Prepandemic Alzheimer Disease Biomarkers and Anxious-Depressive Symptoms During the COVID-19 Confinement in Cognitively Unimpaired Adults

RESEARCH ARTICLE OPEN ACCESS Prepandemic Alzheimer Disease Biomarkers and Anxious-Depressive Symptoms During the COVID-19 Confinement in Cognitively Unimpaired Adults Muge Akinci, MSc, Cleof´ e Peña-G´ omez, PhD, Gregory Operto, PhD, Sherezade Fuentes-Julian, PhD, Carme Deulofeu, PhD, Gonzalo S´ anchez-Benavides, PhD, Marta Mil` a-Alom` a, PhD, Oriol Grau-Rivera, PhD, Nina Gramunt, PhD, Arcadi Navarro, PhD, Carolina Minguill´ on, PhD, Karine Fauria, PhD, Ivonne Suridjan, PhD, Gwendlyn Kollmorgen, PhD, Anna Bayfield, PhD, Kaj Blennow, PhD, Henrik Zetterberg, PhD, Jos´ e Luis Molinuevo, PhD, Marc Su´ arez-Calvet, PhD, Juan Domingo Gispert, PhD, and Eider M. Arenaza-Urquijo, PhD, for the ALFA Study Neurology®2022;99:e1486-e1498. doi:10.1212/WNL.0000000000200948 Correspondence Dr. Arenaza-Urquijo [email protected] Abstract Background and Objectives Increased anxious-depressive symptomatology is observed in the preclinical stage of Alzheimer disease (AD), which may accelerate disease progression. We investigated whether β-amyloid, cortical thickness in medial temporal lobe structures, neuroinflammation, and sociodemographic factors were associated with greater anxious-depressive symptoms during the COVID-19 confinement. Methods This retrospective observational study included cognitively unimpaired older adults from the Alzheimer’s and Families cohort, the majority with a family history of sporadic AD. Participants performed the Hospital Anxiety and Depression Scale (HADS) during the COVID-19 confinement. A subset had available retrospective (on average: 2.4 years before) HADS assessment, amyloid [ 18 F] flutemetamol PET and structural MRI scans, and CSF markers of neuroinflammation (interleukin-6 [IL-6], triggering receptor expressed on myeloid cells 2, and glial fibrillary acidic protein levels). We performed multivariable linear regression models to investigate the associations of prepandemic AD-related biomarkers and sociodemographic factors with HADS scores during the confinement. We further performed an analysis of covariance to adjust by participants’prepandemic anxiety-depression levels. Finally, we explored the role of stress and lifestyle changes (sleep patterns, eating, drinking, smoking habits, and medication use) on the tested associations and performed sex-stratified analyses. Results We included 921 (254 with AD biomarkers) participants. β-amyloid positivity (B = 3.73; 95% CI = 1.1 to 6.36; p= 0.006), caregiving (B = 1.37; 95% CI 0.24–2.5; p= 0.018), sex (women: B = 1.95; 95% CI 1.1–2.79; p< 0.001), younger age (B = −0.12; 95% CI −0.18 to −0.052; p< 0.001), and lower education (B = −0.16; 95% CI −0.28 to −0.042; p= 0.008) were associated with greater anxious-depressive symptoms during the confinement. Considering MORE ONLINE CME Course NPub.org/cmelist COVID-19 Resources For the latest articles, invited commentaries, and blogs from physicians around the world NPub.org/COVID19 From the BarcelonaBeta Brain Research Center (BBRC) (M.A., C.P.-G., G.O., S.F.-J., C.D., G.S.-B., M.M.-A., O.G.-R., A.N., C.M., K.F., J.L.M., M.S.-C., J.D.G., E.M.A.-U.), Pasqual Maragall Foundation, Barcelona; Universitat Pompeu Fabra (M.A., M.M.-A., A.N., J.D.G.), Barcelona; IMIM (Hospital del Mar Medical Research Institute) (G.S.-B., M.M.-A., O.G.-R., C.M., M.S.-C., J.D.G., E.M.A.-U.), Barcelona; Centro de Investigaci´ on Biom´ edica en Red de Fragilidad y Envejecimiento Saludable (CIBERFES) (G.S.-B., O.G.-R., C.M., K.F., M.S.-C., E.M.A.-U.), Madrid; Servei de Neurologia (O.G.-R., M.S.-C.), Hospital del Mar, Barcelona; Fundaci´ o Pasqual Maragall (N.G.), Barcelona; Centre for Genomic Regulation (CRG) (A.N.), Barcelona Institute of Science and Technology (BIST); Institute of Evolutionary Biology (UPF-CSIC) (A.N.), Department of Experimental and Health Sciences, Universitat Pompeu Fabra, Barcelona; Instituci´ o Catalana de Recerca i Estudis Avançats (ICREA) (A.N.), Barcelona, Spain; Roche Diagnostics International Ltd (I.S.), Rotkreuz, Switzerland; Roche Diagnostics GmbH (G.K., A.B.), Penzberg, Germany; Department of Psychiatry and Neurochemistry (K.B., H.Z.), Institute of Neuroscience and Physiology, University of Gothenburg, M¨ olndal; Clinical Neurochemistry Laboratory (K.B., H.Z.), Sahlgrenska University Hospital, M¨ olndal, Sweden; UK Dementia Research Institute at UCL (H.Z.), London; Department of Neurodegenerative Disease (H.Z.), UCL Institute of Neurology, Queen Square, London, United Kingdom; Hong Kong Center for Neurodegenerative Diseases (H.Z.), China; H. Lundbeck A/S (J.L.M.), Denmark; and Centro de Investigaci´ on Biom´ edica en Red Bioingenier ´ ıa (J.D.G.), Biomateriales y Nanomedicina, Madrid, Spain. Go to Neurology.org/N for full disclosures. Funding information and disclosures deemed relevant by the authors, if any, are provided at the end of the article. ALFA coinvestigators are listed at links.lww.com/WNL/C377. