Sage Journals: Discover world-class research

Abstract

Purpose

This study investigated causal relationships and underlying mechanisms between serum DNA repair proteins and liver cancer to identify biomarkers for clinical management.

Methods

A two-sample bidirectional Mendelian randomization (MR) design was employed. We performed inverse variance weighted (IVW) analysis, followed by protein quantitative trait loci (pQTL), colocalization, and pathway enrichment analyses to explore biological mechanisms.

Results

MR analysis revealed significant associations: NBR1 (IVW: OR = 2.28, 95%CI: 1.17–4.45, P = 0.015) and RAD51 (IVW: OR = 2.16, 95%CI: 1.12–4.15, P = 0.021) were risk factors. PARP11 was protective for hepatocellular carcinoma (OR = 0.48, 95%CI: 0.25–0.93, P = 0.030) but a risk for intrahepatic cholangiocarcinoma (OR = 1.86, 95%CI: 1.04–3.33, P = 0.038). Mechanistically, pQTL and colocalization identified rs2793568 as a key regulator of PARP1, which was enriched in base excision repair pathways. Sensitivity analyses confirmed the robustness of these findings.

Conclusion

Serum NBR1, RAD51, and PARP11 are potentially causal in liver cancer pathogenesis. Specifically, the genetic regulation of PARP1 highlights a critical DNA repair mechanism, supporting their utility as predictive biomarkers.

Keywords

Mendelian randomization DNA repair hepatocellular carcinoma biliary tract cancer causality genome-wide association study

Introduction

Liver cancer remains a major global health burden despite advances in cancer detection, molecular classification, and targeted therapy. Worldwide, cancer continues to be a leading cause of morbidity and mortality, with research increasingly focusing on genetic susceptibility, tumor microenvironment, and molecular targets that underlie cancer initiation and progression. Recent progress in precision oncology and high-throughput genomic profiling has revealed complex genetic and epigenetic landscapes across cancers, driving the development of molecularly guided therapies. Nonetheless, hepatocellular carcinoma (HCC) and biliary tract cancers continue to exhibit poor prognoses and limited therapeutic options, underscoring an urgent need for improved understanding of their underlying molecular determinants.^1–3

Liver cancer is the sixth most common cancer, with an estimated 865,350 new cases in 2022 worldwide (4.3% of all cancers), but the third leading cause of cancer-related mortality, with an estimated 757,948 deaths (7.8% of all cancer-related deaths).¹ HCC is a highly lethal invasive carcinoma that accounts for the majority of liver cancers.^2,4 The most important risk factors for HCC include preexisting liver cirrhosis and hepatitis B infection (due to both direct oncogenic effect and its contribution to cirrhosis). Other major risk factors contributing to cirrhosis (and therefore HCC) are hepatitis C infection, alcohol use, and nonalcoholic steatohepatitis.^2,4–6 The 5-year survival rate for HCC remains approximately 18%.⁷ Among non-HCC liver cancers, intrahepatic bile duct carcinoma constitutes a rare, aggressive malignancy (20%-25% of all biliary tract cancers, representing < 1% of all cancers) associated with older age, cholelithiasis, primary sclerosing cholangitis, obesity, and diabetes.^3,8–10

DNA repair is central to eukaryotic life, as mammalian DNA in a single cell can experience approximately 10,000 lesions daily from endogenous and exogenous sources.¹¹ DNA repair pathways preserve genomic integrity by correcting these lesions, enabling accurate gene transmission and normal cellular function.¹² DNA repair factors (i.e., proteins directly involved in these pathways) are critical for genome maintenance.¹³ Improper repair can inactivate tumor suppressor genes or activate oncogenes, promoting malignant transformation.¹⁴ In liver cancer, recurrent genetic alterations involve TERT, TP53, CTNNB1, ARID1A, AXIN1, ARID2, CCND1, and VEGFA.^2,6 In contrast, biliary tract cancers display frequent mutations in KRAS, TP53, and C-ERBB2.¹⁵ However, the potential influence of DNA repair factor expression on liver cancer risk remains insufficiently understood, largely due to the scarcity of large-scale longitudinal data.

To address this research gap, the present study utilized genome-wide association study (GWAS) data to investigate potential causal relationships between genetically predicted serum levels of DNA repair factors and the risk of liver cancer.^16,17 Mendelian randomization (MR) was applied as an analytical framework to infer causality while minimizing confounding and reverse causation.^18–20 This method relies on three assumptions: (1) genetic variants are strongly associated with the exposure, (2) they are independent of confounders, and (3) they affect the outcome only through the exposure.²¹

Therefore, the purpose of this study was to evaluate the causal association between genetically predicted serum DNA repair factors and liver cancer risk using large-scale GWAS data. By elucidating the mechanistic role of DNA repair processes in liver carcinogenesis, this research aimed to provide new molecular insights that may inform early detection strategies and potential therapeutic targets for liver cancer.

Methods

Study design

This study used publicly available data from GWASs to investigate the causal associations between DNA repair factors and liver cancer (Figure 1). Ethical approval was not necessary since the data were from studies that were already in compliance with the Declaration of Helsinki. The assumptions for MR studies are 1) the relevance assumption (i.e., the genetic variant(s) being used as an instrument for the exposure is associated with the exposure), 2) the independence assumption (i.e., there are no common causes of the genetic variant(s) and the outcome of interest), and 3) the no horizontal pleiotropy assumption (i.e., there is no independent pathway between the genetic variant(s) and the outcome other than through the exposure).

Figure 1.

Schematic representation of the study.

Data source

The dataset for HCC, intrahepatic cholangiocarcinoma (IHCC), and malignant neoplasm of the liver was derived from the genomic data of the UK Biobank (UKB), which was used for a previous study. The malignant neoplasm of the liver, primary (PheCode 155.1) (ID GCST90041812) dataset contains 128 patients and 456,220 controls. The intrahepatic bile duct carcinoma (ICD-10 C22.1) (ID GCST90043859) contains 104 cases and 456,244 controls. The HCC (ICD-10 C22.0) (ID GCST90043858) dataset contains 123 cases and 456,225 controls. All individuals were of European ancestry. The exposure data were also from the GWAS catalog: serum levels of protein RAD51 (ID GCST90088117, 5354 individuals), Next to BRCA1 gene 1 (NBR1) protein levels (ID GCST90242103; 3301 individuals), DNA polymerase levels (ID GCST90240928; 3301 individuals), DNA repair protein XRCC4 levels (ID GCST90240935; 3301 individuals), A/G-specific adenine DNA glycosylase levels (ID GCST90240167; 3301 individuals), DNA repair endonuclease XPF measurement (ID GCST90240931; 3301 individuals), Poly (ADP-ribose) polymerase 1 (PARP1) levels (ID GCST90242303; 3301 individuals), and Poly (ADP-ribose) polymerase 11 (parp11) levels (ID GCST90242304; 3301 individuals). Those DNA repair factors were selected because they are available in GWAS datasets.

Instrumental variable selection

The instrumental variables (IVs) included in this study had to meet the following three criteria. 1) The entire genome was screened for SNPs significantly associated with the exposure, i.e., selecting SNPs associated with each exposure at P < 5 × 10⁻⁸, but that threshold yielded too few SNPs to allow an MR analysis, and the threshold had to be relaxed to P < 5 × 10⁻⁶.^22,23 Second, SNPs with a minor allele frequency (MAF) > 0.01 were selected. Third, linkage disequilibrium (LD) between SNPs was removed based on R² < 0.001 and a window size of 10,000 kb.²⁴ Hence, SNPs located within 10,000 kb of each other in the genome sequence and showing a strong association (R² > 0.001) were excluded. If a SNP associated with an exposure was not found in the outcome summary data, SNPs with high LD (R² > 0.8) with the original SNP were selected as a proxy. Finally, to ensure the absence of weak instrumental bias, the F-value was calculated for each SNP using F = R² × (N-2)/(1-R²). F-values > 10 indicated the absence of weak instrumental bias and that the SNP could be kept for analysis.

Mendelian randomization analysis

This primary analysis to assess the causal association between exposure and outcome risk was the inverse variance weighted (IVW) method, which provides results as odds ratios (ORs) and 95% confidence intervals (CIs). Other MR methods were used to test the robustness of the IVW results, including the MR-Egger, weighted median, and weighted mode methods. Statistically significant IVW results, accompanied by statistically significant results from the other tests, were considered robust; otherwise, they were considered with caution. A causal association shown by IVW and not by the other methods, or by one of the other methods and not by IVW, had to be taken cautiously. All analyses were carried out using the “TwoSampleMR” package in R 4.3.1 (The R Project for Statistical Computing, www.r-project.org). The results were depicted using scatter and forest plots.

Sensitivity analysis

Potential heterogeneity among IVs was detected using Cochran's Q test, which was used to assess heterogeneity. A P-value > 0.05 indicated low heterogeneity, meaning that the estimates among IVs were randomly distributed and had little impact on the IVW results. Due to the risk of genetic variation heterogeneity influencing the estimation of association effects, the MR-Egger regression method was employed to detect the presence of horizontal pleiotropy. The absence of heterogeneity was considered when the intercept term of the MR-Egger regression tended toward zero or lacked statistical significance. In addition, an MR pleiotropy residual sum and outlier (MR-PRESSO) analysis was used to detect potential outliers (i.e., SNPs with P < 0.05). Such SNPs were removed to recalibrate the causal associations and correct for horizontal pleiotropy. A leave-one-out analysis was used to assess the robustness and consistency of the results.

Protein quantitative trait loci

A protein quantitative trait loci (pQTL) analysis was performed to determine whether SNPs could be associated with variations in protein levels or expression. Using IVW MR, DNA repair proteins significantly associated with liver cancer (P < 5 × 10⁻⁸) were identified. After extracting the corresponding SNPs and cross-referencing with the plasma pQTL database,²⁵ only the PARP1-related loci with a consistent direction and P < 0.05 were retained. Bayesian colocalization analysis (COLOC, PPH4 ≥ 0.8) confirmed that the PARP1 locus shared causal variants with liver cancer risk. Finally, PARP1 was imported into the KEGG pathway enrichment analysis to determine whether it drove liver cancer development through the base excision repair pathway (hsa03410).

Bioinformatics

The proteins identified as causally associated with liver cancer were investigated using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) annotation analyses.

Results

Instrument variable selection

Our study performed an extensive series of MR analyses, focusing on diverse exposures implicated in DNA damage and repair pathways, using liver cancer as the primary outcome of interest. First, we identified IV sets with varied potency corresponding to each exposure. All F-values were > 10, affirming the exclusion of weak instrumental biases across all selected IVs, thereby reinforcing the reliability of our investigation into the intricate dynamics connecting DNA integrity, its restoration, and the onset of liver cancer (Supplementary Table 1).

