Identify Systemic Lupus Erythematosus (SLE) Patients Treated with Belimumab with Reduced Severe and Severe/Moderate Flares: A Post-hoc Trial Analysis

Identify Systemic Lupus Erythematosus (SLE) Patients Treated with Belimumab with Reduced Severe and Severe/Moderate Flares: A Post-hoc Trial Analysis

We are applying for a small funding from PhRMA of $10,000 for 1 year for Health Outcomes and Informatics. See below:

http://www.phrmafoundation.org/2018-awards/research-starter-grants/health-outcomes%E2%80%A2-informatics/

holder of GSK stocks

Background
This post-hoc study will allow better targeting of patients treated with belimumab who may experience flare reduction. Flares are associated with patient fear, worse health-related quality of life, higher health care costs, and increased organ damage. The results of the post-hoc trial analysis will support targeting right patients and strengthening economic value of belimumab. In the current U.S. and global health care markets, biologics continue to be under scrutiny from payers and the public for value and price.
This study will be different from and build on the study on baseline predictors of SLE flares published in 2013 by Petri et al. who studied flare response in the control groups only in the two pivotal trials BLISS-52 and BLISS-76. We are proposing to study flare response to belimumab treatment aiming to identify who benefited in the way of flare reduction. We propose to pool the 3 belimumab phase 3 trials (two intravenous trials and one subcutaneous trial) and apply state-of-the-art machine learning techniques for high-dimensional data for an exhaustive search of all baseline characteristics available as potential response predictors.
This study will complement the study published by van Vollenhoven et al. in 2012 that predicted response to the primary end point defined by SLE Responder Index (SRI) in the two IV pivotal trials.
In addition to the differences in endpoints, this study will be distinguished in using state-of-the-art machine learning techniques for a complete search for all baseline predictors of flare response, including demographics, disease characteristics, medications, biomarkers, etc. This will maximize predictive accuracy of flare response and identification of flare responders allowed by the trial data. Specifically, we will compare predictive performance of regular logistic regression with Lasso logistic regression and neural network. SLE disease is not thoroughly understood and neither is response to belimumab. Although the true predictors of response may not be measured, some of them may be correlated with baseline endpoints collected in the trials. We hypothesize that multi-layer neural network (ie, deep learning) will be the most powerful prediction algorithm followed by Lasso logistic regression which simultaneously selects and estimates important predictors, both of which will be superior to regular logistic regression with variables selected in an ad hoc way. The gap in prediction performance between deep learning and regular logistical regression may indicate the scientific knowledge gap of flare response mechanism.
Research Aims
(1)   Compare performance of predicting severe and severe/moderate flares using regular logistic regression and machine learning methods including regularized logistic regression, neural network and random forest.
(2)   To identify subgroups of SLE patients treated with belimumab who experienced fewer severe and severe/moderate flares.

[{ "PostingID": 1416, "Title": "GSK-HGS1006-C1056", "Description": "A Phase 3, Multi-Center, Randomized, Double-Blind, Placebo-Controlled, 76-Week Study to Evaluate the Efficacy and Safety of Belimumab (HGS1006, LymphoStat-B™), a Fully Human Monoclonal Anti-BLyS Antibody, in Subjects with Systemic Lupus Erythematosus (SLE)" },{ "PostingID": 1417, "Title": "GSK-HGS1006-C1057", "Description": "A Phase 3, Multi-Center, Randomized, Double-Blind, Placebo-Controlled, 52-Wk Study to Evaluate the Efficacy and Safety of Belimumab (HGS1006, LymphoStat-B™), a Fully Human Monoclonal Anti-BLyS Antibody, in Subjects With Systemic Lupus Erythematosus (SLE)" },{ "PostingID": 14488, "Title": "GSK-HGS1006-C1115", "Description": "A Phase 3, Multi-Center, Randomized, Double-Blind, Placebo-Controlled, 52-Week Study to Evaluate the Efficacy and Safety of Belimumab (HGS1006) Administered Subcutaneously (SC) to Subjects with Systemic Lupus Erythematosus (SLE)" }]

The publication citation will be added after the research is published.