Downstaging hepatocellular carcinoma before liver transplantation: A multicenter analysis of the “all-comers” protocol in the Multicenter Evaluation of Reduction in Tumor Size before Liver Transplantation (MERITS-LT) consortium

Abstract

Patients with hepatocellular carcinoma meeting united network for organ sharing (UNOS)-downstaging (DS) criteria have excellent liver transplantation (LT) outcomes after DS. However, outcomes for “all-comers” (AC) patients with tumors initially exceeding UNOS-DS are poorly understood. Patients meeting AC (n = 82) or UNOS-DS (n = 229) at 7 LT centers in 4 UNOS regions were prospectively followed from 2015–2020. AC patients had a lower probability of successful DS (67% vs 83% within 12 months; P <.001). The 3-year survival was 69% for UNOS-DS vs 58% for AC (P = .05) and reduced to 30% in patients with Child-Pugh B/C cirrhosis or alpha-fetoprotein (AFP) ≥ 500. Five-year LT probability was 42% for AC vs 74% in UNOS-DS (P = .10). Thirty-eight percent were understaged on explant, with the increasing sum of the largest tumor diameter plus the number of lesions before LT (odds ratio 1.3; P =.01) and AFP ≥ 20 (odds ratio 5.9; P =.005) associated with understaging. Post-LT 3-year survival was 91% for AC vs 81% for UNOS-DS (P = .67). In this first prospective multiregional study of AC patients from the multicenter evaluation of reduction in tumor size before liver transplantation (MERITS-LT) consortium, we observed a 65% probability of successful DS. Three-year survival in AC was nearly 60%, though AC with Child-Pugh B/C or AFP ≥ 500 had poor survival. Explant pathology and 3-year post-LT outcomes were similar between cohorts, suggesting that LT is a reasonable goal in selected AC patients.

1. Introduction

The Milan criteria (1 lesion ≤ 5 cm or 2 to 3 lesions ≤3 cm) have long served as the gold standard for achieving excellent post liver transplantation (post-LT) survival in patients with hepatocellular carcinoma (HCC) comparable to that of nonmalignant etiologies.^1–4 In patients with more extensive tumor burden, downstaging (DS) with local regional therapy (LRT) allows for reduction of tumor burden to within acceptable LT criteria (ie, Milan).² LRT also helps distinguish poor candidates for LT as tumor progression despite LRT carries a higher risk for post-LT HCC recurrence.^2,5–7 Rather than depending on tumor morphology alone, DS uses the response to LRT and successful DS as a prognostic marker of favorable tumor biology to select patients who will likely achieve excellent post-LT outcomes.

In recent years, studies applying a uniform DS protocol to patients modestly beyond Milan criteria have shown promising results. In their single-center experience of patients within the University of San Francisco (UCSF)-DS criteria, the UCSF group observed 5-year post-LT and HCC recurrence-free survival of 78% and 91%, respectively.⁸ A follow-up multicenter study in united network for organ sharing (UNOS) Region 5 reported similar 5-year post-LT and recurrence-free survival rates of 80% and 87%, respectively.⁹ Subsequently, UNOS implemented the UCSF-DS (UNOS-DS) DS protocol as a national policy to allow priority listing for patients who are successfully downstaged to within Milan criteria.¹⁰ More recently, the multicenter evaluation of reduction in tumor size before liver transplantation (MERIT-LT) consortium prospectively applied the UNOS-DS criteria on a broader scale at 7 LT centers in 4 UNOS regions and validated the feasibility of this approach by demonstrating a > 80% rate of successful DS and a 3-year intention-to-treat (ITT) survival of 73%.¹¹

Although these studies have demonstrated excellent outcomes for patients within UNOS-DS criteria, the safety and feasibility of successful DS and LT-related outcomes for patients with tumor burden initially exceeding UNOS-DS criteria (termed “all-comers” [AC]) are poorly understood. Expanding transplant criteria requires thoughtful consideration of the potential risks of LRT, including hepatic decompensation, particularly in patients with advanced disease.¹² The application of a uniform, highly structured approach to DS provides the opportunity to evaluate factors beyond tumor burden alone that influence successful DS and eventual transplant.

In this multiregional study on DS from the MERITS-LT consortium, patients were prospectively followed to better elucidate the ITT and post-LT outcomes among patients with tumor burden within UNOS-DS criteria vs those with more advanced disease (AC). We sought to identify factors beyond tumor size that may predict successful transplant and ITT survival with the goal of expanding access to LT for patients with HCC by incorporating response to LRT into transplant listing.

