Abstract
Background:
Liver transplantation (LT) is a well-established treatment for hepatocellular carcinoma (HCC), but there are ongoing debates regarding outcomes and selection. This study examines the experience of LT for HCC at a high-volume centre.
Methods:
A prospectively maintained database was used to identify HCC patients undergoing LT from 2000 to 2020 with more than or equal to 3-years follow-up. Data were obtained from the centre database and electronic medical records. The Metroticket 2.0 HCC-specific 5-year survival scale was calculated for each patient. Kaplan–Meier and Cox-regression analyses were employed assessing survival between groups based on Metroticket score and individual donor and recipient risk factors.
Results:
Five hundred sixty-nine patients met criteria. Median follow-up was 96.2 months (8.12 years; interquartile range 59.9–147.8). Three-year recurrence-free (RFS) and overall survival (OS) were 88.6% (n=504) and 86.6% (n=493). Five-year RFS and OS were 78.9% (n=449) and 79.1% (n=450). Median Metroticket 2.0 score was 0.9 (interquartile range 0.9–0.95). Tumour size greater than 3 cm (P=0.012), increasing tumour number on imaging (P=0.001) and explant pathology (P<0.001) was associated with recurrence. Transplant within Milan (P<0.001) or UCSF criteria (P<0.001) had lower recurrence rates. Increasing alpha-fetoprotein (AFP)-values were associated with more HCC recurrence (P<0.001) and reduced OS (P=0.008). Chemoembolization was predictive of recurrence in the overall population (P=0.043) and in those outside-Milan criteria (P=0.038). A receiver-operator curve using Metroticket 2.0 identified an optimal cut-off of projected survival greater than or equal to 87.5% for predicting recurrence. This cut-off was able to predict RFS (P<0.001) in the total cohort and predict both, RFS (P=0.007) and OS (P=0.016) outside Milan. Receipt of donation after brain death (DBD) grafts (55/478, 13%) or living-donor grafts (3/22, 13.6%) experienced better survival rates compared to donation after cardiac death (DCD) grafts (n=15/58, 25.6%, P=0.009). Donor age was associated with a higher HCC recurrence (P=0.006). Both total ischaemia time (TIT) greater than 6hours (P=0.016) and increasing TIT correlated with higher HCC recurrence (P=0.027). The use of DCD grafts for outside-Milan candidates was associated with increased recurrence (P=0.039) and reduced survival (P=0.033).
Conclusion:
This large two-centre analysis confirms favourable outcomes after LT for HCC. Tumour size and number, pre-transplant AFP, and Milan criteria remain important recipient HCC-risk factors. A higher donor risk (i.e. donor age, DCD grafts, ischaemia time) was associated with poorer outcomes.
Keywords: hepatocellular carcinoma, liver transplantation, recurrent hepatocellular carcinoma
Introduction
Highlights
Five-year recurrence-free (RFS) and overall survival (OS) after liver transplantation (LT) for hepatocellular carcinoma (HCC) were 78.9% (n=449) and 79.1% (n=450), respectively.
Candidates transplanted for HCC inside Milan (P<0.001) and University of California, San Francisco (UCSF) Criteria (P<0.001) had better outcomes than those outside those criteria.
An optimized cut-off of greater than or equal to 87.5% predicted recurrence-free survival using Metroticket 2.0 (P<0.001) in the total cohort and predicted both RFS (P=0.007) and OS (P=0.016) after LT of outside-Milan candidates.
High-risk graft features including donation after cardiac death (DCD) grafts (P=0.009), increasing donor age (P=0.006) and total ischaemia time (P=0.016) predicted unfavourable transplant outcomes. The deleterious effect of DCD grafts was upheld in LT outside Milan (P=0.033) and after removing early LT-associated complications (P=0.046).
Hepatocellular carcinoma (HCC) accounts for 90% of primary liver cancer world-wide and is increasing in incidence and mortality in North America1. Underlying liver cirrhosis is present in a significant proportion of HCC patients and complicates management. Curative options include liver resection, microwave ablation, and liver transplantation (LT); however, LT is more suitable for those with concomitant underlying cirrhosis or multifocal disease2.
LT emerged as a treatment for HCC in 1996 following the publication by Mazzaferro et al. 3 outlining the Milan Criteria. These criteria identified that patients with a single tumour of less than 5 cm in size or 2–3 tumours less than 3 cm had a 75% overall survival (OS) and 83% recurrence-free survival (RFS)3. The Milan criteria have since been incorporated into the United Network for Organ Sharing (UNOS) staging system for HCC and became the mainstay for guiding organ allocation for deceased donor transplantation4.
Since then, several studies have investigated the impact of expanding the Milan Criteria. The University of California, San Francisco (UCSF) expanded criteria was one of the first studies to broaden the size limitations of the Milan Criteria with no significant difference in overall survival5,6. As the treatment of HCC with LT has risen, so has the need to identify the impact of tumour biology, pre-transplant recipient factors and donor characteristics that may impact survival and tumour recurrence. Prognostic models, such as the Metroticket 2.0, have utilized these associations to predict the risk of HCC-specific mortality after LT with the goal to refine current transplantation criteria for HCC7.
The Milan system represents a binary classification, which may under-appreciate the nuanced continuum of risk that could apply to each patient. The Metroticket scale, first described in 2013 and updated in 2018, creates a prognostic model for determining the percent risk of mortality after LT for HCC, and includes both, static (HCC number and size) and biological recipient risk factors (AFP levels)7,8. This model uses staged cut-offs for AFP based on tumour size and incorporates the number of active nodules. This model is preferred to incorporate the so-called “two hit hypothesis” in HCC, referring that two key factors influence recurrence: tumour burden and liver function7.
Based on an increasing number of novel concepts for HCC downstaging therapies with better efficiency and refined acceptance criteria for candidate’s admission to the liver waiting list, it is important to assess additional contributors to tumour recurrence other than recipient and HCC related factors.
This study examines the experience of patients with HCC treated with LT in two high-volume centres. The large cohort and long-term follow-up in this study provides a unique opportunity to investigate the impact of several donor and graft-specific risk factors on overall and recurrence-free survival in the context of different recipient risk profiles.
Methods
This study was a retrospective analysis conducted at two high-volume transplant centres in a large institution. The study aimed to assess the outcomes of LT in patients diagnosed with HCC and to identify variables predictive of tumour recurrence following transplantation. Ethical approval was obtained from the Institutional Review Board on 11 May 2017. The study is also registered on clinicaltrials.gov (NCT06018857, https://clinicaltrials.gov/study/NCT06018857?term=NCT06018857&rank=1). No patients or public support was received for this research. All work has been reported in concordance with the STROCSS criteria for reporting of retrospective cohort studies9.
Data collection and participants
Patients undergoing LT with histologically-confirmed HCC between 2000 and 2020 with more than 3 years of follow-up or who expired before this time were identified from a prospectively maintained database which contains detailed information on patient demographics, tumour characteristics, preoperative imaging and treatments, surgical procedures, and follow-up outcomes. Inclusion criteria consisted of patients with a confirmed diagnosis of HCC who underwent LT. Patients were excluded who did not receive a liver transplant or did not have a diagnosis of HCC. No patients at our centre are treated empirically after liver transplantation with immunotherapy, chemotherapy, or other anti-cancer agents prior to a recurrence. Donor information that was not present in the database was supplemented with United Network for Organ Sharing (UNOS) centre-level data and the electronic medical record. Data for LTs prior to 2010 was limited as this was prior to implementation of the electronic records system, specifically ischaemic times and donor parameters were limited in patients prior to 2010. The Metroticket 2.0 5-year HCC-specific survival predictor was calculated for each patient using the online calculator (http://www.hcc-olt-metroticket.org/index.html)7,10,11.
Variables and definitions
The primary outcome of interest was tumour recurrence after liver transplantation. Recurrence was diagnosed based on radiological evidence of new lesions or histological confirmation using the Liver Imaging Reporting & Data System (LI-RADS) criteria recommended by the American Association for the Study of Liver Diseases (AASLD)12. Both intra-hepatic and extra-hepatic recurrences were considered as positive in evaluation of the primary outcome. Lesion size was recorded, but lesions of any size were considered as positive per LI-RADS. Secondary outcomes included overall survival, disease-free survival, and post-transplant complications.
