Genotype misclassification in genetic association studies of the rs1042522 TP53 (Arg72Pro) polymorphism: a systematic review of studies in breast, lung, colorectal, ovarian, and endometrial cancer.
Dahabreh, Issa.
2011
-
Abstract: Background:
Genetic association studies in oncology often use tumor tissue as the source of
genotyping material for cancer patients. Because tumors exhibit multiple somatic genetic
aberrations, tumor-derived genetic material may not accurately represent the
individual's constitutional genotype. Loss-of-heterozygosity (LOH) that preferentially
affects one variant allele at a locus ... read moremay affect genotyping results and lead to
differential misclassification. We hypothesized that preferential loss of the proline
(Pro)-encoding allele at the rs1042522 locus of the TP53 gene (known as Arg72Pro) may
cause such differential misclassification. Methods: We performed a systematic review of
MEDLINE, the Human Genome Epidemiology Literature Finder, the Genetic Association
Database and two locus-specific databases dedicated to TP53 (last search: March 8th,
2011) to identify studies investigating the association of rs1042522 with breast, lung,
colorectal, ovarian or endometrial cancer. Data on the populations enrolled, aspects of
study design, genotyping methods used and the genotype distribution in cancer cases and
controls were extracted from eligible studies. Meta-analysis was performed using both
maximum likelihood and Bayesian approaches. For Bayesian analyses, informative priors
for the bias effect were derived from a recently published meta-analysis of the same
polymorphism in cervical cancer. Results: We identified 161 case-control studies
reported in 134 papers. Sixty-nine studies (28513 cases/31850 controls) investigated
breast cancer, 42 (16743 cases/16504 controls) lung cancer, 26 (7377 cases/10011
controls) colorectal cancer, 16 (1982 cases/5226 controls) ovarian cancer and 8 (726
cases/1292 controls) endometrial cancer. Twenty-two studies (14%) used tumor tissue as
the source of genotyping material for cases. Using multilevel models incorporating
evidence from all 5 cancers, we found strong evidence that use of tumor tissue was
associated with an underestimation of the Pro allele's effect by approximately 20% (OR
=0.78; 95% confidence interval, CI, 0.69-0.90). Among studies not using tumor tissue for
genotyping we found a significant association between rs1042522 and lung cancer, OR=1.09
(95% CI, 1.02-1.15). In Bayesian analyses, the use of tumor tissue was estimated to lead
to an underestimation of the genetic effect by 25%-30%. The probability that use of
cancer tissue is associated with underestimation of the Pro allele's "true" effect was
more than 95% in all analyses (and higher than 99% in analyses using informative
priors). These results were robust to sensitivity analyses. Conclusions: Use of tumor
tissue as the source of genotyping material for cases is associated with significant
bias in the estimate of the genetic effect in genetic association studies of rs1042522
with beast, lung, colorectal, ovarian and endometrial cancer. Future studies should
avoid the use of tumor tissue as the source for material for genotyping cases,
particularly for loci that may be affected by
LOH.
Thesis (M.S.)--Tufts University, 2011.
Submitted to the Dept. of Clinical & Translational Science.
Advisor: Joseph Lau.
Committee: Thomas Trikalinos, and Christopher Schmid.
Keywords: Genetics, and Biostatistics.read less - ID:
- zg64tz56w
- Component ID:
- tufts:20300
- To Cite:
- TARC Citation Guide EndNote