2. Materials and methods

2.1. Study cohorts and DS protocol

The MERITS-LT consortium was established in 2015 and is comprised of 7 high-volume LT centers in 4 UNOS regions. Through this consortium, consecutive patients with HCC meeting AC or UNOS-DS criteria were enrolled from 2015–2020 at the time of first LRT and prospectively followed at each center. Eligibility criteria for the AC and UNOS-DS cohorts varied by tumor size and number at the time of first LRT and are summarized in Table 1. The UNOS-DS criteria were defined as 1 lesion being > 5 cm and < 8 cm; 2 to 3 lesions each < 5 cm; and 4 to 5 lesions each < 3 cm with a total tumor diameter of < 8 cm. Patients within the AC cohort had an initial tumor burden beyond these criteria without an upper limit. Exclusion criteria for both cohorts included evidence of extrahepatic disease or vascular invasion on cross-sectional imaging. Data were collected through each center’s Institutional Review Board-approved protocols.

Table 1.

All-comers and united network for organ sharing (UNOS) downstaging (DS) protocols.

UNOS-DS protocol	All-Comers protocol
Inclusion criteria
Hepatocellular carcinoma (HCC) exceeds Milan criteria but meets one of the following: Single lesions 5.1 to 8 cm 2 to 3 lesions each ≤ 5 cm with the sum of the maximal tumor diameters ≤ 8 cm 4 to 5 lesions each ≤ 3 cm with the sum of the maximal tumor diameters ≤ 8 cm	HCC exceeding UNOS-DS protocol by any of the following: HCC tumor number HCC tumor size Total HCC tumor diameter
Criteria for successful downstaging
Residual tumor size and diameter within Milan criteria (1 lesion ≤ 5 cm, 2 to 3 lesions ≤ 3 cm)
Criteria for DS failure and exclusion from a liver transplant
1. Tumor progression beyond inclusion/eligibility criteria above 2. Extrahepatic disease and vascular disease on cross-sectional imaging 3. Per current UNOS policy, when alpha-fetoprotein (AFP) ≥ 1000, then transplant cannot be performed unless AFP declines to < 500 ng/mL with LRT
Transplant timing
Minimal observation period of 3 mo between successful DS and transplant demonstrating disease stability	Minimal observation period of 6 mo between DS and transplant demonstrating disease stability and approval by review board

Diagnosis of HCC was based on either quadruple-phase computed tomography (CT) or magnetic resonance imaging (MRI) with gadolinium contrast using standard Liver Imaging Reporting and Data System criteria.^13–15 Percutaneous biopsy was not routinely performed to confirm a diagnosis of HCC. Hepatic nodules < 1 cm were not classified as HCC.

The type of LRTused for DS was determined at the discretion of each center’s multidisciplinary tumor board and was not pre-specified in the research protocols for either cohort. Patients underwent CTor MRI approximately 1 month after LRT to assess treatment response and calculate viable tumor burden. Areas of complete necrosis after LRT were not included in the residual size.¹² Multiple courses of LRT were performed as necessary to achieve complete necrosis prior to LT if possible. Imaging criteria for successful DS required either complete tumor necrosis or reduction in viable tumor size and diameter to within Milan criteria. Disease stability was monitored with CT or MRI at least every 3 months using standard Liver Imaging Reporting and Data System criteria. After successful DS, patients were listed for LT with Model for End-Stage Liver Disease (MELD) exception points after a UNOS-mandated wait period of 6 months.

2.2. Histopathologic analysis

In patients who underwent transplants, we evaluated explant histopathologic features of tumor size, number of tumor nodules, histologic grade of differentiation, and the presence of macro-vascular or microvascular invasion.¹⁶ We also calculated tumor stage based on the maximal diameter and number of viable tumors using UNOS tumor, node, and metastasis criteria.¹⁷

2.3. Study endpoints

The primary outcome evaluated successful DS and protocol dropout due to HCC tumor progression, hepatic decompensation, and/or death in both the UNOS-DS and AC cohorts. Overall survival was calculated from date of first LRT to either date of death or last follow-up for living patients. Secondary outcomes included probability of LT, post-LT survival, HCC recurrence, and explant understaging of pathology. Patients were censored at time of removal from the protocol if they developed a medical contraindication to LT not related to liver disease, withdrew from consideration of LT, or were noncompliant with each center’s transplant policies.

