Sex-dependent grades of haematopoietic modulation in patients with major depressive episodes are associated with suicide attempts

ARTICLE IN PRESS JID: NEUPSY [m6+; June 26, 2020;5:25 ] European Neuropsychopharmacology (2020) 000, 1–14 www.elsevier.com/locate/euroneuro Sex-dependent grades of haematopoietic modulation in patients with major depressive episodes are associated with suicide attempts Patricia Martínez-Botía a , b , 1 , Angela Velasco a , c , d , 1 , Valeria Rolle a , Luis Jiménez-Trevino a , c , d , e , Lorena De la Fuente-Tomás a , c , d , Ángel Bernardo a , f , Tamara Arias a , g , María del Carmen Muñoz-Turrillas a , g , Leticia García-Alvarez a , d , Leticia González-Blanco a , c , d , e , Julia Rodríguez-Revuelta a , c , e , Maria Paz Garcia-Portilla a , c , d , e , Philippe Courtet h , i , Julio Bobes a , c , d , e , Laura Gutiérrez a , b , ∗, Pilar Alejandra Sáiz a , c , d , e , ∗ a Instituto de Investigación Sanitaria del Principado de Asturias (ISPA), Oviedo, Spain b Department of Medicine, Universidad de Oviedo, Spain c Department of Psychiatry, Universidad de Oviedo, Spain d Biomedical Research Networking Centre in Mental Health (CIBERSAM), Oviedo, Spain e Servicio de Salud del Principado de Asturias (SESPA), Oviedo, Spain f Hospital Universitario Central de Asturias, Laboratorio de Diagnóstico Clínico Hematología, Oviedo, Spain g Centro Comunitario de Sangre y Tejidos de Asturias, Oviedo, Spain h Department of Emergency Psychiatry & Post-Acute Care, Lapeyronie Hospital, CHU Montpellier – Inserm U1061, University of Montpellier, Montpellier, France i Fondation FondaMental, Créteil, France Received 19 December 2019; received in revised form 8 May 2020; accepted 9 June 2020 Available online xxx ∗Corresponding authors at: Instituto de Investigación Sanitaria del Principado de Asturias (ISPA) and Universidad de Oviedo, Oviedo, Spain. E-mail addresses: [email protected] (L. Gutiérrez), [email protected] (P.A. Sáiz). 1 These authors contributed equally to this work. https://doi.org/10.1016/j.euroneuro.2020.06.006 0924-977X/ ©2020 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license. ( http://creativecommons.org/licenses/by-nc-nd/4.0/ ) Please cite this article as: P. Martínez-Botía, A. Velasco and V. Rolle et al., Sex-dependent grades of haematopoietic modulation in patients with major depressive episodes are associated with suicide attempts, European Neuropsychopharmacology, https://doi.org/10.1016/j. euroneuro.2020.06.006 2 P. Martínez-Botía, A. Velasco and V. Rolle et al. ARTICLE IN PRESS JID: NEUPSY [m6+; June 26, 2020;5:25 ] KEYWORDS Major depressive episode; Haematopoiesis; Suicide; Inflammation; Sex; LTE Abstract Suicide is the leading cause of non-natural death worldwide, and major depressive disorder (MDD) is the mood disorder with the highest prevalence among individuals with suicidal behaviour (SB). The role of inflammation and immunomodulation in mood disorders has raised interest in recent years, as inflammation biomarkers have been reported to be increased in mood disorder patients, suggesting a role of inflammation in their pathogenesis. The influence of inflammation on the haematopoietic production is well known; however, a comprehensive study of the haematopoietic production in patients with major depressive episodes (MDE) is lacking. We examined global haematopoietic parameters from complete blood counts (CBC) of patients with MDE, in search of prognostic patterns. MDE patients presented differences in several CBC parameters, differences that were clearly pronounced and/or significant in concurrence with suicide attempts (SA). Red and white blood cell lineage parameters were affected, suggesting general haematopoietic modulation or imbalance. We observed distinct haematological parameter changes in women versus men, with men presenting milder alterations than women. Interestingly, we found that the List of Threatening Experiences (LTE) score, but not the Childhood Trauma Questionnaire (CTQ), was associated with the haematopoietic alterations observed exclusively in women and, more importantly, served as a parameter to stratify female MDE patients based on concurrence or non-concurrence with SA. In conclusion, grades of haematopoietic modulation in MDE patients are associated with absence or presence of SA. Haematopoietic manifestations differ between men and women and, in the latter, are markedly influenced by late, and not early, traumatic events. ©2020 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license. ( http://creativecommons.org/licenses/by-nc-nd/4.0/ ) 1. Introduction Suicide is the leading cause of non-natural death worldwide and the second leading cause of mortality