A primer on the use of machine learning to distil knowledge from data in biological psychiatry.

Applications of machine learning in the biomedical sciences are growing rapidly. This growth has been spurred by diverse cross-institutional and interdisciplinary collaborations, public availability of large datasets, an increase in the accessibility of analytic routines, and the availability of powerful computing resources. With this increased access and exposure to machine learning comes a responsibility for education and a deeper understanding of its bases and bounds, borne equally by data scientists seeking to ply their analytic wares in medical research and by biomedical scientists seeking to harness such methods to glean knowledge from data. This article provides an accessible and critical review of machine learning for a biomedically informed audience, as well as its applications in psychiatry. The review covers definitions and expositions of commonly used machine learning methods, and historical trends of their use in psychiatry. We also provide a set of standards, namely Guidelines for REporting Machine Learning Investigations in Neuropsychiatry (GREMLIN), for designing and reporting studies that use machine learning as a primary data-analysis approach. Lastly, we propose the establishment of the Machine Learning in Psychiatry (MLPsych) Consortium, enumerate its objectives, and identify areas of opportunity for future applications of machine learning in biological psychiatry. This review serves as a cautiously optimistic primer on machine learning for those on the precipice as they prepare to dive into the field, either as methodological practitioners or well-informed consumers.

Polygenic risk scores analyses of psychiatric and metabolic traits with antipsychotic-induced weight gain in schizophrenia: an exploratory study.

Given the polygenic nature of antipsychotic-induced weight gain (AIWG), we investigated whether polygenic risk scores (PRS) for various psychiatric and metabolic traits were associated with AIWG. We included individuals with schizophrenia (SCZ) of European ancestry from two cohorts (N = 151, age = 40.3 ± 11.8 and N = 138, age = 36.5 ± 10.8). We investigated associations of AIWG defined as binary and continuous variables with PRS calculated from genome-wide association studies of body mass index (BMI), coronary artery disease (CAD), fasting glucose, fasting insulin, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol (LDL-C), triglycerides, type 1 and 2 diabetes mellitus, and SCZ, using regression models. We observed nominal associations (uncorrected p < 0.05) between PRSs for BMI, CAD, and LDL-C, type 1 diabetes, and SCZ with AIWG. While results became non-significant after correction for multiple testing, these preliminary results suggest that PRS analyses might contribute to identifying risk factors of AIWG and might help to elucidate mechanisms at play in AIWG.

Older molecular brain age in severe mental illness.

Psychiatric disorders are associated with accelerated aging and enhanced risk for neurodegenerative disorders. Brain aging is associated with molecular, cellular, and structural changes that are robust on the group level, yet show substantial inter-individual variability. Here we assessed deviations in gene expression from normal age-dependent trajectories, and tested their validity as predictors of risk for major mental illnesses and neurodegenerative disorders. We performed large-scale gene expression and genotype analyses in postmortem samples of two frontal cortical brain regions from 214 control subjects aged 20-90 years. Individual estimates of "molecular age" were derived from age-dependent genes, identified by robust regression analysis. Deviation from chronological age was defined as "delta age". Genetic variants associated with deviations from normal gene expression patterns were identified by expression quantitative trait loci (cis-eQTL) of age-dependent genes or genome-wide association study (GWAS) on delta age, combined into distinct polygenic risk scores (PRScis-eQTL and PRSGWAS), and tested for predicting brain disorders or pathology in independent postmortem expression datasets and clinical cohorts. In these validation datasets, molecular ages, defined by 68 and 76 age-related genes for two brain regions respectively, were positively correlated with chronological ages (r = 0.88/0.91), elevated in bipolar disorder (BP) and schizophrenia (SCZ), and unchanged in major depressive disorder (MDD). Exploratory analyses in independent clinical datasets show that PRSs were associated with SCZ and MDD diagnostics, and with cognition in SCZ and pathology in Alzheimer's disease (AD). These results suggest that older molecular brain aging is a common feature of severe mental illnesses and neurodegeneration.

