PNPLA3, the history of an orphan gene of the potate tuber protein family that found an organ: the Liver

SILVIA SOOKOIAN; CARLOS JOSE PIROLA

HEPATOLOGY (BALTIMORE, MD.)

JOHN WILEY & SONS INC

Lugar: New York; Año: 2014 vol. 59 p. 2068 - 2071

0270-9139

" To fall in love is to create a religion that has a fallible god" . Jorge Luis Borges 1952. The patatin-like phospholipase domain containing 3 gene (PNPLA3) was an orphan gene looking for a disease until 2008 when the first GWAS on NAFLD showed the association between the risk of liver fat accumulation and the nonsynonymous rs738409 C/G variant (1). Months later, the variant was significantly associated with NAFLD disease severity (2) and ALT levels (3), and the role of PNPLA3 as a modifier of the natural history of NAFLD was unequivocally replicated in different populations around the world, from children to adult (4). The enthusiasm about PNPLA3 led to the search for other biologically plausible liver-disease associations, for instance, alcoholic liver disease (ALD) (5-7) and an increased risk of steatosis in patients with chronic hepatitis C (CHC) (8, 9) and B (10). The variant was not only associated with cirrhosis and the occurrence of hepatocellular carcinoma (HCC) (11), but the G allele was also identified as an independent risk factor for death and poor prognosis (12). At last, PNPLA3 found an organ: the Liver. In this issue, Trepo E et al. (13) explored the strength of the association between the rs738409 and the prevalence of HCC in cirrhotic patients by means of a meta-analysis of 2503 individual participant data (IPD) from Europe. The authors found that the risk allele G was significantly associated with HCC (OR per allele=1.77), after adjusting for sex, age, and obesity. Although the association was also found in patients with HCV-related cirrhosis (OR=1.55), in which the variant showed a small but statistically significant effect on the risk of HCC, the effect was more pronounced among patients with ALD (OR=2.20), suggesting that the variant influence is stronger, if not exclusive, on fat-derived HCC. Altogether, these results suggest some reflections: (a) The rs738409 is the most consistent genetic modifier of the natural history of common chronic liver diseases in which inflammation and fibrogenesis are induced by environmental factors (dietary fat, alcohol, or viruses). (b) The rs738409 is consistently associated with disease progression in liver diseases in which liver fat accumulation matters (NAFLD, ALD, and also CHC). Hence, instead of being ?the cause? of the disease in terms of traditional monogenic disorders, the variant seems to be an important link between environmental stressors and the susceptibility of the liver to develop a more aggressive phenotype. (c) From a theoretical perspective, it can be argued that the rs738409 has per se a substantial role in hepatocarcinogenesis because the variant is involved in related histological outcomes, such as steatosis and inflammation (4). Because the report of Trepo et al. is a cross-sectional study, we cannot rule out the possibility that patients enrolled in this meta-analysis have had fatty liver prior to developing cirrhosis, thus, we do not know whether carriers of the G allele that developed HCC were those who had increased susceptibility to steatosis (in other terms, adjustment by steatosis). Considering that IPD is a unique opportunity to have full access of patients´ record, it would have been interesting to know whether the association between the variant and HCV-HCC remained significant after adjusting for well-known risk factors of HCC development, such as viral load and genotype; undoubtedly, this finding should be confirmed in a better than 0.6 powered sample. Moreover, adjustment by insulin resistance would have been desirable in both cohorts. Also, to be generalized these results should be replicated in non European patients.   Hence, to fall in love with the idea that PNPLA3 is per se involved in hepatocarcinogenesis is to create a hypothesis that has a priori a fallible biological plausibility. We can envision some questions raised by the readers. For example, clinicians may ask:  If I have a patient with cirrhosis who carries the GG genotype, should I make a quick decision to reach an early diagnosis and treatment of HCC?  In other words, are the results from PNPLA3 test clinically applicable? According to Trepo et al., the probability with which HCC will occur in carriers of the G variant is ~64% higher per allele (OR 1.77 converted to absolute risk). In terms of personalized medicine, the rs738409 is still not a feasible predictive test that can accurately discriminate between individuals who will develop HCC and those who will not, because very large ORs (probably over 10-200 to increase the AUC in a ROC curve) are typically needed to improve predictive accuracy above that provided by clinical information. People from the genomic world may ask: Why did previous GWAS on HCC not uncover the PNPLA3 association? Although a comprehensive search of markers encompassing the entire genome was previously performed on the risk of CHC-induced HCC (14, 15), SNPs at PNPLA3 were not identified. There are a number of explanations for this lack of association: (i) small effects for which GWAS are underpowered to detect owing to the large number of tested SNPs and the need of multiple testing correction; (ii) spurious associations; (iii) the ?missing heritability? explained by epistasis and gene-environmental interactions, and so forth. The simplest explanation is that hypothesis-driven candidate gene studies still have a place for the understanding of the genetic architecture of complex traits using approaches such as the one described by Trepo et al. Finally, scientists may wonder whether PNPLA3 has another function in addition to the known triacylglycerol lipase and acylglycerol O-acyltransferase activities that explain a putative role in liver injury and carcinogenesis. The identification of the molecular mechanisms by which the rs738409 is linked to hepatocarcinogenesis is certainly challenging because, disappointedly, the mechanisms by which the variant modulates cancer-linked phenotypes, such as inflammation and fibrogenesis, still remain unknown. In silico analysis of interaction pathways of PNPLA3 suggests that some of the genes on the pathway are involved in the release of arachidonic acid and prostaglandin E2 (Fig. 1), a finding previously associated with the transition from steatosis to NASH (16). Nevertheless, there are many things we do not know; for example, the impact of the variant on chromatin structure, and how these changes might impact on the transcriptional machinery that orchestrates PNPLA3- related network. Putative mechanisms associated with PNPLA3, epigenetic changes and hepatocarcinogenesis are depicted in Fig. 1. Finally, we do not know almost anything about the role of PNPLA3 on the liver function beyond its participation in lipid remodeling. A recent report showed that pnpl3 knocking down in zebrafish decreased the expression of hepatic progenitor cells and significantly impacted on liver growth and global homeostasis (17). This finding allows us to speculate that PNPLA3 might exert a phospholipase D6 (PLD6)-like activity, which may modulate the biogenesis of Piwi-interacting RNAs (piRNAs, small RNAs associated with proteins of the Argonaute family) or even regulate the release of phosphatidic acid on the mitochondrial surface (Fig. 1). In summary, we learned that PNPLA3 is associated with increased liver fat accumulation and, consequently, with inflammation and even fibrosis. We need to re-focus our research on both the physiologic role of the PNPLA3 on the human liver and the pathogenic role of the rs738409 on the progression of liver diseases, including liver cancer.