Currently, given the scarcity of graft resources and competing indications for liver transplantation, patients beyond conventional pre-defined criteria are often not transplanted. Despite achieving successful downstaging to within accepted criteria, patients are not always offered the option of liver transplantation. The question remains whether they should.

One single reference, a 2020 randomized controlled trial by Mazzaferro et al., met the pre-defined PICO criteria and was included for review (Supplementary Table S1) [10]. This study analyzed 74 patients from nine different Italian centers and showed that after an effective and sustained downstaging of tumors originally beyond Milan criteria, liver transplantation improved tumor event-free survival and overall survival compared with non-transplantation therapies [10]. Data supporting that successfully downstaged patients should be considered for liver transplantation.

Many patients with HCC are diagnosed at an advanced stage, falling beyond accepted transplant criteria. However, if the overall tumor burden were to decrease, they could potentially reach a stage for which liver transplantation is usually indicated. Whether this should be actively pursued, treating patients with the goal of lowering their tumor burden so that liver transplantation might become possible, regardless of their initial stage, is still up for debate.

After reviewing 413 references, six observational studies were found to meet the PICO criteria (Supplementary Table S2) [5, 8, 19–21]. All these six studies showed no impact of the original HCC state on post-transplant survival. Although some studies showed a trend towards decreased disease-free survival in patients with advanced HCC (based on size and number) compared to those with less advanced HCC before downstaging, none reached significance [5, 8, 20]. In addition, one study based on waitlist alpha-fetoprotein (AFP) changes even suggested the opposite, utilizing the United States (US) Scientific Registry of Transplant Recipients (SRTR) and including 60 highly selected patients. In the cohort of patients demonstrating a waitlist AFP decrease below 400 ng/mL, those with high original AFP >1,000 ng/mL showed a trend towards better post-transplant survival compared to those with original AFP between 400 and 700, and between 700 and 100 ng/mL (100% vs. ∼75% vs. ∼55%, p = 0.072) [19]. Altogether, the identified studies support the use of downstaging in all patients with HCC beyond conventional criteria (all comers) as long as potentially eligible for transplantation, as the post-transplant survival in case of successful downstaging is not negatively influenced by the original HCC state. Of note, data suggest that a combination of morphological and biological (AFP) criteria should be used to assess the success of downstaging in all comer patients [22]. Also, enough time should be left between a successful downstaging and transplantation (e.g., >6 months) to decrease the risk of post-transplant recurrence [22].

Although the original HCC state has no demonstrated impact on post-transplant survival, several studies showed that patients with advanced HCC are more likely to fail downstaging strategies, confirming the role of downstaging as a selection tool. To illustrate, two studies including 209 and 326 patients reported rates of successful downstaging to within Milan criteria at 39.1% and 38.2% for patients originally beyond UCSF criteria, and at 58% and 45.2% for patients originally between Milan and UCSF (p = 0.042, p = 0.001) [20, 23]. However, as downstaging and palliation involve similar locoregional and systemic treatments, it can generally be argued that it is to the patients’ benefit to keep them in a downstaging strategy.

Macrovascular invasion has historically been a contraindication for liver transplantation in patients with HCC. Although difficult to treat, some patients with macrovascular invasion manage to achieve complete radiologic response after locoregional or systemic treatment. Whether these patients should be considered for liver transplantation is still to be answered.

Of the 85 references found, seven studies met all pre-defined PICO criteria. After reviewing their references, one more study was identified for inclusion, bringing the total to eight studies for further review (Supplementary Table S3) [24–30]. Although several studies demonstrated a 5 years overall survival rate of more than 50% in patients who received downstaging treatments before transplantation, most studies also reported high recurrence rates [24–30]. The largest included study, by Yu et al., analyzed 176 patients with portal vein tumor thrombus (PVTT) type 1–2 and showed a 5 years overall survival of 78.3% in patients with type 1 PVTT compared to 51.6% for those with type 2 PVTT (p = 0.005) [28]. However, recurrence-free survival was about 46% in both groups. Moreover, no subgroup analysis was performed for patients who achieved complete response after pre-operative therapy. This subgroup analysis was also lacking in most of the other included studies [24, 25, 29, 30]. The two studies that did report on outcomes for patients with radiologic (near-to) complete response, by Soin et al. (n = 25) and Serenari et al. (n = 5), showed a 5 years overall survival of 57% and 60%, and a recurrence rate of 24% and 60%, respectively [26, 27]. Consequently, due to insufficient evidence in the contemporary literature, no clear recommendation can be made on whether or not patients with HCC and macrovascular invasion should be considered for transplantation after complete radiologic response. If pursued, this strategy should be carried out within specific clinical trial settings.