in individuals aged 15–29 years (World Health Organization, 2018 ), psychiatric disorders and substance abuse being the acknowledged risk factors. Major depressive disorder (MDD) is the most prevalent mood disorder in individuals with suicidal behaviour (SB), depression itself being associated with a 20-fold increase in suicide mortality rate ( Osby et al., 2001 ). Stressors in childhood and adulthood have been associated with suicide attempts (SA), and there is ongoing debate whether early events have a stronger association with SB in comparison with late or recent events ( Liu et al., 2017 ; Ludwig and Dwivedi, 2018 ). Importantly, individuals who have attempted suicide are at higher risk for future attempts, and a large proportion of them experience selfinflicted injuries or long-term disability, requiring medical attention and follow-up. It is therefore of interest to health systems to develop and rely on more precise personalised algorithms and biomarkers to better predict and prevent SA ( Barrigon and Baca-Garcia, 2018 ). In this direction, efforts have been made to identify different novel biomarkers using high-throughput technologies (GWAS, metabolomics), highlighting the importance of transversal or multidisciplinary approaches to understand mood disorders. The role of inflammation and immunomodulation in different conditions or pathologies, including mood disorders, has raised interest recently ( Pariante, 2017 ). Inflammationrelated biomarkers, such as the neutrophil-to-lymphocyte ratio (NLR) ( Ekinci and Ekinci, 2017 ; Ivkovic et al., 2016 ; Velasco et al., 2020 ) and C-reactive protein (CRP) ( Courtet et al., 2015 ; Gibbs et al., 2016 ), have been reported to be increased in mood disorder patients, suggesting a role of inflammation in their pathogenesis and a potential use for these biomarkers as predictors of SB. More recently, NLR and mean platelet volume (MPV) have been related to the severity of SA ( Orum et al., 2018 ). Inflammatory and infectious diseases, including allergy, asthma, and toxoplasmosis, have been associated with SB ( Postolache et al., 2008 ; Zhang et al., 2012 ). Furthermore, an association has been established between SA, inflammation, and increased risk of morbidity and mortality from natural (cardiometabolic) causes ( Bergen et al., 2012 ). Interestingly, patients treated with pro-inflammatory cytokines demonstrated increased risk of suicidal ideation or SA ( Fragoso et al., 2010 ). A comprehensive model focusing on the influence of the immune system in the pathophysiology of SB has been proposed ( Courtet et al., 2016 ), where childhood abuse, sleep disturbance, infections, and other stressors induce a chronic inflammatory state causing dysregulation of the hypothalamic-pituitary-adrenal axis, increasing cortisol levels and indolamine 2,3-dioxygenase activation, which results in increased N-methyl-D-aspartate (NMDA) agonist and decreased serotonin levels. All these mechanisms may lead to psychological vulnerability and SB. However, the causeconsequence relationship between psychiatric disorders, SB, and systemic inflammation response and feedback is still largely unknown. Please cite this article as: P. Martínez-Botía, A. Velasco and V. Rolle et al., Sex-dependent grades of haematopoietic modulation in patients with major depressive episodes are associated with suicide attempts, European Neuropsychopharmacology, https://doi.org/10.1016/j. euroneuro.2020.06.006 3 ARTICLE IN PRESS JID: NEUPSY [m6+; June 26, 2020;5:25 ] Haematopoiesis occurs in an inter-lineage equilibrium, where production of the different haematopoietic cell types that circulate in the blood is tightly regulated, as they need to be constantly replenished due to their short lifespan. A complete blood count (CBC) test measures a number of variables related to the different circulating blood cell-types, and is a bona fide portrait of an individual’s health status. Hence, complete blood count (CBC) variables follow a given correlation in healthy individuals. When haematopoietic distress, imbalance, or modulation occurs, those correlations are disrupted, because the haematopoietic output shifts depending on the underlying physiological demand. It is well known that in response to external cues (e.g. cytokines during inflammation), the steady state output of the different types of blood cells shifts, usually towards producing higher numbers of innate immune cells (e.g. basophils, neutrophils, etc.), at the expense of erythropoiesis, megakaryopoiesis, and lymphopoiesis ( Zhao and Baltimore, 2015 ). For example, anaemia of inflammation occurs when the body feels a need to increase white blood cell (WBC) production in response to