Predictor variables included demographic characteristics (age, sex), aetiology of liver disease, pre-transplant treatments (trans-arterial chemoembolization, Y-90 therapy, or radiofrequency ablation), tumour characteristics (size, number of lesions, presence of vascular invasion), and laboratory parameters [alpha-fetoprotein levels, Model for End-Stage Liver Disease (MELD) score]. Tumour size refers to preoperative imaging unless specified that we are referring to pathology. This approach was chosen as the imaging represents the information available to clinicians when making decisions regarding donor and recipient selection. Pathologic response was defined by pathologist report of the explanted liver as none, partial or complete, with complete representing the absence of viable tumour as reported by the reading pathologist. The study additionally examined the impact of donor factors including cold ischaemia time (CIT), warm ischaemia time (WIT), graft type, donor age, sex, gender, BMI (kg/m2). Recipient WIT (rWIT) refers to the graft implantation time, and donor WIT (dWIT) in DCD donors refers to the time from withdrawal of life support to the initiation of cold organ preservation (total donor warm ischaemia time)13,14. Total Ischaemia Time (TIT) is defined as dWIT+CIT+rWIT. The Balanced Assessment of Risk (BAR) score (https://www.assessurgery.com/bar-score/) was calculated for each patient15 and the UK-DCD-Risk score was calculated for DCD grafts14.
Patients were divided for sub-group analysis based on whether they were “within Milan” or “outside Milan” criteria using the traditional definition described by Mazzaferro et al. 3. For the purposes of this analysis, expanded criteria here refers to patients who received LT but did not meet Milan criteria (outside Milan). Patients were also categorized into UCSF criteria using the standard definition5. However, because the Milan criteria have been implemented in our centre since study initiation, we felt using these criteria for a primary analysis was most appropriate compared with other criteria described in the literature.
Statistical analysis
Descriptive statistics were used to summarize the demographic and clinical characteristics of the study population. Continuous variables were reported as median with interquartile range (IQR), while categorical variables were presented as frequencies (n) and percentages.
Survival analysis using the Cox-proportional hazards regression model was conducted to evaluate the association between predictor variables and the risk of tumour recurrence. Hazard ratios (HRs) along with 95% CIs were estimated. Kaplan–Meier analysis was employed to construct survival curves and calculate the median time of recurrence-free and overall survival for all patients and for those undergoing expanded-criteria transplantation. The log-rank test was used to compare survival distributions between different subgroups or treatment modalities. A Receiver-Operator Analysis was conducted, aiming to identify the ideal cut-off to maximize the utility of the Metroticket 2.0 predictor of mortality at also predicting recurrence in this population.
All statistical analyses were conducted using SPSS version 28 (IBM Corp.) and a P value less than 0.05 was considered statistically significant.
Results
A total of 569 patients met inclusion criteria. Demographic and medical information, both donor and recipient, is available in Table 1. The median follow-up was 96.2 months (IQR 59.9–147.8; 8.02 years). The median time on the waitlist was 64 days (IQR 28–157). About half of patients (n=264, 46.4%) overall had partial or complete pathologic response to downstaging therapy found on recipient hepatectomy.
Table 1.
Demographic information of patients undergoing LT for HCC overall and by inclusion in Milan criteria.
| Total (N=569) | Inside Milan (N=469) | Outside Milan (N=100) | |
|---|---|---|---|
| Recipient parameter | |||
| Age (year) | 60 (56–65.5) | 62 (57– 67) | 60 (55–65) |
| Male sex, N (%) | 429 (75.4) | 348 (74.2) | 81 (81) |
| BMI (kg/m2) | 29 (25–33) | 29 (15–33) | 28 (25–31.8) |
| Underlying liver disease, N (%) | |||
| Hepatitis B | 35 (6.2) | 30 (6.4) | 5 (5) |
| Hepatitis C | 322 (56.6) | 259 (55.2) | 63 (63) |
| NASH | 82 (14.4) | 70 (14.9) | 12 (12) |
| Alcoholic cirrhosis | 117 (20.6) | 97 (20.7) | 20 (20) |
| Lab MELD (points) | 12 (8–16) | 12 (8.25–17) | 11 (8–18) |
| AFP (ng/ml) | 9 (4.5–31) | 8.6 (4.2–27) | 3 (2.2–4.5) |
| Days on Waitlist | 64 (28–157 | 65 (27–159) | 31 (63–152) |
| Immunosuppression, N (%) | |||
| Tacrolimus | 564 (99.1) | 464 (98.9) | 100 (100) |
| Mycophenolate | 498 (87.5) | 414 (88.3) | 84 (84) |
| Prednisone | 556 (97.7) | 461 (98.3) | 95 (95) |
| Sirolimus | 196 (34.4) | 158 (33.7) | 38 (38) |
| Pre-treatment with Sorafenib, N (%) | 46 (8.1) | 23 (4.9) | 23 (23) |
| Donor and graft parameter | |||
| Donor age (year) | 42.5 (29–54) | 41 (28.5–54) | 45 (30–55.3) |
| Donor BMI (kg/m2) | 27.4 (23.7–31.5) | 27.5 (23.7–31.6) | 26.8 (23.9–29.8) |
| Graft type, N (%) | |||
| DBD | 478 (84.0) | 389 (82.9) | 89 (89) |
| DCD | 58 (10.2) | 48 (10.2) | 10 (10) |
| LDLT | 22 (3.9) | 21 (4.5) | 1 (1) |
| Male sex, N (%) | 345 (6.1) | 284 (60.6) | 61 (61) |
| Hepatitis B donor, N (%) | 40 (7.0) | 31 (6.6) | 9 (9) |
| Diabetes, N (%) | 82 (14.4) | 68 (14.5) | 14 (14) |
| rWIT (min) | 44 (36–51) | 44 (37–510) | 41 (34.3–49) |
| CIT (min) | 396 (336–480) | 392.5 (328.5–480) | 420 (350–480) |
| TIT (min) | 420 (307.5–492.8) | 417.5 (300–488.3) | 433 (365–505) |
Continuous variables are presented as median and IQR, categorial variables are presented as number (n) and %.
AFP, alpha-fetoprotein; CIT, cold ischaemia time; DBD, donation after brain death; DCD, donation after circulatory death; HCC, hepatocellular carcinoma; IQR, interquartile range; LDLT, living-donor liver transplantation; LT, liver transplantation; NASH, Non alcoholic steatohepatitis; rWIT, recipient warm ischaemia time, also known as implantation time; TIT, total ischaemia time.
Seventy-five patients (13.2%) experienced recurrence. Three-year RFS and OS were 88.6% (n=504/569) and 86.6% (n=493/569), respectively. Five-year RFS and OS were 79.1% (n=450/569) and 74.9% (n=426/569). Patients with recurrence were more likely to experience mortality compared to those who did not recur (P=0.029). Of the 75 patients with recurrence, 45 (45/75; 60%) experienced mortality attributable to HCC, 2 (2.7%) died of non-cancerous causes (both secondary to cardio-pulmonary decompensation), and 28 (37.3%) remain alive.
At the time of publication, 206 patients (36.2%) had died. The most common identifiable cause of death was recurrent HCC (n=45/206, 21.8%) followed by cardiac events (n=33/206, 16.0%), other malignancy (n=33/206, 16.0%), liver-related complications (n=24/206, 11.7%), neurological events (n=13/206, 6.3%) or other (n=51/206, 24.8%). Including deaths occurring within the same admission as the index transplant and those directly attributable to LT occurring outside the index admission, 44 patients had early mortality (<6-months) directly attributable to their liver transplantation. These complications (n=44) included recurrent viral hepatitis (n=11/44, 25%), infectious/septic episodes (n=7/44, 15.9%), ischaemic cholangiopathy (n=4/44, 9.1%), renal failure/fluid overload (n=4/44, 9.1%), primary graft nonfunction (n=3/44, 6.8%), cardiac arrest/acute myocardial infarction (n=4/44, 9.1%), chronic rejection (n=2/44, 4.5%), Graft-vs.-host Disease (n=2/44, 4.5%), uncontrolled bile leak (n=1/44, 2.3%), post-operative haemorrhage (n=1/44, 2.3%), Sirolimus-induced lung injury (n=1/44, 2.3%), failure-to-thrive (n=2/44, 4.5%), and other/unknown causes (n=2/44, 4.5%).
Recipient factors
An increasing number of tumour lesions evaluated as continuous variable on preoperative imaging (P=0.001) and on explant pathology (P<0.001) were each associated with a higher recurrence rate. Recurrence rates were 23.5% for tumours greater than or equal to 3 cm (n=32/136) compared to 15.9% in those less than 3 cm (n=40/259; P=0.012). Over half of patients (n=364, 63.9%) received pre-transplant loco-regional therapy (LRT) of some variety. Patients outside of Milan were more likely to receive LRT (n=85, 85%, P<0.001). Patients outside-Milan criteria were more likely to receive TACE (n=80/100, 80.2%) vs. those inside Milan (n=172/327, 52.6%, P<0.001). Similarly, those outside Milan were more likely to receive Y-90 (14% vs. 7%, P=0.033), though there was no correlation between Milan criteria and radiofrequency ablation (RFA) (P=0.478).