Mendelian randomization analysis results

Then, the IVs identified for each exposure were tested for association with the outcomes. The genetic prediction results suggested a causal association between Next to BRCA1 gene 1 protein levels and primary malignant neoplasm of the liver (IVW: OR = 2.28, 95%CI: 1.17–4.45, P = 0.015; weighted median: P = 0.015), between serum levels of protein RAD51 and primary malignant neoplasm of the liver (IVW: OR = 2.16, 95%CI: 1.12–4.15, P = 0.021), between Poly (ADP-ribose) polymerase (PARP) 11 levels and HCC (IVW: OR = 0.48, 95%CU: 0.25–0.93, P = 0.030), and between PARP11 levels and intrahepatic bile duct carcinoma (IVW: OR = 1.86, 95%CI: 1.04–3.33, P = 0.038).

Then, cancers were used as the exposure, and DNA repair factor levels were used as the outcomes. The reverse analysis suggested causal associations between HCC and PARP1 levels (IVW: OR = 0.96, 95%CI: 0.93–0.99, P = 0.007) and between intrahepatic bile duct carcinoma and DNA repair protein XRCC4 levels (IVW: OR = 1.05, 95%CI: 1.02–1.09, P = 0.003) and Poly (ADP-ribose) polymerase 1 levels (IVW: OR = 1.05, 95%CI: 1.01–1.08, P = 0.005 (Table 1 and Figures 2 and 3). The remaining results showed no significant correlations. (All P > 0.05).

Figure 2.

Scatter plots for the Mendelian randomization analysis of (A) next to BRCA1 gene 1 protein (NBR1) levels as exposure and primary malignant neoplasm of the liver as the outcome, (B) serum RAD51 recombinase (RAD51) protein levels as exposure and primary malignant neoplasm of the liver as the outcome, (C) poly(ADP-ribose) polymerase 11 (PARP11) levels as exposure and hepatocellular carcinoma as the outcome, (D) PARP11 levels as exposure and intrahepatic bile duct carcinoma as the outcome, (E) hepatocellular carcinoma as exposure and poly(ADP-ribose) polymerase 1 (PARP1) levels as the outcome, (F) intrahepatic bile duct carcinoma as exposure and DNA repair protein X-ray repair cross complementing 4 (XRCC4) levels as the outcome, and (G) intrahepatic bile duct carcinoma as exposure and PARP1 levels as the outcome.

Figure 3.

Forest plots for the Mendelian randomization analysis of (A) next to BRCA1 gene 1 protein (NBR1) levels as exposure and primary malignant neoplasm of the liver as the outcome, (B) serum RAD51 recombinase (RAD51) protein levels as exposure and primary malignant neoplasm of the liver as the outcome, (C) poly(ADP-ribose) polymerase 11 (PARP11) levels as exposure and hepatocellular carcinoma as the outcome, (D) PARP11 levels as exposure and intrahepatic bile duct carcinoma as the outcome, (E) hepatocellular carcinoma as exposure and poly(ADP-ribose) polymerase 1 (PARP1) levels as the outcome, (F) intrahepatic bile duct carcinoma as exposure and DNA repair protein X-ray repair cross complementing 4 (XRCC4) levels as the outcome, and (G) intrahepatic bile duct carcinoma as exposure and PARP1 levels as the outcome.

Table 1.

Mendelian randomization (mr) results of the causal associations between liver cancer and DNA repair-related genes.