Bridging therapy is commonly used to keep patients with HCC within established transplant criteria. However, it is uncertain whether this also results in improved post-transplant survival and should therefore be standard practice for every patient on the transplant waiting list.

After screening 989 references, eight studies were selected for full review. One was a systematic review and meta-analysis (the studies analyzed herein were not separately reinstated for full review), the remaining seven were observational studies (Supplementary Table S4) [31–38]. Some of the identified studies showed significantly better long-term post-transplant survival outcomes in patients treated with bridging therapy [33, 35, 37, 38]. The largest of these studies, by Xing and Kim, looked at 14.511 transplanted patients within Milan criteria pre-transplant (3.889 with bridging, 10.622 without) and showed a 1, 3, and 5 years post-transplant survival of 95%, 85%, 80% in bridged patients versus 94%, 83%, 78% in patients without bridging (p < 0.001) [37]. In the multivariable analysis, bridging therapy remained associated with a significantly better post-transplant survival with a hazard ratio (HR) of 2.28 (95% CI 1.39–3.14; p = 0.003). Bauschke et al. showed in their cohort of 70 patients, all within Milan criteria, that the survival benefit persists even after 10 years post-transplant (95% bridged vs. 73% without bridging, p = 0.014) [33]. Another study analysing patients classified as within Milan criteria pre-transplant showed that the positive effect of bridging therapy on post-transplant survival even seems to last in a setting of recurrence, where the median survival of recurred bridged patients was 75.9 months versus 53.1 months in patients without bridging treatment (p = 0.001) [35]. Looking specifically at patients within UCSF criteria, two studies were evaluated, one with 134 patients and another with 39 patients, but both failed to report any statistical difference in survival between bridged and non-bridged patients.

It is widely believed that bridging therapy is effective in keeping patients within established transplant criteria, however, whether it actually results in reduced waitlist dropout has yet to be confirmed.

A total of 634 references were identified, of which six observational studies and one systematic review and meta-analysis met the pre-defined PICO criteria (the studies analyzed in the systematic review were not separately reinstated for review) (Supplementary Table S5) [31, 34, 36, 38–41]. Considering the most common transplant criteria (Milan, UCSF, ETC), none of the identified studies showed a decrease in overall or disease-specific waitlist dropout for patients who received bridging treatment compared to those without bridging treatment [31, 34, 36, 38–41]. Although not statistically significant, some of the studies did show a longer waitlist time in the group of patients who received bridging therapy [34, 36, 39, 40]. When specifically focussing on progression-related waitlist dropout, one study—evaluating 265 patients within Milan criteria—showed a statistically significantly lower dropout rate in the bridged patient population (2.58%) versus patients without bridging therapy (8.18%) [38]. However, the all-cause waitlist dropout in this study was higher in the bridged patient group (28.4% vs. 14.5% without bridging). Another study, a 2018 meta-analysis by Kulik and others, evaluating 257 cirrhotic patients classified as T2 HCC (patients within Milan criteria), reported no difference in progression-related waitlist dropout between groups treated with and without bridging treatment (relative risk [RR] 0.32; 95% confidence interval 0.06–1.85) [31]. Whether the type of bridging therapy plays a role in waitlist dropout was evaluated in the study by Györi et al., where they analyzed 84 patients within Milan criteria [34]. A transarterial chemoembolization (TACE)-based group (n = 48) was compared with a percutaneous ethanol injection (PEI)/radiofrequency ablation (RFA) group (n = 32) and a control group consisting of patients without bridging treatment (n = 22). They found no difference in all-cause waitlist dropout between groups: 41.7% TACE-based vs. 31.2% PEI/RFA vs. 36.4% control (p = 0.65) [34]. However, a serious limitation in all these retrospective studies, is the inextricable involvement of selection bias in the indication for bridging. Consequently, bridged and non-bridged populations consistently include non-comparable groups of patients and therefore ineluctably mask any effect that bridging therapy might have on waiting list dropout. Thus, precluding the effect of bridging on waitlist dropout from being inferred.