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal. e1486 Copyright © 2022 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. prepandemic anxiety-depression levels, we further observed an association between lower levels of CSF IL-6 (B = −5.11; 95% CI −10.1 to −0.13; p= 0.044) and greater HADS scores. The results were independent of stress-related variables and lifestyle changes. Stratified analysis revealed that the associations were mainly driven by women. Discussion Our results link AD-related pathophysiology and neuroinflammation with greater anxious-depressive symptomatology during the COVID-19-related confinement, notably in women. AD pathophysiology may increase neuropsychiatric symptomatology in response to stressors. This association may imply a worse clinical prognosis in people at risk for AD after the pandemic and thus deserves to be considered by clinicians. Trial Registration Information ClinicalTrials.gov Identifier NCT02485730. There has been a global increase in anxious-depressive symptomatology with the COVID-19 pandemic and home confinement. 1,2 This will bring long-term implications for mental health and cognitive decline in vulnerable populations. 3-5 In this context, both anxiety 3 and depression 4-6 are associated with an increased risk for developing cognitive impairment 7-9 and Alzheimer disease (AD). The prevalence of AD is higher in women, 10 and both women and caregivers reported higher anxiety and depression, 11,12 particularly during the COVID-19 pandemic. 13 Recent studies suggested an early link between β-amyloid (Aβ) and worsening anxious-depressive symptoms in cognitively unimpaired (CU) adults. 14,15 Moreover, AD pathology may alter brain structures that regulate the brain’s response to stress and increase the proneness to develop anxious-depressive symptoms. 16 Another mechanism linking AD with anxietydepression might be neuroinflammation, which has an early involvement in the pathogenesis of the disease. 17 Notably, the CSF interleukin-6 (IL-6) has been consistently reported to be elevated in both patients with depression and AD. 16,18,19 Altogether, it becomes relevant to investigate the COVID-19 confinement-related anxious-depressive symptomatology in adults at risk for cognitive decline and AD, addressing sex/ gender differences and caregivers’mental health. Therefore, here we focused on CU older adults, the majority with a family history (FH) of clinically diagnosed sporadic AD. Older adults with FH of sporadic AD are at a higher risk for cognitive impairment and dementia 20 and start showing ADrelated pathologic changes early during midlife. 21,22 We investigated the associations of Aβburden, neuroinflammation, and brain structure data acquired approximately 2.4 years before the pandemic with anxious-depressive symptomatology during the COVID-19 confinement. We hypothesized that (1) adults with Aβburden, higher CSF IL-6 values, and/or lower structural integrity in AD-related regions (medial temporal lobe structures) will show greater anxiety-depression during the confinement and (2) these associations will be independent of the preconfinement anxiety-depression levels. We also hypothesized that women and caregivers will present higher anxious-depressive symptoms during the confinement. Methods Participants Participants were recruited from the ALzheimer’s and FAmilies (ALFA) and ALFA+ cohorts established at the Barcelonaβeta Brain Research Center in Barcelona, Spain, as a research platform to characterize preclinical AD. 20 The ALFA cohort includes 2743 CU (Clinical Dementia Rating [CDR] score = 0) older adults aged between 45 and 74 years, enriched for FH of AD (86% had at least 1 parent diagnosed with dementia) and APOE e4 genotypes. At the baseline visit (2013–2014), sociodemographic, clinical, epidemiologic, genetic, and cognitive data were collected. Participants from the ALFA cohort were invited to participate in the ALFA+ study following a genetic risk enrichment strategy (APOE e4 carriership and FH of sporadic AD). Four hundred fifty participants from the ALFA cohort were enrolled in the nested ALFA+ study. These participants underwent advanced MRI and PET, lumbar puncture, clinical interviews, cognitive testing, lifestyle, and risk factors evaluations. The inclusion criteria of ALFA+ participants were as follows: (1) participation in the ALFA study; (2) aged 45–75 years at inclusion in Glossary AD = Alzheimer disease; ALFA = ALzheimer and FAmilies; Aβ=β-amyloid; BRS = Brief Resilience Scale; CDR = Clinical Dementia Rating; CL = Centiloid; CU = cognitively unimpaired; FH = family history; GFAP = glial fibrillary acidic protein; HADS = Hospital Anxiety and Depression Scale; IL-6 = interleukin-6; MNI = Montreal Neurological Institute; PSS = Perceived Stress Scale; sTREM2 = soluble fragment of triggering receptor expressed on myeloid cells 2; SUVR = standardized uptake value ratio. Neurology.org/N Neurology | Volume 99, Number 14 | October 4, 2022 e1487 the ALFA