Pharmacogenetic Implications for Antidepressant Pharmacotherapy in Late-Life Depression: A Systematic Review of the Literature for Response, Pharmacokinetics and Adverse Drug Reactions.

Affecting up to 15% of older adults, late-life depression (LLD) is characterized by the occurrence of depressive symptoms after the age of 50-65 years and maybe pathophysiologically distinct from depression in younger adults. Therefore, LLD is challenging to treat, and predictive genetic testing might be essential to improve treatment in this vulnerable population. The current review aims to provide a summary of the literature exploring genetic associations with antidepressant treatment outcomes in late-life. We conducted a systematic search of three integrated electronic databases. We identified 29 articles investigating genetic associations with antidepressant treatment outcomes, pharmacokinetic parameters, and adverse drug reactions in older adults. Given the small number of investigations conducted in older adults, it is difficult to conclude the presence or absence of genetic associations with the outcomes of interest. In sum, the most substantial amount of evidence exists for the CYP2D6 metabolizer status, SLC6A4 5-HTTLPR, and BDNF rs6265. These findings are consistent in the literature when not restricting to older adults, suggesting that similar treatment recommendations may be provided for older adults regarding genetic variation, such as those outlined for CYP2D6 by the Clinical Pharmacogenetics Implementation Consortium. Nonetheless, further studies are required in well-characterized samples, including genome-wide data, to validate if similar treatment adjustments are appropriate in older adults, given that there appear to be significant effects of genetic variation on antidepressant treatment factors.

Investigation of the Gut Microbiome in Patients with Schizophrenia and Clozapine-Induced Weight Gain: Protocol and Clinical Characteristics of First Patient Cohorts.

BACKGROUND: Emerging evidence suggests an important role of the human gut microbiome in psychiatry and neurodevelopmental disorders. An increasing body of literature based on animal studies has reported that the gut microbiome influences brain development and behavior by interacting with the gut-brain axis. Furthermore, as the gut microbiome has an important role in metabolism and is known to interact with pharmaceuticals, recent evidence suggests a role for the microbiome in antipsychotic-induced metabolic side effects in animals and humans. PURPOSE: Here we present the protocol for a two-phase study investigating the gut microbiome in healthy controls and in patients with schizophrenia treated with antipsychotics. METHODS: Phase I of our study involves humans exclusively. We recruit 25 patients who are chronically treated with clozapine and compare them with 25 healthy controls matched for age, sex, BMI, and smoking status. A second cohort consists of 25 patients newly starting on clozapine, and a third cohort includes 25 antipsychotic-naive patients. The patients in the second cohort and third cohort are prospectively assessed for up to 6 and 12 weeks, respectively. Phase II of this study will incorporate microbiota humanized mouse models to examine the influence of human fecal transplant on metabolic parameters and the gut-brain axis. Progress and Future Directions: We are underway with the first participants enrolled in all phase I treatment cohorts. This study will contribute to elucidating the role of the gut microbiome in schizophrenia and metabolic side effects. In addition, its results may help to explore potential therapeutic targets for antipsychotic-induced metabolic side effects.

Clinical implications of APOE genotyping for late-onset Alzheimer's disease (LOAD) risk estimation: a review of the literature.

Alzheimer's disease is a genetically complex neurodegenerative disorder representing the leading cause of dementia. Advances in personal genomics are increasing the public uptake of genetic susceptibility testing for complex diseases such as late-onset Alzheimer's disease (LOAD). For LOAD, the discovery of the major risk ε4 allele of the APOE gene has prompted a debate on the ethics and utility of presymptomatic (i.e., predictive) testing. Although the mechanistic contribution of APOE to disease onset remains uncertain, presymptomatic genetic testing provides a relative risk of developing LOAD. Presymptomatic testing for complex disorders, such as LOAD is much less conclusive than early-onset Alzheimer's disease (EOAD) which follows a Mendelian inheritance pattern. Given the lack of preventive strategies available for EOAD or LOAD, APOE genotyping offers limited clinical utility, thus, raising ethical and practical questions. We conducted a systematic search of five electronic databases or primary studies published during January 2008-January 2018 which investigated practical and ethical issues of presymptomatic APOE genotyping for LOAD risk estimation. We identified 31 articles which suggested that APOE genotyping for LOAD susceptibility provides potential benefits to at-risk patients and can guide changes in positive health-related behaviors. However, other individuals may experience test-related anxiety, depression and psychological distress. Future research should focus on developing an integrated risk assessment tool to enhance the utility of APOE genotyping. Furthermore, empirical research is required to understand actual psychological and social implications associated with testing.

