Contributions
Abstract: 183
Type: Oral
Abstract Category: Pathology and pathogenesis of MS - Biomarkers
Background: Relapsing-remitting multiple sclerosis (RRMS) is a chronic debilitating disease with highly variable outcomes. Currently there are few tools outside of MRI to directly measure disease activity (DA). While useful, MRI reflects historic damage but not the dynamic biological processes that underlie RRMS. Furthermore, the field is in need of tools to accurately track DA, identify early evidence of relapse, and treatment response. Given the complex biology of RRMS, we designed a study to assess the utility of multiple serum biomarkers (BMs) as the first step in creating a novel prognostic test.
Methods: We curated 300 publications, surveyed biology models and experimental data to identify candidate serum BMs. 220 serum proteins were selected to be run using the RBM MAP platform. Serum samples and clinical data were obtained from the Accelerated Cure Project registry for 60 RRMS patients in exacerbation and 65 RRMS patients in remission (no relapses for > 1 year). Samples were matched according to age, gender, disease duration, co-morbidities criteria.
Results: Statistical analysis identified 130 BMs (p=< 0.05) that were differentially expressed in RRMS compared to healthy individuals. We assessed the ability to separate exacerbation versus non-exacerbation subjects using univariate, multivariate and logistic regression models. 20 BMs produced significant univariate results at p=< .05. Random forest, boosted trees and lasso models were built using 100 iterations and 5-fold cross validation to provide separate training and testing sets. The average area under the curve ROC for the test sets was .80 or greater for models using 20-40 BMs. Analysis controlled for age, gender and disease duration.
Conclusion: This study identified a broad range of serum BMs in RRMS patients which have potential utility in tracking DA and prognosis. Of these, we constructed models to separate subjects in relapse versus remission at a meaningful level of statistical significance. Future longitudinal studies will build on these findings to verify and identify BMs which can track changes in DA, determine early evidence of relapse, and discriminate patients with no evidence of DA (NEDA) versus subclinical disease.
Disclosure:
Robert C Axtell. I have received a research grants from Octave Bioscience and Merck KgAA, consulted for Biogen Idec, and serve on the speakers bureau for EMD Serono.
Lawrence Steinman: consults for Celgene, AbbVie, Teva, EMD Serono, Novartis, Atreca, Raptor, Tolerion, Octave; holds stock or options in Raptor, Tolerion, Atreca, Transparency Life Sciences, Octave Bio
William A. Hagstrom: nothing to disclose
Robert B. Schmidt: nothing to disclose
Michael Walker: Consult to Octave Bioscience
Abstract: 183
Type: Oral
Abstract Category: Pathology and pathogenesis of MS - Biomarkers
Background: Relapsing-remitting multiple sclerosis (RRMS) is a chronic debilitating disease with highly variable outcomes. Currently there are few tools outside of MRI to directly measure disease activity (DA). While useful, MRI reflects historic damage but not the dynamic biological processes that underlie RRMS. Furthermore, the field is in need of tools to accurately track DA, identify early evidence of relapse, and treatment response. Given the complex biology of RRMS, we designed a study to assess the utility of multiple serum biomarkers (BMs) as the first step in creating a novel prognostic test.
Methods: We curated 300 publications, surveyed biology models and experimental data to identify candidate serum BMs. 220 serum proteins were selected to be run using the RBM MAP platform. Serum samples and clinical data were obtained from the Accelerated Cure Project registry for 60 RRMS patients in exacerbation and 65 RRMS patients in remission (no relapses for > 1 year). Samples were matched according to age, gender, disease duration, co-morbidities criteria.
Results: Statistical analysis identified 130 BMs (p=< 0.05) that were differentially expressed in RRMS compared to healthy individuals. We assessed the ability to separate exacerbation versus non-exacerbation subjects using univariate, multivariate and logistic regression models. 20 BMs produced significant univariate results at p=< .05. Random forest, boosted trees and lasso models were built using 100 iterations and 5-fold cross validation to provide separate training and testing sets. The average area under the curve ROC for the test sets was .80 or greater for models using 20-40 BMs. Analysis controlled for age, gender and disease duration.
Conclusion: This study identified a broad range of serum BMs in RRMS patients which have potential utility in tracking DA and prognosis. Of these, we constructed models to separate subjects in relapse versus remission at a meaningful level of statistical significance. Future longitudinal studies will build on these findings to verify and identify BMs which can track changes in DA, determine early evidence of relapse, and discriminate patients with no evidence of DA (NEDA) versus subclinical disease.
Disclosure:
Robert C Axtell. I have received a research grants from Octave Bioscience and Merck KgAA, consulted for Biogen Idec, and serve on the speakers bureau for EMD Serono.
Lawrence Steinman: consults for Celgene, AbbVie, Teva, EMD Serono, Novartis, Atreca, Raptor, Tolerion, Octave; holds stock or options in Raptor, Tolerion, Atreca, Transparency Life Sciences, Octave Bio
William A. Hagstrom: nothing to disclose
Robert B. Schmidt: nothing to disclose
Michael Walker: Consult to Octave Bioscience