Addition of tyrosine kinase inhibitors (TKIs) in patients (pts) with unresectable hepatocellular carcinoma (HCC) who progress on first-line immunotherapy (IO).

e16193 Background: Atezolizumab plus bevacizumab is now standard first-line therapy for advanced HCC. Early phase trials in HCC and phase III trials in other solid tumors testing IO and TKI have shown greater antitumor activity with the combination than with either agent alone. Cancer and vascular endothelial cells release VEGF, which not only increases angiogenesis, promoting tumor growth, but also induces an immunosuppressive tumor microenvironment (TME). Preclinical studies suggest that TKIs with anti-VEGF activity have immunomodulatory potential, enabling the infiltration and activation of effector T cells to the TME. Experimental models of HCC with combination of IO plus TKI also reverted immune escape mechanisms mediated by β-catenin, Wnt, and TGFβ activation and by Treg or neutrophil infiltration. At our institution, pts who were treated with IO alone in the front line often had a TKI added on progression of disease (PD). We aimed to evaluate whether the addition of a TKI could rescue IO failure. Methods: We conducted an IRB-approved retrospective chart review of pts with unresectable HCC at Mount Sinai Health System who received IO from 1/2017 to 6/2021 and who subsequently received a TKI on PD. We assessed the objective response rate (ORR), median progression-free survival (mPFS) and overall survival (mOS). Follow-up CT/MRI imaging studies were conducted every 2-3 months after addition of TKI. Results: Of 277 pts who received first-line IO during the study period, 46 evaluable pts subsequently had TKIs added to IO upon PD. Pts were predominantly male (37/46), with median age 63 years, and 38 had cirrhosis. Etiologies of HCC were as follows: 9 HBV, 22 HCV, 15 alcohol-related, 7 NASH-associated. All pts received first-line nivolumab. TKIs added were lenvatinib (44/46) and sorafenib (2). At the start of IO, most pts (31/46) had Child Pugh class A liver disease and performance status ECOG 0 (31/46). TKIs were added upon PD and after a median of 5.8 months (range 2.1-28.1 months) from IO initiation. ORR of IO plus TKI was 15% (7/46; 2 complete responses and 5 partial responses). 16 pts had stable disease, and 13 had PD. Among the 11/46 pts who initially responded to first-line IO, 2 achieved a response with TKI. There were 5 responses among the 35/46 non-responders to first-line IO. TKI-related grade 3-4 adverse events, including liver toxicity, rash, and encephalopathy, led to discontinuation in 7/46 pts. mOS after adding TKI was 9.5 months (95% CI: 6.1-13.7) and mPFS was 5.4 months (95% CI: 3.6-9.7). The mOS and mPFS of adding TKI after PD did not differ between IO initial responders and non-responders. Overall, mOS from IO initiation was 17.6 months (95% CI: 14.5-24.4). Conclusions: Adding a TKI after PD on single agent IO in pts with advanced HCC did not add a signal of clinical benefit compared with reported outcomes from second-line TKIs alone.