We next looked at the impact of various LRT on patient outcomes for transplants performed inside-Milan criteria. A total of 469 patients underwent transplant within Milan criteria, of which 279 (59.5%) received LRT. Specifically, within Milan Criteria, a tumour size on preoperative imaging of greater than 3 cm (P=0.002) and a lab MELD of greater than 11 points (P<0.001) were associated with an increased use of LRT. No correlation was seen between LRT and an increasing tumour number (P=0.133) or an AFP greater than 400 (P=0.107). In the cohort inside-Milan, the use of any LRT was not associated with lower recurrence rates (P=0.881) or a better three-year RFS (P=0.883). When sub-dividing by type of LRT within the inside-Milan cohort, RFA was not associated with pathologic tumour response (P=0.293), lympho-vascular invasion (P=0.496), recurrence rates (P=0.195) or three-year RFS (P=0.108). In contrast, TACE was associated with pathologic tumour response (P<0.001), but not with lympho-vascular invasion (P=0.749), recurrence rates (P=0.293) or three-year RFS (P=0.857). Y-90 therapy was not associated with pathologic response (P=0.173), lympho-vascular invasion (P=0.096), recurrence rates (P=0.556) or 3-year RFS (P=0.420). Finally, for patients within Milan Criteria, time on waitlist was not associated with recurrence rates (Mann–Whitney U=6587, P=0.993).
As expected, most patients classified within Milan criteria were also inside UCSF criteria (n=324/327, 99.0%). In contrast, the cohort within UCSF criteria included a significant number of patients who were outside of Milan criteria (n=56/100, 56%, P<0.001). Transplant for those inside-Milan criteria had reduced recurrence rates (n=36/278, 12.9%) compared to those outside Milan (29/93, 31.2%, P<0.001). Similarly, LT inside of UCSF criteria had reduced recurrence rates (n=15/385, 15.3%) vs. outside of UCSF (n=16/45, 35.6%, P<0.001). Increasing alpha-fetoprotein (AFP) was associated with an increased recurrence (P<0.001) and a reduced survival (P=0.008). Recipient AFP greater than 400 ng/ml was also associated with a higher rate of recurrence (P=0.003). In contrast, recipient age, gender, BMI, and immunosuppression regimen were not associated with higher recurrence rates [Table 2]. Interestingly, increased recipient time on the waiting list prior to LT was not associated with higher recurrence rates (P=0.547).
Table 2.
Univariate analysis of factors predictive of recurrence and mortality after LT for HCC.
| Recurrence | P | Recurrence within 3 years | P | Mortality within 3 years | P | |
|---|---|---|---|---|---|---|
| Recipient parameter, n (%) | ||||||
| Age>50 years | 4/47 (8.5) | 0.105 | 4/54 (7.4) | 0.377 | 5/41 (12.2) | 0.951 |
| Age<50 years | 68/380 (22) | 51/403 (12.7) | 59/471 (12.5) | |||
| BMI>30 kg/m2 | 24/151 (15.8) | 0.504 | 21/204 (10.3) | 0.892 | 24/204 (11.8) | 0.752 |
| BMI<30 kg/m2 | 51/278 (18.3) | 37/347 (10.7) | 44/347 (12.7) | |||
| Male | 63/338 (18.6) | 0.210 | 50/429 (11.7) | 0.357 | 50/429 (11.7) | 0.111 |
| Female | 12/92 (13) | 12/136 (8.8) | 23/136 (16.9) | |||
| Underlying liver disease, n (%) | ||||||
| HCV | 50/261 (19.2) | 0.244 | 36/322 (11.2) | 0.835 | 44/322 (13.7) | 0.805 |
| NASH | 7/51 (13.7) | 0.456 | 7/82 (8.5) | 0.374 | 10/82 (12.2) | 0.738 |
| EtOH | 13/82 (18.9) | 0.674 | 11/117 (9.4) | 0.440 | 9/117 (7.7) | 0.043 |
| PBC | 2/19 (10.6) | 0.416 | 2/23 (8.7) | 0.675 | 4/23 (17.4) | 0.561 |
| Non-HCV viral hepatitis | 52/169 (18.6) | 0.407 | 38/306 (12.4) | 0.798 | 47/344 (13.7) | 0.336 |
| Creatinine>1.6 mg/dl, n (%) | 4/50 (8) | 0.046 | 4/61 (6.6) | 0.282 | 7/65 (10.7) | 0.586 |
| Creatinine<1.6 mg/dl, n (%) | 69/378 (18.3) | 51/455 (11.2) | 60/455 (13.2) | |||
| MELD>11 points, n (%) | 38/216 (17.6) | 0.700 | 30/201 (14.9) | 0.235 | 28/231 (12.1) | 0.972 |
| MELD<11 points, n (%) | 34/211 (19.2) | 21/208 (10) | 28/229 (12.2) | |||
| No. lesions on imaging, n (%)a | ||||||
| 0 | 4/56 (7.7) | 0.001 | 3/69 (4.3) | <0.001 | 6/69 (8.7) | <0.001 |
| 1 | 30/208 (14.4) | 22/269 (8.2) | 25/269 (9.3) | |||
| 2 | 14/77 (18.2) | 9/100 (9) | 8/100 (8) | |||
| 3 | 11/44 (25) | 8/58 (13.8) | 11/58 (19.0) | |||
| 4+ | 15/44 (34.1) | 14/48 (29.2) | 17/31 (5.5) | |||
| Within Milan, n (%) | ||||||
| Yes | 36/278 (12.9) | <0.001 | 23/327 (7.0) | <0.001 | 32/327 (9.8) | 0.002 |
| No | 29/93 (31.2) | 24/101 (23.8) | 22/101 (21.8) | |||
| Neoadjuvant, n (%) | ||||||
| RFA | 13/57 (22.8) | 0.262 | 17/118 (14.4) | 0.253 | 27/118 (22.9) | <0.001 |
| TACE | 46/218 (21.1) | 0.043 | 41/309 (13.3) | 0.131 | 47/309 (15.2) | 0.157 |
| Y-90 | 7/17 (41.2) | 0.008 | 11/60 (18.3) | |||
| No. lesions (pathology), n (%) | ||||||
| 1 | 26/237 (10.9) | <0.001 | 17/266 (6.4) | <0.001 | 24/266 (9) | 0.007 |
| 2 | 15/89 (16.9) | 10/97 (10.3) | 12/97 (12.4) | |||
| 3 | 6/41 (14.6) | 4/46 (8.7) | 10/46 (21.7) | |||
| 4+ | 25/60 (41.7) | 21/86 (24.4) | 14/86 (16.3) | |||
| Lesion size>3 cm, n (%) | 32/136 (23.5) | 0.012 | 25/153 (16.3) | 0.003 | 20/153 (13.1) | 0.620 |
| Lesion size<3 cm, n (%) | 40/291 (15.9) | 26/339 (7.7) | 39/339 (11.5) | |||
| Vascular invasion, n (%) | 36/93 (38.7) | <0.001 | 30/117 (25.6) | <0.001 | 22/117 (18.8) | 0.012 |
| No invasion, n (%) | 36/333 (18) | 22/377 (5.8) | 38/377 (10.0) | |||
| Immunosuppression, n (%)b | ||||||
| Prograf | 75/425 (17.6) | 0.301 | 65/564 (11.5) | 0.420 | 76/564 (13.5) | 0.378 |
| Mycophenolate | 64/359 (17.8) | 0.636 | 56/498 (11.2) | 0.723 | 73/498 (14.7) | 0.016 |
| Prednisone | 72/418 (17.2) | 0.484 | 63/556 (11.3) | 0.650 | 75/556 (13.5) | 0.544 |
| Sirolimus | 31/157 (19.7) | 0.340 | 24/172 (13.9) | 0.655 | 27/196 (13.8) | 0.831 |
| Induction therapy | ||||||
| Yes | 32/190 (16.8) | 26/239 (10.9) | 24/239 (10.0) | |||
| No | 43/247 (17.4) | 0.876 | 39/330 (11.8) | 0.728 | 52/330 (15.8) | 0.048 |
| Sorafenib, n (%) | 16/43 (37.2) | <0.001 | 14/32 (43.8) | <0.001 | 11/46 (23.9) | 0.028 |
| No sorafenib, n (%) | 59/387 (15.2) | 51/523 (15.8) | 65/523 (12.4) | |||
| Donor and graft parameter, n (%) | ||||||
| Donor Age>40 years | 36/208 (17.3) | 0.888 | 37/268 (13.8) | 0.199 | 37/268 (13.8) | 0.955 |
| Donor Age<40 years | 27/151 (17.9) | 19/193 (4.7) | 27/193 (14.0) | |||
| Donor BMI>30 kg/m2 | 17/115 (14.8) | 0.344 | 18/158 (11.4) | 0.720 | 23/158 (14.6) | 0.762 |
| Donor BMI<30 kg/m2 | 46/244 (18.9) | 38/303 (12.5) | 41/262 (15.6) | |||
| ABO incompatible | 4/16 (25) | 0.417 | 6/19 (31.6) | 0.043 | 6/25 (24) | 0.110 |
| ABO compatible | 371/414 (17) | 59/544 (10.8) | 70/544 (12.9) | |||
| CDC high viral transmission Risk16 c | 14/63 (22.2) | 0.279 | 15/111 (13.5) | 0.440 | 18/111 (16.2) | 0.324 |
| CDC normal risk | 61/367 (16.6) | 50/458 (10.9) | 58/458 (12.7) | |||
| With diabetes | 11/55 (20) | 0.592 | 11/82 (13.4) | 0.540 | 14/82 (17.1) | 0.371 |
| Without diabetes | 64/375 (17.1) | 54/487 (11.1) | 62/487 (12.7) | |||
| Sex, n (%): | ||||||
| Male | 49/260 (18.8) | 0.343 | 45/345 (13.0) | 0.132 | 48/345 (13.9) | 0.628 |
| Female | 26/170 (15.3) | 20/224 (8.9) | 28/224 (12.5) | |||
| HBV positive | 9/35 (25.7) | 0.178 | 10/40 (25) | 0.005 | 7/40 (17.5) | 0.638 |