Exposure	Outcome	Method	N_snp	OR (95% CI)	P
Next to BRCA1 gene 1protein levels	Malignant neoplasm of liver, primary	Inverse variance weighted	7	2.28 (1.17–4.45)	0.015
Next to BRCA1 gene 1protein levels	Malignant neoplasm of liver, primary	MR Egger	7	5.50 (0.68–44.47)	0.171
Next to BRCA1 gene 1protein levels	Malignant neoplasm of liver, primary	Weighted median	7	3.03 (1.24–7.44)	0.015
Next to BRCA1 gene 1protein levels	Malignant neoplasm of liver, primary	Weighted mode	7	3.13 (1.18–8.31)	0.062
Next to BRCA1 gene 1protein levels	Hepatocellular carcinoma	Inverse variance weighted	7	1.02 (0.52–2.01)	0.959
Next to BRCA1 gene 1protein levels	Hepatocellular carcinoma	MR Egger	7	3.44 (0.42–27.86)	0.299
Next to BRCA1 gene 1protein levels	Hepatocellular carcinoma	Weighted median	7	1.18 (0.52–2.68)	0.695
Next to BRCA1 gene 1protein levels	Hepatocellular carcinoma	Weighted mode	7	1.10 (0.43–2.85)	0.844
Next to BRCA1 gene 1protein levels	Intrahepatic bile duct carcinoma	Inverse variance weighted	7	1.18 (0.51–2.73)	0.701
Next to BRCA1 gene 1protein levels	Intrahepatic bile duct carcinoma	MR Egger	7	0.80 (0.05–13.36)	0.885
Next to BRCA1 gene 1protein levels	Intrahepatic bile duct carcinoma	Weighted median	7	1.61 (0.60–4.33)	0.345
Next to BRCA1 gene 1protein levels	Intrahepatic bile duct carcinoma	Weighted mode	7	1.61 (0.58–4.49)	0.396
AG-specific adenine DNA glycosylase levels	Malignant neoplasm of liver, primary	Inverse variance weighted	9	1.57 (0.68–3.58)	0.288
AG-specific adenine DNA glycosylase levels	Malignant neoplasm of liver, primary	MR Egger	9	4.12 (0.77–22.19)	0.143
AG-specific adenine DNA glycosylase levels	Malignant neoplasm of liver, primary	Weighted median	9	1.03 (0.37–2.84)	0.954
AG-specific adenine DNA glycosylase levels	Malignant neoplasm of liver, primary	Weighted mode	9	1.07 (0.27–4.17)	0.924
AG-specific adenine DNA glycosylase levels	Hepatocellular carcinoma	Inverse variance weighted	9	1.32 (0.60–2.92)	0.492
AG-specific adenine DNA glycosylase levels	Hepatocellular carcinoma	MR Egger	9	0.93 (0.16–5.46)	0.941
AG-specific adenine DNA glycosylase levels	Hepatocellular carcinoma	Weighted median	9	0.94 (0.36–2.45)	0.893
AG-specific adenine DNA glycosylase levels	Hepatocellular carcinoma	Weighted mode	9	0.77 (0.20–2.95)	0.715
AG-specific adenine DNA glycosylase levels	Intrahepatic bile duct carcinoma	Inverse variance weighted	9	1.37 (0.61–3.07)	0.438
AG-specific adenine DNA glycosylase levels	Intrahepatic bile duct carcinoma	MR Egger	9	0.65 (0.12–3.58)	0.634
AG-specific adenine DNA glycosylase levels	Intrahepatic bile duct carcinoma	Weighted median	9	2.06 (0.72–5.88)	0.178
AG-specific adenine DNA glycosylase levels	Intrahepatic bile duct carcinoma	Weighted mode	9	2.42 (0.59–9.97)	0.256
DNA polymerase levels	Malignant neoplasm of liver, primary	Inverse variance weighted	13	1.13 (0.59–2.17)	0.703
DNA polymerase levels	Malignant neoplasm of liver, primary	MR Egger	13	1.34 (0.25–7.10)	0.736
DNA polymerase levels	Malignant neoplasm of liver, primary	Weighted median	13	1.01 (0.43–2.36)	0.987
DNA polymerase levels	Malignant neoplasm of liver, primary	Weighted mode	13	1.04 (0.29–3.70)	0.958
DNA polymerase levels	Hepatocellular carcinoma	Inverse variance weighted	13	0.81 (0.43–1.50)	0.499
DNA polymerase levels	Hepatocellular carcinoma	MR Egger	13	0.96 (0.21–4.36)	0.955
DNA polymerase levels	Hepatocellular carcinoma	Weighted median	13	0.69 (0.32–1.50)	0.354
DNA polymerase levels	Hepatocellular carcinoma	Weighted mode	13	0.69 (0.21–2.20)	0.537
DNA polymerase levels	Intrahepatic bile duct carcinoma	Inverse variance weighted	13	1.10 (0.51–2.38)	0.800
DNA polymerase levels	Intrahepatic bile duct carcinoma	MR Egger	13	1.29 (0.18–9.09)	0.803
DNA polymerase levels	Intrahepatic bile duct carcinoma	Weighted median	13	1.54 (0.55–4.34)	0.414
DNA polymerase levels	Intrahepatic bile duct carcinoma	Weighted mode	13	0.31 (0.04–2.60)	0.304
DNA repair protein XRCC4 levels	Malignant neoplasm of liver, primary	Inverse variance weighted	10	1.30 (0.72–2.35)	0.392
DNA repair protein XRCC4 levels	Malignant neoplasm of liver, primary	MR Egger	10	0.77 (0.13–4.65)	0.786
DNA repair protein XRCC4 levels	Malignant neoplasm of liver, primary	Weighted median	10	1.33 (0.61–2.87)	0.474
DNA repair protein XRCC4 levels	Malignant neoplasm of liver, primary	Weighted mode	10	1.29 (0.52–3.21)	0.591
DNA repair protein XRCC4 levels	Hepatocellular carcinoma	Inverse variance weighted	10	1.45 (0.79–2.66)	0.229
DNA repair protein XRCC4 levels	Hepatocellular carcinoma	MR Egger	10	1.16 (0.19–7.24)	0.877
DNA repair protein XRCC4 levels	Hepatocellular carcinoma	Weighted median	10	1.38 (0.59–3.23)	0.459
DNA repair protein XRCC4 levels	Hepatocellular carcinoma	Weighted mode	10	1.60 (0.57–4.47)	0.392
DNA repair protein XRCC4 levels	Intrahepatic bile duct carcinoma	Inverse variance weighted	10	1.07 (0.55–2.06)	0.842
DNA repair protein XRCC4 levels	Intrahepatic bile duct carcinoma	MR Egger	10	1.03 (0.13–8.18)	0.975
DNA repair protein XRCC4 levels	Intrahepatic bile duct carcinoma	Weighted median	10	0.98 (0.40–2.40)	0.967
DNA repair protein XRCC4 levels	Intrahepatic bile duct carcinoma	Weighted mode	10	0.67 (0.25–1.79)	0.446
DNA repair endonuclease XPF measurement	Malignant neoplasm of liver, primary	Inverse variance weighted	10	0.54 (0.26–1.14)	0.108
DNA repair endonuclease XPF measurement	Malignant neoplasm of liver, primary	MR Egger	10	0.83 (0.17–4.09)	0.822
DNA repair endonuclease XPF measurement	Malignant neoplasm of liver, primary	Weighted median	10	0.39 (0.16–0.96)	0.041
DNA repair endonuclease XPF measurement	Malignant neoplasm of liver, primary	Weighted mode	10	0.31 (0.07–1.35)	0.152
DNA repair endonuclease XPF measurement	Hepatocellular carcinoma	Inverse variance weighted	10	0.71 (0.32–1.58)	0.403
DNA repair endonuclease XPF measurement	Hepatocellular carcinoma	MR Egger	10	0.44 (0.08–2.40)	0.371
DNA repair endonuclease XPF measurement	Hepatocellular carcinoma	Weighted median	10	0.60 (0.21–1.68)	0.331
DNA repair endonuclease XPF measurement	Hepatocellular carcinoma	Weighted mode	10	0.26 (0.04–1.80)	0.207
DNA repair endonuclease XPF measurement	Intrahepatic bile duct carcinoma	Inverse variance weighted	10	1.33 (0.63–2.84)	0.455
DNA repair endonuclease XPF measurement	Intrahepatic bile duct carcinoma	MR Egger	10	1.15 (0.22–5.98)	0.875
DNA repair endonuclease XPF measurement	Intrahepatic bile duct carcinoma	Weighted median	10	1.43 (0.52–3.91)	0.485
DNA repair endonuclease XPF measurement	Intrahepatic bile duct carcinoma	Weighted mode	10	2.59 (0.48–13.88)	0.295
Serum levels of protein RAD51	Malignant neoplasm of liver, primary	Inverse variance weighted	17	2.16 (1.12–4.15)	0.021
Serum levels of protein RAD51	Malignant neoplasm of liver, primary	MR Egger	17	0.81 (0.19–3.47)	0.782
Serum levels of protein RAD51	Malignant neoplasm of liver, primary	Weighted median	17	2.08 (0.88–4.89)	0.095
Serum levels of protein RAD51	Malignant neoplasm of liver, primary	Weighted mode	17	1.84 (0.67–5.05)	0.254
Serum levels of protein RAD51	Hepatocellular carcinoma	Inverse variance weighted	17	1.56 (0.72–3.39)	0.262
Serum levels of protein RAD51	Hepatocellular carcinoma	MR Egger	17	1.22 (0.19–7.72)	0.833
Serum levels of protein RAD51	Hepatocellular carcinoma	Weighted median	17	1.51 (0.59–3.86)	0.391
Serum levels of protein RAD51	Hepatocellular carcinoma	Weighted mode	17	2.07 (0.59–7.23)	0.272
Serum levels of protein RAD51	Intrahepatic bile duct carcinoma	Inverse variance weighted	17	1.85 (0.93–3.68)	0.078
Serum levels of protein RAD51	Intrahepatic bile duct carcinoma	MR Egger	17	1.43 (0.28–7.25)	0.673
Serum levels of protein RAD51	Intrahepatic bile duct carcinoma	Weighted median	17	1.52 (0.57–4.07)	0.406
Serum levels of protein RAD51	Intrahepatic bile duct carcinoma	Weighted mode	17	1.29 (0.39–4.23)	0.681
Poly [ADP-ribose] polymerase 1 levels	Malignant neoplasm of liver, primary	Inverse variance weighted	7	0.71 (0.32–1.54)	0.384
Poly [ADP-ribose] polymerase 1 levels	Malignant neoplasm of liver, primary	MR Egger	7	0.74 (0.11–4.91)	0.765
Poly [ADP-ribose] polymerase 1 levels	Malignant neoplasm of liver, primary	Weighted median	7	0.64 (0.24–1.73)	0.379
Poly [ADP-ribose] polymerase 1 levels	Malignant neoplasm of liver, primary	Weighted mode	7	0.61 (0.16–2.25)	0.481
Poly [ADP-ribose] polymerase 1 levels	Hepatocellular carcinoma	Inverse variance weighted	7	0.81 (0.37–1.81)	0.616
Poly [ADP-ribose] polymerase 1 levels	Hepatocellular carcinoma	MR Egger	7	0.32 (0.05–2.21)	0.300
Poly [ADP-ribose] polymerase 1 levels	Hepatocellular carcinoma	Weighted median	7	1.02 (0.36–2.91)	0.963
Poly [ADP-ribose] polymerase 1 levels	Hepatocellular carcinoma	Weighted mode	7	1.09 (0.25–4.73)	0.913
Poly [ADP-ribose] polymerase 1 levels	Intrahepatic bile duct carcinoma	Inverse variance weighted	7	0.75 (0.32–1.79)	0.524
Poly [ADP-ribose] polymerase 1 levels	Intrahepatic bile duct carcinoma	MR Egger	7	1.34 (0.16–10.94)	0.794
Poly [ADP-ribose] polymerase 1 levels	Intrahepatic bile duct carcinoma	Weighted median	7	0.51 (0.17–1.51)	0.223
Poly [ADP-ribose] polymerase 1 levels	Intrahepatic bile duct carcinoma	Weighted mode	7	0.51 (0.11–2.24)	0.403
Poly [ADP-ribose] polymerase 11 levels	Malignant neoplasm of liver, primary	Inverse variance weighted	16	0.57 (0.30–1.08)	0.084
Poly [ADP-ribose] polymerase 11 levels	Malignant neoplasm of liver, primary	MR Egger	16	0.52 (0.09–2.94)	0.470
Poly [ADP-ribose] polymerase 11 levels	Malignant neoplasm of liver, primary	Weighted median	16	0.62 (0.28–1.34)	0.224
Poly [ADP-ribose] polymerase 11 levels	Malignant neoplasm of liver, primary	Weighted mode	16	0.65 (0.17–2.41)	0.529
Poly [ADP-ribose] polymerase 11 levels	Hepatocellular carcinoma	Inverse variance weighted	16	0.48 (0.25–0.93)	0.030
Poly [ADP-ribose] polymerase 11 levels	Hepatocellular carcinoma	MR Egger	16	0.28 (0.05–1.65)	0.180
Poly [ADP-ribose] polymerase 11 levels	Hepatocellular carcinoma	Weighted median	16	0.44 (0.20–0.95)	0.038
Poly [ADP-ribose] polymerase 11 levels	Hepatocellular carcinoma	Weighted mode	16	0.34 (0.08–1.48)	0.171
Poly [ADP-ribose] polymerase 11 levels	Intrahepatic bile duct carcinoma	Inverse variance weighted	16	1.86 (1.04–3.33)	0.038
Poly [ADP-ribose] polymerase 11 levels	Intrahepatic bile duct carcinoma	MR Egger	16	0.71 (0.15–3.37)	0.676
Poly [ADP-ribose] polymerase 11 levels	Intrahepatic bile duct carcinoma	Weighted median	16	1.58 (0.71–3.53)	0.267
Poly [ADP-ribose] polymerase 11 levels	Intrahepatic bile duct carcinoma	Weighted mode	16	1.24 (0.27–5.69)	0.788
Hepatocellular carcinoma	Serum levels of protein RAD51	Inverse variance weighted	5	1.00 (0.98–1.02)	0.837
Hepatocellular carcinoma	Serum levels of protein RAD51	MR Egger	5	1.02 (0.95–1.09)	0.622
Hepatocellular carcinoma	Serum levels of protein RAD51	Weighted median	5	1.00 (0.97–1.03)	0.943
Hepatocellular carcinoma	Serum levels of protein RAD51	Weighted mode	5	1.00 (0.97–1.03)	0.967
Hepatocellular carcinoma	DNA polymerase levels	Inverse variance weighted	5	1.02 (0.98–1.06)	0.455
Hepatocellular carcinoma	DNA polymerase levels	MR Egger	5	1.10 (0.99–1.21)	0.169
Hepatocellular carcinoma	DNA polymerase levels	Weighted median	5	1.01 (0.97–1.06)	0.599
Hepatocellular carcinoma	DNA polymerase levels	Weighted mode	5	1.00 (0.95–1.06)	0.866
Hepatocellular carcinoma	DNA repair protein XRCC4 levels	Inverse variance weighted	5	0.99 (0.96–1.03)	0.626
Hepatocellular carcinoma	DNA repair protein XRCC4 levels	MR Egger	5	1.03 (0.93–1.14)	0.649
Hepatocellular carcinoma	DNA repair protein XRCC4 levels	Weighted median	5	0.99 (0.95–1.03)	0.683
Hepatocellular carcinoma	DNA repair protein XRCC4 levels	Weighted mode	5	0.97 (0.91–1.03)	0.363
Hepatocellular carcinoma	DNA repair endonuclease XPF measurement	Inverse variance weighted	5	1.00 (0.97–1.03)	0.975
Hepatocellular carcinoma	DNA repair endonuclease XPF measurement	MR Egger	5	0.99 (0.90–1.08)	0.762
Hepatocellular carcinoma	DNA repair endonuclease XPF measurement	Weighted median	5	0.99 (0.96–1.03)	0.756
Hepatocellular carcinoma	DNA repair endonuclease XPF measurement	Weighted mode	5	0.99 (0.95–1.04)	0.798
Hepatocellular carcinoma	Poly [ADP-ribose] polymerase 1 levels	Inverse variance weighted	5	0.96 (0.93–0.99)	0.007
Hepatocellular carcinoma	Poly [ADP-ribose] polymerase 1 levels	MR Egger	5	1.04 (0.95–1.13)	0.459
Hepatocellular carcinoma	Poly [ADP-ribose] polymerase 1 levels	Weighted median	5	0.96 (0.93–1.00)	0.052
Hepatocellular carcinoma	Poly [ADP-ribose] polymerase 1 levels	Weighted mode	5	0.99 (0.92–1.05)	0.688
Hepatocellular carcinoma	Poly [ADP-ribose] polymerase 11 levels	Inverse variance weighted	5	1.03 (1.00–1.06)	0.056
Hepatocellular carcinoma	Poly [ADP-ribose] polymerase 11 levels	MR Egger	5	1.03 (0.94–1.12)	0.611
Hepatocellular carcinoma	Poly [ADP-ribose] polymerase 11 levels	Weighted median	5	1.03 (0.99–1.06)	0.187
Hepatocellular carcinoma	Poly [ADP-ribose] polymerase 11 levels	Weighted mode	5	1.02 (0.98–1.07)	0.383
Hepatocellular carcinoma	AG-specific adenine DNA glycosylase levels	Inverse variance weighted	5	1.00 (0.97–1.03)	0.880
Hepatocellular carcinoma	AG-specific adenine DNA glycosylase levels	MR Egger	5	1.01 (0.92–1.10)	0.906
Hepatocellular carcinoma	AG-specific adenine DNA glycosylase levels	Weighted median	5	1.01 (0.97–1.05)	0.637
Hepatocellular carcinoma	AG-specific adenine DNA glycosylase levels	Weighted mode	5	1.01 (0.96–1.06)	0.734
Hepatocellular carcinoma	Next to BRCA1 gene 1protein levels	Inverse variance weighted	5	0.99 (0.96–1.02)	0.428
Hepatocellular carcinoma	Next to BRCA1 gene 1protein levels	MR Egger	5	0.99 (0.90–1.08)	0.806
Hepatocellular carcinoma	Next to BRCA1 gene 1protein levels	Weighted median	5	0.98 (0.94–1.01)	0.212
Hepatocellular carcinoma	Next to BRCA1 gene 1protein levels	Weighted mode	5	0.97 (0.92–1.02)	0.352
Intrahepatic bile duct carcinoma	Serum levels of protein RAD51	Inverse variance weighted	2	0.98 (0.94–1.02)	0.389
Intrahepatic bile duct carcinoma	DNA polymerase levels	Inverse variance weighted	3	1.02 (0.99–1.06)	0.226
Intrahepatic bile duct carcinoma	DNA polymerase levels	MR Egger	3	1.02 (0.95–1.08)	0.710
Intrahepatic bile duct carcinoma	DNA polymerase levels	Weighted median	3	1.03 (0.99–1.07)	0.200
Intrahepatic bile duct carcinoma	DNA polymerase levels	Weighted mode	3	1.03 (0.98–1.07)	0.358
Intrahepatic bile duct carcinoma	DNA repair protein XRCC4 levels	Inverse variance weighted	3	1.05 (1.02–1.09)	0.003
Intrahepatic bile duct carcinoma	DNA repair protein XRCC4 levels	MR Egger	3	1.05 (0.98–1.13)	0.397
Intrahepatic bile duct carcinoma	DNA repair protein XRCC4 levels	Weighted median	3	1.05 (1.00–1.09)	0.041
Intrahepatic bile duct carcinoma	DNA repair protein XRCC4 levels	Weighted mode	3	1.04 (0.98–1.10)	0.318
Intrahepatic bile duct carcinoma	DNA repair endonuclease XPF measurement	Inverse variance weighted	3	1.01 (0.98–1.04)	0.606
Intrahepatic bile duct carcinoma	DNA repair endonuclease XPF measurement	MR Egger	3	0.99 (0.93–1.05)	0.791
Intrahepatic bile duct carcinoma	DNA repair endonuclease XPF measurement	Weighted median	3	1.00 (0.97–1.04)	0.839
Intrahepatic bile duct carcinoma	DNA repair endonuclease XPF measurement	Weighted mode	3	1.00 (0.96–1.05)	0.940
Intrahepatic bile duct carcinoma	Poly [ADP-ribose] polymerase 1 levels	Inverse variance weighted	3	1.05 (1.01–1.08)	0.005
Intrahepatic bile duct carcinoma	Poly [ADP-ribose] polymerase 1 levels	MR Egger	3	1.03 (0.97–1.10)	0.508
Intrahepatic bile duct carcinoma	Poly [ADP-ribose] polymerase 1 levels	Weighted median	3	1.05 (1.00–1.09)	0.031
Intrahepatic bile duct carcinoma	Poly [ADP-ribose] polymerase 1 levels	Weighted mode	3	1.05 (1.00–1.10)	0.199
Intrahepatic bile duct carcinoma	Poly [ADP-ribose] polymerase 11 levels	Inverse variance weighted	3	1.00 (0.96–1.03)	0.882
Intrahepatic bile duct carcinoma	Poly [ADP-ribose] polymerase 11 levels	MR Egger	3	1.00 (0.94–1.07)	0.990
Intrahepatic bile duct carcinoma	Poly [ADP-ribose] polymerase 11 levels	Weighted median	3	0.99 (0.96–1.03)	0.785
Intrahepatic bile duct carcinoma	Poly [ADP-ribose] polymerase 11 levels	Weighted mode	3	0.99 (0.94–1.04)	0.716
Intrahepatic bile duct carcinoma	AG-specific adenine DNA glycosylase levels	Inverse variance weighted	3	1.01 (0.98–1.05)	0.438
Intrahepatic bile duct carcinoma	AG-specific adenine DNA glycosylase levels	MR Egger	3	0.98 (0.92–1.05)	0.680
Intrahepatic bile duct carcinoma	AG-specific adenine DNA glycosylase levels	Weighted median	3	1.00 (0.97–1.04)	0.839
Intrahepatic bile duct carcinoma	AG-specific adenine DNA glycosylase levels	Weighted mode	3	1.00 (0.95–1.05)	0.934
Intrahepatic bile duct carcinoma	Next to BRCA1 gene 1protein levels	Inverse variance weighted	3	1.01 (0.96–1.06)	0.679
Intrahepatic bile duct carcinoma	Next to BRCA1 gene 1protein levels	MR Egger	3	1.01 (0.89–1.15)	0.904
Intrahepatic bile duct carcinoma	Next to BRCA1 gene 1protein levels	Weighted median	3	1.01 (0.96–1.06)	0.756
Intrahepatic bile duct carcinoma	Next to BRCA1 gene 1protein levels	Weighted mode	3	0.99 (0.93–1.06)	0.850
Malignant neoplasm of liver, primary	Serum levels of protein RAD51	Inverse variance weighted	5	1.00 (0.98–1.03)	0.893
Malignant neoplasm of liver, primary	Serum levels of protein RAD51	MR Egger	5	0.99 (0.94–1.05)	0.863
Malignant neoplasm of liver, primary	Serum levels of protein RAD51	Weighted median	5	1.00 (0.97–1.04)	0.813
Malignant neoplasm of liver, primary	Serum levels of protein RAD51	Weighted mode	5	1.01 (0.97–1.05)	0.745
Malignant neoplasm of liver, primary	DNA polymerase levels	Inverse variance weighted	4	1.03 (0.99–1.07)	0.155
Malignant neoplasm of liver, primary	DNA polymerase levels	MR Egger	4	0.96 (0.87–1.06)	0.504
Malignant neoplasm of liver, primary	DNA polymerase levels	Weighted median	4	1.03 (0.99–1.08)	0.161
Malignant neoplasm of liver, primary	DNA polymerase levels	Weighted mode	4	1.05 (0.99–1.12)	0.179
Malignant neoplasm of liver, primary	DNA repair protein XRCC4 levels	Inverse variance weighted	4	0.99 (0.96–1.02)	0.505
Malignant neoplasm of liver, primary	DNA repair protein XRCC4 levels	MR Egger	4	0.99 (0.90–1.09)	0.819
Malignant neoplasm of liver, primary	DNA repair protein XRCC4 levels	Weighted median	4	0.99 (0.95–1.03)	0.564
Malignant neoplasm of liver, primary	DNA repair protein XRCC4 levels	Weighted mode	4	0.99 (0.95–1.05)	0.861
Malignant neoplasm of liver, primary	DNA repair endonuclease XPF measurement	Inverse variance weighted	4	0.99 (0.95–1.02)	0.423
Malignant neoplasm of liver, primary	DNA repair endonuclease XPF measurement	MR Egger	4	0.94 (0.85–1.03)	0.320
Malignant neoplasm of liver, primary	DNA repair endonuclease XPF measurement	Weighted median	4	0.98 (0.95–1.02)	0.391
Malignant neoplasm of liver, primary	DNA repair endonuclease XPF measurement	Weighted mode	4	0.97 (0.92–1.03)	0.374
Malignant neoplasm of liver, primary	Poly [ADP-ribose] polymerase 1 levels	Inverse variance weighted	4	1.01 (0.98–1.05)	0.557
Malignant neoplasm of liver, primary	Poly [ADP-ribose] polymerase 1 levels	MR Egger	4	1.08 (0.98–1.19)	0.276
Malignant neoplasm of liver, primary	Poly [ADP-ribose] polymerase 1 levels	Weighted median	4	1.01 (0.97–1.05)	0.745
Malignant neoplasm of liver, primary	Poly [ADP-ribose] polymerase 1 levels	Weighted mode	4	1.00 (0.95–1.06)	0.883
Malignant neoplasm of liver, primary	Poly [ADP-ribose] polymerase 11 levels	Inverse variance weighted	4	1.01 (0.97–1.04)	0.625
Malignant neoplasm of liver, primary	Poly [ADP-ribose] polymerase 11 levels	MR Egger	4	0.94 (0.85–1.03)	0.330
Malignant neoplasm of liver, primary	Poly [ADP-ribose] polymerase 11 levels	Weighted median	4	1.00 (0.96–1.04)	0.983
Malignant neoplasm of liver, primary	Poly [ADP-ribose] polymerase 11 levels	Weighted mode	4	0.99 (0.94–1.05)	0.796
Malignant neoplasm of liver, primary	AG-specific adenine DNA glycosylase levels	Inverse variance weighted	4	1.01 (0.98–1.05)	0.476
Malignant neoplasm of liver, primary	AG-specific adenine DNA glycosylase levels	MR Egger	4	0.99 (0.90–1.09)	0.832
Malignant neoplasm of liver, primary	AG-specific adenine DNA glycosylase levels	Weighted median	4	1.00 (0.97–1.04)	0.823
Malignant neoplasm of liver, primary	AG-specific adenine DNA glycosylase levels	Weighted mode	4	1.00 (0.95–1.05)	0.958
Malignant neoplasm of liver, primary	Next to BRCA1 gene 1protein levels	Inverse variance weighted	4	1.00 (0.97–1.04)	0.892
Malignant neoplasm of liver, primary	Next to BRCA1 gene 1protein levels	MR Egger	4	0.97 (0.88–1.07)	0.584
Malignant neoplasm of liver, primary	Next to BRCA1 gene 1protein levels	Weighted median	4	1.01 (0.97–1.05)	0.643
Malignant neoplasm of liver, primary	Next to BRCA1 gene 1protein levels	Weighted mode	4	1.02 (0.96–1.07)	0.626