Bridging therapies are used in several patients within conventional transplant criteria to delay tumor progression and to minimize the risk of de-listing while on the waiting-list (dropout). Despite the strong belief that the type of response to bridging is able of influencing the rate of post-transplant tumor recurrence, this, and the weight that tumor response may have on post-transplant survival, have yet to be determined.

Given the high rate of overestimation of treatment response of radiology over pathology, the literature review focused on pathologic responses only. After the identification of 423 references, nine references were included for further review (Supplementary Table S6) [35, 42–49]. All but one study analyzed outcomes achieved after both bridging and downstaging therapies, with TACE being the most commonly used treatment modality. In all studies, patients with complete pathologic response at explant pathology showed better overall survival and recurrence-free survival rates compared with those without complete pathological response [35, 42–49]. Allard et al. found that the favorable prognostic effect of response induced by TACE on explant pathology in 189 patients was confirmed not just for complete necrosis but also for “near to complete responses” (>90%), suggesting a “nearly all - or nothing” rule [48]. This data was later confirmed by the largest single-center US experience (n = 501) published by Agopian in the same year, updated in 2020 in a multicentric fashion including 3,439 patients undergoing liver transplantation from 2002 to 2013 in 20 US centers and all receiving bridging and/or downstaging therapies pre-transplant (with 802 patients showing complete pathological response) [43, 49]. All data supporting the need to pursue a complete (or close to complete) radiological tumor response in patients with HCC listed for liver transplantation.

Many different types of locoregional therapy for HCC exist. In the context of liver transplantation, locoregional therapy is used in the attempt to effectively control the patient’s tumor burden until a suitable liver donor becomes available for transplantation. Consequently, adequate short-term disease control is desired. What type of locoregional therapy best achieves this remains to be determined.

As treatment allocation in clinical practice is subjected to both confounding factors and selection bias, only randomized controlled trials (RCTs) on the application of locoregional therapies outside a transplant setting were included. This approach allows for the least biased comparison between therapeutic modalities. Of the 2,944 unique references found, 40 RCTs comparing at least two treatment modalities were included for further review (Supplementary Table S7) [50–88]. Treatment comparisons were grouped according to lesion size and number combinations.

RCTs on uninodular lesions with size up to 3 cm (BLCL 0, A; within Milan) have compared: radiofrequency ablation (RFA) to percutaneous ethanol injection (PEI) [52, 57, 68, 74, 77], RFA to percutaneous laser ablation (PLA) [80], RFA to percutaneous acetic acid injection (PAAI) [57, 64], RFA to cryoablation [55], RFA to microwave ablation (MWA) [50, 58, 59, 69, 73, 76], and RFA to RFA combinatorial approaches [61, 65, 70, 71, 78, 82]. RFA appeared to induce higher frequencies of radiological complete responses (rCR) and improved 1 year local recurrence (LR) rate compared to PEI and PLA. Compared to PAAI, RFA induced similar rCR. However, 3 years LR rate was improved in RFA-treated versus PAAI-treated patients (RR = 0.41, 95% CI: 023–0.91) [57]. Cryoablation has been shown to have equal rCR, 1 year LR rate, 1-year overall survival, and 1 year disease-free survival as RFA, albeit in a single RCT [55]. In a meta-analysis on RCTs among RFA- and MWA-treated lesions no difference in radiological complete response rates was observed (RR: 1.01, 95% CI: 0.99–1.02) [67]. Moreover, 1 year disease-free and overall survival rates were similar. No difference in adverse events (Aes) could be observed between RFA and MWA-treated patients. RCTs on combination of RFA with TACE [65, 70, 82] or other therapeutic regimen (PEI [71], Iodine-125 [78], Interferon alpha [61]) did not show or report any difference in rCR in these tumor lesions compared to RFA only.

