A semi-parametric analysis for identifying scleroderma patients responsive to an anti-fibrotic agent

In a recently concluded multi-center two-armed clinical trial for Scleroderma patients, results from a variety of analyses based on parametric models showed that late phase patients benefited from the bovine collagen treatment at 15 months and the placebo group did not. However, the cutoff time for defining early or late phase patients was somewhat arbitrary and sample size for late phase patients was small in the two groups. We apply a combination of semi-parametric and parametric approaches to further ascertain treatment efficacy and address modeling issues related to the choice of the cutoff and the scientific hypothesis of interest. We compare results with those obtained from parametric models and demonstrate how semi-parametric methods can help to better identify Scleroderma patients responsive to an anti-fibrotic agent.