We read with great interest the letter by Liu etal., who raise several methodological concerns about our recently published meta-analysis.1 Although their points are theoretically valid for any meta-analysis, we have found them inapplicable to our article. The first concern raised by Liu etal. is the searching strategy, which they claim should include the use of local databases and manual searching. They should have provided actual proof of missing studies; we are quite confident that no other reports with reliable data on the influence of rs738409 on nonalcoholic fatty liver disease had been published by the time of our search. For instance, manual searching with Google Scholar did not reveal any other eligible studies in English or other languages. In addition, local databases do not always include only peer-reviewed articles, so the data are not necessarily reliable. Moreover, with respect to clinical practice, not all studies in local databases are considered equally strong, and they can be almost impossible to find. Hence, this concern is overcome. Liu etal.'s second concern is the Hardy-Weinberg equilibrium (HWE) in the control groups. We are very glad to report that a χ2 test was used to determine whether the observed genotype frequencies conformed to the HWE expectations. The data extracted from all the articles (except two) showed no deviations from the HWE with very stringent criteria (Pearson's goodness-of-fit χ2 < 0.05, degree of freedom = 1). The publications whose results did not follow the HWE2, 3 showed borderline P values. Hence, this concern is overcome. Finally, Liu etal. claim that we did not perform a stratified analysis based on the sources of the controls to confirm the validity of our results. To prove that the results are not biased, we have performed the suggested analysis, and we have found that the effects of rs738409 on nonalcoholic fatty liver disease are very similar with controls selected from the general population and with controls selected from a hospital-based setting; the effects are even greater in the random model with general population controls (Fig. 1). In conclusion, methodological considerations are always very welcome. Nevertheless, criticisms must be accompanied and justified by actual facts so that the readers can determine whether they are valid. In this particular case, all the concerns raised by the authors are generic considerations that are not applicable to our article, and we regret the title of their letter.

thumbnail image

Figure 1. Stratified analysis of the controls. References for the listed studies are provided elsewhere.1 Abbreviations: CI, confidence interval; GP, general population; HB, hospital-based setting; T2D, type 2 diabetes.

Download figure to PowerPoint


  1. Top of page
  • 1
    Sookoian S, Pirola CJ. Meta-analysis of the influence of I148M variant of patatin-like phospholipase domain containing 3 gene (PNPLA3) on the susceptibility and histological severity of nonalcoholic fatty liver disease. HEPATOLOGY 2011.
  • 2
    Kantartzis K, Peter A, Machicao F, Machann J, Wagner S, Konigsrainer I, etal. Dissociation between fatty liver and insulin resistance in humans carrying a variant of the patatin-like phospholipase 3 gene. Diabetes 2009; 58: 2616-2623.
  • 3
    Wagenknecht LE, Palmer ND, Bowden DW, Rotter JI, Norris JM, Ziegler J, et al. Association of PNPLA3 with non-alcoholic fatty liver disease in a minority cohort: the Insulin Resistance Atherosclerosis Family Study. Liver Int 2011; 31: 412-416.

Silvia Sookoian M.D., Ph.D.*, Carlos J. Pirola Ph.D.†, * Departments of Clinical and Molecular Hepatology, Alfredo Lanari Institute of Medical Research, University of Buenos Aires/National Council of Scientific and Technological Research, Buenos Aires, Argentina, † Molecular Genetics and Biology of Complex Diseases, Alfredo Lanari Institute of Medical Research, University of Buenos Aires/National Council of Scientific and Technological Research, Buenos Aires, Argentina.