Harm and Benefits of Primary Liver Resection and Salvage Transplantation for Hepatocellular Carcinoma

Primary transplantation offers longer life‐expectancy in comparison to hepatic resection (HR) for hepatocellular carcinoma (HCC) followed by salvage transplantation; however, livers not used for primary transplantation can be reallocated to the remaining waiting‐list patients, thus, the harm caused to resected patients could be balanced, or outweighed, by the benefit obtained from reallocation of livers originating from HCC patients first being resected. A Markov model was developed to investigate this issue based on literature data or estimated from the United Network for Organ Sharing database. Markov model shows that primary transplantation offers longer life‐expectancy in comparison to HR and salvage transplantation if 5‐year posttransplant survival remains higher than 60%. The balance between the harm for resected patients and the benefit for the remaining waiting list depends on (a) the proportion of HCC candidates, (b) the percentage shifted to HR and (c) the median expected time‐to‐transplant. Faced with a low proportion of HCC candidates, the harm caused to resected patients was higher than the benefit that could be obtained for the waiting‐list population from re‐allocation of extra livers. An increased proportion of HCC candidates and/or an increased median time‐to‐transplant could lead to a benefit for waiting‐list patients that outweighs this harm.


Introduction
Hepatocellular carcinoma (HCC) represents one of the most common tumors worldwide, and its incidence is increasing in western countries (1,2). Two therapies are currently considered as potentially curative: hepatic resection (HR) and liver transplantation (LT). LT should be considered as the optimal strategy because it removes not only the tumor but also the underlying cirrhotic liver that is at risk for the development of de novo HCC; however, the shortage of donor organs represents the major problem in applying primary transplantation to all patients. HR remains an important therapeutic option but the long-term prognosis is undermined by a high incidence of HCC recurrence, up to 50-70% of cases 5 years after surgery. The combination of both treatments can be a reasonable strategy: HCC patients, within Milan criteria (3) (single nodule ≤5 cm or two or three nodules <3 cm) and with preserved liver function, can successfully undergo HR, limiting the transplantation option to cases of tumor recurrence or hepatic decompensation. In 2000, Majno et al. reported that this strategy, called salvage transplantation, can be considered reasonable; nevertheless, considering a median time-totransplant of 6 months, and a monthly waiting list drop-out rate of 3%, primary transplantation proved the best strategy to adopt, providing a longer life-expectancy in comparison to HR and salvage transplantation (4).
During the last decade, the clinical scenario has substantially changed. Significant improvements in the surveillance of cirrhotic patients (5,6) and those in postoperative care (7,8) have led to a significant improvement of survival after resection; on the other hand, the allocation policy for LT has moved from a Child-Turcotte-Pugh (CTP) based allocation system to the more accurate model for end-stage liver disease (MELD) system, reducing waiting list mortality (9,10). The new liver allocation policy proposed by the United Network for Organ Sharing (UNOS) assigns priority to HCC patients on the basis of their risk of becoming ineligible for transplantation: this policy initially led to a significant reduction of the waiting list drop-out rate from 25.9% in the pre-MELD era to 6.7% in the post-MELD era (9), resulting in an actual median time-to-transplant for HCC patients of 3 months (11,12) and a monthly drop-out rate of about 2% (10,12 (13,14). We considered intrahepatic recurrence as a T2 recurrence.

Variables considered after HR
The outcome of partial hepatectomy in cirrhotic patients has improved remarkably in recent years with improved surgical techniques and perioperative care (7,8): mortality is reported to range between 0% and 5% in tertiary centers and postoperative hepatic decompensation to be about 4% (8,(20)(21)(22). The annual mortality rate assumed was 5%, which is consistent with data reported for CTP Class A patients of 1-year survival of about 95% (23) and with data reported by OPTN for MELD <15 patients (12). The assumed annual decompensation rate from CTP Class A to CTP Class B/C was 7%, which is consistent with the probability reported by OPTN to move from MELD <15 to MELD ≥15 of 7% per year (12) and within the range of 5-10% per year reported in the literature (23 (20,(24)(25)(26). Regarding transplantability of HCC recurrence, the literature reports that tumor recurrence suitable for salvage transplantation occurs with a median of 60%, ranging from 23% to 89% (20,22,(27)(28)(29)(30).

Variables considered during the waiting list period and after transplantation
Calculated variables used in the model during the waiting list period are detailed in Table 1 and are comparable to those used in previously published models of the liver transplant organ allocation system (11,12,15,31,32). Regarding survival after transplantation, we assumed that the procedurerelated mortality of salvage transplantation was the same as that of primary transplantation, as recently reported (28,29,33); in addition, the same posttransplantation survival rates were used for all waiting-list patients, as these do not vary substantially by MELD score (16,17) and it is unclear whether MELD score could predict posttransplantation survival over the long-term (34).

On an intention-to-treat basis, the Markov model showed that, with the UNOS policy of liver allocation to HCC patients, primary transplantation gives a longer lifeexpectancy in comparison to HR and salvage transplantation: 10 years since intention-to-treat, the calculated
life-expectancy for patients undergoing primary transplantation was 6.78 years, and for patients undergoing HR was 6.20 years (Figure 2A) Figure 2B). Patients listed for primary transplantation experienced a tumor progression that led to a drop-out rate from the waiting list of 6.6% in the first year after listing and then decreased; whereas a larger proportion of resected patients developed recurrence or tumor progression that led to nontransplantability with a maximum of 10% 3 years after HR and then decreased ( Figure 2C). Sensitivity analysis showed that none of the variables used in the present model modified the results within their respective plausible ranges considered, except for annual mortality rate after transplantation ( Figure 3A): in fact, HR became the best strategy to adopt if the 5year survival rate after LT was lower than 60% ( Figure 3B). Simulations for different age populations and two-way sensitivity analysis did not change the overall results.

Figure 4 reports the thresholds required to obtain a balance between the harm caused to resected patients and the benefit (gain in life-expectancy) obtained from reallocation of livers to the remaining waiting-list patients. In the base-case scenario, with a proportion of HCC candidates of 10% and a median time-to-transplant of 3 months, the
percentage of patients to be resected was higher than 100% and thus not possible to achieve ( Figure 4A). However, faced with an increased proportion of HCC patients on the waiting list and/or an increased median time-totransplant, HR and salvage transplantation could produce a gain in life-expectancy for the waiting-list population that outweighs the harm caused to resected patients. Considering, for example, a proportion of HCC candidates of 30% and a median time-to-transplant of 6 months, submitting 35% of these patients to HR resulted in a balance between harm and benefit for the remaining waiting-list candidates ( Figure 4A); if the median time-to-transplant increases to 12 months, the proportion of patients to be submitted to HR decreases to 18% ( Figure 4A). Over the thresholds reported in Figure 4A, gain in life-expectancy for waitinglist patients increased proportionally to the percentage of patients submitted to HR. As already reported, in the basecase scenario, the increase of this percentage did not achieve any benefit for waiting-list candidates ( Figure 4B); on the contrary, faced with a proportion of HCC candidates of 30% and a median time-to-transplant of 6 months, the gain in life-expectancy for waiting-list candidates progressively increased ( Figure 4C) and was more pronounced if median time-to-transplant was 12 months ( Figure 4D).  (29)(30)(31)(32)(33), this policy could be followed. It must also be considered that recent evidence has shown an increase of HCC incidence in western countries due to the epidemic of hepatitis C (39); in addition, the improved surveillance of cirrhotic patients has led to diagnosis of HCC at earlier stages, submitting more patients to potentially curative treatments such as LT (5,6