infection, at the expense of red blood cell (RBC) production; in this case, the anaemia is caused by a rise in IL-6 and hepcidin levels, with subsequent inhibition of iron absorption, ultimately resulting in hampered erythropoiesis ( Nemeth and Ganz, 2014 ). In a transgenic mouse model of anaemia of chronic disease, characterised by enhanced CD27-mediated co-stimulation and a strong increase in the production of IFNγ-producing effector T cells, progressive anaemia is due to inhibition of erythropoiesis through the IRF-1-dependent activation of PU.1 transcription factor in haematopoietic precursors ( Libregts et al., 2011 ). This exemplifies how this balance is self-regulated through transcriptional reprogramming of haematopoietic precursors. Contrariwise, recurrent infections are typical of chronic anaemia patients, as, in these cases, the necessity to increase RBC production negatively influences the commitment and differentiation of lymphoid precursors ( Jonker et al., 2017 ). It has been reported that inflammation influences haematopoiesis at the haematopoietic stem cell level ( King and Goodell, 2011 ). Hence, it is plausible that the sub-clinical inflammation ( Köhler et al., 2017 ) reported in patients with MDE may be accompanied by other alterations in the haematopoietic production, which, we hypothesise, will manifest at different levels depending on SA co-occurrence, sex, and early or late stressors. In the present study, we examine global haematopoietic parameters from CBCs of MDE patients, with or without SA, and evaluate potential haematopoietic imbalance based on SA co-occurrence, sex, and early or late stressors, with the objective of defining clinical patterns that would contribute to a better prognosis and SB management in these patients. 2. Experimental procedures 2.1. Study sample We performed a cross-sectional study, including 172 Caucasian participants aged ≥18 years recruited in the area of Oviedo, Spain, from April 2016 to September 2018. All participants gave informed consent. The study was conducted according to the Declaration of Helsinki ( World Medical Association, 2013 ). The cohort consisted of 79 patients recruited at the Mental Health Services of Oviedo, diagnosed with MDE according to the Diagnostic and Statistical Manual of Mental Disorders (DSM5) ( American Psychiatric Association, 2013 ). The exclusion criteria were comorbid psychiatric diagnoses, acute infection, active or chronic inflammatory or autoimmune diseases, smoking ≥20 cigarettes/day, obesity (BMI > 30 kg/m 2 ), current treatment with anti-inflammatory or immunosuppressant drugs, acute coronary syndrome, history of chronic renal, hepatic, or cerebrovascular disease, and reported haematological disorders. Our control population consisted of 93 healthy active blood donors approved by the regional blood bank (Centro Comunitario de Sangre y Tejidos de Asturias, CCSTA). 2.2. Clinical assessment Patients were assessed by well-trained interviewers using an ad hoc protocol for sociodemographic and clinical data. Psychometric evaluation included the Spanish versions of different scales and questionnaires. We employed the Hamilton Depression Rating Scale (HDRS) ( Bobes et al., 2003 ) to determine severity of depression; the Childhood Trauma Questionnaire–Short Form (CTQ-SF) ( Hernandez et al., 2013 ), a self-report questionnaire, to detect early life stressors (emotional, physical and sexual abuse, and emotional and physical neglect); the List of Threatening Experiences (LTE) to detect twelve stressful life events in the six months prior to the evaluation ( Motrico et al., 2013 ), and the Medical Damage Scale (MDS) ( Beck et al., 1975 ) to score medical damage due to SA, ranging from 0 (none) to 8 (dead). SA was defined as a “self-initiated sequence of behaviours by an individual who, at the time of initiation, expected that the set of actions would lead to his or her own death” ( American Psychiatric Association, 2013 ). For every patient, seven days prior to assessment, mean daily antidepressant doses were calculated and standardised to fluoxetine equivalents ( Hayasaka et al., 2015 ). 2.3. CBC analysis Fasting blood samples were collected in the morning in EDTA tubes. CBCs were performed the same day using a Sysmex XN-10/XN-20 haematology analyser. 