Genetic Similarities between Compulsive Overeating and Addiction Phenotypes: A Case for "Food Addiction"?

There exists a continuous spectrum of overeating, where at the extremes there are casual overindulgences and at the other a 'pathological' drive to consume palatable foods. It has been proposed that pathological eating behaviors may be the result of addictive appetitive behavior and loss of ability to regulate the consumption of highly processed foods containing refined carbohydrates, fats, salt, and caffeine. In this review, we highlight the genetic similarities underlying substance addiction phenotypes and overeating compulsions seen in individuals with binge eating disorder. We relate these similarities to findings from neuroimaging studies on reward processing and clinical diagnostic criteria based on addiction phenotypes. The abundance of similarities between compulsive overeating and substance addictions puts forth a case for a 'food addiction' phenotype as a valid, diagnosable disorder.

The Effects of Video Games on Cognition and Brain Structure: Potential Implications for Neuropsychiatric Disorders.

Video games are now a ubiquitous form of entertainment that has occasionally attracted negative attention. Video games have also been used to test cognitive function, as therapeutic interventions for neuropsychiatric disorders, and to explore mechanisms of experience-dependent structural brain changes. Here, we review current research on video games published from January 2011 to April 2014 with a focus on studies relating to mental health, cognition, and brain imaging. Overall, there is evidence that specific types of video games can alter brain structure or improve certain aspects of cognitive functioning. Video games can also be useful as neuropsychological assessment tools. While research in this area is still at a very early stage, there are interesting results that encourage further work in this field, and hold promise for utilizing this technology as a powerful therapeutic and experimental tool.

CYP2D6 Phenotype Influences Pharmacokinetic Parameters of Venlafaxine: Results from a Population Pharmacokinetic Model in Older Adults with Depression.

In this study, we aimed to improve upon a published population pharmacokinetic (PK) model for venlafaxine (VEN) in the treatment of depression in older adults, then investigate whether CYP2D6 metabolizer status affected model-estimated PK parameters of VEN and its active metabolite O-desmethylvenlafaxine. The model included 325 participants from a clinical trial in which older adults with depression were treated with open-label VEN (maximum 300 mg/day) for 12 weeks and plasma levels of VEN and O-desmethylvenlafaxine were assessed at weeks 4 and 12. We fitted a nonlinear mixed-effect PK model using NONMEM to estimate PK parameters for VEN and O-desmethylvenlafaxine adjusted for CYP2D6 metabolizer status and age. At both lower doses (up to 150 mg/day) and higher doses (up to 300 mg/day), CYP2D6 metabolizers impacted PK model-estimated VEN clearance, VEN exposure, and active moiety (VEN + O-desmethylvenlafaxine) exposure. Specifically, compared with CYP2D6 normal metabolizers, (i) CYP2D6 ultra-rapid metabolizers had higher VEN clearance; (ii) CYP2D6 intermediate metabolizers had lower VEN clearance; (iii) CYP2D6 poor metabolizers had lower VEN clearance, higher VEN exposure, and higher active moiety exposure. Overall, our study showed that including a pharmacogenetic factor in a population PK model could increase model fit, and this improved model demonstrated how CYP2D6 metabolizer status affected VEN-related PK parameters, highlighting the importance of genetic factors in personalized medicine.

A cross-disease resource of living human microglia identifies disease-enriched subsets and tool compounds recapitulating microglial states.

