Adam M. Chekroud

Cross-trial prediction of treatment outcome in depression: a machine learning approach

BACKGROUND Antidepressant treatment efficacy is low, but might be improved by matching patients to interventions. At present, clinicians have no empirically validated mechanisms to assess whether a patient with depression will respond to a specific antidepressant. We aimed to develop an algorithm to assess whether patients will achieve symptomatic remission from a 12-week course of citalopram. METHODS We used patient-reported data from patients with depression (n=4041, with 1949 completers) from level 1 of the Sequenced Treatment Alternatives to Relieve Depression (STAR*D; ClinicalTrials.gov, number NCT00021528) to identify variables that were most predictive of treatment outcome, and used these variables to train a machine-learning model to predict clinical remission. We externally validated the model in the escitalopram treatment group (n=151) of an independent clinical trial (Combining Medications to Enhance Depression Outcomes [COMED]; ClinicalTrials.gov, number NCT00590863). FINDINGS We identified 25 variables that were most predictive of treatment outcome from 164 patient-reportable variables, and used these to train the model. The model was internally cross-validated, and predicted outcomes in the STAR*D cohort with accuracy significantly above chance (64·6% [SD 3·2]; p<0·0001). The model was externally validated in the escitalopram treatment group (N=151) of COMED (accuracy 59·6%, p=0.043). The model also performed significantly above chance in a combined escitalopram-buproprion treatment group in COMED (n=134; accuracy 59·7%, p=0·023), but not in a combined venlafaxine-mirtazapine group (n=140; accuracy 51·4%, p=0·53), suggesting specificity of the model to underlying mechanisms. INTERPRETATION Building statistical models by mining existing clinical trial data can enable prospective identification of patients who are likely to respond to a specific antidepressant. FUNDING Yale University.

Reevaluating the Efficacy and Predictability of Antidepressant Treatments: A Symptom Clustering Approach

Importance Depressive severity is typically measured according to total scores on questionnaires that include a diverse range of symptoms despite convincing evidence that depression is not a unitary construct. When evaluated according to aggregate measurements, treatment efficacy is generally modest and differences in efficacy between antidepressant therapies are small. Objectives To determine the efficacy of antidepressant treatments on empirically defined groups of symptoms and examine the replicability of these groups. Design, Setting, and Participants Patient-reported data on patients with depression from the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) trial (n = 4039) were used to identify clusters of symptoms in a depressive symptom checklist. The findings were then replicated using the Combining Medications to Enhance Depression Outcomes (CO-MED) trial (n = 640). Mixed-effects regression analysis was then performed to determine whether observed symptom clusters have differential response trajectories using intent-to-treat data from both trials (n = 4706) along with 7 additional placebo and active-comparator phase 3 trials of duloxetine (n = 2515). Finally, outcomes for each cluster were estimated separately using machine-learning approaches. The study was conducted from October 28, 2014, to May 19, 2016. Main Outcomes and Measures Twelve items from the self-reported Quick Inventory of Depressive Symptomatology (QIDS-SR) scale and 14 items from the clinician-rated Hamilton Depression (HAM-D) rating scale. Higher scores on the measures indicate greater severity of the symptoms. Results Of the 4706 patients included in the first analysis, 1722 (36.6%) were male; mean (SD) age was 41.2 (13.3) years. Of the 2515 patients included in the second analysis, 855 (34.0%) were male; mean age was 42.65 (12.17) years. Three symptom clusters in the QIDS-SR scale were identified at baseline in STAR*D. This 3-cluster solution was replicated in CO-MED and was similar for the HAM-D scale. Antidepressants in general (8 of 9 treatments) were more effective for core emotional symptoms than for sleep or atypical symptoms. Differences in efficacy between drugs were often greater than the difference in efficacy between treatments and placebo. For example, high-dose duloxetine outperformed escitalopram in treating core emotional symptoms (effect size, 2.3 HAM-D points during 8 weeks, 95% CI, 1.6 to 3.1; P < .001), but escitalopram was not significantly different from placebo (effect size, 0.03 HAM-D points; 95% CI, −0.7 to 0.8; P = .94). Conclusions and Relevance Two common checklists used to measure depressive severity can produce statistically reliable clusters of symptoms. These clusters differ in their responsiveness to treatment both within and across different antidepressant medications. Selecting the best drug for a given cluster may have a bigger benefit than that gained by use of an active compound vs a placebo.

