Recurrence of Hepatocellular Carcinoma After Liver Transplantation: The Blind Spot of HCC Management

ABSTRACT

Recurrence of hepatocellular carcinoma (HCC) after liver transplantation (LT) significantly impacts transplant outcomes, responsible for half of all deaths in the first 5 years after LT for HCC, with a 12–15‐month median overall survival after recurrence. Recent advances in post‐LT risk stratification and efficacy data of radical local treatments with curative intent support risk‐adapted tailored surveillance. To date, only immunosuppressive regimen minimisation has been recognised as a potential preventive measure, although the respective roles of calcineurin inhibitor minimisation and mTOR inhibitor introduction remain inconclusive. Retrospective studies highlight the considerable heterogeneity between patients with recurrent HCC after LT in terms of timing, anatomical distribution, and applicability of treatments. Selected patients may benefit in a durable manner from local approaches with a curative intent, while tyrosine kinase inhibitors remain the first line systemic treatments. The use of immune checkpoint inhibitors is a major challenge associated with major risks of graft rejection and related mortality, that should be evaluated in prospective clinical trials. The impact on recurrent HCC of recent changes of pre‐LT management, such as expanded selection criteria or the increasing use of downstaging strategies including post‐ICI LT, has not been evaluated yet. Recurrent HCC after LT is a major unmet need, calling for a prospective and multicentre effort to improve outcomes for this special population.

1. When and How Does HCC Recur After Liver Transplantation?

Liver transplantation (LT) is the best curative treatment for hepatocellular carcinoma (HCC), with overall post‐LT survival around 70% and 50% at 5 and 10 years, respectively [1]. HCC accounts for about a third of all indications for LT [2].

Because of the chronic shortage of allografts in most Western countries, models have been developed to select patients with the lowest risk of tumour recurrence. They usually integrate size and number of lesions (such as the Milan, the Up‐To‐Seven, or the UCSF criteria), sometimes including the circulating level of alpha‐fetoprotein (AFP), such as in the ‘AFP score’, implemented in France since 2013, or the Metroticket 2.0 developed in Italy [3, 4]. There is significant heterogeneity in allocation policies and pre‐LT selection worldwide [5]. However, despite adherence to any stringent selection criteria, HCC recurrence still recurs in about 15% of patients, regardless of the criteria used [2] (Figure 1a).

FIGURE 1.

Clinical course and management of HCC recurrence after liver transplantation. (a) Timeline illustrating the temporal incidence density of recurrence events and overall survival at 5 years post‐transplantation. Post‐transplantation survival without recurrence is 80%–85%, while approximately 15% of patients experience recurrence despite stringent selection criteria. Early recurrence (≤ 24 months) is an independent poor prognostic factor, with median survival of only 10–13 months after recurrence detection. (b) The types of recurrence (extrahepatic, combined intra‐and‐extrahepatic, and intrahepatic) with common metastatic sites (lungs, liver, and bones). (c) Major risk factors for recurrence. (d) Treatment algorithm stratified by disease pattern at recurrence. AFP: alpha foetoprotein, CNI: calcineurin inhibitors.

Recurrence of HCC significantly impairs post‐LT survival. Half of all deaths occurring within the first 5 years after transplantation for HCC are attributable to tumour recurrence [6]. Median survival after recurrence is 10–15 months according to several European and North American studies, with 20%–40% 5‐year post‐transplant survival [6, 7].