study; and (3) long-term commitment to follow-up visits, assessments, and study procedures. ALFA+ exclusion criteria were as follows: (1) cognitive impairment (CDR score >0, Mini-Mental State Examination <27, and semantic fluency <12); (2) any unstable medical condition or significant systemic illness that could interfere with protocol compliance; (3) any contraindication to the tests or study procedures; and (4) FH of monogenic AD. 20 In the current study, sociodemographic, genetic, clinical, and neuroimaging data collected between 2016 and 2019 from the ALFA and/or ALFA+ participants were used and are referred to as preconfinement measurements. On May 8, 2020, during the de-escalation phases of the COVID-19 confinement, the invitation to participate in the current study was sent via email to 2,582 ALFA participants. On March 14, 2020, during the first wave of the COVID-19 pandemic in Spain, the Spanish government declared a state of emergency and started a national lockdown to control the increasing number of COVID19 cases in the country. From March 15, all residents were confined to their homes except to make necessary purchases, work, and emergencies. 23 On May 2, the government started to implement de-escalation phases to ease the confinement restrictions. BetweenMay 8 and August 31, the period referred to as confinement hereafter, 967 ALFA participants agreed to take part in the current study and completed an online assessment battery that included the Hospital Anxiety and Depression Scale (HADS), the Perceived Stress Scale (PSS), the Brief Resilience Scale (BRS), and an ad hoc evaluation on caregiving and changes in lifestyle patterns (sleep patterns, eating, drinking, smoking habits, and medication use). Of these, 265 were from the ALFA+ study and referred to as biomarker sample in this article. The average time from preconfinement to confinement assessments was 2.4 (±0.8) years. Standard Protocol Approvals, Registrations, and Patient Consents The ALFA and ALFA+ study protocols have been approved by the Independent Ethics Committee “Parc de Salut Mar,”Barcelona, and registered at ClinicalTrials.gov (ALFA Identifier: NCT01835717; ALFA+ Identifier: NCT02485730). The COVID-19 protocol (CovidImpact_BBRC2020) has been approved by the Independent Ethics Committee “Parc de la Salud”on March 16, 2020 (Identifier: 2020/9255). All participants signed informed consent that had also been approved by the Independent Ethics Committee “Parc de la Salud”. Clinical Measurements Anxiety and Depression Anxiety and depression were measured with the HADS consisting of 7-item anxiety and 7-item depression subscales. Each subscale has a possible total score ranging from 0 to 21 (≤7normal, 8–10 borderline, and ≥11 probable anxiety or depression). 24 Stress-Related Measurements We measured self-perceived stress using the 10-item PSS, 25 with higher scores indicating greater stress perception. We also assessed the ability to resist or recover from stress with the 6-item BRS. 26 Higher scores reflect greater stress resilience. Caregiver Status and Changes in Lifestyle Patterns Caregiver status was defined with the following question in the ad hoc evaluation: “Are you a caregiver for a dependent person?”Furthermore, we investigated the changes in lifestyle patterns during the confinement reflecting neuropsychiatriclike behaviors (sleep patterns, eating, drinking and smoking habits, and medication use). Participants answered a questionnaire aimed at evaluating change in sleep (hours), caloric food, alcohol and tobacco consumption, use of anxiolytics/ antidepressants, sleeping pills, and analgesics during the confinement compared with preconfinement (eFigure 1, links.lww.com/WNL/C226). The sleep variables were coded to reflect less or more than 7 hours of sleep before and during the confinement. Then, we classified participants under the categories of “No change,”“Decreased,”or “Increased”sleep hours. We categorized the responses to the questions of the rest of the variables as follows: “Decreased”(I have stopped consuming or I have decreased the consumption), “No change”(I have not changed the consumption), and “Increased”(I have increased the consumption moderately or I have increased the consumption significantly). APOE Genotyping The APOE genotype was obtained from the allelic combination of the rs429358 and rs7412 variants. APOE status was determined based on the APOE e4 allele, and the participants were classified as APOE e4carriersorAPOE e4 noncarriers. Neuroimaging and CSF Biomarker Measurements MRI Acquisitions and MRI-Based AD Signature Anatomic 3D T1-weighted fast field echo sequence MRIs were obtained with a 3T scanner (Ingenia CX, Philips, the Netherlands) at the Barcelonaβeta Brain Imaging Center with the following parameters: voxel size = 0.75 mm 3 isotropic, field of view = 240 × 240 × 180 mm 3 ,flip angle = 8°, repetition time = 9.9 ms, echo time = 4.6 ms, and inversion time = 900 ms in sagittal acquisition. FreeSurfer version 6.0 was used to determine the cortical thickness of regions vulnerable to AD. The so-called AD signature was calculated as the surface-area weighted average of the individual thickness values of the following regions: entorhinal, inferior temporal, middle temporal, and fusiform cortices in both hemispheres. 