Keywords
We thank Long et al. for their positive comments on the novelty and the potential clinical impact of our recent study on the use of polygenic risk scores (PRSs) to infer causality between hepatic fat and carcinogenesis, and to predict hepatocellular carcinoma (HCC) development in individuals with dysmetabolism and non-alcoholic fatty liver disease (NAFLD).
[1]
,[2]
As highlighted by Long et al.,
[2]
the PRS prediction for HCC in a large European NAFLD cohort was largely superior to that of the single genetic risk variants. However, the AUC was relatively low (<0.7) and, in fact, we refrained from proposing the clinical use of PRS as a standalone diagnostic test. Additionally, in our study, PRS and classical risk factors were independently associated with HCC and together conferred a significant improvement in diagnostic accuracy compared to each of them alone.[1]
We observed a non-linear relationship between PRS and HCC risk, and in line with previous data, a synergistic effect between PRS and severity of insulin resistance in determining liver disease.
[1]
,[3]
,[4]
Therefore, we identified the optimal thresholds to discriminate HCC risk and validated these thresholds in participants with obesity and type 2 diabetes from the UK Biobank cohort. In this population-based study, PRS alone had and AUROC of 0.70 for detecting HCC. A positive PRS test was observed in approximately 1 in 10 individuals and it was able to predict HCC with 90% specificity. We reasoned that, as part of the ability of a PRS to predict HCC is mediated by the life-long burden of genetic variants on liver damage,[1]
the PRS may be optimal to identify young at-risk individuals for whom preventive measures and surveillance programs can be implemented long before cirrhosis development.[1]
As Long et al. suggested, we now report the fraction of HCC variability explained by genetic predisposition to NAFLD in order to assess the relative burden compared to classical risk factors. The population attributable fraction (PAF) of cirrhosis and HCC accounted for by PRS compared to the single genetic variants in the UK Biobank population-based cohort is shown in Table 1.
[5]
Remarkably, genetic predisposition to NAFLD jointly accounted for 30.39% and 60.95% of cirrhosis and HCC variability, respectively (17.28 and 57.49% for the PRS). Considering HCC, this compares to 58.57% accounted for by male sex, and 33.35% and 41.78% by obesity and type 2 diabetes, respectively. It is important to bear in mind that sex, a major determinant of HCC risk, is also genetically determined, and that adiposity and type 2 diabetes also share a large inherited fraction predisposing to liver disease.[3]
In our paper, the PRS was based on PNPLA3-TM6SF2-GCKR-MBOAT7-HSD17B13 variants and in agreement with Long et al.[2]
we believe the PRS will be improved by the inclusion of novel genetic determinants,[6]
including those recently identified in novel exome-wide association studies.[6]
,[7]
Nevertheless, the estimated combined PAF should be interpreted cautiously, mainly because protective variants (such as HSD17B13) can result in counter-intuitive large PAF estimates. Furthermore, part of the impact of genetic predisposition to NAFLD is mediated through the interaction with environmental risk factors or mediated by dysmetabolic features.Table 1PAF of PRSs of hepatic fat accumulation for HCC and cirrhosis in the UK Biobank cohort (n = 358,342 unrelated European individuals). The PAF of the single genetic variants included in the score and the PAF of classical risk factors are shown as comparison.
