Comments to author on gene-based association analysis

In this manuscript, Jiang et al report a gene-based association analysis with meta-

analysis p-values of two independent HCC GWASs. Further, the gene- based- p-values
were evaluated within multiple gene sets defined according to knowledge of HCC and
significant SNPs were selected for replication in two independent HCC case/control
samples.
The results are interesting because the molecular mechanisms of many susceptibility
SNPs defined by GWASs in cancer heritability and in promoting cancer risk remain
elusive.
However, in my opinion, the data and conclusions are not solid enough to warrant
publication at this stage.

Major deficiencies of the manuscript are

1) Authors have performed genome wide meta-analysis using p-values of SNPs from
two existing Chinese HCC GWASs (Jiang, D. K. et al. Genetic variants in STAT4
and HLA-DQ genes confer risk of hepatitis B virus-related hepatocellular
carcinoma. Nat Genet 45, 72-5 (2013) and Chan, K. Y.et al. Genome-wide
association study of hepatocellular carcinoma in Southern Chinese patients with
chronic hepatitis B virus infection. PLoS One 6, e28798 (2011)). The reasons for
selecting these two studies are not mentioned. There are many other GWASs studies
has been published in recent years, which could have been involved (for example- Qu
LS, et.al. Nine susceptibility loci for hepatitis B virus-related hepatocellular
carcinoma identified by a pilot two-stage genome-wide association study. Oncol Lett.
2016 Jan;11(1):624-632.) or authors should mention specific reasons for the selection
of these two studies. The other issue with selection in these two studies is number of
samples, in one study the number is 2514 and in other study number is 192, which are
not statistically significant.
2) Authors have used KGG version 3.5 and Molecular signature Database (MSigDB V
4.0), although new and more recently introduced version is available.
3) Authors claim that, the Systemic lupus erythematosus pathway, passed the
Significance level after the gene-set based association p –value by the Wilcoxon test on KGG,
however no explanation has given how this pathway is involved in HBV associated HCC.

4) Authors claim that the germline mutations in genes frequently intergraded by HBV
may not contribute to the risk of HCC. However experimental proof is lacking to
support this statement.
5) Authors have used human proteome atlas which shows elevated levels of 433 genes
in liver compared to other tissue types. This limit the proper control reason being,
liver tissue Vs other tissue, ideal control may be liver proteome of chronic HBV
infection and HBV-HCC.
6) Authors have tested the SNPs in two independent HCC case control samples in two
batches, but the details of chronic hepatitis B are missing like viral load, genotype,
mutation, eAg pos/eAg negative, seroconversion, treated/untreated etc.
7) In the first independent samples (sample A) authors screened 21 SNPs and they
selected 2 SNPs (rs17343667 and rs7612684) and in sample B they replicated
potential associations of these two SNPs and another batch of 15 SNPs. It is not clear
why they did not take all SNPs (21+15=36) in sample A and after selection from A,
They could have replicated potential SNPs in cohort B.
8) Authors claim five statistically significant genes including, RNY, GOLGA8M,
LINC01207, WHAMMP2 and SLC39A8 may be relevant to the development
of HCC, however the mechanistic detail function of these genes are missing in
the manuscript and there is no experimental data to prove these findings.
9) The references used in the manuscripts are old; authors should use recent
references for more updated data.