Sequence kernel association test for survival outcomes in the presence of a non-susceptible fraction.

Autor: Lakhal-Chaieb, Lajmi; Simard, Jacques; Bull, Shelley

Publication year: 2020

Biostatistics (Oxford, England)

issn:1468-4357 1465-4644

doi: 10.1093/biostatistics/kxy075

Abstract:

In this work, we propose a single nucleotide polymorphism set association test for survival phenotypes in the presence of a non-susceptible fraction. We consider a mixture model with a logistic regression for the susceptibility indicator and a proportional hazards regression to model survival in the susceptible group. We propose a joint test to assess the significance of the genetic variant in both logistic and survival regressions simultaneously. We adopt the spirit of SKAT and conduct a variance-component test treating the genetic effects of multiple variants as random. We derive score-type test statistics, and we investigate several approaches to compute their $p$-values. The finite-sample properties of the proposed tests are assessed and compared to existing approaches by simulations and their use is illustrated through an application to ovarian cancer data from the Consortium of Investigators of Modifiers of BRCA1 and BRCA2.

Language: eng

Pmid: 30590388

Tags: Humans; Female; Polymorphism, Single Nucleotide; *Models, Statistical; *Models, Genetic; *Disease Susceptibility; *Survival Analysis; Association test; BRCA2 Protein/genetics; Censored trait; Logistic regression; Non-susceptible fraction; Ovarian Neoplasms/genetics/mortality; Proportional hazards Cox model; Ubiquitin-Protein Ligases/genetics

Link: https://pubmed.ncbi.nlm.nih.gov/30590388/