Affordable Access

deepdyve-link
Publisher Website

An exploration of genetic association tests for disease risk and age at onset.

Authors
  • Martin, Eden R1, 2
  • Gao, Xiaoyi R3
  • Li, Yi-Ju4, 5
  • 1 John P. Hussman Institute for Human Genetics, Miller School of Medicine, University of Miami, Miami, Florida, USA.
  • 2 John T. MacDonald Department of Human Genetics, Miller School of Medicine, University of Miami, Miami, Florida, USA.
  • 3 Departments of Ophthalmology and Visual Science, Department of Biomedical Informatics, Division of Human Genetics, The Ohio State University, Columbus, Ohio, USA.
  • 4 Duke Molecular Physiology Institute, School of Medicine, Duke University, Durham, North Carolina, USA.
  • 5 Department of Biostatistics and Bioinformatics, School of Medicine, Duke University, Durham, North Carolina, USA.
Type
Published Article
Journal
Genetic Epidemiology
Publisher
Wiley (John Wiley & Sons)
Publication Date
Apr 01, 2021
Volume
45
Issue
3
Pages
249–279
Identifiers
DOI: 10.1002/gepi.22368
PMID: 33075194
Source
Medline
Keywords
Language
English
License
Unknown

Abstract

Risk genes influence the chance of an individual developing disease over their lifetime, although the age at onset (AAO) genes influence disease timing. These two categories are not disjoint; a gene that influences AAO might also appear to influence the risk. When an allele influences both AAO and risk, a reasonable question is whether we would have more power to detect association using a statistical test based on risk or AAO. To address this question, we compared power analytically for the Cochran-Armitage trend case-control test for risk and a linear regression case-only test for AAO. We also used simulations to compare the power of these tests with a 2-degree of freedom joint test (which combines the risk and AAO statistics) and the Cox proportional hazards survival model testing AAO (with censored data in controls). We found that when there is little heterogeneity, the case-control risk test has more power than the case-only AAO test (with equivalent sample sizes), but when the model is complex (e.g., with heterogeneity or reduced penetrance), the relationship reverses. The joint test generally outperforms the risk or AAO test alone and ultimately is our recommendation as a powerful alternative in many scenarios. © 2020 Wiley Periodicals LLC.

Report this publication

Statistics

Seen <100 times