Human microglia play a pivotal role in neurological diseases, but we still have an incomplete understanding of microglial heterogeneity, which limits the development of targeted therapies directly modulating their state or function. Here, we use single-cell RNA sequencing to profile 215,680 live human microglia from 74 donors across diverse neurological diseases and CNS regions. We observe a central divide between oxidative and heterocyclic metabolism and identify microglial subsets associated with antigen presentation, motility and proliferation. Specific subsets are enriched in susceptibility genes for neurodegenerative diseases or the disease-associated microglial signature. We validate subtypes in situ with an RNAscope-immunofluorescence pipeline and high-dimensional MERFISH. We also leverage our dataset as a classification resource, finding that induced pluripotent stem cell model systems capture substantial in vivo heterogeneity. Finally, we identify and validate compounds that recapitulate certain subtypes in vitro, including camptothecin, which downregulates the signature of disease-enriched subtypes and upregulates a signature previously associated with Alzheimer's disease.

User Experiences of Pharmacogenomic Testing and Opinions among Psychiatry Patients.

Pharmacogenomic testing (PGx) is a tool used to guide physicians in selecting an optimal medication for clients based on their genetic profile. The objective of this qualitative study is to understand patients' experiences with PGx testing as well as their opinions regarding the clinical adoption of such tests in psychiatry. A focus group was conducted to assess the needs of clients who had experience using a PGx test. Participants were recruited from a large study on PGx testing that offered physicians an opportunity to use PGx reports to guide psychotropic prescriptions. The focus group discussions were recorded, transcribed, and coded using NVivo to identify core themes. A total of 11 people participated in the focus group. Our analysis revealed that many participants were in favour of implementing PGx testing in psychiatric practice, and all expressed important considerations for patient-centred optimization of PGx testing. The main themes captured were: education and awareness among clinicians, cost considerations, PGx results-sharing and accessibility, and prospective benefits. The results of this study suggest that patients are keen to see PGx testing in widespread clinical care, but they report important opportunities to improve knowledge mobilization of PGx testing.

Exploratory genome-wide analyses of cortical inhibition, facilitation, and plasticity in late-life depression.

Late-life depression (LLD) is a heterogenous mood disorder influenced by genetic factors. Cortical physiological processes such as cortical inhibition, facilitation, and plasticity may be markers of illness that are more strongly associated with genetic factors than the clinical phenotype. Thus, exploring the relationship between genetic factors and these physiological processes may help to characterize the biological mechanisms underlying LLD and improve diagnosis and treatment selection. Transcranial magnetic stimulation (TMS) combined with electromyography was used to measure short interval intracortical inhibition (SICI), cortical silent period (CSP), intracortical facilitation (ICF), and paired associative stimulation (PAS) in 79 participants with LLD. We used exploratory genome-wide association and gene-based analyses to assess for genetic correlations of these TMS measures. MARK4 (which encodes microtubule affinity-regulating kinase 4) and PPP1R37 (which encodes protein phosphatase 1 regulatory subunit 37) showed genome-wide significant association with SICI. EGFLAM (which encodes EGF-like fibronectin type III and laminin G domain) showed genome-wide significant association with CSP. No genes met genome-wide significant association with ICF or PAS. We observed genetic influences on cortical inhibition in older adults with LLD. Replication with larger sample sizes, exploration of clinical phenotype subgroups, and functional analysis of relevant genotypes is warranted to better characterize genetic influences on cortical physiology in LLD. This work is needed to determine whether cortical inhibition may serve as a biomarker to improve diagnostic precision and guide treatment selection in LLD.

ABCB1 Gene Variants and Antidepressant Treatment Outcomes: A Systematic Review and Meta-Analysis Including Results from the CAN-BIND-1 Study.

