The Treatment Effect of Liver Transplantation versus Liver Resection for HCC: A Review and Future Perspectives

Abstract

For patients presenting with hepatocellular carcinoma within the Milan criteria, either liver resection or liver transplantation can be performed. However, to what extent either of these treatment options is superior in terms of long-term survival is unknown. Obviously, the comparison of these treatments is complicated by several selection processes. In this article, we comprehensively review the current literature with a focus on factors accounting for selection bias. Thus far, studies that did not perform an intention-to-treat analysis conclude that liver transplantation is superior to liver resection for early-stage hepatocellular carcinoma. In contrast, studies performing an intention-to-treat analysis state that survival is comparable between both modalities. Furthermore, all studies demonstrate that disease-free survival is longer after liver transplantation compared to liver resection. With respect to the latter, implications of recurrences for survival are rarely discussed. Heterogeneous treatment effects and logical inconsistencies indicate that studies with a higher level of evidence are needed to determine if liver transplantation offers a survival benefit over liver resection. However, randomised controlled trials, as the golden standard, are believed to be infeasible. Therefore, we suggest an alternative research design from the causal inference literature. The rationale for a regression discontinuity design that exploits the natural experiment created by the widely adopted Milan criteria will be discussed. In this type of study, the analysis is focused on liver transplantation patients just within the Milan criteria and liver resection patients just outside, hereby ensuring equal distribution of confounders.

Keywords: Milan criteria; hepatocellular carcinoma; liver resection; liver transplantation; regression discontinuity; survival.

Figure 1

Study selection flowchart. (A) systematic studies and (B) of original articles published in or after 2017. Abbreviations: number of studies (N); liver transplantation (LT); liver resection (LR); intention-to-treat (ITT); propensity score matching (PSM).

Figure 2

Regression discontinuity for single lesion Milan criteria. The solid points resemble the observed 5-year survival of fictive patients with single lesion HCC. The hollow points resemble long-term survival that could have been observed if the patient was treated with the alternative treatment. In this illustration, we assume that a smaller tumour is preferred over a bigger tumour, and that this relationship is linear. Furthermore, we chose to depict LT as the superior treatment, with a gain in survival T.

Figure 3

Regression discontinuity for the Milan criteria. Illustration of the average 5-year survival of patients for different combinations of tumour number and size. The black area on the bottom describes the region in which patients are within the Milan criteria and are treated with liver transplantation. The area in grey marks the region outside the Milan criteria where patients are treated with liver resection. Lines a, b, d depict the treatment effect of liver transplantation vs. liver resection. The boundary marked by c describeds the comparison between LT and LR in which the LT group only consists of patients with single lesions, whereas the LR group only consists of patients with two or more tumours. Lastly, the asterisks indicate the regions where the smoothness assumption can be tested.