Recurrence usually occurs within the first 2 years after LT and 20%–40% of patients relapse during the first year after transplantation (Figure 1a), according to several multicentric retrospective series [7, 8, 9]. Recurrent HCC after LT is characterised by a specific anatomical distribution, which is different from the classical presentation of HCC in non‐transplanted patients (Figure 1b). In those same retrospective series, about half of the cases present as an exclusively extrahepatic disease, while intrahepatic lesions without extrahepatic metastasis only accounts for 5–30% of cases. The most common tumour sites are the lungs (30%–50%) and bones (20%–60%) [7, 8, 9]. Median time from LT to recurrence was longer for intrahepatic (20.6 months; IQR 9.8–32.1) than for extrahepatic recurrences (11.1 months; IQR 4.2–14.4) in a retrospective study of 151 patients [10]. However, in a recent analysis of 431 patients with recurrent HCC, there was no significant difference in terms of anatomical pattern between early and late recurrence when using 1 year as a cut‐off [9].

Early recurrence (with cut‐offs varying across studies from 12 to 24 months post‐transplant) is an independent prognostic factor for post‐recurrence survival [7]. These recurrences are more often multiple, extra‐hepatic and of greater tumour volume. Bone recurrence and serum AFP have also been associated with a poorer prognosis [7, 11].

Important gaps of understanding remain regarding the natural history and prognosis of recurrent HCC after LT. In particular, the risk of performing LT for patients that have undergone extensive downstaging, sometimes with immune‐checkpoint‐inhibitor‐based combinations, and/or have been selected using extended criteria is poorly known. Whether these new strategies lead to more aggressive recurrence with poorer prognosis will need to be determined by future studies.

2. Can We Predict HCC Recurrence and How to Monitor After LT?

Despite adequate selection of candidates to LT for HCC, the persistent risk of recurrence theoretically warrants post‐LT surveillance. Most international guidelines underline the limited evidence to support any specific surveillance protocol in terms of imaging modality, screening interval, or duration of surveillance. Recently, the latest ILCA‐ILTS consensus suggested to perform surveillance in patients with intermediate‐ or high‐risk of recurrence, paving the way for personalised management, albeit with several levels of uncertainties [12], leading to no specific recommendation on selection model, surveillance duration or modalities.

The first question is whether early diagnosis of recurrence associates with improved post‐recurrence survival. Recent advances on the efficacy of radical treatment of oligometastatic recurrent disease suggest such association, that is yet to be demonstrated prospectively. Cost‐effectiveness studies are also limited to a single North American study, that analysed the association between post‐recurrence survival and exposure to surveillance in 223 patients with post‐LT recurrence. A minimum of three scans within the first 24 months after LT was necessary to obtain a significant benefit on post‐recurrence survival [13], but the benefit has not been evaluated beyond that timepoint.

Another burning question is whether all patients should undergo surveillance for post‐LT recurrence. Development of stratification models of post‐LT recurrence risk is the first step to address such questions, usually based on tumour‐related risk factors (Figure 1c). The most relevant models evaluate the risk of recurrence based on pathological findings on the explant [14, 15, 16, 17] (Table 1). Those models rely on risk factors that are only available at that time (such as microvascular invasion or tumour differentiation), and on the pathological reassessment of tumour burden due to the limitations of radio‐pathological correlation. The most extensively validated model is the Risk Estimation of Tumour Recurrence After Transplantation (RETREAT) score, based on the tumour characteristics on the explant (microvascular invasion, the sum of the diameter of the largest viable nodule and the number of viable nodules) and the last AFP before LT, first developed in a cohort of 1601 patients and internally validated in 340 patients, to define 6 groups of increasing 5‐year recurrence risk, below 3% for a score of 0 and greater than 75% for a score of 5 and above [16]. The RETREAT score has been recently prospectively validated in a multicentric study including 1166 patients with HCC who underwent LT, outperforming other models such as the MORAL or the RELAPSE scores. RFS in patients with a RETREAT score of 0 was 99.4% at 3 years, suggesting that surveillance in these patients could be unnecessary [18]. Importantly, the RETREAT score was developed only in patients with HCC meeting the Milan criteria by imaging at the time of transplantation. Whether this score can be transposed in other allocation systems outside North America remains to be determined. Another model, the R3‐AFP score, including the number of nodules, the size of the largest nodule, microvascular invasion, nuclear grade and the last pre‐LT AFP, was therefore developed in 1359 European patients and validated in 1085 Latin American patients [14]. The score defines 4 levels of risk, ranging from 5.5% to 73.9% at 5 years, for patients selected for transplantation according to different criteria including the AFP model, that might help alleviate surveillance in the lowest‐risk group. The R3‐AFP score has been validated by leveraging the prospectively collected data of the SILVER study, a randomised phase 3 trial of sirolimus‐based immunosuppressive regimen to prevent HCC recurrence in LT recipients [19]. Whether these scores are applicable regardless of the immunosuppressive regimen remains inconclusive.

