Comments on ‘Association of FcϵRIβ polymorphisms with risk of asthma and allergic rhinitis: evidence based on 29 case–control studies’

Dear editor,

We researched the relevant studies about the association between the high-affinity IgE receptor β chain (FcεRIβ) polymorphisms and allergic diseases risk in Medline, Embase, Web of Science, Chinese National Knowledge Infrastructure, and Wanfang databases. No limit of start year and month was set, and the updated time was August 2019. The terms, search strategies, and inclusion/exclusion criteria were the same as reported by Guo et al. [1]. Comparing our retrieved studies with the ones in Table 1 of Guo et al.’s paper [1], it seems that some errors or mistakes should be corrected.

First, several relevant studies that met the inclusion criteria were missed in Guo et al.’s paper [2–15]. Of the 14 missed studies, 5 articles were published before January 2000 [2–6], which was the start time of published paper restricted in Guo et al.’s literature searching strategy [1]; 3 reports were from Japan [2,6,11], 4 studies were from China [9,13–15], 1 each was from South Africa [3], Switzerland [4], Australia [5], India [7], South Korea [8], the U.S.A. [11], and Hungary [12], respectively. In Green et al.’s study, black and white populations were recruited, respectively [3]. In Undarmaa et al.’s report, children and adult populations were collected, respectively [10].

Second, several studies published by the same research group were included in Guo et al.’s report [1]. According to the inclusion and exclusion criteria, when more than two studies were reported by the same research group, only the paper with the largest sample size was included in the analysis. We think Cui et al.’s study [16], published in 2004, with 106 adult asthmatics and 106 controls, were incorporated into their another paper, published in 2003, with 216 (number including adults and children) cases and 198 controls [17]. Similarly, the study populations in Hua et al.’s papers [18,19] and the Chinese Han case/control populations in Ramphul et al.’s article [20], were recruited by the same research group, the two smaller sample-size studies should be excluded from the analysis [18,20].

Third, one study reported by Laprise et al. [21], with atopic/non-atopic contrast groups, not all the subjects in atopic group met with the diagnosis criteria of asthma, should be excluded from the analysis.

Fourth, the reported genotype frequency for the C-109T or E+237G polymorphisms of FcεRIβ gene in two studies of Guo et al.’s paper [1] were not in agreement with the ones in their original papers [22,23]. In Sharma and Ghosh’s study, the CC, CT, and TT genotype frequency for C-109T polymorphism in case/control groups were (89, 108, and 40)/(34, 118, and 69), respectively [22], which were wrongly counted as (87, 113, and 37)/(39, 108, and 74), respectively, in Guo et al.’s paper [1]. In Amo et al.’s published article, the EE, EG, and GG genotype frequency in control group for E+237G polymorphism were 487, 39, and 0, respectively [23], which were wrongly counted as 144, 277, and 105, respectively [1].

Considering the above-listed mistakes or errors in Guo et al.’s published paper, it seems that the findings and conclusions of Guo et al.’s study were not entirely reliable [1]. To overcome the limitations, we performed an updated meta-analysis to re-assess the associations of C-109T and E+237G polymorphisms in the FcεRIβ gene with allergic disease (asthma and allergic rhinitis) risk. The statistical analysis methods and software used in this comment were the same as reported by Guo et al., unless otherwise indicated [1].

The main characteristics of the eligible studies [2–17,19,20,22–42], including the first author, publication year, country where individual study was conducted, ethnicity of study population, atopic disease category, sample size of case/control groups, the detailed genotype frequency, and the P-values for Hardy–Weinberg Equilibrium (HWE) test, were shown in Table 1. There were 36 case–control studies about the association between E+237G variant and allergic diseases risk [2–4,6–15,17,19,20,23–28,30–33,36,38,39,41,42], and 15 were about the correlation of C-109T polymorphism with allergic diseases risk [5,8,12,14,17,19,22,23,29,34,35,37,38,40]. Of the 15 case–control studies about C-109T polymorphism and allergic disease risk (14 ones according to ethnicity or HWE classification), 10 were performed in Asians [8,14,17,19,20,22,29,34,35,38] and 4 were conducted in Caucasians [5,23,37,40], respectively; 13 studies were about asthma risk [5,8,17,19,20,22,29,34,37,38,40], 1 was about allergic rhinitis risk [23], and 1 about asthma and rhinitis risk [23], respectively; genotype frequency distribution in control groups of 13 studies were in agreement with HWE [5,8,14,17,19,20,22,23,34,35,37,38,40] and 1 was not [29], respectively. Of the 36 case–control studies about E+237G variant with allergic diseases risk (35 ones according to ethnicity or HWE classification), 25 were carried out in Asians [2,6–10,13–15,17,19,20,24,26–29,32,33,36,38,39,42], 8 were performed in Caucasians [3,4,12,23,25,30,31,41], 1 in Africans [3] and 1 in mixed populations [11], respectively; 31 studies were about asthma risk [2–4,6–15,17,19,20,24–26,28,29,31–33,36,38,39,41,42], 2 were on rhinitis risk [23,27], and 3 were concerned with asthma/rhinitis risk [23,30,31], respectively; genotype frequency distribution in control groups of 32 studies were in line with HWE [2–4,6–11,13–15,17,19,20,23–26,28–33,36,38,39,41,42] and 3 were not [12,13,27], respectively.

