Gene Mutations After Liver Transplantation May Reveal Which Liver Cancer Patients Face Recurrence

Liver transplantation is often described as the most effective curative option for patients with hepatocellular carcinoma, the most common form of primary liver cancer and the third leading cause of cancer death worldwide. The operation removes the tumor entirely while also resolving the cirrhotic, pro-cancerous liver disease that produced it in the first place. Yet the procedure carries a persistent shadow: even in carefully selected patients, the cancer can return, and recurrence after transplantation remains one of the most important causes of death in this population. A new study published in the journal Heliyon by Xinqiang Li, Bin Wu and colleagues at the Affiliated Hospital of Qingdao University now suggests that the answer to predicting which patients will relapse may lie not in the tumor’s size or number, but in its genome.

The research team carried out whole-exome sequencing on tumor samples and matched blood controls from 37 patients with hepatocellular carcinoma who underwent liver transplantation at their center between January 2019 and July 2021, all selected according to the Hangzhou criteria, an expansion of the classic Milan criteria that allows somewhat larger or more numerous tumors to qualify for transplant. Sequencing was performed on the Illumina NovaSeq6000 platform at an average depth of 500-fold for tumor tissue and 100-fold for blood, a level of coverage deep enough to detect low-frequency somatic mutations with confidence. The team then asked a deceptively simple question: do the mutations carried by tumors that came back within a year differ systematically from those that did not?

The clinical picture was sobering. Twelve of the 37 patients, or 32.4 percent, experienced recurrence within one year of transplantation, a figure broadly consistent with previous reports regardless of whether Milan or Hangzhou criteria were used. Patients whose disease recurred early had dramatically worse overall survival than those who did not, with the difference reaching high statistical significance. Strikingly, however, when the researchers compared the two groups on standard clinical variables, age, preoperative alpha-fetoprotein levels, tumor size, tumor number, differentiation grade, and prior treatment before transplant, none of the differences reached statistical significance. Microvascular invasion, the presence of tumor cells within blood vessels, was more than twice as common in the recurrence group, but even this well-known risk factor fell just short of significance in the direct comparison.

At the genomic level, the mutational landscape of the cohort mirrored what is known about hepatocellular carcinoma more broadly. The most frequently altered genes were BCLAF1, mutated in 73 percent of patients, followed by MUC4 at 57 percent, TP53 at 49 percent, FMN2 at 49 percent, and TTC7A at 46 percent. These genes are not random bystanders: BCLAF1 has been shown to promote tumor angiogenesis by regulating the transcription factor HIF-1α, MUC4 mutations have been linked to poorer responses to immunotherapy combined with targeted therapy, and TP53 is the most commonly mutated gene in liver cancer, shaping both progression and prognosis. The most common mutation types were missense mutations, which alter a single amino acid, followed by splice-site and nonsense mutations, which can disrupt or truncate the encoded protein.

When the researchers compared mutation frequencies between the recurrence and non-recurrence groups across 9,658 tested genes, they identified 32 genes with nominal differences, 25 of which were detected only in the recurrence group. TP53 and the giant structural gene TTN showed the highest mutation frequencies among these differential genes. It is important to note that after applying the Benjamini–Hochberg correction for multiple testing across the full gene panel, none of these frequency differences retained formal statistical significance, a caveat the authors acknowledge transparently. Tumor mutational burden and tumor neoantigen burden, two measures often associated with immunotherapy response, also did not differ significantly between the groups, and neither was associated with recurrence-free or overall survival in this cohort.

The picture changed when the team turned to survival analysis. Univariate Cox regression identified 304 genes whose mutations were significantly associated with poorer recurrence-free survival, and after adjusting for age, alpha-fetoprotein, tumor mutational burden group, and microvascular invasion in multivariate models, 205 of these remained. For overall survival, 811 genes were significant in univariate analysis and 482 survived multivariate adjustment. Among the clinical variables, elevated alpha-fetoprotein and microvascular invasion were each associated with poorer recurrence-free survival in univariate analysis, and microvascular invasion with poorer overall survival, but no clinical factor proved to be an independent predictor once the models were fully adjusted. The genomic data, in other words, carried prognostic information that the standard clinical variables could not.

