F]FEOBV PET refleja la actividad colinérgica del cerebro a través de la señalización del transportador vesicular de acetilcolina, siendo un marcador sensible a la denervación de los terminales colinérgicos en la EP (Okkels, et al., 2023; Schumacher et al., 2023). Otra modalidad bien establecida para la investigación de los cambios funcionales del cerebro durante la progresión de la enfermedad es el [18F]FDG PET. Esta modalidad permite evaluar el metabolismo celular y se posiciona como una de las pruebas de neuroimagen más útiles para la investigación de la disfunción metabólica cerebral relacionada con los procesos neuropatológicos que ocurren en la EP y otras enfermedades neurodegenerativas (Garcia- Garcia et al., 2012). Por otra parte, las modalidades estructurales de neuroimagen permiten evaluar y cuantificar in vivo la integridad de los distintos tejidos nerviosos. Particularmente, la imagen por resonancia magnética (MRI) con secuencias ponderadas T1 permite evaluar la integridad de las estructuras anatómicas del cerebro y, junto a los nuevos métodos computacionales, permite cuantificar distintas métricas e índices de neurodegeneración, como el volumen de sustancia gris o el espesor de la corteza (Lewis et al., 2016). Otra modalidad estructural a tener en cuenta es la imagen por tensores de difusión (DTI), la cual también permite evaluar la integridad estructural de las regiones anatómicas a través de 27 medidas más sensibles a cambios microestructurales en los tejidos nerviosos, como la difusividad media o la anisotropía fraccional (Schulz et al., 2018). La disponibilidad de modalidades de neuroimagen es amplia y la selección de ellas debe dirigirse por las hipótesis planteadas para el estudio de los mecanismos subyacentes en la sintomatología de la progresión de la EP. Como se ha indicado anteriormente, tanto las DILs como el deterioro cognitivo son unos de los fenómenos clínicos más incapacitantes y con mayor impacto en la calidad de vida de los pacientes con EP y, por ello, se posicionan cómo objeto de estudio en esta tesis a través de diversas modalidades de neuroimagen. Biomarcadores de neuroimagen de las DILs en la EP De acuerdo con el modelo fisiopatológico de las DILs (figura 11), estudios de neuroimagen han podido evidenciar el papel crítico que desempeña la denervación dopaminérgica del cuerpo estriado en el desarrollo de DILs a través del uso de imágenes moleculares de la densidad de los niveles de los DAT, específicamente [18F]FP-CIT PET o [123I]FP-CIT SPECT. Se ha demostrado que particularmente niveles bajos de los DAT en el putamen predicen el desarrollo de DILs en pacientes con EP de novo, en contraste con otras regiones del estriado como el caudado o el estriado ventral (Hong et al., 2014). De hecho, el agotamiento regional de los niveles de los DAT en el estriado puede predecir el momento de inicio de las DILs (Yoo et al., 2018). Además, un índice de asimetría más alto de la región del putamen posterior también se ha asociado con cambios más lentos en las dosis de L-DOPA (Chung et al., 2018). Sin embargo, todavía existe una controversia considerable con respecto a las subregiones exactas del estriado que están más estrechamente involucradas en el desarrollo de las DILs. Por ejemplo, un estudio longitudinal reciente encontró que el agotamiento dopaminérgico involucrado en el desarrollo de DILs no se limita a las regiones putaminales sino que también involucra áreas del caudado (Jeong et al., 2018). Otro estudio diferente encontró que el índice de asimetría del caudado predice un mayor riesgo de desarrollo de DILs, pero no el índice de asimetría del putamen (Eusebi et al., 2018). Curiosamente, un estudio reciente en pacientes con EP de novo no pudo encontrar diferencias significativas en los niveles de los DAT del putamen posterior, putamen anterior o caudado entre pacientes que posteriormente desarrollaron DILs en comparación con aquellos que no las desarrollaron. Esto puede sugerir que la cuantificación de los DAT en las divisiones anatómicas del cuerpo estriado puede no 28 ser lo suficientemente sensible para detectar de manera confiable cambios sutiles en los niveles de los DAT asociados con las DILs en etapas tempranas de la enfermedad (A.-A. Roussakis et al., 2019). Aunque la neuroanatomía del cuerpo estriado se divide clásicamente en caudado, putamen y núcleo accumbens, experimentos en modelos animales sobre las proyecciones axonales del estriado han demostrado que subregiones específicas relacionadas con funciones motoras, sensitivas, límbicas y ejecutivas pueden ser discriminados en función de sus distintos perfiles de conectividad cortical (Alexander et al., 1991; DeLong & Wichmann, 2007; Parent & Hazrati, 1995). Estudios recientes de MRI en humanos sobre la conectividad del cerebro in vivo han demostrado una arquitectura funcional similar del cuerpo estriado en función de los perfiles específicos de conectividad cortical (Draganski et al., 2008; E. Y. Choi et al., 2012). Un estudio realizado por Tziortzi y colaboradores utilizó dicha información de conectividad estriatocortical derivada de imágenes de DTI para desarrollar un atlas que parcela el cuerpo estriado en subregiones funcionales según su perfil de conectividad cortical (Tziortzi et al., 2014a). Es importante destacar que a través de un estudio farmacológico de imágenes PET, los autores validaron la relevancia funcional de su atlas de conectividad del estriado al demostrar que la distribución espacial de la liberación de dopamina inducida por D-anfetamina se correspondía más estrechamente con las subregiones funcionales del estriado derivadas de su conectividad cortical en comparación con las subdivisiones estructurales clásicas. En base a los distintos perfiles de conectividad cortical, el atlas distingue tres subdivisiones funcionales del estriado relacionadas con el sistema límbico, ejecutivo y sensitivo-motor. Además, este atlas subdivide la región sensitivo-motora en 3 subregiones más precisas específicamente relacionadas con las áreas rostrales y caudales de la corteza motora y áreas corticales parietales. El uso de este atlas del estriado permitiría la evaluación y cuantificación de los niveles de los DAT dentro de subregiones definidas según su perfil funcional de conectividad cortical. Esto permitiría investigar qué sistemas funcionales del estriado están específicamente asociados al desarrollo de las DILs en la EP, así como su sensibilidad en comparación con las subdivisiones neuroanatómicas clásicas. De acuerdo con el modelo fisiopatológico de las DILs y su manifestación clínica, el desarrollo de las DILs puede estar específicamente relacionado con cambios en los niveles de los DAT de las subregiones del estriado que están 29 específicamente asociadas con funciones sensitivo-motoras a diferencia de las subregiones relacionadas con funciones cognitivas. Biomarcadores de neuroimagen del deterioro cognitivo en la EP De acuerdo con estudios neuropatológicos reportados previamente (Liu et al. 2015), los índices de neuroimagen estructural sobre la integridad del sistema colinérgico han proporcionado evidencia indirecta in vivo del papel crítico que juega la degeneración colinérgica en el deterioro cognitivo asociado a la EP (Bohnen et al., 2018). Aunque los núcleos colinérgicos no son directamente visibles en los contrastes estructurales de la MRI, el análisis volumétrico de las regiones de interés (ROI) centrado en el PBc reveló una notable atrofia de estas estructuras anatómicas tanto en pacientes con EP como en pacientes con demencia por cuerpos de Lewy (DLB) (Hanyu et al., 2005; S. H. Choi et al., 2012; Grothe et al., 2014). Sin embargo, la degeneración del PBc en fases previas a la demencia de la EP y su relevancia para la aparición de déficits cognitivos sutiles aún no han sido exploradas rigurosamente. Un estudio de neuroimagen usando volumetría manual encontró que los volúmenes reducidos de PBc en pacientes con EP con deterioro cognitivo leve predicen la progresión a la demencia, mientras otros no encontraron diferencias grupales entre pacientes con deterioro cognitivo leve y aquellos con cognición normal (S. H. Choi et al., 2012; Lee et al., 2014). Estudios más recientes utilizaron técnicas automatizadas de morfometría del PBc basadas en mapeos estereotácticos de los núcleos colinérgicos, las cuales pueden proporcionar un análisis in vivo más preciso y potencialmente más sensible de la degeneración del PBc (Barrett et al., 2019; Ray et al., 2018; Schulz et al., 2018). Utilizando este enfoque se pudo demostrar que, aunque los volúmenes del PBc no están reducidos significativamente en pacientes con EP de novo a nivel de grupo, se pueden detectar volúmenes reducidos del PBc en un subconjunto de pacientes. Estos pacientes también muestran un riesgo significativamente mayor de deterioro cognitivo inminente, lo que sugiere que los cambios relevantes para la cognición en el volumen del PBc ya están presentes en etapas tempranas de la enfermedad (Barrett et al., 2019; Ray et al., 2018). 30 A parte de la reciente investigación in vivo sobre la degeneración del PBc en la EP, estudios previos de neuroimagen han demostrado la implicación de otras estructuras cerebrales relevantes para el deterioro cognitivo relacionado con la enfermedad, particularmente el sistema de memoria del hipocampo (Burton et al., 2004; Carlesimo et al., 2012; Melzer et al., 2012). Sin embargo, las contribuciones relativas de estos diferentes sistemas neuroanatómicos en el deterioro cognitivo de la EP siguen siendo en gran medida desconocidos. Además, estudios recientes multimodales de neuroimagen indicaron que las alteraciones de la difusión microestructural medidas mediante imágenes con DTI pueden ser marcadores más sensibles de procesos neurodegenerativos relevantes para la cognición en la EP en comparación con las mediciones volumétricas en la MRI (Carlesimo et al., 2012; Sampedro et al., 2019; Schulz et al., 2018). Por otra parte, estudios de autopsias en otras enfermedades neurodegenerativas han descrito relaciones estrechas entre la degeneración del PBc, el agotamiento colinérgico cortical y la patología cerebral neurodegenerativa (Cullen et al., 1997; Lippa et al., 1999). Por ejemplo, en la enfermedad de Alzheimer (EA) se ha descubierto que la pérdida de neuronas del PBc está estrechamente relacionada con la perdida de sustancia gris cortical (Cullen et al., 1997), cuya asociaciones in vivo se han demostrado posteriormente utilizando mediciones precisas de MRI estructural de la degeneración del PBc y la atrofia cortical (Grothe et al., 2016; Kilimann et al., 2016; Schmitz et al., 2018). La atrofia cortical asociada a la degeneración del PBc se localizó en áreas de asociación heteromodales, las cuales también muestran una denervación colinérgica más pronunciada en la EA. Estas denervaciones colinérgicas fueron evidenciadas mediante PET con trazadores específicos para el sistema colinérgico (Schmitz et al., 2018). Las áreas reportadas incluían el lóbulo temporal medial, la circunvolución del cíngulo, la corteza fronto-parietal y regiones mediales de la corteza temporal. Curiosamente, la atrofia de estas áreas corticales también se ha relacionado consistentemente con el deterioro cognitivo en la EP. Estudios recientes han mostrado que, como grupo, los pacientes con EP en estadios iniciales no muestran grandes diferencias en el espesor cortical en comparación con los controles sanos, pero la reducción específica del espesor cortical en ciertas regiones heteromodales sí muestra una notable asociación significativa con los déficit cognitivos de estos pacientes (Pereira et al., 2014; Filippi et al., 2020). 31 En conjunto, todos estos hallazgos resaltan la variabilidad heterogénea en la degeneración del PBc y la atrofia cortical entre los pacientes con EP, lo cual parece estar estrechamente relacionado con el rendimiento cognitivo de los pacientes. Sin embargo, aún no se ha estudiado la relación entre la degeneración del PBc y la atrofia cortical junto con su impacto en el deterioro cognitivo en la EP. Además, dado que estas asociaciones del PBc también han sido reportadas en la EA y existe una alta comorbilidad de esta patología en pacientes con EP que muestran declive cognitivo (Siderowf et al., 2010), surge la pregunta de si la posible conexión entre la degeneración del PBc y la atrofia cortical asociada en la EP puede estar impulsada principalmente por pacientes con patología de la EA comórbida o si es una característica intrínseca a la patología de la EP. Finalmente, si bien estos estudios previos han contribuido a la investigación de los volúmenes del PBc como biomarcadores de la integridad del sistema colinérgico en la EP, aún es necesario un examen más profundo del impacto de la degeneración de estos núcleos en la disfunción cortical y los déficits cognitivos resultantes que manifiestan los pacientes con EP. Curiosamente, estudios previos de [18F]FDG PET también han demostrado que la disfunción metabólica en diversas regiones corticales cognitivamente relevantes está consistentemente relacionada con el deterioro cognitivo asociado a la EP (Garcia-Garcia et al., 2012; González- Redondo et al., 2014). En este sentido, una creciente acumulación de evidencias sugiere que la disfunción cortical fisiológica en enfermedades que se encuentran dentro del espectro de los trastornos por cuerpos de Lewy puede estar influenciada más por la pérdida de proyecciones colinérgicas provenientes del PBc que por la atrofia cortical regional o la presencia local de cuerpos de Lewy (Nestor, 2007). De hecho, varios estudios en animales han demostrado que las lesiones selectivas de PBc interrumpen la neurotransmisión colinérgica cortical e inducen hipometabolismo en regiones corticales denervadas colinérgicamente, asociándose a su vez con el consiguiente deterioro cognitivo (Browne et al., 2001; Gelfo et al., 2013). Sin embargo, aún no se ha investigado si el hipometabolismo cortical dentro de áreas cognitivamente relevantes puede atribuirse a una degeneración prominente del PBc en pacientes con EP. 32 HIPÓTESIS / HYPOTHESIS 39 Objetivo 2.2. Evaluar las asociaciones entre la difusividad media del PBc y el deterioro cognitivo en pacientes con EP sin demencia de la cohorte multicéntrica COPPADIS mediante el uso de DTI. Objetivo 3. Estudiar la relación entre la degeneración del PBc, la atrofia cortical y el declive cognitivo en pacientes con EP durante las etapas iniciales de la enfermedad. Objetivo 3.1. Evaluar las asociaciones entre los cambios longitudinales del volumen del PBc y el grosor cortical en pacientes con EP de la cohorte PPMI mediante el uso de MRIs estructurales seriadas. Objetivo 3.2. Evaluar los efectos de mediación de la reducción del grosor cortical en las asociaciones entre los cambios longitudinales del volumen del PBc y el declive cognitivo en pacientes con EP de la cohorte PPMI. Objetivo 3.3. Evaluar los efectos de la patología de EA comórbida en las asociaciones entre la degeneración del PBc y la atrofia cortical en pacientes con EP de la cohorte PPMI. Objetivo 4. Estudiar la relación entre la degeneración del PBc, el hipometabolismo cortical y el deterioro cognitivo en pacientes con EP. Objetivo 4.1. Evaluar las asociaciones entre el volumen del PBc y el hipometabolismo cortical cuantificado en pacientes con EP de la cohorte HUVR mediante el uso de MRI estructural y [18F]FDG PET. Objetivo 4.2. Evaluar los efectos de mediación del hipometabolismo cortical en las asociaciones entre el volumen del PBc y el deterioro cognitivo en pacientes con EP de la cohorte HUVR. Objetivo 4.3. Evaluar la correspondencia espacial entre el patrón de hipometabolismo cortical asociado a la degeneración del PBc y la denervación colinérgica cortical cuantificada mediante [18F]FEOBV PET. 40 Así, las hipótesis planteadas fueron evaluadas mediante el desarrollo del objetivo correspondiente en un estudio científico y concluyéndose en la producción científica de un manuscrito para su publicación en revistas internacionales revisadas por pares: • Hipótesis 1 → Objetivo 1 → Estudio 1 → Publicación 1: Labrador-Espinosa MA, Grothe MJ, Macías-García D, Jesús S, Adarmes-Gómez A, Muñoz-Delgado L, Fernández-Rodríguez P, Martín-Rodríguez JF, Huertas I, García-Solís D, Mir P. Levodopa-Induced Dyskinesia in Parkinson’s Disease Specifically Associates With Dopaminergic Depletion in Sensorimotor-Related Functional Subregions of the Striatum. Clin Nucl Med. 2021 Jun 1;46(6):e296- e306. doi: 10.1097/RLU.0000000000003609. PMID: 33782308. • Hipótesis 2 → Objetivo 2 → Estudio 2 → Publicación 2: Grothe MJ*, Labrador-Espinosa MA*, Jesús S, Macías-García D, Adarmes-Gómez A, Carrillo F, Camacho EI, Franco-Rosado P, Lora FR, Martín-Rodríguez JF, Barberá MA, Pastor P, Arroyo SE, Vila BS, Foraster AC, Martínez JR, Padilla FC, Morlans MP, Aramburu IG, Ceberio JI, Vara JH, de Fábregues- Boixar O, de Deus Fonticoba T, Pascual-Sedano B; COPPADIS Study Group; Kulisevsky J, Martínez- Martín P, Santos-García D, Mir P. In vivo cholinergic basal forebrain degeneration and cognition in Parkinson’s disease: Imaging results from the COPPADIS study. Parkinsonism Relat Disord. 2021 Jul;88:68-75. doi: 10.1016/j.parkreldis.2021.05.027. Epub 2021 May 31. PMID: 34144230. • Hipótesis 3 → Objetivo 3 → Estudio 3 → Publicación 3: Labrador-Espinosa MA, Silva-Rodríguez J, Reina-Castillo MI, Mir P, Grothe MJ. Basal Forebrain Atrophy, Cortical Thinning, and Amyloid-β Status in Parkinson’s disease-Related Cognitive Decline. Mov Disord. 2023;10.1002/mds.29564. doi:10.1002/mds.29564. • Hipótesis 4 → Objetivo 4 → Estudio 4 → Manuscrito 4 (enviado): Labrador-Espinosa MA, Silva-Rodriguez J, Okkels N, Horsager J, Castro-Labrador S, Reina-Castillo MI, Muñoz-Delgado L, Franco-Rosado P, Castellano-Guerrero AM, San-Eufrasio M, Macías-García D, Jesús S, Adarmes-Gómez A, Carrillo F, Martín-Rodríguez JF, Roldan-Lora F, García-Solís D, Borghammer P, Mir P, Grothe MJ. Cholinergic basal forebrain atrophy associates with cortical hypometabolism in Parkinson’s disease-related cognitive impairment. Submitted. Oct 2023. 41 OBJECTIVES The overall aim of this PhD thesis was to advance the knowledge of the pathological mechanisms underlying LID and cognitive impairment in PD through the study of quantitative multimodal neuroimaging measures. For this purpose, two main objectives were addressed: (I) to investigate the functional implications of striatal dopamine depletion in the development of LID in PD; and (II) to investigate the associations between cBF degeneration and PD-related cognitive impairment. A local patient cohort from the Hospital Universitario Virgen del Rocío (HUVR) in Seville, a national multicentre cohort from the COPPADIS consortium, and an international multicentre PPMI cohort were used to achieve these aims. Several individual objectives were proposed according to the specific hypotheses: Objective 1. To investigate associations between dopamine depletion in functional striatal subregions and the development of LID in PD patients. Objective 1.1. To assess differences in dopamine levels within functionally-defined striatal subregions between PD patients with and without LID from the cross-sectional HUVR cohort using [{I]FP-CIT SPECT data. Objective 1.2. To validate the observed effects in early stages of the disease by examining longitudinal changes in dopamine levels within functionally-defined striatal subregions between de novo PD patients who did or did not develop LID from the longitudinal PPMI cohort using [123I]FP-CIT SPECT data. Objective 2. To investigate associations between cholinergic basal forebrain degeneration and cognitive impairment in non-demented PD patients. Objective 2.1. To assess associations between cognitive impairment and cholinergic basal forebrain volume in PD patients from the multicentric COPPADIS cohort using MRI data. Objective 2.2. To assess associations between cognitive impairment and mean diffusivity measures of the cholinergic basal forebrain in PD patients from the multicentric COPPADIS cohort using DTI data. 42 Objective 3. To investigate associations between cholinergic basal forebrain atrophy, cortical thinning and cognitive decline in de novo PD patients from the PPMI cohort using longitudinal MRI data and longitudinal cognitive assessments. Objective 3.1. To assess associations between longitudinal changes in cholinergic basal forebrain volume and cortical thinning in PD patients using MRI data. Objective 3.2. To assess the mediating effects of cortical thinning on associations between basal forebrain atrophy and cognitive decline in PD patients. Objective 3.3. To assess the impact of comorbid Alzheimer’s disease on associations between cholinergic basal forebrain degeneration and cortical atrophy in PD patients. Objective 4. To investigate associations between cholinergic basal forebrain degeneration, cortical hypometabolism and cognitive impairment in PD patients. Objective 4.1. To assess associations between cholinergic basal forebrain volume and cortical hypometabolism in PD patients from the HUVR cohort using MRI and [18F]FDG PET. Objective 4.2. To assess the mediating effects of cortical hypometabolism on associations between cholinergic basal forebrain volume and cognitive impairment in PD patients from the HUVR cohort. Objective 4.3. To assess whether cortical hypometabolism associated with cholinergic basal forebrain degeneration in PD patients corresponds to areas of cortical cholinergic denervation in PD as measured by [18F]FEOBV PET. 43 Thus, each hypothesis was evaluated through the development of the corresponding main objective in a scientific study and concluded with the scientific production of a manuscript for publication in peer-reviewed international journals. • Hypothesis 1 → Objective 1 → Study 1 → Publication 1: Labrador-Espinosa MA, Grothe MJ, Macías-García D, Jesús S, Adarmes-Gómez A, Muñoz-Delgado L, Fernández-Rodríguez P, Martín-Rodríguez JF, Huertas I, García-Solís D, Mir P. Levodopa-Induced Dyskinesia in Parkinson’s Disease Specifically Associates With Dopaminergic Depletion in Sensorimotor-Related Functional Subregions of the Striatum. Clin Nucl Med. 2021 Jun 1;46(6):e296- e306. doi: 10.1097/RLU.0000000000003609. PMID: 33782308. • Hypothesis 2 → Objective 2 → Study 2 → Publication 2: Grothe MJ*, Labrador-Espinosa MA*, Jesús S, Macías-García D, Adarmes-Gómez A, Carrillo F, Camacho EI, Franco-Rosado P, Lora FR, Martín-Rodríguez JF, Barberá MA, Pastor P, Arroyo SE, Vila BS, Foraster AC, Martínez JR, Padilla FC, Morlans MP, Aramburu IG, Ceberio JI, Vara JH, de Fábregues- Boixar O, de Deus Fonticoba T, Pascual-Sedano B; COPPADIS Study Group; Kulisevsky J, Martínez- Martín P, Santos-García D, Mir P. In vivo cholinergic basal forebrain degeneration and cognition in Parkinson’s disease: Imaging results from the COPPADIS study. Parkinsonism Relat Disord. 2021 Jul;88:68-75. doi: 10.1016/j.parkreldis.2021.05.027. Epub 2021 May 31. PMID: 34144230. • Hypothesis 3 → Objective 3 → Study 3 → Publication 3: Labrador-Espinosa MA, Silva-Rodríguez J, Reina-Castillo MI, Mir P, Grothe MJ. Basal Forebrain Atrophy, Cortical Thinning, and Amyloid-β Status in Parkinson’s disease-Related Cognitive Decline. Mov Disord. 2023;10.1002/mds.29564. doi:10.1002/mds.29564. • Hypothesis 4 → Objective 4 → Study 4 → Manuscript 4 (submitted): Labrador-Espinosa MA, Silva-Rodriguez J, Okkels N, Horsager J, Castro-Labrador S, Reina-Castillo MI, Muñoz-Delgado L, Franco-Rosado P, Castellano-Guerrero AM, San-Eufrasio M, Macías-García D, Jesús S, Adarmes-Gómez A, Carrillo F, Martín-Rodríguez JF, Roldan-Lora F, García-Solís D, Borghammer P, Mir P, Grothe MJ. Cholinergic basal forebrain atrophy associates with cortical hypometabolism in Parkinson’s disease-related cognitive impairment. Submitted. Oct 2023. 44 METODOLOGÍA 45 MÉTODOS En este apartado se resumen brevemente los métodos principales empleados en los trabajos de investigación presentados en esta tesis para abordar los objetivos planteados conforme a las hipótesis de estudio. Una exposición detallada y exhaustiva de estos métodos se indica en el apartado de publicaciones. En primer lugar, se describe brevemente la metodología común empleada para extraer las medidas cuantitativas de las diferentes modalidades de imagen utilizadas en los trabajos. Después, se detallan brevemente los métodos específicos de cada uno de los estudios desarrollados. Métodos comunes Cuantificación de los niveles de los DAT del estriado usando [123I]FP-CIT SPECT Las imágenes [123I]FP-CIT SPECT utilizadas en esta tesis doctoral fueron adquiridas siguiendo los protocolos estandarizados para esta modalidad en los escáneres correspondientes disponibles en las cohortes de estudio (ver publicación del estudio 1). El procesamiento de las imágenes [123I]FP-CIT SPECT se llevó a cabo utilizando procedimientos de procesamiento automatizados y estandarizados mediante SPM12 (Wellcome Centre for Human Neuroimaging, Institute of Neurology, UCL, London, United Kingdom) ejecutado en MATLAB 2018a (MathWorks, Natick, MA). En primer lugar, las imágenes se reorientaron estableciendo la comisura anterior como el origen del sistema de coordenadas. Luego, cada exploración se normalizó espacialmente al espacio estándar estereotáctico del Instituto Neurológico de Montreal (MNI) utilizando una plantilla de [123I]FP-CIT desarrollada previamente por (García-Gómez et al., 2013). Las imágenes resultantes se re-escalaron a una matriz de 91x109x91 de voxels de 2x2x2 mm³. La ratio de unión específica de [123I]FP-CIT (SBR, de sus siglas en inglés “Specific Binding Ratio”) se calculó para cada voxel de la imagen cerebral utilizando la siguiente fórmula: SBR = [(valor de captación de radioligando del voxel - valor medio de captación de radioligando del lóbulo occipital) / valor medio de captación de radioligando del lóbulo occipital] (Innis et al., 2007). Los niveles de los DAT se cuantificaron mediante el SBR medio en las diferentes subregiones funcionales del estriado, tal como se mapeó en el atlas espacial en el espacio MNI 46 desarrollado por Tziortzi (Tziortzi et al., 2014). Este atlas subdivide el estriado en subregiones según sus patrones diferenciales de conectividad cortical con áreas límbicas, ejecutivas y sensitivo-motoras. Así, la subregión estriatal límbica está conectada con el giro orbital, el giro recto y el giro subcalloso así como la corteza insular anterior ventral; la subregión ejecutiva con el lóbulo frontal superior y la corteza prefrontal dorsal; y las regiones del estriado relacionadas con la función sensitivo-motora se subdivide aún más en una subregión motora rostral con conectividad con el área rostral 6, el área motora suplementaria y la región del campo ocular frontal; una subregión motora caudal conectada con el giro precentral; y una subregión parietal conectada con el lóbulo parietal. Cuantificación del volumen de regiones de interés y del grosor cortical usando MRI Las MRIs utilizadas en esta tesis doctoral fueron adquiridas siguiendo los protocolos estandarizados para esta modalidad en secuencias ponderadas T1 para los escáneres de resonancia magnética disponibles en las cohortes de estudio (ver publicaciones de los estudios 2, 3 y 4). Estas exploraciones de MRI estructural ponderadas en T1 fueron utilizadas para evaluar el volumen de las regiones de interés (ROIs) de estudio, particularmente el PBc y el hipocampo, utilizando procedimientos de procesamientos automatizados y estandarizados para morfometría basada en voxel implementados en la Computational Anatomy Toolbox (CAT12, Hospital Universitario de Jena) ejecutándose bajo el paquete de Statistical Parametric Mapping (SPM12, Centro Wellcome Trust para Neuroimagen) en Matlab R2018a (MathWorks, Natick, MA). En resumen, las exploraciones de MRI se segmentaron en particiones de sustancia gris (GM), sustancia blanca (WM) y líquido cefalorraquídeo (CSF). Los mapas cerebrales de estas segmentaciones 3D tisulares fueron deformados y modulados para su alineación en alta dimensionalidad al espacio estándar MNI. El volumen intracraneal total (TIV) se calculó como medida del tamaño de la cabeza del sujeto mediante la suma total de las particiones de GM, WM y CSF. El volumen cerebral se calculó mediante la suma de las particiones de GM y WM. En caso de trabajar con MRI seriadas longitudinalmente se realizó el procedimiento estandarizado análogo para procesamiento longitudinal de MRI usando las mismas herramientas. 