2.4. Statistical analysis

Chi-squared and Wilcoxon rank-sum tests were used to compare baseline and disease-related characteristics between the UNOS-DS and AC cohorts. The Kaplan–Meier method was used to estimate probability of dropout, successful DS, liver transplant, and ITT survival. Time zero was at date of first LRT in all analyses except for post-LT outcomes of recurrence and survival, for which the date of LT served as time zero.

Factors associated with dropout, successful DS, and ITT survival were estimated as hazard ratios (HRs) using Cox proportional hazards regression models. Variables with a P-value <.1 in unadjusted analysis were included in a multivariable analysis, and covariates not attaining significance at the 0.05 level were sequentially eliminated via backward elimination selection until all covariates were significantly associated with the outcome. Analyses were performed using SAS (version 9.4, SAS Institute Inc).

3. Results

3.1. Baseline and waitlist characteristics

A total of 82 patients in the AC protocol and 229 patients in the UNOS-DS cohort were enrolled during the study period. Patient and disease characteristics are presented in Table 2. Patient age at time of LRT did not differ significantly between the AC and UNOS-DS cohorts (median 63.7 vs 62.6; P = .46), and the vast majority of study participants were male (83.3% vs 85.3%; P =.72). There were no significant differences in liver disease etiology between the AC and UNOS-DS cohorts (P = .57), with the proportions of patients with hepatitis C (47.6% vs 55.5%), hepatitis B (9.8% vs 7.3%), and nonalcoholic fatty liver disease (NAFLD) (19.5% vs 12.4%) in the AC and UNOS-DS cohorts, respectively.

Table 2.

Demographics and clinical characteristics at study entry (time of first local regional therapy)

Variable	AC (N = 82)	UNOS-DS (N = 229)	P-value
Male (%)	65 (83.3)	191 (85.3)	.72
Median age (IQR)	63.7 (58.2–67.5)	62.6 (58.0–66.6)	.46
Race/ethnicity (%)			.18
Caucasian	48 (58.5)	130 (57.3)
Hispanic	14 (17.1)	56 (24.7)
Asian	11 (13.4)	24 (10.6)
African American	2 (2.4)	10 (4.4)
Other	7 (8.5)	9 (3.1)
Liver disease etiology (%)			.57
Hepatitis C	39 (47.6)	121 (55.5)
Alcohol	11 (13.4)	35 (16.1)
NAFLD	16 (19.5)	27 (12.4)
Hepatitis B	8 (9.8)	16 (7.3)
Other	6 (7.3)	13 (6.0)
> 1 Diagnosis	2 (2.4)	6 (2.8)
CTP (%)			.78
A	54 (72.0)	155 (72.8)
B	20 (26.7)	52 (24.4)
C	1 (1.3)	6 (2.8)
Median MELD (IQR)	9 (7–11)	9 (7–12)	.99
Median AFP (IQR)	20.2 (6.0–187)	12.1 (5.0–67.0)	.15
Median # of HCC lesions	3 (2–4)	2 (1–2)	<.001
Total tumor diameter (cm; IQR)	10.0 (8.3–12.9)	6.2 (5.5–7.1)	<.001

Abbreviations: AFP, alpha-fetoprotein; AC, all-comers; CTP, Child-Turcotte-Pugh; HCC, hepatocellular carcinoma; IQR, interquartile range; MELD, Model for End-Stage Liver Disease; NAFLD, nonalcoholic fatty liver disease; UNOS-DS, United Network for Organ Sharing-downstaging.

At the time of the first DS procedure, both AC and UNOS-DS had similar MELD (median 9 vs 9; P = .99) and Child-Pugh (CP) scores (CPA 72.0% vs 72.8%; P = .78). By definition, those in the AC cohort had a larger initial tumor diameter (median 10.0 cm vs 6.2 cm) and a larger number of lesions (median 3 vs 2) compared with those in UNOS-DS. In the AC group, 7.3% had a single HCC lesion, whereas the remainder had multifocal disease, compared with 34.9% with unifocal disease in UNOS-DS. Alpha-fetoprotein (AFP) was similar in groups (AC median 20.2 vs UNOS-DS median 12.1 ng/mL, P = .15) prior to first LRT. LRT details are described in Table 3. Transarterial chemoembolization was the most common form of LRT, comprising 55.6% and 63.0% of first DS procedures in AC and UNOS-DS, respectively.