Sensitivity analyses

Heterogeneity and horizontal pleiotropy can affect MR results, and their presence was investigated. No heterogeneity or pleiotropy was observed (all P > 0.05) (Table 2 and Figure 4). Outliers and single SNPs driving the results can also affect the ME results. The MR-PRESSO analysis revealed no outliers (Table 3). The leave-one-out analyses suggested that no individual SNP drove the results (Figure 5).

Figure 4.

Funnel plots for the Mendelian randomization analysis of (A) next to BRCA1 gene 1 protein (NBR1) levels as exposure and primary malignant neoplasm of the liver as the outcome, (B) serum RAD51 recombinase (RAD51) protein levels as exposure and primary malignant neoplasm of the liver as the outcome, (C) poly(ADP-ribose) polymerase 11 (PARP11) levels as exposure and hepatocellular carcinoma as the outcome, (D) PARP11 levels as exposure and intrahepatic bile duct carcinoma as the outcome, (E) hepatocellular carcinoma as exposure and poly(ADP-ribose) polymerase 1 (PARP1) levels as the outcome, (F) intrahepatic bile duct carcinoma as exposure and DNA repair protein X-ray repair cross complementing 4 (XRCC4) levels as the outcome, and (G) intrahepatic bile duct carcinoma as exposure and PARP1 levels as the outcome.

Figure 5.

Leave-one-out analysis plots for the Mendelian randomization analysis of (A) next to BRCA1 gene 1 protein (NBR1) levels as exposure and primary malignant neoplasm of the liver as the outcome, (B) serum RAD51 recombinase (RAD51) protein levels as exposure and primary malignant neoplasm of the liver as the outcome, (C) poly(ADP-ribose) polymerase 11 (PARP11) levels as exposure and hepatocellular carcinoma as the outcome, (D) PARP11 levels as exposure and intrahepatic bile duct carcinoma as the outcome, (E) hepatocellular carcinoma as exposure and poly(ADP-ribose) polymerase 1 (PARP1) levels as the outcome, (F) intrahepatic bile duct carcinoma as exposure and DNA repair protein X-ray repair cross complementing 4 (XRCC4) levels as the outcome, and (G) intrahepatic bile duct carcinoma as exposure and PARP1 levels as the outcome.

Table 2.

Sensitivity and horizontal pleiotropy analyses of the causal associations between liver cancer and DNA repair-related genes.