| Predictor | Cirrhosis | HCC | ||||||
|---|---|---|---|---|---|---|---|---|
| p-adj ∗ Logistic regression models adjusted for age, sex, BMI, type 2 diabetes, first 10 genomic principal components and genotyping array batch. HCC, hepatocellular carcinoma; PAF, population attributable fraction; PRS-HFC, polygenic risk score hepatic fat content; PRS-5, polygenic risk score of fatty liver disease. | OR | 95% CI | % PAF | p-adj ∗ Logistic regression models adjusted for age, sex, BMI, type 2 diabetes, first 10 genomic principal components and genotyping array batch. HCC, hepatocellular carcinoma; PAF, population attributable fraction; PRS-HFC, polygenic risk score hepatic fat content; PRS-5, polygenic risk score of fatty liver disease. | OR | 95% CI | % PAF | |
| PRS-5 | 1.54E-46 | 3.97 | 3.29-4.79 | 17.28 | 1.40E-32 | 16 | 10.1-25.2 | 57.49 |
| PRS-HFC | 3.96E-42 | 3.73 | 3.09-4.52 | 17.28 | 7.90E-29 | 15 | 9.31-24.1 | 57.49 |
| PRS-5 ≥0.495 | 2.67E-32 | 1.94 | 1.73-2.16 | 8.03 | 3.81E-27 | 3.88 | 3.04-4.97 | 21.87 |
| PRS-HFC ≥0.532 | 1.61E-34 | 1.96 | 1.76-2.19 | 8.43 | 4.59E-26 | 3.77 | 2.95-4.83 | 21.64 |
| rs738409 | 4.79E-23 | 1.51 | 1.39-1.64 | 15.72 | 3.44E-09 | 1.97 | 1.57-2.47 | 26.92 |
| rs58542926 | 1.02E-05 | 1.27 | 1.14-1.42 | 3.77 | 1.98E-06 | 1.88 | 1.45-2.44 | 11.76 |
| rs641738 | 0.0619 | 1.09 | 0.996-1.19 | 6.16 | 0.026 | 1.34 | 1.04-1.73 | 19.25 |
| rs1260326 | 0.252 | 1.05 | 0.965-1.14 | 1.36 | 0.0723 | 1.24 | 0.98-1.58 | 10.49 |
| rs72613567 | 1.81E-03 | 0.877 | 0.808-0.953 | 7.28 | 7.03E-03 | 0.731 | 0.582-0.918 | 16.21 |
| Combined PAF (genetic risk) | 30.39 | 60.95 | ||||||
| Age, >45 years | 4.83E-14 | 2.00 | 1.67-2.39 | 54.85 | 1.17E-04 | 4.38 | 2.07-9.28 | 81.33 |
| Sex, Male | 1.42E-53 | 1.97 | 1.81-2.15 | 37.1 | 2.54E-17 | 3.19 | 2.44-4.17 | 58.57 |
| Obesity, yes | 2.21E-30 | 1.66 | 1.52-1.81 | 23.85 | 2.69E-07 | 1.85 | 1.47-2.35 | 33.35 |
| Type 2 diabetes, yes | 4.70E-167 | 3.97 | 3.6-4.38 | 24.89 | 2.79E-46 | 5.97 | 4.67-7.63 | 41.78 |
∗ Logistic regression models adjusted for age, sex, BMI, type 2 diabetes, first 10 genomic principal components and genotyping array batch. HCC, hepatocellular carcinoma; PAF, population attributable fraction; PRS-HFC, polygenic risk score hepatic fat content; PRS-5, polygenic risk score of fatty liver disease.
Lastly, Long et al. point out that, although the fatty liver PRS was developed in a US general population where it predicted liver damage in African/Hispanic/European Americans,
[8]
it was validated for HCC risk assessment in large multicenter cohorts consisting exclusively of Europeans. This remark is in line with our thought that additional studies will be necessary in multi-ethnic populations prior to clinical implementation of this or similar diagnostic scores. An advantage of our approach is that we chose only the more robust common causal variants of NAFLD already validated in all major ethnicities.[4]
,[9]
Furthermore, the PNPLA3 variant alone accounts for 50% of inter-ethnic variability in the susceptibility to fatty liver.[10]
At the price of reducing the number of considered genetic markers, this choice would theoretically render the instrument more robust in different ethnic groups. However, it is important to note that, as remarked in the manuscript,[1]
in specific populations some rarer variants may be more prevalent or the presence of specific environmental risk factors may modulate the impact of genetic variants.In conclusion, we have shown that genetic predisposition to hepatic fat accumulation accounts for a large fraction of HCC susceptibility. After all, our study represents a proof of concept that will hopefully encourage new research efforts, leading to improvements in the current PRS instruments, validation in multi-ethnic cohorts and optimization of their possible use to improve the prevention and clinical management of liver disease.