Table 3.

Treatment and explant characteristics.

Variable	AC (N = 82)	UNOS-DS (N = 229)	P-value
Type of LRT, n (%)
First LRT			.34
TACE	45 (55.6)	143 (63.0)
Y-90	29 (35.8)	66 (29.1)
Ablation	1 (1.2)	8 (3.5)
Other	6 (7.4)	10 (4.4)
Second LRT			.14
TACE	44 (61.1)	119 (67.6)
Y-90	18 (25.0)	26 (14.8)
Ablation	4 (5.6)	20 (11.4)
Other	6 (8.3)	11 (6.3)
Median # LRT until downstaging (DS) (IQR)	2 (1–3)	1 (1–2)	<.001
Median mo from 1st LRT to DS (IQR)	3.9 (2.4–7.8)	2.4 (1.4–4.6)	<.001
Liver transplant, n (%)	20 (24.4)	85 (37.1)	.04
Explant stage, n (% * )			.89
No viable tumor	2 (10.0)	14 (16.7)
Within Milan	10 (50.0)	38 (45.2)
Outside Milan	8 (40.0)	32 (38.1)
Explant microvascular invasion, n (% * )	4 (20.0)	13 (15.5)	.74
HCC recurrence after LT, n (% * )	1 (5.0)	7 (8.2)	>.99

Abbreviations: AFP, alpha-fetoprotein; CI, confidence interval; HCC, hepatocellular carcinoma; MELD, Model for End-Stage Liver Disease; NAFLD, nonalcoholic fatty liver disease; LRT, local regional therapy.Abbreviations: AC, all-comers; CTP, Child-Turcotte-Pugh; HCC, hepatocellular carcinoma; IQR, interquartile range; LRT, local regional therapy; LT, liver transplantation; MELD, Model for End-Stage Liver Disease; NAFLD, nonalcoholic fatty liver disease; TACE, transarterial chemoembolization; UNOS-DS, United Network for Organ Sharing-downstaging; Y-90, Yttrium-90.

^*Proportion calculated based on those who received a liver transplant.

3.2. Tumor DS

Of the 82 patients in the AC cohort, 53 (64.6%) were successfully downstaged to within Milan criteria, compared with 83.0% in the UCSF-DS group (P < .001). AC patients required more LRT (median 2 vs 1; P <.001) and required a longer time period (median 3.9 months vs 2.4 months; P < .001) to be successfully downstaged. Type of first and second LRT did not differ significantly between groups (Table 3). The cumulative probability of successful DS was 45.6% at 6 months and 67.3% at 1 year for the AC cohort, compared with 72.0% and 82.6% for UNOS-DS (P <.001) (Fig. 1).

Figure 1.

The Kaplan–Meier probability of successful downstaging (DS) by AC vs UNOS-DS.

Abbreviations: AC, all-comers; LRT, LRT, local regional therapy; UNOS-DS, United Network for Organ Sharing-downstaging.

Type of first LRT (transarterial chemoembolization [TACE] vs Yttrium-90 [Y-90]) and center both did not significantly affect the probability of successful DS. Using Cox proportional hazards regression models of the AC cohort, increasing the size of the largest lesion plus the number of lesions was negatively associated with DS (HR, 0.86 per unit; 95% confidence interval [CI], 0.75–0.99; P =.04) (Table 4). Other studied factors such as age, gender, pretreatment number of lesions, MELD score, CP score, use of systemic therapy, AFP at all tested cutoffs, and etiology of liver disease were not significant predictors of DS.

Table 4.

Unadjusted and multivariable analyses of predictors of downstaging in the all-comers cohort (n = 82).