Outcome	Exposure	Heterogeneity (IVW)		Pleiotropy
Outcome	Exposure	Q	Q_pval	MR-Egger intercept	P value
Malignant neoplasm of liver, primary	Next to BRCA1 gene 1protein levels	5.961	0.428	−0.272	0.424
Hepatocellular carcinoma	Next to BRCA1 gene 1protein levels	3.219	0.781	−0.377	0.282
Intrahepatic bile duct carcinoma	Next to BRCA1 gene 1protein levels	7.767	0.256	0.119	0.789
Malignant neoplasm of liver, primary	AG-specific adenine DNA glycosylase levels	10.460	0.234	−0.291	0.241
Hepatocellular carcinoma	AG-specific adenine DNA glycosylase levels	9.187	0.327	0.104	0.675
Intrahepatic bile duct carcinoma	AG-specific adenine DNA glycosylase levels	7.955	0.438	0.225	0.361
Malignant neoplasm of liver, primary	DNA polymerase levels	13.556	0.330	−0.032	0.833
Hepatocellular carcinoma	DNA polymerase levels	5.647	0.933	−0.032	0.815
Intrahepatic bile duct carcinoma	DNA polymerase levels	15.638	0.208	−0.029	0.868
Malignant neoplasm of liver, primary	DNA repair protein XRCC4 levels	3.252	0.953	0.138	0.567
Hepatocellular carcinoma	DNA repair protein XRCC4 levels	7.923	0.542	0.059	0.807
Intrahepatic bile duct carcinoma	DNA repair protein XRCC4 levels	8.677	0.468	0.009	0.975
Malignant neoplasm of liver, primary	DNA repair endonuclease XPF measurement	11.228	0.260	−0.088	0.572
Hepatocellular carcinoma	DNA repair endonuclease XPF measurement	12.146	0.205	0.101	0.542
Intrahepatic bile duct carcinoma	DNA repair endonuclease XPF measurement	9.312	0.409	0.031	0.843
Malignant neoplasm of liver, primary	Serum levels of protein RAD51	18.227	0.311	0.185	0.164
Hepatocellular carcinoma	Serum levels of protein RAD51	24.683	0.076	0.046	0.779
Intrahepatic bile duct carcinoma	Serum levels of protein RAD51	16.322	0.431	0.049	0.732
Malignant neoplasm of liver, primary	Poly [ADP-ribose] polymerase 1 levels	1.398	0.966	−0.009	0.964
Hepatocellular carcinoma	Poly [ADP-ribose] polymerase 1 levels	2.298	0.890	0.196	0.345
Intrahepatic bile duct carcinoma	Poly [ADP-ribose] polymerase 1 levels	3.800	0.704	−0.121	0.581
Malignant neoplasm of liver, primary	Poly [ADP-ribose] polymerase 11 levels	21.511	0.121	0.022	0.904
Hepatocellular carcinoma	Poly [ADP-ribose] polymerase 11 levels	22.394	0.098	0.119	0.521
Intrahepatic bile duct carcinoma	Poly [ADP-ribose] polymerase 11 levels	14.324	0.501	0.206	0.213
Serum levels of protein RAD51	Hepatocellular carcinoma	1.522	0.823	−0.028	0.646
DNA polymerase levels	Hepatocellular carcinoma	7.076	0.132	−0.137	0.208
DNA repair protein XRCC4 levels	Hepatocellular carcinoma	4.919	0.296	−0.063	0.534
DNA repair endonuclease XPF measurement	Hepatocellular carcinoma	0.842	0.933	0.028	0.740
Poly [ADP-ribose] polymerase 1 levels	Hepatocellular carcinoma	3.974	0.410	−0.143	0.156
Poly [ADP-ribose] polymerase 11 levels	Hepatocellular carcinoma	0.938	0.919	0.007	0.933
AG-specific adenine DNA glycosylase levels	Hepatocellular carcinoma	2.662	0.616	−0.006	0.940
Next to BRCA1 gene 1protein levels	Hepatocellular carcinoma	3.319	0.506	0.001	0.991
Serum levels of protein RAD51	Intrahepatic bile duct carcinoma	1.837	0.175
DNA polymerase levels	Intrahepatic bile duct carcinoma	0.192	0.909	0.012	0.884
DNA repair protein XRCC4 levels	Intrahepatic bile duct carcinoma	1.328	0.515	−0.002	0.979
DNA repair endonuclease XPF measurement	Intrahepatic bile duct carcinoma	0.560	0.756	0.047	0.600
Poly [ADP-ribose] polymerase 1 levels	Intrahepatic bile duct carcinoma	0.593	0.743	0.039	0.653
Poly [ADP-ribose] polymerase 11 levels	Intrahepatic bile duct carcinoma	0.445	0.801	−0.007	0.930
AG-specific adenine DNA glycosylase levels	Intrahepatic bile duct carcinoma	1.314	0.518	0.074	0.459
Next to BRCA1 gene 1protein levels	Intrahepatic bile duct carcinoma	4.125	0.127	0.001	0.997
Serum levels of protein RAD51	Malignant neoplasm of liver, primary	1.191	0.880	0.007	0.804
DNA polymerase levels	Malignant neoplasm of liver, primary	3.584	0.310	0.130	0.287
DNA repair protein XRCC4 levels	Malignant neoplasm of liver, primary	0.650	0.885	0.003	0.979
DNA repair endonuclease XPF measurement	Malignant neoplasm of liver, primary	1.612	0.657	0.100	0.385
Poly [ADP-ribose] polymerase 1 levels	Malignant neoplasm of liver, primary	2.940	0.401	−0.124	0.306
Poly [ADP-ribose] polymerase 11 levels	Malignant neoplasm of liver, primary	2.378	0.498	0.140	0.263
AG-specific adenine DNA glycosylase levels	Malignant neoplasm of liver, primary	1.098	0.777	0.047	0.655
Next to BRCA1 gene 1protein levels	Malignant neoplasm of liver, primary	1.298	0.730	0.067	0.536

Table 3.

Results for pleiotropy in the causal relationships between liver cancer and DNA repair-associated genes.

Exposure	Outcome	Raw		Outlier corrected		Global P	Outliers, n	Distortion P
Exposure	Outcome	OR (CI%)	P	OR (CI%)	P	Global P	Outliers, n	Distortion P
Poly [ADP-ribose] polymerase 1 levels	Malignant neoplasm of liver, primary	0.71 (0.48–1.03)	0.12	NA	NA	0.967	NA	NA
DNA repair protein XRCC4 levels	Malignant neoplasm of liver, primary	1.30 (0.91–1.85)	0.19	NA	NA	0.953	NA	NA
Next to BRCA1 gene 1protein levels	Malignant neoplasm of liver, primary	2.28 (1.18–4.44)	0.05	NA	NA	0.469	NA	NA
AG-specific adenine DNA glycosylase levels	Malignant neoplasm of liver, primary	1.57 (0.68–3.58)	0.32	NA	NA	0.265	NA	NA
DNA polymerase levels	Malignant neoplasm of liver, primary	1.13 (0.59–2.17)	0.71	NA	NA	0.356	NA	NA
Serum levels of protein RAD51	Malignant neoplasm of liver, primary	2.16 (1.12–4.15)	0.03	NA	NA	0.375	NA	NA
Poly [ADP-ribose] polymerase 11 levels	Malignant neoplasm of liver, primary	0.57 (0.30–1.08)	0.1	NA	NA	0.134	NA	NA
DNA repair endonuclease XPF measurement	Malignant neoplasm of liver, primary	0.54 (0.26–1.14)	0.14	NA	NA	0.28	NA	NA
Poly [ADP-ribose] polymerase 1 levels	Hepatocellular carcinoma	0.81 (0.50–1.34)	0.45	NA	NA	0.886	NA	NA
Poly [ADP-ribose] polymerase 11 levels	Hepatocellular carcinoma	0.48 (0.25–0.93)	0.05	NA	NA	0.103	NA	NA
Serum levels of protein RAD51	Hepatocellular carcinoma	1.56 (0.72–3.39)	0.28	NA	NA	0.089	NA	NA
DNA repair protein XRCC4 levels	Hepatocellular carcinoma	1.45 (0.82–2.56)	0.23	NA	NA	0.583	NA	NA
DNA repair endonuclease XPF measurement	Hepatocellular carcinoma	0.71 (0.32–1.58)	0.42	NA	NA	0.219	NA	NA
DNA polymerase levels	Hepatocellular carcinoma	0.81 (0.53–1.24)	0.34	NA	NA	0.94	NA	NA
AG-specific adenine DNA glycosylase levels	Hepatocellular carcinoma	1.32 (0.60–2.92)	0.51	NA	NA	0.318	NA	NA
Next to BRCA1 gene 1protein levels	Hepatocellular carcinoma	1.02 (0.62–1.67)	0.95	NA	NA	0.811	NA	NA
AG-specific adenine DNA glycosylase levels	Intrahepatic bile duct carcinoma	1.37 (0.62–3.07)	0.46	NA	NA	0.494	NA	NA
Serum levels of protein RAD51	Intrahepatic bile duct carcinoma	1.85 (0.93–3.68)	0.1	NA	NA	0.438	NA	NA
DNA repair protein XRCC4 levels	Intrahepatic bile duct carcinoma	1.07 (0.56–2.04)	0.84	NA	NA	0.455	NA	NA
DNA polymerase levels	Intrahepatic bile duct carcinoma	1.10 (0.51–2.38)	0.8	NA	NA	0.219	NA	NA
Poly [ADP-ribose] polymerase 11 levels	Intrahepatic bile duct carcinoma	1.86 (1.05–3.29)	0.05	NA	NA	0.504	NA	NA
Next to BRCA1 gene 1protein levels	Intrahepatic bile duct carcinoma	1.18 (0.51–2.73)	0.71	NA	NA	0.355	NA	NA
DNA repair endonuclease XPF measurement	Intrahepatic bile duct carcinoma	1.33 (0.63–2.84)	0.47	NA	NA	0.396	NA	NA
Poly [ADP-ribose] polymerase 1 levels	Intrahepatic bile duct carcinoma	0.75 (0.38–1.50)	0.45	NA	NA	0.719	NA	NA
Hepatocellular carcinoma	Serum levels of protein RAD51	1.00 (0.99–1.02)	0.76	NA	NA	0.844	NA	NA
Malignant neoplasm of liver, primary	Serum levels of protein RAD51	1.00 (0.99–1.01)	0.82	NA	NA	0.891	NA	NA
Malignant neoplasm of liver, primary	DNA polymerase levels	1.03 (0.99–1.07)	0.25	NA	NA	0.4	NA	NA
Hepatocellular carcinoma	DNA polymerase levels	1.02 (0.98–1.06)	0.5	NA	NA	0.189	NA	NA
Hepatocellular carcinoma	DNA repair protein XRCC4 levels	0.99 (0.96–1.03)	0.65	NA	NA	0.314	NA	NA
Malignant neoplasm of liver, primary	DNA repair protein XRCC4 levels	0.99 (0.97–1.00)	0.25	NA	NA	0.897	NA	NA
Hepatocellular carcinoma	DNA repair endonuclease XPF measurement	1.00 (0.99–1.01)	0.95	NA	NA	0.942	NA	NA
Malignant neoplasm of liver, primary	DNA repair endonuclease XPF measurement	0.99 (0.96–1.01)	0.35	NA	NA	0.677	NA	NA
Hepatocellular carcinoma	Poly [ADP-ribose] polymerase 1 levels	0.96 (0.93–0.99)	0.05	NA	NA	0.473	NA	NA
Malignant neoplasm of liver, primary	Poly [ADP-ribose] polymerase 1 levels	1.01 (0.98–1.04)	0.59	NA	NA	0.466	NA	NA
Malignant neoplasm of liver, primary	Poly [ADP-ribose] polymerase 11 levels	1.01 (0.98–1.04)	0.62	NA	NA	0.559	NA	NA
Hepatocellular carcinoma	Poly [ADP-ribose] polymerase 11 levels	1.03 (1.01–1.04)	0.02	NA	NA	0.919	NA	NA
Hepatocellular carcinoma	AG-specific adenine DNA glycosylase levels	1.00 (0.98–1.03)	0.86	NA	NA	0.629	NA	NA
Malignant neoplasm of liver, primary	AG-specific adenine DNA glycosylase levels	1.01 (0.99–1.03)	0.32	NA	NA	0.805	NA	NA
Malignant neoplasm of liver, primary	Next to BRCA1 gene 1 protein levels	1.00 (0.98–1.02)	0.85	NA	NA	0.755	NA	NA
Hepatocellular carcinoma	Next to BRCA1 gene 1 protein levels	0.99 (0.96–1.02)	0.43	NA	NA	0.498	NA	NA

