We thank the reader very much for his interesting comments on the paper. However, we think the reader should refer to ‘Identification of relevant studies and eligibility criteria’ part in ‘Materials and Methods’ section. We have clearly stated the inclusion criteria for studies recruited in the meta-analysis. These studies mentioned by the reader [Q192R, Topić et al. , Voetsch et al. , and Ranade et al. ] were excluded from the present meta-analysis because they did not describe the separate ischemic stroke data on genotype distribution [1, 3] or the data of the study  were consistent with an earlier published article in 2002 by the same author . We only included the later in the present meta-analysis according to inclusion criteria. Furthermore, the study (Voetsch et al.)  was not included also because the genotype distribution violates Hardy–Weinberg equilibrium (HWE) in controls for PON1 T107C polymorphism. Likewise, Ref. 34, 36 for 192QR [4, 5], Ref. 41, 46 for L55M [6, 7], and Ref. 53 for G148A  were excluded because of a deviation from HWE in controls.
As for another study (Wang et al) , we have had a description in ‘Introduction’ section. Please see ‘By using meta-analysis, researchers investigated the relationship of certain PON gene polymorphisms (Q192R, L55M of PON1, and PON2 C311S) and IS … …’ part. Moreover, we also included some eligible studies included in Wang's study  under our inclusion criteria.
For PON2 C311S, the reader thinks the allele distribution is in error for two studies (Ref. 46 and 52) [7, 10]. However, the data on genotype distribution were extracted from eligible studies in the present meta-analysis. In addition, Pasdar (Ref. 46)  kindly provided the information on the genotype distribution in his study. Various population studies have reported the inter-ethnic differences in the allele frequencies for PON2 C311S. For example, in Pakistan (C: 0.612) , United Kingdom (C: 0.74)  and United States (C: 0.767)  healthy population, the C allele is more common compared with that of population in China or Japan. Some readers might consider the minor alleles as risk alleles; however, for single nucleotide polymorphism, the major alleles do not necessarily play a protective role in diseases, and vice versa.
We analyzed the pooled data again after adjusting the data from the two studies on C311S genotype distribution [7, 10], namely, the data on C allele and S allele exchanged each other. However, there were still no statistically significant differences among all comparisons (odds ratio, 95% confidence interval: 0.96, 0.77–1.20, P = 0.71 in allelic comparison, 0.99, 0.70–1.39, P = 0.95 for Serine/Serine genotype (SS) vs. Cysteine/Serine genotype (CS) plus Cysteine/Cysteine genotype (CC), and 0.92, 0.58-1.48, P = 0.74 for CC vs. CS+SS, respectively) (figures omitted).
Taken together, the evidence above verified the reliability and stability of the results in the present meta-analysis.