[1]
Financial support
This work was supported by project grants from MyFirst Grant AIRC n.16888, Ricerca Finalizzata Ministero della Salute RF-2016-02364358 (“Impact of whole exome sequencing on the clinical management of patients with advanced nonalcoholic fatty liver and cryptogenic liver disease”), Ricerca corrente Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, the European Union (EU) Programme Horizon 2020 (under grant agreement No. 777377) for the project LITMUS- “Liver Investigation: Testing Marker Utility in Steatohepatitis”, Programme “Photonics” under grant agreement “101016726” for the project “REVEAL: Neuronal microscopy for cell behavioural examination and manipulation”, Gilead_IN-IT-989-5790 “Developing a model of care for risk stratification and management of diabetic patients with non-alcoholic fatty liver disease (NAFLD)”, Fondazione IRCCS Ca’ Granda “Liver BIBLE” PR-0391, Fondazione IRCCS Ca’ Granda core COVID-19 Biobank (RC100017A) to LV; Amgen and Sanofi-Aventis, the Swedish Research Council [Vetenskapsrådet (VR), 2016-01527], the Swedish state under the agreement between the Swedish government and the county councils (the ALF-agreement) [SU 2018-04276], the Novo Nordisk Foundation Grant for Excellence in Endocrinology [Excellence Project, 9321-430], the Swedish Diabetes Foundation [DIA 2017-205], the Swedish Heart Lung Foundation [20120533], the Wallenberg Academy Fellows from the Knut and Alice Wallenberg Foundation [KAW 2017.0203] (SR); European Community’s Seventh Framework Programme (FP7/2001-2013) under grant agreement HEALTH-F2-2009-241762 for the project FLIP. Cancer Research UK (CR UK) centre grant C9380/A18084; programme grant C18342/A23390 and Accelerator award C9380/A26813 to HR.
Authors' contributions
All authors were involved in the writing of this commentary and reviewed it prior to submission.
Conflict of interest
The authors declare that they have no conflict of interest relevant to the present study. LV has received speaking fees from MSD, Gilead, AlfaSigma and AbbVie, served as a consultant for Gilead, Pfizer, AstraZeneca, Novo Nordisk, Intercept, Diatech Pharmacogenetics and Ionis Pharmaceuticals, and received research grants from Gilead. SR has served as a consultant for AstraZeneca, Celgene, Sanofi, Amgen, Akcea Therapeutics, Camp4, Ambys, Medacorp, Novartis and Pfizer in the past 5 years, and received research grants from AstraZeneca, Sanofi and Amgen.
Please refer to the accompanying ICMJE disclosure forms for further details.
Supplementary data
The following is the supplementary data to this article:
References
- Non-invasive stratification of hepatocellular carcinoma risk in non-alcoholic fatty liver using polygenic risk scores.J Hepatol. 2020; 74: 775-782
- Polygenic risk score: a promising predictor for hepatocellular carcinoma in the population with nonalcoholic fatty liver disease.J Hepatol. 2021; 74: 1493-1494
- Adiposity amplifies the genetic risk of fatty liver disease conferred by multiple loci.Nat Genet. 2017; 49: 842-847
- Update on NAFLD genetics: from new variants to the clinic.J Hepatol. 2020; 72: 1196-1209
- Meta-analysis of genome-wide association studies identifies ten loci influencing allergic sensitization.Nat Genet. 2013; 45: 902-906
- Exome-wide association study on alanine aminotransferase identifies sequence variants in the GPAM and APOE associated with fatty liver disease.Gastroenterology. 2020;
- Association of genetic variation with cirrhosis: a multi-trait genome-wide association and gene-environment interaction study.Gastroenterology. 2020;
- Causal relationship of hepatic fat with liver damage and insulin resistance in nonalcoholic fatty liver.J Intern Med. 2018; 283: 356-370
- Leveraging human genetics to identify potential new treatments for fatty liver disease.Cell Metab. 2020; 31: 35-45
- Genetic variation in PNPLA3 confers susceptibility to nonalcoholic fatty liver disease.Nat Genet. 2008; 40: 1461-1465
Article info
Publication history
Published online: March 04, 2021
Accepted:
February 25,
2021
Received in revised form:
February 25,
2021
Received:
February 19,
2021
Footnotes
Author names in bold designate shared co-first authorship
Identification
Copyright
© 2021 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved.