Multisite prediction of 4-week and 52-week treatment outcomes in patients with first-episode psychosis: a machine learning approach

BACKGROUND At present, no tools exist to estimate objectively the risk of poor treatment outcomes in patients with first-episode psychosis. Such tools could improve treatment by informing clinical decision-making before the commencement of treatment. We tested whether such a tool could be successfully built and validated using routinely available, patient-reportable information. METHODS By applying machine learning to data from 334 patients in the European First Episode Schizophrenia Trial (EUFEST; International Clinical Trials Registry Platform number, ISRCTN68736636), we developed a tool to predict poor versus good treatment outcome (Global Assessment of Functioning [GAF] score ≥65 vs GAF <65, respectively) after 4 weeks and 52 weeks of treatment. To enable the unbiased estimation of the predictive systems generalisability to new patients, we used repeated nested cross-validation to prevent information leaking between patients used for training and validating the models. In pursuit of everyday clinical applicability, we retrained the 4-week outcome predictor with only the top ten predictors of the pooled prediction system and then tested this tool in 108 independent patients with 4-week outcome labels. Discontinuation and readmission to hospital events in patients with predicted poor versus good outcomes were assessed with Kaplan-Meier log-rank analyses, whereas generalised linear mixed-effects models were used to investigate the GAF-based predictions against several clinically meaningful outcome indicators, including treatment adherence, symptom remission, and quality of life. FINDINGS The generalisability of our outcome predictions were estimated with cross-validation (test-fold balanced accuracy [BAC] of 75·0% for 4-week outcomes and 73·8% for and 52-week outcomes), and leave-site-out validation across 44 European sites (BAC of 72·1% for 4-week outcomes and 71·1% for 52-week outcomes). We identified a smaller group of ten predictors still providing a BAC of 71·7% in 108 patients never used for model discovery. Unemployment, poor education, functional deficits, and unmet psychosocial needs predicted both endpoints, whereas previous depressive episodes, male sex, and suicidality additionally predicted poor 1-year outcomes. 52-week predictions identified patients at risk for symptom persistence, non-adherence to treatment, readmission to hospital and poor quality of life. Specifically among these patients, amisulpride and olanzapine showed superior efficacy versus haloperidol, quetiapine, and ziprasidone. INTERPRETATION Our results suggest that prognostic models operating on brief, patient-reportable pre-treatment data might provide useful insight into individualised outcome trajectories, optimising treatment selection, and targeted clinical trial designs. To embed these tools into real-world care, replication is needed in external first-episode samples with overlapping variables, which are not available in the field at present. FUNDING The European Group for Research in Schizophrenia.

Patterns in the human brain mosaic discriminate males from females

In their PNAS article, Joel et al. (1) demonstrate extensive overlap between the distributions of females and males for many brain characteristics, measured across multiple neuroimaging modalities and datasets. They pose two requirements for categorizing brains into distinct male/female classes: ( i ) gender differences should appear as dimorphic form differences between male and female brains, and ( ii ) there should be internal consistency in the degree of “maleness–femaleness” of different elements within a single brain. Based on these criteria, the authors convincingly establish that there is little evidence for this strict sexually dimorphic view of human brains, counter to the popular lay conception of a “male” and “female” brain. This … [↵][1]2To whom correspondence should be addressed. Email: adam.chekroud{at}yale.edu. [1]: #xref-corresp-1-1