| HBV negative | 66/395 (16.7) | 55/529 (16.7) | 69/529 (13.0) | |||
| Race donor, n (%) | ||||||
| White | 60/325 (18.5) | 0.545 | 53/435 (12.2) | 0.534 | 58/435 (13.3) | 0.839 |
| Black | 12/78 (15.4) | 10/104 (9.6) | 13/104 (12.5) | |||
| Other | 3/27 (11.1) | 2/30 (6.7) | 5/30 (16.7) | |||
| Graft type, n (%) | ||||||
| DBD | 64/371 (17.3) | 0.765 | 55/478 (115) | 0.590 | 55/423 (13) | 0.009 |
| DCD | 9/44 (20.5) | 7/51 (13.7) | 15/58 (25.9) | |||
| LDLT | 1/9 (11) | 1/22 (4.5) | 3/22 (13.6) | |||
| tWITd>1 h, n (%) | 6/45 (13.3) | 0.136 | 5/41 (12.2) | 0.226 | 1/55 (1.8) | 0.086 |
| tWITd <1 h, n (%) | 16/235 (6.81) | 25/360 (6.9) | 29/346 (8.4) | |||
| CIT>6 h, n (%) | 46/240 (19.2) | 0.188 | 41/300 (13.7) | 0.155 | 40/300 (13.3) | 0.495 |
| CIT<6 h, n (%) | 16/118 (13.6) | 16/173 (9.2) | 27/173 (15.6) | |||
| Total ischaemia time>6 h, n (%) | 50/227 (22.0) | 0.356 | 48/354 (13.6) | 0.016 | 52/354 (14.7) | 0.292 |
| Total ischaemia time<6 h, n (%) | 19/132 (14.4) | 13/192 (6.8) | 22/192 (11.5) | |||
Continuous variables are presented as median and IQR, categorial variables are presented as number (n) and %; Note that denominator provided due to missing data in the database in some cases.
Statistical significance (P < 0.05) values are in bold.
CDC, Centers for disease control; CIT, cold ischaemia time; DBD, donation after brain death; DCD, donation after circulatory death; ETOH, Alcohol; HBV, Hepatitis B Virus; HCC, hepatocellular carcinoma; HCV, Hepatitis C Virus; IQR, interquartile range; LDLT, living-donor liver transplantation; LT, liver transplantation; MELD, model for end-stage liver disease; NASH, Non alcoholic Steatohepatitis; PBC, Primary Biliary Cirrhosis; RFA, radiofrequency ablation; TACE, trans-arterial chemoembolization; TIT, total ischaemia time; tWIT, total Warm Ischemia Time.
Last imaging. Before transplantation.
Primary immunosuppression immediately after transplantation.
High Risk for Viral Transmission by CDC guidelines16.
Duration of hepatectomy in the donor and recipient liver implantation time.
Increasing likelihood of survival calculated by the pre-operative Metroticket 2.0 5-year HCC-specific calculator was associated with a reduction in recurrence rates (P<0.001) and improvement in three-year survival (P=0.04), though the same survival benefit was not found at 5 years (P=0.304) in this cohort. A receiver-operator curve (ROC) Analysis was performed demonstrating an area under-the curve (AUC) of 0.724 (95% CI 0.659–0.788) [Fig. 1A]. A repeat ROC analysis was performed stratified by Hepatitis C status, yielding similar results in both the Hepatitis C Virus-positive and Hepatitis C Virus-negative groups [Fig. 1B] Overall analysis identified the optimum cut-off of 0.875, or a predicted five-year survival of 87.5% according to the Metroticket 2.0 scale, with an associated sensitivity of 0.792 and specificity of 0.573 at this cut-off. This yielded a positive predictive value (PPV) of 0.632 and negative predictive value (NPV) of 0.792. Using this established cut-off, a predicted 5-year survival greater than or equal to 87.5% according to the Metroticket 2.0 scale was associated with a longer time-to-recurrence (298 days) compared with a predicted survival less than 87.5% (243 days, Mann–Whitney U test=23454.5, P<0.001). A similar analysis did not demonstrate a significant correlation between predicted survival according to Metroticket 2.0 and actual survival time in days (P=0.094). Five-year RFS for those with Metroticket 2.0 greater than or equal to 87.5% was 83.2% (n=362/435) vs. 65.7% (n=88/134) in those with Metroticket 2.0 less than or equal to 87.5% (P<0.001). Five-year OS for those with Metroticket 2.0 greater than or equal to 87.5% was 77.9% (n=339/435) vs. 64.9% (n=87/134) in those with Metroticket 2.0 less than or equal to 87.5% (P=0.002).
Figure 1.
Metroticket 2.0 model performance for predicting recurrence in hepatocellular carcinoma (HCC) after liver transplantation (LT). (A) Receiver-operator curve examining the sensitivity and 1-specificity of the Metroticket 2.0 HCC-specific Five-year mortality calculator at predicting recurrence-free survival after LT for HCC. (B) Receiver-operator curve examining the sensitivity and 1-specificity of the Metroticket 2.0 HCC-specific 5-year mortality calculator at predicting recurrence-free survival after LT for HCC stratified by patient hepatitis C status, demonstrating that the model is applicable to patients with (green) and without (blue) HCV.
Donor and transplant factors
Transplantation with a DCD graft was associated with a reduced survival at 3 years (n=15/58, 25.6%) vs. DBD (n=55/423, 13%) or living-donor (LDLT: n=3/22, 13.6%, P=0.009), although graft type was not associated with overall recurrence rate (P=0.765) [Fig. 2]. Increasing donor age was associated with an increased chance of recurrence (P=0.006). Donor gender (P=0.343), race (P=0.545), BMI (P=0.344), diabetes (P=0.592) or Centers for disease control-high-risk features (P=0.279) did not influence recurrence rates. TIT of greater than 6 h was associated with increased recurrence (P=0.016). Donor WIT of greater than 20 min (P=0.326), total Warm Ischemia Time (tWIT) of greater than 45 min (P=0.829), tWIT greater than 60 min (P=0.136), and cold ischaemia time greater than 6 h (P=0.188) were not significantly correlated with recurrence rates when assessed individually. However, increasing TIT using Mann–Whitney testing was associated with increased recurrence (P=0.027). Increasing tWIT was not associated with increased recurrence (P=0.301).
Figure 2.
Survival analysis based on Metroticket 2.0 predicted survival after liver transplantation for hepatocellular carcinoma. Kaplan–Meier survival curves for patients split based on Metroticket 2.0 predicted survival greater than or equal to 87.5% (Green) or less than 87.5% (Red), using the cut-off established using the ROC analysis described and demonstrated in Fig. 1. (A) Recurrence-free survival (RFS) based on Metroticket 2.0. Metroticket 2.0 predicted survival greater than or equal to 87.5% is associated with improved recurrence free (RFS) (P<0.001). (B) Overall survival (OS) based on Metroticket 2.0. Metroticket 2.0 predicted survival greater than or equal to 87.5% approached ability to predict OS but was not significant (P=0.094). (C) RFS for patients outside of Milan Criteria. Metroticket 2.0 predicted survival greater than or equal to 87.5% is associated with improved recurrence free (RFS) (P=0.007). (D) OS for patients outside of Milan Criteria. Metroticket 2.0 predicted survival greater than or equal to 87.5% is associated with improved recurrence free (OS) (P=0.016).
A BAR score of greater than 9 points (n=41) was not associated with an increased recurrence rate in the overall population (P=0.076), or when sub-divided by DBD (P=0.061) or DCD grafts (P=0.428). A BAR of greater than 9 points was also not correlated with overall (P=0.810) or recurrence-free survival (0.755). Similarly, the UK DCD risk score was not associated with an increased recurrence rate (P=0.283) or reduced survival (P=0.743) in the DCD population.