17,27 PET Imaging Acquisitions and Preprocessing [ 18 F] Flutemetamol PET scans were acquired in a Siemens Biograph mCT (Munich, Germany) after performing a cranial CT scan for attenuation correction. One hundred eighty-five MBq (range 166.5–203.5 MBq) of [ 18 F] flutemetamol was injected to the participants, and 4 frames of 5 minutes each were acquired following the waiting period of 90 minutes. An OSEM3D algorithm with 8 iterations and 21 subsets was used to reconstruct the images with point spread function and time-of-flight corrections into a 1.02 × 1.02 × 2.03 mm matrix. e1488 Neurology | Volume 99, Number 14 | October 4, 2022 Neurology.org/N The acquired images were preprocessed using SPM12. Averaged PET images were coregistered to corresponding MRI scans. Following the segmentation of MRIs, the images were normalized to Montreal Neurological Institute (MNI) space together with the PET images. The standardized uptake value ratio (SUVR) was calculated in MNI space from the standard regions (bilateral frontal and parietotemporal areas) and the whole cerebellum asthe reference region. We thentransformed the SUVR values to the centiloid (CL) scale. Amyloid positivity was ascertained using CL values. 28 We defined the cutoffvalue for CL with a threshold of 12 to classify the participants as Aβnegative (<12 CL) or Aβ-positive (≥12 CL). CSF Measurements IL-6 was measured with the Roche NeuroToolKit, a panel of automated Elecsys and prototype immunoassays (Roche Diagnostics International Ltd, Rotkreuz, Switzerland) on a cobas e411 or e601 instrument at the Clinical Neurochemistry Laboratory, Sahlgrenska University Hospital, M¨olndal, Sweden. Glial fibrillary acidic protein (GFAP) and soluble fragment of triggering receptor expressed on myeloid cells 2 (sTREM2), measured with the Roche NeuroToolKit, and also reflecting neuroinflammatory processes, 29 were used to test the specificity of the associations between IL-6 and anxiety-depression. We used the log10‐transformed versions of the IL-6 and GFAP measurements for the analyses as they were not normally distributed. 29 All participants were blinded to the results of their CSF and amyloid PET assessments. Statistical Analysis The characteristics of the samples were defined with means and SDs or medians and ranges for continuous variables and frequencies and percentages for categorical variables. With descriptive purposes, we investigated the differences by sex, caregiver status, Aβpositivity in preconfinement HADS scores, and stress-related measurements (PSS and BRS) with ttests. We also reported raw HADS change scores by group and explored the differences between these groups in lifestyle changes during the confinement with χ 2 analyses. In our main analyses, we performed 2 sets of multivariable linear regression models with HADS total scores during the confinement as the outcome variable. First, we investigated whether AD-related biomarkers showed cross-sectional associations with HADS total scores in the biomarker sample. To this end, we included Aβ-positivity, CSF IL-6, and cortical thickness in AD signature regions as independent variables and demographics, APOE e4 status, and caregiver status as covariates in the model. We also investigated the factors associated with HADS total scores in the whole sample considering the demographics, APOE e4 status, and caregiver status as independent variables. In a second step, we performed the models adjusting by preconfinement anxietydepression levels in the subgroup of participants with available preconfinement HADS scores (see the Results section). In these models, we also controlled for the interindividual variability in the time lag between preconfinement and confinement HADS assessments. As sensitivity analysis, we explored whether anxiety or depression (HADS-Anxiety and HADS-Depression as dependent variables) drove the results of the main analyses. In addition, considering the higher prevalence of anxiety and depression in women, 11 we performed sex-stratified analyses to investigate whether the tested associations were driven by women. All analyses were performed using RStudio v1.4.1103-4 and SPSS 27 (IBM, Armonk, NY) statistical software. Statistical significance was considered when the results yielded a 2-tailed pvalue lower than 0.05. Data Availability The data supporting the findings of the current study may be available on a reasonable request from the ALFA study management team. Results From the ALFA participants who accepted to participate in the study, those with complete HADS evaluation during the confinement were included in the present study (N = 921/2,582, 35.67%). Of these, 254 participants had available AD biomarker data (biomarker