The ATG16L1 gene variant rs2241880 (p.T300A) is associated with susceptibility to HCC in patients with cirrhosis

Protein and organelle turnover by autophagy is a key component to maintain cellular homeostasis. Loss of the autophagy protein ATG16L1 is associated with reduced bacterial killing and aberrant interleukin‐1β production, perpetuating inflammation and carcinogenesis. Here we hypothesized that the functional p.T300A gene variant in ATG16L1 is associated with an increased risk for hepatocellular carcinoma (HCC) in cirrhosis.


| INTRODUC TI ON
The development of hepatocellular carcinoma (HCC) is a frequent complication in cirrhosis, accounting for up to 8% of all cancers in males and up to 6% in females. 1 The incidence of HCC has been increasing affecting currently up to 4.8 per 100 000 inhabitants who have a poor overall survival. 2 Curative treatment options, including surgical resection, curative radiofrequency ablation and liver transplantation, are only available to <30% of patients due to advanced tumour stage at diagnosis. 3 Therefore, the early detection of HCC in surveillance programmes is crucial for successful treatment. Currently, ultrasound surveillance at six month intervals is recommended for all patients at high risk, but patients who might benefit most from intensified surveillance programmes are insufficiently characterized. 4 The identification of genetic and environmental risk factors for HCC may improve such surveillance strategies. Whereas genomewide association studies (GWAS) have identified genetic risk loci for HCC in large Asian populations with predominantly viral hepatitis, [5][6][7] genetic risk factors for non-viral HCC are less well characterized. 8 Given the close relation between inflammation and hepatic carcinogenesis, variants in genes encoding for chemokines and cytokines have often been the subject of candidate gene association studies on HCC. [9][10][11][12] In a recent large candidate-gene study, Stickel et al 13 confirmed an increased risk for HCC in patients with alcohol-related cirrhosis carrying genetic variants of Patatin-like phospholipase domain-containing protein 3 (PNPLA3) and Transmembrane 6 superfamily 2 (TM6SF2)-genes related to hepatic lipid and fatty acid turnover. 13 Autophagy is a central mechanism for maintaining homeostasis, metabolism and survival, 14 thereby stabilizing the cell against metabolic stress and apoptosis. 15 In carcinogenesis, autophagy plays a dual role: although maintaining autophagy is important for the survival of already established tumours, autophagy can also suppress the initiation and development of malignant tumours and decrease the susceptibility to carcinogens, especially under inflammatory conditions. 16,17 In addition there is a role for autophagy in limiting immune activation and inflammation. 18 Absence of autophagy-related protein 16-1 (ATG16L1) or a loss of function mutation promotes TLR4-and NOD2-dependent inflammation resulting in an increased release of IL-1β by innate immune cells. [19][20][21][22] In addition, the coding ATG16L1 gene variant p.T300A affects intestinal barrier function and impairs antibacterial autophagy. 23 We hypothesized a role of ATG16L1 in hepatic carcinogenesis and investigated the association of the ATG16L1 gene variant p.T300A with HCC in a case-control study including 440 patients with cirrhosis from three German academic centres.

| Study design
For this candidate gene case-control study, we recruited two patient cohorts. The derivation cohort consisted of 208 patients with cirrhosis (107 cases and 101 controls) that were consecutively recruited between January 2011 and January 2016 at the Jena University Hospital. Patients were stratified for HCC according to current guidelines either by histological criteria or contrast-enhanced imaging technologies (CT, MRI, contrast-enhanced ultrasound). Patients included as cases with HCC were eligible for this study at the time of diagnosis and or referral to our centre. 54 (50%) patients with HCC were participating in HCC surveillance programmes for a median of 4 years prior to diagnosis (range: 0.5 to 23), whereas 53 presented with HCC outside of surveillance programmes ( Figure S1). Patients included as cirrhotic controls without HCC underwent surveillance in the majority of cases (95/101; 94%) for a median of 24 months (range: 1 to 96 months) with a median total imaging procedures of 2 (range: 1 to 15) before inclusion and a subsequent median 36 months (range 0 to 92) of HCC surveillance with a median of total imaging procedures of 3 (range 0 to 17) in order to exclude HCC. One patient, who was originally assigned to the control group, developed an HCC in follow-up after 84 months and was analysed as HCC case for final analysis.
The independent validation cohort consisted of 232 patients

| Genotyping
Genomic DNA was extracted from whole blood using the QIAmp A subset of patients in the validation cohort has been reported in a previous case-control study. 13 To validate the rs2241880 genotyping assay, Sanger sequencing of 15 randomly selected patient samples was performed (five homozygous wild type, five homozygous mutant and five heterozygous for rs2241880). ATG16L1 genomic DNA was amplified using Platinum Taq DNA Polymerase (ThermoFisherScientific, Germany) as described by the manufacturer and amplification primers FW: ACAGGTTAGTGTGCAGGAGA and REV: CAGTCAGCTCTGCCATTACA (95ºC 2 minutes, 33 cycles of 95ºC 30 seconds-60ºC 20 seconds-72ºC 30 seconds). The PCR product was cleaned using NucleoSpin Gel and PCR Clean-up (Macherey-Nagel, Germany) (NTI-Buffer-Dilution 1:7). Sanger sequencing was performed by Macrogen (Macrogen Europe, Netherlands) using the sequencing primer TGCAGGAGAGTAAGGCATGT ( Figure 1).