Protein quantitative trait loci analysis

The pQTL analysis showed that the rs2793568 SNP was associated with the expression levels of the PARP1 protein (Figure 6A). PARP1 was associated with several GO terms related to DNA replication and DNA repair (Figure 6B). PARP1 was also associated with the KEGG terms of diabetic cardiomyopathy, necroptosis, apoptosis, NF-κB signaling, and base excision repair (Figure 6C).

Figure 6.

Protein quantitative trait loci analysis. (A) protein quantitative trait loci (pQTL) analysis showing that the rs2793568 single-nucleotide polymorphism (SNP) was associated with Poly(ADP-ribose) polymerase 1 (PARP1) protein levels. (B) Gene Ontology (GO) analysis of PARP1 and DNA replication and repair. (C) Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of PARP1.

Discussion

This study aimed to explore the causal association between genetically predicted DNA repair factors and liver cancer. The results suggest that Next to BRCA1 gene 1 protein levels, serum levels of protein RAD51, PARP1 levels, PARP11 levels, and DNA repair protein XRCC4 levels may be involved in the pathogenesis of liver cancers. Attention should be given to those factors in future studies.

Next to BRCA1 gene 1 protein is encoded by the NBR1 gene and is an ubiquitin-binding autophagy adapter that participates in various processes, including host defense and intracellular homeostasis.^26,27 It acts as a shuttle, bringing ubiquitinated proteins to autophagosomes and participating in the formation of protein aggregates.^28,29 Next to BRCA1 gene 1 protein also regulates the innate immune response by modulating type I interferon production and targeting ubiquitinated IRF3 for autophagic degradation.³⁰ In response to oxidative stress, Next to BRCA1 gene 1 protein promotes an increase in SQSTM1, phosphorylation, and body formation by preventing its autophagic degradation. In turn, it activates the KEAP1-NRF2/NFE2L2 antioxidant pathway. Next to BRCA1 gene 1 protein is involved in the metastatic progression of ovarian cancer³¹ and breast cancer.³² In liver cancer, Next to BRCA1 gene 1 protein is involved in the specific degradation of MHC-1, thereby reducing CD8⁺ T cell-mediated antitumor immunity.^33,34 In the present study, Next to BRCA1 gene 1 protein levels showed a causal association with primary malignant neoplasm of the liver, as supported by the above studies.

RAD51 plays a key role in DNA repair, especially in homologous strand exchange through homologous recombination.^35–37 RAD51 binds to single-stranded DNA in an ATP-dependent manner to form nucleoprotein filaments essential for the homology search and strand exchange processes during DNA repair to form a joint molecule between a processed DNA break and the repair template.^35,36,38 RAD51 is also involved in the resolution of stalled replication forks during replication stress, is part of a PALB2-scaffolded HR complex containing BRCA2 and RAD51C, plays a role in DNA repair,^39,40 is involved in regulating mitochondrial DNA copy number under conditions of oxidative stress in the presence of RAD51C and XRCC3,⁴¹ and plays a role in interstrand crosslink repair.⁴² Several cancer types, including liver cancer, overexpress RAD51.⁴³ Inhibiting RAD51 in HCC displays antitumor effects.⁴⁴ The present MR study suggests a causal association between serum levels of the protein RAD51 and primary malignant neoplasms of the liver.

PARP is a family of proteins involved in many cellular processes, such as DNA repair, genomic stability, and programmed cell death.⁴⁵ The primary role of Poly (ADP-ribose) polymerases is to detect and initiate an immediate cellular response to single-strand DNA breaks by signaling the enzymatic machinery involved in repair.⁴⁶ Although the involvement of Poly (ADP-ribose) polymerase 1 in the pathogenesis of cancer is relatively well-known,⁴⁷ the role of Poly (ADP-ribose) polymerase 11 remains elusive.⁴⁸ The present MR analysis suggested a negative causal association between Poly(ADP-ribose) polymerase 11 and HCC, but a positive association with intrahepatic bile duct carcinoma. Hence, Poly (ADP-ribose) polymerase 11 exhibited strikingly divergent associations with HCC and IHCC, reflecting potential tissue-specific mechanisms in hepatocytes versus bile duct epithelial cells. HCC and IHCC, although both originating from the liver, arise from distinct cellular lineages (hepatocytes and biliary epithelial cells (BECs)) that differ transcriptionally, metabolically, and immunologically. This cellular divergence provides a biologically plausible framework for understanding the opposing statistical associations of Poly (ADP-ribose) polymerase 11 (protective in HCC, risk-enhancing in IHCC).^49,50 In hepatocytes, Poly(ADP-ribose) polymerase 11 may play a homeostatic or protective role by maintaining genomic integrity and controlling lipid metabolism. Poly (ADP-ribose) polymerases are central to single-strand DNA break repair and replication fork stabilization. In hepatic tissue, particularly under oxidative stress, moderate PARP activity supports survival by repairing damage and avoiding necroinflammatory cascades. Excessive activation of poly (ADP-ribose) polymerase 1/11, however, may shift metabolism toward NAD + depletion; conversely, controlled Poly(ADP-ribose) polymerase 11 activity can protect hepatocytes from oncogenic transformation. Conversely, in bile duct epithelial cells, Poly(ADP-ribose) polymerase 11 expression may potentiate oncogenic pathways, possibly through an interplay with biliary-specific transcriptional regulators (HNF1β, SOX9) and distinct immune microenvironments that favor chronic inflammation and fibrosis, which are well-known drivers of cholangiocarcinogenesis.^49–52 Several molecular mechanisms may explain the opposite effects, including the context of DNA repair pathways,^53,54 immune modulation,⁵⁵ and the epigenetic and metabolic contexts.⁵⁶ In addition, both the IVW and weighted median methods support the causal association between Poly(ADP-ribose) polymerase 11 and HCC, whereas the weighted mode estimator did not reach statistical significance. This pattern is common in MR analyses and arises from the differing statistical assumptions and power characteristics of these estimators. The potential reasons for the discrepancy include weaker instrument strength, reduced number of valid clusters, sampling variability and heterogeneity, and low total IV strength.

XRCC4 plays important roles in DNA repair: DNA non-homologous end joining core factor required for double-strand break repair and V(D)J recombination,^57,58 recruitment of other proteins to DNA double-strand breaks, formation of alternating helical filaments that bridge DNA like a bandage,^59,60 and direct interaction with DNA ligase IV.⁶¹

Many proteins are involved in DNA repair and cancer, and the relationships often vary among cancer types.^62–64 Although some results were negative in the present study, it does not indicate that those proteins are not involved in liver cancer. MR analyses are based on the genetic prediction of protein expression, and other factors can influence the expression and activity of proteins. Importantly, the present study was conducted using serum levels of DNA repair factors, and whether these serum indicators accurately reflect what occurs in cells remains unknown. In addition, the investigated proteins are nuclear proteins, and their half-life, once released into circulation, may not accurately represent their actual fate within the cells. Nuclear proteins are typically found in the blood circulation due to cell death and the release of their contents.^30,65 Those factors could explain why some results were positive and others were negative. MR studies do not reveal the specific mechanisms responsible for the observed associations. Nevertheless, this study suggests that the serum levels of some DNA repair factors are associated with liver cancer, but whether the proteins themselves are involved in liver cancer or their release through cell death indicates that the development of liver cancer remains to be investigated.

The integration of DNA repair-related serum proteins, such as BRCA1, RAD51, XRCC4, Poly(ADP-ribose) polymerase 1, and Poly(ADP-ribose) polymerase 11, into established HCC and IHCC screening workflows could markedly improve early detection and risk stratification. These biomarkers reflect homologous recombination repair (HRR) dysfunction and DNA damage response dysregulation, both of which are increasingly recognized in hepatobiliary tumorigenesis.^66–68 Alpha-fetoprotein (AFP) remains a cornerstone of liver cancer screening, though it has limited sensitivity and specificity for early-stage HCC or IHCC. Recent studies show that combining AFP with alternative molecules (such as AFP-L3, des-γ-carboxy prothrombin (DCP), or CA19-9) improves diagnostic accuracy. Serum proteins like RAD51 and Poly (ADP-ribose) polymerase 1 could be evaluated alongside AFP to identify molecular subsets of HCC that exhibit HRR defects, improving sensitivity where AFP is negative or borderline.^69–71 Routine imaging (ultrasound, computed tomography, and magnetic resonance imaging) can miss small or atypical lesions, particularly in cirrhotic livers. Serum panels incorporating Poly(ADP-ribose) polymerase 1, Poly(ADP-ribose) polymerase 11, and RAD51 may serve as pre-selection or triage tools, flagging high-risk patients for intensive imaging. Elevated DNA repair-related protein levels could also help interpret indeterminate imaging findings.^54,72 Serum testing is minimally invasive and easy to implement in surveillance programs, particularly for high-risk groups such as chronic hepatitis B/C or cirrhosis patients. It aligns well with established paradigms of semiannual blood-based and imaging surveillance.^68,72 Serial monitoring of RAD51 and Poly (ADP-ribose) polymerase 1 levels may reveal evolving genomic instability before lesion detectability by imaging, thus supporting longitudinal tracking of disease dynamics.^66,69 Biomarkers such as Poly(ADP-ribose) polymerase 1 and RAD51 not only hold diagnostic potential but also predict therapeutic responsiveness to Poly(ADP-ribose) polymerase inhibitors and DNA damage-targeted agents, linking detection with precision treatment opportunities.^53,67