TABLE 1.

Summary of the main post‐LT recurrence risk stratification models based on explant tumour burden.

Risk model	Population	Tumour burden	Bio marker	Microvascular invasion	Tumour differentiation	Risk groups	Proportion of patients in lowest risk group	Lowest risk	Prospective validation
MORAL (2017) [15]	US	Largest size and number	AFP, NLR	Yes	Yes	4	39.9%	97.4% 5‐year RFS	No
RETREAT (2017) [16]	US	Largest size and number	AFP	Yes	No	3	21.6%	2.9% recurrence at 5 years	Validation in dedicated study [18]
R3‐AFP (2022) [14]	Europe/Latin America	Largest size and number	AFP	Yes	Yes	4	39.3%	5.5% recurrence at 5 years	Validation in prospectively collected data [19]
RELAPSE (2023) [17]	US/Europe	Largest size and number	AFP	Yes	Yes	4	Na	7.5% recurrence at 5 years	No

Once patients have been selected to undergo surveillance, evidence on the optimal surveillance imaging modality, interval and duration are limited. The most widely used approach is biannual thoracic and abdominal CT scans, as recommended by most international guidelines based on expert opinions. The additional value of adding biological biomarkers such as AFP monitoring is uncertain although it is also proposed by expert opinions. Consequently, surveillance protocols significantly vary between and within transplant centres. In a national survey in 48 American transplant centres, most protocols (74%) included cross‐sectional thoraco‐abdominal imaging, while 21% included only abdominal imaging. Only 3% included bone scans, 65% included AFP monitoring. Forty‐eight per cent of centres reported 5 years of surveillance after LT, while 18% discontinued monitoring after 2 years [20]. With regard to the duration of monitoring, in a study evaluating the combination of CT scan and AFP assay every 6 months, monitoring beyond 2 years was not considered cost‐effective [21]. Other studies have highlighted the importance of later recurrences and suggest monitoring with cross‐sectional imaging and AFP every three to 6 months after transplantation for at least 5 years [22].

Finally, the practical use of post‐LT risk scores is also limited by the lack of validated adjuvant therapies after liver transplantation. The most frequently evaluated treatment has been sorafenib, but only in retrospective and monocentric studies with a small sample size, leading to inconclusive data on efficacy and concerns on safety [23]. Therefore, adapting the immunosuppressive regimen has been the main focus of interest in studies to prevent HCC recurrence after LT.

3. Preventing Recurrence: The Key Role of Immunosuppression

Immunosuppressive treatment is a double‐edged sword in transplant patients, necessary for graft survival, but associated with infections and cancer risk, the main drivers of post‐transplant morbidity and mortality.

Currently, there is no evidence suggesting any role of induction immunosuppression on the risk of recurrence in patients undergoing LT for HCC [24]. In a US study of 22,535 LTs for HCC, neither non‐depletive nor depletive induction was associated with tumour recurrence, when compared with no induction.

