Genome-wide association study for depression susceptibility genes and associations with brain structural changes in major depressive disorder
Principal Investigators: Roy H. Perlis, MD, MSc1,2, Jordan W. Smoller, MD1,2 and Dan V. Iosifescu, MD, MSc1
1 Department of Psychiatry, Massachusetts General Hospital
2 Center for Human Genetic Research, Massachusetts General Hospital
What questions did we ask in this DBP?
Treatment-resistant major depressive disorder (TRD), in which patients fail to achieve recovery despite at least one antidepressant treatment trial, is prevalent in clinical populations. A major multi-center study supported by the NIMH found that, of nearly 4,000 patients, around a third failed to reach recovery in spite of multiple carefully adjusted antidepressant treatment trials. TRD contributes disproportionately to the morbidity and mortality, as well as the high costs to society, associated with major depressive disorder. Unfortunately, TRD is difficult to study in randomized, controlled trials, which may be prohibitively expensive and require many years to complete.
The central question in this DBP is, can we use data mining techniques applied to longitudinal medical records to rapidly identify a cohort of patients with TRD, and a matched cohort of patients with major depressive disorder (MDD) responsive to standard antidepressants (selective serotonin reuptake inhibitors, or SSRIs)? Can we then use these cohorts to conduct a genetic association study to identify genes conferring risk for TRD, and build prediction systems to stratify individuals’ risk for TRD?
The pathophysiology of TRD is very poorly understood. One intriguing and potentially important finding from previous studies is that many depression patients have white matter abnormalities on MRI, which may be associated with treatment-resistance. Therefore, a second key question is, can we use similar data mining techniques to identify existing magnetic resonance images (MRIs) from MDD patients treated long-term with SSRIs and MDD subjects treated with non-SSRI antidepressants? Can we compare MRIs in these groups to detect whether long-term treatment with SSRIs is associated with increased white matter lesions? Can we also use genome-wide associations to detect genetic moderators of white matter abnormalities in depressed subjects?
How did i2b2 help us answer these questions?
The i2b2 team provided the technical expertise to identify, process, and mine the electronic medical records. In addition to standard tools, the team developed natural language processing (NLP) approaches to examine unstructured text fields in the medical record. For the depressed subjects thus selected the i2b2 team collaborated with the NAMIC NCBC to provide the tools to extract existing MRIs already acquired in the process of clinical treatment, as well as from age- and gender-matched healthy volunteers. A statistics-based selection algorithm was developed that used both structured and NLP-derived attributes to define the final cohort. Finally, the team will apply its expertise to assist in analysis of the genetic association study, using both standard and novel techniques.
What tools were developed from our work that will be of value to others?
NLP tools were developed to establish psychiatric diagnosis, treatment history and treatment outcomes from a longitudinal data set (HiTEX suite). The statistics-based selection algorithm (adaptive Lasso) can now be applied to other large health care system data sets, and facilitate the identification of other cohorts for examining treatment response in psychiatry. We also developed tools to analyze pooled existing MRI data. As part of this process we developed methods to analyze MRI data quality, to convert into a single standard MRI data acquired with multiple protocols. We also validated the feasibility of image analysis on pools of legacy data and the appropriate statistical analysis of such data. Given that many research groups have valuable existing MRI data, the tools developed as part of this project could be of large interest for the scientific community.
What new clinical discoveries arose from our work?
The ultimate goal of this project is twofold: to better characterize the pathophysiology underlying TRD, and to develop prediction rules that would allow identification of individuals at high risk for TRD early in treatment. The first steps towards these goals focused on identification of novel risk genes for TRD, and confirmation of existing putative risk genes. We continue to focus on identification of brain white-matter changes associated with long term treatment with SSRIs.
New Grants arising from this work:
R01MH08602 Innovative Approaches to Personalizing the Treatment of Depression, PI Roy Perlis (4/09-2/12)
R01MH085542 International Cohort Collection for Bipolar disorder, PI Jordan Smoller (9/08-5/13)
Publications arising from this work:
- Castro V, Gallagher P, Murphy SN, Gainer V, Fava M, Weilburg J, Churchill S, Kohane I, Iosifescu D, Smoller J, Perlis R. Using electronic medical records to enable large-scale studies in Psychiatry: Treatment Resistant Depression as a model. Psychological Med. 2011; June 10:1-10.
- Castro V, Gallagher PJ, Clements CC, Murphy SN, Gainer VS, Weilburg JB, Fava M, Churchill SE, Kohane IS, Smoller JW, Iosifescu DV, Perlis RH. Incident user cohort study of risk for gastrointestinal bleed and stroke in individuals with Major Depressive Disorder treated with antidepressants. Brit Med J Open. 2012 Mar 30;2(2):e000544. PMID:22466034.
- Hoogenboom WS, Perlis RH, Smoller JW, Zeng0-Treitler Q, Gainer VS, Murphy SN, Churchill SE, Kohane IS, Shenton ME, Iosifescu DV. Limbic system white matter microstructure and long-term treatment outcome in Major Depressive Disorder: A diffusion tensor imaging study using legacy data. World J Biol Psychiatry. 2012 Apr 30. PMID:22540406.
- Gallagher PJ, Castro V, Fava M, Weilburg JB, Murphy SN, Gainer VS, Churchill SE, Kohane IS, Iosifescu DV, Smoller JW, Perlis RH. Antidepressant response in individuals with Major Depressive Disorder exposed to NSAIDS: a pharmacovigilance study. Am J Psychiatry (in press).
- Hoogenboom WS, Perlis RH, smoller JW, Zeng0Treitler Q, Gainer VS, Murphy SN, Churchill SE, Kohane IS, Shenton ME, Iosifescu DV. Feasibility of styding brain morphology in Major Depressive Disorder with structural magnetic resonance imagine and clinical data from the electronic medical record: A pilot study. Psych. 2012: accepted.