Roy H Perlis - US grants

DESCRIPTION (provided by applicant): One-third or more of individuals treated for major depressive disorder (MDD) do not experience remission of symptoms despite at least two adequate antidepressant trials. Such treatment-resistant depression (TRD) contributes disproportionately to the tremendous costs of MDD, in terms of health care costs, functional impairment, and diminished quality of life. The promise of personalized medicine for individuals at high risk for TRD is apparent. If these individuals could be recognized early in their disease course, they could be triaged to more intensive or targeted interventions to improve their likelihood of remission. For example, they might receive earlier addition of cognitive-behavioral therapy, earlier use of combination medication treatments, or earlier referral for electroconvulsive therapy. With the proliferation of treatment options in MDD, individuals can spend months or years in and out of treatment before receiving these next-step treatments. Moreover, the ability to identify these individuals would facilitate the development of new personalized interventions: rather than the requiring multiple failed prospective trials, high-risk individuals could immediately be offered study participation. At present, there are two primary obstacles to translating personalized medicine into clinical practice. First, no large and generalizable cohorts have been collected in which to build risk models. Second, no validation cohorts exist to demonstrate that such models perform well in clinical settings. The present study proposes to address these two obstacles directly. Previous investigations, including work in the large multicenter Systematic Treatment Alternatives to Relieve Depression (STAR*D) study, have identified putative clinical or genetic predictors of treatment response. However, in the absence of replication, such associations are hypothesis-generating at best. An ongoing study will collect data from 1,000 individuals treated in a New England health system for whom prospective treatment outcomes are available (the Dep1 cohort), including 500 individuals with TRD and 500 with SSRI-responsive MDD, with completion of a genome wide association study expected by spring 2009. The proposed study will first use cutting-edge modeling techniques to construct and cross-validate models of TRD using sociodemographic, clinical, and genetic predictors in the existing Dep1 cohort. In parallel, it will collect an additional 1,000 MDD subjects with 6-month treatment outcomes from the same health system. This second cohort (Dep2) will be used to validate the TRD risk stratification model. To identify these patient cohorts, this study will take advantage of computerized administrative data systems, data-mining, and natural language processing techniques that have been successfully applied to support population-based research. This approach allows identification of clinical features, such as comorbidities, medication treatments, as well as longitudinal outcomes, based on claims, pharmacy data, and medical records. The resulting patient data is far more representative of clinical populations, and far less expensive to generate, than that which could be obtained using more traditional approaches. Therefore, beyond facilitating personalized treatment of MDD, the proposed study would establish the methodology for using large clinical populations to personalize treatment in psychiatry as a whole. Public Health Relevance: A third or more of people with major depression do not get well despite two or more different treatments, and identifying these people early in treatment might allow more personalized approaches with greater chances of success. This study will use statistical techniques to try to predict who is at risk for this treatment- resistant depression, based on clinical differences and genetic variations. Then, it will examine a second group of patients to see how well this technique might work if it is applied in a large health system.

DESCRIPTION (provided by applicant): Major depressive episodes contribute substantially to morbidity and mortality in bipolar disorder. While multiple medications have demonstrated efficacy for treating these episodes, a majority of patients do not reach complete symptomatic remission or have difficulty tolerating these medications. Studies of the mechanism of action of treatments known to be effective in bipolar may provide new treatment targets. Work by our group and others strongly implicates Wnt/GSK3 signaling in the mechanism of action of lithium, which remains a first-line treatment for bipolar depression as well as prevention of recurrence. We therefore have utilized high-throughput cell-based screening in neuronal cells to identify compounds that showed potential additivity or synergy with lithium in terms of effects on Wnt/GSK3 signaling. Among the active F.D.A.-approved drugs with safety profiles compatible with long-term use, we identified multiple statins that acted synergistically with Wnt3a treatment and show further additivity with lithium treatment, including simvastatin, one of the most potent statins known to be capable of crossing the blood-brain barrier. We have validated 3-hydroxy-3methylglutaryl coenzyme A (HMG-CoA) reductase as the relevant target. Statins have not been directly examined in the treatment of bipolar disorder. A recent rodent study found evidence that a statin augmented the antidepressant-like effects with fluoxetine. Intriguingly, multiple population-based studies also suggest that statins may be associated with a statistically significant decrease in depressive symptoms, and a decrease in the likelihood of adverse psychiatric outcomes. Another study examining statin treatment of dementia indicated a decrease in depressive symptoms compared to placebo. We now propose to conduct a randomized, double-blind, placebo-controlled, proof-of-concept investigation of simvastatin as add-on treatment to lithium in outpatients with bipolar I disorder in a major depressive episode. In parallel, we will collect fibroblasts and derive induced pluripotent stem cells (iPSCs) and neuronal progenitor (NP) cells. The function of the Wnt signaling pathway in patient-specific iPSC-derived NP cells will then be quantified in cell-based assays, with and without treatment with lithium and simvastatin, to enable examination of the association between Wnt/GSK3 signalling and magnitude of improvement in depressive symptoms. These experiments are expected to serve as a crucial first step in the development of new bipolar pharmacotherapies. By predicting the benefit of an adjunctive therapy for bipolar disorder based upon its effects on Wnt/GSK3 signaling pathway and attempting to correlate these responses using patient-specific neuronal cell models, our proposed study will provide a critical test of the importance of Wnt/GSK3 signaling in regulating neuroplasticity in bipolar disorder and depression. At a minimum, these studies will also provide a well-phenotyped, patient-derived cellular resource for future investigation of lithium response and bipolar disorder that can be applied in future studies toward high-throughput screening for lithium-like drugs.