Table 2 listed the summary odds ratios (ORs) of the association of FcεRIβ C-109T polymorphism with allergic diseases risk. Overall, no significant associations between C-109T polymorphism and allergic diseases risk were observed (OR = 1.001, 95% confidence interval (CI): 0.909–1.102 for CC+CT vs. TT and OR = 1.015, 95% CI: 0.788–1.307 for CC vs. CT+TT, respectively). When subgroup analyses by ethnicity (Asian and Caucasian), allergic disease classification (asthma, rhinitis, and both) and HWE (in and not) were performed, we did not find any statistically significant associations of C-108T polymorphism with allergic diseases risk (Table 2). No any publication and other small study related biases were observed in overall and subgroup analyses (Table 2).

Table 3 showed the summary ORs for the association between FcεRIβ E237G variant and allergic diseases risk. Overall, we observed FcεRIβ 237G allele was associated with increased risk of allergic diseases in total population (OR = 1.178, 95% CI: 1.022–1.357 for G vs. E and OR = 1.207, 95% CI: 1.031–1.411 for GG+EG vs. EE, respectively) (Table 3 and Figure 1). When restricted the analysis to the studies with control groups’ genotype frequency distribution were met with HWE, we observed an elevated risk of allergic diseases among subjects carrying EG or GG genotypes, in comparison with EE genotype carriers (OR = 1.225, 95% CI: 1.041–1.442) (Table 3 and Figure 1). When stratified analyses were conducted by ethnicity, we found an increased risk of allergic diseases in subjects carrying EG or GG genotypes, compared with EE genotype carries in Asians (OR = 1.189, 95% CI: 1.001–1.412) (Table 3 and Figure 2). No significant association of E237G polymorphism with allergic diseases risk was observed in Caucasians (OR = 1.544, 95% CI: 0.884–2.697 for G allele vs. E allele and OR = 1.547, 95% CI: 0.895–2.673 for EG+GG vs. EE, respectively) (Table 3 and Figure 2). In subgroup analyses by allergic diseases classification (asthma, allergic rhinitis, and both), we did not observe significant association of E237G with any allergic diseases categories (Table 3 and Figure 3).

We also performed a cumulative meta-analysis which accumulated the evidence about association of E237G variant with allergic diseases risk in the order of publication year of individual study. We observed that the association of EG/GG genotypes with increased allergic diseases started to become significant for the first time when Zeng et al.’s [28] study published in the year of 2001 (OR = 1.374, 95% CI: 1.013–1.864) and the summary OR became very similar to the OR estimated in this report (OR = 1.207) when Wang et al.’s [39] study published in the year of 2009 (OR = 1.299, 95% CI: 1.026–1.644) (Figure 4).The overall tendency of summary OR variation seemed alarming at the top of forest plot of the cumulative meta-analysis. It should be noted that Shirakawa et al. [2] (study 1) reported the first positive association of 237EG+GG with asthma risk with OR being equal to 2.693 in a Japanese (Asian) population and the second included study [3] with two independent case–control studies [one in black (study 2) and one in white (study 3)] reported different associations, one is negative association of 237EG+GG with asthma risk in Black (OR = 0.763) and the other is similar to Shirakawa et al.’s result in White population (OR = 3.693). When merging the result of study 1 with study 2, the overall OR significantly decreased. And when summarizing the results of the first three studies (i.e. study 1, 2, and 3) the overall OR looked bigger than the second. As a result, the overall ORs would generate a V-form change in the forest plot of cumulative meta-analysis. The alarming changes in the plot of cumulative meta-analysis also indicated that there was high heterogeneity of the results of the included individual studies.

Sensitivity analysis was performed by sequentially omitting each individual study in the order of publication year and the pooled ORs were estimated repeatedly, which was used to evaluate the stability of the results of present meta-analysis. The sensitivity analysis showed that the association of EG and GG genotypes with increased risk of allergic diseases maintained statistically significant when removing any each individual study (Figure 5). Egger’s regression test and Begg’s rank correlation test were used to evaluate the small-study effects and potential publication bias in current meta-analysis. Both tests indicated that the significant association of G allele or EG+GG genotypes with elevated risk of allergic diseases might strongly influenced by small-study effect or publication bias (Table 3). The Egger’s funnel plots for the association between E237G polymorphism and allergic diseases risk also showed that the OR distributions for both G allele vs. E allele (Figure 6A) and EG+GG vs.EE (Figure 6B) were obviously asymmetrical.

There are some inherent limitations of meta-analysis which should be taken into consideration when using the results of this comment. First, there was high heterogeneity in this meta-analysis, especially in the case of association of E237G variant with allergic diseases risk. Although, subgroup analyses were performed on the basis of ethnicity, allergic disease category and HWE, heterogeneity among the included studies still be statistically significant in all subgroups. Second, publication bias tests indicated that the probable existence of publication bias, i.e. some unpublished negative results studies thus could not be included in this analyses might result in an over-estimated association of E237G with allergic disease risk.

In conclusion, the results of Guo et al.’s study [1] should be interpreted with caution. To make an asserted conclusion, well-designed studies with large number of homogeneous population are required. We do hope that this comment will be helpful to clarify the results presented by Guo et al. [1].

The authors declare that there are no competing interests associated with the manuscript.

This work was supported by the Key Program of Science and Technology Research Project from the Department of Education of Hubei Province [grant number D20192001].

H.Y.: designed the study, performed the statistical analysis and edited the manuscript. L.Z.: conducted literature search and extracted data from individual studies. Y.Z. and M.Y.: prepared and reviewed the manuscript. S.W.: conducted literature search and data checking. All authors approved the final manuscript.

CI

confidence interval

FcεRIβ

high-affinity IgE receptor β chain

HWE

Hardy–Weinberg equilibrium

OR

odds ratio