Calcineurin inhibitors (CNIs) are the backbone of immunosuppressive regimens after transplantation. However, they have been associated in vitro and in animal models with accelerated tumour growth [25]. Several clinical studies indirectly confirm this increase in cancer risk (Figure 1c). In a retrospective evaluation of 219 patients receiving LT for HCC meeting the Milan criteria, higher exposure to CNIs during the first month post‐LT (mean tacrolimus trough concentrations > 10 ng/mL or ciclosporin > 300 ng/mL), but not thereafter, was associated with an increased risk of HCC recurrence (27.7% vs. 14.7% at 5 years; p = 0.007), even after adjusting for tumour characteristics on the explant [26]. Cumulative exposure to tacrolimus, defined as the area under the curve of trough levels (in the first 3 and 12 months after transplantation) was the only IS‐related predictor of post‐LT de novo cancer in another study of 2495 transplanted patients [27].

Consequently, IS minimisation in recipients of LT for HCC is key. While acknowledging the moderate level of evidence provided by these studies, the ILTS and expert consensus recommend reducing exposure to CNIs to the minimally effective dose to reduce the risk of recurrence of HCC and de novo cancer [12, 28, 29]. Short‐term IS minimisation should aim at reducing CNIs trough levels under the aforementioned threshold. Beyond 3 months after transplantation, general measures applicable to all LT recipients, such as steroid withdrawal or selection of adequate candidates for CNI monotherapy, should be applied. However, IS minimisation strategies should be balanced against the risk of rejection and immune activation in consideration to recipient characteristics, aetiology of primary liver disease and episodes of rejection [30].

The PI3K/AKT/mTOR signalling pathway is aberrantly overexpressed in approximately 20%–30% of HCCs. The use of mTOR inhibitors as immunosuppressants for their potential antitumour properties is a matter of debate. Multiple cohort studies have suggested a possible reduction in the risk of recurrence of HCC after transplantation [31, 32]. However, the only randomised controlled trial conducted to address this question was negative [33]. The SiLVER study was an open‐label phase 3 study that included 525 patients receiving LT for HCC and compared recurrence‐free survival using immunosuppression regimens with or without mTOR inhibitors introduced four to six weeks after LT. The primary endpoint was increase in 5‐year recurrence‐free survival. Recurrence‐free survival was 70.2% with the sirolimus‐based regimen and 64.5% with the standard regimen without sirolimus at study end; this difference was not significant (p = 0.28; hazard ratio [HR], 0.84; 95% confidence interval [95% CI], 0.62; 1.15). However, in a pre‐planned exploratory analysis, patients with sirolimus‐based regimen had better RFS at 3 year (HR, 0.7; 95% CI, 0.48–1.00). Similarly, overall survival did not differ between the two groups. Interestingly, this beneficial effect was only observed in patients with the lowest risk of recurrence, as well as in patients treated without CNI. Currently, based on this negative study, the use of mTOR inhibitors specifically for the prevention of HCC recurrence is not recommended.

4. Local Strategies for Recurrent HCC

There is currently no consensus on the treatment of recurrent HCC after liver transplantation (Figure 1d). Proposed strategies range from aggressive local treatments in the case of unifocal or oligometastatic recurrence to systemic treatments for all. The latter strategy is based on the premise that a recurrence after LT is a metastatic disease per se [34].

Several retrospective studies advocate for favouring local approaches such as resection and/or ablation in selected patients. Most of them found a statistically significant association between applying these treatment modalities on all observable lesions, feasible in about 25%–35% of patients with recurrence, and better overall post‐recurrence survival. When treatments with curative intent are applied, 5‐year post‐recurrence survival could be as high as 40% [9]. In terms of disease‐free survival, a Korean study of 52 patients undergoing resection for lung single or oligo‐metastases reported a 20% 5‐year survival free of any new recurrence [35].

However, caution is warranted when analysing these data. First, there are no currently validated selection criteria for radical approaches. Retrospective studies are inherently biased by treatment selection by expert teams, and do not report outcomes by number, size or types of organ involvement. Whether these approaches are beneficial regardless of the delay of recurrence is currently unknown. Second, some metastatic localisations, such as bone lesions, are more difficult to treat, and the definition of oligometastatic disease amenable to radical treatment may differ site by site according to anatomical considerations and functional reserve. For example, liver resection of an allograft could be associated with surgical challenges due to local adhesions and the general context of immunodepression [36]. Finally, it is currently not known whether patients may benefit from a combination of local and systemic treatments.