2.4. Statistical analysis Descriptive parameters were shown as mean and standard deviation. In addition, normal distribution was assessed by the Shapiro– Wilk test. A Chi-square test was used for comparison of categorical variables, whereas an unpaired Mann–Whitney U test or a Kruskal– Wallis test were used to compare continuous variables among two or more groups, respectively. False discovery rate (FDR) correction was applied to account for multiple comparisons. Correlation analyses were performed by Spearman correlation, followed by FDR correction. Statistical significance was considered p (or q value in the case of FDR correction) < 0.05. All data were analysed using R version 3.6 (R Core Team, 2019 ), and figures were produced using the ggplot2 package ( Wickham, 2009 ). Principal component analysis (PCA) was performed after scale-to-interval normalisation of variable values using Perseus Software version 1.5.2.6. Please cite this article as: P. Martínez-Botía, A. Velasco and V. Rolle et al., Sex-dependent grades of haematopoietic modulation in patients with major depressive episodes are associated with suicide attempts, European Neuropsychopharmacology, https://doi.org/10.1016/j. euroneuro.2020.06.006 4 P. Martínez-Botía, A. Velasco and V. Rolle et al. ARTICLE IN PRESS JID: NEUPSY [m6+; June 26, 2020;5:25 ] 3. Results 3.1. Patient cohort: socio-demographics and psychiatric evaluation The study population consisted of 79 MDE patients [mean age (SD): 52.28 (10.56) years; females: 46 (58.2%)], most with MDD [ n = 71 (89.9%)]. MDE patients were stratified into two major groups based on history of SA (MDE SA) [ n = 48 (60.8%)] or absence of SA (MDE noSA) [ n = 31 (39.2%)]. The control group consisted of 93 healthy active blood donors [mean age (SD): 48.22 (11.44) years; females: 40 (43.0%)]. No significant differences between patient groups were identified with regard to sociodemographic characteristics or sex. Age differed considerably between patient and control groups, although post-hoc pairwise analysis (Duncan) revealed significant differences only in the MDE noSA group compared with the control group ( Table 1 and Supplementary Table 1). 3.2. CBC of MDE patients: grades of haematopoietic modulation associated with SA The CBCs of all patients were within normal range for all variables analysed, including WBC and differential count (neutrophils, lymphocytes, monocytes, eosinophils, and basophils), RBC count, mean corpuscular volume (MCV), red cell distribution width (RDW), haematocrit (HCT), and haemoglobin-related variables (haemoglobin (Hb), mean corpuscular haemoglobin (MCH), mean corpuscular haemoglobin concentration (MCHC)), and platelets and mean platelet volume (PLT, MPV). From these data, the neutrophil-to-lymphocyte (NLR), monocyte-to-lymphocyte (MLR), and platelet-to-lymphocyte (PLR) ratios were calculated ( Table 2 and Supplementary Table 1). Principal component analysis (PCA) was performed on the CBC variables as presented in Supplementary Table 1 (taking into consideration only the WBC differential counts and not the percentages) in order to visualise the clustering of the patients in the different groups and the controls. This analysis revealed that while the clusters of MDE patients without SA (MDE noSA) and the controls overlapped to a large extent, the cluster of MDE patients with SA (MDE SA) presented only a partial overlap with the cluster of controls, with a number of MDE SA patients located very distant from the nucleus (or centre) of the control cluster ( Fig. 1 ). Of note, the MDE SA patients who clearly separated from the control cluster nucleus were females. This data suggests differential clustering of MDE patients, in concurrence with or without SA, based on their CBC measurements. To identify the variables responsible for this differential clustering, we next compared potential haematological differences in blood samples between the two major subgroups of MDE patients (SA and noSA) and controls. Descriptive analysis revealed significant changes in RBC lineage-related parameters, e.g. RBC count (reduced), MCV (increased), RDW (reduced), MCH (increased), and MCHC (reduced), with no Hb or HCT changes, suggesting signs of stress erythropoiesis in MDE patients. There were no significant changes in WBC counts; however, there was a significant reduction in monocytes and an increase in eosinophils and basophils, suggestive of systemic inflammation. Platelet-related parameters showed a tendency toward thrombocytosis (not significant) with a significant reduction in MPV in MDE patients versus controls. The difference in these variables was significant and/or greater in MDE patients in concurrence with SA, supporting the PCA results. Furthermore, a tendency toward increased NLR and significantly reduced MLR were observed in MDE SA patients ( Tables 2 and 3 and Fig. 2 ). 