sample, Figure 1). The majority of participants were residing in the northeast region of Spain, Catalonia (eFigure 2, links.lww.com/WNL/C226). Participants who accepted to participate in the current study (compared with those who declined [N = 1,661]) had significantly higher years of education (t 2580 =3.87;p< 0.001) and lower preconfinement HADS scores (t 1754 =−2.68; p=0.007). Table 1 shows the demographic, biological, imaging, and clinical data of the biomarker sample and whole sample included in the study. A total of 253 (99.35%) in the biomarker sample and 767 (83.28%) in the whole sample had preconfinement HADS scores. In brief, there were 61.7% of women (N = 568), 14.5% of caregivers (N = 134), 99.1% of White Caucasian (0.9% Latinos), 10.2% (N = 26) of Aβpositive participants (biomarker sample only), and 10.7% (N = 99) with a self-reported clinical diagnosis of anxiety or depression during the confinement (Table 1). Fifty-one percent of all participants (N = 473) completed the HADS in May, 46.5% (N = 428) in June, 1.4% (N = 13) in July, and 0.8% (N = 7) in August. Thirty-six percent (N = 330) of all participants completed the HADS during the de-escalation phase 0, 13.1% (N = 121) during phase 1, 35.9% (N = 332) during phase 2, 11% (N = 102) during phase 3, and 3.9% (N = 36) during phase 4. The month or phase of the confinement when the HADS was completed did not show any effect on the total anxiety-depression scores (eFigure 3, links.lww.com/WNL/C226). Neurology.org/N Neurology | Volume 99, Number 14 | October 4, 2022 e1489 Participants in the whole sample had significantly higher preconfinement HADS scores (t 766 = 40.8; p< 0.001), were younger (t 920 = 255.7; p< 0.001), and had higher years of education (t 920 = 122.4; p< 0.001) than participants in the biomarker sample. The biomarker sample included a higher number of APOE e4 carriers (X 2 = 32.7; p< 0.001). Association of Confinement HADS Total Scores With Confinement-Related Variables In the absence of a control condition, we used proxies of length and intensity of the confinement to assess its association with HADS measurements. Participants who started the confinement at an earlier date or who were confined in smaller-size dwellings did not show higher HADS scores. However, anxietydepression scores were higher in participants who went outdoors less frequently (F = 21.4, p< 0.001) and in those who did not have any open-air space (e.g., garden, terrace, and balcony) at their dwellings (F = 4.24, p= 0.04). Preconfinement and Confinement HADS Measurements In the preconfinement evaluation, the majority of participants scored within the normal ranges of HADS-Anxiety (76.7%) and HADS-Depression (96%). 24 During the confinement, 16.6% of the participants showed a significant increase (p< 0.001) in anxious symptomatology (10.8% changed from normal to borderline, 2.5% from borderline to probable, and 3.3% from normal to probable), and 9.9% showed a significant increase (p< 0.001) in depressive symptomatology (6.1% changed from normal to borderline, 0.9% from borderline to probable, and 2.9% from normal to probable). The change in clinical HADS categories from preconfinement to confinement is provided by sex and caregiver status in eTable 1, links. lww.com/WNL/C226. In the preconfinement evaluations, women had significantly higher total anxiety-depression scores than men (biomarker sample: t 245.4 = 3.87; p< 0.001, whole sample: t 765 = 4.65; p< 0.001). Caregivers also showed higher HADS scores at the preconfinement as compared to noncaregivers (biomarker sample: t 235 = 2.67; p= 0.008, whole sample: t 832 = 3.21; p< 0.001). Aβ-positive or Aβ-negative participants did not show any difference in preconfinement HADS total scores (t 203 = 1.84; p= 0.067). Raw mean change in HADS scores (biomarker sample: 1.5, whole sample: 1.32) was higher in women (1.55 ± 6) vs men (0.72 ± 4.6), younger (1.68 ± 5.7) vs older (0.76 ± 5.3) adults, noncaregivers (1.4 ± 5.5) vs caregivers (0.29 ± 5), Aβ-positive (1.81 ± 7.7) vs Aβ-negative (1.6 ± 5.9) participants, APOE e4 noncarriers (1.35 ± 5.1) vs APOE e4 carriers (1.04 ± 6.1), participants with lower CSF IL-6 levels (1.85 ± 5.9) vs higher IL-6 levels (1.09 ± 5.6), and those with lower years of education (1.4 ± 5.3) vs higher years of education (1.04 ± 5.6). The mean HADS total, HADS-Anxiety, and HADSDepression scores by sex, caregiver, and Aβstatus during preconfinement and confinement are shown in eFigure 4, links.lww.com/WNL/C226. Figure 1 Flow Diagram Illustrating the Recruitment and Number of Participants Included in Cross-sectional and Longitudinal Analyses HADS = Hospital Anxiety and Depression Scale. e1490 Neurology | Volume 99, Number 14 | October 4, 2022 Neurology.org/N Differences in Stress-Related Measurements and Lifestyle Changes by Sex, Caregiver Status, and Amyloid Status During the confinement, women had higher PSS scores than men (biomarker sample: t 240 = 1.97; p= 0.05, whole sample: t 863 = 3.47; p< 0.001). The 2 groups did