Serum IGF-1 Scores and Clinical Outcomes in the Phase III IMbrave150 Study of Atezolizumab Plus Bevacizumab versus Sorafenib in Patients with Unresectable Hepatocellular Carcinoma

Abstract

Purpose: Child-Turcotte-Pugh class A (CTP-A) in unresectable hepatocellular carcinoma (HCC) is the standard criterion for active therapy and clinical trial enrollment. We hypothesized that insulin-like growth factor-1 (IGF-1) derived scores may provide improved survival prediction over CTP classification. This study aimed to evaluate the potential prognostic and predictive effects of IGF-1 derived scores in the phase III IMbrave150 study.

Patients and methods: Baseline and on-treatment serum IGF-1 levels from 371 patients were subjected to central analysis. Patients' IGF-1 score (1/2/3) and IGF-CTP score (A/B/C) were determined based on pre-specified cutoffs. Outcomes were analyzed by baseline and by on-treatment changes of the IGF-1 and IGF-CTP scores within and between the two treatment arms. The interaction between these scores and outcomes was assessed using univariate and multivariate analyses.

Results: Baseline IGF-CTP score (A vs B/C) showed prognostic significance for OS in both the atezolizumab-bevacizumab (hazard ratio [HR], 0.33; 95% confidence interval [CI], 0.20-0.56; P<0.001) and sorafenib (HR, 0.32; 95% CI, 0.16-0.65; P=0.002) arms. Baseline IGF-1 score (1 vs 2/3) also showed prognostic significance for OS in both the atezolizumab-bevacizumab (HR, 0.33; 95% CI, 0.20-0.55; P<0.001) and sorafenib (HR, 0.48; 95% CI, 0.26-0.89; P=0.02) arms. HRs for PFS were consistent with those for OS. No significant predictive effects were observed for either score between the two arms. Kinetic analysis revealed that patients with increased IGF-1 score (1-> 2/3) at 3 weeks post treatment had shorter OS than patients with stable IGF-1 score of 1 in both the atezolizumab-bevacizumab (HR, 3.70; 95% CI, 1.56-8.77; P=0.003) and sorafenib (HR, 5.83; 95% CI, 1.88-18.12; P=0.0023) arms.

Conclusion: Baseline and kinetic change of IGF-CTP and IGF-1 scores are independent prognostic factors for patients with unresectable HCC treated with atezolizumab-bevacizumab or sorafenib. These novel scores may provide improved patient stratification in future HCC clinical trials. IMbrave150 ClincialTrials.gov number, NCT03434379.

Keywords: HCC; IGF-CTP score; immunotherapy; prognostic biomarker.

Figure 1

Association of serum insulin-like growth factor-1 (IGF-1) levels with clinicopathological features. Box plots show the associations between serum IGF-1 levels and (A) serum levels of albumin (ALB), bilirubin, alkaline phosphatase (ALP), alanine transaminase (ALT), and aspartate transaminase (AST); (B) tumor occupancy of ≥50% of liver; (C) high-risk status; (D) portal vein tumor thrombosis (PVTT); and (E) etiology. The upper and lower borders of each box represent maximum and minimum values, respectively, and the line inside each box represents the median of each data set. The points are test values of individual patients. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. High risk was defined as tumor invasion into the main trunk of the portal vein and/or the portal vein branch contralateral to the primarily involved lobe (VP4), and/or bile duct invasion and/or tumor occupancy of ≥50% of liver.

Figure 2

Prognostic and predictive effects of baseline IGF Child-Turcotte-Pugh (CTP) score. The prognostic effects of IGF-CTP scores are shown with Kaplan-Meier (KM) curves of OS (A) and PFS (B) stratified by IGF-CTP A vs B/C for patients within the atezolizumab-bevacizumab (Atezo+Bev) or sorafenib treatment arms. The predictive effect of IGF-CTP scores on Atezo+Bev benefit over sorafenib is shown with KM curves of OS (C) and PFS (D) stratified by treatment arms among patients with IGF-CTP A scores or with B/C scores. The HRs (95% CIs) and P values shown in each graph are adjusted for known prognostic factors described in the Patients and Methods section. NE, not established.

Figure 3

Prognostic and predictive effects of baseline IGF levels (Point 1 vs Point 2/3). The prognostic effects of IGF-1 levels are shown with KM curves of OS (A) and PFS (B) stratified by IGF-1 Point 1 vs Point 2/3 for patients within the Atezo+Bev or sorafenib arms. The predictive effect of IGF-1 levels on Atezo+Bev benefit over sorafenib is shown with KM curves of OS (C) and PFS (D) stratified by treatment arms within patients with Point 1 or with Point 2/3 IGF-1 levels. The HRs (95% CIs) and P values shown in each graph are adjusted for known prognostic factors described in the Patients and Methods section. NE, not established.

Figure 4

Association of kinetics of IGF-1 levels or IGF-CTP scores with survival outcomes. KM curves of OS stratified after 1 cycle of Atezo+Bev (A) or sorafenib (B) treatment IGF-1 levels stayed at Point 1 (Stable Point 1), decreased from Point 1 to Point 2/3 (Deteriorated) or stayed at Point 2/3 (Stable Point 2/3). KM curves of OS stratified after 1 cycle of Atezo+Bev (C) or sorafenib (D) treatment IGF-CTP classes stayed at A (Stable A), changed to B/C (Deteriorated) or stayed at B/C (Stable Class B/C).