An official website of the United States government

Publication Information

PubMed ID
Public Release Type
Journal
Publication Year
2024
Affiliation
Department of Statistics and Data Science, University of Texas at Austin 2Department of Biostatistics, Harvard University 3Department of Biomedical Data Science, Stanford University
Authors
Cai T, Parast L, Tian L
Studies

Abstract

Clinical studies of chronic diseases often require long-term patient follow-up. It is known that such long durations impact our ability to get e ective treatments to patients in need, increase study costs, and substantially contribute to participant bur- den and noncompliance. Over the past 30 years, an incredible amount of progress has been made in the development of statistical methods to identify surrogate markers i.e, measurements that could replace long-term outcomes. However, available methods are generally not applicable to studies with a small sample size. These methods tend to either rely on nonparametric kernel smoothing with requires a relatively large sam- ple size or rely on strict model assumptions that are unlikely to hold in practice and empirically dicult to verify with a small sample. In this paper, we propose a novel nonparametric rank-based approach to evaluate a surrogate marker in a small sample size setting using a rank-based approach and relying on the theory of U-statistics. The methods developed in this paper are motivated by and applied to a small study of chil- dren with nonalcoholic fatty liver disease (NAFLD), a a diagnosis for a range of liver conditions in individuals without signi cant history of alcohol intake. Speci cally, we examine change in alanine aminotransferase (ALT; measured in blood) as a surrogate marker for change in NAFLD activity score (obtained by biopsy) in the TONIC trial which compared Vitamin E (n = 50) versus placebo (n = 46) among children with NAFLD. 2