Predictor	Hazard ratio (95% CI)	P-value
Unadjusted analysis
Age (per y)	1.00 (0.97–1.04)	.84
Female (vs male)	0.88 (0.37–2.08)	.78
Etiology of liver disease
NAFLD (vs hepatitis C)	0.92 (0.44–1.90)	.82
Alcohol (vs hepatitis C)	0.46 (0.18–1.19)	.11
Other including hepatitis B or > 1 diagnosis (vs hepatitis C)	0.73 (0.35–1.51)	.39
MELD score (per point)	1.02 (0.95–1.10)	.54
Child’s B/C cirrhosis (vs A)	1.03 (0.52–2.04)	.92
AFP before LRT (ng/mL)
≥ 20 (vs < 20)	0.60 (0.34–1.03)	.07
≥ 100 (vs < 100)	0.70 (0.37–1.30)	.25
≥ 500 (vs < 500)	0.32 (0.1–1.03)	.06
≥ 1000 (vs < 1000)	0.19 (0.03–1.40)	.10
Total tumor burden (per cm)	0.95 (0.88–1.04)	.27
Number of lesions + largest lesion diameter (per unit)	0.88 (0.77–1.01)	.06
Number of HCC lesions	1.04 (0.89–1.22)	.59
Systemic therapy (vs none)	0.35 (0.11–1.1)	.07
Multivariable analysis
Number of lesions + largest lesion diameter (per unit)	0.86 (0.75–0.99)	.04

Abbreviations: AFP, alpha-fetoprotein; CI, confidence interval; HCC, hepatocellular carcinoma; MELD, Model for End-Stage Liver Disease; NAFLD, nonalcoholic fatty liver disease; LRT, local regional therapy.Abbreviations: AC, all-comers; CTP, Child-Turcotte-Pugh; HCC, hepatocellular carcinoma; IQR, interquartile range; LRT, local regional therapy; LT, liver transplantation; MELD, Model for End-Stage Liver Disease; NAFLD, nonalcoholic fatty liver disease; TACE, transarterial chemoembolization; UNOS-DS, United Network for Organ Sharing-downstaging; Y-90, Yttrium-90.

3.3. Dropout from DS protocol

In the AC cohort, 25 patients (30.5%) experienced protocol dropout before ever achieving successful DS into Milan criteria, whereas 14 dropped out of consideration for LTafter initial DS. Of these 39 AC patients who experienced protocol dropout, HCC progression (n = 20; 51.3%) was the most common reason for dropout, followed by hepatic decompensation/medical decline precluding transplant (n = 13; 33.3%), and nonadherence or loss of follow-up (n = 3; 7.7%). Median time from first DS procedure to dropout was 7.0 months in AC (interquartile range [IQR] 5.1–14.1) vs 9.5 months (IQR 5.3–15.1) in UNOS-DS (P = .55).

Dropout due to HCC tumor progression or liver-related death did not differ between groups, with a cumulative probability of 17.1% at 1 year and 52.2% at 3 years for AC compared with 12.5% and 45.1% for UNOS-DS (P = .31). In unadjusted analysis of the AC cohort, only CP score B/C (vs A) was associated with increased risk of dropout (HR 2.58, 95% CI 1.27–5.25, P = .01). Center-specific differences in waitlist dropout were not observed.

3.4. ITT survival

ITT survival at 3 years was 69.4% for UNOS-DS vs. 58.1% for AC (P = .05) (Fig. 2A). No center-specific differences were noted to significantly affect survival. The type of first LRT (TACE vs Y-90) was also not significantly associated with survival. Predictors of worse ITT survival in the multivariable analysis of the AC cohort (Table 5) included NAFLD (vs hepatitis C; HR 2.77, 95% CI 1.05–7.29, P = .04), AFP ≥ 500 (HR 3.50, 95% CI 1.27–9.65, P = .02), and CP B/C cirrhosis (vs CP A; HR 3.35, 95% CI 1.39–8.11, P = .007). In AC patients with CP B/C cirrhosis, ITT survival was markedly reduced to 23.2% at 3 years (vs 65.6% for AC patients with CP A cirrhosis; P = .004). In AC patients without either AFP > 500 or CP B/C cirrhosis, ITT survival was 67.7% at 3 years, compared with 62.8% at 1 year and 29.9% at 3 years for those with either risk factor (P = .003) (Fig. 2B). Finally, ITT survival in AC patients at 3 years was 65.8% in those who achieved successful initial DS compared with only 26.8% in those who were never able to be downstaged (P = .005).

Figure 2.

(A) The Kaplan–Meier probability of intention-to-treat (ITT) survival by AC vs. UNOS-DS cohort. (B) The probability of ITT survival is determined by a number of risk factors.

Abbreviations: AC, all-comers; AFP, alpha-fetoprotein; DS, downstaging; LRT, local regional therapy.

Table 5.

Unadjusted and multivariable analyses of predictors of survival in the all-comers cohort (n = 82).