The P-glycoprotein efflux pump, encoded by the ABCB1 gene, has been shown to alter concentrations of various antidepressants in the brain. In this study, we conducted a systematic review and meta-analysis to investigate the association between six ABCB1 single-nucleotide polymorphisms (SNPs; rs1045642, rs2032582, rs1128503, rs2032583, rs2235015, and rs2235040) and antidepressant treatment outcomes in individuals with major depressive disorder (MDD), including new data from the Canadian Biomarker and Integration Network for Depression (CAN-BIND-1) cohort. For the CAN-BIND-1 sample, we applied regression models to investigate the association between ABCB1 SNPs and antidepressant treatment response, remission, tolerability, and antidepressant serum levels. For the meta-analysis, we systematically summarized pharmacogenetic evidence of the association between ABCB1 SNPs and antidepressant treatment outcomes. Studies were included in the meta-analysis if they investigated at least one ABCB1 SNP in individuals with MDD treated with at least one antidepressant. We did not find a significant association between ABCB1 SNPs and antidepressant treatment outcomes in the CAN-BIND-1 sample. A total of 39 studies were included in the systematic review. In the meta-analysis, we observed a significant association between rs1128503 and treatment response (T vs. C-allele, odds ratio = 1.30, 95% confidence interval = 1.15-1.48, P value (adjusted) = 0.024, n = 2,526). We did not find associations among the six SNPs and treatment remission nor tolerability. Our findings provide limited evidence for an association between common ABCB1 SNPs and antidepressant outcomes, which do not support the implementation of ABCB1 genotyping to inform antidepressant treatment at this time. Future research, especially on rs1128503, is recommended.

Effects of CYP2C19 and CYP2D6 gene variants on escitalopram and aripiprazole treatment outcome and serum levels: results from the CAN-BIND 1 study.

Cytochrome P450 drug-metabolizing enzymes may contribute to interindividual differences in antidepressant outcomes. We investigated the effects of CYP2C19 and CYP2D6 gene variants on response, tolerability, and serum concentrations. Patients (N = 178) were treated with escitalopram (ESC) from weeks 0-8 (Phase I), and at week 8, either continued ESC if they were responders or were augmented with aripiprazole (ARI) if they were non-responders (<50% reduction in Montgomery-Åsberg Depression Rating Scale from baseline) for weeks 8-16 (Phase II). Our results showed that amongst patients on ESC-Only, CYP2C19 intermediate and poor metabolizers (IM + PMs), with reduced or null enzyme function, trended towards significantly lower symptom improvement during Phase II compared to normal metabolizers (NMs), which was not observed in ESC + ARI. We further showed that CYP2D6 NMs and IM + PMs had a higher likelihood of reporting a treatment-related central nervous system side effect in ESC-Only and ESC + ARI, respectively. The differences in the findings between ESC-Only and ESC + ARI may be due to the altered pharmacokinetics of ESC by ARI coadministration in ESC + ARI. We provided evidence for this postulation when we showed that in ESC-Only, CYP2C19 and CYP2D6 IM + PMs demonstrated significantly higher ESC concentrations at Weeks 10 and 16 compared to NMs. In contrast, ESC + ARI showed an association with CYP2C19 but not with CYP2D6 metabolizer group. Instead, ESC + ARI showed an association between CYP2D6 metabolizer group and ARI metabolite-to-drug ratio suggesting potential competition between ESC and ARI for CYP2D6. Our findings suggest that dosing based on CYP2C19 and CYP2D6 genotyping could improve safety and outcome in patients on ESC monotherapy.

Corrigendum: Potential genetic overlap between insomnia and sleep symptoms in major depressive disorder: A polygenic risk score analysis.

[This corrects the article DOI: 10.3389/fpsyt.2021.734077.].

Identifying the Common Genetic Basis of Antidepressant Response.