3.3. Disruption of haematopoietic equilibrium in MDE patients We next evaluated the haematological equilibrium of CBC variables by studying their correlations in MDE patients versus controls, with the hypothesis that the identified differences and differential clustering observed in the PCA analysis could be explained by the loss of canonical or appearance of de novo correlations, supporting the notion of haematopoietic distress or production imbalance. In fact, the significant correlations identified in the control group, and thus, the normal equilibrium of haematopoietic production, were gradually disrupted in MDE patients, with increasing severity associated with concurrence of SA ( Fig. 3 ). Interestingly, we observed new significant correlations of haematopoietic variables that would indicate an imbalance of haematopoiesis not present in the control group. This imbalance would suggest an incipient shift favouring WBC lineage at the expense of RBC lineage production ( Fig. 3 ). 3.4. Sex and haematopoietic modulation in MDE It is well known that some haematological parameters from a CBC depend on age and sex ( Cheng et al., 2004 ; Qiao et al., 2014 ); therefore we performed multiple regression analyses, which allowed us to identify the haematological parameters dependent on or influenced by these variables in our cohort of patients and controls ( Table 2 ). As is already known, females presented slightly reduced RBC count, Hb, HCT, and MCH compared with males. Contrariwise, females displayed significantly higher platelet numbers. This also explains the differential clustering of male and female individuals in the PCA ( Fig. 1 ). Additionally, we observed agerelated differences that, although small, were still significant; i.e. the RDW % was higher in older individuals while the opposite was true for lymphocyte counts ( Table 2 ). It was therefore important to study whether the haematopoietic imbalance observed in our cohort was maintained when the sexes were studied separately and whether it would impact males and females equally. Therefore, we next performed the same comparative analysis after stratification of our cohort based on sex. Of note, the difference observed in age in the global population (MDE noSA vs controls) disappeared in women after stratifying our cohort, but persisted in men (data not shown). We observed increased SA-associated haematopoietic imbalance both in men and women; however, both sexes presented particularities and separate haematopoietic modulation profiles. While females experienced the most eviPlease cite this article as: P. Martínez-Botía, A. Velasco and V. Rolle et al., Sex-dependent grades of haematopoietic modulation in patients with major depressive episodes are associated with suicide attempts, European Neuropsychopharmacology, https://doi.org/10.1016/j. euroneuro.2020.06.006 5 ARTICLE IN PRESS JID: NEUPSY [m6+; June 26, 2020;5:25 ] Table 1 Sociodemographic and clinical data. SD: Standard deviation; HDRS: Hamilton Depression Rating Scale; MDE: Major Depressive Episode; SA: Suicide attempt; noSA: No suicide attempt; MDS: Medical Damage Score; LTE: List of Threatening Experiences; CTQ: Childhood Trauma Questionnaire; SSRI: Selective Serotonin Reuptake Inhibitor; SNRI: Serotonin and Norepinephrine Reuptake Inhibitor; NDRI: Norepinephrine Dopamine Reuptake Inhibitor; 1 Statistically different (Post-hoc Duncan test). MDE MDE SA MDE noSA Healthy controls MDE MDE + Controls n = 79 n = 48 n = 31 n = 93 X 2 ∗/ T test ( p ) X 2 ∗/ANOVA ( p ) Sex [ n (%)] Males 33 (41.8%) 18 (37.5%) 15 (48.4%) 53 (57.0%) 0.918 ∗(0.338) 4.849 ∗(0.089) Females 46 (58.2%) 30 (62.5%) 16 (51.6%) 40 (43.0%) 1.414 (0.162) 3.810 (0.024) Age [Mean (SD)] 52.28 (10.56) 50.94 (9.71) 54.35 (11.61) 1 48.22 (11.44) 1 Diagnosis [ n (%)] Major Depressive Disorder 71 (89.9%) 43 (89.6%) 28 (90.3%) 0.011 ∗(0.915) Bipolar Disorder 8 (10.1%) 5 (10.4%) 3 (9.7%) HDRS [Mean (SD)] 21.15 (4.21) 21.04 (4.56) 21.32 (3.67) 0.288 (0.774) Number of SA [Mean (SD)] –2.58 (2.53) –– Age of 1st SA [Mean (SD)] – 39.71 (13.64) –– MDS [Mean (SD)] – 3.17(1.45) –– LTE [Mean (SD)] 2.99 (2.44) 3.25 (2.99) 2.58 (1.12) −1.193 (0.237) CTQ Total Score [Mean (SD)] 51.09 (17.90) 53.94 (19.93) 46.77 (13.45) −1.894 (0.062) CTQ Physical Abuse [Mean (SD)] 6.64 (3.39) 7.13 (4.07) 5.90 (1.78) −1.817 (0.074) CTQ Sexual Abuse [Mean (SD)] 6.82 (4.34) 7.32 (4.81) 6.06 (3.42) −1.344 (0.183) CTQ Emotional Abuse [Mean (SD)] 8.88 (4.62) 9.45 (4.86) 8.03 (4.16) −1.331 (0.187) CTQ Physical Neglect [Mean (SD)] 8.41 (3.83) 8.85 (3.71) 7.74 (3.98) −1.256 (0.213) CTQ Emotional Neglect [Mean (SD)] 11.03 (5.40) 11.64 (5.48) 10.10 (5.21) −1.239 (0.219) Pharmacological Treatment [ n (%)] 74 (93.7%) 46 (95.8%) 28 (90.3%) 0.965 ∗(0.326) Antidepressant equivalent to fluoxetine 40 mg/day [Mean (SD)] 45.21 (19.08) 45.74 (20.71) 44.36 (16.41) −0.316 (0.766) SSRI [ n (%)] 26 (32.9%) 17 (35.4%) 9 (29.0%) 0.348 ∗(0.555) SNRI [ n (%)] 