When to call it off: defining transplant candidacy limits in liver donor liver transplantation for hepatocellular carcinoma

Living donor liver transplantation (LDLT) is an acceptable treatment option for hepatocellular carcinoma (HCC). Traditional transplant criteria aim at best utilization of donor organs with low risk of post transplant recurrence. In LDLT, long term recurrence free survival (RFS) of 50% is considered acceptable. The objective of the current study was to determine preoperative factors associated with high recurrence rates in LDLT. Between April 2012 and December 2019, 898 LDLTs were performed at our center. Out of these, 242 were confirmed to have HCC on explant histopathology. We looked at preoperative factors associated with ≤ 50%RFS at 4 years. For survival analysis, Kaplan Meier curves were used and Cox regression analysis was used to identify independent predictors of recurrence. Median AFP was 14.4(0.7–11,326.7) ng/ml. Median tumor size was 2.8(range = 0.1–11) cm and tumor number was 2(range = 1–15). On multivariate analysis, AFP > 600 ng/ml [HR:6, CI: 1.9–18.4, P = 0.002] and microvascular invasion (MVI) [HR:5.8, CI: 2.5–13.4, P <  0.001] were independent predictors of 4 year RFS ≤ 50%. When AFP was > 600 ng/ml, MVI was seen in 88.9% tumors with poor grade and 75% of tumors outside University of California San Francisco criteria. Estimated 4 year RFS was 78% for the entire cohort. When AFP was < 600 ng/ml, 4 year RFS for well-moderate and poor grade tumors was 88 and 73%. With AFP > 600 ng/ml, RFS was 53% and 0 with well-moderate and poor grade tumors respectively (P <  0.001). Patients with AFP < 600 ng/ml have acceptable outcomes after LDLT. In patients with AFP > 600 ng/ml, a preoperative biopsy to rule out poor differentiation should be considered for patient selection.


Background
Liver transplantation is an established treatment modality in patients with hepatocellular carcinoma (HCC) and cirrhosis [1,2]. Milan criteria and University of California San Francisco (UCSF) criteria are the most widely accepted transplant criteria for patient selection [3,4]. These criteria have been primarily derived in the deceased donor liver transplant (DDLT) setting and aim at best utilization of donor organs. In living donor liver transplantation (LDLT), issues such as prolonged waiting times and competition for donor organs are not encountered. Thus, Milan and UCSF criteria appear restrictive. As a result, most LDLT centers in Asia perform LDLT for HCC outside these criteria [5].
The minimum acceptable recurrence free survival(RFS) in the setting of LDLT remains debatable. Considering the operative risk to a living donor in LDLT, long term RFS of 50% is considered the benchmark [6,7]. Other than tumor size and number; certain factors like AFP level, microvascular invasion (MVI) and poor grade have a significant impact on post transplant recurrence [8,9]. With a keen desire to increase patient pool eligible for transplantation, there is a need to identify patients who should be denied LDLT due to unacceptable recurrence risk.
We are a high volume LDLT center, and perform a significant number of transplants for HCC; some of whom are outside traditional transplant criteria. This provides a unique opportunity to assess outcomes in advanced tumors managed with LDLT.
In the setting of LDLT, long term RFS < 50% should be considered unacceptable and transplant in such patients should be considered futile. The objective of the current study was to identify preoperative factors, in the presence of which, LDLT can be potentially declined.