In the context of reverse MR, the observed associations of HCC with Poly (ADP-ribose) polymerase 1 and IHCC with XRCC4 likely represent downstream biological effects of the cancer process rather than primary causal mechanisms. Nevertheless, these reverse associations provide insight into tumor-driven feedback pathways, cellular stress signaling, and systemic alterations that accompany disease progression, rather than indicating direct etiologic roles. Reverse MR linking hepatocellular carcinoma to increased Poly (ADP-ribose) polymerase 1 levels likely reflects feedback upregulation of the DNA damage response. Poly (ADP-ribose) polymerase 1 is a nuclear enzyme essential for recognizing and repairing DNA strand breaks. During hepatocarcinogenesis, genomic instability, oxidative stress, and chronic inflammation trigger the excessive activation of Poly(ADP-ribose) polymerase 1, which maintains cell survival under genotoxic conditions.^73–75 Elevated serum Poly(ADP-ribose) polymerase 1 levels are often observed in advanced or aggressive tumors due to increased cell turnover and necrosis, leading to the passive release of nuclear proteins into the circulation. Therefore, this reverse association plausibly represents both tumor-induced transcriptional activation of Poly(ADP-ribose) polymerase 1 within malignant hepatocytes and protein leakage into the serum during hepatic injury.^74,76 The association between IHCC and XRCC4 levels, identified by reverse MR, may also denote secondary molecular responses to DNA damage or stromal remodeling. XRCC4 is a key scaffolding factor in non-homologous end joining (NHEJ) repair, often upregulated in biliary tumors exposed to oxidative DNA insults or chemoresistance pressure.^77,78 In cholangiocarcinoma, chronic inflammation and bile acid-induced genotoxicity activate XRCC4-mediated repair pathways, which support the survival of malignant cells but may concurrently lead to detectable elevations of XRCC4 in serum due to tumor leakage or exosome-mediated secretion.^78,79 Taken together, the elevated circulating PARP1 and XRCC4 levels in reverse MR analyses should be interpreted as markers of cancer-associated tissue stress or damage response activation, rather than as drivers of tumor initiation. They likely reflect both intracellular feedback regulation sustaining malignant viability and extracellular release due to progression-related cell death.

This study had several strengths. It used a two-sample MR design and GWAS data from thousands of patients. It also evaluated the causal association between DNA repair proteins and liver cancers. On the other hand, the study also had limitations. The GWAS data in the present study primarily originated from a European population, and their generalizability to other populations remains unknown. Liver cancer etiologies exhibit pronounced geographic diversity, primarily driven by differences in viral hepatitis prevalence, alcohol consumption, and metabolic disease patterns worldwide. HCC accounts for about 90% of primary liver cancers, and its underlying causes vary significantly between regions and populations. In Asia and Africa, hepatitis virus B or C infection is the most common etiology, while in Europe and North America, alcohol and metabolic liver diseases are the most common culprits.^80–82 Second, some participants could have been included in both the exposure and the outcome, but it is impossible to determine the exact number. Still, using stringent statistics should minimize the impact of an eventual overlap. Third, several SNPs could be selected as IVs, affecting the causal associations. Finally, the DNA repair proteins we studied were selected based on their established roles in DNA repair and oncogenesis, as documented in the literature. MR studies are entirely statistical studies based on available datasets, and no experiments were performed in the present study. The mechanistic roles of the DNA repair factors identified in this study will need to be confirmed in future studies.

In conclusion, next to BRCA1 gene 1 protein levels, serum levels of proteins RAD51, XRCC4, Poly(ADP-ribose) polymerase 1, and Poly(ADP-ribose) polymerase 11 may be involved in the pathogenesis of liver cancers. Attention should be given to those factors in future studies.

Supplemental Material

sj-docx-1-cbm-10.1177_18758592261429328 - Supplemental material for Bidirectional Mendelian randomization analysis reveals significant associations between Serum DNA repair proteins and liver cancer

Supplemental material, sj-docx-1-cbm-10.1177_18758592261429328 for Bidirectional Mendelian randomization analysis reveals significant associations between Serum DNA repair proteins and liver cancer by Sihao Lin, Kunpeng Fang and Li Geng in Cancer Biomarkers

Supplemental Material

sj-docx-2-cbm-10.1177_18758592261429328 - Supplemental material for Bidirectional Mendelian randomization analysis reveals significant associations between Serum DNA repair proteins and liver cancer

Supplemental material, sj-docx-2-cbm-10.1177_18758592261429328 for Bidirectional Mendelian randomization analysis reveals significant associations between Serum DNA repair proteins and liver cancer by Sihao Lin, Kunpeng Fang and Li Geng in Cancer Biomarkers

Supplemental Material

sj-docx-3-cbm-10.1177_18758592261429328 - Supplemental material for Bidirectional Mendelian randomization analysis reveals significant associations between Serum DNA repair proteins and liver cancer

Supplemental material, sj-docx-3-cbm-10.1177_18758592261429328 for Bidirectional Mendelian randomization analysis reveals significant associations between Serum DNA repair proteins and liver cancer by Sihao Lin, Kunpeng Fang and Li Geng in Cancer Biomarkers

Footnotes

Acknowledgements

Not applicable.

ORCID iDs

Sihao Lin

Li Geng

Ethical statement

Not applicable.

Consent for publication

Not applicable.

Author contributions

Conception and design: Li Geng

Administrative support: Li Geng

Provision of study materials or patients: Kunpeng Fang

Collection and assembly of data: Sihao Lin

Data analysis and interpretation: Sihao Lin

Manuscript writing: All authors

Final approval of manuscript: All authors.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Availability of data and materials

All data generated or analyzed during this study are included in this published article.

Supplemental material

Supplemental material for this article is available online.

References

Bray

Laversanne

Sung

, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024; 74: 229–263. 2024/04/04.

Villanueva

. Hepatocellular carcinoma. N Engl J Med 2019; 380: 1450–1462. 2019/04/11.

Vogel

Bridgewater

Edeline

, et al. Biliary tract cancer: ESMO clinical practice guideline for diagnosis, treatment and follow-up. Ann Oncol 2023; 34: 127–140. 2022/11/14.

NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines). Hepatocellular Carcinoma. Version 1.2024. Fort Washington: National Comprehensive Cancer Network, 2024.

Marrero

Kulik

Sirlin

, et al. Diagnosis, staging, and management of hepatocellular carcinoma: 2018 practice guidance by the American association for the study of liver diseases. Hepatology 2018; 68: 723–750. 2018/04/07.

Vogel

Cervantes

Chau

, et al. Hepatocellular carcinoma: ESMO clinical practice guidelines for diagnosis, treatment and follow-up. Ann Oncol 2018; 29: iv238–iv255. 2018/10/05.

Siegel

Miller

Jemal

. Cancer statistics, 2019. CA Cancer J Clin 2019; 69: 7–34. 2019/01/09.

Zhao

Wang

Wong

, et al. Impact of aging on cardiovascular diseases: from chronological observation to biological insights: JACC family series. JACC Asia 2024; 4: 345–358. 2024/05/20.

Khan

Dageforde

. Cholangiocarcinoma. Surg Clin North Am 2019; 99: 315–335. 2019/03/09.

10.

Hickman

Contreras

. Gallbladder cancer: diagnosis, surgical management, and adjuvant therapies. Surg Clin North Am 2019; 99: 337–355. 2019/03/09.

11.

Hoeijmakers

. DNA Damage, aging, and cancer. N Engl J Med 2009; 361: 1475–1485. 2009/10/09.

12.

Chatterjee

Walker

. Mechanisms of DNA damage, repair, and mutagenesis. Environ Mol Mutagen 2017; 58: 235–263. 2017/05/10.

13.

D'Amico

Vasquez

. The multifaceted roles of DNA repair and replication proteins in aging and obesity. DNA Repair (Amst) 2021; 99: 103049. 20210121.

14.

Sharma

Lewis

Wlodarski

. DNA Repair syndromes and cancer: insights into genetics and phenotype patterns. Front Pediatr 2020; 8: 570084. 2020/11/17.

15.

Sharma

Gupta

, et al. Gallbladder cancer epidemiology, pathogenesis and molecular genetics: recent update. World J Gastroenterol 2017; 23: 3978–3998. 2017/06/28.

16.

Visscher

Wray

Zhang

, et al. 10 Years of GWAS discovery: biology, function, and translation. Am J Hum Genet 2017; 101: 5–22. 2017/07/08. DOI: 10.1016/j.ajhg.2017.06.005.

17.

Sud

Kinnersley

Houlston

. Genome-wide association studies of cancer: current insights and future perspectives. Nat Rev Cancer 2017; 17: 692–704. 2017/10/14.

18.

Smith

Ebrahim

. ‘Mendelian randomization': can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol 2003; 32: 1–22. 2003/04/12.

19.

Davey Smith

Hemani

. Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Hum Mol Genet 2014; 23: R89–R98. 2014/07/30.

20.

Sekula

Del Greco

Pattaro

, et al. Mendelian randomization as an approach to assess causality using observational data. J Am Soc Nephrol 2016; 27: 3253–3265. 2016/11/02.

21.

Haycock

Burgess

Wade

, et al. Best (but oft-forgotten) practices: the design, analysis, and interpretation of Mendelian randomization studies. Am J Clin Nutr 2016; 103: 965–978. 2016/03/11.

22.

Guo

Cao

, et al. Causal association between inflammatory bowel disease and psoriasis: a two-sample bidirectional Mendelian randomization study. Front Immunol 2022; 13: 916645. 2022/06/28.

23.

Choi

Stein

Nishimi

, et al. An exposure-wide and Mendelian randomization approach to identifying modifiable factors for the prevention of depression. Am J Psychiatry 2020; 177: 944–954.

24.

Raghavan

Vardarajan

Mayeux

. Genomic variation in educational attainment modifies Alzheimer disease risk. Neurol Genet 2019; 5: e310. 2019/03/14.

25.

Sun

Maranville

Peters

, et al. Genomic atlas of the human plasma proteome. Nature 2018; 558: 73–79. 2018/06/08.

26.

Zeng

Wei

, et al. The PB1 protein of influenza A virus inhibits the innate immune response by targeting MAVS for NBR1-mediated selective autophagic degradation. PLoS Pathog 2021; 17: e1009300. 2021/02/13.

27.

Walinda

Morimoto

Sugase

, et al. Solution structure of the ubiquitin-associated (UBA) domain of human autophagy receptor NBR1 and its interaction with ubiquitin and polyubiquitin. J Biol Chem 2014; 289: 13890–13902. 2014/04/03.

28.

Nicot

Lo Verso

Ratti

, et al. Phosphorylation of NBR1 by GSK3 modulates protein aggregation. Autophagy 2014; 10: 1036–1053. 2014/06/01.

29.

Turco

Savova

Gere

, et al. Reconstitution defines the roles of p62, NBR1 and TAX1BP1 in ubiquitin condensate formation and autophagy initiation. Nat Commun 2021; 12: 5212. 2021/09/03.

30.

Cai

Zhu

Zheng

, et al. NBR1 Mediates autophagic degradation of IRF3 to negatively regulate type I interferon production. Biochem Biophys Res Commun 2022; 623: 140–147. 2022/08/02.

31.

Marsh

Kenific

Suresh

, et al. Autophagic degradation of NBR1 restricts metastatic outgrowth during mammary tumor progression. Dev Cell 2020; 52: 591–604.e596. 2020/02/23.

32.

Marsh

Debnath

. Autophagy suppresses breast cancer metastasis by degrading NBR1. Autophagy 2020; 16: 1164–1165. 2020/04/09.