5. Systemic Treatments: After Transplantation, a Patient Like Any Other?

Immune checkpoint inhibitor (ICI)‐based combinations of atezolizumab (anti‐PDL1) and bevacizumab (anti‐VEGF), of tremelimumab (anti‐CTLA4) and durvalumab (anti‐PDL1), and of ipilimumab (anti‐CTLA4) and nivolumab (anti‐PD1) are the recommended first‐line therapies in advanced HCC [2]. However, a history of transplantation has always been an exclusion criterion from the pivotal trials that defined the HCC standard of care. The major risk of ICI use is graft rejection. In extrahepatic malignancies, such as melanoma, the use of ICIs in allograft recipients has been associated with an estimated 30%–40% risk of acute rejection [37]. Rejections usually occur early after ICI initiation, with a median delay of 3 weeks, i.e., as early as after a single injection. Graft loss was 65% in a pharmacovigilance study of 23 rejections documented in the WHO VigiBase database, leading to rejection‐related mortality in almost 90% of cases with graft loss (15 out of 17 deaths) [38]. In recurrent HCC, data are currently limited to single‐agent anti‐PD(L)1 (with nivolumab or pembrolizumab) with similar rates of rejection [39]. Data on the use of atezolizumab‐bevacizumab are limited to case reports. These risks mean that immunotherapy must be used extremely cautiously in transplanted patients, with a specific immunosuppressive regimen that is not defined, and with an unknown impact on treatment efficacy. In this context, we advocate for reporting the use of ICI‐based therapies for recurrent HCC in LT recipients in prospective clinical trials, rigorously assessing the safety and efficacy of such treatments in a controlled setting (such as NCT06254248). Outside of clinical trials, patients should be clearly wary of the uncertainties around the risk/benefit ratio of such treatments after transplantation, that should restrict their use as a salvage option after failure or ineligibility to other systemic therapies.

Tyrosine kinase inhibitors (TKIs) are therefore the mainstay of systemic treatment for recurrent HCC after LT. Sorafenib is the most documented TKI in this context, through several retrospective series, although there is no prospective randomised trial. In a 2013 retrospective Italian study comparing 15 patients treated with sorafenib to 14 patients treated with best supportive care (BSC), the median overall survival with sorafenib was 10.6 versus 2.2 months, respectively. The benefit of sorafenib persisted in multivariate analysis (HR 4.01 for BSC, p < 0.001) [40]. In a 2015 meta‐analysis evaluating 113 patients treated with sorafenib, the RECISTv1.1 objective response rate was 4.8%, median progression‐free survival was 5.6 months and overall survival 10.5 months [41].

The largest study evaluating regorafenib is a multicentre European, Argentinian and Canadian study including 36 patients treated as second‐line therapy after sorafenib, compared to a historical cohort of 45 patients treated with BSC (unavailability of regorafenib prior to 2017). Overall survival on regorafenib was 13.1 months (95% CI: 7–19.2), significantly superior to 5.5 months (95% CI: 2.3–8.7, p = 0.002) with BSC. Overall survival of patients in whom sorafenib and regorafenib could be sequenced was 28.8 months from the date of sorafenib initiation [42].

An interesting 20% ORR and median overall survival ranging from 14.5 to 19.6 months have been reported with the use of lenvatinib in small retrospective Western and Asian studies [9, 43, 44, 45]. Two of these studies, including one using a propensity‐score‐based adjustment, also compared survival outcomes of lenvatinib to sorafenib, suggesting a superiority of lenvatinib [9, 43]. However, caution should be applied as patients and context of use of the 2 TKIs were vastly different, and propensity‐score‐adjustment may fail to account for such large variations in such studies with small sample size.