36 (45.6%) 21 (43.6%) 15 (48.4%) 0.163 ∗(0.686) NDRI [ n (%)] 7 (8.9%) 4 (8.3%) 3 (9.7%) 0.042 ∗(0.837) Others [ n (%)] 36 (45.6%) 23 (47.9%) 13 (41.9%) 0.272 ∗(0.602) Mood Stabilisers [ n (%)] 10 (12.7%) 7 (14.6%) 3 (9.7%) 0.410 ∗(0.522) Antipsychotics [ n (%)] 32 (40.5%) 23 (47.9%) 9 (29.0%) 2.787 ∗(0.095) Benzodiazepines [ n (%)] 67 (84.8%) 43 (89.6%) 24 (77.4%) 2.163 ∗(0.141) Others [ n (%)] 12 (15.18%) 6 (12.5%) 6 (19.4%) 0.687 ∗(0.407) Tobacco consumption [ n (%)] Yes 36 (45.6%) 20 (41.7%) 16 (51.6%) 0.751 ∗(0.386) No 43 (54.4%) 28 (58.3%) 15 (48.4%) No. of cigarettes/day [Mean (SD)] 15.19 (9.56) 13.35 (5.76) 17.50 (12.70) 1.307 (0.200) dent changes, these were not always significantly altered in males ( Table 3 ). Regarding RBC lineage, a significant increase in RBC count with no change in RDW was observed in MDE females; however, in males, the MCV and RDW changes were significant only in concurrence with SA ( Fig. 4 ). RDW is affected by age, as our analysis showed ( Table 2 ), but since age was different in the MDE noSA group and the significant difference was detected in the MDE SA group, we consider our results valid. These data suggest that females with MDE are more prone to experience haematological imbalance, affecting both the RBC and WBC lineages, while men present overall milder alterations. 3.5. CTQ and LTE scores and haematopoietic modulation in MDE Further analyses were conducted to evaluate the associations between stressors during childhood and adulthood, as assessed using CTQ and LTE scores, respectively, and the Please cite this article as: P. Martínez-Botía, A. Velasco and V. Rolle et al., Sex-dependent grades of haematopoietic modulation in patients with major depressive episodes are associated with suicide attempts, European Neuropsychopharmacology, https://doi.org/10.1016/j. euroneuro.2020.06.006 6 P. Martínez-Botía, A. Velasco and V. Rolle et al. ARTICLE IN PRESS JID: NEUPSY [m6+; June 26, 2020;5:25 ] Table 2 Complete blood parameters in the patient and control groups. Variables are summarised as mean (standard deviation). Differences between the MDE SA and MDE noSA groups and the healthy controls were assessed by one-way Kruskal–Wallis test. Variables with significant differences ( p < 0.05) are highlighted in bold. MDE: Major Depressive Episode; SA: Suicide attempt; noSA: No suicide attempt; WBC: white blood cell count; RBC: red blood cell count; MCV: mean corpuscular volume; MCH: mean corpuscular haemoglobin; MCHC: mean corpuscular haemoglobin concentration; RDW: red cell distribution width; MPV: mean platelet volume; NLR: neutrophil-to-lymphocyte ratio; PLR: platelet-to-lymphocyte ratio; MLR: monocyte-to-lymphocyte ratio. MDE MDE SA MDE noSA Healthy controls One-way Kruskal–Wallis Sex adjusted Age adjusted N = 79 N = 48 N = 31 N = 93 p p p WBC (10 3 / μL) 7.47 (2.25) 7.82 (2.38) 6.93 (1.94) 7.11 (1.66) 0.205 1 1 RBC (10 6 / μL) 4.73 (0.46) 4.69 (0.43) 4.79 (0.49) 4.90 (0.38) 0.017 < 0.001 1 Haemoglobin (g/dL) 14.31 (1.54) 14.17 (1.60) 14.51 (1.44) 14.56 (1.11) 0.460 < 0.001 1 Haematocrit (%) 42.49 (4.03) 42.25 (4.12) 42.86 (3.93) 42.34 (2.79) 0.945 < 0.001 1 MCV (fL) 89.76 (5.22) 89.87(6.35) 89.59 (2.75) 86.52 (3.94) < 0.001 1 1 MCH (pg) 31.83 (9.84) 32.80 (12.54) 30.33 (1.29) 29.73 (1.50) 0.006 < 0.001 1 MCHC (g/dL) 33.73 (1.29) 33.65 (1.45) 33.85 (1.00) 34.37 (1.00) 0.003 1 1 RDW (%) 13.39 (1.24) 13.41 (1.41) 13.36 (0.96) 13.70 (1.01) 0.022 1 < 0.001 Neutrophils (10 3 / μL) 4.23 (1.84) 4.54 (2.02) 3.74 (1.39) 3.95 (1.29) 0.109 1 1 Lymphocytes (10 3 / μL) 2.38 (0.84) 2.41 (0.87) 2.35 (0.81) 2.31 (0.78) 0.835 1 < 0.001 Monocytes (10 3 / μL) 0.56 (0.17) 0.57 (0.16) 0.55 (0.19) 0.64 (0.18) 0.012 1 1 Eosinophils (10 3 / μL) 0.24 (0.15) 0.25 (0.15) 0.21 (0.14) 0.19 (0.12) 0.043 1 1 Basophils (10 3 / μL) 0.05 (0.02) 0.06 (0.02) 0.05 (0.02) 0.03 (0.01) < 0.001 1 1 Platelets (10 3 / μL) 252.16 (68.97) 260.08 (80.56) 239.90 (44.09) 244.25 (60.63) 0.540 < 0.001 1 MPV (fL) 10.83 (0.80) 10.80 (0.81) 10.89 (0.78) 11.29 (0.94) 0.008 1 1 NLR 1.97 (1.35) 2.17 (1.66) 1.68 (0.57) 1.87 (0.80) 0.510 1 1 PLR 117.26 (53.49) 122.39 (64.51) 109.30 (28.50) 117.99 (54.77) 0.952 1 1 MLR 0.26 (0.11) 0.26 (0.12) 0.25 (0.08) 0.30 (0.13) 0.017 1 1 Fig. 1 Principal component analysis (PCA) of haematological parameters in MDE patients and healthy controls shows distinct clustering of samples depending on co-occurrence of SA. PCA of haematological parameters performed on MDE patients (MDE noSA, MDE SA, and controls) is depicted. Please cite this article as: P. Martínez-Botía, A. Velasco and V. Rolle et al., Sex-dependent grades of haematopoietic modulation in patients with major depressive episodes are associated with suicide