A Cox-proportional hazard model consisting of Milan Criteria, tumour size greater than 3 cm, number of lesions, graft type, TIT and donor age was conducted, demonstrating that Inside-Milan Criteria (HR=0.424, P=0.009) and increasing number of tumours (HR=1.106, P=0.020) predicted recurrence. Additional model variables were not significant [Table 3].
Table 3.
Cox-proportional hazard model of factors significantly associated with recurrence after LT for HCC across all 569 patients undergoing LT for this indication.
| Hazard ratio (95% CI) | P | |
|---|---|---|
| Within Milan criteria | 0.424 (0.096–0.752) | 0.009 |
| HCC lesion size>3 cm | 1.422 (0.846–1.998) | 0.221 |
| No. lesions on preoperative imaging | 1.106 (1.063–1.149) | 0.020 |
| Total ischaemia time>6 h | 0.898 (0.538–1.258) | 0.764 |
| Graft type | 1.090 (0.53–1.65) | 0.877 |
| Donor age>40 years | 0.897 (0.337–1.186) | 0.708 |
HCC, hepatocellular carcinoma; LT, liver transplantation.
Statistical significance (P < 0.05) values are in bold.
Extended-criteria cohort
A total of 100 patients received LT for HCC that were outside of Milan, also referred to as extended criteria, at time of transplant. Three-year RFS and OS were 76% (n=76/100) and 79% (n=79/100). Five-year RFS and OS were 69% (n=69/100) and 68% (n=68/100). Liver transplantation with DCD grafts were associated with a reduced RFS at both, 3 years (n=5/10 vs. n=71/90, P=0.042) and 5 years (n=4/10 vs. n=65/90, P=0.037) post LT [Fig. 3A]. DCD grafts were also associated with a reduction in OS in this subpopulation (P=0.033) [Fig. 3B]. TIT did not affect recurrence (P=0.511) or survival (P=0.223) for transplants outside Milan. Pre-transplant treatments were also analyzed for their impact on recurrence and survival. Neoadjuvant TACE was associated with an increase in recurrence (n=27/74, 36.5%, P=0.038). In contrast, RFA (n=4/15, 26.7%, P=0.541) and Y-90 (n=4/14, 28.6%, P=0.289) were not associated with recurrence. No treatment modality was associated with survival in the extended-criteria cohort. As with the total cohort, increasing wait time was not correlated with an increase in recurrence rate (P=0.292) or OS (P=0.399) for patients outside-Milan criteria (P=0.292). Patients outside of Milan that died were more likely to die of HCC recurrence than their counterparts within Milan (P<0.0001).
Figure 3.
Survival analysis based on graft type after LT for HCC. Kaplan–Meier curves demonstrating the association between graft type and recurrence free (RFS) or overall survival (OS) after liver transplant for hepatocellular carcinoma. Graft types included brain dead (DBD, Blue), cardiac death (DCD, Green) or living donor (LDLT, Red). (A) RFS for all included patients by graft type. Graft type was not associated with reduction in recurrence in the complete cohort (P=0.765) (B) OS for all included patients by graft type. DCD was associated with a reduced survival (0.009). (C) RFS for LT outside of Milan criteria. LT outside of Milan had improved RFS with DBD vs. DCD grafts (P=0.039). (D) OS for LT outside of Milan criteria. LT outside of Milan had improved OS with DBD vs. DCD grafts (P=0.033).
Patients in the extended-criteria population who experienced disease recurrence had a higher predicted likelihood of recurrence according to the Metroticket 2.0 scale compared with those who did not experience recurrence (P=0.003). Application of the greater than or equal to 87.5% cut-off described in the complete cohort demonstrated continued significance in the extended-criteria cohort. Specifically, a Metroticket 2.0 predicted survival greater than or equal to 87.5% was associated with both an improved recurrence-free (P=0.007) and overall survival (P=0.016). Five-year RFS in the expanded-criteria cohort was 85.2% (n=23/27) for Metroticket 2.0 greater than or equal to 87.5% vs. 63.0% (n=46/73) for those less than 87.5% (P=0.033). Five-year OS in the expanded-criteria cohort was 92.6% (n=25/27) for Metroticket 2.0 greater than or equal to 87.5% vs. 58.9% (n=43/73) for those less than 87.5% (P=0.001) (Fig. 2).
Sub-analysis excluding early LT-associated mortality
We next removed the 44 patients with mortality within 6-months of transplant directly attributable to non-HCC-related complications, and a sub-analysis was performed using the Metroticket 2.0 criteria and the graft type in this remaining population of n=525 patients. As with the whole cohort, graft type was not associated with RFS (P=0.603). However, in this cohort, a DCD-graft was associated with a reduced OS compared with DBD or living-donor grafts (P=0.046) [Fig. 4A, B]. A Metroticket 2.0 score of greater than or equal to 0.875 remained predictive of 3-year RFS (P<0.001) and OS (P=0.004) [Fig. 4C, D].
Figure 4.
Survival analysis excluding early liver transplantation (LT)-associated mortality. Kaplan–Meier analysis for the subset of patients (n=525) that excludes patients who died of non-hepatocellular carcinoma causes directly attributable to early complications of their LT. (A) RFS for all included patients by graft type. Graft types included brain dead (DBD, Blue), cardiac death (DCD, Green) or living donor (LDLT, Red). Graft type was not associated with reduction in recurrence in this cohort (P=0.603). (B) OS for all included patients by graft type. DCD was associated with a reduced survival (0.046). (C) RFS for LT split based on Metroticket 2.0 predicted survival greater than or equal to 87.5% (Green) or less than 87.5% (Red). After excluding early LT-related complications, Metroticket predicted survival was able to predict RFS (P<0.001). (D) OS for LT split based on Metroticket 2.0 predicted survival. Again, Metroticket was predictive of survival (P=0.004).
Effect of downstaging
Seventy-six patients presented outside of Milan criteria and underwent attempted downstaging therapy. Forty-eight patients (63.2%) were successfully downstaged (Table 4). The most common treatment approach was trans-arterial chemoembolization (TACE, n=45/48, 93.8%) followed by Y-90 (n=7, 14.6%) and radiofrequency ablation (RFA, n=6, 12.5%). The 5-year recurrence-free and overall survival rate for the population that underwent successful downstaging was 70.8% (n=34) and 72.9% (n=35) respectively. The 5-year RFS and OS for those who were not successfully downstaged (n=28) was 53.5% (n=15) and 53.5% (n=15).
Table 4.
Neo-adjuvant therapy approach for patients (N=48) initially outside-Milan criteria who were successfully downstaged.
| Therapy | N (%) | 5-year RFS, N (%) | P | 5-year OS, N (%) | P |
|---|---|---|---|---|---|
| Total | 48 (100) | 34 (70.8) | — | 35 (72.9) | — |
| Trans-arterial chemoembolization (TACE) | 45 (93.8) | 31 (68.9) | 0.251 | 32 (71.1) | 0.276 |
| Radiofrequency ablation (RFA) | 6 (12.5) | 5 (83.3) | 0.481 | 6 (100) | 0.111 |
| Y-90 | 7 (14.6) | 4 (57.1) | 0.389 | 5 (71.4) | 0.924 |
P values listed compare the relevant outcome to that same outcome across all patients successfully downstaged.
OS, overall survival; RFS, recurrence-free survival.
Patients presenting with an HCC outside Milan who were successfully downstaged (n=48) did not experience a different 3-year (P=0.104) or 5-year (P=0.147) RFS compared to those that were not successfully downstaged (n=28) but still underwent transplantation. The population who was downstaged did experience an improvement in three-year overall survival (P=0.026) compared to those who failed downstaging but still received transplant. This survival benefit was not significant at 5-years post-transplant (P=0.066).
In the 48 patients who were successfully downstaged, the method of downstaging was not correlated with alterations in RFS or OS at 3 or 5 years. Interestingly, an AFP greater than 400 mg/dl was not predictive in this cohort. The Metroticket 2.0 cut-off of 87.5% was correlated with 5-year OS (P=0.046), but not 3-year RFS (P=0.146), 5-year RFS (P=0.146) or 3-year OS (P=0.066). Liver transplantation with a DCD-graft was associated with an improved 3-year RFS (P=0.010) and OS (0.022), although this was not significant at 5 years for RFS (P=0.078) or OS (P=0.055).
