Timing and allocation trade-offs in liver transplantation for colorectal liver metastases: A decision-analytic study

Background: Liver transplantation (LT) has re-emerged as a potentially curative strategy for highly selected patients with liver-only, unresectable colorectal liver metastases (CRLM). However, the survival benefit observed in recent trials was achieved under conditions of rapid prioritized access to transplantation. Whether this benefit persists when LT is progressively delayed, and what consequences graft diversion may impose on standard transplant candidates, remain unclear.
Methods: We developed a decision-analytic state-transition modelling framework to evaluate the timing- and allocation-dependent implications of deceased-donor LT for CRLM. A monthly-cycle cloned intention-to-treat Markov strategy was used to simulate increasing planned time-to-transplantation, incorporating time-dependent failure-to-transplant, post-dropout survival, and post-transplant survival. CRLM-specific inputs were derived from the TransMet and CAIRO5 trials. Pre-transplant dropout was modelled using a Weibull-cure function reconstructed from TransMet failure-to-transplant events, while post-dropout and post-transplant survival were modelled using parametric Weibull functions. Restricted mean survival time (RMST) at 60 months was used as the primary measure of transplant benefit.
Allocation-related waiting-list harm was evaluated by simulating diversion of 5%, 10%, and 20% of deceased-donor grafts toward CRLM candidates, generating additional delays of 1, 3, and 6 months for standard transplant candidates. Displacement scenarios included hepatocellular carcinoma (HCC), general waitlist, non-HCC, and high-MELD (≥40) candidates. Registry-derived estimates from the United States and Italy were used to explore different allocation environments. Net survival benefit was defined as the balance between survival gained in CRLM recipients and survival lost among displaced standard candidates.
Results: Immediate transplantation reproduced the TransMet per-protocol benchmark, with 5-year overall survival (OS) of 73.2% and a 60-month RMST of 51.0 months. Short delays of 1–3 months produced limited erosion of benefit. However, clinically relevant decline became evident beyond 6 months. At 12 months, only 69.0% of patients remained transplant-eligible, 5-year OS decreased to 54.1%, and RMST gain versus the TransMet intention-to-treat comparator declined to 4.6 months. At 24 months, fewer than half of initially eligible patients ultimately underwent LT, and RMST benefit versus the ITT comparator became negative.
Allocation-related harm increased proportionally with both graft diversion and waiting time. Harm remained relatively limited when lower-risk candidates were displaced, but increased markedly in high-MELD scenarios. Under 20% graft allocation and a 6-month delay, the estimated loss reached 102.3 5-year post-transplant survivors per 1000 grafts in the US high-MELD scenario.
Conclusions: The benefit of LT for CRLM is strongly time-dependent and largely reflects early prioritized transplantation rather than transplantation itself. The acceptability of this strategy depends on preserving rapid access to LT while minimizing displacement of the highest-risk standard transplant candidates.