PROJECT 3 ABSTRACT: Major depressive disorder (MDD) remains one of the most prevalent and costly medical disorders. A third or more of patients may not achieve symptomatic remission despite multiple medication treatments; many other individuals are simply unable or unwilling to initiate prescription pharmacologic or psychosocial treatment. Complementary and alternative medications (CAM) represent an important option for such patients. In addition to understanding which CAM antidepressant strategies are efficacious for depression as well as for stress, the aim of Project 1, it would be highly valuable to understand the mechanisms by which certain CAM treatments exert their therapeutic effect. This understanding could increase the acceptability of current treatments, allow better matching of patients with effective treatments, and facilitate the investigation and development of novel CAM treatments. For standard antidepressant treatments, multiple hypotheses regarding mechanisms of action have been developed. These include (i) promotion of neuroplasticity, (ii) modulation of inflammation, and (iii) promotion of neurogenesis. To date, investigation of these latter hypotheses has been hampered by a lack of direct models of human neurobiology ? and particularly neuropathology - amenable to rapid screening and quantitative functional assessment. That is, it has not been possible to examine whether these hypotheses are supported in neural tissue from patients with the particular disease targeted by these interventions. Progress in stem cell technology and developmental neurobiology allows a novel strategy that forms the focus of Project 3. Dermal fibroblasts from 180 patient participants in the randomized trial of Project One will be reprogrammed (transdifferentiated) to induced neurons (iNs). Transcriptional profiles of these iN's will be compared after exposure to n-3 fatty acids, S- adenosyl L-methionine (SAMe), or vehicle to test whether the two CAMs regulate genes related to neuroplasticity (Aim 1a), and whether degree of modulation of neuroplasticity is associated with treatment efficacy. In parallel, a subset (10 per treatment arm) of patient-derived fibroblasts will be reprogrammed to induced pluripotent stem cells. These cell lines will then be differentiated into neuronal precursors and ultimately to mature neurons. Work by this group and others indicates that it is possible to generate such cells and incorporate them in high-throughput, quantitative functional assays to characterize phenotypes relevant to antidepressant mechanism. Specifically, the hypothesis that n-3 fatty acids and SAMe modulate inflammatory markers on neural-lineage cells (Aim 2a), and promote neurogenesis (Aim 2b), will be tested using validated assays. The hypothesis that these mechanisms are associated with treatment efficacy will also be tested. In addition to examining these primary hypotheses, this project will establish a critical resource for future investigation of CAM compounds, a biobank of 180 fibroblasts and 30 pluripotent stem cells and neuronal precursor cells, all derived from patients with MDD participating in Project 1's placebo-controlled investigations. NARRATIVE Major depressive disorder is a major contributor to morbidity worldwide, and existing treatments fail to yield symptomatic remission in ~1/3 of patients. While Complementary and Alternative Medicine (CAM) compounds are increasingly used to treat depression, and well-accepted by patients, their mechanisms of effect have not been fully characterized. The proposed investigation will test specific hypotheses extending preliminary data using unique patient-derived stem cells and induced neurons, while at the same timing establishing biomarkers, biosignatures, and a biobank to facilitate future CAM studies.

? DESCRIPTION (provided by applicant): Convergent genetic and epidemiologic evidence suggests the importance of understanding psychiatric illness from a dimensional rather than solely a categorical perspective. The limitations of traditional diagnostic categories motivated a major NIMH-supported effort to identify measures of psychopathology that more closely align with underlying disease biology. At present, however, the available large clinical data sets, whether health claims, registries, or electronic health records, do not include such dimensional measures. Even with the integration of structure clinician and patient-reported outcomes, generating such cohorts could require a decade or more. Moreover, coded data does not systematically capture clinically-important concepts such as health behaviors or stressors. While such cohorts are developed, natural language processing can facilitate the application of existing electronic health records to enable precision medicine in psychiatry. Specifically, while traditional natural language tools focus on extracting individual terms, emerging methods including those in development by the investigators allow extraction of concepts and dimensions. The present investigation proposes to develop a toolkit for natural language processing of narrative patient notes to extract measures of psychopathology, including estimated RDoC domains. In preliminary investigations in a large health system, these tools have demonstrated both face validity and predictive validity. This toolkit also allows extraction o complex concepts from narrative notes, such as stressors and health behaviors. In the proposed study, these natural language processing tools will be applied to a large psychiatric inpatient data set as well as a large general medical inpatient data set, to derive measures of psychopathology and other topics. The resulting measures will then be used in combination with coded data to build regression and machine-learning-based models to predict clinical outcomes including length of hospital stay and risk of readmission. The models will then be validated in independent clinical cohorts. By combining expertise in longitudinal clinical investigation, natural language processing, and machine learning, the proposed study brings together a team with the needed skills to develop a critical toolkit for understanding health records dimensionally The resulting models can be applied to facilitate investigation of dimensions of psychopathology and related topics, allowing stratification of clinical risk to enable development of targeted interventions.

