Early evidence from meta-analyses of genetic risk factors is unreliable

Article type
Authors
Trikalinos T, Ntzani E, Ioannidis J, Contopoulos-Ioannidis D
Abstract
Background: Numerous genetic association studies for complex diseases are performed. Although many investigators place emphasis on formal statistical significance (P-values<0.05), empirical evidence suggests that early findings are sometimes not replicated by subsequent research.

Objectives: To assess the predictive ability of characteristics of early-published studies for the eventual establishment of a genetic association.

Methods: We scrutinized 55 cumulative meta-analyses of genetic associations (579 studies). We estimated whether having statistical significance in the earliest (first) published study or in at least half among several (33) early-published studies had any predictive ability for establishing or refuting the presence of the genetic association in subsequent research.

Results: In 35 associations, a first study was statistically significant ("positive") and in 15 associations more than half of the early-published reports were "positive".

The average publication rate of subsequent studies increased 1.71-fold (95% confidence interval [95% CI] = 1.08, 1.27) with a "positive" first report. However, it was independent of whether most of the early-published studies were "positive" or not (rate ratio 1.02 [95% CI = 0.82, 1.26]).

When compared against the summary results of subsequent research, sensitivity and specificity were 0.65 (95% CI = 0.43, 0.84) and 0.38 (95% CI = 0.21, 0.56) for the first reports, and 0.40 (95% CI = 0.16, 0.68) and 0.73 (95% CI = 0.54, 0.87), respectively, when at least three early studies were considered.

The first study of 19 meta-analyses claimed an attributable fraction (AF) of at least 2% (based on the coverage of the corresponding 95% CI), whereas the first study in the remaining 36 could not have such certainty. The synthesis of the subsequent studies claimed an AF of at least 2% in 14 meta-analyses, while another 9 meta-analyses eventually excluded an AF of 2%.

Conclusions: Although "positive" findings in the very first reports provide strong incentive for conducting more studies on a putative genetic association, the statistical significant or even the magnitude of the effect of early studies cannot predict eventual establishment of an association. Conversely, many genuine associations of gene variants with disease risk would be missed, if research were abandoned after early underpowered "negative" studies.

Acknowledgements: Supported by a PENED grant from the General Secretariat for Research and Technology, Greece and the European Union.