Predictor	Hazard ratio (95% CI)	P-value
Unadjusted analysis
Age (per y)	1.06 (1.00–1.12)	.048
Female (vs male)	2.06 (0.89–4.78)	.09
Etiology of liver disease
NAFLD (vs hepatitis C)	2.44 (0.94–6.33)	.07
Alcohol (vs hepatitis C)	2.05 (0.70–6.03)	.19
Other including hepatitis B or > 1	1.39 (0.47–4.07)	.55
diagnosis (vs hepatitis C)
MELD score (per point)	1.07 (0.97–1.18)	.18
Child’s B/C cirrhosis (vs A)	3.01 (1.40–6.47)	.005
AFP (ng/mL)
≥ 20 (vs < 20)	0.74 (0.35–1.59)	.44
≥ 100 (vs < 100)	1.31 (0.59–2.89)	.51
≥ 500 (vs < 500)	2.40 (0.96–5.99)	.06
≥ 1000 (vs < 1000)	4.89 (1.82–13.18)	.002
Total tumor burden (per cm)	1.02 (0.90–1.16)	.72
Number of lesions + largest lesion diameter (per unit)	0.90 (0.75–1.08)	.26
Number of HCC lesions	0.86 (0.70–1.06)	.15
Systemic therapy (vs none)	1.43 (0.49–4.13)	.51
Multivariable analysis
Etiology of liver disease: NAFLD (vs hepatitis C)	2.77 (1.05–7.29)	.04
AFP ≥ 500 (vs < 500)	3.50 (1.27–9.65)	.02
Child’s B/C cirrhosis (vs A)	3.35 (1.39–8.11)	.007

Abbreviations: AFP, alpha-fetoprotein; CI, confidence interval; HCC, hepatocellular carcinoma; MELD, Model for End-Stage Liver Disease; NAFLD, nonalcoholic fatty liver disease.

3.5. LT outcomes

At the last follow-up, 20 (24.4%) patients in the AC cohort had undergone liver transplants, with a median time from successful DS to LTof 7.1 months. The cumulative probability of LTat 1 and 5 years from the first DS procedure was 13.0% and 42.2% for AC, compared with 9.8% and 73.6% for UNOS-DS (P = .10). The type of first LRT (TACE vs Y-90) was not associated with the probability of achieving LT.

Of the 20 AC patients who received LT, 2 (10.0%) had complete tumor necrosis on explant pathology, whereas 10 (50.0%) had residual tumors within stage T1/T2 criteria. Eight (40.0%) had tumor burden beyond T2/Milan criteria, consistent with understaging on last pre-LT imaging. The UNOS-DS cohort had similar rates of understaging, with 38.1% beyond Milan (P = .89) (Table 3). In combined multivariable logistic regression of the 105 combined AC and UNOS-DS patients who underwent LT, predictors of understaging were increasing size of largest lesion plus number of lesions prior to LT (OR 1.35 per unit, 95% CI 1.08–1.69, P = .01) and AFP ≥ 20 (OR 5.89, 95% CI 1.71–20.28, P = .005). Microvascular invasion was noted in 20.0% of AC compared with 15.5% of UNOS-DS (P = .74). Among patients with viable tumors on explant, most had moderately differentiated tumors (AC 63.2% and UNOS-DS 54.3%), with no significant differences in tumor grade between cohorts (P = .66).

Kaplan–Meier post-LT 3-year survival was 90.9% for AC vs 80.6% for UNOS-DS (P = .66). HCC recurrence developed in 5.0% of AC compared with 8.2% of UNOS-DS (P > .99) with a median post-LT follow-up in the AC and UNOS-DS cohorts of 18.5 months and 18.6 months, respectively.