Analysis by Era
The highest proportion of patients outside-Milan criteria were transplanted from 2005 to 2010 (n=55/230, 24.9%), followed by 2010–2015 (n=32/193, 16.6%), before 2005 (n=5/33) and finally after 2015 (n=8/112, 7.1%, P=0.002) [Fig. 5A]. The highest proportions of patients with Metroticket 2.0 greater than 0.875 were transplanted before 2005 (n=12/33, 36.4%) followed by 2005–2010 (n=65/230, 28.3%), 2010–2015 (n=42/193, 21.8%) and after 2015–2020 (n=14/112, 12.5%, P=0.003) [Fig. 5B]. Interestingly, there was no difference in RFS (P=0.162) or OS (P=0.149) [Fig. 5C-D], despite the mentioned differences in baseline risk of each population. The impact of Milan and Metroticket greater than or equal to 0.875 criteria were also investigated within each era. Transplants outside Milan had higher recurrence rate vs. transplants within Milan from 2005 to 2010 (60% vs. 40%, P<0.001), 2010 to 2015 (39% vs. 61%, P<0.001) and 2015 to 2020 (0% vs. 100%, P=0.003). Transplants outside Metroticket 2.0 greater than or equal to 0.875 similarly had increased recurrence rates from 2005 to 2010 (60% vs. 40%, P<0.001), 2010 to 2015 (53% vs. 47%, P=0.001) and 2015 to 2020 (75% vs. 25%, P<0.001).
Figure 5.
Risk profile and outcomes of liver transplantation (LT) for HCC by era. (A) Proportion of patients inside (green) and outside (blue) Milan criteria by each era of our program divided into half-decade blocks, from 2000 to 2005, 2005 to 2010, 2011 to 2015, 2016 to 2020. (B) Proportion of patients inside (green) and outside (blue) Metroticket greater than or equal to 0.875 criteria by each era of our program divided into half-decade blocks. (C) Recurrence-free survival was not different (P=0.162) between eras despite significant differences in risk profile. (D) Overall survival was not different (P=0.149) between eras despite significant differences in risk profile.
Discussion
This large, two-centre study contains granular, patient-level detail and, with a median follow-up over 8 years, this represents a very long-term observation period after LT for HCC. Our study has the following main findings: First, the RFS and OS in our cohort were slightly improved compared with previous studies at both, 3 and 5 years, although the range could be considered comparable in context of ongoing medical advances17–19. Second, our study supports prior findings that both the Milan and UCSF criteria can predict recurrence-free and overall survival3,4,6,11. Third, increasing tumour size, number, and alpha-fetoprotein (AFP) were associated with higher recurrence rates and reduced time-to-recurrence. Like Nutu et al. 20, we found that pre-LT chemoembolization was associated with a higher rate of recurrence. Fourth, this study is unique in its thorough evaluation of donor-specific factors with the finding that DCD grafts demonstrated reduced survival overall for those who received riskier grafts and who are extended-criteria recipients. Finally, the Metroticket 2.0 score was able to successfully predict recurrence overall and to predict both recurrence and survival in those outside of Milan criteria7.
The allocation of liver grafts in the setting of HCC has been debated since the advent of liver transplant and came into clear focus with Mazzaferro’s seminal 1996 work on the Milan Criteria3. Since that time, efforts to expand criteria and increase the number of eligible recipients has been plentiful, including the presentation of UCSF criteria with expanded size criteria5. Our study, like others, confirms that the Milan Criteria and the UCSF criteria can predict recurrence and survival after LT for HCC6,21. Transplants occurring outside of Milan or UCSF criteria were associated with worse outcomes, and patients in this study who died after LT outside Milan were more likely to die of HCC recurrence compared to their standard criteria counterparts. We further found a 3-year RFS of less than 70% in candidates with an HCC classified outside of Milan, compared to those candidates with a tumour inside Milan with greater than 80%, which is consistent with prior works citing outside-Milan RFS below 70% and inside-Milan RFs of 80–90%6,22,23. Further, our study supports many prior works in finding that MELD score was not associated with recurrence or survival, again providing evidence against the use of this selection tool in LT for HCC24,25. Interestingly, patients undergoing successful downstaging from outside to inside Milan did not seem to experience a survival benefit compared to those transplanted outside Milan or those who presented outside Milan, failed downstaging but still underwent transplantation. Also of note, the increased risk-per-LT in the earlier years did not translate to program-wide differences in recurrence-free or overall survival in different eras of our program, demonstrating how the preponderance of lower risk cases can help adjust for overall programmatic risk of some high-risk cases.
Similarly, patients inside of Milan criteria did not seem to benefit from neoadjuvant therapy with TACE RFA or Y-90. This seemingly differs from studies such as those by DiNorcia and Agopian and colleagues, which indicated that pathologic complete response (pCR) to LRT was associated with improved outcomes for patients within Milan criteria26,27. However, our cohort has significant differences from those described in the DiNorcia and Agopian Studies. Specifically, patients in our study receiving any LRT had larger tumours and higher MELD scores compared with the un-treated cohort, while the favourable cohort in the DiNorcia study had smaller tumours and lower MELD scores26,27. No definite conclusions can be drawn from this finding, though it may reinforce that sicker patients have higher recurrence rates regardless of LRT, further emphasizing the importance of mitigating other risk factors of organ transplant where possible.
Our study upholds the prior works citing the utility of the Metroticket 2.0 system at predicting recurrence, with a previously cited overall accuracy of 0.721 and an accuracy of 0.733 in our study7. The utility of Metroticket 2.0 was further upheld in the expanded-criteria cohort, which supports the findings of the 2022 study by Barreto et al. 28. It is notable that the five-year OS was nearly equivalent in the Metroticket-poor group regardless of whether the analysis was overall or within Milan, perhaps indicating the Metroticket could be considered instead of or in concert with the more conventional criteria. In addition to OS, our study demonstrated the utility of Metroticket at predicting RFS, specifically identifying an optimized predicted risk of 0.875 for prediction of recurrence. Metroticket was historically developed to predict survival, with subsequent analyses confirming its ability to predict HCC-specific mortality when combined with mRECIST10 However, the robustness of this score at predicting overall recurrence and/or recurrence-free survival is less well established. Pinero et al. 29 previously reported a combined approach for predicting recurrence using Metroticket and AFP cut-offs. However, to our knowledge, this represents the first study establishing a single-score clinically simple cut-off that is highly predictive of recurrence and survival. Thus, the 0.875 cut-off may help supplement the standard criteria in prediction of recurrence and survival, and surgeons may consider caution in offering LT to those with predicted survival less than 87.5% as these patients have much higher recurrence and lower survival rates.
Unlike previous large studies on the topic, this work focused heavily on donor-specific factors that may influence recurrence and survival after LT for HCC. Our study found that donor age is associated with recurrence. This diverges from the study of Cusumano and colleagues but supports Orci and colleagues’ finding that donor age older than 60 years is a risk factor for recurrence30,31. Similarly, the impact of donor warm ischaemia time as seen with the use of DCD liver grafts on HCC recurrence is controversially discussed in the literature. Prior works, by Croome and Silverstein and colleagues have demonstrated that DCD grafts are not associated with increased recurrence or mortality in most cases, though Silverstein did demonstrate increased risk in very high-risk donors32,33. Croome and colleagues have for example demonstrated reduced survivals with DCD livers in the UNOS database (2013) then subsequently found no impact of DCD grafts based on a multicenter study in 201533,34. This does, perhaps, highlight that the risk profile of DCDs is not binary, but rather a spectrum of risk based on various graft-specific variables. Interestingly, some previous works had even argued that HCC-recipients should preferentially receive DCD grafts due to the non-urgent nature of the operation and the low laboratory MELD scores35. To our knowledge, this study represents the first large study to examine DCD-graft use in expanded-criteria recipients, finding a higher rate of recurrence and a poorer survival with DCD grafts in this population. While no definitive mechanistic evaluation can be made, it is possible that higher-risk patients may disproportionately experience the deleterious effects of a riskier graft. We further found reduced OS with DCD grafts in the population at-large and after removing early LT-associated mortality, again representing a slight divergence from prior literature. DCD grafts were also associated with reductions in 3-year outcomes in those undergoing downstaging, who represent a different higher-risk cohort. Overall, this would not support the works of Vining and colleagues, instead indicating that perhaps DCD grafts could be avoided in this population when possible or when the HCC recipient risk crosses certain thresholds. This is particularly true given the lack of association between time on waitlist and recurrence, perhaps suggesting it would be prudent to wait for an ideal DBD-graft in HCC patients, particularly those at high risk.
In similar vein to the discussion of higher-risk donor features, we also found that a total ischaemia time, defined as the sum of donor WIT, cold storage time, and recipient WIT, greater than 6 hours was associated with a higher recurrence rate. Nagai et al. 36 had previously shown that longer ischaemic times are associated with early recurrence and, though CIT and WIT were not independently predictive in our cohort, our work does support the correlation of increasing ischaemia time and recurrence. TIT represents a modifiable risk factor; wherein improved logistics could reduce the risk of recurrence. Interestingly, our study did not find an increased risk of recurrence with increased WIT. This is a distinct contrast from prior studies demonstrating poorer outcomes in all-comer and HCC-specific LT with increased WIT and CIT37,38. While the exact correlation between ischaemia time and risk profile in this context is not established, various methods of increased ischaemic time do seem to introduce potential for increased acute and ongoing inflammation and thus poorer outcomes. A leading proposed mechanism for this is the pro-inflammatory microenvironment associated with ischaemic reperfusion injury (IRI) that comes with longer ischaemic time, and it is proposed that pharmacological, logistical, or even surgical techniques to reduce IRI reduce the risk of recurrence39. This same inflammatory environment is present in post-transplant regeneration of split-donor or living-donor grafts, and thus it is not surprising that the higher rate of regeneration in small-for-size grafts has been shown to increase the risk of HCC recurrence40. This risk was most pronounced in LT outside Milan, a finding that could be extrapolated to the higher risk associated with DCD grafts outside Milan in our study.