Methods
Between April 2012 and December 2019, a total of 898 living donor liver transplants were performed at our center. For this study we included adult patients with a diagnosis of HCC on explant histopathology who underwent LDLT between April 2012 and June 2019 (n = 242).
Details of diagnostic workup and patient selection for transplantation have been reported elsewhere [10,11]. The diagnosis of HCC was made on a liver dynamic CT scan. A dynamic MRI scan of the liver was performed if CT scan findings were not conclusive. Biopsy was reserved for cases where diagnosis could not be established on dynamic imaging. All patients were discussed in multi disciplinary team meeting. Extra-hepatic metastasis and main portal vein tumor thrombosis were considered absolute contraindications for transplantation. Preoperative treatments including trans-arterial chemoembolization (TACE), radio-frequency ablation (RFA) and percutaneous ethanol injection (PEI) were discussed with all patients. With accumulating experience, patients with tumor size > 10 cm, segmental or lobar portal vein tumor thrombus, or AFP > 1000 ng/ml were considered for down-staging, if feasible. In case there was an anticipated delay, these treatments were used as a bridge for transplantation.
We looked at patient demographics, AFP levels, etiology of liver failure, Model for end stage liver disease(-MELD), and Child Turcot Pugh (CTP) scores, tumor variables, and preoperative treatment. Prognostic variables including Milan and UCSF criteria for transplantation, tumor grade, microvascular invasion, AFP, MELD and CTP score, graft to recipient weight ratio (GRWR), and preoperative treatment were assessed to determine impact on RFS. We used different AFP cutoff values for prognostication based on previous reports [11][12][13]. We classified 4 year RFS > 50% as acceptable while ≤50% as unacceptable RFS based on previous recommendations [6,7].
Primary objective of the study was to identify preoperative factors associated with unacceptable recurrence rates in the setting of LDLT. The RFS was calculated by subtracting date of recurrence from date of transplantation. Univariate analysis was performed using Kaplan Meier survival curves and Log rank test was used to determine significance. Variables that were associated with RFS < 50% at 4 years were included in the multivariate Cox proportional hazard model. To determine AFP cutoff for recurrence, we used receiver operator curves (ROC). To increase transplant eligible patients, we aimed for high specificity on ROC curves as previously shown [14]. Prognostic groups were developed based on preoperative prognostic factors associated with < 50% RFS at 4 years. Chi square and Fischer test were used for categorical variables. A P value < 0.05 was considered statistically significant. All analysis was performed on Statistical package for the social sciences (SPSS version 20). The study was approved by the hospital ethics committee.

Patient characteristics
Median age was 53

Prognostic factors associated with inferior 4 year RFS
Estimated 4 year RFS was 78% for the entire study cohort. Table 2 demonstrates 4 year RFS for various prognostic factors. An AFP cutoff of 40 ng/ml had sensitivity of 64% and specificity of 74% for recurrence. While AFP cutoff of 600 ng/ml had a sensitivity of 35% and specificity of 97% for recurrence (AUC = 0.69, P < 0.0001) (Fig. 1). AFP level   A multivariate analysis including prognostic factors associated with 4 year RFS < 50% was performed as shown in Table 3. An AFP > 600 ng/ml and MVI were independent predictors of RFS. Risk of recurrence was significantly increased with AFP > 600 ng/ml [HR:6, CI: 1.9-18.4, P = 0.002] and MVI [HR:5.8, CI: 2.5-13.4, P < 0.001].The 4 year RFS in patients with combined AFP > 600 ng/ml and MVI was 0 versus 83% (P < 0.001) as shown in Fig. 2.

Preoperative factors associated with MVI
Since MVI cannot be accurately determined preoperatively, we looked for factors associated with high risk (> 50%) of MVI. When combined with AFP > 600 ng/ml, tumors outside UCSF criteria and poor grade were significant factors associated with high risk of MVI as shown in Table 4.
Since AFP and MVI were the only independent predictors of < 50% RFS on multivariate analysis, we developed prognostic groups associated with high risk of MVI based on AFP > 600 ng/ml. The highest risk of MVI was seen in patients with AFP > 600 ng/ml and poor grade (88.9%) and AFP > 600 ng/ml/UCSF out tumors (75%) as shown in Table 5. Table 6 demonstrates actual recurrence rates in various prognostic groups. All patients with AFP < 600 ng/ml within and outside UCSF criteria, irrespective of tumor grade, had an acceptable 4 year RFS (> 50%) as shown in   year RFS in patients with AFP > 600 ng/ml irrespective of whether tumors were within or outside UCSF criteria was < 50% (Fig. 3a). In patients with AFP > 600 ng/ml, 4 year RFS was 53% for well-moderately differentiated tumors while it was 0 and not reached with poorly differentiated tumors (Fig. 3b). When patients with AFP > 600 ng/ml and poor differentiation were excluded, the 4 year RFS of our patient cohort was 82%.

