The growing expansion of the approved multiple sclerosis (MS) disease-modifying treatments (DMTs) and the variable responses to MS treatment have created an unmet medical need to provide individually tailored therapy. Efforts to bring precision medicine to provide individualized MS treatment selection have been impeded by our limited understanding of the factors that determine treatment response. While genomics hold the promise for closing this knowledge gap, the insufficient number of patients with detailed treatment response data and the modest effect size of genetic variants that influence treatment responses are the main limiting factors in pharmacogenomics studies. As electronic health records (EHR) become widely adopted and increasingly standardized and as we implement sophisticated computational and statistical methods to harness the EHR data, EHR systems can become cost-effective platforms to perform large-scale treatment response studies in real-life settings. Our team with a history of productive collaborations and diverse expertise (led by PI Dr. Xia) previously developed robust algorithms to identify 5,495 MS patients from the Partners HealthCare EHR systems and then model MS disease activity in these patients using EHR data. The Partners EHR system contains longitudinal clinical information on thousands of MS patients from two large academic medical centers and is linked to a well-characterized MS patient research registry and biobanks with existing genomics data. For the proposed study, we will test the hypothesis that meaningful phenotypes of MS disease activity can be extracted from EHR data to inform treatment response, and that additional common genetic variants exist in the population and can predict therapeutic response in MS when combined with clinical features derived from EHR data. The proposed study has three aims with the overall goal to produce a computational and analytic approach capable of identifying MS disease activity in relation to treatment history using EHR data and integrate with genomics profile to develop a predictive model of therapeutic response to commonly prescribed DMTs in this cohort of 5,495 MS patients, including injectable (interferon-?, glatiramer acetate) and oral (fingolimod, dimethyl fumarate) options. Specifically, we will (1) leverage narrative electronic health records data (e.g., clinical notes, radiology reports) and natural language processing (NLP) to ascertain individualized response to DMTs (n=600 for each DMT); (2) Identify clinical features from electronic health record data (e.g., diagnoses, exposures) that predict response to DMTs using a systematic phenome-wide approach; (3) Develop and test a comprehensive predictive model of individualized response to DMTs that incorporates clinical and genetic predictors. This research has the potential impact to be transformative by contributing to a major knowledge gap regarding the factors that influence treatment response and bringing precision medicine closer to individualized MS treatment selection.
AND PUBLIC HEALTH RELEVANCE Multiple sclerosis (MS) is a chronic neurological condition that affects over 400,000 individuals in the United States and creates a high socioeconomic burden as a leading cause of neurological disability in young adults. Because not all patients respond the same way to a specific medication, physicians and MS patients often lose precious time searching for effective treatment with serially testing of costly medications. An individually tailored treatment can ensure early start of effective medication that can prevent relapse and progression of disability. Ultimately, this project will help gain insights into the factors that determine treatment response and enable physicians to match an individual MS patient's clinical and genomic profile with uniquely tailored therapy to maximize effectiveness, delay disease progression and reduce overall cost.
| Manners, Jody; Jadhav, Ashutosh P; Xia, Zongqi (2017) Mystery Case: A 61-year-old woman with lower extremity paralysis and sensory loss. Neurology 89:e257-e263 |
| Bargiela, David; Bianchi, Matthew T; Westover, M Brandon et al. (2017) Selection of first-line therapy in multiple sclerosis using risk-benefit decision analysis. Neurology 88:677-684 |
| Xia, Zongqi; Steele, Sonya U; Bakshi, Anshika et al. (2017) Assessment of Early Evidence of Multiple Sclerosis in a Prospective Study of Asymptomatic High-Risk Family Members. JAMA Neurol 74:293-300 |