RCTs comparing RFA to PEI [52, 74], RFA to PLA [80], RFA to PAAI [64], RFA to cryoablation [55], and RFA to MWA [50, 58, 59, 69, 73, 76] have included uninodular lesions, ranging 3–5 cm as well. As RFA and MWA in these trials have shown to be clinically effective one might suggest that these techniques are preferred as first line regimen. Yet, locoregional ablative therapies tend to become less effective if tumor lesion size increases.

In case of increased tumor burden, intra-arterial therapies or radiotherapy provide an alternative. Different RCTs on uninodular lesions ranging 3–5 cm (BCLC A; within Milan) and uni-/multinodular lesions ≥5 cm (BCLC A, outside Milan; BCLC B, outside Milan, resp.) have compared: TACE to transarterial or “bland” embolization (TAE) [54, 60, 81, 84], TACE/RFA to TACE combined with RFA [51, 56, 65, 66, 70, 72, 86], TACE to transarterial radio-embolization (TARE) [53, 62, 83, 88], TACE to transarterial ethanol ablation (TAEA) [75], TACE to transarterial chemo-infusion (TACI) [63], and TACE to radiotherapy [79, 85]. Hyperselective TACE (tend to) induced higher frequencies of rCR or radiological partial response (rPR) compared to “bland” embolization. 1 year disease-free and overall survival was either non-significantly different among the groups or tended to be increased in TACE-treated patients. When combining TACE with ablative therapies, combination regimen appeared to induce higher rCR (i.e., + PEI [66], + RFA [86], and + cryoablation [51]), 1 year disease-free survival [66], and 1-year overall survival [51, 86], although studied in relatively small cohorts. RCTs comparing TACE to TARE have shown conflicting results. Whereas Raoul et al. reported no difference in rCR/PR when using Iodine-131 radioembolisation [53], other trials have shown a trend to higher radiological response rates in Yttrium-90 (Y-90) radioembolization cohorts compared to TACE [83, 88]. Moreover, Salem et al. have observed that Y-90 appeared to have lower 1 year LR rate [62]. Generally, treatment-related or grade ≥3 AEs were either equal or reduced in favor of TARE. Conformably, in the prospective, multi-center, non-randomized MERITS-LT trial both TACE and Y90-TARE showed equal efficacy in downstaging towards liver transplantation [87]. Though not statistically significant, explanted livers of TARE-treated patients demonstrated higher frequencies of tumor necrosis (30.8% vs. 20.5%) and lower frequencies microvascular invasion (7.7% vs. 20.5%) hinting towards improved local tumor control. Nowadays, TARE has been accepted as an effective alternative in case TACE is contraindicated (e.g., portal thrombosis). To this end, no clear benefit of TAEA, TACI, or radiotherapy (i.e., proton-beam, brachytherapy) over TACE in RCTs was observed. Yet, recent prospective cohort studies strongly hint to safe and superior efficacy of stereotactic body radiotherapy over TACE as bridge to transplant [89, 90]. Any conclusive results on these therapies are expected from ongoing phase III RCTs (i.e., NCT03960008).

Although this data provides valuable insight in the potential of each locoregional treatment in a non-transplant setting, their results cannot directly be translated to waitlisted patients. Therefore, no recommendations can be made. Nonetheless, these comparisons can provide guidance in determining the kind of treatment to pursue (Table 2).

Immunotherapy has recently become part of the standard treatment for advanced unresectable HCC who are not amenable to curative or locoregional therapy. Due to its promising results, interest has emerged in the use of immunotherapy in a neoadjuvant setting. Whether patients receiving immunotherapy prior to liver transplantation are at risk for rejection has yet to be determined.