Perhaps most pertinent to today’s discussion is the impact of ex-situ machine perfusion, which has known powerful interactions with the peri- and post-transplant immune cascade41. Multiple strategies for reducing IRI with perfusion have been described, with end-ischaemic perfusion representing the most common strategy currently employed42. During hypothermic oxygenated perfusion (HOPE) highly oxygenated perfusate is distributed to re-activate aerobic mitochondrial metabolism and metabolize accumulated succinate and NADH during ischaemia to prevent later reactive oxygen species (ROS) released at normothermic reperfusion on a device or at implantation43. One recent study by Rigo et al. 44 demonstrated the protection of candidates transplanted for HCC by hypothermic oxygenated perfusion (HOPE), specifically allowing for the use of extended-criteria donor (ECD) grafts without increasing the risk of HCC recurrence. Further, Mueller et al. 45 showed a reduced risk of HCC recurrence with HOPE, again highlighting potential benefits of ex-situ perfusion in HCC cases. While the findings of this study cannot offer definitive evidence regarding the future of ex-situ perfusion in preventing HCC recurrence, it is reasonable to think that HOPE may help mitigate donor- or graft-specific risk factors and thus reduce recurrence in the higher-risk grafts. Based on such early clinical studies and described underlying mechanisms, randomized controlled clinical trials are currently starting on the impact of HOPE on HCC recurrence after transplantation in Europe. Future works should also focus on the impact of perfusion in higher-risk grafts in LT for HCC, specifically focusing on patients with higher identified risk using the Metroticket 2.0 or other comparable scales, and at least on trial on this topic is currently in development. Finally, HCC has a pioneer role in transplantation oncology, and allows us a continuing opportunity with a high volume of patients to examine the role of both tumour, recipient and donor-specific risk factors that may impact donor-recipient matching and transplant candidacy. These findings may also be applicable when considering expanding transplant to other primary or metastatic cancers.
Limitations to the study include the retrospective nature and thus only correlation could be established. The study population underwent transplantation from 2000 to 2020, a time during which changes in operative technique and perioperative management may create a non-homogenous population. The increasing use of ex-situ perfusion may limit the future applicability of analyses regarding the role of ischaemic times. The analysis of preoperative decision making is limited by potentially confounding variables, such that reasons why certain treatments were selected may bias their likelihood of success. Interpretation of outcomes with and without downstaging was limited to people receiving transplant, and it is likely that many who failed downstaging were never transplanted, which would be an important comparator if available. Limits in prior documentation prevented thorough evaluation of IRI or early allograft dysfunction (EAD) in examination of high-risk transplants. These limits also prevented thorough analysis of some variables that were not collected in the database and were not available in the EMR for some cases conducted prior to institutional implementation of the EMR in 2007. We do acknowledge that the rather small sample size of downstaged patients limits the power of the comparison of outcomes in these cohorts. We do not have an external cohort with which to validate the Metroticket 0.875 model, but we recommend external validation prior to clinical application.
Conclusion
This large, two-centre study supports the ongoing use of established selection criteria such as Milan or UCSF criteria for liver transplantation of candidates with hepatocellular carcinoma. A predicted mortality of less than 87.5% according to the Metroticket 2.0 scale is predictive of recurrence and survival in both, overall and in the high-risk population. Donor-specific risk factors appear most deleterious in patients already at high-risk from their HCC standpoint. Future work should focus on the ability of machine perfusion to mitigate risk in high-risk patients.
Ethical approval
Institutional Review Board (IRB) approval was obtained for this study, number 17-009.
The ethical approval board is the IRB. This study is published on clinicaltrials.gov NCT06018857.
Hyperlink: https://clinicaltrials.gov/ct2/show/NCT06018857.
Consent
Not applicable.
Source of funding
No funding was obtained for this research.
Author contribution
C.J.W., A.S. designed the study from inception and conceptualized approaches. C.J.W., R.R., M.M., S.S. and A.S. led data collection including maintaining and subsequently completing the institutional databases. C.J.W., A.S. were primarily responsible analyzing the data and generating the figures and tables. C.J.W., A.P., F.A., K.H., A.S. were responsible for primary editing of the manuscript. Revision and critical discussion: all coauthors.
Conflicts of interest disclosure
There are no conflicts of interest.
Research registration unique identifying number (UIN)
This study is registered on clinicaltrials.gov NCT06018857.
Hyperllink https://clinicaltrials.gov/ct2/show/NCT06018857.
Guarantor
Chase Wehrle, Andrea Schlegel.
Data availability statement
Data are available upon reasonable request.
Provenance and peer review
This was not invited, although this work has been accepted for presentation at the ISLS 2023 meeting.
Provenance and peer review
Not commissioned, externally peer-reviewed.
Data statement
The authors certify that all data are accurate to the best of their knowledge. Full de-identified data will be made available on reasonable request and after completion of proper regulatory and legal requirements. All data collection was approved by our regulatory oversight board and was conducted according to best practices for research.
Apologies if this is not the proper data statement. I was not able to find the full data statement guidelines, but we are happy to comply with them!
Footnotes
Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.
Published online 22 January 2024
Contributor Information
Chase J. Wehrle, Email: wehrlec@ccf.org.
Roma Raj, Email: roma.raj@hotmail.com.
Marianna Maspero, Email: mariannamaspero@gmail.com.
Sangeeta Satish, Email: satishs2@ccf.org.
Bijan Eghtesad, Email: eghtesb@ccf.org.
Alejandro Pita, Email: pitaa@ccf.org.
Jaekeun Kim, Email: kimj30@ccf.org.
Mazhar Khalil, Email: khalilmazher@gmail.com.
Esteban Calderon, Email: caldere@ccf.org.
Danny Orabi, Email: orabid@ccf.org.
Bobby Zervos, Email: zervosx@ccf.org.
Jamak Modaresi Esfeh, Email: modarej@ccf.org.
Maureen Whitsett Linganna, Email: linganm@ccf.org.
Teresa Diago-Uso, Email: diagout@clevelandclinicabudhabi.ae.
Masato Fujiki, Email: fujikim@ccf.org.
Cristiano Quintini, Email: quintic2@clevelandclinicabudhabi.ae.
Choon David Kwon, Email: kwonc2@ccf.org.
Charles Miller, Email: millerc8@ccf.org.
Antonio Pinna, Email: pinnaa@ccf.org.
Federico Aucejo, Email: aucejof@ccf.org.
Koji Hashimoto, Email: hashimk@ccf.org.
Andrea Schlegel, Email: schlegela4@ccf.org.
