How does tumour viability influence the predictive capability of the Metroticket model? Comparing predicted-to-observed 5-year survival after liver transplant for hepatocellular carcinoma

• Published in: Canadian Journal of Surgery Vol. 64; p. S152
• Main Authors: Smith, V, Rivers-Bowerman, M, Costa, A, Stueck, A, Campbell, N, Allen, S, Gala-Lopez, B
• Format: Journal Article
• Language: English
• Published: Ottawa CMA Impact, Inc 01-12-2021
• Subjects: Liver cancer; Liver transplants; Tumors
• ISSN: 0008-428X; 1488-2310

Background: The Metroticket project produced prognostic calculators for patients undergoing liver transplant (LT) for hepatocellular carcinoma (HCC). Radiology-based and pathology-based calculators predict 5-year HCC-specific and overall survival, respectively. Our objective was to evaluate how viable tumour burden at explant affects the predictive capability of the Metroticket model. Methods: A retrospective cohort analysis of HCC LT patients from 1996 to 2019 was conducted. Locoregional therapy (LRT) data, radiographic parameters and explant pathology findings including tumour viability were collected. Metroticket predicted survival was calculated. Tumour response to LRT was assessed. Radiographic total tumour volume (TTV) and explant total viable tumour volume (TVV) were correlated. HCC-specific survival of these subgroups was assessed using Kaplan-Meier curves and compared via log-rank testing. Finally, predicted versus observed survival was compared by patient subgroups. Results: Eighty patients were included. TTV and TVV correlated strongly (Pearson r = 0.98, p < 0.01), with imaging overestimating TVV by 42.1%. There was no significant difference in HCC-specific survival if patients underwent LRT (p = 0.50), regardless of tumour response (p = 0.85), nor if tumours were viable (p = 0.10), regardless of viable tumour burden (p = 0.74). Similarly, the presence of microvascular invasion (p = 0.73) or satellitosis (p = 0.99) did not influence HCC-specific survival. Observed 5-year overall survival was significantly lower than predicted by the Metroticket model for patients with viable tumours (66.3% v. 61.8, p = 0.03). This finding was amplified in patients without vascular invasion (78.9 v. 66.7%, p < 0.01). Patients with viable tumours and presence of microvascular invasion demonstrated overall survival significantly greater than predicted by the Metroticket model (60.6 v. 51.7%, p < 0.01). Conclusion: In our study, HCC-specific survival does not appear to be affected by LRT or the burden of viable tumours. However, tumour viability does appear to influence the predictive capability of the Metroticket model, varying with the presence of microvascular invasion. Integrating tumour viability in the Metroticket model may augment its efficacy.