Background: Antidepressants are a first-line treatment for depression. However, only a third of individuals experience remission after the first treatment. Common genetic variation, in part, likely regulates antidepressant response, yet the success of previous genome-wide association studies has been limited by sample size. This study performs the largest genetic analysis of prospectively assessed antidepressant response in major depressive disorder to gain insight into the underlying biology and enable out-of-sample prediction. Methods: Genome-wide analysis of remission (n remit = 1852, n nonremit = 3299) and percentage improvement (n = 5218) was performed. Single nucleotide polymorphism-based heritability was estimated using genome-wide complex trait analysis. Genetic covariance with eight mental health phenotypes was estimated using polygenic scores/AVENGEME. Out-of-sample prediction of antidepressant response polygenic scores was assessed. Gene-level association analysis was performed using MAGMA and transcriptome-wide association study. Tissue, pathway, and drug binding enrichment were estimated using MAGMA. Results: Neither genome-wide association study identified genome-wide significant associations. Single nucleotide polymorphism-based heritability was significantly different from zero for remission (h 2 = 0.132, SE = 0.056) but not for percentage improvement (h 2 = -0.018, SE = 0.032). Better antidepressant response was negatively associated with genetic risk for schizophrenia and positively associated with genetic propensity for educational attainment. Leave-one-out validation of antidepressant response polygenic scores demonstrated significant evidence of out-of-sample prediction, though results varied in external cohorts. Gene-based analyses identified ETV4 and DHX8 as significantly associated with antidepressant response. Conclusions: This study demonstrates that antidepressant response is influenced by common genetic variation, has a genetic overlap schizophrenia and educational attainment, and provides a useful resource for future research. Larger sample sizes are required to attain the potential of genetics for understanding and predicting antidepressant response.

Genome-wide analysis suggests the importance of vascular processes and neuroinflammation in late-life antidepressant response.

Antidepressant outcomes in older adults with depression is poor, possibly because of comorbidities such as cerebrovascular disease. Therefore, we leveraged multiple genome-wide approaches to understand the genetic architecture of antidepressant response. Our sample included 307 older adults (≥60 years) with current major depression, treated with venlafaxine extended-release for 12 weeks. A standard genome-wide association study (GWAS) was conducted for post-treatment remission status, followed by in silico biological characterization of associated genes, as well as polygenic risk scoring for depression, neurodegenerative and cerebrovascular disease. The top-associated variants for remission status and percentage symptom improvement were PIEZO1 rs12597726 (OR = 0.33 [0.21, 0.51], p = 1.42 × 10-6) and intergenic rs6916777 (Beta = 14.03 [8.47, 19.59], p = 1.25 × 10-6), respectively. Pathway analysis revealed significant contributions from genes involved in the ubiquitin-proteasome system, which regulates intracellular protein degradation with has implications for inflammation, as well as atherosclerotic cardiovascular disease (n = 25 of 190 genes, p = 8.03 × 10-6, FDR-corrected p = 0.01). Given the polygenicity of complex outcomes such as antidepressant response, we also explored 11 polygenic risk scores associated with risk for Alzheimer's disease and stroke. Of the 11 scores, risk for cardioembolic stroke was the second-best predictor of non-remission, after being male (Accuracy = 0.70 [0.59, 0.79], Sensitivity = 0.72, Specificity = 0.67; p = 2.45 × 10-4). Although our findings did not reach genome-wide significance, they point to previously-implicated mechanisms and provide support for the roles of vascular and inflammatory pathways in LLD. Overall, significant enrichment of genes involved in protein degradation pathways that may be impaired, as well as the predictive capacity of risk for cardioembolic stroke, support a link between late-life depression remission and risk for vascular dysfunction.

Potential Genetic Overlap Between Insomnia and Sleep Symptoms in Major Depressive Disorder: A Polygenic Risk Score Analysis.