Data on the use of cabozantinib are limited to a preliminary multicentric report including 16 patients treated in second or third‐line, with a reported median overall survival of 15.6 months from the start of cabozantinib. Grade ≥ 3 adverse events were observed in 38% of patients, the most frequent being proteinuria.

Cytotoxic chemotherapy has been reported as an option though it has never shown to improve overall survival in advanced HCC in randomised controlled trials and is not a recommended treatment of HCC.

Combining TKIs with mTOR inhibitor‐based immunosuppression for the treatment of recurrent HCC has been evaluated in several small retrospective studies. In a two‐centre Italian study of 50 patients treated with sorafenib, a statistically significant increase in overall survival (p = 0.03) was observed in 37 patients with mTOR inhibitors compared to 13 without [46]. However, the study included patients treated with sorafenib alone and others treated with combination sorafenib and local treatments (such as surgery). In another unicentric French study evaluating 43 patients treated with sorafenib, a benefit in overall survival was observed for patients treated with the combination of everolimus‐sorafenib versus sorafenib alone, but only when everolemia trough levels were above 5 ng/mL (p = 0.021) [47].

The safety profile of TKIs in combination with immunosuppressants has been questioned due to potential drug interactions. However, there is no conclusive evidence of such an increase in toxicity, whether in combination with CNIs, anti‐metabolites, or mTOR inhibitors. Rates of grade 3/4 adverse events are 40%–70%, of dose reduction for toxicity 40%, and of treatment discontinuation 18%–32% [48], and these data do not appear to differ from safety profiles reported in randomised trials evaluating TKIs for advanced HCC in non‐transplanted patients. An interesting pharmacokinetic study found no correlation between circulating concentrations of tacrolimus and sorafenib in four patients treated with these two compounds [49]. Concerning mTOR inhibitors, a unicentric French study found no statistical (or numerical) difference in terms of adverse events with sorafenib alone (40%) or combined with everolimus (42%) [47].

In consequence, there is no solid evidence to support combining systemic therapy with specific IS regimens such as mTOR inhibitor introduction, despite this being a commonly used approach in real‐world practice, and future prospective randomised trials are warranted.

6. Conclusions and Future Directions

In contrast to decades of gradual improvement of patient selection before transplantation, management of recurrent HCC, occurring after 15% of LTs for HCC, is surprisingly an area of limited evidence and a major unmet need. To date, retrospective studies indicate that post‐transplant HCC is a different disease than HCC in non‐transplanted patients, with significant heterogeneity between patients.

The landscape of post‐transplant HCC management presents several critical research priorities. Effectiveness of tailored post‐LT surveillance strategies based on individual risk assessment will need to be determined, taking into account tumour‐related and possibly non‐tumour related risk factors for recurrence. The management of immunosuppression requires rigorous investigation as the balance between preventing allograft rejection and potentially promoting tumour growth remains poorly defined. The use of neoadjuvant or adjuvant IO‐based combinations appears unfeasible due to the risk of allograft rejection, and new therapeutic strategies will need to be developed specifically for this population in which adjuvant TKIs were ineffective. Prospective studies should be conducted to determine patient selection criteria for local radical treatments once HCC has recurred, which could be applied more aggressively than in non‐transplanted patients. For the treatment of recurrence, IO‐based regimens should be investigated cautiously in well‐designed trials to properly mitigate the risk of rejection.

Conflicts of Interest

The authors declare no conflicts of interest.

Ningarhari, M. , Lassailly G., Dharancy S., Moreno C., and Trépo E.. 2025. “Recurrence of Hepatocellular Carcinoma After Liver Transplantation: The Blind Spot of HCC Managemento.” United European Gastroenterology Journal: 1069–1076. 10.1002/ueg2.70078.

Data Availability Statement

The authors have nothing to report.