47 El volumen del PBc se extrajo automáticamente sumando los valores de los voxeles de GM modulados dentro de ROIs definidas por información combinada de atlas estereotácticos existentes de los núcleos del PBc en el espacio MNI. Este atlas se basa en un estudio metodológico previo que caracteriza subdivisiones funcionalmente homogéneas dentro del PBc humano, permitiendo cuantificar el volumen total del PBc, la región anatómica anteriormedial (PBc anterior o región anterior del PBc, que corresponde principalmente al núcleo septal medial y a la banda diagonal de Broca) y la región anatómica posterior-lateral (PBc posterior o región posterior del PBc, que corresponde principalmente al núcleo basal de Meynert) (Fritz et al., 2018). El volumen del hipocampo también fue cuantificado automáticamente mediante la suma de los valores de los voxeles de GM modulados dentro de la ROI del hipocampo definida en el atlas de Desikan-Killiany (Desikan et al., 2006). Estas exploraciones de MRI estructural ponderadas en T1 también fueron utilizadas para evaluar el espesor cortical regional utilizando flujos de procesamientos estandarizados de neuroimagen para morfometría basada en superficies (SBM) implementada también en CAT12. Brevemente, se estimaron mapas de grosor cortical a nivel de vértices en una malla de alta resolución de 164k vértices y se suavizaron utilizando un kernel gaussiano de 15 mm isométrico. Cuantificación de la difusividad media de regiones de interés usando DTI Las DTI utilizadas en esta tesis doctoral fueron adquiridas siguiendo los protocolos estandarizados para esta modalidad en los escáneres de resonancia magnética disponibles en las cohortes de estudios (ver publicación del estudio 2). El procesamiento de los datos de DTI se implementó utilizando la caja de herramientas DTI de FSL (Versión 5.0, FMRIB, Oxford, Reino Unido) e incluyó la corrección de los efectos de corriente inducida, el movimiento de la cabeza, la separación entre cráneo y tejido cerebral y el cálculo de los tensores de difusión para derivar mapas escalares de difusividad media. Los mapas de difusividad media se normalizaron espacialmente al espacio estándar MNI utilizando los campos de deformación calculados en la normalización de las exploraciones de MRI ponderadas en T1, co-registradas anteriormente con las imágenes de DTI. 48 Los índices individuales de difusividad media del PBc y del hipocampo se calcularon extrayendo los valores promedio de los voxeles en los mapas normalizados dentro de las respectivas ROIs anteriormente descritas (Fritz et al., 2018; Desikan et al., 2006). Cuantificación del metabolismo cortical usando [18F]FDG PET Las imágenes [18F]FDG PET utilizadas en esta tesis doctoral fueron adquiridas siguiendo los protocolos estandarizados para esta modalidad en los escáneres disponibles de la cohorte de estudio (ver publicación del estudio 4). El procesamiento de las imágenes [18F]FDG PET se llevó a cabo utilizando procedimientos de procesamientos automatizados y estandarizados mediante SPM12 (Wellcome Centre for Human Neuroimaging, Institute of Neurology, UCL, London, United Kingdom) ejecutado en MATLAB 2018a (MathWorks, Natick, MA). Cada exploración de [18F]FDG PET se co-registró con la exploración de MRI estructural del sujeto correspondiente y se corrigió para efectos de volumen parcial (PVEs) utilizando la herramienta PETPVE12 en SPM12 (Gonzalez-Escamilla et al., 2017). Las exploraciones de [18F]FDG PET corregidas por PVE fueron deformadas y alineadas al espacio de referencia MNI utilizando los campos de flujo de deformación derivados del alineamiento al espacio MNI de las exploraciones de las MRI correspondientes. Finalmente, las intensidades de los voxeles de las imágenes [18F]FDG PET normalizadas se reescalaron en proporción a los valores de captación del puente troncoencefálico para calcular las ratios de valor de captación estandarizados en cada voxel (SUVR, por sus siglas en inglés “Standardized Uptake Value Ratio”). Finalmente, las imágenes se suavizaron con un kernel gaussiano de 8 mm isométrico. Material y métodos de cada estudio individual Estudio 1. “Levodopa-induced dyskinesia in Parkinson’s disease specifically associates with dopaminergic depletion in sensorimotor-related functional subregions of the striatum” Con el propósito de llevar a cabo el primer objetivo de esta tesis doctoral, se planteó este primer estudio que incluía dos cohortes independientes de pacientes con EP. La primera cohorte incluía un total de 185 pacientes con EP de la cohorte del HUVR en función de la disponibilidad de una exploración [123I]FP-CIT SPECT, la cual fue adquirida 2.67±1.9 años 55 volumen de la PBc. Además, se cuantificaron los índices de R2 descompuestos para cada variable del modelo completo. También se utilizaron análisis de mediación para investigar si la asociación entre la atrofia del PBc y el deterioro cognitivo en la EP está mediada por el hipometabolismo cortical regional. Primero, se evaluó la asociación de la atrofia del PBc y el deterioro cognitivo. Luego, se calculó el hipometabolismo cortical promedio relacionado con el PBc para cada paciente y se evaluó si el hipometabolismo cortical relacionado con el PBc también se asocia con el deterioro cognitivo. Todas estas asociaciones se realizaron utilizando modelos de regresión lineal ajustados por sexo, edad, educación, TIV (cuando se utilizaron variables de volumen basadas en MRI) y escáner PET (cuando se utilizaron variables de metabolismo). Se calcularon modelos de mediación utilizando el volumen del PBc como variable causal, la puntuación de la PD-CRSTotal como variable objetivo y el hipometabolismo cortical como variable mediadora. También se calcularon modelos de mediación adicionales para las subpuntuaciones fronto-subcorticales y posterior-corticales de la PD-CRS. Además, se realizó un análisis de sensibilidad para evaluar la solidez de las asociaciones significativas utilizando modelos de regresión completamente controlados por varios posibles factores clínicos de confusión (años de duración de la enfermedad y UPDRS-III-off) y modelos de regresión excluyendo a los pacientes con EP y demencia. Finalmente, la correspondencia espacial entre el patrón de hipometabolismo cortical asociado al PBc y el patrón de denervación colinérgica en la EP fue evaluada a través de la correlación de Spearman de los valores promedios cuantificados en las 100 regiones corticales funcionales. 56 RESULTADOS 57 RESULTADOS En este apartado se resumen brevemente los resultados principales obtenidos en los trabajos de investigación presentados en esta tesis para abordar los objetivos planteados e hipótesis de estudio. Una exposición detallada y más amplia se expone en el apartado de publicaciones. Resumen de resultados del estudio 1: “Levodopa-induced dyskinesia in Parkinson’s disease specifically associates with dopaminergic depletion in sensorimotor-related functional subregions of the striatum” En este estudio se investigó las diferencias en los niveles de los DAT en subregiones funcionales del estriado entre pacientes con EP con DILs y sin ellas con diversos grados de progresión de la enfermedad utilizando datos transversales de [123I]FP-CIT SPECT de la cohorte de pacientes del HUVR. Además, se validaron los efectos específicos en etapas tempranas de la enfermedad mediante la evaluación de los cambios longitudinales de los niveles de los DAT en estas subregiones funcionales entre pacientes recien diagnosticados con EP que luego desarrollaron DILs y los que no utilizando datos longitudinales de [123I]FPCIT SPECT de la cohorte de pacientes de PPMI. Los resultados principales obtenidos en la cohorte transversal del HUVR mostraron que: - Los niveles medios de los DAT en subregiones del estriado relacionadas con la función sensitivo-motora fueron significativamente más bajos en el grupo de pacientes con EP y DILs en comparación con el grupo de pacientes sin DILs: región sensitivo-motora completa (F=5.99, p=0.016), región motora rostral (F=4.51, p=0.035) , región motora caudal (F=5.70, p=0.018) y región parietal (F=6.73, p=0.01). - Los niveles medios de los DAT en la subregión del estriado relacionada con la función ejecutiva también fueron significativamente más bajos en pacientes con DILs en comparación con pacientes sin DILs (F=5.30, p=0.023). - Las diferencias grupales en los niveles de los DAT en la subregión del estriado relacionada con la función límbica alcanzaron sólo significación estadística a nivel de tendencia (F=3.33, p=0.07). 58 - Comparativamente con las regiones anatómicas del estriado, los niveles de los DAT en el putamen mostraron diferencias significativas entre grupos (F=6.30, p=0.01), pero las diferencias en caudado sólo alcanzaron significación estadística a nivel de tendencia (F=3.27, p=0.07). Los resultados principales obtenidos en la cohorte longitudinal de PPMI mostraron que: - Los niveles de los DAT en la subregión del estriado relacionada con la función sensitivo-motora fueron significativamente más bajos en pacientes con EP que desarrollaron DILs en comparación con pacientes que no la desarrollaron (efecto de grupo: β=-0.06, t=-2.05, p=0.041) durante el seguimiento longitudinal. - Los niveles de los DAT en las subregiones del estriado relacionadas con las funciones ejecutivas y límbicas no fueron significativamente diferentes entre pacientes con EP que desarrollaron DILs en comparación con pacientes que no la desarrollaron durante el seguimiento longitudinal. - Entre las diferentes subregiones del estriado relacionadas con la función sensitivomotora, los niveles de los DAT en las subregion motora rostral (efecto de grupo: β=-0.08, t=-2.14, p=0.033) y motora caudal (efecto de grupo: β=-0.06, t=-1.98, p=0.049) fueron significativamente menores en pacientes con EP que desarrollaron DILs que en pacientes que no las desarrollaron, pero las diferencias grupales en la subregión parietal solo fueron significativas a nivel de tendencia (efecto de grupo: β=-0.04, t=-1.86, p=0.064). - Los niveles de los DAT en todas las subregiones estriatales disminuyeron significativamente con el tiempo (efecto temporal para todas las subregiones: p<0.001), pero el cambio longitudinal no difirió entre los grupos de pacientes con EP que desarrollaron DILs y aquellos que no las desarrollaron en ninguna de las subregiones funcionales del estriado (efecto interacción grupo-tiempo para todas las subregiones: p > 0.32). - Comparativamente con las regiones anatómicas del estriado, los niveles de los DAT en el caudado y el putamen no mostraron diferencias significativas entre grupos durante el seguimiento longitudinal. - 59 Resumen de resultados del estudio 2: “In vivo cholinergic basal forebrain degeneration and cognition in Parkinson’s disease: Imaging results from the COPPADIS study” En este estudio se investigó la relación entre los volúmenes del PBc basados en medidas de MRI y el deterioro cognitivo en una cohorte multicéntrica bien caracterizada de pacientes con EP sin demencia. Además, se exploró más a fondo la utilidad de los índices de integridad microestructural basadas en la difusividad media de exploraciones de DTI para detectar cambios tisulares relevantes relacionados con el déficit cognitivo asociada a la EP. Estas medidas multimodales también fueron cuantificadas en el hipocampo para su uso como región cognitiva control en los resultados. Los análisis principales para la evaluación de estas asociaciones mostraron que: - La reducción del volumen del PBc se asocia significativamente con una menor puntuación total de la PD-CRS (rparcial = 0.37, p < 0.001), incluso tras corregir por variables clínicas de confusión. - Entre las asociaciones de las puntuaciones individuales de los dominios cognitivos de la PD-CRS con el volumen del PBc, el volumen del PBc se asoció notablemente con la memoria inmediata (rparcial=0.37, p<0.001), pero también con la memoria diferida (rparcial=0.31, p<0.001), la memoria de trabajo (rparcial=0.33, p<0.001) y la fluidez verbal alternante (rparcial=0.27, p<0.001). - En los análisis específicos para subregiones del PBc, todas las asociaciones con las puntuaciones individuales de los dominios cognitivos de la PD-CRS fueron ligeramente más pronunciadas para la subregión posterior del PBc que para la subregión anterior del PBc, pero ninguna de las diferencias entre estas subregiones fue estadísticamente significativa (p > 0.83). - El volumen del hipocampo también se asoció con la puntuación total de la PD-CRS (rparcial=0.24, p=0.002) y con las puntuaciones de varios dominios cognitivos individuales, pero ninguna de estas asociaciones se mantuvo tras corregir por variables clínicas de confusión. 60 - La difusividad media del PBc no se asoció significativamente con la puntuación total de la PD-CRS (rparcial=−0.10, p=0.42) o puntuaciones individuales de los dominios cognitivos (p>0.27). - La combinación multimodal de mediciones volumétricas y de difusividad de la integridad del PBc en un modelo de regresión multimodal para su asociación con la puntuación total de la PD-CRS no aumentó el ajuste en comparación con el modelo basado únicamente en el volumen del PBc (ΔR2=0.003, p=0.60). - La difusividad media del hipocampo se asoció significativamente con la puntuación total de la PD-CRS (rparcial=−0.43 , p<0.001), principalmente dirigida por las asociaciones con las puntuaciones de fluidez verbal alternante (rparcial=−0.44, p=0.001). - La combinación multimodal de mediciones volumétricas y de difusividad de la integridad del hipocampo en un modelo de regresión multimodal para su asociación con la puntuación total de la PD-CRS no aumentó significativamente el ajuste en comparación con el modelo basado únicamente en la difusividad del hipocampo (ΔR2=0.034, p=0.23). 61 Resumen de resultados del estudio 3. “Basal forebrain atrophy, cortical thinning, and amyloid-β status in Parkinson’s diseaserelated cognitive decline” En este estudio se investigó las asociaciones longitudinales in vivo entre la degeneración del PBc, la atrofia cortical y el declive cognitivo en una gran muestra de pacientes con EP de novo de la cohorte PPMI. La integridad del PBc y de la corteza cerebral fue cuantificada mediante el volumen del PBc y el espesor cortical, respectivamente, usando MRI. Además, se evaluó si estas asociaciones dependen de la presencia de la patología de EA comórbida, la cual se identificó mediante un biomarcador de β-amiloide en el LCR. Los principales resultados del trabajo mostraron que: - Como grupo, los pacientes con EP mostraron una disminución longitudinal significativa en el volumen del PBc (t=5.41, p<0.001), el espesor cortical regional y el rendimiento cognitivo global (MoCA: t=2.89, p=0.004) así como en el rendimiento de memoria (memoria diferida: t=2.41, p=0.017) y atención/memoria de trabajo (SDM: t=4.35, p<0.001; LNS: t=2.35, p=0.02). En cambio, los sujetos sanos no mostraron una disminución longitudinal significativa en ninguna de estas métricas. - La disminución longitudinal del volumen del PBc en los pacientes con EP se asoció significativamente con la reducción del espesor cortical longitudinal en áreas generalizadas, con efectos pronunciados en la corteza prefrontal medial y lateral, el cíngulo posterior, el precúneo, las áreas temporo-parietales laterales y el lóbulo temporal medial, incluida la corteza entorrinal, el giro fusiforme y el giro lingual. - La degeneración longitudinal del PBc en los pacientes con EP se asoció significativamente con el declive cognitivo global en el rendimiento de MoCA (r=0.30, p<0.001), así como con el declive cognitivo de dominios específicos de memoria (memoria diferida: r=0.27, p<0.001). 62 - La reducción del espesor cortical en áreas relacionadas con la degeneración del PBc también se asoció significativamente con el declive cognitivo global, con efectos pronunciados en áreas frontales mediales y laterales, áreas temporoparietales laterales, corteza entorrinal, cúneo, precúneo y cíngulo posterior (tamaño del efecto promedio: r=0.37, p<0.001). - La reducción media del espesor cortical en estas áreas relacionadas con la cognición y el PBc medió los efectos de la degeneración del PBc sobre el deterioro cognitivo global (efecto total: β=0.16 [0.08, 0.25], p<0.001, ADE: β=0.08 [0.01, 0.17], p=0.064; ACME: β=0.08 [0.01, 0.16], p=0.02; proporción mediada: 48,5%). - Tanto los pacientes positivos como los pacientes negativos para β-amiloide mostraron asociaciones significativas entre la degeneración longitudinal del PBc y la atrofia cortical paralela. Algunos de los efectos regionales fueron significativamente más pronunciados en pacientes con EP y positivos para βamiloide en comparación con los pacientes con EP y negativos para β-amiloide, particularmente en las áreas del cíngulo posterior, lóbulo parietal superior y lóbulo frontal lateral. 63 Resumen de resultados del estudio 4. “Cholinergic basal forebrain atrophy associates with cortical hypometabolism in Parkinson’s disease-related cognitive impairment” En este estudio se investigó las asociaciones multimodales in vivo entre la degeneración del PBc y el hipometabolismo cortical regional en una gran muestra de pacientes con EP con diversos grados de deterioro cognitivo mediante el uso de MRI y [18F]FDG PET. Además, se evaluó si el hipometabolismo cortical asociado al PBc media los efectos de la atrofia del PBc en el deterioro cognitivo asociado a la EP. Por último, se examinó la correspondencia espacial entre los patrones de hipometabolismo cortical asociado al PBc y la denervación colinérgica cortical medida por la diferencia del [18F]FEOBV PET entre controles sanos y pacientes con EP. Los principales resultados del trabajo mostraron que: - El volumen del PBc posterior, pero no el del PBc anterior, se asoció significativamente con un hipometabolismo cortical extenso, con efectos pronunciados en las regiones fronto-parietales laterales y mediales, lóbulo temporal lateral y áreas posteriores, incluidas las regiones del cíngulo, cúneo, corteza lingual y corteza occipital lateral (tamaño del efecto promedio: rparcial=0.45, p<0.001). - Por el contrario, el volumen del hipocampo se asoció selectivamente con el hipometabolismo del propio hipocampo y regiones del circuito límbico (áreas temporales y frontales mediales). - En los análisis adicionales que evaluaron la influencia de la atrofia cortical local, combinar tanto el volumen del PBc posterior como el volumen local de GM cortical en un modelo de regresión multimodal para la asociación con el hipometabolismo cortical incrementó ligeramente el ajuste en comparación con el modelo basado únicamente en el volumen del PBc (R2 global: 0.468 vs 0.503, ΔR2=0.035 (7.5%), p=0.01). Sin embargo, aunque el volumen local de la GM cortical local fue un predictor del hipometabolismo cortical (varianza parcial explicada: t=2.51, p=0.01, R2 descompuesto=0.16, 31% de la varianza total), el volumen posterior del PBc 64 mostró una influencia mayor en comparación (varianza parcial explicada: t=4.42, p<0.001, R2 descompuesto=0.26, 52% de la varianza total). - El deterioro cognitivo global en los pacientes con EP se asoció significativamente con la atrofia del PBc posterior (rparcial=0.36, p=0.001) así como con el hipometabolismo cortical promedio asociado con el PBc posterior (rparcial=0.40, p<0.001). El hipometabolismo cortical asociado con el PBc medió los efectos de la degeneración del PBc posterior en el deterioro cognitivo global (efecto total: β=0.63 [0.44, 0.82], p<0.001; ADE: β =0.30, [0.09,0.52], p=0.008; ACME: β=0.31 [0.14,0.51], p=0.002; proporción mediada: 50.1%, p=0.006). - Los efectos mediadores del hipometabolismo cortical asociado al PBc también fueron observados para las asociaciones entre la degeneración del PBc posterior y los déficits cognitivos en las subpuntuaciones fronto-subcorticales y posteriorcorticales, sin observarse diferencias significativas entre las asociaciones de ambas subpuntuaciones (p≥0.94). - En análisis de sensibilidad adicionales para evaluar la solidez de las asociaciones frente a la gravedad de la enfermedad, se observaron asociaciones similares utilizando modelos de regresión que controlaron adicionalmente por los años de progresión de la enfermedad y la gravedad de los síntomas motores medidos por UPDRS-III en “off”, incluso excluyendo a los pacientes con EP y demencia. - Las diferencias de [18F]FEOVB PET entre controles sanos y pacientes con EP mostraron un patrón de denervación colinérgica similar al hipometabolismo cortical asociado al PBc. Los efectos más pronunciados del agotamiento regional de [18F]FEOBV se observaron en las áreas parieto-temporo-occipitales y en el lóbulo frontal superior lateral. El análisis formal de la correspondencia espacial reveló una correlación significativa elevada entre el patrón del hipometabolismo cortical asociado al PBc y el patrón de agotamiento regional de [18F]FEOBV (r de Spearman= 0.57, p <0.001). 71 PD + LID with trend-level statistical significance. PD + LID patients had significantly higher L-DOPA doses (P< 0.001) and higher total LED (P< 0.001), but did not differ from PD-LID in LED of dopaminergic agonists. There were no significant differences between groups in any other clinical features. Differences in DAT Levels of Functional Striatal Subregions Between PD-LID and PD + LID Patients in the Local PD Cohort Mean DAT levels for each group in limbic, executive, and whole sensorimotor, as well as rostral-motor, caudal-motor, and parietal striatal subregions, are summarized in Table 2. As hypothesized, mean DAT levels in sensorimotor-related striatal subregions were significantly lower in the PD + LID group compared with the PD-LID group: whole sensorimotor (F= 5.99, P= 0.016), rostral-motor (F= 4.51, P= 0.035), caudal-motor (F= 5.70, P= 0.018), and parietal (F= 6.73, P= 0.01). Moreover, mean DAT levels in the executive-related striatal subregion were also significantly lower in PD + LID compared with PD-LID (F= 5.30, P= 0.023). Group differences in DAT levels in the limbic-related striatal subregion reached only trend-level statistical significance (F= 3.33, P= 0.07). Demographics and Clinical Features of PD-LID and PD + LID Patients in the PPMI Cohort Demographics and clinical characteristics of the validation cohort of de novo PD patients are summarized in Table 3. The mean age at onset of PD was significantly earlier in the PD + LID group compared with the PD-LID group (59.8 ± 9.4 vs 62.3 ± 9.8, P= 0.015). PD + LID patients also had significantly higher baseline UPDRS-III score (22.1 ± 8.3 vs 19.4 ± 9.2, P= 0.003) and H&Y stage (1.5 [interquartile range {IQR}, 1–2] vs 2 with [IQR, 1–2]; P= 0.033) than PD-LID patients. There were no significant differences between groups in any other demographic or clinical features at baseline. Longitudinal changes in motor symptoms and dopaminergic therapy of the PD + LID and PD-LID groups are illustrated in Figure 3. In mixed linear models, the UPDRS-III score was on average significantly higher in PD + LID patients compared with PD-LID patients (effect of group: β= 3.14, t= 2.96, P= 0.003). The UPDRS-III scores significantly increased over time in both groups (effect of time: β= 2.64, t= 12.85, P< 0.001), but this increase was less pronounced for the PD + LID group (group !time interaction: β=−0.69, t=−2.35, P= 0.018). There was a significant group effect on L-DOPA doses and total LED, being significantly higher in PD + LID patients compared with PD-LID patients since first year of drug initiation (L- DOPA doses: β= 95.19, t= 2.27, P= 0.023; total LED: β= 112.80, t= 2.65, P= 0.008). Levodopa doses and total LED significantly increased over time in both groups (L-DOPA doses: β= 67.10, t= 8.39, P< 0.001; total LED: β= 88.38, t= 10.99, P< 0.001), butthis increase was significantlyhigher in the PD + LID group compared with the PD-LID group (group !time interaction: L-DOPA doses: β= 32.7, t= 2.89, P= 0.004; total LED: β= 26.96, t= 2.37, P= 0.018). L-DOPA equivalent doses of dopaminergic agonists were not significantly different between groups. Differences in DAT Levels of Functional Striatal Subregions Between PD-LID and PD + LID Patients in the PPMI Cohort Baseline and longitudinal measurements of subregional striatal DAT measurements in the PD + LID and PD-LID groups from the PPMI cohort are summarized in Table 4 and illustrated in Figure 4. After correcting for sex, age at SPECT, UPDRS-III, and L-DOPA doses, DAT levels in the sensorimotor-related striatal subregion were significantly lower in PD + LID patients compared with PD-LID patients (group effect: β=−0.06, t=−2.05, P= 0.041), but no significant group effects were observed for the limbic- and executive-related striatal subregions. Among the TABLE 1. Demographics and Clinical Characteristics of the Local PD Cohort Variables PD-LID (n= 112) PD + LID (n= 73) Statistics (P) Gender (female), n (%) 40 (36) 31 (42) χ 2 = 0.59 (0.443) Age 65.6 ± 11.35 62.76 ± 10.37 t= 1.75 (0.081) Age at disease onset, y 58.6 ± 11.24 56.16 ± 10.63 t= 1.48 (0.139) Disease duration, y 7.00 ± 1.37 6.59 ± 2.35 t= 1.33 (0.186) Disease duration at SPECT, y 2.75 ± 1.81 2.57 ± 2.03 t= 0.64 (0.524) H&Y stage 2 (2–2.5) 2 (2–2) U= 3492 (0.07) Levodopa doses 619.48 ± 338.05 946.3 ± 395.19 t=−5.6 (<0.001) LED by agonists 266.89 ± 121.92 306.88 ± 146.72 t=−1.46 (0.127) Total LED 720.68 ± 461.38 1066.81 ± 519.46 t=−4.62 (<0.001) The descriptive values presented are number (%) for female sex, median (IQR) for H&Y stage, and mean ± SD for all other continuous variables. PD-LID, PD patient group who did not develop LID; PD + LID, PD patient group who developed LID; LED, Levodopa equivalent dose. TABLE 2. Mean Regional Striatal DAT Binding in the Local PD Cohort Striatal Subregions PD-LID (n= 112) PD + LID (n= 73) Cohen dF(P) Limbic 2.19 ± 0.34 2.1 ± 0.3 0.27 3.33 (0.07) Executive 2.22 ± 0.42 2.1 ± 0.36 0.31 5.30 (0.023) Sensorimotor 1.86 ± 0.33 1.74 ± 0.26 0.38 5.99 (0.016) Rostral motor 2.12 ± 0.42 1.99 ± 0.35 0.33 4.51 (0.035) Caudal motor 1.85 ± 0.33 1.73 ± 0.26 0.38 5.70 (0.018) Parietal 1.66 ± 0.28 1.55 ± 0.22 0.41 6.73 (0.01) The descriptive values presented are mean SBR ± SD, the Cohen deffect size for (unadjusted) group differences between PD-LID and PD + LID, and analysis of covariance statistics. PD-LID, PD patient group who did not develop LID; PD + LID, PD patient group who developed LID. Labrador-Espinosa et al Clinical Nuclear Medicine •Volume 46, Number 6, June 2021 e300 www.nuclearmed.com © 2021 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved. 72 different sensorimotor-related striatal subregions, DAT levels in rostral-motor (β=−0.08, t=−2.14, P= 0.033) and caudal-motor (β=−0.06, t=−1.975, P= 0.049) subregions were significantly lower in PD + LID than PD-LID patients, but group differences in the parietal subregion were only trend-level significant (β=−0.04, t=−1.86, P=0.064). Dopamine transporter levels in all striatal subregions significantly decreased over time (all time effects, P< 0.001), but longitudinal change did not differ between the PD + LID and PD-LID groups in any of the striatal subregions (all group !time interactions, P> 0.32). DISCUSSION We investigated DAT changes measured with 123 I-FP-CIT SPECT within functionally defined striatal subregions between PD patients who developed LID and those who did not. In a primary study on our local monocentric PD cohort with varying degrees of disease progression, we found that PD patients who developed LID showed significantly lower DAT levels across large parts of the striatum and specifically in subregions associated with sensorimotor functions. The regional specificity of LID-associated DAT depletion in sensorimotor-related subregions of the striatum could be corroborated in an independent study cohort of longitudinally followed de novo PD patients. Interestingly, analysis of longitudinal 123 I-FP-CIT SPECT data in this cohort indicated that the LID-associated differences in sensorimotor striatal DAT levels were already present at the time of initial PD diagnosis, on average 4 years before the development of LID, and group differences were largely constant over time. Together, these results indicate that the development of LID in PD is specifically related to dopaminergic denervation in striatal subregions that are associated with sensorimotor functions and that these changes can be detected at a very early disease stage. Dopamine transporter depletion inbroadanatomicallydefined caudate and putaminal regions 32,33 has been related to the development of LID in PD patients in several previous studies. 