attempts, European Neuropsychopharmacology, https://doi.org/10.1016/j. euroneuro.2020.06.006 7 ARTICLE IN PRESS JID: NEUPSY [m6+; June 26, 2020;5:25 ] Table 3 Differences between patient groups, stratified by severity of MDE (SA/noSA) and sex, and healthy controls were evaluated by the Mann–Whitney U test, and the False Discovery Rate (FDR, q value) is indicated. Variables with significant differences are highlighted in bold. Ratios show the relationship between the means of each patient group and the healthy controls for each haematological parameter. MDE: Major Depressive Episode; SA: Suicide attempt; noSA: No suicide attempt; WBC: white blood cell count; RBC: red blood cell count; MCV: mean corpuscular volume; MCH: mean corpuscular haemoglobin; MCHC: mean corpuscular haemoglobin concentration; RDW: red blood cell distribution width; MPV: mean platelet volume; NLR; neutrophil-to-lymphocyte ratio; PLR: platelet-to-lymphocyte ratio; MLR: monocyte-to-lymphocyte ratio. Global Male Female MDE SA MDE noSA MDE SA MDE noSA MDE SA MDE noSA q Ratio q Ratio q Ratio q Ratio q Ratio q Ratio WBC (10 3 / μL) 0.236 0.722 0.687 0.811 0.726 0.134 0.85 RBC (10 6 / μL) 0.02 0.96 0.218 0.638 0.811 0.190 0.309 Haemoglobin (g/dL) 0.342 0.676 0.856 0.914 0.797 0.630 Haematocrit (%) 0.884 0.947 0.687 0.811 0.698 0.576 MCV (fL) < 0.001 1.04 < 0.001 1.04 0.04 1.03 0.153 0.005 1.04 0.003 1.04 MCH (pg) 0.02 1.10 0.125 0.056 0.365 0.182 0.409 MCHC (g/dL) 0.02 0.98 0.077 0.130 0.811 0.190 0.02 0.98 RDW (%) 0.02 0.98 0.218 0.04 0.96 0.351 0.660 0.630 Neutrophils (10 3 / μL) 0.193 0.558 0.587 0.811 0.726 0.128 Lymphocytes (10 3 / μL) 0.656 0.947 0.995 0.811 0.732 0.521 Monocytes (10 3 / μL) 0.03 0.88 0.089 0.587 0.860 0.153 0.031 0.71 Eosinophils (10 3 / μL) 0.03 1.34 0.558 0.242 0.365 0.190 0.978 Basophils (10 3 / μL) < 0.001 2.05 < 0.001 1.81 0.001 1.87 0.009 1.88 < 0.001 2.27 0.031 1.78 Platelets (10 3 / μL) 0.386 0.993 0.812 0.994 0.726 0.921 MPV (fL) 0.020 0.96 0.111 0.242 0.145 0.153 0.693 NLR 0.545 0.676 0.812 0.811 0.855 0.309 PLR 0.998 0.917 0.964 0.811 0.883 0.921 MLR 0.02 0.88 0.218 0.638 0.990 0.112 0.128 haematopoietic modulation observed in MDE patients, with or without stratification based on sex. First, we studied the distribution of CTQ and LTE scores in our patient cohort. Although there was a tendency towards a progressive increase in those scores from lower in MDE noSA patients to higher in MDE SA patients, there were no significant changes in CTQ nor LTE scores (data not shown). In order to analyse the association of CTQ and LTE scores with the haematopoietic imbalance observed in MDE patient subgroups, we generated new ratio variables calculated as “haematopoietic parameter value vs CTQ or LTE score” per patient ( Table 4 ) in the global MDE cohort and after stratification based on sex. This analysis showed that LTE (and not CTQ) was associated with the haematological alterations observed in females only, and that the ratios calculated with all the RBC lineage parameters and those calculated with lymphocytes and monocytes as part of the WBC differential clearly separated SA from noSA in females ( Fig. 5 ). These results suggest that females are more susceptible to developing signs of haematopoietic distress, that this imbalance is more severe in concurrence with SA, and that the correlation of all RBC lineage parameters and some of the WBC parameters with the LTE score clearly separates SA from noSA in female patients with MDE. 4. Discussion We examined global haematological parameters from CBCs of MDE patients, with or without SA, and evaluated potential haematopoietic modulation, imbalance or distress due to co-occurrence of SA, and how sex and early or late stressors may influence potential haematopoietic modulation manifestations in MDE patients. We suggest that the haematopoietic imbalance observed in these patients goes beyond the inflammatory status previously reported by others ( Courtet et al., 2016 , 2015 ; Ekinci and Ekinci, 2017 ; Gibbs et al., 2016 ; Glaus et al., 2018a , b ; Ivkovic et al., 2016 ; Kayhan et al., 2017 ; Liu et al., 2012 ; Mazza et al., 2019 ; Miller and Raison, 2016 ; Orum et al., 2018 ). Haematopoietic imbalance, modulation, or distress refers to the physiological response of the haematopoietic system to a certain stimulus, which causes loss of the haematopoietic production equilibrium. This loss of equilibrium may not affect single variables (as the normal ranges for some of them are considerably wide), and hence, CBCs as such may be within the normal range. However, since the haematopoietic production is so tightly regulated, correlations amongst variables start to be lost even when shifts of variables are still