AFP Model and Liver Transplantation.

Introduction Recurrence of hepatocellular carcinoma (HCC) after liver transplantation (LT) is a catastrophic event, with limited therapeutics and poor survival [1]. Prior to 1990s, transplantation for HCC was a limited option with high recurrence and poor recipient survival, largely because of a lack of accurate selection of transplant candidates and a late diagnosis.

The Milan criteria have remained as the gold standard selection criteria, with post-transplant 5-year recurrence less than 15% and 5-year survival over 70% [2]. These criteria do not contemplate a relevant histological variable and a risk marker of recurrence, microvascular invasion (mvi). Presence of mvi in the explanted liver has been identified as an independent variable related to recurrence that categorizes patients with low and high risk of HCC recurrence [3]. That is why other selection criteria, beyond Milan criteria, have tried to find pre-transplant surrogate markers of mvi or other related risk factors for recurrence [4-6].

Serum alpha-fetoprotein (AFP) is a marker of renewed interest as a selection criterion, which has been identified as an independent risk variable associated with recurrence and as a surrogate marker of mvi [4,5,7,8]. Evaluation of mvi before LT by a tumor biopsy has many caveats; on the one hand, the risk of tumor seeding, bleeding and other complications and on the other hand, the absence of it does not exclude of its presence in the explanted liver analysis.

Consequently, while it is known that the presence of mvi defines a greater risk, its absence generates uncertainty regarding the risk of recurrence in this population. The Up-to 7 criteria showed that the occurrence of mvi at any size and tumor number was paralleled by a significant lower patient survival and higher recurrence [3]. However, the Metroticket calculator excluded important pre transplant variables such as serum AFP and tumor treatment or progression while on the waitlist [9]. A recently predicting model, the RETREAT score, showed that when considering pre and post transplant variables, the AFP at different cut-offs, presence of mvi and the sum of the largest tumor diameter and tumor number were associated with HCC recurrence [7]. While it has been published earlier in Italian and French cohorts that the prevalence of mvi is 24% [4,11], in another Latin American validation cohort of the AFP model the prevalence of mvi was 22.2% [10].

Our objective therefore, is to evaluate the AFP model against Milan and San Francisco criteria (UCSF) in the prediction of HCC recurrence and patient survival specifically in patients without mvi in the explanted liver and evaluate if the AFP model can assure and identify good candidates for LT that are beyond Milan and UCSF in this subgroup of patients.

Patients and Methods Study design, setting and participating centers This study is going to include three multicenter cohorts of consecutive adult patients (>17 years of age) who underwent a first LT between January 1 2005 and December 31 2016 [4,10,11]. Participating centers will appoint a study coordinator responsible for data collection. In cases of conflicting or missing data, central revision and resubmission were requested. All the requested variables are going to be included in a written CRF.

Eligibility criteria and study variables Criteria for inclusion required patients to be adult cirrhotic or non-cirrhotic recipients with confirmed HCC in the explanted liver. Patients are going to be excluded if: 1) other tumors than HCC are confirmed in the explanted liver, or 2) there are missing relevant information; 3) extrahepatic or macrovascular tumor invasion were observed during pre transplant evaluation or in the explanted liver 4) incidental HCC, 5) had a previous liver transplant and 6) patients with microvascular invasion in the explanted liver (this latter criteria is going to be assessed for the final inclusion criteria).

Recipient characteristics, pre-transplant tumor characteristics and serum α-fetoprotein (AFP) levels are going to be recorded at listing and at last pre-LT evaluation. Subjects with HCC diagnosis prior to transplant based on imaging criteria are going to be classified according to Milan (MC) and University of California San Francisco (UCSF) [12] and the AFP model [4], depending on size and number of lesions detected on pre-LT computerized tomography (CT) or magnetic resonance images (MRI) and on serum pre-LT AFP levels ng/ml including the following cut-offs values ≤100 ng/ml, 101-1000 ng/ml, and >1000 ng/ml [4].

Tumor treatment before transplantation is going to be recorded (date and type of treatments performed) including trans-arterial chemoembolization (TACE), radiofrequency ablation (RFA), percutaneous ethanol injection (PEI) and liver resection (LR).

Explanted liver findings will include macroscopic and microscopic evaluation of each nodule, number and diameter (cm) of each, presence of mvi, and degree of tumor differentiation according to Edmonson-Steiner grading system. Nodules of largest diameter are going to be identified as the major nodule. Necrotic nodules are going to be also measured including necrotic and viable tumor diameter. Finally, Milan and Up-to seven criteria are going to also be applied to the explanted liver specimen.

Study end-points Primary end-points analyzed are going to be 5-year patient survival and HCC recurrence. Recurrence is going to be determined on the basis of imaging criteria plus serum AFP or by biopsy. Time to recurrence (TTR) is going to be calculated considering it a robust clinical outcome measure and calculated as the time in months elapsed between transplantation and diagnosis of recurrence.

Secondary aims include the development of tumor progression while on the waitlist according to RECIST 1.1 criteria.

Statistical Analysis Categorical data were compared using Fisher's exact test or Chi-Square (Χ2) test (2-tailed). Continuous variables were compared with Student's T test or Wilcoxon rank-sum test according to their distribution, respectively. First of all, in order to evaluate if microvascular invasion in the explant is an independent risk variable for recurrence, a multivariate Cox regression analysis, with hazard ratios (HR) and 95% confidence intervals (95% CI) for identifying explanted liver risk variables for 5-year recurrence is going to be carried out evaluating potential confounding variables. After calculation of each adjusted HR separately in each overall cohort (French, Italian and LATAM), cumulative incidence of recurrence by a competing risk regression analysis with death is going to be compared among patients with or without mvi ("high" and "low" risk patients, respectively).

Second, after evaluating the effect of mvi as a selection criteria of "low" and "high" risk patients, the following analysis is going to be done after excluding patients with mvi in each cohort in order to evaluate the performance of the AFP model in these "low-risk" patients.

A multivariate Cox regression analysis (HR, 95% CI) is going to be done in order to identify the effect of each pre-LT model on prediction of HCC 5-year recurrence adjusted with potential confounding variables. Variables with a P value <0.05 after the univariate analysis are going to be included in the multivariate model, generated by stepwise forward elimination evaluating P values (Wald test) and considering adjusted HR with confounding variables (>20% of change in crude HR). For each multivariate model, 1 variable per 10 events were included. Adjustment of each final model is going to be evaluated with proportional hazards through graphic and statistical evaluation (Schoenfeld residual test). Calibration is going to be assessed by comparison of observed and predicted curves and evaluation of the goodness of fit of the model by Harrell's c-statistic index.

Finally, a competing risk analysis with death and recurrence is going to be done with calculation of subharzard ratios (SHR) and 95% CI. Kaplan Meier survival curves were compared using the log-rank test (Mantel-Cox). Collected data were analyzed with STATA 10.0.

Agenda.

1. Collection and analysis of data: until 1-January-2018.
2. Full statistical analysis and overall results: until 1-January to February-2019.
3. Submission of overall results to all co-authors included in the 3 cohorts: March 2019.
4. Manuscript, figures and tables done: March 2019.