12,14,15,34 TABLE 3. Demographics and Clinical Characteristics of the PPMI Cohort PD-LID (n= 182) PD + LID (n = 161) Variables BL 1st Visit 2nd Visit 4th Visit BL 1st Visit 2nd Visit 4th Visit Gender 64 (35%) ———53 (33%) ——— Age 62.87 (9.93) 63.57 (10.05) 64.78 (10.2) 67.20 (10.23) 60.25 (9.45) 61.26 (9.34) 62.84 (9.49) 65.09 (8.90) Disease duration years 0.57 (0.55) 1.73 (0.61) 2.65 (0.53) 4.6 (0.55) 0.49 (0.5) 1.57 (0.52) 2.58 (0.55) 4.61 (0.53) UPDRS III 19.37 (9.15) 23.15 (10.04) 27.35 (11.81) 28.11 (10.6) 22.11 (8.25) 25.73 (10.51) 26.32 (9.87) 30.59 (10.84) Levodopa doses —63.49 (165.27) 136.52 (224.29) 269.18 (269.58) —177.86 (306.56) 318.04 (391.34) 484.31 (450.16) Dopamine agonist LED —46.46 (85.2) 74.44 (117.1) 89.46 (116.88) —44.95 (85.12) 66.17 (116.71) 84.14 (136.52) Total LED —153.36 (189.18) 273.88 (231.13) 430.57 (265.7) —268.68 (303.04) 461.01 (392.23) 634.49 (454.18) The descriptive values presented are number (%) for female sex and mean (SD) for all other continuous variables. PD-LID, PD patient group who did not develop LID; PD + LID, PD patient group who developed LID; BL, baseline visit; LED, Levodopa equivalent dose. FIGURE 3. Longitudinal changes in UPDRS-III scores and LEDs for PD + LID and PD-LID groups in the PPMI cohort. A, Longitudinal LEDs and (B) total UPDRS-III score over follow-up time illustrated by the mean value and standard error for each visit. Levodopa doses were significantly higher in PD + LID patients compared with PD-LID patients since first year of drug initiation, and the dosage increase over time was significantly higher in PD + LID patients. The UPDRS-III score was on average significantly higher in PD + LID patients compared with PD-LID patients, although the UPDRS-III score increase over time was less pronounced for the PD + LID group. PD-LID, PD patient group who did not develop LID; PD + LID, PD patient group who developed LID; BL, baseline visit. Clinical Nuclear Medicine •Volume 46, Number 6, June 2021 Levodopa-Induced Dyskinesia in Parkinson Disease © 2021 Wolters Kluwer Health, Inc. All rights reserved. www.nuclearmed.com e301 Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved. 73 However, these studies had not differentiated between distinct functionally defined striatal subregions, which may be differentially implicated in different aspects of the clinical symptoms in PD. 35–37 In agreement with previous experimental studies in animal models, 38,39 our in vivo imaging findings confirm that dopaminergic degeneration associated with the occurrence of LID specifically implicates striatal areas connected to the cortical sensorimotor system. Furthermore, in an independent validation study using data from the PPMI cohort, we found that excess DAT reductions in de novo PD patients who later developed LID were limited to the sensorimotor striatal region and that this difference was already present at study baseline (coinciding with initial PD diagnosis) and remained largely constant over an average of 1.8 years of follow-up with SPECT imaging. Our results agree with previous studies that have shown the predictive role of decreased putaminal DAT levels for the development of LID in de novo PD patients using both the PPMI 17 and other cohorts. 14,15 However, these findings contrast with other studies that could not fully reproduce this predictive effect in de novo PD patients, 19 including a recent study using data from the PPMI cohort. 18 This latter study investigated a wide range of possible risk factors for the development of LID, and although they could confirm several previously reported risk factors for LID, they did not find significant differences in baseline DAT uptake levels of the caudate or putamen between patients who did or did not develop LID. However, somewhat surprisingly, and in contrast to other studies on LID-associated DAT changes, 16 the caudate asymmetry index, but not the putamen asymmetry index, was found to predict an increased risk of LID development in this study. Possible explanations for the discrepancy with our current findings on LID development in the PPMI cohort include the use of longitudinal SPECT data and its statistical modeling using mixed linear models, whereas only baseline SPECT-derived DAT values were used in this previous study. However, another explanation may be that the rather broad anatomical divisions of the striatum (caudate and putamen) used in this previous study may not be sensitive enough to reliably detect subtle LID-associated DAT changes in this early disease stage (see Supplemental Data for a complementary analysis of standard caudate and putamen regions of interest in our data sets that corroborate this notion (Supplemental Digital Content 1, http://links.lww.com/CNM/A315). Taken together, our findings across 2 independent cohorts of PD patients at varying stages of disease evolution suggest that the development of LID may be specifically associated with reduced DAT levels in functionally defined sensorimotor-related striatal subregions. These subregions are defined by their specific cortical connectivity pattern with cortical motor areas including rostral Brodmann area 6, pre-supplementary motor area, precentral gyrus, and the frontal eye field region. 27 Interestingly, recent diffusion tensor imaging and resting-state functionalMRIstudieshaveshownabnormalstriatocortical connectivity patterns in PD patients who suffer from LID, especially affecting connections between the putamen and cortical sensorimotor areas. 40–42 Moreover, in a pharmacodynamic functional neuroimaging approach that mapped the effect of a single dose of L-DOPA on connectivity in cortico - basal ganglia motor loops, it could be shown that the dopaminergic modulationoffeedbackconnectionsfromthe putamen to cortical motor areas was strongly involved in the development and severity of LID, but not the forward connections from cortical motor areas to the putamen. 43 Our findings on specific LID-associated DAT level depletion in functional subregions of the striatum that are connected to sensorimotor cortical areas may represent a neurodegenerative correlate of this abnormal dopaminergic modulation of cortical motor areas through striatal projections. However, additional multimodal imaging studies are necessary to investigate the relation between regionally specific DAT depletion and striatocortical connectivity changes in the development of LID. Although the LID-associated differences in striatal DAT levels showed highest effect size in the sensorimotor striatal subregions, it should be noted that in our primary study cohort these differences also extended into the executive striatal subregion. These results could be related to recent findings reported by Yoo et al, 44 who demonstrated that LID development was closely associated with the progression of cognitive decline, especially with frontal executive dysfunction. Interestingly, a recent multimodal 123 I-FP-CIT SPECT and [ 18 F]FDG-PET imaging study found that reduced DAT levels TABLE 4. Mean Regional Striatal DAT Binding in Serial SPECT Data From the PPMI Cohort Descriptive Values Mixed Linear Model Statistics Striatal Subregions Group BL 1st Visit 2nd Visit 4th Visit (β,t,P) Limbic PD-LID 2.19 (0.37) 2.12 (0.34) 2.05 (0.36) 1.93 (0.33) a. β=−0.03, t=−0.84, P= 0.40 b. β=−0.05, t=−7.38, P< 0.001 c. β=−0.004, t=−0.53, P= 0.60 PD + LID 2.18 (0.33) 2.08 (0.36) 1.99 (0.34) 1.86 (0.32) Executive PD-LID 2.23 (0.39) 2.15 (0.36) 2.04 (0.39) 1.89 (0.35) a. β=−0.05, t=−1.23, P= 0.22 b. β=−0.07, t=−9.38, P< 0.001 c. β=−0.0004, t=−0.05, P= 0.96 PD + LID 2.21 (0.36) 2.09 (0.39) 1.97 (0.37) 1.83 (0.36) Sensorimotor PD-LID 1.84 (0.29) 1.78 (0.27) 1.70 (0.29) 1.60 (0.25) a. β=−0.06, t=−2.05, P= 0.041 b. β=−0.05, t=−10.27, P< 0.001 c. β= 0.005, t= 0.77, P= 0.44 PD + LID 1.80 (0.25) 1.71 (0.25) 1.63 (0.24) 1.53 (0.23) Rostral motor PD-LID 2.13 (0.37) 2.03 (0.35) 1.93 (0.37) 1.81 (0.33) a. β=−0.08, t=−2.14, P= 0.033 b. β=−0.07, t=−10.96, P< 0.001 c. β= 0.005, t= 0.57, P= 0.57 PD + LID 2.05 (0.33) 1.95 (0.33) 1.83 (0.32) 1.70 (0.32) Caudal motor PD-LID 1.82 (0.29) 1.77 (0.27) 1.69 (0.30) 1.58 (0.26) a. β=−0.06, t=−1.975, P= 0.049 b. β=−0.05, t=−9.53, P< 0.001 c. β= 0.01, t= 0.74, P= 0.46 PD + LID 1.78 (0.25) 1.69 (0.25) 1.61 (0.25) 1.51 (0.23) Parietal PD-LID 1.64 (0.23) 1.59 (0.21) 1.54 (0.24) 1.45 (0.20) a. β=−0.04, t=−1.86, P= 0.064 b. β=−0.04, t=−8.70, P< 0.001 c. β= 0.0, t= 0.99, P= 0.32 PD + LID 1.61 (0.20) 1.55 (0.21) 1.48 (0.20) 1.41 (0.17) The descriptive values presented are mean SBR (SD). Mixed-linear model statistics are (a) group effect, (b) time effect, and the (c) group !time interaction. PD-LID, PD patient group who did not develop LID; PD + LID, PD patient group who developed LID; BL, baseline visit. Labrador-Espinosa et al Clinical Nuclear Medicine •Volume 46, Number 6, June 2021 e302 www.nuclearmed.com © 2021 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved. 74 FIGURE 4. Longitudinal DAT binding changes in striatal subregions for PD + LID and PD-LID groups in the PPMI cohort. DAT changes in the (A) limbic, (B) executive, and (C) sensorimotor striatal region over follow-up time illustrated by the mean value and standard error for each visit. DAT levels in the sensorimotor-related striatal subregion were significantly lower in PD + LID patients compared with PD-LID patients, but no significant differences were observed for the limbic- and executive-related striatal subregions. PD-LID, PD patient group who did not develop LID; PD +LID,PDpatientgroupwhodevelopedLID;BL,baselinevisit. Clinical Nuclear Medicine •Volume 46, Number 6, June 2021 Levodopa-Induced Dyskinesia in Parkinson Disease © 2021 Wolters Kluwer Health, Inc. All rights reserved. www.nuclearmed.com e303 Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved. 75 in the cognitive part of the striatum (combined executive + limbic subregions), but not in the sensorimotor part, are associated with frontomedial hypometabolism in PD patients, which likely represents a neurofunctional correlate of impaired executive functions. 35 Together with our findings of a selective association of LID with DAT depletion in sensorimotor regions in drug-naive de novo PD patients, these data could indicate a sequence of LID-associated DAT changes progressing from sensorimotor to executive striatal subregions with corresponding clinical-cognitive changes. Molecular neuroimaging studies over longer follow-up intervals will be necessary to study the regionally progressive neurodegenerative changes underlying LID-associated cognitive changes in more detail. As expected, in both study cohorts, we found higher doses of L-DOPA and total LED in PD patients who suffered LID compared with PD patients who did not develop this complication. Moreover, PD patients with LID also had more severe motor symptoms than PD patients without LID. These results are fully consistent with previous findings on common risk factors of LID, indicating a narrowing of the therapeutic window of L-DOPA treatment with progression of motor symptoms in PD. 45,46 Interestingly, in the PPMI cohort, we also observed a higher longitudinal increase of L-DOPA doses in PD patients who develop LID, whereas DAT levels in the sensorimotor-related striatal subregion were already significantly reduced at baseline compared with the PD-LID group and remained relatively constant over time. This corroborates previously reported interactive effects of L-DOPA treatment and dopaminergic depletion on LID, where high doses of L-DOPA specifically associate with dyskinesia in patients with higher levels of dopaminergic depletion. 47,48 Our data suggest that the proposed method of measuring DAT depletion in a functionally defined sensorimotor-related striatal region may provide a more sensitive imaging biomarker for detecting LID-associated dopaminergic degeneration in an early disease stage and may thus improve the individual prognosis of, and clinical decision making for, this common medication-related complication in PD. 14,15,18,49 Taking into account other known clinical risk factors, the assessment of an individual patient’sLIDriskthroughthe specific measurement of sensorimotor-related striatal dopaminergic depletion at an early disease stage could help the clinician in therapeutic decision making and potentially improve prevention and management of this complication through individually adjusted therapeutic strategies. 50–52 Anextsteptowardsuccessfulclinical translation of our current research findings to such a precision medicine approach will involve the development of accurate clinical decision support systems that integrate the information from the proposed molecular imaging biomarker with other clinical and biological sources of information within multivariate predictive models. 53,54 However, it also has to be noted that our current findings were obtained in a controlled research environment and in patients from highly specialized tertiary care centers, so a further validation in less selected patient cohorts is required. Moreover, the automated analysis methods we use for measuring DAT levels in specific functionally defined striatal subregions require expertise in computational image processing and analysis, which could pose a limitation for the implementation of this measurement in the wider health care system. Nevertheless, we believe that translation of this method outside of dedicated research centers will be feasible through increasingly specialized software solutions that are becoming available for medical image analysis in clinical settings. 55,56 Awiderclinicalaccessibility of our proposed regional DAT quantification method through such user-friendly software solutions will allow testing the prognostic potential of this method as a molecular imaging biomarker for increased LID risk through a diagnostic trial in a real-world clinical setting. Aprincipallimitationofourstudyisthatthedatafromourlocal cohort were collected in a retrospective manner. The occurrence of LID was evaluated by clinical examination through neurologists specialized in movement disorders, but no additional standardized scales for the assessment of motor complications in PD (such as the UPDRS-IV or Abnormal Involuntary Movement Scale) were available, so that more detailed analyses of dyskinesia severity or type of dyskinetic complication could not be performed. 48,57,58 Moreover, quantification of specific DAT binding reductions in spatially detailed striatal subregions may benefit from partial volume correction of 123 I-FP-CIT SPECT signal using anatomic information from high-resolution structural MRI, 59 but these data were not available for our retrospective cohort. In general, the spatial resolution of 123 I-FP-CIT SPECT scans may cast doubt on the ability of this imaging modality to discern signal from relatively small striatal subregions, particularly with respect to the distinct sensorimotor-related subdivisions defined in the employed striatal connectivity atlas. 27 However, we observed reproducible differences in relation to LID-associated DAT depletion in the larger sensorimotor-related region compared with the cognition-related regions across 2 independent cohorts. Finally, other neuronal systems 60 and the influence of other neurotransmitters, 61 such as alterations in serotonin levels, 62,63 could be involved in the pathophysiology of LID, but could not be taken into account in our current study. In summary, we provide evidence that the development of LID in PD specifically associates with dopaminergic depletion in distinct sensorimotor-related subregions of the striatum. Measuring DAT depletion in these functionally defined regions of interest may provide a more sensitive tool to detect LID-associated dopaminergic changes in an early disease stage, and could thus improve individual prognosis of, and clinical decision making for, this common complication in PD symptoms. Studying the relation of region-specific striatal dopaminergic denervation with functional changes in striatocortical signaling loops and their interaction with others pathophysiologic factors measurable by neuroimaging, such as nondopaminergic neurotransmitter deficits or region-specific atrophic brain changes, provides an exciting venue for future research into the complex pathophysiologic mechanisms underlying LID in PD. ACKNOWLEDGMENTS The authors thank all the patients and their caregivers for their efforts to participate in this study. In addition to our local university hospital cohort, data used in the preparation of this article were obtained from the Parkinson Progression Marker’sInitiativedatabase (www.ppmi-info.org/data). Parkinson Progression Marker’sInitiative —a public-private partnership—is funded by the Michael J. Fox Foundation for Parkinson’s Research. Corporate funding partners include AbbVie, Allergan, Amathus, Avid, Biogen, BioLegend, Bristol-Myers Squibb, Celgene, Denali, GE Healthcare, Genentech, GlaxoSmithKline, Janssen Neurosciene, Lilly, Lundbeck, Merck, Meso Scale Discovery, Pfizer, Piramal, Prevail, Roche, Sanofi Genzyme, Servier, Takeda, Teva, UCB, Verily, and Voyager, and philanthropic funding partners include GOLUB CAPITAL. REFERENCES 1. Poewe W, Seppi K, Tanner CM, et al. Parkinson disease. Nat Rev Dis Prim. 2017;3:1–21. 2. Marsden CD, Parkes JD. “On-off ”effects in patients with Parkinson’s disease on chronic levodopa therapy. Lancet. 1976;307:292–296. 3. Péchevis M, Clarke CE, Vieregge P, et al. Effects of dyskinesias in Parkinson’s disease on quality of life and health-related costs: a prospective European study. Eur J Neurol. 2005;12:956–963. 4. Ahlskog JE, Muenter MD. Frequency of levodopa-related dyskinesias and motor fluctuations as estimated from the cumulative literature. Mov Disord. 2001;16:448–458. Labrador-Espinosa et al Clinical Nuclear Medicine •Volume 46, Number 6, June 2021 e304 www.nuclearmed.com © 2021 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved. 76 5. Rascol O, Brooks DJ, Korczyn AD, et al. A five-year study of the incidence of dyskinesia in patients with early Parkinson’s disease who were treated with ropinirole or levodopa. N Engl J Med. 2000;342:1484–1491. 6. Espay AJ, Morgante F, Merola A, et al. Levodopa-induced dyskinesia in Parkinson disease: current and evolving concepts. Ann. Neurol.2018;797–811. 7. Warren Olanow C, Kieburtz K, Rascol O, et al. Factors predictive of the development of levodopa-induced dyskinesia and wearing-off in Parkinson’s disease. Mov Disord. 2013;28:1064–1071. 8. Grandas F, Galiano ML, Tabernero C. Risk factors for levodopa-induced dyskinesias in Parkinson’s disease. J Neurol. 1999;246:1127–1133. 9. Fahn S, Oakes D, Shoulson I, et al. Levodopa and the progression of Parkinson’s disease. N Engl J Med. 2004;351:2498–2508. 10. Porras G, de Deurwaerdere P, Li Q, et al. L-DOPA–induced dyskinesia: beyond an excessive dopamine tone in the striatum. Sci Rep. 2014;4. 11. De la Fuente-Fernández R, Sossi V, Huang Z, et al. Levodopa-induced changes in synaptic dopamine levels increase with progression of Parkinson’s disease: implications for dyskinesias. Brain. 2004;127:2747–2754. 12. De la Fuente-Fernndez R, Lu JQ, Sossi V, et al. Biochemical variations in the synaptic level of dopamine precede motor fluctuations in Parkinson’s disease: PET evidence of increased dopamine turnover. Ann Neurol. 2001; 49:298–303. 13. Picconi B, Hernández LF, Obeso JA, et al. Motor complications in Parkinson’s disease: striatal molecular and electrophysiological mechanisms of dyskinesias. Mov Disord. 2018;33:867–876. 14. Hong JY, Oh JS, Lee I, et al. Presynaptic dopamine depletion predicts levodopa-induced dyskinesia in de novo Parkinson disease. Neurology. 2014;82:1597–1604. 15. Yoo HS, Chung SJ, Chung SJ, et al. Presynaptic dopamine depletion determines the timing of levodopa-induced dyskinesia onset in Parkinson’s disease. Eur J Nucl Med Mol Imaging. 2018;45:423–431. 16. Chung SJ, Yoo HS, Lee HS, et al. The pattern of striatal dopamine depletion as a prognostic marker in de novo Parkinson disease. Clin Nucl Med. 2018; 43:787–792. 17. Jeong EH, Sunwoo MK, Song YS. Serial I-123-FP-CIT SPECT image findings of Parkinson’s disease patients with levodopa-induced dyskinesia. Front Neurol. 2018;9:1–7. 18. Eusebi P, Romoli M, Paoletti FP, et al. Risk factors of levodopa-induced dyskinesia in Parkinson’s disease: results from the PPMI cohort. npj Park Dis [Internet]. 2018;4:4. http://dx.doi.org/10.1038/s41531-018-0069-x. 19. Roussakis A-A, Gennaro M, Lao-Kaim NP, et al. Dopamine transporter density in de novo Parkinson’s disease does not relate to the development of levodopa-induced dyskinesias. J Neuroinflamm Neurodegener Dis [Internet]. 2019;3:10000. http://www.pubmedcentral.nih.gov/articlerender.fcgi? artid=PMC6901354. 20. Alexander G. Parallel organization of functionally segregated circuits linking basal ganglia and cortex. Annu Rev Neurosci. 1986;9:357–381. 21. Haber SN. The primate basal ganglia: parallel and integrative networks. J Chem Neuroanat. 2003;26:317–330. 22. Alexander GE, Crutcher MD, Delong MR. Chapter 6 basal ganglia-thalamocortical circuits: parallel substrates for motor, oculomotor, “prefrontal”and “limbic”functions. Prog Brain Res. 1991;85:119–146. 23. Parent A, Hazrati LN. Functional anatomy of the basal ganglia. I. The cortico-basal ganglia-thalamo-cortical loop. Brain Res Rev. 1995;91–127. 24. DeLong MR, Wichmann T. Circuits and circuit disorders of the basal ganglia. Arch Neurol. 2007;64:20–24. 25. Choi EY, Thomas Yeo BT, Buckner RL. The organization of the human striatum estimated by intrinsic functional connectivity. J Neurophysiol. 2012; 108:2242–2263. 26. Draganski B, Kherif F, Klöppel S, et al. Evidence for segregated and integrative connectivity patterns in the human basal ganglia. J Neurosci. 2008; 28:7143–7152. 27. Tziortzi AC, Haber SN, Searle GE, et al. Connectivity-based functional analysis of dopamine release in the striatum using diffusion-weighted MRI and positron emission tomography. Cereb Cortex. 2014;24:1165–1177. 28. Postuma RB, Berg D, Stern M, et al. MDS clinical diagnostic criteria for Parkinson’s disease. Mov Disord. 2015;30:1591–1601. 29. Seibyl J, Marek K, Zubal IG. The role of the core imaging laboratory in multicenter trials. Semin Nucl Med. 2010;338–346. 30. García-Gómez FJ, García-Solís D, Luis-Simón FJ, et al. Elaboration of the SPM template for the standardization of SPECT images with 123 I-Ioflupane. Rev Esp Med Nucl Imagen Mol English Ed. 2013;32:350–356. 31. Innis RB, Cunningham VJ, Delforge J, et al. Consensus nomenclature for in vivo imaging of reversibly binding radioligands. J Cereb Blood Flow Metab. 2007;27:1533–1539. 32. Morrish PK, Sawle GV, Brooks DJ. Regional changes in [ 18 F]dopa metabolism in the striatum in Parkinson’s disease. Brain. 