33.

Chao

Qian

Wang

, et al. Autophagy and liver cancer. Clin Mol Hepatol 2020; 26: 606–617. 2020/10/16.

34.

Akkoc

Gozuacik

. Autophagy and liver cancer. Turk J Gastroenterol 2018; 29: 270–282. 2018/05/15.

35.

Inano

Sato

Katsuki

, et al. RFWD3-Mediated Ubiquitination promotes timely removal of both RPA and RAD51 from DNA damage sites to facilitate homologous recombination. Mol Cell 2017; 66: 622–634.e628. 2017/06/03.

36.

Ameziane

May

Haitjema

, et al. A novel fanconi anaemia subtype associated with a dominant-negative mutation in RAD51. Nat Commun 2015; 6: 8829. 2015/12/19.

37.

Brannvoll

Xue

Kwon

, et al. The ZGRF1 helicase promotes recombinational repair of replication-blocking DNA damage in human cells. Cell Rep 2020; 32: 107849. 2020/07/09.

38.

Wan

Han

Liu

, et al. Scaffolding protein SPIDR/KIAA0146 connects the bloom syndrome helicase with homologous recombination repair. Proc Natl Acad Sci U S A 2013; 110: 10646–10651. 2013/03/20.

39.

Park

Singh

Nassar

, et al. Breast cancer-associated missense mutants of the PALB2 WD40 domain, which directly binds RAD51C, RAD51 and BRCA2, disrupt DNA repair. Oncogene 2014; 33: 4803–4812. 2013/10/22.

40.

Pellegrini

, et al. Insights into DNA recombination from the structure of a RAD51-BRCA2 complex. Nature 2002; 420: 287–293. 2002/11/21.

41.

Sage

Gildemeister

Knight

. Discovery of a novel function for human Rad51: maintenance of the mitochondrial genome. J Biol Chem 2010; 285: 18984–18990. 2010/04/24.

42.

Wang

Kim

Wagner

, et al. A dominant mutation in human RAD51 reveals its function in DNA interstrand crosslink repair independent of homologous recombination. Mol Cell 2015; 59: 478–490. 2015/08/09.

43.

Liu

Weng

. A pan-cancer bioinformatic analysis of RAD51 regarding the values for diagnosis, prognosis, and therapeutic prediction. Front Oncol 2022; 12: 858756. 2022/04/02.

44.

Pan

Sha

Qiu

, et al. RAD51 Inhibition shows antitumor activity in hepatocellular carcinoma. Int J Mol Sci 2023; 24: 7905. 2023/05/13. DOI: 10.3390/ijms24097905.

45.

Herceg

Wang

. Functions of poly(ADP-ribose) polymerase (PARP) in DNA repair, genomic integrity and cell death. Mutat Res 2001; 477: 97–110. 2001/05/30.

46.

Isabelle

Moreel

Gagne

, et al. Investigation of PARP-1, PARP-2, and PARG interactomes by affinity-purification mass spectrometry. Proteome Sci 2010; 8: 22. 2010/04/15.

47.

Marti

Fernandez-Cortes

Serrano-Saenz

, et al. The Multifactorial Role of PARP-1 in Tumor Microenvironment. Cancers (Basel) 2020; 12: 739. 2020/04/05. DOI: 10.3390/cancers12030739.

48.

Richard

Burgess

O'Byrne

, et al. Beyond PARP1: the potential of other members of the poly (ADP-ribose) polymerase family in DNA repair and cancer therapeutics. Front Cell Dev Biol 2021; 9: 801200. 2022/02/01.

49.

Limaye

Bowen

Orr

, et al. Expression of hepatocytic- and biliary-specific transcription factors in regenerating bile ducts during hepatocyte-to-biliary epithelial cell transdifferentiation. Comp Hepatol 2010; 9: 9. 2010/12/04.

50.

Hrncir

Gracz

. Cellular and transcriptional heterogeneity in the intrahepatic biliary epithelium. Gastro Hep Adv 2023; 2: 108–120. 2023/01/04.

51.

Lee

Konstantinopoulos

. PARP Inhibition and immune modulation: scientific rationale and perspectives for the treatment of gynecologic cancers. Ther Adv Med Oncol 2020; 12: 1758835920944116. 2020/08/13.

52.

Zong

Zhu

, et al. PARP Mediated DNA damage response, genomic stability and immune responses. Int J Cancer 2022; 150: 1745–1759. 2021/12/26.

53.

Mohan

Quingalahua

Gunchick

, et al. PARP Inhibitor therapy in patients with IDH1 mutated cholangiocarcinoma. Oncologist 2024; 29: 725–730.

54.

Keggenhoff

Castven

Becker

, et al. PARP-1 selectively impairs KRAS-driven phenotypic and molecular features in intrahepatic cholangiocarcinoma. Gut 2024; 73: 1712–1724. 2024/06/11.

55.

Brooks

Anand

Cohen

. Immunomodulatory roles of PARPs: shaping the tumor microenvironment, one ADP-ribose at a time. Curr Opin Chem Biol 2023; 77: 102402.

56.

Lin

H-Y

Jeon

A-J

Chen

, et al. The epigenetic basis of hepatocellular carcinoma – mechanisms and potential directions for biomarkers and therapeutics. Br J Cancer 2025; 132: 869–887.

57.

Roy

de Melo

, et al. XRCC4/XLF Interaction is variably required for DNA repair and is not required for ligase IV stimulation. Mol Cell Biol 2015; 35: 3017–3028. 2015/06/24.

58.

Zhang

Karnak

Tan

, et al. FBXW7 Facilitates nonhomologous End-joining via K63-linked polyubiquitylation of XRCC4. Mol Cell 2016; 61: 419–433. 2016/01/18.

59.

Brouwer

Sitters

Candelli

, et al. Sliding sleeves of XRCC4-XLF bridge DNA and connect fragments of broken DNA. Nature 2016; 535: 566–569. 2016/07/21.

60.

Normanno

Negrel

de Melo

, et al. Mutational phospho-mimicry reveals a regulatory role for the XRCC4 and XLF C-terminal tails in modulating DNA bridging during classical non-homologous end joining. Elife 2017; 6: e22900. 2017/05/14. DOI: 10.7554/eLife.22900.

61.

Recuero-Checa

Dore

Arias-Palomo

, et al. Electron microscopy of Xrcc4 and the DNA ligase IV-Xrcc4 DNA repair complex. DNA Repair (Amst) 2009; 8: 1380–1389. 2009/10/20.

62.

Alhmoud

Woolley

Al Moustafa

, et al. DNA Damage/Repair Management in Cancers. Cancers (Basel) 2020; 12: 1050. 2020/04/29. DOI: 10.3390/cancers12041050.

63.

Groelly

Fawkes

Dagg

, et al. Targeting DNA damage response pathways in cancer. Nat Rev Cancer 2023; 23: 78–94. 2022/12/06.

64.

Kehm

Santella

, et al. DNA Repair phenotype and cancer risk: a systematic review and meta-analysis of 55 case-control studies. Sci Rep 2022; 12: 3405. 2022/03/03.

65.

Pisetsky

. The translocation of nuclear molecules during inflammation and cell death. Antioxid Redox Signal 2014; 20: 1117–1125. 20130320.

66.

Xiong

Chen

, et al. Identification of Rad51 as a prognostic biomarker correlated with immune infiltration in hepatocellular carcinoma. Bioengineered 2021; 12: 2664–2675. 2021/06/12.

67.

Wang

Tang

Geng

, et al. Rational combination therapy for hepatocellular carcinoma with PARP1 and DNA-PK inhibitors. Proc Natl Acad Sci U S A 2020; 117: 26356–26365. 2020/10/07.

68.

Ghantus

Ciocan

Schlanger

, et al. Novel Biomarkers in Hepatocellular Carcinoma from Embryogenic Antigens to cfDNA. Biomedicines 2025; 13: 1020. 2025/05/28. DOI: 10.3390/biomedicines13051020.

69.

Liu

Jiang

, et al. The diagnostic performance of AFP, AFP-L3, DCP, CA199, and their combination for primary liver cancer. J Hepatocell Carcinoma 2025; 12: 513–526. 2025/03/11.

70.

Hanif

Ali

Susheela

, et al. Update on the applications and limitations of alpha-fetoprotein for hepatocellular carcinoma. World J Gastroenterol 2022; 28: 216–229. 2022/02/04.

71.

Reichl

Mikulits

. Accuracy of novel diagnostic biomarkers for hepatocellular carcinoma: an update for clinicians (review). Oncol Rep 2016; 36: 613–625. 2016/06/10.

72.

Mocan

Grapa

Crăciun

, et al. Unveiling novel serum biomarkers in intrahepatic cholangiocarcinoma: a pilot proteomic exploration. Front Pharmacol 2024; 15: 1440985. 2024/09/17.

73.

Conceição

CJF

Moe

Ribeiro

, et al. PARP1: a comprehensive review of its mechanisms, therapeutic implications and emerging cancer treatments. Biochim Biophys Acta Rev Cancer 2025; 1880: 189282. 2025/02/14.

74.

Tian

Chen

, et al. PARP-1 in liver diseases: Molecular mechanisms, therapeutic potential and emerging clinical applications (Review). Mol Med Rep 2025; 32: 324. 2025/09/26. DOI: 10.3892/mmr.2025.13689.

75.

Demény

Virág

. The PARP Enzyme Family and the Hallmarks of Cancer Part 1. Cell Intrinsic Hallmarks. Cancers (Basel) 2021; 13: 2042. 2021/05/01. DOI: 10.3390/cancers13092042.

76.

Valanejad

Cast

Wright

, et al. PARP1 Activation increases expression of modified tumor suppressors and pathways underlying development of aggressive hepatoblastoma. Commun Biol 2018; 1: 67. 2018/10/03.

77.

Feng

Yuan

, et al. Identification of biomarkers associated with exhausted CD8+T cells in the tumor microenvironment of intrahepatic cholangiocarcinoma based on Mendelian randomization and bioinformatics analysis. Discov Oncol 2025; 16: 1092. 2025/06/14.

78.

Chen

, et al. Genetic evidence for the causal influence of inflammatory factors on intrahepatic cholangiocarcinoma risk. World J Gastrointest Oncol 2025; 17: 108455. 2025/07/23.

79.

Ning

Gao

Chen

, et al. Causality of blood metabolites on hepatocellular carcinoma and cholangiocarcinoma: a metabolome-wide Mendelian randomization study. BMC Cancer 2025; 25: 389. 2025/03/05.

80.

Cano

Foucher

Musso

. Geographic diversity of human liver cancers mirrors global social inequalities. Front Oncol 2025; 15: 1565692. 2025/06/02.

81.

Liu

. Changes in the Epidemiology of Hepatocellular Carcinoma in Asia. Cancers (Basel) 2022; 14: 4473. 2022/09/24. DOI: 10.3390/cancers14184473.

82.

Pinheiro

Jones

Medina

, et al. Incidence of etiology-specific hepatocellular carcinoma: diverging trends and significant heterogeneity by race and ethnicity. Clin Gastroenterol Hepatol 2024; 22: 562–571.e568.