A Novel Prognostic Nomogram Accurately Predicts Hepatocellular Carcinoma Recurrence after Liver Transplantation: Analysis of 865 Consecutive Liver Transplant Recipients Vatche G Agopian, MD, Michael Harlander-Locke, MPH, Ali Zarrinpar, MD, PhD, Fady M Kaldas, MD, FACS, Douglas G Farmer, MD, FACS, Hasan Yersiz, MD, Richard S Finn, Myron Tong, MD, Jonathan R Hiatt, MD, FACS, Ronald W Busuttil, MD, PhD, FACS

MD,

Although radiologic size criteria (Milan/University of California, San Francisco [UCSF]) have led to improved outcomes after liver transplantation (LT) for hepatocellular carcinoma (HCC), recurrence remains a significant challenge. We analyzed our 30-year experience with LT for HCC to identify predictors of recurrence. STUDY DESIGN: A novel clinicopathologic risk score and prognostic nomogram predicting post-transplant HCC recurrence was developed from a multivariate competing-risk Cox regression analysis of 865 LT recipients with HCC between 1984 and 2013. RESULTS: Overall patient and recurrence-free survivals were 83%, 68%, 60% and 79%, 63%, and 56% at 1-, 3-, and 5-years, respectively. Hepatocellular carcinoma recurred in 117 recipients, with a median time to recurrence of 15 months, involving the lungs (59%), abdomen/pelvis (38%), liver (35%), bone (28%), pleura/mediastinum (12%), and brain (5%). Multivariate predictors of recurrence included tumor grade/differentiation (G4/poor diff hazard ratio [HR] 8.86; G2-3/mod-poor diff HR 2.56), macrovascular (HR 7.82) and microvascular (HR 2.42) invasion, nondownstaged tumors outside Milan criteria (HR 3.02), nonincidental tumors with radiographic maximum diameter 5 cm (HR 2.71) and 10 (HR 0.64, p ¼ 0.031), and significantly more likely to recur in patients with greater pretransplantation total cholesterol (HR 1.003 per mg/dL, p ¼ 0.033), AFP (HR 1.91 per log unit, p < 0.001), maximum AFP (HR 1.82 per log unit, p < 0.001), and NLR (HR 2.14 per log unit, p ¼ 0.002). No other recipient, laboratory, or donor or operative variables conferred an increased risk of HCC recurrence. Radiographic predictors of recurrence included number of lesions (1lesion, HR 2.67, p ¼ 0.004; 2, HR 4.67, p < 0.001; 3 or more, HR 7.73, p < 0.001), maximum (HR 13.6 per log unit, p < 0.001) and cumulative tumor diameter (HR 15.0 per log unit, p < 0.001), nondownstaged tumors outside MC (HR 10.2, p < 0.001), bilobar lesions (HR 2.88, p < 0.001) and the need for 3 or more tumor treatments (HR 2.66, p < 0.001). Pathologic predictors of recurrence included number of lesions (2 lesions, HR 2.04, p ¼ 0.008; 3 or more, HR 3.63, p < 0.001), maximum (HR 34.4 per log unit, p < 0.001) and cumulative (HR 62.6 per log unit,

Vol. 220, No. 4, April 2015

Agopian et al

Liver Transplantation for Hepatocellular Carcinoma

421

p < 0.001) tumor diameter, microvascular (HR 2.98, p < 0.001) and macrovascular (HR 17.3, p < 0.001) invasion, nuclear grade 3/poor differentiation (HR 5.28, p < 0.001) and nuclear grade 4/poor differentiation (HR 13.2, p < 0.001), multifocality (HR 3.10, p < 0.001), and AJCC T stage (T2, HR 2.88, p ¼ 0.001; T3a, HR 9.59, p < 0.001; T3b, HR 31.7, p < 0.001; T4, HR 91.8, p < 0.001). Multivariate analysis and risk models Multivariate predictors of recurrence (Fig. 5) include nuclear grade 4/poor differentiation (HR 8.86, 95% CI 3.37 to 23.3, p < 0.001), macrovascular invasion (HR 7.82, 95% CI 4.69 to 13.1, p < 0.001), nondownstaged tumors beyond MC (HR 3.02, 95% CI 1.66 to 5.49, p < 0.002), nonincidental tumors with maximum radiologic diameter 5 cm (HR 2.71, 95% CI 1,31 to 5.62, p ¼ 0.007), nuclear grade 2e3/moderate to poor differentiation (HR 2.56, 95% CI 1.26 to 5.17, p < 0.001), microvascular invasion (HR 2.42, 95% CI 1.49 to 3.94, p < 0.001), nonincidental tumors with maximum radiologic diameter < 5 cm (HR 1.55, 95% CI 0.87 to 2.75, p ¼ 0.136), NLR (HR 1.77 per log unit, 95% CI 1.08 to 2.89, p ¼ 0.023), pretransplantation maximum AFP (HR 1.21 per log unit, 95% CI 1.01 to 1.45, p ¼ 0.037), and total cholesterol (HR 1.003 per mg/dL increase, 95% CI 0.999 to 1.006, p ¼ 0.161). Two risk prediction models based on these multivariate models were developed. A pretransplantation model included only ascertainable pretransplantation variables and is shown in Table 6. Significant predictors of HCC recurrence included tumors beyond MC that were not downstaged to MC (HR 4.96, 95% CI 2.92 to 8.44, p < 0.001), nonincidental tumors with maximum radiologic diameter 5 cm (HR 1.85, 95% CI 0.9 to 3.79, p ¼ 0.093) and 10 Total cholesterol, per mg/dL Log pre-LT AFP, per log unit Log pre-LT AFP max, per log unit Log NLR, per log unit Donor and operative variables Nonstandard graft* Donor age, per year Donor male Cold ischemia time >10 h Warm ischemia time >45 min

J Am Coll Surg

Liver Transplantation for Hepatocellular Carcinoma

Hazard ratio

95% CI

p Value

0.62 1.32 0.81 0.86 0.44 0.50 1.21 0.86

0.43e0.89 0.85e2.03 0.52e1.25 0.55e1.34 0.11e1.83 0.07e3.81 0.81e1.80 0.37e1.98

0.009 0.215 0.336 0.504 0.261 0.504 0.344 0.716

1.00 0.56 0.73 0.34 0.89

ref 0.34e0.90 0.36e1.48 0.08e1.41 0.34e2.35

ref 0.017 0.383 0.137 0.811

0.64

0.43e0.96

0.031

1.003

1.000e1.006

0.033

1.91

1.62e2.24

A novel prognostic nomogram accurately predicts hepatocellular carcinoma recurrence after liver transplantation: analysis of 865 consecutive liver transplant recipients.

Although radiologic size criteria (Milan/University of California, San Francisco [UCSF]) have led to improved outcomes after liver transplantation (LT...
877KB Sizes 0 Downloads 9 Views