1996;119:2097–2103. 33. Nandhagopal R, Kuramoto L, Schulzer M, et al. Longitudinal progression of sporadic Parkinson’s disease: a multi-tracer positron emission tomography study. Brain. 2009;132:2970–2979. 34. De la Fuente-Fernández R, Pal PK, Vingerhoets FJG, et al. Evidence for impaired presynaptic dopamine function in parkinsonian patients with motor fluctuations. J Neural Transm. 2000;107:49–57. 35. Apostolova I, Lange C, Frings L, et al. Nigrostriatal degeneration in the cognitive part of the striatum in Parkinson disease is associated with frontomedial hypometabolism. Clin Nucl Med. 2020;45:95–99. 36. Khan AR, Hiebert NM, Vo A, et al. Biomarkers of Parkinson’s disease: striatal sub-regional structural morphometry and diffusion MRI. Neuroimage Clin. 2019;21:101597. 37. Kübler D, Schroll H, Buchert R, et al. Cognitive performance correlates with the degree of dopaminergic degeneration intheassociativepartofthestriatumin non-demented Parkinson’spatients.JNeuralTransm.2017;124:1073–1081. 38. Wong MY, Borgkvist A, Choi SJ, et al. Dopamine-dependent corticostriatal synaptic filtering regulates sensorimotor behavior. Neuroscience. 2015;290: 594–607. 39. Bamford NS, Robinson S, Palmiter RD, et al. Dopamine modulates release from corticostriatal terminals. J Neurosci. 2004;24:9541–9552. 40. Herz DM, Haagensen BN, Christensen MS, et al. The acute brain response to levodopa heralds dyskinesias in Parkinson disease. Ann Neurol. 2014;75: 829–836. 41. Herz DM, Haagensen BN, Nielsen SH, et al. Resting-state connectivity predicts levodopa-induced dyskinesias in Parkinson’s disease. Mov Disord. 2016;31:521–529. 42. Wang L, Wang M, Si Q, et al. Altered brain structural topological properties in Parkinson’s disease with levodopa-induced dyskinesias. Park Relat Disord. 2019;67:36–41. 43. Herz DM, Haagensen BN, Christensen MS, et al. Abnormal dopaminergic modulation of striato-cortical networks underlies levodopa-induced dyskinesias in humans. Brain. 2015;138:1658–1666. 44. Yoo HS, Chung SJ, Lee YH, et al. Levodopa-induced dyskinesia is closely linked to progression of frontal dysfunction in PD. Neurology. 2019;92: E1468–E1478. 45. Angela Cenci M. Presynaptic mechanisms of L-DOPA–induced dyskinesia: the findings, the debate, the therapeutic implications. Front Neurol. 2014;5: 1–15. 46. Hauser RA, McDermott MP, Messing S. Factors associated with the development of motor fluctuations and dyskinesias in Parkinson disease. Arch Neurol. 2006;63:1756–1760. 47. Francardo V, Recchia A, Popovic N, et al. Impact of the lesion procedure on the profiles of motor impairment and molecular responsiveness to L-DOPA in the 6-hydroxydopamine mouse model of Parkinson’s disease. Neurobiol Dis. 2011;42:327–340. 48. Putterman DB, Munhall AC, Kozell LB, et al. Evaluation of levodopa dose and magnitude of dopamine depletion as risk factors for levodopa-induced dyskinesia in a rat model of Parkinson’s disease. J Pharmacol Exp Ther. 2007;323:277–284. 49. Schapira AHV. A pragmatic, personalised approach to treatment initiation in Parkinson’s disease. Lancet Neurol. Elsevier Ltd; 2020;19:376–378. 50. De Bie RMA, Clarke CE, Espay AJ, et al. Initiation of pharmacological therapy in Parkinson’s disease: when, why, and how. Lancet Neurol. 2020;19: 452–461. 51. Fox SH, Katzenschlager R, Lim SY, et al. International Parkinson and movement disorder society evidence-based medicine review: update on treatments for the motor symptoms of Parkinson’s disease. Mov Disord. 2018;33:1248–1266. 52. Worth PF. When the going gets tough: how to select patients with Parkinson’s disease for advanced therapies. Pract Neurol. 2013;13:140–152. 53. Klucken J,KrugerR,SchmidtP,etal.ManagementofParkinson’sdisease20years from now: towards digital health pathways. JParkinsonsDis.2018;8:S85–S94. 54. Zubair R, Francisco G, Rao B. Artificial intelligence for clinical decision support. Brain. 2018;102:210–211. 55. López-González FJ, Silva-Rodríguez J, Paredes-Pacheco J, et al. Intensity normalization methods in brain FDG-PET quantification. Neuroimage. 2020;222:117229. Clinical Nuclear Medicine •Volume 46, Number 6, June 2021 Levodopa-Induced Dyskinesia in Parkinson Disease © 2021 Wolters Kluwer Health, Inc. All rights reserved. www.nuclearmed.com e305 Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved. 77 56. Brogley JE. DatQuant: the future of diagnosing Parkinson disease. J Nucl Med Technol. 2019;47:21–26. 57. Colosimo C, Martínez-Martín P, Fabbrini G, et al. Task force report on scales to assess dyskinesia in Parkinson’s disease: critique and recommendations. Mov Disord. 2010;25:1131–1142. 58. Pavese N,EvansAH,TaiYF,etal.Clinicalcorrelatesoflevodopa-induceddopa- mine release in Parkinson disease: a PET study. Neurology.2006;67:1612–1617. 59. Trnka J, Dusek P, Samal M, et al. MRI-guided voxel-based automatic semi-quantification of dopamine transporter imaging. Phys Med. 2020;75: 2–10. 60. Calabresi P, Picconi B, Tozzi A, et al. Direct and indirect pathways of basal ganglia: a critical reappraisal. Nat Neurosci.2014;17:1022–1030. 61. Beaudoin-Gobert M, Météreau E, Duperrier S, et al. Pathophysiology of levodopa-induced dyskinesia: insights from multimodal imaging and immunohistochemistry in non-human primates. Neuroimage. 2018;183:132–141. 62. Carta M, Carlsson T, Kirik D, et al. Dopamine released from 5-HT terminals is the cause of L-DOPA–induced dyskinesia in parkinsonian rats. Brain. 2007;130:1819–1833. 63. Roussakis AA, Politis M, Towey D, et al. Serotonin-to-dopamine transporter ratios in Parkinson disease. Neurology. 2016;86:1152–1158. Labrador-Espinosa et al Clinical Nuclear Medicine •Volume 46, Number 6, June 2021 e306 www.nuclearmed.com © 2021 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved. 78 1 Supplementary data. Complementary analysis of specific [123I]-FP-CIT binding ratio in standard anatomically-defined caudate and putamen regions. In this complementary analysis, dopamine transporter (DAT) levels were quantified by the mean specific binding ratio (SBR) of [123I]- FP-CIT in standard anatomically-defined caudate and putamen regions (1). The same methodology of the main study to compare DAT levels between groups (ANCOVA for local cohort and linear-mixed model for PPMI cohort) was used. The local cohort results showed that mean DAT levels in the putamen region were significantly lower in the PD+LID group compared to the PD-LID group, whereas group differences in DAT levels in the caudate region reached only trend-level statistical significance (supplementary table 1). However, in the PPMI cohort no significant group effects were observed for any of these two regions (supplementary table 2). 1. Tziortzi AC, Searle GE, Tzimopoulou S, Salinas C, Beaver JD, Jenkinson M, et al. Imaging dopamine receptors in humans with [11C]-(+)-PHNO: Dissection of D3 signal and anatomy. Neuroimage. 2011; 79 2 Supplementary Table 1. Mean DAT binding in standard anatomical striatal regions for each group in HUVR cohort. Striatal regions PD-LID (! = 112) PD+LID (! = 73) Stats (F value, p value) Caudate 1.01 ± 0.35 0.95 ± 0.32 F = 3.27, p = 0.07 Putamen 1.19 ± 0.41 1.04 ± 0.34 F = 6.30, p = 0.01 The descriptive values presented are mean specific binding ratio (SBR) (standard deviation). Abbreviation: PD-LID = PD patient group who did not develop LID; PD+LID = PD patient group who developed LID; 80 3 Supplementary Table 2. Mean striatal DAT in connectivity-based functional regions for each group in longitudinal cohort. Striatal subregions PD-LID (! = 182) PD+LID (! = 161) Stats (Beta, t value, P value) BL 1st Visit 2nd Visit 4th Visit BL 1st Visit 2nd Visit 4th Visit a. Group effect. b. Time effect. c. Group x time interaction. Caudate 2.00 (0.34) 1.94 (0.31) 1.86 (0.34) 1.73 (0.31) 2.00 (0.32) 1.91 (0.35) 1.80 (0.33) 1.69 (0.32) a. β = -0.03, t = -1.03, p = 0.30 b. β = -0.05, t = -7.75, p < 0.001 c. β = -0.0008, t = 0.11, p = 0.91 Putamen 2.19 (0.37) 2.11 (0.35) 2.01 (0.37) 1.87 (0.33) 2.15 (0.33) 2.03 (0.35) 1.93 (0.34) 1.79 (0.32) a. β = -0.05, t = -1.47, p = 0.14 b. β = -0.07, t = -10.33, p < 0.001 c. β = 0.0001, t = -0.02, p = 0.99 The descriptive values presented are mean specific binding ratio (SBR) (standard deviation). Mixed-linear model stats are a) group effect, b) time effect, and the c) group x time interaction. Abbreviation: PD-LID = PD patient group who did not develop LID; PD+LID = PD patient group who developed LID; 87 Parkinsonism and Related Disorders 88 (2021) 68–75 73 measured by volumetric changes on MRI may only occur at a later stage of PD-related cognitive decline. However, more subtle neuroimaging measures may increase the sensitivity to detect early neurodegenerative tissue changes in the hippocampus already at a predementia stage in PD. Thus, a previous multimodal MRI/DTI imaging study found that hippocampal diffusivity changes correlated with memory performance among non-demented PD patients, but no significant associations were observed for hippocampal volume [14]. In our DTI substudy we could confirm that hippocampal diffusivity measures were more closely associated with cognition in non-demented PD compared to volume, although this association showed a rather unexpected specificity for verbal fluency instead of memory measures. Interestingly, in contrast to the findings for the hippocampus, DTI- based diffusivity measures of the CBF were not associated with cognition in our sample. This is partly in contrast to a previous MRI/DTI study in de novo PD patients from the PPMI cohort [11], which found both CBF volume and MD to be significant predictors of prospective cognitive decline. Similarly, another recent DTI study found significant cross-sectional correlations between diffusivity changes in CBF ROIs and cognitive performance in non-demented PD patients, but CBF volumes were not assessed in that study [26]. These different findings could at least partly relate to differences in patient characteristics, as the proportion of PD-MCI patients was notably higher (20 out of 52 patients; 38%) in this previous study compared to our current study sample (16 out of 73 patients with PD-CRS total ≤81; 22%). However, they may also reflect a generally higher variability in DTI-based diffusion measures, which are known to be more susceptible to noise compared to standard volumetric measures on structural MRI [27]. This study has several limitations. First, although the multicentric design of the study allowed us to assess CBF-cognition associations in a relatively large and clinically diverse sample of PD patients, the extracted imaging indices may have been affected by the multicentric acquisition of the imaging data, which is particularly a concern for the DTI signal measurements [27]. In addition to a generally higher susceptibility to inter-scanner related variance, DTI signal measurements may potentially also be more prone to other signal confounders, such as for example contamination by CSF signal that borders the ventral limit of the basal forebrain. Thus, besides modality-specific sensitivities for distinct neuropathological processes, the observed differences in cognition associations between these imaging modalities may also relate to technical differences in the MRI- and DTI-based measurements. However, it has to be noted that we controlled all our statistical analyses for the different acquisition sites, and the multicentric design also allowed us to demonstrate the robustness of the CBF-cognition associations across centers (see center-specific meta-analysis in the supplementary material), which is a key requirement of potential imaging markers that are to be used in wider clinical setting. Another limitation is the cross-sectional nature of our study. It remains to be determined whether the observed cognition-relevant differences in CBF volume among PD patients also translate to a higher risk for progression to dementia. Furthermore, the COPPADIS cohort represents a rather selective research cohort that may not be reflective of the general PD population, and thus our results should be replicated in less preselected cohorts from community-based studies. Another important consideration is that the employed MRI markers of CBF integrity can only serve as indirect markers of cholinergic degeneration. The analyzed CBF region corresponds to the localization of the cholinergic nuclei in the human brain as determined by cytoarchitectonic mappings from combined histology and post-mortem MRI [20], and imaging-pharmacologic [6] and imaging-neurophysiologic [28] studies provide evidence for a link between CBF integrity on MRI and cholinergic system functioning. However, one cannot exclude that the imaging indices might also reflect change in other neuronal or glial components. Finally, the correlation coefficients for the association between CBF volume and cognition are of rather moderate effect size, which, however, lies well in the range of previous studies assessing associations between regional imaging measures and cognition in PD [4,10,14]. Associations between neuroimaging measures and clinical outcomes can rarely be expected to be strictly linear and will generally depend on the functional consequences of the neurodegenerative process that is indexed by the macro- and microstructural neuroimaging alterations. A recent imaging-neuropathology association study found that CBF volume on structural MRI correlated with the extent of Lewy body pathology in the nucleus basalis of Meynert, but not with semi-quantitative assessments of cell loss in this region [29]. This indicates that the neuroimaging measures may also pick up relatively subtle tissue changes in response to neurodegenerative pathology [30]. In conclusion, we demonstrate in a well-characterized multicentric sample of PD patients without dementia that cognitive differences in this non-demented spectrum are more closely related to structural MRI measures of CBF degeneration than hippocampal degeneration. DTI- based diffusion measures could increase the sensitivity to detect Table 3 Linear model stats for the association of diffusivity imaging markers with cognitive scores. Cognitive scores Mean Diffusivity Cholinergic Basal Forebrain Hippocampus Part. r T value (P value) P (FDR) Part. r T value (P value) P(FDR) PD-CRS total −0.10 −0.82 (0.418) 0.418 −0.43 −3.98 (<0.001) <0.001 −0.29 −2.40 (0.02) 0.029 Fronto-subcortical items Sustained attention 0.01 0.05 (0.958) 0.958 0.02 0.20 (0.843) 0.843 Working memory −0.10 −0.84 (0.404) 0.95 −0.26 −2.22 (0.030) 0.049 −0.01 −0.08 (0.939) 0.94 Clock drawing −0.08 −0.63 (0.528) 0.95 −0.26 −2.27 (0.026) 0.049 −0.15 −1.17 (0.247) 0.371 Alternating verbal fluency 0.04 0.37 (0.710) 0.958 −0.49 −4.72 (<0.001) <0.001 −0.44 −3.90 (<0.001) 0.001 Action verbal fluency −0.01 −0.10 (0.924) 0.958 −0.30 −2.58 (0.012) 0.036 −0.20 −1.63 (0.108) 0.323 Immediate verbal memory −0.24 −2.01 (0.048) 0.216 −0.18 −1.53 (0.131) 0.147 −0.14 −1.10 (0.274) 0.274 Delayed verbal memory −0.27 −2.37 (0.021) 0.186 −0.25 −2.18 (0.033) 0.049 −0.14 −1.14 (0.257) 0.274 −0.03 −0.24 (0.812) 0.94 Posterior-cortical items Naming −0.02 −0.20 (0.846) 0.958 −0.21 −1.80 (0.077) 0.099 Clock copy −0.09 −0.77 (0.445) 0.95 −0.32 −2.85 (0.006) 0.026 −0.18 −1.42 (0.161) 0.323 Linear model stats are presented as (partial) r and T value (P value). For each cognitive item score, results from single predictor linear regression models (controlled for acquisition site) are indicated in the first row, and for significant models fully covariate-controlled regression results are indicated in the second row. Abbreviations: PD-CRS =PD-Cognitive Rating Scale. M.J. Grothe et al. 88 Parkinsonism and Related Disorders 88 (2021) 68–75 74 cognition-relevant tissue changes in the hippocampus, but at least in the context of unharmonized multicentric imaging acquisitions appear to be inferior to standard volumetric measures for capturing cognitionrelevant CBF changes. These findings may have important implications for the development of clinically useful imaging biomarkers aiding in the early detection of PD patients at increased risk for cognitive decline. Authors’ contribution Michel J. Grothe: conception of the manuscript, study design, analysis, writing of the first draft of the manuscript, review and critique. Miguel A. Labrador-Espinosa: conception of the manuscript, study design, analysis, review and critique. Silvia Jesús: review and critique; recruitment and/or assessment of participants. Daniel Macías-García: review and critique; recruitment and/or assessment of participants. Astrid Adarmes-G´ omez: review and critique; recruitment and/or assessment of participants. Elena Iglesias Camacho: review and critique. Pablo Franco-Rosado: review and critique. Florinda Rold´ an Lora: review and critique; acquisition of neuroimaging data. Juan Francisco Martín- Rodríguez: review and critique. Miquel Aguilar Barber´ a: review and critique; recruitment and/or assessment of participants. Pau Pastor: review and critique; recruitment and/or assessment of participants. Sonia Escalante Arroyo: review and critique; recruitment and/or assessment of participants. Berta Solano Vila: review and critique; recruitment and/or assessment of participants. Anna Cots Foraster: review and critique; recruitment and/or assessment of participants. Javier Ruiz Martínez: review and critique; recruitment and/or assessment of participants. Francisco Carrillo Padilla: review and critique; recruitment and/or assessment of participants. Mercedes Pueyo Morlans: review and critique; recruitment and/or assessment of participants. Isabel Gonz´ alez Aramburu: review and critique; recruitment and/or assessment of participants. Jon Infante Ceberio: review and critique; recruitment and/or assessment of participants. Jorge Hern´ andez Vara: review and critique; recruitment and/or assessment of participants. Oriol de F´ abregues- Boixar: review and critique; recruitment and/or assessment of participants. Teresa de Deus Fonticoba: review and critique; recruitment and/ or assessment of participants. Berta Pascual-Sedano: review and critique; recruitment and/or assessment of participants. Jaime Kulisevsky: review and critique. Overall supervision. Pablo Martínez-Martín: review and critique. Overall supervision. Diego Santos-García: conception, organization, and execution of the project; recruitment and/or assessment of participants; review and critique. Pablo Mir: conception of the manuscript, study design, review and critique. Collaborators (COPPADIS SUDY GROUP) Almeria M 4 , Alonso-C´ anovas A 15 , Alonso-Losada MG 16 , ´ Alvarez Sauco M 17 , Arn´ aiz S 18 , Arribas S 4 , Ascunce Vidondo A 19 , Bernardo N 5 , Bejr-Kasem H 12 , Bl´ azquez Estrada M 20 , Botí M 4 , Borrue C 21 , Cabello Gonz´ alez C 19 , C´ amara Lorenzo A 22 , Casas E 18 , Catal´ an MJ 23 , Clavero Ibarra P 19 , Cortina Fern´ andez A 11 , Crespo Cuevas A 24 , Cubo E 18 , Díez- Fairen M 4 , Erro E 19 , Fern´ andez Guill´ an N 23 , G´ amez P 18 , Gallego M 25 , García Caldentey J 26 , García Campos C 27 , García Moreno JM 27 , Gast´ on I 19 , G´ omez Garre MP 1,2 , Gonz´ alez Aloy J 6 , Gonz´ alez Ardura J 28 , Gonz´ alez García B 25 , Gonz´ alez Palm´ as MJ 29 , Gonz´ alez Toledo GR 8 , Golpe Díaz A 11 , Grau Sol´ a M 30 , Guardia G 4 , Horta-Barba A 12 , Infante J 9 , Kurtis M 31 , Labandeira C 16 , Lacruz Bescos F 19 , Lage Castro M 29 , Legarda I 32 , L´ opez Díaz LM 33 , L´ opez Manzanares L 25 , L´ opez Seoane B 11 , L´ opez Ariztegui N 34 , Macías Y 35 , Martí Andres G 19 , Martí MJ 22 , Martínez- Castrillo JC 15 , Mata M 21 , McAfee D 36 , Meitín MT 33 , Men´ endez Gonz´ alez M 20 , M´ endez Del Barrio C 1,2 , Miranda Santiago J 18 , Moreno Di´ eguez A 11 , Morales MI 34 , Nogueira V 33 , Novo Amado A 11 , Novo-Ponte S 37 , Ord´ as C 38 , Pagonabarraga J 12 , Pascual-Sedano B 12 , Paz Gonz´ alez JM 14 , Pare´ es I 31 , P´ erez Fuertes A 11 , P´ erez Noguera R 27 , Planellas L 22 , Prats MA 6 , Prieto Jurczynska C 38 , Puente V 24 , Redondo Rafales N 25 , Rodríguez M´ endez L 11 , Rodríguez P´ erez AB 17 , Rold´ an F 1 , Ruíz De Arcos M 27 , S´ anchez Alonso P 37 , S´ anchez-Carpintero M 11 , S´ anchez Díez G 15 , S´ anchez Rodríguez A 9 , Santacruz P 22 , Seijo M 29 , Serarols A 29 , Segundo Rodríguez JC 34 , Sierra Pe˜ na M 9 , Su´ arez-Castro E 11 , Tartari JP 4 , Valero C 39 , Vargas L 1 , V´ azquez G´ omez R 29 , Vela L 35 , Villanueva C 23 , Vives B 32 , Villar MD 8 . 15. Hospital Universitario Ram´ on y Cajal. Madrid. 16. Hospital Meixoeiro, Complejo Hospitalario Universitario de Vigo. Vigo. 17. Hospital General Universitario de Elche, Elche, Spain. 18. Complejo Asistencial Universitario de Burgos, Burgos, Spain. 19. Complejo Hospitalario de Navarra. Pamplona. Navarra. 20. Hospital Universitario Central de Asturias, Oviedo, Spain. 21. Hospital Infanta Sofía, Madrid, Spain. 22. Hospital Clınic de Barcelona, Barcelona, Spain. 23. Hospital Clínico San Carlos, Madrid, Spain. 24. Hospital del Mar, Barcelona, Spain. 25. Hospital La Princesa, Madrid, Spain. 26. Centro Neurologico Oms 42, Palma de Mallorca. 27. Hospital Universitario Virgen Macarena, Sevilla. 28. Hospital Universitario Lucus Augusti (HULA), Lugo, Spain. 29. Complejo Hospitalario Universitario de Pontevedra (CHOP), Pontevedra, Spain. 30. Consorci Sanitari Integral, Hospital Moises Broggi, SantJoan Despı, Barcelona, Spain. 31. Hospital Ruber Internacional, Madrid, Spain. 32. Hospital Universitario Son Espases, Palma de Mallorca, Spain. 33. Hospital Da Costa de Burela, Lugo, Spain. 34. Complejo Hospitalario de Toledo. Toledo. 35. Fundacion Hospital de Alcorcon, Madrid, Spain. 36. University of Pennsylvania, Philadelphia, PA, USA. 37. Hospital Universitario Puerta del Hierro. Madrid. 38. Hospital Rey Juan Carlos, Madrid, Spain, Madrid, Spain. 39. Hospital Arnau de Vilanova, Valencia, Spain. Declaration of competing interest None. Acknowledgments This work was supported by the Alzheimer Forschung Initiative e.V. (AFI International Training Grant to MJG), the Instituto de Salud Carlos III-Fondo Europeo de Desarrollo Regional (ISCIII-FEDER) [PI14/01823, PI16/01575, PI18/01898, PI19/01576, PI20/00613], the Consejería de Economía, Innovaci´ on, Ciencia y Empleo de la Junta de Andalucía [CVI- 02526, CTS-7685], the Consejería de Salud y Bienestar Social de la Junta de Andalucía [PI-0471-2013, PE-0210-2018, PI-0459-2018, PE-0186- 2019], the Fundaci´ on Alicia Koplowitz and the Fundaci´ on “Curemos el Parkinson” (https://www.curemoselparkinson.org). MJG is supported by the “Miguel Servet” program [CP19/00031], MALE by the University of Seville [USE-20046-J], JFM by the “Sara Borrell” program [CD13/ 00229] and VI-PPIT-US from the University of Seville [USE-18817-A], SJ by the “Acci´ on B-Clínicos-Investigadores” program [B-0007-2019], and DMG by the “Río Hortega” program [CM18/00142]. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.parkreldis.2021.05.027. References [1] P. Svenningsson, E. Westman, C. Ballard, D. Aarsland, Cognitive impairment in patients with Parkinson’s disease: diagnosis, biomarkers, and treatment, the Lancet, Neurology 11 (8) (2012) 697–707. M.J. Grothe et al. 89 Parkinsonism and Related Disorders 88 (2021) 68–75 75 [2] N.I. Bohnen, M.J. Grothe, N.J. Ray, M. Muller, S.J. Teipel, Recent advances in cholinergic imaging and cognitive decline-Revisiting the cholinergic hypothesis of dementia, Current geriatrics reports 7 (1) (2018) 1–11. [3] H. Hanyu, Y. Tanaka, S. Shimizu, H. Sakurai, T. Iwamoto, K. Abe, Differences in MR features of the substantia innominata between dementia with Lewy bodies and Alzheimer’s disease, J. Neurol. 252 (4) (2005) 482–484. [4] S.H. Choi, T.M. Jung, J.E. Lee, S.K. Lee, Y.H. Sohn, P.H. Lee, Volumetric analysis of the substantia innominata in patients with Parkinson’s disease according to cognitive status, Neurobiol. Aging 33 (7) (2012) 1265–1272. [5] M.J. Grothe, C. Schuster, F. Bauer, H. Heinsen, J. Prudlo, S.J. Teipel, Atrophy of the cholinergic basal forebrain in dementia with Lewy bodies and Alzheimer’s disease dementia, J. Neurol. 261 (10) (2014) 1939–1948. [6] H. Hanyu, S. Shimizu, Y. Tanaka, K. Hirao, T. Iwamoto, K. Abe, MR features of the substantia innominata and therapeutic implications in dementias, Neurobiol. Aging 28 (4) (2007) 548–554. [7] J.E. Lee, K.H. Cho, S.K. Song, H.J. Kim, H.S. Lee, Y.H. Sohn, P.H. Lee, Exploratory analysis of neuropsychological and neuroanatomical correlates of progressive mild cognitive impairment in Parkinson’s disease, J. Neurol. Neurosurg. Psychiatr. 85 (1) (2014) 7–16. [8] M. Grothe, H. Heinsen, S.J. Teipel, Atrophy of the cholinergic Basal forebrain over the adult age range and in early stages of Alzheimer’s disease, Biol. Psychiatr. 71 (9) (2012) 805–813. [9] N.J. Ray, S. Bradburn, C. Murgatroyd, U. Toseeb, P. Mir, G.K. Kountouriotis, S. J. Teipel, M.J. Grothe, In vivo cholinergic basal forebrain atrophy predicts cognitive decline in de novo Parkinson’s disease, Brain : J. Neurol. 141 (1) (2018) 165–176. [10] M.J. Barrett, S.A. Sperling, J.C. Blair, C.S. Freeman, J.L. Flanigan, M.E. Smolkin, C. A. Manning, T.J. Druzgal, Lower volume, more impairment: reduced cholinergic basal forebrain grey matter density is associated with impaired cognition in Parkinson disease, J. Neurol. Neurosurg. Psychiatr. 90 (11) (2019) 1251–1256. [11] J. Schulz, G. Pagano, J.A. Fern´ andez Bonfante, H. Wilson, M. Politis, Nucleus basalis of Meynert degeneration precedes and predicts cognitive impairment in Parkinson’s disease, Brain : J. Neurol. 141 (5) (2018) 1501–1516. [12] E.J. Burton, I.G. McKeith, D.J. Burn, E.D. Williams, J.T. O’Brien, Cerebral atrophy in Parkinson’s disease with and without dementia: a comparison with Alzheimer’s disease, dementia with Lewy bodies and controls, Brain : J. Neurol. 127 (Pt 4) (2004) 791–800. [13] T.R. Melzer, R. Watts, M.R. MacAskill, T.L. Pitcher, L. Livingston, R.J. Keenan, J. C. Dalrymple-Alford, T.J. Anderson, Grey matter atrophy in cognitively impaired Parkinson’s disease, J. Neurol. Neurosurg. Psychiatr. 83 (2) (2012) 188–194. [14] G.A. Carlesimo, F. Piras, F. Assogna, F.E. Pontieri, C. Caltagirone, G. Spalletta, Hippocampal abnormalities and memory deficits in Parkinson disease: a multimodal imaging study, Neurology 78 (24) (2012) 1939–1945. [15] F. Sampedro, S. Martínez-Horta, J. Marín-Lahoz, J. Pagonabarraga, J. Kulisevsky, Longitudinal intracortical diffusivity changes in de-novo Parkinson’s disease: a promising imaging biomarker, Park. Relat. Disord. 