Discussion
The current study reports outcomes of a significantly large cohort of patients who underwent LDLT for HCC [9,14]. A small percentage of patients (< 10%)had AFP > 600 ng/ml. This group was further assessed for poor differentiation. Overall, < 5% patients had AFP > 600 ng/ ml with poor differentiation. We believe this is the group of patients with very high risk of recurrence and needs to be identified on preoperative biopsy. We identified AFP and MVI as important prognostic variables for recurrence post transplantation. The prognostic impact of AFP and MVI is well known [8,[14][15][16][17][18][19]. The challenge remains in preoperatively identifying patients likely to have MVI. Various imaging modalities including computed tomography (CT), magnetic resonance imaging(MRI), positron emission tomography (PET) scan and tumor marker cutoffs have been assessed but remain inconsistent in detection of MVI [18][19][20][21].Preoperative biopsy is also not accurate in detecting MVI but carries sinister risk of tumor seeding [18]. This limits the widespread application of preoperative biopsy in transplant candidates and a cautious approach is warranted. Based on results of the current study, only 22/242(9%) patients would mandate a preoperative biopsy. Poor grade in this group would be a surrogate marker of MVI as 88.9% patients had MVI when AFP > 600 ng/ml and poor grade were present together.
Role of preoperative biopsy in the diagnosis of HCC, in particular before transplantation remains less clear. The obvious advantage includes pertinent diagnostic and  prognostic information [22]. It has been shown that the diagnostic sensitivity and specificity of needle biopsy is 94 and 100% respectively [23]. The drawbacks include an invasive procedure, with risks of bleeding and tumor seeding, and inaccurate information. In recent times, with technical improvements, the risk of tumor seeding (0.001%) and major complications appears (0.004%) to have tremendously reduced and diagnostic errors are rare [24]. Moreover, better knowledge of molecular and immuno histochemical properties of HCC has led to renewed interest in the role of biopsy in patients with HCC [22,25]. Poor grade has been used by the Toronto and Hangzhou group to select patients unlikely to benefit from transplantation [8,17]. With Toronto criteria, 108/242(44.6%) patients in the current study would require preoperative biopsy. With AFP > 600 ng/ml as the only indication, irrespective of tumor size and number, we have potentially limited preoperative biopsy to very few patients, increased the number of patient pool eligible for transplantation, and identified a subgroup which despite AFP > 600 ng/ml can be transplanted with acceptable risk of recurrence. In this group, LDLT should only be offered if well to moderate grade is confirmed preoperatively. A strict tumor size or number cutoff to select patients for preoperative biopsy can miss out on patients with high AFP and poor grade leading to unacceptable recurrence risk. We suggest that if poor grade is present in patients with AFP > 600 ng/ml, LDLT should be potentially declined.
In recent years, increasingly complex criteria incorporating tumor size and number, yet relying heavily on biological factors have been proposed [26][27][28][29]. These are primarily based on DDLT experience and attempt to increase eligible transplant pool without compromising outcomes when compared with Milan criteria. However, it has been suggested that post transplant outcomes in patients who fulfill these criteria are not comparable to Milan criteria [28]. In the current study, we have attempted to identify eligible candidates with acceptable   [26,[30][31][32][33][34]. To improve identification of high risk HCC patients for LDLT, some centers have used des-gamma-carboxy prothrombin (DCP) and PET scan. Imaging modalities such as PET scans and tumors markers other than AFP still need validation in terms of their clinical applicability. It is important that patients with HCC in whom LDLT is essentially futile are identified using simpler models with easily applicable tools that have been previously validated to impact prognosis.

Conclusion
The current study uses well established preoperative variables in a large cohort of HCC patients who underwent LDLT, to identify patients at high risk of post transplant recurrence. Judicious use of preoperative biopsy in patients with AFP > 600 ng/ml can identify patients not suitable for transplantation. We believe, it is more relevant to LDLT setting, where liberal cutoffs are used on tumor dimensions, waiting time is short and competition for donor organs is absent. These results need to be validated in similar settings with longer follow up to determine applicability of current findings.