4. Discussion

DS allows for both expanded criteria and incorporation of response to LRT as a marker of tumor biology, with the goal of selecting a subset of patients who may achieve posttransplant survival rates similar to those always within Milan criteria.² A recent multiregional study by the MERITS-LT consortium demonstrated excellent outcomes in the UNOS-DS cohort, with a successful DS rate of > 80% and 3-year ITT survival of 73%.¹¹ We built on this work and sought to explore the DS and transplant outcomes in patients with more advanced cancer with tumor burden beyond UNOS-DS in the MERITS-LT consortium. In this study, we observed a significantly lower probability of successful DS at 1 year in AC compared with UNOS-DS (67% vs 83%). This finding echoes that of a prior single-center study comparing AC vs UNOS-DS, which showed a significantly lower rate of successful tumor DS and higher cumulative probability of waitlist dropout in the AC group.¹⁸ AC patients who were successfully downstaged, however, had significantly higher ITT survival (66% at 3 years vs 27% in those who were unable to be downstaged). These findings highlight the importance of ongoing efforts to improve DS in AC patients. With recent advances in systemic therapy, one such approach may involve incorporation of systemic therapy to maintain target tumor stage by delaying tumor progression, thus serving as a bridge to transplant. Although further prospective analysis on the safety and efficacy of systemic therapy in this population is needed, expanded transplant criteria that incorporate response to both LRT and systemic therapy may improve waitlist outcomes in patients at high risk of dropout.

Tumor burden, specifically size of the largest lesion plus number of lesions, emerged as the key predictor for failure to be downstaged in the AC cohort. For every unit increase in tumor burden, the risk of successful DS dropped by 14%. This finding is supported by the single-center analysis by Sinha et al¹⁸ of the AC cohort, which found a < 50% probability of successful DS in those with a sum of the largest lesion plus number of lesions >12. Mazzaferro et al¹⁹ also previously used the sum of tumor number and largest tumor diameter as a valuable marker of post-LT survival in the Metroticket 2.0 calculator. The present study reiterates the role of this summative variable in LT prognostication compared with tumor number or size alone.

A recent randomized controlled trial of patients with HCC initially beyond Milan undergoing tumor DS demonstrated excellent 5-year ITT survival of 77% for patients in the transplantation group vs 31% in a control group, further emphasizing the utility and feasibility of LT for patients who are successfully downstaged.²⁰ Most studies to date, however, have not specifically addressed whether acceptable ITT survival can be achieved with LT in a group presenting beyond conventional DS criteria. In the present multicenter, prospective study, ITT survival in the AC group was 58% at 3 years, which was not too dissimilar from the UNOS-DS group (69%; P =.05). ITT survival at 3 years, however, markedly dropped to 30% in AC with either a risk factor of CP B/C cirrhosis or an AFP > 500. LRT itself is certainly not without risk, as patients with decompensated cirrhosis are at increased risk of LRT-related mortality and hepatic decompensation that would preclude further treatment.^2,12 Thus, as expected, CP B/C cirrhosis was also associated with an increased risk of waitlist dropout in the AC cohort. Similarly, elevated AFP is correlated with more aggressive tumor biology and an increased risk of waitlist dropout.^12,21 Our study suggests that DS is safe and feasible with acceptable ITT survival in selected patients with extensive tumor burden, though likely only in those with well-compensated liver disease and favorable tumor biology.

Though longer follow-up is still needed, observed posttransplant survival and early HCC recurrence were similar between the AC and UNOS-DS cohorts. Furthermore, HCC recurrence has been observed in only 5% of ACs to date. A recent retrospective study found comparable post-LT survival at 1, 5, and 10 years between these 2 cohorts as well, though it did note higher recurrence rates in patients with tumor burden beyond UNOS-DS.²² A prior pooled analysis also suggested higher posttransplant HCC recurrence rates of at least 16% in patients who underwent DS.²⁴ Our low observed rate of recurrence to date may reflect the beneficial effects of a uniform, prospective DS protocol as well as a mandatory wait time prior to LT listing. In carefully selected AC patients with favorable tumor biology, as indicated by an objective response to LRT, stability during a mandatory wait time, and low AFP, it appears that similar post-LToutcomes to those of the more widely adopted UNOS-DS criteria can be achieved.

Although post-LT outcomes and explant pathology were similar between AC and UNOS-DS cohorts, it is important to note that nearly 40% were understaged on explant, with multiple prior studies reporting an association between understaging and increased post-LT HCC recurrence and mortality.^25–27 Moreover, 1 retrospective study of the UNOS database found a 33% to 41% understaging rate in patients requiring DS compared with 17% among those who were always within Milan.²³ In the present combined UNOS-DS and AC cohort, the sum of the largest lesion plus the number of lesions before LTwas associated with explant tumor beyond Milan criteria, with the odds of understaging increasing by 35% per unit increase in this sum. AFP ≥ 20 ng/mL was also associated with understaging. Interestingly, both of these variables were associated with HCC-specific deaths in the Metroticket 2.0 calculator.¹⁹ Our findings also support those of Mahmud et al²⁵ who described pre-LT AFP, multiple tumors, and the need for DS with LRTas risk factors for being understaged on pathology in a national explant analysis. Together, these studies highlight the need for ongoing efforts to improve understaging. Especially in patients requiring DS, radiographic evaluation of treatment response is often not straightforward after LRT, though the use of MRI may be helpful. Additionally, continuation of LRT after successful DS until complete radiographic response and normalization of AFP is a common clinical practice, particularly at centers with long wait times.¹²