present within the normal range, but deviate from the values and correlations measured in a given individual under healthy conditions. A simple cold, which is transitory, will induce such distress. Furthermore, studying a given variable, like the red blood cell count, in a certain patient cohort, such as the one as we present in this manuscript, may show a skewed distribution towards a minimum threshold -still within the normal range-, while in healthy donors, the distribution is evenly dispersed within the top and bottom limits. While it is important to acPlease cite this article as: P. Martínez-Botía, A. Velasco and V. Rolle et al., Sex-dependent grades of haematopoietic modulation in patients with major depressive episodes are associated with suicide attempts, European Neuropsychopharmacology, https://doi.org/10.1016/j. euroneuro.2020.06.006 8 P. Martínez-Botía, A. Velasco and V. Rolle et al. ARTICLE IN PRESS JID: NEUPSY [m6+; June 26, 2020;5:25 ] Fig. 2 Haematological parameters are affected in MDE patients and changes are more pronounced in co-occurrence with SA. (a) Box plots of selected haematological parameters from the CBCs of MDE patients (SA and noSA) and healthy controls. Top row depicts some parameters of the RBC lineage. Middle row depicts some parameters of the WBC lineage differential. Lower row depicts the mean platelet volume. (b) Box plots of calculated ratios. RBC, red blood cell; MCV, mean corpuscular volume; MCHC, mean corpuscular haemoglobin concentration; MPV, mean platelet volume; NLR, neutrophil-lymphocyte ratio; and MLR, monocytelymphocyte ratio. ∗p < 0.05 and ∗∗∗ p < 0.001. NS: not significant. knowledge that CBCs of all patients were within the normal range, we observed clear signs of stress erythropoiesis (reduced RBC and RDW with increased MCV) accompanying inflammation (decreased monocytes and increased basophils and eosinophils). The tendency toward increased platelet counts, although not significant, along with reduced MPV, are suggestive of reactive thrombocytosis and platelet vesiculation, which is normally present in acute phase response or systemic inflammation. All these changes were more pronounced or significant in MDE patients in cooccurrence with SA. Furthermore, the equilibrium of the haematopoietic production, as exemplified by haematological variable correlation matrixes, showed a disappearance of many correlations and the appearance of de novo correlations in MDE patients versus controls. These data, taken together, suggest increasing grades of haematopoietic modulation/distress in MDE patients (associated with SA), indicative of inflammation (as previously reported), but also Please cite this article as: P. Martínez-Botía, A. Velasco and V. Rolle et al., Sex-dependent grades of haematopoietic modulation in patients with major depressive episodes are associated with suicide attempts, European Neuropsychopharmacology, https://doi.org/10.1016/j. euroneuro.2020.06.006 9 ARTICLE IN PRESS JID: NEUPSY [m6+; June 26, 2020;5:25 ] Fig. 3 Correlation of haematological parameters is disturbed in MDE patients. Correlation matrixes of haematological parameters in MDE patients (noSA and SA) and healthy controls are depicted. Significant correlations are shown by different sized circles, indicating the strength of the slope. Direct correlations are coloured blue, while negative correlations are coloured red. Note that many correlations identified in the healthy control group disappear in MDE patients. De novo correlations are framed in a red-line square. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) indicative of underlying or onset of stress erythropoiesis. Changes in haematological parameters in mood disorders have been previously reported, including the appearance of specific correlations as we report in this study ( Wysokinski and Szczepocka, 2016 , 2018 ). In some reports, the tendencies observed are the opposite (increased RDW vs decreased RDW as we observed) ( Demircan et al., 2016 ). Heterogeneity among study cohorts (probably due to sociodemographic and environmental factors) and sample size may condition this discrepancy, separating inflammation-associated erythroid stress due to iron deficiency (or malabsorption) from other types of dyserythropoiesis. Furthermore, the variables affected by sex (i.e. RBC, HB, HCT, MCH, and PLT) or by age (RDW and lymphocyte count) suggest that correction or Please cite this article as: P. Martínez-Botía, A. Velasco and V. Rolle et al., Sex-dependent grades of haematopoietic modulation in patients with major depressive episodes are associated with suicide attempts, European Neuropsychopharmacology, https://doi.org/10.1016/j. euroneuro.2020.06.006