68 (2019) 22–25. [16] D. Santos-Garcia, P. Mir, E. Cubo, L. Vela, M.C. Rodriguez-Oroz, M.J. Marti, J. M. Arbelo, J. Infante, J. Kulisevsky, P. Martinez-Martin, COPPADIS-2015 (COhort of Patients with Parkinson’s DIsease in Spain, 2015), a global–clinical evaluations, serum biomarkers, genetic studies and neuroimaging–prospective, multicenter, non-interventional, long-term study on Parkinson’s disease progression, BMC Neurol. 16 (2016) 26. [17] J. Pagonabarraga, J. Kulisevsky, G. Llebaria, C. Garcia-Sanchez, B. Pascual-Sedano, A. Gironell, Parkinson’s disease-cognitive rating scale: a new cognitive scale specific for Parkinson’s disease, Mov. Disord. : official journal of the Movement Disorder Society 23 (7) (2008) 998–1005. [18] D. Wolf, M. Bocchetta, G.M. Preboske, M. Boccardi, M.J. Grothe, Reference Standard Space hippocampus Labels According to the EADC-ADNI Harmonized Protocol: Utility in Automated Volumetry, Alzheimers Dement, 2017. [19] M.J. Grothe, I. Kilimann, L. Grinberg, H. Heinsen, S. Teipel, In vivo volumetry of the cholinergic basal forebrain, in: R. Perneczky (Ed.), Biomarkers for Preclinical Alzheimer’s Disease, Springer Science+Business Media, LLC, 2018, pp. 213–232. [20] H.J. Fritz, N. Ray, M. Dyrba, C. Sorg, S. Teipel, M.J. Grothe, The Corticotopic Organization of the Human Basal Forebrain as Revealed by Regionally Selective Functional Connectivity Profiles, Hum. Brain Mapp. 40 (3) (2019 Feb 15) 868–878, https://doi.org/10.1002/hbm.24417. Epub 2018 Oct 11. PMID: 30311315. [21] R.A. Armstrong, When to use the Bonferroni correction, Ophthalmic Physiol. Opt. : the journal of the British College of Ophthalmic Opticians (Optometrists) 34 (5) (2014) 502–508. [22] A.K. Liu, R.C. Chang, R.K. Pearce, S.M. Gentleman, Nucleus basalis of Meynert revisited: anatomy, history and differential involvement in Alzheimer’s and Parkinson’s disease, Acta Neuropathol. 129 (4) (2015) 527–540. [23] N.I. Bohnen, M.L. Müller, V. Kotagal, R.A. Koeppe, M.A. Kilbourn, R.L. Albin, K. A. Frey, Olfactory dysfunction, central cholinergic integrity and cognitive impairment in Parkinson’s disease, Brain : J. Neurol. 133 (Pt 6) (2010) 1747–1754. [24] S. van der Zee, M.L.T.M. Müller, P. Kanel, T. van Laar, N.I. Bohnen, Cholinergic denervation patterns across cognitive domains in Parkinson’s disease, Mov. Disord. : official journal of the Movement Disorder Society 36 (3) (2021 Mar) 642–650, https://doi.org/10.1002/mds.28360. Epub 2020 Nov 2. PMID: 33137238. [25] A.S. Mihaescu, M. Masellis, A. Graff-Guerrero, J. Kim, M. Criaud, S.S. Cho, C. Ghadery, M. Valli, A.P. Strafella, Brain degeneration in Parkinson’s disease patients with cognitive decline: a coordinate-based meta-analysis, Brain imaging and behavior 13 (4) (2019) 1021–1034. [26] F. Gargouri, C. Gallea, M. Mongin, N. Pyatigorskaya, R. Valabregue, C. Ewenczyk, M. Sarazin, L. Yahia-Cherif, M. Vidailhet, S. Leh´ ericy, Multimodal magnetic resonance imaging investigation of basal forebrain damage and cognitive deficits in Parkinson’s disease, Mov. Disord. : official journal of the Movement Disorder Society 34 (4) (2019) 516–525. [27] S.J. Teipel, S. Reuter, B. Stieltjes, J. Acosta-Cabronero, U. Ernemann, A. Fellgiebel, M. Filippi, G. Frisoni, F. Hentschel, F. Jessen, S. Kl¨ oppel, T. Meindl, P.J. W. Pouwels, K.H. Hauenstein, H. Hampel, Multicenter stability of diffusion tensor imaging measures: a European clinical and physical phantom study, Psychiatr. Res. 194 (3) (2011) 363–371. [28] J. Peter, J. Lahr, L. Minkova, E. Lauer, M.J. Grothe, S. Teipel, L. K¨ ostering, C. P. Kaller, B. Heimbach, M. Hüll, C. Normann, C. Nissen, J. Reis, S. Kl¨ oppel, Contribution of the cholinergic system to verbal memory performance in mild cognitive impairment, J. Alzheim. Dis. : JAD 53 (3) (2016) 991–1001. [29] S.J. Teipel, H.C. Fritz, M.J. Grothe, Neuropathologic features associated with basal forebrain atrophy in Alzheimer disease, Neurology 95 (10) (2020) e1301–e1311. [30] S. Teipel, A. Drzezga, M.J. Grothe, H. Barthel, G. Chetelat, N. Schuff, P. Skudlarski, E. Cavedo, G.B. Frisoni, W. Hoffmann, J.R. Thyrian, C. Fox, S. Minoshima, O. Sabri, A. Fellgiebel, Multimodal imaging in Alzheimer’s disease: validity and usefulness for early detection, the Lancet, Neurology 14 (10) (2015) 1037–1053. M.J. Grothe et al. 90 Supplementary data Grothe et al. In vivo cholinergic basal forebrain degeneration and cognition in Parkinson’s disease: imaging results from the COPPADIS study Inclusion and exclusion criteria of the COPPADIS-2015 cohort Patients were recruited from a total of five different clinical research centers in Spain between January 2016 and October 2017. Patients were included if they had idiopathic PD according to UK Parkinson’s Disease Society Brain Bank criteria, were aged between 30 and 75 years, and did not meet screening criteria for dementia (Mini Mental State Examination (MMSE) ≥ 26). Patients were excluded from the study if they were not capable of completing the questionnaires adequately, had other disabling concomitant neurological disease, other severe and disabling concomitant non-neurological disease, known chronic anemia and/or hyperuricemia, or were receiving active treatment with continuous infusion of levodopa and/or apomorphine and/or with deep brain stimulation. Figure S1. Cholinergic basal forebrain region of interest Coronal brain sections from anterior (left) to posterior (right) showing the cholinergic basal forebrain (CBF) region of interest, separated into its functionally-defined anterior-medial (aCBF, green) and posterior-lateral (pCBF, red) subdivisions. When compared to cytoarchitectonically-defined CBF subdivisions, the aCBF largely corresponds to the rostral nuclei of the medial septum/diagonal band and anterior medial parts of the Nucleus basalis of Meynert (NBM), whereas the pCBF covers the remaining anterior-lateral, intermediate, and posterior parts of the NBM [1]. Numbers above sections indicate Montreal Neurological Institute (MNI) y-coordinates. 91 Supplementary Table S1. Acquisition parameters of structural MRI and DTI data Abbreviations: TR = Repetition Time; TE = Echo Time; HC = healthy controls; PD = Parkinson’s disease Sequence Parameters Research center 1 (Signa HDxt GE Medical Systems) Research center 2 (Achieva Philips Healthcare) Research center 3 (Ingenia Philips Healthcare) Research center 4 (Signa HDxt GE Medical Systems) Research center 5 (TrioTim Siemens) T1- weighted N (HC/PD) 25/91 10/14 10/46 0/9 0/20 Magnetic Field (T) 1.5 3 3 3 3 Sequence Name AX 3D FSPGR TRA 3D T1 3D TFE SENSE 3D FSPGR MPRAGE Acquistion mode Axial Axial Sagittal Sagittal Sagittal TR (ms) 10.5 8.4 8.2 5.5 5000 TE (ms) 3.2 3.9 3.75 1.3 2.9 Bandwidth (Hz/pixel) 122 299 191 122 238 Flip Angle (º) 20 8 8 12 0 Slice Thickness (mm) 1 0.92 1 1 1 In-plane resolution (mm) 0.94x0.94 0.83x0.83 0.94x0.94 0.51x0.51 1.00x1.00 Matrix 256x256x144 288x288x140 256x256x180 512x512x196 256x240x176 Diffusionweighted N (HC/PD) 0/44 0/9 0/20 Sequence Name DTI high iso SENSE EP – Axial Tensor EP2D Acquisition mode Axial Axial Axial N Gradient Directions B0+32 B0+33 B0+30 B value 800 1000 800 TR (s) 4.2 14 9.2 TE (ms) 100 72 90 Flip Angle (º) 90 90 90 Slice Thickness (mm) 2 2.4 2 In-plane resolution (mm) 1.75x1.75 1.02x1.02 1.96x1.96 Matrix 128x128x60 256x256x64 122x122x65 92 Association between CBF subregion volumes and cognition in PD In subregion-specific analyses all associations between CBF volumes and cognitive scores were slightly more pronounced for the pCBF than for the aCBF (Supplementary Table S2). However, none of the differences in correlation strength between subregions were statistically significant according to Steiger’s Z-test (p>0.83). Different CBF subdivisions are characterized by differential cortical cholinergic innervation patterns and have been linked to different aspects of cognition in animal models [2], but few studies so far have analysed the implications of different CBF subdivisions in PD- related cognitive decline [3, 4]. Previous neuropathological [3] and structural MRI studies [5] had specifically implicated posterior parts of the CBF in early and most pronounced neurodegeneration in Alzheimer’s disease, but a similar subregion-specific CBF involvement has not yet been established for PD [3]. One recent longitudinal MRI study found that NBM volumes (similar to our pCBF ROI) were already reduced at baseline in PD patients who later progressed to PD dementia, but anterior sectors of the CBF showed increased volume loss over time, suggesting a posterior-anterior pattern of CBF atrophy in the progression to dementia in PD [6]. It remains to be determined whether this progressive atrophy pattern is reflected by a sequential decline in different neuropsychological functions. Supplementary Table S2. Linear model stats for the association of cholinergic basal forebrain subregion volumes with cognitive scores Linear model stats are presented as (partial) r and T value (P value) for those PD-CRS items that showed a significant association with total cholinergic basal forebrain volume. For each cognitive item score, results from single predictor linear regression models (controlled for acquisition site and TIV) are indicated in the first row, and for significant models fully covariate-controlled regression results are indicated in the second row. Abbreviations: PD-CRS = PD-Cognitive Rating Scale; aCBF = anterior-medial cholinergic basal forebrain; pCBF = posterior-lateral cholinergic basal forebrain. Cognitive scores Cholinergic Basal Forebrain volume aCBF pCBF Partial r T value (P value) P(FDR) Partial r T value (P value) P(FDR) PD-CRStotal 0.3 4.20 (<0.001) <0.001 0.39 5.50 (<0.001) <0.001 0.17 2.18 (0.031) 0.031 0.24 3.24 (0.001) 0.003 Working memory 0.29 4.05 (<0.001) <0.001 0.31 4.34 (<0.001) <0.001 0.17 2.24 (0.026) 0.052 0.18 2.3 (0.023) 0.027 Alternating verbal fluency 0.22 3.03 (0.003) 0.003 0.29 3.98 (<0.001) <0.001 0.12 1.5 (0.135) 0.135 0.17 2.23 (0.027) 0.027 Immediate verbal memory 0.34 4.73 (<0.001) <0.001 0.35 4.92 (<0.001) <0.001 0.26 3.44 (0.001) 0.003 0.24 3.24 (0.001) 0.006 Delayed verbal memory 0.25 3.40 (0.001) 0.001 0.33 4.64 (<0.001) <0.001 0.12 1.53 (0.128) 0.135 0.2 2.65 (0.009) 0.017 93 Supplementary Table S3. Linear model stats for the association of volumetric imaging markers with cognitive scores in the patient subsample with DTI data Linear model stats are presented as (partial) r and T value (P value). For each cognitive item score, results from single predictor linear regression models (controlled for acquisition site and TIV) are indicated in the first row, and for significant models fully covariate-controlled regression results are indicated in the second row. Abbreviations: PD-CRS = PD-Cognitive Rating Scale. Cognitive scores Volume Cholinergic Basal Forebrain Hippocampus Partial r T value (P value) P(FDR) Partial r T value (P value) P(FDR) PD-CRStotal 0.45 4.20 (<0.001) <0.001 0.19 1.57 (0.121) 0.121 0.24 1.91 (0.061) 0.061 Fronto-subcortical items Sustained attention 0.17 1.45 (0.152) 0.171 0.05 0.41 (0.684) 0.769 Working memory 0.40 3.57 (0.001) 0.002 0.20 1.67 (0.099) 0.223 0.15 1.21 (0.232) 0.348 Clock drawing 0.17 1.45 (0.151) 0.171 0.06 0.47 (0.636) 0.769 Alternating verbal fluency 0.38 3.38 (0.001) 0.003 0.16 1.32 (0.192) 0.346 0.26 2.14 (0.036) 0.108 Action verbal fluency 0.24 2.01 (0.048) 0.072 0.03 0.24 (0.812) 0.812 0.742 Immediate verbal memory 0.43 3.94 (<0.001) 0.002 0.21 1.75 (0.084) 0.223 0.31 2.57 (0.013) 0.076 Delayed verbal memory 0.39 3.51 (0.001) 0.002 0.26 2.23 (0.029) 0.223 0.20 1.60 (0.115) 0.229 0.12 0.93 (0.357) 0.357 Posterior-cortical items Naming 0.16 1.31 (0.193) 0.193 0.08 0.62 (0.536) 0.769 Clock copy 0.24 2.05 (0.045) 0.072 0.21 1.78 (0.08) 0.223 0.04 0.33 (0.742) 0.742 0.16 1.24 (0.221) 94 Supplementary Table S4a. Meta-analysis of the association between volumetric imaging markers and total PD-CRS Volume Cholinergic Basal Forebrain Hippocampus Acquisition site n Partial r [C.I.] Partial r [C.I.] Research center 1 91 0.28 [0.08; 0.46] 0.31 [0.11; 0.48] Research center 2 14 0.48 [-0.07; 0.81] 0.3 [-0.27; 0.72] Research center 3 46 0.45 [0.18; 0.65] 0.02 [-0.27; 0.31] Research center 4 9 0.2 [-0.54; 0.76] 0.35 [-0.41; 0.82] Research center 5 20 0.55 [0.14; 0.8] 0.36 [-0.1; 0.69] Meta-analysis Pooled partial r (P value) 0.37 [0.23; 0.49] (<0.001) 0.24 [0.1; 0.38] (0.001) Between-study heterogeneity I2 index 0% 0% 𝜏2 0.0083 0.0078 Q value (P value) 2.6 (0.628) 3.11 (0.54) Partial r reflects the correlation between regional brain volumes and PD-CRStotal score, after accounting for total intracranial volume 95 Supplementary Table S4b. Meta-analysis of the association between diffusivity imaging markers and total PD-CRS Mean Diffusivity Cholinergic Basal Forebrain Hippocampus Acquisition site n r [C.I.] r [C.I.] Research center 3 44 -0.05 [-0.34; 0.25] -0.49 [-0.69; -0.23] Research center 4 9 -0.31 [-0.81; 0.45] -0.18 [-0.75; 0.55] Research center 5 20 -0.06 [-0.49; 0.39] -0.44 [-0.74; 0] Meta-analysis Pooled r (P value) -0.08 [-0.31; 0.17] (0.532) -0.45 [-0.62; -0.24] (<0.001) Between-study heterogeneity I2 index 0% 0% 𝜏2 0.0027 0.0062 Q value (P value) 0.39 (0.823) 0.66 (0.719) 96 References [1] H.-C.J. Fritz, N. Ray, M. Dyrba, C. Sorg, S. Teipel, M.J. Grothe, The corticotopic organization of the human basal forebrain as revealed by regionally selective functional connectivity profiles, Human brain mapping (2018). [2] L. Záborszky, P. Gombkoto, P. Varsanyi, M.R. Gielow, G. Poe, L.W. Role, M. Ananth, P. Rajebhosale, D.A. Talmage, M.E. Hasselmo, H. Dannenberg, V.H. Minces, A.A. Chiba, Specific Basal Forebrain-Cortical Cholinergic Circuits Coordinate Cognitive Operations, The Journal of neuroscience : the official journal of the Society for Neuroscience 38(44) (2018) 9446-9458. [3] A.K. Liu, R.C. Chang, R.K. Pearce, S.M. Gentleman, Nucleus basalis of Meynert revisited: anatomy, history and differential involvement in Alzheimer's and Parkinson's disease, Acta neuropathologica 129(4) (2015) 527-40. [4] N.I. Bohnen, M.J. Grothe, N.J. Ray, M. Muller, S.J. Teipel, Recent advances in cholinergic imaging and cognitive decline-Revisiting the cholinergic hypothesis of dementia, Current geriatrics reports 7(1) (2018) 1-11. [5] M. Grothe, H. Heinsen, S.J. Teipel, Atrophy of the cholinergic Basal forebrain over the adult age range and in early stages of Alzheimer's disease, Biological psychiatry 71(9) (2012) 805-13. [6] J.B. Pereira, S. Hall, M. Jalakas, M.J. Grothe, O. Strandberg, E. Stomrud, E. Westman, D. van Westen, O. Hansson, Longitudinal degeneration of the basal forebrain predicts subsequent dementia in Parkinson's disease, Neurobiology of disease 139 (2020) 104831. 103 correction, significant effects only remained for MoCA performance (P<0.001)anddelayedrecall(P=0.002). Cortical areas related to cBF degeneration were also significantly associated with global cognitive decline. Thus, among cBF-associated cortical regions, decline in MoCA score correlated with cortical thinning in medial superior and lateral frontal areas, lateral temporoparietal areas, entorhinal cortex, cuneus, precuneus, and posterior cingulate (average effect size: r=0.37, P< 0.001; Figs. 2a,b, S6.1 in Data S1). Interestingly, path analyses showed that approximately half of the total effect of cBF degeneration on cognitive decline (total effect: β=0.16 [0.08, 0.25], P< 0.001) was mediated by cortical thinning in the cognition-related areas (ADE: β=0.08 [!0.01, 0.17], P=0.064; ACME: β=0.08 [0.01, 0.16], P=0.02; mediated proportion: 48.5%) (Fig. 2c–e). When adding slopes of cBF decline to the regression model between cortical thinning and MoCA decline, average cortical thinning in these regions remained significantly associated with MoCA decline (β=0.29 vs. β’=0.23), although the total effect was reduced by 20%. All associations have been confirmed using robust regression models that account for possible outlier observations (Supplementary data 7 in Data S1). Similar mediation effects of cortical thinning were also observed for the association between longitudinal cBF degeneration and cognitive decline in delayed recall (Supplementary data 8 in Data S1). Effects of Comorbid Amyloid-βPathology on the Observed Associations Interestingly, both amyloid-β-positive and -negative patients showed similar degrees of longitudinal cBF FIG. 2. Mediation analysis of associations between cholinergic basal forebrain (cBF) degeneration, cortical thinning, and cognitive decline. (a) Brain renderings of cortical clusters showing significant correlations between vertex-wise slopes of cortical thickness change and the slope of change in Montreal Cognitive Assessment (MoCA) score. Vertex-wise regression analyses were restricted to cortical areas showing significant associations with cBF degeneration (as depicted in Fig. 1) and corrected for multiple comparisons using a false discovery rate (FDR)-corrected vertex-level threshold of P< 0.05. Effects are depicted on cortical surfaces showing medial, lateral, superior, and inferior brain renderings. (b) Average individual slopes of thickness change in cognition-related cortical areas (y-axis) are plotted against individual slopes of change in MoCA score (x-axis). Red line represents linear trend. (c) Individual slopes of change in cBF volume (x-axis) are plotted against individual slopes of change in MoCA score (y-axis). Blue line represents linear trend. (d) Individual slopes of change in cBF volume (x-axis) are plotted against average individual slopes of thickness change in cognition-related cortical areas (y-axis). Orange line represents linear trend. (e) Path diagram of causal mediation model where the effect of cBF degeneration on cognitive decline is plotted in path Z (blue arrow), the effect of cBF degeneration on cortical thinning is plotted in path X (orange arrow), and the effect of cortical thinning on cognitive decline is plotted in path Y (red arrow). Betas [95% confidence intervals] and Pvalues of regression analyses are indicated for each association separately. The total effect of path Zis composed of the sum of the average direct effect (Z ADE ) and the average causal mediation effect (Z ACME ) (purple arrows). [Color figure can be viewed at wileyonlinelibrary.com] 6Movement Disorders, 2023 LABRADOR-ESPINOSA ET AL 15318257, 0, Downloaded from https://movementdisorders.onlinelibrary.wiley.com/doi/10.1002/mds.29564 by Readcube (Labtiva Inc.), Wiley Online Library on [06/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 104 atrophy with no significant difference between groups (β=!2.11 [!17.24, 13.03]; t=!0.26; P=0.79). Similarly, significant associations between longitudinal cBF degeneration and parallel cortical thinning were observed independently of amyloid-βstatus (Fig. 3). However, some of the regional effects were significantly more pronounced in amyloid-β-positive compared to amyloid-β-negative PD patients, particularly in the posterior cingulate, superior parietal, and lateral frontal areas (Supplementary data 9 in Data S1). Discussion In the present study we analyzed in vivo associations between longitudinal cBF degeneration, cortical thinning, and cognitive decline in a large sample of de novo PD patients, and further investigated whether these associations depended on the presence of comorbid AD pathology as assessed by a CSF amyloid-βbiomarker. We found that longitudinal cBF degeneration was associated with parallel cortical thinning in cognitionrelevant cortical areas, including frontal areas, lateral FIG. 3. Associations between longitudinal cholinergic basal forebrain (cBF) degeneration and regional cortical thinning in amyloid-β-negative and -positive Parkinson’sdisease(PD)patients.Thefigure shows regional effects of regression analyses of the slope of change in cBF volume on vertex-wise slopes of change in cortical thickness, controlled for whole brain atrophy, in amyloid-β-negative (a) and amyloid-β-positive (b) PD patients separately. Results were corrected for multiple comparisons using a false discovery rate (FDR)-corrected vertex-level threshold of P FDR < 0.05. Effects are depicted on cortical surfaces showing medial, lateral, superior, and inferior brain renderings. [Color figure can be viewed at wileyonlinelibrary.com] Movement Disorders, 2023 7 BASAL FOREBRAIN AND CORTICAL THINNING IN PD 15318257, 0, Downloaded from https://movementdisorders.onlinelibrary.wiley.com/doi/10.1002/mds.29564 by Readcube (Labtiva Inc.), Wiley Online Library on [06/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 105 temporoparietal areas, and posterior regions. The parallel cortical thinning in these areas mediated the association between cBF degeneration and cognitive decline. Interestingly, significant associations between cBF degeneration and cortical thinning were observed independent of amyloid-βstatus, although associations with posterior cortical areas were significantly more pronounced in amyloid-β-positive patients. Given that the loss of cholinergic neurons is a wellknown neuropathologic feature of dementia in PD, 6 recent in vivo MRI morphometry studies have investigated the cognitive implications of cBF degeneration in predementia phases of PD. These studies could demonstrate that cBF degeneration, as measured by lower GM volume on MRI, is robustly associated with cognitive impairments even in non-demented PD patients. 9,10 Furthermore, reduced cBF volume was also found to predict future cognitive decline in PD patients with intact cognition at baseline. 8,11,12 In line with these cross-sectional cBF measurements, in our current serial MRI study we found that longitudinal cBF degeneration was paralleled by global and domain-specific cognitive decline, particularly in the memory domain. In another line of research, several brain morphometry studies using MRI have reported associations between cognitive impairment and cortical atrophy in PD, particularly in frontotemporal and posterior cortical areas. 18,19 Together, this suggests that cBF degeneration and cortical atrophy may be interrelated phenomena linked to declining cognition in pathophysiologic PD progression. In fact, neuropathologic studies of both PD and AD patients have described close relationships between loss of cholinergic BF neurons and pathologic changes in cortical areas. 13,14 Moreover, recent neuroimaging studies of AD could also demonstrate in vivo associations between cBF degeneration and atrophy in highly innervated cortical target areas of the cBF projections. 15-17 Here, we extend these findings to PD, demonstrating for the first time that a similar in vivo association between longitudinal cBF degeneration and parallel cortical thinning can also be observed in PD patients and that it links with cognitive decline even from an early disease stage. It should be noted that although the effect of cBF degeneration on cognitive decline is largely mediated through its association with cortical thinning, this does not exclude that cortical thinning may also contribute to cognitive decline in PD independently from cBF degeneration. Our additional analyses of cBF-independent effects of cortical degeneration on cognitive decline indicate that cortical thinning indeed appears to be the closest structural correlate of cognitive decline in PD, and parts of this process are linked to degeneration of the cortically-projecting cBF, mediating its impact on cognitive decline. The exact pathophysiologic mechanisms underlying these associations remain largely unknown and cannot be inferred from our association analyses. However, a likely mechanism involves aggravation of cortical pathologic processes in cholinergically denervated areas, which may be caused by disrupted clearance mechanisms. Emerging evidence from basic science research emphasizes the important role of cholinergic innervations in a range of housekeeping and metabolic functions of cortical target neurons. 4 Furthermore, in experimental animal models of AD, selective cBF lesions lead to a rapid accumulation of cortical amyloid-βpathology. 26 Similarly, and with respect to our findings of the cortical mediation of cBF- associated cognitive decline, selective cBF lesions in animal models have been found to induce neuronal dysfunction in denervated cortical areas, where the degree of lesion-induced cognitive deficits depended on the severity of cortical dysfunction. 27 Of note, we cannot infer directionality from the observed associations, and an alternative explanation could be a retrograde degeneration of the cBF secondary to cortical atrophy, which may be caused by lack of trophic support to the cBF. Thus, due to their large axons that project to distant cortical and subcortical areas, cholinergic neurons are highly dependent on trophic support for cellular maintenance and the transport of biomolecules, and these mechanisms may be disrupted by primary degenerative processes in cortical target areas. 28 However, based on the assumed sequence of pathology progression as reflected by the Braak Lewy body staging scheme, 29 it appears unlikely that cortical atrophy would precede cBF degeneration in these early stages of PD. Given the prominent cBF degeneration in AD and the high comorbidity of AD pathology in PD patients experiencing cognitive decline, 30 the question arises whether cBF degeneration and the associated cortical atrophy in PD is mainly driven by patients with comorbid AD pathology or whether it is an intrinsic feature of PD pathology. Thus, previous CSF or PET-based imaging studies have shown that amyloid-β-positive PD patients show more pronounced cortical atrophy in AD-related areas compared to amyloidβ-negative PD patients. 31 However, to our knowledge, cBF atrophy in PD has not yet been assessed in relation to biomarkers of comorbid AD pathology. In our current study, we explicitly assessed this question using a CSF biomarker of amyloid-βpathology as an index of comorbid AD pathology. We found that both longitudinal cBF degeneration and its association with cortical thinning were largely independent of amyloid-βstatus, indicating that these processes are proper features of PD pathophysiology. Further molecular neuroimaging studies are needed to better understand the complex interactive effects of AD and Lewy body pathology on system-specificneu- rodegeneration and cognitive decline in PD. 32 Our study emphasizes the important role of cholinergic system degeneration in the development of cognitive deficits in early-stage PD, and may warrant a reevaluation of initiating cholinergic treatment at predementia stages of PD. 2 While clinical trials evaluating cholinergic treatments for predementia level cognitive impairment in PD have provided mixed results, 33 8Movement Disorders, 2023 LABRADOR-ESPINOSA ET AL 15318257, 0, Downloaded from https://movementdisorders.onlinelibrary.wiley.com/doi/10.1002/mds.29564 by Readcube (Labtiva Inc.), Wiley Online Library on [06/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 106 this may at least partially relate to the heterogeneity of cholinergic degneration among non-demented PD patients. 