Background: The prevalence of insomnia and hypersomnia in depressed individuals is substantially higher than that found in the general population. Unfortunately, these concurrent sleep problems can have profound effects on the disease course. Although the full biology of sleep remains to be elucidated, a recent genome-wide association (GWAS) of insomnia, and other sleep traits in over 1 million individuals was recently published and provides many promising hits for genetics of insomnia in a population-based sample. Methods: Using data from the largest available GWAS of insomnia and other sleep traits, we sought to test if sleep variable PRS scores derived from population-based studies predicted sleep variables in samples of depressed cases [Psychiatric Genomics Consortium - Major Depressive Disorder subjects (PGC MDD)]. A leave-one-out analysis was performed to determine the effects that each individual study had on our results. Results: The only significant finding was for insomnia, where p-value threshold, p = 0.05 was associated with insomnia in our PGC MDD sample (R 2 = 1.75-3, p = 0.006). Conclusion: Our results reveal that <1% of variance is explained by the variants that cover the two significant p-value thresholds, which is in line with the fact that depression and insomnia are both polygenic disorders. To the best of our knowledge, this is the first study to investigate genetic overlap between the general population and a depression sample for insomnia, which has important treatment implications, such as leading to novel drug targets in future research efforts.

Validation study of microRNAs previously associated with antidepressant response in older adults treated for late-life depression with venlafaxine.

BACKGROUND: MicroRNAs (miRNAs) are small 22 nucleotides long, non-coding RNAs that are potential biomarkers for antidepressant treatment response. We aimed to replicate previous associations of miRNAs with antidepressant treatment response in a sample of older adults diagnosed with late-life depression. METHODS: Our sample included 184 older adults diagnosed with moderately severe depression that received open-label venlafaxine (up to 300 mg/day) for approximately 12 weeks. We quantified miRNA expression levels at baseline and week 12 for miRNAs miR-1202, miR-135a-5p, miR-16-5p, miR-146a-5p, miR-146b-5p, miR-425-3p, and miR-24-3p to explore their association with remission status, response trajectories, and time-to-remission. RESULTS: At T0 and T12, there were no differences in miRNA expression levels between remitters and non-remitters. However, remitters showed a trend toward higher baseline miR-135a-5p (Median = 11.3 [9.9, 15.7], p = .083). Prior to correction, baseline miR-135a-5p expression levels showed an association with remission status (OR = 1.8 [1.0, 3.3], p = .037). Individuals with higher baseline miR-135a-5p showed better response trajectories (F = 4.5, FDR-corrected p = 4.4 × 10-4), particularly at weeks 10 and 12 (p < .05). In addition, individuals with higher miR-135a-5p expression reached remission faster than those with lower expression (HR = 0.6 [0.4, 0.9], FDR-corrected p = .055). LIMITATIONS: Although the sample size was relatively modest, our findings are consistent with the literature suggesting that higher miR-135a-5p levels may be associated with better antidepressant treatment response. CONCLUSIONS: However, the miRNA signature of antidepressant response in older adults may be different as compared to younger adults.

GWAS-based machine learning approach to predict duloxetine response in major depressive disorder.

Major depressive disorder (MDD) is one of the most prevalent psychiatric disorders and is commonly treated with antidepressant drugs. However, large variability is observed in terms of response to antidepressants. Machine learning (ML) models may be useful to predict treatment outcomes. A sample of 186 MDD patients received treatment with duloxetine for up to 8 weeks were categorized as "responders" based on a MADRS change >50% from baseline; or "remitters" based on a MADRS score ≤10 at end point. The initial dataset (N = 186) was randomly divided into training and test sets in a nested 5-fold cross-validation, where 80% was used as a training set and 20% made up five independent test sets. We performed genome-wide logistic regression to identify potentially significant variants related to duloxetine response/remission and extracted the most promising predictors using LASSO regression. Subsequently, classification-regression trees (CRT) and support vector machines (SVM) were applied to construct models, using ten-fold cross-validation. With regards to response, none of the pairs performed significantly better than chance (accuracy p > .1). For remission, SVM achieved moderate performance with an accuracy = 0.52, a sensitivity = 0.58, and a specificity = 0.46, and 0.51 for all coefficients for CRT. The best performing SVM fold was characterized by an accuracy = 0.66 (p = .071), sensitivity = 0.70 and a sensitivity = 0.61. In this study, the potential of using GWAS data to predict duloxetine outcomes was examined using ML models. The models were characterized by a promising sensitivity, but specificity remained moderate at best. The inclusion of additional non-genetic variables to create integrated models may improve prediction.