Comments on “Correlation between PNPLA3 rs738409 polymorphism and hepatocellular carcinoma: a meta-analysis of 10,330 subjects”

Dear Editor,

Huang et al. ¹ recently published a meta-analysis on the association between rs738409 polymorphism and hepatocellular carcinoma (HCC). As their main finding, they indicated “a significant association with the likelihood of HCC was detected for the PNPLA3 rs738409 polymorphism in dominant (P = 0.0001; odds ratio (OR) 0.66; 95% confidence interval (CI) 0.53, 0.82), and recessive (P < 0.0001; OR 2.32; 95% CI 1.76, 3.06) comparisons.” As seen the results are not aligned for these genetic models. However, there are some issues that need to be addressed in the concept of their genetic models and meta-analysis, which affects the results and interpretation and should be noticed. We aim to comment on these issues in this letter.

There are two different points of views for genetic models of rs738409, based on which allele (C or G) is chosen as the reference for definitions of genetic models. If the authors aim to compare C vs. G in allelic model (as described in their study), associated dominant and recessive models are CC+CG vs. GG and CC vs. CG+GG, respectively. Actually, the dominant model for the C allele means that even with the presence of one C allele the effect should be present, thus CC and CG are the attributed genotypes for this model; for the recessive model only the presence of both C alleles (CC genotype) makes the effect. However, in the Huang et al. ¹ meta-analysis in the text, Table 2, and supplementary, GG vs. CC+CG has been attributed to the recessive model, which should be modified to CC vs. CG+GG. Accordingly, the dominant model comparison in the text, Table 2 (overall and subgroups), and supplementary of their study should be modified to the recessive model, which is related to CC vs. CG+GG. In addition, we have shown the modified analysis related to the dominant model comparison (CC+CG vs. GG) (Fig. 1). Our results show the protective effects of the C allele (OR=0.43; 95% CI=0.33, 0.57). We also performed the subgroups analysis based on the ethnicity in the correct dominant model (Table 1).

Figure 1.

Forest plot of rs7384409 polymorphism and risk HCC under the dominant model.

Table 1.

Overall and subgroup meta-analysis for rs7384409 polymorphism and risk of HCC.

Population	Sample size	Dominant comparison (CC+CG vs. GG)			Homozygote comparison (CC vs. GG)
		P-value	OR (95% CI)	I2	P-value	OR (95% CI)	I2
Overall	3154/7175	<0.0001	0.43 (0.33-0.57)	73%	<0.0001	0.37 (0.26-0.53)	80%
Asian	856/1792	0.22	0.69 (0.38-1.25)	73%	0.30	0.66 (0.30-1.46)	79%
Caucasian	204/4889	<0.0001	0.37 (0.27-0.49)	67%	<0.0001	0.30 (0.21-0.44)	75%

In a further meta-analysis, we found the most significant association (OR=0.37; 95% CI=0.26, 0.53) in homozygote model (CC vs. GG), which is important in interpreting the results (Figure 2).

Figure 2.

Forest plot of rs7384409 polymorphism and risk of HCC under homozygote model.

In conclusion, when we combine our meta-analysis with the results from the Huang et al. ¹ meta-analysis it seems to be clear that the C allele of PNPLA3 rs738409 polymorphism has a significantly protective effect against the risk of HCC.