not show any difference in BRS scores (biomarker sample: t 250 =−0.58; p= 0.562; whole sample: t 858.3 = 0.69; p= 0.493). Compared with noncaregivers, caregivers had higher PSS scores (biomarker sample: t 228 = 2.07; p= 0.04, whole sample: t 181.5 = 4.56; p< 0.001). Nevertheless, caregivers had higher BRS scores than noncaregivers (whole sample only: t 826 = 2.23; p= 0.05). There were no significant differences between Aβ-positive or Aβ-negative participants in PSS (t 193 = 1.24; p= 0.216) or BRS (t 202 = 1.56; p= 0.12) scores. Regarding the analyses investigating the changes in lifestyle patterns, we observed sex differences in hours of sleep and Table 1 Demographic, Biological, Imaging, and Clinical Characteristics of the Study Participants Variable Biomarker sample (N = 254) Whole sample (N = 921) Age, mean (SD), y 63.5 (4.78) 62.7 (6.36) Caucasian, no. (%) 252 (99.2) 912 (99.1) Female, no. (%) 154 (60.6) 568 (61.7) Education, mean (SD), y 13.4 (3.49) 13.8 (3.41) APOE «4 carrier, no. (%) a 146 (57.5) 342 (37.1) Caregiver, no. (%) b 39 (15.4) 134 (14.5) Amyloid positivity (>12CL), no. (%) c 26 (10.2) — AD signature (Cth, mm), mean (SD) d 2.42 (0.096) — IL-6 (pg/mL), median (range) e 3.6 (12.1) — GFAP (ng/mL), median (range) f 7.2 (23.8) — sTREM2 (ng/mL), mean (SD) g 7.92 (2.25) — Preconfinement HADS total scores, mean (SD) h 6.57 (4.81) 7.59 (5.14) HADS-Anxiety scores, mean (SD) h 4.73 (3.22) 5.39 (3.38) HADS-Depression scores, median (range) h 1 (9) 1 (13) Confinement HADS total scores, mean (SD) 8.07 (5.98) 8.91 (6.23) HADS-Anxiety scores, mean (SD) 5.19 (3.48) 5.56 (3.55) HADS-Depression scores, median (range) 2 (14) 2 (17) PSS scores, mean (SD) i 16 (8.68) 16.9 (8.65) BRS scores, mean (SD) j 3.16 (0.39) 3.15 (0.37) Time from preconfinement to confinement evaluations, mean (SD), y 2.37 (0.77) 2.41 (0.76) Currently diagnosed/under treatment for anxiety-depression, no. (%) k 26 (10.2) 99 (10.7) Abbreviations: AD = Alzheimer disease; BRS = Brief Resilience Scale; CL = Centiloid; Cth = cortical thickness, IL-6 = interleukin-6; GFAP = glial fibrillary acidic protein; HADS = Hospital Anxiety and Depression Scale; PSS = Perceived Stress Scale; sTREM2 = soluble fragment of triggering receptor expressed on myeloid cells 2. a Whole sample N = 913. b Biomarker sample N = 238, whole sample N = 834. c Biomarker sample N = 206. d Biomarker sample N = 246. e Biomarker sample N = 234. f Biomarker sample N = 236. g Biomarker sample N = 236. h Biomarker sample N = 253, whole sample N = 767. i Biomarker sample N = 242, whole sample N = 865. j Biomarker sample N = 252, whole sample N = 904. k Biomarker sample N = 252, whole sample N = 907. Neurology.org/N Neurology | Volume 99, Number 14 | October 4, 2022 e1491 food consumption (women>men, X 2 = 7.52; p= 0.023; X 2 = 37.5; p< 0.001, respectively). In addition, we observed a significant difference between caregivers and noncaregivers in food consumption (caregivers>noncaregivers, X 2 = 6.41; p= 0.041). There were no significant differences between Aβ-positive or Aβ-negative participants in any of the investigated lifestyle domain (eTable 2, links.lww.com/WNL/ C226). Factors Associated With Total AnxietyDepression During the Confinement In the biomarker sample, Aβpositivity (B = 3.73; 95% CI 1.1–6.36; p= 0.006), but not thickness in AD signature (B = −5.31; 95% CI −14.5 to 3.87; p= 0.255) or CSF IL-6 (B = −5.13; 95% CI −10.4 to 0.11; p= 0.055), showed a crosssectional association with greater total anxiety-depression scores independent of age, sex, and years of education. When the model was adjusted by preconfinement HADS scores, higher preconfinement anxiety-depression scores were associated with greater confinement HADS scores and the association between Aβpositivity and HADS total scores remained significant. In addition, lower levels of CSF IL-6 were associated with greater HADS total scores irrespective of the preconfinement anxiety-depression level (Figure 2A and Table 2). In the whole sample, younger age (B = −0.12; 95% CI −0.18 to −0.052; p< 0.001), being a woman (B = 1.95; 95% CI 1.1–2.79; p≤0.001), lower years of education (B = −0.16; 95% CI −0.28 to −0.042; p= 0.008), and being a caregiver (B = 1.37; 95% CI 0.24–2.5; p= 0.018) showed crosssectional associations with greater HADS total scores during the confinement. Higher preconfinement HADS scores were associated with greater confinement HADS scores. The associations of age and sex with total anxiety-depression scores were still significant after controlling for the preconfinement HADS scores (Figure 2B and Table 2). Models with HADS subscales (anxiety and depression) as dependent variables showed similar associations in both samples. Among the AD-related biomarkers, only β-amyloid positivity showed a specific association with greater HADSAnxiety scores irrespective of the preconfinement anxiety level (eTable 3, links.lww.com/WNL/C226). Sensitivity Analyses Amyloid Results Given the relatively small percentage of Aβ-positive subjects in our