Trajectories of relapse in randomised, placebo-controlled trials of treatment discontinuation in major depressive disorder: an individual patient-level data meta-analysis

Background Understanding patterns of relapse in antidepressant treatment responders can inform strategies for preventing relapse. Methods We re-analyzed individual-patient data from four double-blind discontinuation clinical trials of duloxetine or fluoxetine vs. placebo in major depression (N=1462). Trajectories of depression severity (Hamilton Depression Rating Scale scores) were identified in the entire sample, and separately in arms where antidepressant had been continued or discontinued. Predictors of trajectory membership were assessed. Findings We identified similar “relapse” trajectories and two trajectories of stable depression scores in the normal range on active medication and on placebo. Active treatment (OR=0.47, 95% CI: (0.37, 0.61)) significantly lowered the odds of membership in the “relapse” trajectory whereas female sex (OR=1.56, 95% CI: (1.23, 2.06)), shorter length of time with clinical response (OR=1.10, 95% CI: (1.06, 1.15)) and higher Clinical Global Impressions score at baseline (OR=1.28, 95% CI: (1.01, 1.62)) increased the odds. Overall, the protective effect of antidepressant medication relative to placebo on the risk of being classified as a relapser was about 13% (46% vs. 33%). Interpretation The existence of similar relapse trajectories on active medication and on placebo suggests that there is no specific relapse signature associated with antidepressant discontinuation. Furthermore, continued treatment offers only a modest protection against relapse. These data highlight the need for incorporating treatment strategies that prevent relapse as part of the treatment of depression.

Personalised pharmacotherapy: an interim solution for antidepressant treatment?

Healy’s engaging historical perspective on the rise of selective serotonin reuptake inhibitor (SSRI) antidepressants questions the efficacy and biological plausibility of these drugs.1 However, his focus on low serotonin levels is irrelevant to whether SSRIs are effective treatments for depression. Importantly, the response to SSRIs seems to be heterogeneous rather than universally poor. Unbiased trajectory based analysis of more than 2500 patients treated with SSRIs or placebo indicated that most patients (>75%) given an SSRI showed a superior response to those given placebo. However, nearly a quarter of …

Association between physical exercise and mental health in 1·2 million individuals in the USA between 2011 and 2015: a cross-sectional study

BACKGROUND Exercise is known to be associated with reduced risk of all-cause mortality, cardiovascular disease, stroke, and diabetes, but its association with mental health remains unclear. We aimed to examine the association between exercise and mental health burden in a large sample, and to better understand the influence of exercise type, frequency, duration, and intensity. METHODS In this cross-sectional study, we analysed data from 1 237 194 people aged 18 years or older in the USA from the 2011, 2013, and 2015 Centers for Disease Control and Prevention Behavioral Risk Factors Surveillance System survey. We compared the number of days of bad self-reported mental health between individuals who exercised and those who did not, using an exact non-parametric matching procedure to balance the two groups in terms of age, race, gender, marital status, income, education level, body-mass index category, self-reported physical health, and previous diagnosis of depression. We examined the effects of exercise type, duration, frequency, and intensity using regression methods adjusted for potential confounders, and did multiple sensitivity analyses. FINDINGS Individuals who exercised had 1·49 (43·2%) fewer days of poor mental health in the past month than individuals who did not exercise but were otherwise matched for several physical and sociodemographic characteristics (W=7·42 × 1010, p<2·2 × 10-16). All exercise types were associated with a lower mental health burden (minimum reduction of 11·8% and maximum reduction of 22·3%) than not exercising (p<2·2 × 10-16 for all exercise types). The largest associations were seen for popular team sports (22·3% lower), cycling (21·6% lower), and aerobic and gym activities (20·1% lower), as well as durations of 45 min and frequencies of three to five times per week. INTERPRETATION In a large US sample, physical exercise was significantly and meaningfully associated with self-reported mental health burden in the past month. More exercise was not always better. Differences as a function of exercise were large relative to other demographic variables such as education and income. Specific types, durations, and frequencies of exercise might be more effective clinical targets than others for reducing mental health burden, and merit interventional study. FUNDING Cloud computing resources were provided by Microsoft.