References
- 1.Kulik L, El-Serag HB. Epidemiology and Management of Hepatocellular Carcinoma. Gastroenterology 2019;156:477–491.e1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Reig M, Forner A, Rimola J, et al. BCLC strategy for prognosis prediction and treatment recommendation: The 2022 update. J Hepatol 2022;76:681–693. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Mazzaferro V, Regalia E, Doci R, et al. Liver transplantation for the treatment of small hepatocellular carcinomas in patients with cirrhosis. N Engl J Med 1996;334:693–699. [DOI] [PubMed] [Google Scholar]
- 4.Mazzaferro V, Bhoori S, Sposito C, et al. Milan criteria in liver transplantation for hepatocellular carcinoma: an evidence-based analysis of 15 years of experience. Liver Transpl 2011;17(suppl 2):S44–S57. [DOI] [PubMed] [Google Scholar]
- 5.Yao F, et al. Liver transplantation for hepatocellular carcinoma: expansion of the tumor size limits does not adversely impact survival. Hepatology 2001;33:1394–1403. [DOI] [PubMed] [Google Scholar]
- 6.Bento de Sousa JH, Calil IL, Tustumi F, et al. Comparison between Milan and UCSF criteria for liver transplantation in patients with hepatocellular carcinoma: a systematic review and meta-analysis. Transl Gastroenterol Hepatol 2021;6:11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Mazzaferro V, Sposito C, Zhou J, et al. Metroticket 2.0 model for analysis of competing risks of death after liver transplantation for hepatocellular carcinoma. Gastroenterology 2018;154:128–139. [DOI] [PubMed] [Google Scholar]
- 8.Lei JY. “Metroticket” predictor for assessing liver transplantation to treat hepatocellular carcinoma: a single-center analysis in mainland China. World J Gastroenterol 2013;19:8093–8098. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Mathew G, Agha R, Albrecht J, et al. STROCSS 2021: Strengthening the reporting of cohort, cross-sectional and case-control studies in surgery. Int J Surg 2021;96:106165. [DOI] [PubMed] [Google Scholar]
- 10.Cucchetti A, Serenari M, Sposito C, et al. Including mRECIST in the Metroticket 2.0 criteria improves prediction of hepatocellular carcinoma-related death after liver transplant. J Hepatol 2020;73:342–348. [DOI] [PubMed] [Google Scholar]
- 11.Mazzaferro V, Llovet JM, Miceli R, et al. Predicting survival after liver transplantation in patients with hepatocellular carcinoma beyond the Milan criteria: a retrospective, exploratory analysis. Lancet Oncol 2009;10:35–43. [DOI] [PubMed] [Google Scholar]
- 12.Singal AG, Llovet JM, Yarchoan M, et al. AASLD Practice Guidance on prevention, diagnosis, and treatment of hepatocellular carcinoma. Hepatology 2023;78:1922–1965. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hong JC, Yersiz H, Kositamongkol P, et al. Liver transplantation using organ donation after cardiac death: a clinical predictive index for graft failure-free survival. Arch Surg 2011;146:1017–1023. [DOI] [PubMed] [Google Scholar]
- 14.Schlegel A, Kalisvaart M, Scalera I, et al. The UK DCD Risk Score: a new proposal to define futility in donation-after-circulatory-death liver transplantation. J Hepatol 2018;68:456–464. [DOI] [PubMed] [Google Scholar]
- 15.Dutkowski P, Schlegel A, Slankamenac K, et al. The use of fatty liver grafts in modern allocation systems: risk assessment by the balance of risk (BAR) score. Ann Surg 2012;256:861–868; discussion 868-9. [DOI] [PubMed] [Google Scholar]
- 16.Brown CS, Wakam GK, Englesbe MJ. Increased-risk donors and solid organ transplantation: current practices and opportunities for improvement. Curr Opin Organ Transplant 2020;25:139–143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Shetty K, Timmins K, Brensinger C, et al. Liver transplantation for hepatocellular carcinoma validation of present selection criteria in predicting outcome. Liver Transpl 2004;10:911–918. [DOI] [PubMed] [Google Scholar]
- 18.Bechstein WO, Guckelberger O, Kling N, et al. Recurrence-free survival after liver transplantation for small hepatocellular carcinoma. Transpl Int 1998;11(suppl 1):S189–S192. [DOI] [PubMed] [Google Scholar]
- 19.Chagas AL, Felga GEG, Diniz MA, et al. Hepatocellular carcinoma recurrence after liver transplantation in a Brazilian multicenter study: clinical profile and prognostic factors of survival. Eur J Gastroenterol Hepatol 2019;31:1148–1156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Nutu A, Justo I, Marcacuzco A, et al. Liver transplantation for hepatocellular carcinoma using grafts from uncontrolled circulatory death donation. Sci Rep 2021;11:13520. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Piros L, Fehérvári I, Görög D, et al. Examinations of factors influencing survival of liver transplantation for hepatocellular carcinoma: a single-center experience from Budapest. Transplant Proc 2015;47:2201–2206. [DOI] [PubMed] [Google Scholar]
- 22.Unek T, Karademir S, Arslan NC, et al. Comparison of Milan and UCSF criteria for liver transplantation to treat hepatocellular carcinoma. World J Gastroenterol 2011;17:4206–4212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Toniutto P, Fumolo E, Fornasiere E, et al. Liver transplantation in patients with hepatocellular carcinoma beyond the Milan criteria: a comprehensive review. J Clin Med 2021;10:3932. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Morgul MH, Felgendreff P, Kienlein A, et al. Milan criteria in the MELD era-is it justifiable to extend the limits for orthotopic liver transplantation? World J Surg Oncol 2020;18:158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Kim SJ, Kim JM. Prediction models of hepatocellular carcinoma recurrence after liver transplantation: a comprehensive review. Clin Mol Hepatol 2022;28:739–753. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Agopian VG, Morshedi MM, McWilliams J, et al. Complete pathologic response to pretransplant locoregional therapy for hepatocellular carcinoma defines cancer cure after liver transplantation: analysis of 501 consecutively treated patients. Ann Surg 2015;262:536–545; discussion 543-5. [DOI] [PubMed] [Google Scholar]
- 27.DiNorcia J, Florman SS, Haydel B, et al. Pathologic response to pretransplant locoregional therapy is predictive of patient outcome after liver transplantation for hepatocellular carcinoma: analysis from the US Multicenter HCC Transplant Consortium. Ann Surg 2020;271:616–624. [DOI] [PubMed] [Google Scholar]
- 28.Barreto SG, Strasser SI, McCaughan GW, et al. Expansion of liver transplantation criteria for hepatocellular carcinoma from Milan to UCSF in Australia and New Zealand and Justification for Metroticket 2.0. Cancers (Basel) 2022;14:2777. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Piñero F, Costentin C, Degroote H, et al. AFP score and metroticket 2.0 perform similarly and could be used in a “within-ALL” clinical decision tool. JHEP Rep 2023;5:100644. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Cusumano C, De Carlis L, Centonze L, et al. Advanced donor age does not increase risk of hepatocellular carcinoma recurrence after liver transplantation: a retrospective two-centre analysis using competing risk analysis. Transpl Int 2021;34:1948–1958. [DOI] [PubMed] [Google Scholar]
- 31.Orci LA, Silverstein J, Roll G, et al. Donor characteristics and risk of hepatocellular carcinoma recurrence after liver transplantation. Br J Surg 2015;102:1250–1257. [DOI] [PubMed] [Google Scholar]
- 32.Silverstein J, Roll G, Dodge JL, et al. Donation after circulatory death is associated with similar posttransplant survival in all but the highest-risk hepatocellular carcinoma patients. Liver Transpl 2020;26:1100–1111. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Croome KP, Lee DD, Burns JM, et al. The use of donation after cardiac death allografts does not increase recurrence of hepatocellular carcinoma. Am J Transplant 2015;15:2704–2711. [DOI] [PubMed] [Google Scholar]
- 34.Croome KP, Wall W, Chandok N, et al. Inferior survival in liver transplant recipients with hepatocellular carcinoma receiving donation after cardiac death liver allografts. Liver Transpl 2013;19:1214–1223. [DOI] [PubMed] [Google Scholar]
- 35.Vining CC, Ecker BL, Abt PL, et al. Donation after cardiac death in the hepatocellular carcinoma patient: Same indication? Liver Transpl 2017;23(S1):S27–s33. [DOI] [PubMed] [Google Scholar]
- 36.Nagai S, Yoshida A, Facciuto M, et al. Ischemia time impacts recurrence of hepatocellular carcinoma after liver transplantation. Hepatology 2015;61:895–904. [DOI] [PubMed] [Google Scholar]
- 37.Kornberg A, Witt U, Kornberg J, et al. Extended ischemia times promote risk of hcc recurrence in liver transplant patients. Dig Dis Sci 2015;60:2832–2839. [DOI] [PubMed] [Google Scholar]
- 38.Kalisvaart M, Croome KP, Hernandez-Alejandro R, et al. Donor Warm Ischemia Time in DCD Liver Transplantation-Working Group Report From the ILTS DCD, Liver Preservation, and Machine Perfusion Consensus Conference. Transplantation 2021;105:1156–1164. [DOI] [PubMed] [Google Scholar]
- 39.Parente A, Cho HD, Kim KH, et al. Association between Hepatocellular Carcinoma Recurrence and Graft Size in Living Donor Liver Transplantation: A Systematic Review. Int J Mol Sci 2023;24:6224. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Parente A, Cho HD, Kim KH, et al. Association between hepatocellular carcinoma recurrence and graft size in living donor liver transplantation: a systematic review. Int J Mol Sci 2023;24:6224. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Panconesi R, Flores Carvalho M, Dondossola D, et al. Impact of machine perfusion on the immune response after liver transplantation - a primary treatment or just a delivery tool. Front Immunol 2022;13:855263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Chen H, Lu D, Yang X, et al. One shoot, two birds: alleviating inflammation caused by ischemia/reperfusion injury to reduce the recurrence of hepatocellular carcinoma. Front Immunol 2022;13:879552. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Schlegel A, Muller X, Mueller M, et al. Hypothermic oxygenated perfusion protects from mitochondrial injury before liver transplantation. EBioMedicine 2020;60:103014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Rigo F, De Stefano N, Patrono D, et al. Impact of hypothermic oxygenated machine perfusion on hepatocellular carcinoma recurrence after liver transplantation. J Pers Med 2023;13:703. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Mueller M, Kalisvaart M, O’Rourke J, et al. Hypothermic oxygenated liver perfusion (HOPE) prevents tumor recurrence in liver transplantation from donation after circulatory death. Ann Surg 2020;272:759–765. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Data are available upon reasonable request.