These findings also bring up the important question of whether AC patients warrant higher priority for LT after successful DS compared with those with earlier-stage HCC. Not only do patients initially beyond Milan criteria have increased transplant survival benefit,²⁸ but they also have a higher urgency for LT (ie, increased waitlist dropout) as well. For example, using the UNOS database, Huang et al²⁹ recently showed a significantly higher risk of 2-year waitlist dropout after initial successful DS in the “AC-DS” (30%) group compared with the UNOS-DS (25%) and Milan (19%) groups. For these reasons, it has been proposed that DS patients receive the highest priority for LT.³⁰ However, this must be balanced with the observation that AC patients have higher risk of explant understaging and worse post-LT survival in the UNOS database.²³ One possible way to balance these competing aspects of pre-LTurgency and post-LTutility in AC patients is a recent proposal³¹ to give increased priority (eg, award median MELD at transplant (MMAT) after a 6-month wait rather than MMAT-3) based on the increased risk of waitlist dropout that takes into account AFP, tumor burden, and CP/MELD score. However, increased priority would be given only up to the point when expected post-LT survival begins to fall. For example, patients requiring DS who have AFP > 100 ng/mL or progressive disease (within Milan criteria) during the observation period have poor post-LT survival^23,32 and thus it would not be appropriate for such patients to receive additional priority.

The present study describes the results of the first multicenter prospective study evaluating outcomes of patients with HCC beyond UNOS-DS criteria. Importantly, none of the beyond UNOS-DS/AC patients in this study have been included in any publication on the topic to date. There are, however, some important limitations to note. First, as patients were prospectively enrolled from 2015–2020, regression analysis of post-LT outcomes between DS cohorts is not yet possible given the relatively short post-LT follow-up. Additionally, the type of LRT used for DS was left to the discretion of each center’s tumor board and was not prespecified in the study protocol. Type of initial LRT, however, was not associated with successful DS, receipt of LT, or ITT survival. Prior pooled subgroup analysis also did not find differences in post-LT HCC recurrence rates when comparing transarterial chemoembolization vs transarterial radioembolization.²⁴ Finally, lead-time bias may explain differences in survival between the cohorts; patients in the UNOS-DS cohort may appear to have longer survival because their disease was detected earlier and LRT was begun prior to tumor progression to AC criteria.

In conclusion, in this prospective multiregional study comparing patients within UNOS-DS vs AC cohorts in the MERITS-LT consortium, we observed an overall 65% probability of successful DS in AC, with increasing pretreatment tumor burden associated with decreased likelihood of DS. Three-year ITT survival in AC was nearly 60%, though AC patients with either AFP > 500 or CP B/C had a particularly poor survival of 30%. Explant pathology, post-LT survival, and post-LT HCC recurrence were similar between cohorts. Together, these findings suggest that LT is a reasonable goal for selected AC patients who achieve successful DS, especially those with compensated liver disease and favorable tumor biology, as suggested by low AFP. In these patients who do achieve transplant, the survival benefit is significant, which highlights that expanded transplant criteria should look beyond size alone.

Data availability

The data that support the findings of this study are available on request from the corresponding author.

Abbreviations:

AC

all-comers

AFP

alpha-fetoprotein

CI

confidence interval

CP

Child-Pugh

CT

computed tomography

CTP

Child-Turcotte-Pugh

DS

downstaging

HR

hazard ratio

HCC

hepatocellular carcinoma

IQR

interquartile range

ITT

intention-to-treat

LRT

local regional therapy

LT

liver transplantation

MELD

model for end-stage liver disease

MERITS-LT

multicenter evaluation of reduction in tumor size before liver transplantation

MRI

magnetic resonance imaging

NAFLD

nonalcoholic fatty liver disease

TACE

transarterial chemoembolization

UCSF

University of California, San Francisco

UNOS

united network for organ sharing

Y-90

Yttrium-90