cohort, we confirmed the robustness of the results with continuous CL levels that showed an association with HADS total scores irrespective of preconfinement HADS scores (B = 0.055; 95% CI 0.005–0.1; p= 0.031). Models Excluding the Participants Who Are Currently Diagnosed With or Under the Treatment of AnxietyDepression We performed our main analyses without the participants with self-reported clinical diagnosis and/or under the treatment of anxiety-depression. Our main results in the biomarker sample or the whole sample remained significant in these analyses. In addition, IL-6 showed a cross-sectional association with greater HADS total scores during the confinement (B = −5.13; 95% CI −9.98 to −0.27; p= 0.039). Associations With Neuroinflammation Markers We investigated whether the observed association between neuroinflammation and HADS total scores was specific to IL6 as hypothesized. To this end, we performed our main model Figure 2 Forest Plots Showing the Multivariable Linear Associations With HADS Total Scores During the Confinement The figure shows the estimated amount of change (95% CI) in HADS total scores for a given difference in each factor. (A) Biomarker sample. (B) Whole sample. Both models are adjusted by preconfinement HADS scores and the individual variability between preconfinement and confinement HADS assessments. The colors on the figure represent: black = nonsignificant pvalue, brown = p< 0.05, orange ≤0.01, and red ≤0.001. AD = Alzheimer disease; HADS = Hospital Anxiety and Depression Scale; IL-6 = interleukin-6. e1492 Neurology | Volume 99, Number 14 | October 4, 2022 Neurology.org/N in the biomarker sample including the sTREM2 and GFAP while taking into account of the preconfinement HADS scores. Results showed that the association between neuroinflammation and anxious-depressive symptoms was restricted to IL-6 because STREM2 and GFAP did not show any association with HADS (sTREM2, B = −0.31, p= 0.13; GFAP, B = 1.63, p= 0.618). Adjustments by Measurements of Stress The models considering the preconfinement anxietydepression levels were adjusted by PSS and BRS scores to evaluate whether stress-related variables had any confounding effect on the reported results. The main results remained significant after this adjustment (eTable 4, links. lww.com/WNL/C226). Stratified Analyses by Sex The results of the sex-stratified analyses adjusted by preconfinement HADS levels are reported in Table 3. These analyses revealed that women are driving the associations reported above (Table 3). Adjustments by Changes in Lifestyle Patterns We observed differences in hours of sleep and/or food consumption by sex and caregiver status during the confinement. Therefore, we adjusted our main analyses by hours of sleep and food consumption to control for a potential confounding effect of these variables on the associations tested. Following these adjustments, the previously observed associations did not change. Changes in sleep hours, however, showed an association with HADS total scores (F = 6.23; p= 0.002). Discussion The main results of the present study in CU adults at an increased risk for developing AD were as follows: (1) Aβ positivity and lower CSF IL-6 levels measured 2.4 years before the pandemic were associated with greater anxious-depressive symptomatology during the COVID-19 confinement; (2) the results were mainly seen in women and were independent of demographics, stress-related measurements, and changes in lifestyle patterns during the confinement; and (3) women and caregivers presented higher anxious-depressive symptoms during the confinement. Our sample consisted of adults with low burden of anxietydepression before the COVID-19 pandemic. Even so, and in line with previous reports, 1 16.6% and 9.9% of the participants showed clinically significant increases in anxiousdepressive symptoms, respectively. The change in HADS total scores in the biomarker sample (1.5) is considered a significant difference in clinical settings. 30 Furthermore, we did not find any effect of specificconfinement phase or month on the anxious-depressive symptoms. However, we observed that participants going outdoors less frequently (once a week or less during the confinement and deescalation phases) and those spending the confinement in a dwelling without any open-air space showed higher anxietydepression. Overall, these results support that our sample showed modest but clinically meaningful changes associated with the COVID-19 confinement. β-amyloid positivity was associated with greater anxiousdepressive symptoms during the confinement irrespective of the preconfinement anxiety-depression level. This association was driven by anxiety symptoms. These results are consistent with cross-sectional 31,32 and longitudinal 33,34 data showing that Aβburden is associated with neuropsychiatric symptoms in CU adults. The results were independent of self-reported perceived stress and stress resilience. However, these associations might be mediated by the physiologic stress response. 