34 Our findings in de novo PD patients show that there is a high interindividual heterogeneity not only in trajectories of cognitive decline, 35 but also in the degree of cBF degeneration and associated cortical atrophy at this early disease stage. This leads us to hypothesize that the identification of patients with cholinergic degeneration may be crucial even at relatively early disease stages to provide effective treatment to those who could benefit most from such therapies. Consequently, neuroimaging biomarkers of cholinergic system degeneration may help to enrich clinical trial populations for those PD patients that are at highest risk for cognitive decline and most likely to benefit from cholinergic treatments. 2,8,36 In addition, cognitive/ clinical outcomes may not be sensitive enough to detect significant treatment effects over relatively short followup intervals in non-demented patients. 33 In a randomized placebo-controlled study of donepezil in amnestic MCI patients, a possible prodrome to AD dementia, it was found that although donepezil treatment did not have a significant effect on cognitive outcomes over 12 months, it did lead to a significantly reduced atrophy not only of the cBF, but also of relevant cortical target areas. 37,38 Whether similar treatment effects may be achieved in PD patients is fully hypothetical at the moment and additional research is necessary to assess the possible advantages of stratifying PD patients according to individual degrees of cholinergic system degeneration when assessing the efficacy and potential neuroprotective benefits of cholinergic therapy at predementia stages of PD. Our work also presents some limitations that should be considered. First, serial MRI data in the healthy control group was only available for 1 year of follow-up, limiting the assessment of the specificity of the observed effects for PD compared to the normal aging process. However, in a sensitivity analysis using only the 1-year follow-up interval, PD patients continued to show longitudinal decline in cBF volume, cortical thickness, and cognition, whereas no such changes were observed in the healthy controls (Supplementary data 3 in Data S1). Second, cortical cholinergic signaling has not been assessed directly, and thus its role in the observed associations remains unclear. While cBF volume on MRI is only an indirect structural marker of cholinergic system degeneration, there is accumulating evidence from multimodal biomarker studies that it is tightly related to cholinergicsystemfunctioningasassessed by neurophysiologic or molecular imaging markers of cholinergic system integrity. 39,40 Finally, cholinergic system degeneration in PD does not occur in isolation but together with pronounced degenerative changes in the dopaminergic and other neurotransmitter systems, 41 which have not been assessed in our study. Multimodal neuroimaging studies are needed to better understand the unique and interactive contributions of transmitter-specificneu- rodegeneration to cognitive impairments in PD. In conclusion, our study findings emphasize the important role of cholinergic systemdegenerationinthedevelopment of cognitive deficits in PD, and provide novel insights into its relation to regional cortical atrophy and amyloid-βpathology as two established biomarkers that have previously been linkedtocognitivedeclinein PD. The detected relationships may have important implications for biomarker-based disease prognosis and clinical trial stratification, and may warrant a re-evaluation of initiating cholinergic treatment at predementia stages of PD. Acknowledgments: The authors thank the Instituto de Biomedicina de Sevilla and the Hospital Universitario Virgen del Rocío for the research resources used in this study. Data used in the preparation of this manuscript were obtained from the Parkinson’s Progression Markers Initiative database (www.ppmi-info.org/data). For up-to-date information on the study visit www.ppmi-info.org. Parkinson’s Progression Markers Initiative—a public– private partnership—is funded by The Michael J. Fox Foundation for Parkinson’s Research. Corporate funding partners include AbbVie, Allergan, Amathus, Avid, Biogen, BioLegend, Bristol-Myers Squibb, Celgene, Denali, GE Healthcare, Genentech, GlaxoSmithKline, Janssen Neurosciene, Lilly, Lundbeck, Merck, Meso Scale Discovery, Pfizer, Piramal, Prevail, Roche, SanofiGenzyme, Servier, Takeda, Teva, UCB, Verily, and Voyager, and the philanthropic funding partner is Golub Capital. Data Availability Statement Data used in the preparation of this study were obtained from the Parkinson’s Progression Markers Initiative database. All data are full access and available at www.ppmi-info.org. References 1. Aarsland D, Batzu L, Halliday GM, et al. Parkinson diseaseassociated cognitive impairment. Nat Rev Dis Prim 2021;7(1):1–21. https://doi.org/10.1038/s41572-021-00280-3 2. Bohnen NI, Yarnall AJ, Weil RS, et al. Cholinergic system changes in Parkinson’s disease: emerging therapeutic approaches. Lancet Neurol 2022;21(4):381–392. https://doi.org/10.1016/S1474-4422(21)00377-X 3. Ballinger EC, Ananth M, Talmage DA, Role LW. Basal forebrain cholinergic circuits and signaling in cognition and cognitive decline. Neuron 2016;91(6):1199–1218. https://doi.org/10.1016/j.neuron. 2016.09.006 4. Ovsepian SV, O’Leary VB, Zaborszky L. Cholinergic mechanisms in the cerebral cortex: beyond synaptic transmission. Neuroscientist 2016;22(3):238–251. https://doi.org/10.1177/1073858415588264 5. Jellinger K. Quantitative changes in some subcortical nuclei in aging, Alzheimer’s disease and Parkinson’s disease. Neurobiol Aging 1987; 8(6):556–561. https://doi.org/10.1016/0197-4580(87)90134-5 6. Liu AKL, Chang RCC, Pearce RKB, Gentleman SM. Nucleus basalis of Meynert revisited: anatomy, history and differential involvement in Alzheimer’s and Parkinson’s disease. Acta Neuropathol 2015; 129(4):527–540. https://doi.org/10.1007/s00401-015-1392-5 7. Grothe M, Heinsen H, Teipel SJ. Atrophy of the cholinergic basal forebrain over the adult age range and in early stages of Alzheimer’s disease. Biol Psychiatry 2012;71(9):805–813. https://doi.org/10. 1016/j.biopsych.2011.06.019 8. Ray NJ, Bradburn S, Murgatroyd C, et al. In vivo cholinergic basal forebrain atrophy predicts cognitive decline in de novo Parkinson’s disease. Brain 2018;141(1):165–176. https://doi.org/10.1093/brain/ awx310 9. Grothe MJ, Labrador-Espinosa MA, Jesús S, et al. In vivo cholinergic basal forebrain degeneration and cognition in Parkinson’s disease: Movement Disorders, 2023 9 BASAL FOREBRAIN AND CORTICAL THINNING IN PD 15318257, 0, Downloaded from https://movementdisorders.onlinelibrary.wiley.com/doi/10.1002/mds.29564 by Readcube (Labtiva Inc.), Wiley Online Library on [06/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 107 imaging results from the COPPADIS study. Park Relat Disord 2021; 88:68–75. https://doi.org/10.1016/j.parkreldis.2021.05.027 10. Barrett MJ, Sperling SA, Blair JC, et al. Lower volume, more impairment: reduced cholinergic basal forebrain grey matter density is associated with impaired cognition in Parkinson disease. J Neurol Neurosurg Psychiatry 2019;90(11):1251–1256. https://doi.org/10. 1136/jnnp-2019-320450 11. Pereira JB, Hall S, Jalakas M, et al. Longitudinal degeneration of the basal forebrain predicts subsequent dementia in Parkinson’s disease. Neurobiol Dis 2019;2020(139):104831. https://doi.org/10.1016/j. nbd.2020.104831 12. Schulz J, Pagano G, Fern! andez Bonfante JA, Wilson H, Politis M. Nucleus basalis of Meynert degeneration precedes and predicts cognitive impairment in Parkinson’s disease. Brain 2018;141(5):1501– 1516. https://doi.org/10.1093/brain/awy072 13. Cullen KM, Halliday GM, Double KL, Brooks WS, Creasey H, Broe GA. Cell loss in the nucleus basalis is related to regional cortical atrophy in Alzheimer’s disease. Neuroscience 1997;78(3):641– 652. https://doi.org/10.1016/S0306-4522(96)00569-6 14. Lippa CF, Smith TW, Perry E. Dementia with Lewy bodies: choline acetyltransferase parallels nucleus basalis pathology. J Neural Transm 1999;106(5–6):525–535. https://doi.org/10.1007/s007020050176 15. Grothe MJ, Heinsen H, Amaro E, Grinberg LT, Teipel SJ. Cognitive correlates of basal forebrain atrophy and associated cortical hypometabolism in mild cognitive impairment. Cereb Cortex 2016; 26(6):2411–2426. https://doi.org/10.1093/cercor/bhv062 16. Kilimann I, Hausner L, Fellgiebel A, et al. Parallel atrophy of cortex and basal forebrain cholinergic system in mild cognitive impairment. Cereb Cortex 2017;27(3):1841–1848. https://doi.org/10.1093/cercor/bhw019 17. Schmitz TW, Mur M, Aghourian M, Bedard MA, Spreng RN. Longitudinal Alzheimer’s degeneration reflects the spatial topography of cholinergic basal forebrain projections. Cell Rep 2018;24(1):38–46. https://doi.org/10.1016/j.celrep.2018.06.001 18. Filippi M, Canu E, Donzuso G, et al. Tracking cortical changes throughout cognitive decline in Parkinson’s disease. Mov Disord 2020;35(11):1987–1998. https://doi.org/10.1002/mds.28228 19. Pereira JB, Svenningsson P, Weintraub D, et al. Initial cognitive decline is associated with cortical thinning in early Parkinson disease. Neurology 2014;82(22):2017–2025. https://doi.org/10.1212/ WNL.0000000000000483 20. Marek K, Chowdhury S, Siderowf A, et al. The Parkinson’sprogression markers initiative (PPMI)–establishing a PD biomarker cohort. Ann Clin Transl Neurol 2018;5(12):1460–1477. https://doi.org/10.1002/acn3.644 21. Stebbins GT, Goetz CG, Burn DJ, Jankovic J, Khoo TK, Tilley BC. How to identify tremor dominant and postural instability/gait difficulty groups with the Movement Disorder Society Unified Parkinson’s Disease Rating Scale: comparison with the Unified Parkinson’s Disease Rating Scale. Mov Disord 2013;28(5): 668–670. https://doi.org/10.1002/mds.25383 22. Irwin DJ, Fedler J, Coffey CS, et al. Evolution of Alzheimer’s disease cerebrospinal fluid biomarkers in early Parkinson’s disease. Ann Neurol 2020;88(3):574–587. https://doi.org/10.1002/ana.25811 23. Fritz HCJ, Ray N, Dyrba M, Sorg C, Teipel S, Grothe MJ. The corticotopic organization of the human basal forebrain as revealed by regionally selective functional connectivity profiles. Hum Brain Mapp 2019;40(3):868–878. https://doi.org/10.1002/hbm.24417 24. Desikan RS, Ségonne F, Fischl B, et al. An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. Neuroimage 2006;31(3):968–980. https://doi.org/10.1016/j.neuroimage.2006.01.021 25. Imai K, Keele L, Yamamoto T. Identification, inference and sensitivity analysis for causal mediation effects. Stat Sci 2010;25(1):51–71. https://doi.org/10.1214/10-STS321 26. Ramos-Rodriguez JJ, Pacheco-Herrero M, Thyssen D, et al. Rapid β-amyloid deposition and cognitive impairment after cholinergic denervation in APP/PS1 mice. J Neuropathol Exp Neurol 2013; 72(4):272–285. https://doi.org/10.1097/NEN.0b013e318288a8dd 27. Browne SE, Lin L, Mattsson A, Georgievska B, Isacson O. Selective antibody-induced cholinergic cell and synapse loss produce sustained hippocampal and cortical hypometabolism with correlated cognitive deficits. Exp Neurol 2001;170(1):36–47. https://doi.org/10.1006/exnr. 2001.7700 28. Mufson EJ, Ginsberg SD, Ikonomovic MD, DeKosky ST. Human cholinergic basal forebrain: chemoanatomy and neurologic dysfunction. J Chem Neuroanat 2003;26(4):233–242. https://doi.org/10. 1016/S0891-0618(03)00068-1 29. Braak H, Del Tredici K, Rüb U, De Vos RAI, Jansen Steur ENH, Braak E. Staging of brain pathology related to sporadic Parkinson’s disease. Neurobiol Aging 2003;24(2):197–211. https://doi.org/10. 1016/S0197-4580(02)00065-9 30. Siderowf A, Xie SX, Hurtig H, et al. CSF amyloid β1-42 predicts cognitive decline in Parkinson disease. Neurology 2010;75(12): 1055–1061. https://doi.org/10.1212/WNL.0b013e3181f39a78 31. McMillan CT, Wolk DA. Presence of cerebral amyloid modulates phenotype and pattern of neurodegeneration in early Parkinson’s disease. J Neurol Neurosurg Psychiatry 2016;87(10):1112–1122. https://doi.org/10.1136/jnnp-2015-312690 32. Yoo HS, Lee S, Chung SJ, et al. Dopaminergic depletion, β-amyloid burden, and cognition in Lewy body disease. Ann Neurol 2020; 87(5):739–750. https://doi.org/10.1002/ana.25707 33. Baik K, Kim SM, Jung JH, et al. Donepezil for mild cognitive impairment in Parkinson’s disease. Sci Rep 2021;11(1):1–9. https:// doi.org/10.1038/s41598-021-84243-4 34. Bohnen NI, Müller MLTM, Kotagal V, et al. Heterogeneity of cholinergic denervation in Parkinson’s disease without dementia. J Cereb Blood Flow Metab 2012;32(8):1609–1617. https://doi.org/ 10.1038/jcbfm.2012.60 35. Schrag A, Siddiqui UF, Anastasiou Z, Weintraub D, Schott JM. Clinical variables and biomarkers in prediction of cognitive impairment in patients with newly diagnosed Parkinson’s disease: a cohort study. Lancet Neurol 2017;16(1):66. https://doi.org/10.1016/S1474- 4422(16)30328-3 36. Teipel SJ, Cavedo E, Grothe MJ, et al. Predictors of cognitive decline and treatment response in a clinical trial on suspected prodromal Alzheimer’s disease. Neuropharmacology 2016;108:128–135. https://doi.org/10.1016/j.neuropharm.2016.02.005 37. Cavedo E, Grothe MJ, Colliot O, et al. Reduced basal forebrain atrophy progression in a randomized donepezil trial in prodromal Alzheimer’s disease. Sci Rep 2017;7(1):1–10. https://doi.org/10. 1038/s41598-017-09780-3 38. Cavedo E, Dubois B, Colliot O, et al. Reduced regional cortical thickness rate of change in donepezil-treated subjects with suspected prodromal Alzheimer’s disease. J Clin Psychiatry 2016;77(12): e1631–e1638. https://doi.org/10.4088/JCP.15m10413 39. Aghourian M, Aumont ! E, Grothe MJ, Soucy JP, Rosa-Neto P, Bedard MA. FEOBV-PET to quantify cortical cholinergic denervation in AD: relationship to basal forebrain volumetry. J Neuroimaging 2021;31(6):1077–1081. https://doi.org/10.1111/jon.12921 40. Peter J, Mayer I, Kammer T, et al. The relationship between cholinergic system brain structure and function in healthy adults and patients with mild cognitive impairment. Sci Rep 2021;11(1):1–7. https://doi.org/10.1038/s41598-021-95573-8 41. Bohnen NI, Albin RL, Müller MLTM, et al. Frequency of cholinergic and caudate nucleus dopaminergic deficits across the predemented cognitive spectrum of Parkinson disease and evidence of interaction effects. JAMA Neurol 2015;72(2):194–200. https://doi. org/10.1001/jamaneurol.2014.2757 Supporting Data Additional Supporting Information may be found in the online version of this article at the publisher’s web-site. 10 Movement Disorders, 2023 LABRADOR-ESPINOSA ET AL 15318257, 0, Downloaded from https://movementdisorders.onlinelibrary.wiley.com/doi/10.1002/mds.29564 by Readcube (Labtiva Inc.), Wiley Online Library on [06/10/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License 108 1 Supplementary data Supplementary figure 1. Flowchart of the complete analytical approach. ................................................................................................................................................................................................................... 2 Supplementary data 2. Longitudinal change estimations of volumetric, cortical thickness and cognitive measures in PD patients. .................................................................................................................... 4 Supplementary table 2.1. Longitudinal change of volumetric measures in PD patients. .................................................................................................................................................................................................................................................................. 4 Supplementary figure 2.1. Longitudinal change in vertex-wise cortical thickness in PD patients. .................................................................................................................................................................................................................................................. 5 Supplementary table 2.2. Longitudinal change in cognitive performance of PD patients. ............................................................................................................................................................................................................................................................... 6 Supplementary data 3. Longitudinal change estimations of volumetric, cortical thickness, and cognitive measures in healthy controls and in comparison to PD patients. ................................................. 7 Supplementary figure 3.1. Longitudinal change in cortical thickness over 1 year follow-up in healthy controls and PD patients. ................................................................................................................................................................................................ 8 Supplementary table 3.1. Longitudinal change in cognitive performance in healthy controls. ...................................................................................................................................................................................................................................................... 10 Supplementary data 4. Summary of regional effect sizes for the associations between longitudinal cBF degeneration and region-wise cortical thinning in PD. .................................................................. 11 Supplementary figure 4.1. Effect sizes of cBF-cortical thickness correlations within cortical regions defined in the Desikan-Killiany brain atlas. ................................................................................................................................................................... 11 Supplementary data 5. Differences in longitudinal cortical thinning between PD patients with and without fast cBF volume decline during the four years of follow-up. .................................................. 12 Supplementary data 6. Summary of regional effect sizes for the associations between cortical thinning in cBF-related regions and global cognitive decline. ...................................................................... 13 Supplementary figure 6.1. Effect sizes of MoCA-cortical thickness correlations within cortical regions defined in the Desikan-Killiany brain atlas. ............................................................................................................................................................... 13 Supplementary data 7. Robust regression analyses of the associations between cBF degeneration, cortical thinning, and cognitive decline. ................................................................................................... 14 Supplementary table 7.1. Results of robust associations and previously reported conventional associations. .............................................................................................................................................................................................................................. 15 Supplementary figure 7.1. Penalised cases in the robust regression models .................................................................................................................................................................................................................................................................................. 16 Supplementary data 8. Mediation analysis of associations between cBF degeneration, cortical thinning, and memory decline. ........................................................................................................................ 17 Supplementary figure 8.1. Cortical mediation of the association between longitudinal cBF degeneration and decline in delayed recall scores. ......................................................................................................................................................................... 17 Supplementary data 9. Regional differences between amyloidb -positive and –negative PD patients in the association between longitudinal cBF degeneration and cortical thinning. ............................ 19 Supplementary figure 9.1 Regional effect sizes of differences between amyloid-b-negative and -positive PD patients. .............................................................................................................................................................................................................. 19 Supplementary figure 9.2. Regional effect sizes of differences between amyloid-b-negative and -positive PD patients summarized across cortical regions defined in the Desikan-Killiany brain atlas. .............................................................................. 20 109 2 Supplementary figure 1. Flowchart of the complete analytical approach. Longitudinal processing pipeline [CAT12] Surface-based morphometry (SBM) 1. Cortical surface and thickness estimation 2. Resample Freesurfer [364k] 3. Smoothing with Gaussian kernel Voxel-based morphometry (VBM) 1. GM Segmentation 2. Registration (Dartel) 3. Warping & Modulation . . . . . . . . . PD patient 1 PD patients PD patient N Visits Baseline 1st Year 2nd Year 4th Year Longitudinal MRI Longitudinal cognitive scales PD patient 1 PD patients Visits Baseline 1st Year 2nd Year 4th Year 3rd Year PD patient N . . . PD patients Visits Longitudinal dataset for a cognitive scale (e.g. MoCA) model_cognition <-lmer(cognitive_score ~follow_up_visit + sex + age + years_of_education + (follow_up_visit|PATNO_ID), data) get_individual_slopes(model_cognition) PD patient 1 PD patient N PD patient 1 PD patient N Baseline 1st Year 2nd Year 4th Year Baseline 1st Year 2nd Year 4th Year Longitudinal GM volumes Longitudinal cortical thickness PD patients Visits Longitudinal dataset for cBF volume PD patients Vertices (364K) model_cbf<-lmer(cbf_volume ~follow_up_visit + sex + age + years_of_education + TIV + (follow_up_visit|PATNO_ID), data) get_individual_slopes(model_cbf) Visits for(vertex_i in vertices){ model_vertex_i<-lmer(vertex_i ~follow_up_visit + sex + age + years_of_education + (follow_up_visit|PATNO_ID), data) get_individual_slopes(model_vertex_i) } FDR_correction(global_vertex_analyses) cBF atlas from Fritz et al. HBM. 2019 Automated cBF volume extraction Vertex extraction in dataframes Individual slopes computation for cBF volume Individual slopes computation for cognitive scores Individual slopes computation for each vertex of cortical thickness INDIVIDUAL SLOPES FOR A COGNITIVE SCALE (e.g. MoCA) INDIVIDUAL SLOPES FOR cBF VOLUME . . . . . . PD patient 1 PD patient 2 PD patient N PD patient N PD patient 1 PD patient 2 Correlation analyses PD patient 1 PD patient 2 PD patient N . . . PD patient 3 INDIVIDUAL SLOPES FOR A COGNITIVE SCALE (e.g. MoCA) INDIVIDUAL SLOPES FOR cBF VOLUME INDIVIDUAL SLOPES FOR REGIONAL CORTICAL THICKNESS . . . . . . . . . PD patient 1 PD patient 2 PD patient N PD patient N PD patient 1 PD patient 2 PD patient 1 PD patient 2 PD patient N Vertex-wise regression Correlation analyses Vertex-wise regression Correlation analyses Correlation analyses WEIGHTED INDIVIDUAL SLOPES FOR CORTICAL THICKNESS SIMPLIFICATION FOR MEDIATION ANALYSES Weighted average of vertex-wise slopes ACME ADE STEP 1 STEP 2 STEP 3 STEP 4 110 3 The figure shows a complete view of the main analytical processes of the study, and a summary is presented that divides the analyses into four overview steps. STEP 1: Longitudinal MRIs of each subject were segmented and high-dimensionally registered to MNI standard space. GM volumes of the cBF were then automatically extracted using an existing stereotactic atlas of the cBF nuclei in MNI space. Regional cortical thickness measurements were obtained using the longitudinal processing pipeline for surface-based morphometry in the CAT12 toolbox. Longitudinal cognitive scores were extracted for each subject for global (MoCA) and domain-specific cognitive performances; STEP 2: Longitudinal changes in cBF volume, vertex-wise cortical thickness, and cognitive performance over follow-up were calculated within PD group using linear mixed-effects models controlling for sex, age, and years of education (and TIV for volumetric measures). Years of follow-up and all covariates were included as fixed effects predictors for all models. These covariates were included in the fixed effects as control variables for all the time. Participant ID and their effect over time were included as random effects. The individual slopes (rate of change) for each participant were estimated for the cognitive and imaging variables that showed a significant longitudinal change over follow-up (i.e., a significant time effect in the model). For vertex-wise individual slopes of cortical thickness, all vertices (approximately 364k) of cortical thickness for each longitudinal image were extracted into a data frame and the model for each vertex was computed separately. The individual slopes for each vertex and its cortical surface from each subject were used to construct the brain images of cortical individual slopes. These vertex-wise slopes results of cortical thinning were corrected for multiple comparisons at the vertex-level using the False Discovery Rate (FDR) and results were assessed at a peak-level threshold of P(FDR)<0.05; STEP 3: The association between longitudinal cBF degeneration and regional cortical thinning was assessed using regression analyses of the slope of change in cBF volume on the vertex-wise slopes of change in cortical thickness. Moreover, the association between cBF-related cortical thinning and cognitive decline was also assessed using regression analyses of vertex-wise slopes of change in cortical thickness on the slope of change in cognitive scores. All vertex-wise regression models were corrected for multiple comparisons at the vertex-level using FDR with a peak-level threshold of P(FDR)<0.05; STEP 4: The association between longitudinal cBF degeneration and cognitive decline was assessed using regression analyses of the respective slopes of change. Then, mediation analyses were used to further test whether the association between longitudinal cBF degeneration and cognitive decline in PD is mediated by regional cortical thinning. To simplify the analyses, mean cortical thinning of these cognition-related regions was calculated using the weighted average of vertex-wise slopes (using the effect size of the association with cognition as weights for each vertex, i.e. Pearson’s r values). Mediation models were calculated using the slopes of cBF volume change as the causal variable, the slopes of cognitive performance change as the outcome variable, and the weighted average slope of cognitionrelated cortical thickness change as the mediator variable. Path analyses of the average direct effect (ADE), the average causal mediation effect (ACME), and the mediated proportion were assessed. Abbreviations: GM = Grey matter; MNI = Montreal Neurological Institute; TIV = Total intracranial volume. 