The Role of microRNA Expression in Cortical Development During Conversion to Psychosis

In a recent report of the North American Prodrome Longitudinal Study (NAPLS), clinical high-risk individuals who converted to psychosis showed a steeper rate of cortical gray matter reduction compared with non-converters and healthy controls, and the rate of cortical thinning was correlated with levels of proinflammatory cytokines at baseline. These findings suggest a critical role for microglia, the resident macrophages in the brain, in perturbations of cortical maturation processes associated with onset of psychosis. Elucidating gene expression pathways promoting microglial action prior to disease onset would inform potential preventative intervention targets. Here we used a forward stepwise regression algorithm to build a classifier of baseline microRNA expression in peripheral leukocytes associated with annualized rate of cortical thinning in a subsample of the NAPLS cohort (N=74). Our cortical thinning classifier included nine microRNAs, p=3.63 × 10-08, R2=0.358, permutation-based p=0.039, the gene targets of which were enriched for intracellular signaling pathways that are important to coordinating inflammatory responses within immune cells (p<0.05, Benjamini–Hochberg corrected). The classifier was also related to proinflammatory cytokine levels in serum (p=0.038). Furthermore, miRNAs that predicted conversion status were found to do so in a manner partially mediated by rate of cortical thinning (point estimate=0.078 (95% CIs: 0.003, 0.168), p=0.03). Many of the miRNAs identified here have been previously implicated in brain development, synaptic plasticity, immune function and/or schizophrenia, showing some convergence across studies and methodologies. Altered intracellular signaling within the immune system may interact with cortical maturation in individuals at high risk for schizophrenia promoting disease onset.

The perilous path from publication to practice

Precision medicine has long been heralded as the future of clinical practice, discussed at least since the 1940s by the likes of Paul Meehl and Claude Bernard. Fundamentally, precision medicine seeks to shift focus away from the average effectiveness of a given treatment to identifying the optimal treatment for an individual patient, through understanding features of the phenotype that inform the individual’s response to particular intervention. However, even once we have evidence that a particular feature or group of features might guide treatment, we remain a disappointingly long way away from realizing this potential. In this Perspective, we would like to draw the readers’ attention to three kinds of obstacles between publication and practice.

Multivariate Pattern Analysis of Genotype–Phenotype Relationships in Schizophrenia

Genetic risk variants for schizophrenia have been linked to many related clinical and biological phenotypes with the hopes of delineating how individual variation across thousands of variants corresponds to the clinical and etiologic heterogeneity within schizophrenia. This has primarily been done using risk score profiling, which aggregates effects across all variants into a single predictor. While effective, this method lacks flexibility in certain domains: risk scores cannot capture nonlinear effects and do not employ any variable selection. We used random forest, an algorithm with this flexibility designed to maximize predictive power, to predict 6 cognitive endophenotypes in a combined sample of psychiatric patients and controls (N = 739) using 77 genetic variants strongly associated with schizophrenia. Tenfold cross-validation was applied to the discovery sample and models were externally validated in an independent sample of similar ancestry (N = 336). Linear approaches, including linear regression and task-specific polygenic risk scores, were employed for comparison. Random forest models for processing speed (P = .019) and visual memory (P = .036) and risk scores developed for verbal (P = .042) and working memory (P = .037) successfully generalized to an independent sample with similar predictive strength and error. As such, we suggest that both methods may be useful for mapping a limited set of predetermined, disease-associated SNPs to related phenotypes. Incorporating random forest and other more flexible algorithms into genotype-phenotype mapping inquiries could contribute to parsing heterogeneity within schizophrenia; such algorithms can perform as well as standard methods and can capture a more comprehensive set of potential relationships.