Of the 1,560 references identified, nine studies on liver transplantation in patients previously treated with immune checkpoint inhibitors were included, representing 27 cases (Supplementary Table S8) [91–99]. The first case reported resulted in fatal hepatic necrosis at day 8th and patient loss [91]. The ICI was given within 4 weeks before transplantation. A minimum washout period (4 weeks) prior to transplantation given the half-life of 27 days was proposed. Subsequent reports have shown successful results [92, 94–99]. In total, four cases of severe rejection were reported with two successful re-transplantations [91, 93, 97, 98]. Since drug type, pre-transplant treatment and dosage, tumor burden, and response vary from case to case, further investigations are needed to explore the safety and long term oncologic outcomes in a pre-transplant setting.

To optimize the use of immunotherapy treatment in patients with HCC and to be able to evaluate its effect in a (neo)adjuvant setting, it is imperative that tumor response after immunotherapy can be adequately assessed. However, the best way to do this has yet to be determined.

After an extensive review of 6,800 references, seven studies were selected for inclusion (Supplementary Table S9) [11, 13–15, 100–102]. Radiologic evaluation of response after immunotherapy is primarily derived from the recent trials on immunotherapy within advanced HCC where survival benefit was associated with objective response and significant reduction in tumor burden [11, 100–102]. In these studies, the objective response rate by mRECIST ranged from 22% to 34%, whereas complete response was reported in 2.2%–5.5% of the cases [11, 100–102]. Unfortunately, these studies lack confirmation of actual response through pathological assessment. Three recent trials that published on the use of neoadjuvant therapy prior to resection in HCC did report on both response seen on imaging and determined by pathologic assessment. Complete pathologic response ranged from 8% to 25% and major pathologic response (>70% necrosis) was seen in 20%–42%, while pre-operative imaging according to RECIST 1.1 reported partial and complete response in only 8%–15% and 0%, respectively [13–15]. Although data on imaging-pathology response correlations in a transplantation setting are lacking, encouraging pathologic response rates have been reported. In a study of 9 patients who underwent ICI in combination with locoregional therapy, downstaging was successful in 4/5 patients and major pathologic response (>70% necrosis) was noted in 6/9 patients [96]. Improved imaging techniques and biomarkers are needed to define response ahead of pathologic assessment and oncologic outcomes. Given the high rate of explants exceeding Milan criteria post transplantation, significant limitations occur with the current contrast enhanced computed tomography (CT) and magnetic resonance imaging techniques (MRI) in predicting treatment response [87]. In addition, with the use of immunotherapies, the immunologic changes within the tumor and tumor microenvironment may impact the relation between the degree of pathologic and radiographic response [14]. Moreover, the vasoconstrictive and antiangiogenic effects of the drugs may induce a false positive assessment of response by mRECIST [103, 104].

A combined treatment of immunotherapy and locoregional therapy may be more effective than each treatment separately. However, it remains to be seen whether such combined treatment approach is safe in the context of transplantation.

Since no data was found on combined treatment with locoregional and immunotherapy in the setting of transplantation, data outside transplant setting was assessed. In this context, a total of 450 references were identified, whereas 14 were eventually included for further review (Supplementary Table S10) [105–119]. Two of these were systematic reviews and meta-analyses [105, 106]. The first, including 19 studies and comparing TACE or RFA with immunotherapy, did not evaluate safety profiles [105]. The second, including four studies comparing TACE with dendritic cells therapy, reported that patients in the TACE-DC-CIK group were more likely to suffer a fever than the ones in the control group (p = 0.001). In the five prospective studies, one randomized controlled trial and four non-randomized trials, no safety difference between arms was reported [110–115]. However, the small sample sizes limited the robustness of their conclusion. Finally, of the seven non-randomized retrospective studies, five focused on early-death or severe complications with none of the studies reporting any major complication or death associated with the treatment evaluated [108, 117–120]. In the remaining two retrospective studies safety was not reported [107, 109]. Although these data provide valuable insight into the safety and long-term oncologic outcomes of combined treatments of locoregional therapy and immunotherapy in a non-transplant setting, they cannot be extrapolated to a transplant/waitlist-setting. Therefore, no recommendation can be made.