111 4 Supplementary data 2. Longitudinal change estimations of volumetric, cortical thickness and cognitive measures in PD patients. Longitudinal changes in ROI volumes, cortical thickness, and cognitive performance over follow-up were assessed through linear mixed-effects models controlled for sex, age, and education (and TIV in the case of volumetric measures) in PD patients. The individual slopes (rates of change) for each participant were estimated for those variables that showed a significant main effect in longitudinal change over follow-up. PD patients showed a significant longitudinal decline in cBF volume (t=-5.41, P<0.001), whole brain volume (t=-12.8, P<0.001) (supplementary table 1.1), as well as in vertexwise cortical thickness in widespread cortical areas, most notably posterior parieto-temporal areas (supplementary figure 1.1). Significant longitudinal cognitive decline was also observed in global cognitive performance (MoCA: t=-2.89, P=0.004), as well as in specific tests of memory (delayed recall: t=-2.41, P=0.017) and attention/working memory (SDM test: t=-4.35, P<0.001; LNS test: t=-2.35, P=0.02). Supplementary table 2.1. Longitudinal change of volumetric measures in PD patients. Volumetric measures BL 1st Visit 2nd Visit 4th Visit Beta [Conf. interval] T value (P value) N 162 120 113 62 cBF (mm3) 469 (33) 469 (37) 465 (34) 464 (33) !=-4.71 [-6.44, -2.97] t=-5.41 (<0.001) Whole brain volume (cm3) 785 (35) 784 (34) 779 (33) 777 (39) !=-5.81 [-6.70, -4.91] t=-12.8 (<0.001) The descriptive values presented are ROI volume (standard deviation). Mixed-linear model stats are beta [95% confidence interval] and T value (P value). Abbreviations: cBF = Cholinergic Basal Forebrain; BL = Baseline. 112 5 Supplementary figure 2.1. Longitudinal change in vertex-wise cortical thickness in PD patients. The figure shows regional effects of vertex-wise analyses of longitudinal cortical thickness change during the first four follow-up years in PD patients. Effects are thresholded at an FDR-corrected vertex-level statistical threshold of P<0.05, and are depicted on cortical surfaces showing medial, lateral, superior, and inferior brain renderings. Pronounced effects were observed in orbitofrontal areas, medial and lateral superior frontal areas, posterior and isthmus cingulate, precuneus, lateral temporo-parietal areas, and lateral occipital areas. 0 -1 -2 -3 -4 -5 -6 T-value Corrected P value < 0.05 LR 119 12 Supplementary data 5. Differences in longitudinal cortical thinning between PD patients with and without fast cBF volume decline during the four years of follow-up. PD patients were classified into cBF- and cBF+ based on whether or not they presented reduced cBF volume during the available four years of follow-up. Thus, for each time point we calculated the patients’ w-scores of cBF volume relative to the healthy control values at baseline, adjusting for age and head size1. Forty-four PD patients with longitudinal cBF volumes below the 68% confidence limit of the healthy control range were classified as cBF- and 118 PD patients were classified as cBF+. We then used vertex-wise analyses (2-sample t-test) of the previously calculated individual slopes of cortical thickness change to assess whether PD patients with longitudinally reduced cBF volume also had faster cortical atrophy compared to PD patients with normal cBF volume. The figure shows regional Cohen’s d values of vertex-wise analyses of longitudinal cortical thickness change between cBF- and cBF+ PD patients. Effects are shown on cortical surfaces with medial, lateral, superior, and inferior brain renderings. Corroborating our continuous association analyses, we found that cBF- patients also showed faster cortical atrophy than cBF+ patients, with pronounced effects in superior frontal regions, parieto-temporal areas, the posterior cingulate and occipital areas. However, these vertex-wise effects did not persist after applying a strict correction for multiple comparisons (FDR). 1 Ray NJ, Bradburn S, Murgatroyd C, et al. In vivo cholinergic basal forebrain atrophy predicts cognitive decline in de novo Parkinson’s disease. Brain. 2018;141(1):165-176. doi:10.1093/brain/awx310 0.00 Uncorrected Pvalue < 0.05 0.10 0.20 0.30 0.40 Cohen’s d cBF- > cBF+ LR 0.50 120 13 Supplementary data 6. Summary of regional effect sizes for the associations between cortical thinning in cBF-related regions and global cognitive decline. Supplementary figure 6.1. Effect sizes of MoCA-cortical thickness correlations within cortical regions defined in the Desikan-Killiany brain atlas. The figure shows a descriptive summary of the regional effects of the regression analyses of the slope of change in MoCA score on vertex-wise slopes of change in cortical thickness in PD patients. On the left side, the vertex-wise Pearson’s r values within the significant bilateral ROIs defined in the Desikan-Killiany atlas are depicted as boxplots, and on the right side the vertex-wise average and standard deviation of the Pearson’s r values within the respective bilateral ROIs are displayed in tabular format. The color scheme ranges from green representing a smaller effect size (lowest Pearson's r mean value of 0.18), yellow representing a medium effect size, to red representing a larger effect size (highest Pearson's r mean value of 0.25). Abbreviation: ROI = Region of interest. Bilateral ROI Mean correlation (SD) Pars opercularis 0.247 (0.052) Superior frontal 0.239 (0.062) Paracentral 0.239 (0.043) Cuneus 0.237 (0.042) Supramarginal 0.235 (0.053) Inferior temporal 0.233 ( 0.040) Superior parietal 0.230 ( 0.045) Transverse temporal 0.228 (0.038) Precuneus 0.226 (0.035) Precentral 0.224 (0.047) Pars triangularis 0.221 (0.013) Inferior parietal 0.220 ( 0.032) Temporal pole 0.220 ( 0.033) Fusiform 0.219 (0.039) Entorhinal 0.218 (0.021) Isthmus cingulate 0.218 (0.026) Pericalcarine 0.214 ( 0.030) Lateral orbito frontal 0.212 (0.022) Lateral occipital 0.212 (0.031) Superior temporal 0.205 (0.022) Insula 0.202 (0.029) Rostral middle frontal 0.201 (0.023) Posterior cingulate 0.200 (0.022) Middle temporal 0.198 (0.022) Postcentral 0.197 (0.018) Lingual 0.192 ( 0.020) Pars orbitalis 0.190 (0.016) Caudal anterior cingulate 0.181 (0.006) Boxplot of vertex correlations by ROIs Pearson’s r 121 14 Supplementary data 7. Robust regression analyses of the associations between cBF degeneration, cortical thinning, and cognitive decline. Robust regression analyses were performed to assess whether the observed associations remain valid after removing presence of outliers. To account for the presence of influential observations on our association analysis, we used robust regression models that employ penalty functions to minimize and remove the impact of extreme observations using iterated re-weighted least squares1,2. The penalty function used in our models was the Huber weighting method. In addition, we also estimated robust correlation coefficients using the Winsorized correlation method3. Using these robust models and correlations, we removed possible contaminated effects and assessed the results more accurately not only in the associations between changes in MoCA score and cortical thickness (figure 7.1a), but also between changes in cBF volume and MoCA score (figure 7.1b), as well as between changes in cBF volume and cortical thickness (figure 7.1c). All analyses were carried out in R using the MASS and WSR2 libraries. We present the summarized results in the comparison table 2. In summary, robust regression models consistently revealed similar associations among changes in cBF volume, cortical thickness, and MoCA score as obtained by our original method. Notably, the Winsorised correlation coefficient for the association between changes in cBF volume and cortical thickness was 0.45 (compared to the Pearson's correlation coefficient of 0.47), and between cBF volume and MoCA score was 0.30 (same effect size as obtained earlier for Pearson correlation). The association between changes in cortical thickness and MoCA score had a robust association with a significant Winsorised correlation coefficient of 0.25 (compared to the Pearson's correlation coefficient of 0.37). In conclusion, using these robust methods, we found comparable associations to the conventional correlations, while not detecting significant confounding effects that could have affected the reported results. 1 Beaumont C, Hoaglin DC, Mosteller F, Tukey JW. Exploring Data Tables, Trends and Shapes. Stat. 1986;35(3). doi:10.2307/2987762 2 Huber PJ, Ronchetti EM. Robust statistics john wiley & sons. New York. 1981;1(1). 3 Wilcox R. Introduction to Robust Estimation and Hypothesis Testing, Third Edition.; 2011. doi:10.1016/C2010-0-67044-1 122 15 Supplementary table 7.1. Results of robust associations and previously reported conventional associations. Robust associations Robust regression models Winsorized correlations between Beta [C.I.] T value P value Correlation coefficients P value cBF volume and Cortical thickness 0.22 [0.13,0.32] 7.19 <0.001 0.45 <0.001 cBF volume and MoCA scores 0.13 [0.05,0.21] 3.73 <0.001 0.30 <0.001 Cortical thickness and MoCA scores 0.17 [0.03,0.38] 3.66 0.001 0.25 0.001 Conventional associations Linear regression models Pearson’s correlations between Beta [C.I.] T value P value Correlation coefficients P value cBF volume and Cortical thickness 0.34 [0.24,0.45] 6.72 <0.001 0.47 <0.001 cBF volume and MoCA scores 0.16 [0.08,0.25] 3.92 <0.001 0.30 <0.001 Cortical thickness and MoCA scores 0.28 [0.18,0.40] 5.08 <0.001 0.37 <0.001 Abbreviations: cBF = cholinergic basal forebrain, MoCA = Montreal Cognitive Assessment, CI = Confident Interval. 123 16 Supplementary figure 7.1. Penalised cases in the robust regression models. The figure shows the adjusted robust regression models and identify any cases that may have been penalized, using residual plots, q-q plots, and weighting-residuals plots. The penalised cases have been visualized by red colour in the associations (figures a.1, b.1 and c.1), the distance of the influencer cases according to the residuals and fitted values in the regression model (figures a.2, b.2 and c.2), the positions of these cases in the fitted distribution (figures a.3, b.3 and c.3) and the weights that were used to penalise each case and graduate the contribution in the model (figures a.4, b.4 and c.4). −6 −5 −4 −3 −2 −1 0 −7−6−5−4−3−2 cBF individual_w_slope −3.5 −3.0 −2.5 −2.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 −2.5 −2.0 −1.5 −1.0 −0.5 0.0 m3.fitted.values m3.residuals −6 −5 −4 −3 −2 −1 0 1 −3−2−10123 qqnorm.ds.individual_w_slope..x qqnorm.ds.individual_w_slope..y 0.00 0.25 0.50 0.75 1.00 −4.0−3.5−3.0−2.5−2.0−1.5−1.0−0.5 0.0 0.5 1.0 rmodel3.residuals rmodel3.w Residual plot Q-Q plot Weights-residuals plot Contribution in the model Low High Residuals Fitted values Sample quantiles Theorical quantiles Weights Residuals a) Plots of robust association between changes in cBF volume and cortical thickness Longitudinal cBF volume change over time (mm3/year) Longitudinal cognition-related cortical thickness change over time (mm/year) Association between changes in cBF volume and cortical thickness a.1) a.2) a.3) a.4) Penalized Non-penalized −3 −2 −1 0 −7−6−5−4−3−2 cBF moca −3.0 −2.5 −2.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 −0.75 −0.50 −0.25 0.00 0.25 m3.fitted.values m3.residuals −3 −2 −1 0 1 −3−2−10123 qqnorm.ds.moca..x qqnorm.ds.moca..y 0.00 0.25 0.50 0.75 1.00 −3.0 −2.5 −2.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 rmodel3.residuals rmodel3.w Residual plot Q-Q plot Weights-residuals plot Contribution in the model Low High Residuals Fitted values Sample quantiles Theorical quantiles Weights Residuals b) Plots of robust association between changes in cBF volume and MoCA score Longitudinal cBF volume change over time (mm3/year) Longitudinal MoCA score change change over time (mm/year) Association between changes in cBF volume and MoCA score b.1) b.2) b.3) b.4) Penalized Non-penalized −6 −5 −4 −3 −2 −1 0 −3−2−10 moca individual_w_slope −2.5 −2.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 1.5 −1.5 −1.0 −0.5 0.0 0.5 m3.fitted.values m3.residuals −3 −2 −1 0 1 −3−2−10123 qqnorm.ds.moca..x qqnorm.ds.moca..y 0.00 0.25 0.50 0.75 1.00 −2.5 −2.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 rmodel3.residuals rmodel3.w Residual plot Q-Q plot Weights-residuals plot Contribution in the model Low High Residuals Fitted values Sample quantiles Theorical quantiles Weights Residuals c) Plots of robust association between changes in cortical thickness and MoCA score Longitudinal MoCA score change change over time (mm/year) Longitudinal cognition-related cortical thickness change over time (mm/year) Association between changes in cortical thickness and MoCA score c.1) c.2) c.3) c.4) Penalized Non-penalized 124 17 Supplementary data 8. Mediation analysis of associations between cBF degeneration, cortical thinning, and memory decline. Cortical thinning in areas related to cBF degeneration was significantly associated with decline in delayed recall performance, specifically in medial superior and lateral frontal areas, lateral temporo-parietal areas, precuneus, and posterior cingulate (average effect size: r=0.36, P<0.001; supplementary figure 5.1.a,b). Path analyses showed that the effect of cBF degeneration on decline in delayed recall performance (total effect: b=0.07 [0.02,0.11], P=0.002) was significantly mediated by mean cortical thinning in these areas (ADE: b=0.03 [-0.01,0.08], P=0.118; ACME: b=0.04 [0.01,0.06], P=0.006; mediated proportion: 52.6%, P=0.008) (supplementary figure 5.1.c-e). Supplementary figure 8.1. Cortical mediation of the association between longitudinal cBF degeneration and decline in delayed recall scores. (a) Brain renderings of cortical clusters showing significant correlations between vertex-wise slopes of cortical thickness change and the slope of change in delayed recall score. Vertex-wise regression analyses were restricted to cortical areas showing significant associations with cBF degeneration (as depicted in Figure 1) and corrected for multiple comparisons using an FDR-corrected vertex-level threshold of P<0.05. Effects are depicted on cortical surfaces showing medial, lateral, superior, and inferior brain renderings; (b) Average individual slopes of change in delayed recall-related cortical thickness XY Z = ZADE+ZACME ZACME=X·Y ZADE β=0.31 [0.21,0.40] P<0.001 β=0.14 [0.08,0.19] P<0.001 β=0.07 [0.02,0.11] P=0.002 cBF volume change Delayed recallrelated cortical thickness change Delayed recall score change −1.0 −0.8 −0.6 −0.4 −0.2 0.0 0.2 0.4 −7−6−5−4−3−2 cBF HVLTRDLY Z: r=0.27 (P<0.001) Longitudinal cBF volume change over time (mm3/year) Longitudinal change of delayed recall score over time (score/year) c) d) e) a) b) Pearson’s r −6 −5 −4 −3 −2 −1 0 −1.0−0.8−0.6−0.4−0.20.00.20.4 HVLTRDLY individual_w_slope_DLYRECALL Y: r=0.36 (P<0.001) Longitudinal delayed recall-related cortical thickness change over time (mm/year) Longitudinal change of delayed recall score over time (score/year) ZACME : β=0.04 [0.01,0.06] (P=0.006)ZADE : β=0.03 [-0.01,0.08] (P=0.118) −6 −5 −4 −3 −2 −1 0 −7−6−5−4−3−2 cBF individual_w_slope_DLYRECALL X: r=0.46 (P<0.001) Longitudinal cBF volume change over time (mm3/year) Longitudinal delayed recall-related cortical thickness change over time (mm/year) LR 125 18 (y-axis) are plotted against individual slopes of change in delayed recall score (x-axis). Red line represents linear trend; (c) Individual slopes of change in cBF volume (x-axis) are plotted against individual slopes of change in delayed recall score (yaxis). Blue line represents linear trend; (d) Individual slopes of change in cBF volume (x-axis) are plotted against average individual slopes of change in delayed recall-related cortical thickness (y-axis). Orange line represents linear trend; (e) Path diagram of causal mediation model where the effect of cBF degeneration on decline in delayed recall performance is plotted in path Z (blue arrow), the effect of cBF degeneration on cortical thinning is plotted in path X (orange arrow), and the effect of cortical thinning on decline in delayed recall performance is plotted in path Y (red arrow). Betas [95% confidence intervals] and P values of regression analyses are indicated for each association separately. Total effect of association Z is composed by sum of average direct effect (Z ADE ) and average causal mediation effect (Z ACME ) (purple arrows). 126 19 Supplementary data 9. Regional differences between amyloidb -positive and –negative PD patients in the association between longitudinal cBF degeneration and cortical thinning. Supplementary figure 9.1 Regional effect sizes of differences between amyloidb -negative and -positive PD patients. The figure shows vertex-wise differences in the association between longitudinal cBF degeneration and parallel cortical thinning in amyloid-b-positive compared to -negative PD patients as assessed by Fisher’s Z tests of vertex-wise correlation coefficients. Effects are depicted on cortical surfaces showing medial, lateral, superior, and inferior brain renderings. The amyloid-positive patients showed significantly higher correlations in the posterior cingulate, occipito-parietal, and pre- and post-central cortices. 0.00 0.05 2.50 P < 0.05 1.00 1.50 2.00 3.50 Z value 3.00 LR 127 20 Supplementary figure 9.2. Regional effect sizes of differences between amyloidb -negative and -positive PD patients summarized across cortical regions defined in the Desikan-Killiany brain atlas. The figure shows a descriptive summary of the regional differences in the strength of correlation between cBF degeneration and cortical thinning in amyloid-b-positive compared to -negative PD patients. On the left side, the vertex-wise Fisher’s Z values within the significant bilateral ROIs defined in the Desikan-Killiany atlas are depicted as boxplots, and on the right side the vertex-wise average and standard deviation of the Z values within the respective bilateral ROIs are displayed in tabular format. The color scheme ranges from green representing a smaller effect size (lowest mean Z value of 1.96), yellow representing a medium effect size, to red representing a larger effect size (highest mean Z value of 2.58). Abbreviation: ROI = Region of interest. Bilateral ROI Mean Z value (SD) Superior parietal 2.580 (0.381) Postcentral 2.489 (0.314) Isthmus cingulate 2.421 (0.307) Precentral 2.401 (0.327) Lingual 2.237 (0.177) Superior frontal 2.212 (0.145) Lateral occipital 2.129 (0.113) Precuneus 2.082 (0.098) Posterior cingulate 2.076 (0.077) Medial orbitofrontal 2.061 (0.085) Paracentral 2.024 (0.071) Supramarginal 1.985 (0.021) Superior temporal 1.979 (0.013) Rostral middle frontal 1.978 (0.015) Caudal anterior cingulate 1.960 (0.007) Boxplot of vertices' Z values by ROI Z value 128 Cholinergic basal forebrain atrophy associates with cortical hypometabolism in Parkinson’s disease-related cognitive impairment Este trabajo fue enviado a publicación con revisión por pares como: Labrador-Espinosa MA, Silva-Rodriguez J, Okkels N, Muñoz-Delgado L, Horsager J, Castro-Labrador S, Franco-Rosado P, Castellano-Guerrero AM, Iglesias-Camacho E, San-Eufrasio M, Perez-Calvo C, Macías-García D, Jesús S, Adarmes-Gómez A, Ojeda-Lepe E, Carrillo F, Martín-Rodríguez JF, Roldan- Lora F, García-Solís D, Borghammer P, Mir P, Grothe MJ. Cholinergic basal forebrain atrophy associates with cortical hypometabolism in Parkinson’s disease-related cognitive impairment. Submitted. Oct 2023. 135 7 Table 1. Demographic, clinical, and cognitive characteristics of the 95 patients with Parkinson’s disease. The descriptive values presented are: counts (percentage) for categorical variables; median (IQR) for Hoehn & Yahr stage; and mean (standard deviation) for all other continuous variables. Higher education refers to more than 12 years of formal education. Abbreviations: H&Y = Hoehn & Yahr; UPDRS-III-off = Unified Parkinson’s disease rating scale - Part III in “Off” state; PD-CRS = Parkinson’s disease - Cognitive rating scale; MoCA = Montreal cognitive assessment; PD-NC = Patients with normal cognition; PD-MCI = Patients with mild cognitive impairment; PDD = Patients with dementia. Posterior cBF atrophy associates with cortical hypometabolism in Parkinson’s disease In voxel-wise regression analyses adjusted for multiple comparisons, posterior cBF volume was significantly associated with extensive cortical hypometabolism, covering lateral and medial fronto-parietal regions, the lateral temporal lobe, inferior and medial parietal areas, and occipital regions (Figure 1). By contrast, the anterior cBF volume was only associated with hypometabolism in the anterior cBF itself and closely adjacent regions, and these effects did not survive correction for multiple comparisons (supplementary figure S2). Similarly, hippocampal volume was selectively associated with hypometabolism in the hippocampus itself and related medial temporal and medial frontal regions of the limbic circuitry (supplementary figure S3), but none of these effects were significant after adjusting for multiple comparisons. In additional analyses assessing the influence of local cortical atrophy, combining both posterior cBF volume and local cortical GM volume into a multimodal regression model for cortical hypometabolism slightly increased the fit compared to the model based on cBF volume alone (Global R2: 0.468 vs. 0.503, ΔR2=0.035 (7.5%), p = 0.01). However, although local cortical GM volume was also a significant predictor of cortical hypometabolism in this model (t=2.51, p=0.01, decomposed R2=0.16, 31% of total variance), posterior cBF volume showed a considerably stronger influence (t=4.42, p<0.001, decomposed R2=0.26, 52% of total variance). cBF-associated cortical hypometabolism mediates the association between cBF atrophy and cognitive impairment in Parkinson’s disease In covariate-controlled regression models, lower posterior cBF volume was significantly associated with worse PD-CRSTotal scores (rpartial=0.36, p=0.001). By design, lower posterior cBF volume was also significantly associated with average cortical hypometabolism in cBF-associated areas (rpartial=0.45, p<0.001) (figure 2a-b). In addition, cBF-associated cortical hypometabolism was also significantly associated with worse PD-CRSTotal scores (rpartial=0.40, p<0.001). Interestingly, path analyses showed that around half of the total effect of posterior cBF atrophy on cognitive impairment (total effect: b=0.63 [0.44,0.82], p<0.001) was mediated by cBF-associated cortical hypometabolism (ADE: b=0.30 [0.09,0.52], p=0.008; ACME: b=0.31 [0.14,0.51], p=0.002; mediated proportion: 50.1%, p=0.006) (figure 2c- d). Significant mediator effects of cBF-associated cortical hypometabolism were also observed for Descriptive statistics Gender, Female/Male (F%) 24/71 (25%) Age, years 65.1 (9.4) Higher education, Yes/No (Yes %) 35/60 (37%) Age at onset, years 57.8 (8.7) Disease duration, years 7.2 (4.3) H&Y, stages 2 (1 - 4) UPDRS-III-Off, scores 24.9 (12.6) MoCA, scores 21.1 (6) PD-CRS, total scores 76.9 (24.9) Fronto-subcortical subscore 50.3 (21.9) Posterior-cortical subscore 26.6 (4) Cognitive category, PD-CN/PD-MCI/PDD 41/27/27 136 8 associations between posterior cBF atrophy and cognitive deficits in the fronto-subcortical and the posterior-cortical subscores of the PD-CRS (supplementary data S1). In further sensitivity analyses evaluating the potential relation of the observed effects with disease severity, similar associations were observed using regression models that additionally controlled for years of disease progression and motor symptom severity. Moreover, all the observed associations remained significant when using regression models restricted to patients without dementia (supplementary data S2). Figure 1. Associations between posterior cBF atrophy and cortical hypometabolism in Parkinson’s disease. The figure shows regional effects of voxel-wise regression analyses of posterior cBF volume on cortical hypometabolism in patients with Parkinson’s disease, controlled for anterior cBF volume, sex, age, education, TIV, PET scanner, and cognitive category. Effects are depicted on cortical surfaces showing medial, lateral, superior, and inferior brain renderings; coronal slices from posterior to anterior; sagittal slices from left to right; and axial slices from inferior to superior. The black vertical bar in the colour scale of partial r values denotes the effect size corresponding to a statistical threshold of p(FDR)<0.05. 