Schizophrenia (SCZ) and major depressive disorder (MDD) are highly heritable, debilitating diseases with lifetime prevalences of ~1% and 15%, respectively. Both disorders carry substantial morbidity and mortality and are associated with severe societal and personal costs. Despite the availability of efficacious treatments for both disorders, ~1/3 of individuals will not achieve symptomatic improvement even after multiple rounds of medication. Identifying individuals at greater risk for such treatment nonresponse, or treatment resistance, could facilitate more targeted interventions for these individuals. A burgeoning literature has identified genomic variation associated with treatment response. IN particular, antidepressant response has been suggested to be highly heritable; convergent data from rodent studies likewise suggest that antipsychotic and antidepressant response phenotypes are influenced by genetic variation. However, treatment studies to date have had minimal success in identifying variants associated with psychotropic response, likely as a result of limited sample sizes: prior efforts required sequential treatment trials and prospective assessment to characterize outcomes. Longitudinal electronic health records (EHR) data provide an opportunity to efficiently characterize treatment response in many individuals in real-world settings. Coupled with large and expanding biobanks, these cohorts allow for low- cost, large-scale genomic studies that finally achieve sufficient power to detect realistic effect sizes. The investigators now propose to apply these approaches to the EHRs of two large regional health systems, each linked to a large biobank, to investigate treatment resistance in SCZ and MDD. They will apply canonical indicators of treatment resistance - clozapine treatment for SCZ, and electroconvulsive therapy (ECT) for MDD - to identify coded and uncoded clinical features associated with high probability of treatment resistance in EHR data. These predictors will themselves provide a useful baseline for identifying high risk individuals. Then, they will apply these to study the entire affected population of each biobank, extending existing genomic data with additional genome-wide association, yielding more than 26,000 antidepressant-treated individuals and 2,500 antipsychotic-treated individuals. Rather than simply conducting a case-control study, they will examine treatment resistance as a quantitative trait, applying a method developed by the investigators and shown to substantially increase power for such traits. The project combines expertise in clinical informatics, machine learning, and analysis of large scale genomics, as well as domain-specific expertise in psychiatric treatment resistance. Spanning two distinct health systems, the algorithms and methods developed have maximal portability, facilitating next- step investigations. Successful identification of risk variants will facilitate efforts at clinical risk stratification as well as investigation of the biology underlying treatment resistance.

Abstract Major depressive disorder contributes substantially to morbidity, mortality, and health care cost. Standard treatments are ineffective for up to a third of patients, so new treatment options are needed along with strategies to make more effective use of existing treatments. However, progress in expanding therapeutic options has been hindered by heterogeneity in clinical presentation and course of depression. In other disorders such as inflammatory bowel disease, cancer, and dementia, identifying disease subtypes has led to therapeutic discoveries. In major depressive disorder, efforts to identify subtypes based on clinical observation have yielded limited success, primarily because of the lack of availability of adequate cohorts for replication, and because those features most apparent to clinicians may not be the most relevant for differentiating subgroups. Efforts to leverage large electronic health record data sets for subtyping address some of these challenges, but standard approaches may not yield human-interpretable features nor those with value in prediction. The investigators have developed methods for engineering features that balance utility in prediction with interpretability. Preliminary work by the investigators during a year of R56 support yielding 4 publications demonstrates that this approach indeed yields coherent topics without sacrificing predictive validity; electronic health records contain meaningful data that facilitates identification of interpretable patient subgroups. The present study draws on very large cohorts of individuals with major depression, defined by a validated algorithm, in electronic health records from two health systems. It will first apply methods developed by the investigators to identify MDD subtypes. These subtypes will then be examined in terms of predictive validity as well as interpretability by clinicians. The study builds on a productive collaboration between a team experienced in mood disorder phenotyping and clinical investigation, analysis of large-scale longitudinal electronic health records, and development and application of innovative methods in machine learning that yield interpretable models rather than black boxes. Data-driven disease subtyping will facilitate clinically useful risk stratification as well as biological study of mood disorders.