Development and validation of the HALT-HCC score to predict mortality in liver transplant recipients with hepatocellular carcinoma: a retrospective cohort analysis.

BACKGROUND

Tumour morphological criteria for determining the appropriateness of liver transplantation in patients with hepatocellular carcinoma poorly estimate post-transplantation mortality. The aim of this study was to develop and assess the utility of a continuous risk score in predicting overall survival following liver transplantation for hepatocellular carcinoma.

METHODS

We did a retrospective cohort analysis to develop a continuous multivariable risk score for assessment of overall survival following liver transplantation for hepatocellular carcinoma. We used data from 420 patients with hepatocellular carcinoma who underwent liver transplantation between Jan 1, 2002, and Oct 31, 2014, at the Cleveland Clinic Foundation (CCF), Cleveland, OH, USA. The model we developed (Hazard Associated with Liver Transplantation for Hepatocellular Carcinoma; HALT-HCC) assessed the association of the following previously reported variables of interest with overall survival by use of multivariate Cox regression: MELD-sodium (MELD-Na), tumour burden score (TBS), alpha-fetoprotein (AFP), year of transplantation, underlying cause of cirrhosis, neutrophil-lymphocyte ratio, history of locoregional therapy, and Milan criteria status. Once the risk equation was generated, validation and calibration of risk assessment was done with nationwide data for the same time period from the Scientific Registry of Transplant Recipients (SRTR; n=13 717).

FINDINGS

The risk equation was generated as (1·27 × TBS) + (1·85 × lnAFP) + (0·26 × MELD-Na) and the HALT-HCC score ranged from 2·40 to 46·42 in the CCF cohort. In the validation cohort, prognosis worsened with increasing HALT-HCC score (5-year overall survival of 78·7% [95% CI 76·9-80·4] for quartile 1, 74·5% [72·6-76·2] for quartile 2, 71·8% [70·1-73·5] for quartile 3, and 61·5% [59·6-63·3] for quartile 4; p<0·0001). Multivariate Cox modelling showed that HALT-HCC was significantly associated with overall survival (hazard ratio [HR] 1·06 per point, 95% CI 1·05-1·07), even after adjustment for risk factors not related to hepatocellular carcinoma. Assessment of discrimination revealed a C-index of 0·613 (95% CI 0·602-0·623). Calibration coefficients for linear regressions of observed versus predicted mortality were 1·001 (95% CI 0·998-1·007) at 1 year and 0·982 (0·980-0·987) at 2 years after transplantation. Patients within and outside the Milan criteria showed similar risk of death when stratified by HALT-HCC score. Among the 12 754 patients who met the Milan criteria, 2714 were shown to have poor prognosis after transplantation after stratification by HALT-HCC score with a cutoff of 17; conversely, among the 963 patients who did not meet the Milan criteria, 287 had demonstrably good prognosis.

INTERPRETATION

The HALT-HCC score might enable clinicians to accurately assess post-transplantation survival in patients with hepatocellular carcinoma by use of individualised, preoperatively assessed characteristics. However, further studies are needed before adoption.

FUNDING

None.