35 The dysregulation of the hypothalamic-pituitary-adrenal axis may result in a chronic stress response and in elevated adrenal glucocorticoids. This may increase the Aβdeposition and Table 2 Results From the Analyses of Covariance With HADS Total Scores During the Confinement as the Dependent Variable HADS total Biomarker sample, Adj. R 2 = 0.242 B value (95% CI) pValue Age −0.37 (−0.55 to −0.18) <0.001 Sex (women) 1.99 (0.27 to 3.71) 0.024 Years of education −0.27 (−0.5 to −0.03) 0.028 APOE «4 carriers −0.21 (−1.9 to 1.47) 0.803 Caregivers −1.37 (−3.62 to 0.87) 0.228 Amyloid positivity 2.6 (0.074 to 5.12) 0.044 AD signature −3.67 (−12.5 to 5.17) 0.413 IL-6 −5.11 (−10.1 to −0.13) 0.044 Time difference 0.21 (−0.7 to 1.12) 0.645 Preconfinement HADS 0.41 (0.23 to 0.58) <0.001 Whole sample, Adj. R 2 = 0.307 Age −0.09 (−0.15 to −0.033) 0.002 Sex (women) 1.34 (0.55 to 2.13) 0.001 Years of education −0.1 (−0.22 to 0.011) 0.075 APOE «4 carriers −0.26 (−1.3 to 0.51) 0.511 Caregivers −0.54 (−1.6 to 0.51) 0.315 Time difference −0.064 (−0.44 to 0.32) 0.742 Preconfinement HADS 0.63 (0.55 to 0.71) <0.001 Abbreviations: AD = Alzheimer disease; HADS = Hospital Anxiety and Depression Scale; IL-6 = interleukin-6. The unstandardized B represents the variation in HADS total confinement scores with 1-unit variation in a given predictor. Biomarker sample N = 179, whole sample N = 693. Neurology.org/N Neurology | Volume 99, Number 14 | October 4, 2022 e1493 accumulation of tau, which ultimately can cause damage of brain structure and function. 16,36,37 Contrary to our expectations, but in line with this temporality of events, brain integrity in AD-related regions was not associated with anxietydepression. This may indicate that the increase in neuropsychiatric symptomatology in preclinical AD might precede brain atrophy. 38 However, these associations may exist with other brain regions not included in our AD signature such as the insula. 39 Previous findings suggest neuroinflammation as a mechanism by which anxiety and depression are linked to AD pathophysiology. 16,40 Our results showed that the associations between neuroinflammation and anxiety-depression are specific to IL-6 but not to other neuroinflammation markers such as GFAP and sTREM2. Elevated IL-6 levels have been reported previously in subjects with depression 19 and in patients with AD. 16,18 Moreover, long-lasting stressful events, such as the pandemic, may induce the expression of IL-6. 41 Unexpectedly, however, we observed that participants with lower CSF IL-6 levels showed higher anxiety-depression during the confinement irrespective of their anxiety-depression level at the preconfinement. A possible explanation is that the levels of IL-6 might be lower in preclinical AD as reported in early onset AD, 42 and therefore, the associations with anxiety-depression might be different in preclinical AD. Future studies with longitudinal data are required to replicate our results and investigate whether the association of IL-6 with neuropsychiatric symptomatology is different throughout the AD continuum. Our findings showed that the associations between anxietydepression and Aβwere driven by women. We also observed greater changes in sleep patterns in women than men, which is a factor associated with increasing amyloid levels. 43 Although Table 3 Results From the Stratified Analyses of Covariance by Sex in the Biomarker Sample and Whole Sample Women HADS total Biomarker sample Whole sample B (95% CI) pValue B (95% CI) pValue Age −0.49 (−0.77 to −0.22) <0.001 −0.15 (−0.24 to −0.072) <0.001 Years of education −0.28 (−0.61 to 0.056) 0.102 −0.15 (−0.31 to 0.006) 0.059 APOE «4 carriers −0.2 (−2.67 to 2.26) 0.871 −0.35 (−1.45 to 0.75) 0.53 Caregivers −1.73 (−4.77 to 1.29) 0.259 −0.51 (−1.95 to 0.93) 0.484 Amyloid positivity 5.17 (1.28–9.06) 0.01 —— AD signature −1.15 (−13.7 to 11.4) 0.855 —— IL-6 −5.1 (−11.7 to 1.67) 0.14 —— Time difference −0.62 (−0.8 to 2.05) 0.391 −0.074 (−0.62 to 0.47) 0.788 Preconfinement HADS 0.32 (0.1 to 0.55) <0.001 0.62 (0.51 to 0.73) <0.001 Men HADS total Biomarker sample Whole sample B (95% CI) pValue B (95% CI) pValue Age −0.23 (−0.46 to 0.009) 0.059 −0.001 (−0.076 to 0.074) 0.976 Years of education −0.22 (−0.54 to 0.11) 0.188 −0.055 (−0.21 to 0.1) 0.486 APOE «4 carriers 0.063 (−2.13 to 2.26) 0.955 −0.2 (−1.23 to 0.83) 0.707 Caregivers −1.61 (−4.97 to 1.75) 0.342 −1.07 (−2.61 to 0.46) 0.170 Amyloid positivity −0.39 (−3.43 to 2.65) 0.797 —— AD signature −3.59 (−15.9 to 8.77) 0.563 —— IL-6 −4.14 (−11.4 to 3.15) 0.26 —— Time difference −0.003 (−1.08 to 1.08) 0.995 −0.065 (−0.56 to 0.43) 0.798 Preconfinement HADS 0.61 (0.31 to 0.91) <0.001 0.66 (0.55 to 0.77) <0.001 Abbreviations: AD = Alzheimer disease; HADS = Hospital Anxiety and Depression Scale; IL-6 = interleukin-6. The unstandardized B represents the variation in HADS total scores with 1-unit variation a given predictor. e1494 Neurology | Volume 99, Number 14 | October 4, 2022 Neurology.org/N