Methodological concerns about a recent meta-analysis of the influence of the I148M variant of patatin-like phospholipase domain containing 3 on the susceptibility and histological severity of nonalcoholic fatty liver disease


  • Potential conflict of interest: Nothing to report.

To the Editor:

We read the meta-analysis by Sookoian and Pirola1 with great interest. Their meta-analysis suggests that patatin-like phospholipase domain containing 3 rs738409 C/G is a strong modifier of the natural history of nonalcoholic fatty liver disease. However, several points should be mentioned here.

The perfect searching strategy and the use of more related databases allow researchers to include an extensive number of potentially eligible studies, and this is crucial for a meta-analysis. Although the Medical Literature Analysis and Retrieval System Online (MEDLINE) database is one of the most comprehensive databases for health care information, its coverage is not complete.2 Lemeshow et al.3 and Seminara et al.4 suggested that at least MEDLINE, another electronic database, and hand searching should be used for a thorough search. In this meta-analysis, only the MEDLINE database was searched for eligible studies. In addition, Sookoian and Pirola1 limited the search to publications written in English. Using this approach, they may have neglected some eligible studies, and this may have resulted in selection or publication bias. Moreover, local databases also should have been searched.

The Hardy-Weinberg equilibrium should be evaluated in a control group. The deviation from the Hardy-Weinberg equilibrium presents the probability of genotyping errors, selection bias, or other bias.5 However, the Hardy-Weinberg equilibrium test was not performed in this meta-analysis. According to the sources of the controls, a case-control study is usually categorized as a hospital-based case-control (HCC) study (the controls are hospitalized patients) or a population-based case-control study (the controls are healthy people). A meta-analysis based on an HCC study may be biased because HCC controls always have some kind of disease, unhealthy life habit, or risk genotype.6 It is routine in a meta-analysis for a stratified analysis to be performed according to the sources of the controls to confirm the validity of the results rather than bias.6 Similarly, to retain the homogeneity and make the results more reliable, Sookoian and Pirola1 should perform a subgroup analysis for this meta-analysis and thus confirm the validity of the results.

Liu Liu M.D.*, Kai Wang M.D.*, Fu-Zhou Hua M.D.†, Jiang-Hua Shao M.D.*, * Departments of Gastrointestinal Surgery, Nanchang University, Nanchang City, Jiangxi, China, † Anesthesiology, Second Affiliated Hospital, Nanchang University, Nanchang City, Jiangxi, China.