| Statistical analysis
Statistical analysis was performed using SPSS 16

| Patient characteristics
In both cohorts, patients with HCC were more frequently male (83.2% vs. 70.3%, P = .003; 80.6% vs. 66.7%, P = .015) and were older (median 66 vs. 59 years, P < .001; 65 vs. 57 years, P < .001) as F I G U R E 1 ATG16L1 rs2241880 Genotyping. A, Principle of end-point quantitative TaqMan PCR-based genotyping assay. TaqMan polymerase chain reaction (PCR) was performed using wild-type and mutant-specific individually labelled TaqMan-probes (VIC and FAM fluorescent dye). Normalized fluorescent signals were blotted against PCR cycle number and genotypes were called dependent on amplification signals (WT: VIC-signal but no FAM-signal, HET: VIC-signal and FAM-signal, MUT: no VIC signal but FAM-signal). B, Genotyping Assay Validation by Sanger Sequencing. ATG16L1 deoxyribonucleic acid (DNA) from five randomly selected patients was amplified by PCR using ATG16L1-specific primers. DNA sequence was revealed by Sanger Sequencing (rs2241880 highlighted in light blue) and compared to genotypes identified by TaqMan quantitative PCR  There were no differences in the Child-Pugh stage between cases with HCC and controls without HCC at study inclusion in the derivation cohort. In the validation cohort, however, controls without HCC presented with a more advanced stage of liver disease as compared to patients with HCC (Table 1).

| Allele frequencies of ATG16L1 p.T300A (rs2241880)
In both cohorts, the genotypic frequencies for rs2241880 did not deviate from the Hardy-Weinberg equilibrium in cases and controls (P > .10). The frequencies of the rs2241880 G allele in patients with cirrhosis without HCC were 0.51 in the derivation cohort and 0.50 in the validation cohort ( In contrast, patients with HCC presented with a significantly higher G allele frequency as compared to patients without HCC. The In the combined analysis comprising 231 cases with HCC and 209 controls without HCC, the association between carriage of the G allele in rs2241880 and HCC was confirmed (

| HCC phenotypes
To analyse the impact of ATG16L1 genotypes on the HCC pheno-  (Table 3). Patients carrying 3 or more of the investigated risk alleles in ATG16L1 and PNPLA3 were at a significantly increased risk for HCC as compared to patients carrying both wild-type alleles ( Figure 2).

| D ISCUSS I ON
We herein report that the frequent genetic polymorphism ATG16L1 p.T300A (rs2241880) was associated with the development of HCC in two independent cohorts of patients with cirrhosis, and remained significant after adjusting for potential confounders like age, sex, aetiology of cirrhosis and the PNPLA3 genotype. As autophagy is an important mechanism for maintaining cell homeostasis and metabolism, it plays a role in stabilizing non-malignant as well as tumour cells against apoptosis, metabolic and inflammatory stress. 15,25 The patterns. [19][20][21][22] An association of ATG16L1 polymorphisms with cancer has been observed in thyroid carcinoma, 28 gastric cancer 29,30 and oral squamous cell carcinoma 31 and affects the outcome in colorectal cancer 32 and lung cancer. 33 Notably, the loss of function mutation p.T300A was protective in some but deleterious in other studies underlining that intact autophagy facilitates and suppresses tumorigenesis depending on the context. Concerning hepatic carcinogenesis, increased levels of proinflammatory cytokines and gene polymorphisms in the IL1-β gene have been shown to be associated with HCC in patients with viral hepatitis. 34,35 As autophagy can limit excessive inflammation, autophagy has been proposed as a protective mechanism in the inflammation-related HCC development. 17 The development and growth of HCC are associated with proinflammatory mediators in the tumour microenvironment, 36  frequencies were identical in patients with and without participation arguing against surveillance as a biasing factor in our analysis.
We herein demonstrate an independent association of a frequent ATG16L1 germ-line mutation with the risk of developing HCC. The multicentre setting and the composition of aetiologies is likely to be representative also for other Central European countries. Taken together, our data suggest that the ATG16L1 genotype can be employed in addition to clinical data and other genetic risk factors to develop personalized strategies to at risk of HCC, which might benefit from undergoing more intensive tumour surveillance.

CO N FLI C T O F I NTE R E S T
The authors thank Kathrin Schulze for technical assistance. All authors declare that they have no conflict of interest.

AUTH O R S ' CO NTR I B UTI O N S
PAR wrote the study protocol, recruited patients, performed statisti-