From this analysis the researchers distilled a 13-gene signature comprising BOD1L1, C14orf159, CSPG4, EML6, KCNB2, OR4D10, PACS2, SPRED3, TAF1, VRK3, FOXO3, NDUFS7 and TTN. These genes were selected by intersecting those with high mutation frequencies in the recurrence group with the independent prognostic factors identified in the multivariate models. Ten of the thirteen genes carried variants found exclusively in patients who recurred within one year. When patients were classified as high-risk if they carried a mutation in any of the 13 genes and low-risk if all 13 were wild-type, the high-risk group showed significantly worse recurrence-free survival and overall survival, with the model retaining independent significance in multivariate Cox regression for both endpoints.

The model’s performance metrics were encouraging for a study of this size. A receiver operating characteristic analysis of recurrence at any time during follow-up yielded an area under the curve of 0.73, while leave-one-out cross-validation, a demanding internal validation approach in which each patient is successively held out and predicted by the remaining data, produced a mean AUC of 0.864. The team also validated the signature against the TCGA-LIHC dataset, an independent public cohort of liver cancer genomes, where nine of the key genes showed supporting evidence. Machine learning approaches, including LASSO regression with ten-fold cross-validation, random forest, and XGBoost, were used to construct and assess the model, and the proportional hazards assumption was verified using Schoenfeld residual tests.

Several of the signature genes have plausible biological connections to cancer behavior. TAF1 has been proposed as a driver gene in hepatocellular carcinoma development, FOXO3, a transcription factor governing stress responses and autophagy, has been linked in a meta-analysis of more than 1,000 cases to tumor development and shorter survival, and TTN mutations have been associated with altered immune infiltration and poor prognosis in liver cancer. Genes such as KCNB2, a potassium channel, and CSPG4, a cell-surface proteoglycan that has attracted interest as an immunotherapy target, hint at mechanisms ranging from ion signaling to tumor-stroma interaction that may shape post-transplant recurrence, though the authors stress that functional studies are still needed to establish causality.

The study’s limitations are clear and the authors state them candidly. With only 37 patients, and just 12 in the recurrence subgroup, the statistical power is limited and the risk of false negatives is real; the single-center, retrospective design introduces potential selection bias, and no functional experiments were performed to demonstrate how these mutations drive recurrence. The survival analyses also rely on nominal P values rather than genome-wide correction, and the mutation-frequency differences did not survive multiple-testing adjustment. Even so, the work represents one of the first systematic attempts to connect exonic tumor variants with post-transplant recurrence in hepatocellular carcinoma, and it delivers a concrete, testable hypothesis: a 13-gene panel that could, if validated in larger prospective cohorts, help clinicians identify high-risk patients before surgery and tailor surveillance, adjuvant therapy, and transplant selection accordingly. For a disease in which recurrence after transplantation remains so difficult to predict and so devastating when it occurs, that would be a meaningful step forward.

Subject of Research: Genomic predictors of hepatocellular carcinoma recurrence after liver transplantation

Article Title: Genomic alterations in hepatocellular carcinoma patients undergoing liver transplantation predict recurrence and prognosis

Article References: Li, X., Cai, H., Wang, C., Qi, Y., Yu, T., Zhang, Q., Liu, H., Luo, N., Chen, J., Cheng, S., Cai, J., & Wu, B. (2026). Genomic alterations in hepatocellular carcinoma patients undergoing liver transplantation predict recurrence and prognosis. Heliyon, 12(15), Article e45555. https://doi.org/10.1016/j.heliyon.2026.e45555

Image Credits: AI Generated

DOI: Not provided

Keywords: hepatocellular carcinoma, liver transplantation, whole-exome sequencing, cancer recurrence, gene signature, TP53, TTN, tumor mutational burden, microvascular invasion, prognosis, genomics, oncology