137 9 cBF-associated cortical hypometabolism corresponds to areas of pronounced cholinergic denervation in Parkinson’s disease Regional differences in [18F]FEOBV uptake between patients with Parkinson's disease and healthy controls in cohort 2 are illustrated in Figure 3-a. Differences were particularly notable in lateral and medial posterior parieto-temporo-occipital areas and in the lateral superior frontal lobe. Visual comparison revealed a regional pattern remarkably similar to the pattern of cBF-associated cortical hypometabolism in our main cohort (Figure 3-b). Formal topographic correspondence analysis revealed a strong spatial correlation between both patterns (Spearman’s r=0.57, p<0.001) (Figure 3- c). Discussion In the present multimodal MRI-PET imaging study, we investigated in vivo associations between cBF atrophy, cortical hypometabolism, and cognitive impairment in a comparably large sample of patients with Parkinson’s disease. We also examined the spatial correspondence of the observed pattern of cBF-associated cortical hypometabolism with areas of most pronounced cortical cholinergic denervation in Parkinson’s disease as measured by [18F]FEOBV PET. We found that posterior cBF atrophy was closely associated with extensive cortical hypometabolism in fronto-parietal, lateral temporal, and occipital regions, which partially mediated the association between cBF and cognitive impairment. Interestingly, these remote associations were highly specific for the cortically-projecting posterior cBF as no comparable associations were observed for the anterior cBF or the hippocampus. Moreover, cortical areas showing hypometabolism associated with posterior cBF atrophy overlapped spatially with those areas estimated to undergo most pronounced cholinergic denervation in Parkinson’s disease. Recent in vivo MRI studies have revealed robust associations between cBF degeneration and cognitive decline in patients with Parkinson’s disease5–9. In another line of research, cortical hypometabolism as measured by [18F]FDG PET has long been known to closely track with cognitive Figure 2. Mediation analysis of associations between cBF atrophy, cortical hypometabolism, and cognitive impairment. (a) Residualized posterior cBF volume (x-axis) plotted against PD-CRSTotal score (y-axis). Blue line represents linear trend; (b) Residualized posterior cBF volumes (x-axis) plotted against average cortical hypometabolism in cBF-related cortical areas (y-axis). Orange line represents linear trend; (c) Average cortical hypometabolism in cBF- related cortical areas (y-axis) plotted against PD-CRSTotal score (x-axis). Red line represents linear trend; Pearson’s r values obtained from the direct correlation between variables are displayed. (d) Path diagram of the causal mediation model where the effect of cBF atrophy on cognitive impairment is plotted in path Z (blue arrow), the effect of cBF atrophy on cortical hypometabolism is plotted in path X (orange arrow), and the effect of cortical hypometabolism on cognitive impairment is plotted in path Y (red arrow). Standardized betas [95% confidence intervals], T values and P values of the regression analyses are indicated separately for each association. The total effect of path Z is composed by the sum of the average direct effect (ZADE) and the average causal mediation effect (ZACME) (purple arrows). 138 10 impairment in Parkinson’s disease.32,11,12 Taken together with the strong influence of cholinergic innervation on cortical synaptic function and cognition3,33, these findings suggest that cBF atrophy and cortical hypometabolism may be interrelated phenomena in Parkinson’s disease that are pathophysiologically linked to cognitive impairment. Research in experimental animal models has demonstrated that selective cholinergic lesions in the basal forebrain can induce hypometabolism in denervated cortical areas that correlates with cognitive deficits.15,16 In humans, neuropathologic studies of patients with Parkinson’s or Alzheimer’s disease have described strong relationships between the loss of cholinergic cBF neurons and degenerative changes in cortical target areas.34,35 Moreover, more recent neuroimaging studies of patients with Alzheimer’s disease could demonstrate in vivo associations between cBF degeneration and hypometabolism in highly innervated cortical target areas of the cBF projections.25 Here, we demonstrate that a similar in vivo association between cBF atrophy and cortical hypometabolism is also present in patients with Parkinson’s disease and that these processes are interrelated in their contribution to cognitive impairment. Although not directly assessed in our study, an underlying assumption of our analyses is that the associations between cBF atrophy and cortical hypometabolism may represent a functional consequence of cortical cholinergic denervation. In support of this hypothesis, we observed that Figure 3. Topographic correspondence between patterns of cBF-associated cortical hypometabolism and cholinergic denervation in Parkinson’s disease. The figure shows the formal topographic correspondence between patterns of posterior cBF-associated cortical hypometabolism and cholinergic denervation in patients with Parkinson’s disease. Mean partial r values of the association between cBF atrophy and cortical hypometabolism (a) and T values reflecting the difference in [18F]FEOBV uptake between groups of healthy controls and patients with Parkinson’s disease (b) are depicted on cortical surfaces displaying medial and lateral brain renderings. Regional mean values were calculated across 100 ROIs covering the entire cortical gray matter.31 An accompanying scatter plot illustrates the correlation between both measures across all cortical ROIs, demonstrating more pronounced cBF-associated hypometabolism (partial r values, x-axis) in regions with more severe cholinergic denervation (T value of [18F]FEOBV uptake differences, y-axis) (c). 139 11 cortical hypometabolism was specifically associated with atrophy in the posterior cBF ROI, which largely corresponds to the nucleus basalis of Meynert representing the main source of cholinergic projections to the neocortex.23,33 By contrast, atrophy of the anterior cBF ROI, corresponding to the medial septum and diagonal band of Broca, did not show significant associations with cortical hypometabolism. Similarly, atrophy of the hippocampus, which was used as a cognition-relevant control region, was found to be associated with hypometabolism in the medial temporal lobe and related areas of the limbic circuit but did not show comparable associations with widespread neocortical hypometabolism. Interestingly, recent studies combining the same MRI-based cBF volume measurements used here with PET-based imaging of cholinergic neurotransmission in Parkinson’s disease demonstrated that degeneration of the cBF, and particularly its posterior subregion, is indeed closely linked to a depletion of cholinergic activity in parieto-temporo-occipital areas.36,37 Here, we add to this literature by demonstrating in a spatial correlation analysis that cBF-associated cortical hypometabolism is predominantly observed in those cortical areas that undergo most pronounced cholinergic denervation in Parkinson’s disease, as assessed by using novel [18F]FEOBV PET imaging.14 However, it is important to note that the [18F]FDG and [18F]FEOVB PET patterns used here were derived from two different cohorts, and combined multi-tracer [18F]FDG and [18F]FEOVB PET studies are needed to substantiate these data using individual-level analyses of the relation between cortical cholinergic denervation and hypometabolism. While our analysis shows that cBF atrophy is strongly associated with cortical hypometabolism, this does of course not imply that cortical hypometabolism in Parkinson’s disease is exclusively related to cholinergic degeneration. Particularly, cortical hypometabolism may also reflect local neurodegenerative processes in the cortex itself, and previous studies have suggested that cortical hypometabolism in Parkinson’s disease is closely followed by macrostructural cortical atrophy as measured by reduced GM volumes on MRI.12 However, in our additional analyses assessing the relative contributions of cBF degeneration and local cortical atrophy on cortical hypometabolism we observed that remote cBF atrophy had a much stronger influence on cortical hypometabolism compared to co-localized cortical GM atrophy. This finding aligns with a growing body of evidence suggesting that cortical dysfunction in Lewy body disorders may be influenced more by the loss of cholinergic innervation and other neurochemical disturbances than by local neurodegenerative processes in the cortex itself.14,38–41 In this line, a recent multi-tracer PET study found that cortical hypometabolism in Lewy body disorders by far exceeds the degree of cortical synaptic loss as estimated by PET imaging with the synaptic vesicle glycoprotein 2A (SV2A) radiotracer [11C]UCB-J, reinforcing that cortical hypometabolism is at least partly driven by remote processes rather than by local synaptic degeneration.42 This also likely explains the typical dissociation between marked cortical hypometabolism and comparably preserved cortical GM structure in Parkinson’s disease patients with cognitive impairment that is observed across neuroimaging studies,43 and may have critical implications for the development of clinically useful neuroimaging biomarkers for predicting cognitive decline in Parkinson’s disease.44 In our study, both cBF atrophy and cortical hypometabolism were found to be associated with cognitive impairment in patients with Parkinson’s disease, and further path analyses could demonstrate that the link between cBF atrophy and cognitive impairment was mediated by cortical hypometabolism. These findings support previous evidence linking both cortical hypometabolism and cholinergic degeneration with cognitive impairment in Parkinson’s disease.5,11,36 In additional analyses, we found similar associations of cBF atrophy and mediating effects of cortical 140 12 hypometabolism with both the fronto-subcortical and posterior-cortical PD-CRS summary measures. This contrasts with predictions of the dual syndrome hypothesis, which suggests that cholinergic deficits in Parkinson's disease may be more closely associated with deficits in posteriorcortical cognitive domains than with frontosubcortical cognitive domains, which are supposed to be more dependent on dopamine levels.1,45 However, consistent with our current findings, previous PET-based studies of cholinergic degeneration in non-demented Parkinson’s disease patients have also shown a link between cortical cholinergic loss and impaired cognitive function across multiple cognitive domains, including classic fronto-subcortical domains such as attention and executive function.36,46,47 A possible explanation for these discrepancies could be that the relative impact of cholinergic degeneration on cognitive functions may change during disease progression. Thus, more subtle cholinergic dysfunction may affect executive and attentional functions in early disease stages, whereas effects on more posterior cortical functions such as visuospatial abilities may only become apparent with more pronounced cholinergic degeneration as the disease progresses towards dementia. Interestingly, we observed similar associations between cBF atrophy, cortical hypometabolism, and cognitive impairment when limiting our sample to patients without dementia, indicating that these processes are already relevant for early cognitive disturbances in Parkinson’s disease. Together with other recent clinical research on cholinergic system degeneration in non-demented Parkinson’s disease patients,5,7,10,36,37 these findings may warrant a re-evaluation of the role of cholinergic treatment in predementia stages of Parkinson’s disease.2 Treatment with acetylcholinesterase inhibitors is well known to improve cognitive symptoms in patients with PDD and related Lewy body dementias, and this improvement has also been found to be linked to restoration of cortical metabolism.48–50 However, clinical trials assessing the efficacy of cholinergic treatments for cognitive impairments in the predementia stage of Parkinson’s disease have presented inconsistent results,51 which could be related to the high variability in the degree of cholinergic degeneration in this population.52 Consequently, neuroimaging markers of cholinergic system degeneration and associated cortical hypometabolism may help to select participants for clinical trials in Parkinson’s disease, focusing on those patients at greatest risk for cognitive decline and most likely to benefit from cholinergic therapy. Our study also presents some limitations that should be considered. First, the lack of multimodal MRI-PET neuroimaging data of a healthy control group restricts the specificity of the observed effects to patients in the Parkinson’s disease continuum. However, previous research had already demonstrated that variation in cBF volume due to the normal aging process is not associated with comparable effects on cortical hypometabolism or cognitive function.25 Second, assumptions about the chain of events between cBF atrophy, cortical hypometabolism, and cognitive impairment are based on cross-sectional correlations from which causality cannot be inferred. Third, cortical cholinergic signaling has not been directly assessed in this study and its role in the observed associations remains to be investigated in more detail. cBF volume on MRI is only an indirect structural marker of cholinergic system degeneration, although accumulating evidence from multimodal biomarker studies indicates that it is indeed closely related to cholinergic system function as assessed by neurophysiologic53 or molecular imaging markers of cholinergic system integrity.36,37 Finally, cholinergic system degeneration in Parkinson’s disease does not occur in isolation but together with degenerative changes in the dopaminergic and other ascending neurotransmitter systems, which may also affect cortical hypometabolism but were not specifically assessed in our study.41-42,54–56 Multi-tracer PET studies are needed to better understand the unique and interactive contributions 141 13 of neurotransmitter disturbances to cortical dysfunction and cognitive impairment in Parkinson’s disease. In conclusion, in this multimodal imaging study we provide evidence for a close link between in vivo cBF degeneration and cortical hypometabolism in Parkinson’s disease-related cognitive decline. The detected associations may have important implications for the design of clinically useful imaging biomarkers for disease prognosis and clinical trial stratification. Acknowledgments The authors would like to thank the patients who participated in this study for their contribution to science and the Instituto de Biomedicina de Sevilla and the Hospital Universitario Virgen del Rocío for the provided research resources. Funding This work was supported by the Spanish Ministry of Science and Innovation (RTC2019-007150-1, PID2021-127034OA-I00), the Instituto de Salud Carlos III-Fondo Europeo de Desarrollo Regional (ISCIII-FEDER) (PI16/01575, PI18/01898, PI19/01576, PI20/00613, PI21/01875, PI22/01704), the Consejería de Economía, Innovación, Ciencia y Empleo de la Junta de Andalucía (CVI-02526, CTS-7685), the Consejería de Salud y Bienestar Social de la Junta de Andalucía (PI-0471-2013, PE-0210-2018, PI- 0459-2018, PE-0186-2019), the Consejería de Transformación Económica, Industria, Conocimiento y Universidades de la Junta de Andalucía (PY20_00896, P20_00903), and the Fundación Alicia Koplowitz. Several authors of this publication are members of the European Reference Network for Rare Neurological Diseases (project ID 739510). M.A.L.E. is supported by a PhD scholarship (VI-PPIT-US) from the University of Seville (USE-19094-G). J.S.R. is supported by the “Sara Borrell” program (CD21/00067) of the ISCIII-FEDER. M.J.G. is supported by the “Miguel Servet” program (CP19/00031) of the ISCIII-FEDER. D.M.G. is supported by the “Juan Rodés” program (JR22/00073) of the ISCIII-FEDER. L.M.D. is supported by the “Río Hortega” program (CM21/00051) of the ISCIII-FEDER. Competing interests The authors report no competing interests. References 1. Aarsland D, Batzu L, Halliday GM, et al. Parkinson disease-associated cognitive impairment. Nature Reviews Disease Primers. 2021;7(1). doi:10.1038/s41572-021-00280-3 2. Bohnen NI, Yarnall AJ, Weil RS, et al. Cholinergic system changes in Parkinson’s disease: emerging therapeutic approaches. Lancet Neurol. 2022;21(4):381-392. 3. Ballinger EC, Ananth M, Talmage DA, Role LW. Basal forebrain cholinergic circuits and signaling in cognition and cognitive decline. Neuron. 2016;91(6):1199-1218. 4. Ovsepian SV, O’Leary VB, Zaborszky L. Cholinergic Mechanisms in the Cerebral Cortex: Beyond Synaptic Transmission. Neuroscientist. 2015;22(3):238-251. 5. Grothe MJ, Labrador-Espinosa MA, Jesús S, et al. In vivo cholinergic basal forebrain degeneration and cognition in Parkinson’s disease: Imaging results from the COPPADIS study. Parkinsonism and Related Disorders. 2021;88:68-75. doi:10.1016/j.parkreldis.2021.05.027 6. Barrett MJ, Sperling SA, Blair JC, et al. Lower volume, more impairment: reduced cholinergic basal forebrain grey matter density is associated with impaired cognition in Parkinson disease. J Neurol Neurosurg Psychiatry. 2019;90(11):1251-1256. 7. Ray NJ, Bradburn S, Christopher UM, et al. In vivo cholinergic basal forebrain atrophy predicts cognitive decline in de novo Parkinson’s disease. Brain. 2018;141(1):165- 176. 8. Schulz J, Pagano G, Bonfante JAF, Wilson H, Politis M. Nucleus basalis of Meynert degeneration precedes and predicts cognitive impairment in Parkinson’s disease. Brain. 2018;141(5):1501-1516. 142 14 9. Pereira JB, Hall S, Jalakas M, et al. Longitudinal degeneration of the basal forebrain predicts subsequent dementia in Parkinson’s disease. Neurobiol Dis. 2020;139(104831):104831. 10. Labrador-Espinosa MA, Silva-Rodríguez J, Reina-Castillo MI, Mir P, Grothe MJ. Basal Forebrain Atrophy, Cortical Thinning, and Amyloid-β Status in Parkinson’s disease- Related Cognitive Decline. Movement Disorders. Published online 2023. doi:10.1002/mds.29564 11. Garcia-Garcia D, Clavero P, Carmen IGS, et al. Posterior parietooccipital hypometabolism may differentiate mild cognitive impairment from dementia in Parkinson’s disease. Eur J Nucl Med Mol Imaging. 2012;39(11):1767- 1777. 12. González-Redondo R, David PGG, Clavero, et al. Grey matter hypometabolism and atrophy in Parkinson’s disease with cognitive impairment: a two-step process. Brain. 2014;137(Pt 8):2356-2367. 13. Rodriguez-Oroz MC, Gago B, Clavero P, Delgado-Alvarado M, Garcia-Garcia D, Jimenez-Urbieta H. The Relationship Between Atrophy and Hypometabolism: Is It Regionally Dependent in Dementias? Curr Neurol Neurosci Rep. 2015;15(7):44. doi:10.1007/s11910-015-0562-0 14. Nestor PJ. The Lewy body, the hallucination, the atrophy and the physiology. Brain. 2007;130(10). doi:10.1093/brain/awm132 15. Browne SE, Lin L, Mattsson A, Georgievska B, Isacson O. Selective antibody-induced cholinergic cell and synapse loss produce sustained hippocampal and cortical hypometabolism with correlated cognitive deficits. Exp Neurol. 2001;170(1):36-47. 16. Gelfo F, Petrosini L, Alessandro PG, et al. Cortical metabolic deficits in a rat model of cholinergic basal forebrain degeneration. Neurochem Res. 2013;38(10):2114-2123. 17. Postuma RB, Berg D, Stern M, et al. MDS clinical diagnostic criteria for Parkinson’s disease. Mov Disord. 2015;30(12):1591-1601. 18. Okkels N, Horsager J, Labrador-Espinosa M, et al. Severe cholinergic terminal loss in newly diagnosed dementia with Lewy bodies. Brain. Published online June 2023. doi:10.1093/brain/awad192 19. Skorvanek M, Goldman JG, Jahanshahi M, et al. Global scales for cognitive screening in Parkinson’s disease: Critique and recommendations. Movement Disorders. 2018;33(2):208-218. doi:10.1002/mds.27233 20. Mizutani R, Saiga R, Takekoshi S, et al. A method for estimating spatial resolution of real image in the Fourier domain. J Microsc. 2015;261(1):57-66. doi:10.1111/jmi.12315 21. Felix Carbonell, Alex Zijdenbos, Barry Bedell. A Data-driven Approach for Estimating the Spatial Resolution of Brain PET images (4301). Neurology. 2020;94(15 Supplement):4301. 22. Ashburner J. A fast diffeomorphic image registration algorithm. Neuroimage. 2007;38(1):95-113. 23. Fritz HCJ, Ray N, Dyrba M, Sorg C, Teipel S, Grothe MJ. The corticotopic organization of the human basal forebrain as revealed by regionally selective functional connectivity profiles. Published online 2018. doi:10.1002/hbm.24417 24. Desikan RS, Ségonne F, Fischl B, et al. An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. NeuroImage. 2006;31(3):968-980. doi:10.1016/J.NEUROIMAGE.2006.01.021 25. Grothe MJ, Heinsen H, Amaro E, Grinberg LT, Teipel SJ. Cognitive Correlates of Basal Forebrain Atrophy and Associated Cortical Hypometabolism in Mild Cognitive Impairment. Cerebral Cortex. 2016;26(6):2411-2426. doi:10.1093/cercor/bhv062 26. Gonzalez-Escamilla G, Lange C, Teipel S, Buchert R, Grothe MJ. PETPVE12: an SPM toolbox for Partial Volume Effects correction in brain PET – Application to amyloid imaging with AV45-PET. Neuroimage. 2017;147:669- 677. 27. Okkels N, Horsager J, Labrador-Espinosa MA, et al. Distribution of cholinergic nerve terminals in the aged human brain measured 143 15 with [ 18 F]FEOBV PET and its correlation with histological data. NeuroImage. 2023;269:119908. doi:10.1016/j.neuroimage.2023.119908 28. Carlesimo GA, Piras F, Assogna F, Pontieri FE, Caltagirone C, Spalletta G. Hippocampal abnormalities and memory deficits in Parkinson disease: A multimodal imaging study. Neurology. 2012;78(24):1939-1945. doi:10.1212/WNL.0b013e318259e1c5 29. Grömping U. Relative Importance for Linear Regression in R : The Package relaimpo. J Stat Soft. 2006;17(1). doi:10.18637/jss.v017.i01 30. Imai K, Keele L, Yamamoto T. Identification, inference and sensitivity analysis for causal mediation effects. Stat Sci. 2010;25(1):51-71. 31. Schaefer A, Kong R, Gordon EM, et al. Local- Global Parcellation of the Human Cerebral Cortex from Intrinsic Functional Connectivity MRI. Cerebral Cortex. 2018;28(9):3095-3114. doi:10.1093/cercor/bhx179 32. Bohnen NI, Koeppe RA, Minoshima S, et al. Cerebral Glucose Metabolic Features of Parkinson Disease and Incident Dementia: Longitudinal Study. J Nucl Med. 2011;52(6):848-855. doi:10.2967/jnumed.111.089946 33. Mesulam MM. The cholinergic innervation of the human cerebral cortex. In: Vol 145. ; 2004:67-78. doi:10.1016/S0079- 6123(03)45004-8 34. Jellinger K. Quantitative changes in some subcortical nuclei in aging, Alzheimer’s disease and Parkinson’s disease. Neurobiology of Aging. 1987;8(6):556-561. doi:10.1016/0197-4580(87)90134-5 35. Liu AKL, Chang RCC, Pearce RKB, Gentleman SM. Nucleus basalis of Meynert revisited: anatomy, history and differential involvement in Alzheimer’s and Parkinson’s disease. Acta Neuropathol. 2015;129(4):527- 540. doi:10.1007/s00401-015-1392-5 36. Schumacher J, Kanel P, Dyrba M, Storch A, Teipel S, Grothe MJ. Structural and molecular cholinergic imaging markers of. Brain. Published online 2023. doi:10.1093/brain/awad226/7219381 37. Ray NJ, Kanel P, Bohnen NI. Atrophy of the Cholinergic Basal Forebrain can Detect Presynaptic Cholinergic Loss in Parkinson’s Disease. Annals of Neurology. 2023;93(5):991- 998. doi:10.1002/ana.26596 38. Khundakar AA, Hanson PS, Erskine D, et al. Analysis of primary visual cortex in dementia with Lewy bodies indicates GABAergic involvement associated with recurrent complex visual hallucinations. acta neuropathol commun. 2016;4(1):66. doi:10.1186/s40478- 016-0334-3 39. Patterson L, Firbank MJ, Colloby SJ, Attems J, Thomas AJ, Morris CM. Neuropathological Changes in Dementia With Lewy Bodies and the Cingulate Island Sign. Journal of Neuropathology & Experimental Neurology. 2019;78(8):717-724. doi:10.1093/jnen/nlz047 40. Higuchi M, Tashiro M, Arai H, et al. Glucose Hypometabolism and Neuropathological Correlates in Brains of Dementia with Lewy Bodies. Experimental Neurology. 2000;162(2):247-256. doi:https://doi.org/10.1006/exnr.2000.7342 41. Borghammer P. Perfusion and metabolism imaging studies in Parkinson’s disease. Dan Med J. 2012;59(6):B4466. 42. Andersen KB, Hansen AK, Schacht AC, et al. Synaptic Density and Glucose Consumption in Patients with LEWY BODY DISEASES : An [ 11 C ] UCB‐J and [ 18 F ] FDG PET Study. Movement Disorders. 2023;38(5):796-805. doi:10.1002/mds.29375 43. Albrecht F, Ballarini T, Neumann J, Schroeter ML. FDG-PET hypometabolism is more sensitive than MRI atrophy in Parkinson’s disease: A whole-brain multimodal imaging meta-analysis. NeuroImage: Clinical. 2019;21:101594. doi:10.1016/j.nicl.2018.11.004 44. Lanskey JH, McColgan P, Schrag AE, et al. Can neuroimaging predict dementia in Parkinson’s disease? Brain. Published online 2018. 45. Svenningsson P, Westman E, Ballard C, Aarsland D. Cognitive impairment in patients with Parkinson’s disease: diagnosis, biomarkers, and treatment. The Lancet 144 16 Neurology. 2012;11(8):697-707. doi:10.1016/S1474-4422(12)70152-7 46. Bohnen NI, Muller MLTM, Kotagal V, et al. Olfactory dysfunction, central cholinergic integrity and cognitive impairment in Parkinson’s disease. Brain. 2010;133(6):1747- 1754. doi:10.1093/brain/awq079 47. Van Der Zee S, Müller MLTM, Kanel P, Van Laar T, Bohnen NI. Cholinergic Denervation Patterns Across Cognitive Domains in Parkinson’s Disease. Movement Disorders. 2021;36(3):642-650. doi:10.1002/mds.28360 48. Mori T, Ikeda M, Fukuhara R, Nestor PJ, Tanabe H. Correlation of visual hallucinations with occipital rCBF changes by donepezil in DLB. Neurology. 2006;66(6):935-937. doi:10.1212/01.wnl.0000203114.03976.b0 49. Suantio AM, Huang HL, Kwok CSN, Teo DCH, Nguyen MH. FDG-PET in suspected dementia with Lewy bodies: a case report. BMC Geriatr. 2019;19(1):150. doi:10.1186/s12877-019-1166-3 50. Satoh M, Ishikawa H, Meguro K, Kasuya M, Ishii H, Yamaguchi S. Improved Visual Hallucination by Donepezil and Occipital Glucose Metabolism in Dementia with Lewy Bodies: The Osaki-Tajiri Project. Eur Neurol. 2010;64(6):337-344. doi:10.1159/000322121 51. Baik K, Kim SM, Jung JH, et al. Donepezil for mild cognitive impairment in Parkinson’s disease. Sci Rep. 2021;11(1):4734. doi:10.1038/s41598-021-84243-4 52. Bohnen NI, Müller MLTM, Kotagal V, et al. Heterogeneity of Cholinergic Denervation in Parkinson’s Disease without Dementia. J Cereb Blood Flow Metab. 2012;32(8):1609- 1617. doi:10.1038/jcbfm.2012.60 53. Peter J, Mayer I, Kammer T, et al. The relationship between cholinergic system brain structure and function in healthy adults and patients with mild cognitive impairment. Scientific Reports. 2021;11(1):1-7. doi:10.1038/s41598-021-95573-8 54. Frey KA, Bohnen NILJ. Molecular Imaging of Neurodegenerative Parkinsonism. PET Clinics. 2021;16(2):261-272. doi:10.1016/j.cpet.2020.12.002 55. Apostolova I, Lange C, Frings L, Klutmann S, Meyer PT, Buchert R. Nigrostriatal Degeneration in the Cognitive Part of the Striatum in Parkinson Disease Is Associated with Frontomedial Hypometabolism. Clinical Nuclear Medicine. 2020;45(2):95-99. doi:10.1097/RLU.0000000000002869 56. Orso B, Arnaldi D, Girtler N, et al. Dopaminergic and Serotonergic Degeneration and Cortical [ 18 F ]Fluorodeoxyglucose Positron Emission Tomography in De Novo Parkinson’s Disease. Movement Disorders. 2021;36(10):2293-2302. doi:10.1002/mds.28654