Brivanib and FOLFOX in Hepatocellular Carcinoma: Finding the Common Themes Among Negative Trials

Abstract

This editorial accompanies three articles on three randomized, phase III trials of one disease: advanced hepatocellular carcinoma (HCC). The studies explored folinic acid, fluorouracil, and oxaliplatin (FOLFOX) versus doxorubicin in first-line therapy (EACH; Qin et al); brivanib versus sorafenib in first-line therapy (BRISK-FL; Johnson et al); and brivanib versus placebo after failure of sorafenib (BRISK-PS; Llovet et al). Despite a perceived low bar for success, none of these trials met its prespecified target for efficacy, joining ranks with disappointed contemporaries that include sunitinib, linifanib, and the combination of sorafenib plus erlotinib. In stark contrast with the postsorafenib era in renal cell carcinoma, which has seen six new US Food and Drug Administration approvals since 2005, these results underscore the enormous unmet need for active systemic therapies in HCC. Why is the failure rate so high in HCC? Pessimists might take the stance that nothing really works well in HCC owing to indiscriminate chemoresistance, with upregulation of multidrug resistance genes and efflux mechanisms long credited with the failures of conventional cytotoxic chemotherapeutic agents in this disease. Beyond a chemoresistant phenotype, however, there are other common themes across chemotherapy and targeted therapy trials alike, including high rates of toxicity when treating patients at the recommended phase II doses in this population with prevalent liver dysfunction, uninformative surrogate imaging end points, and a remarkable degree of prognostic heterogeneity within clinical and biologic subsets. There remains a distinct possibility that these challenges inherent to HCC could make identifying active therapies difficult in some cases or, conversely, suggest activity when none exists. With these considerations in mind, the outcomes of the recent negative trials in HCC merit careful examination so that we might learn from our failings. In the EACH trial, 371 patients were randomly assigned to open-label FOLFOX versus systemic doxorubicin (50 mg/m intravenously every 3 weeks) between 2007 and 2009, across 38 centers in four Asian countries. The majority of patients was positive for hepatitis B virus (HBV) infection (approximately 90% in both study arms) and had Barcelona Clinic Liver Cancer (BCLC) stage C disease (approximately 80% in both arms). There were noteworthy trends toward higher numbers of prior transarterial chemoembolization cycles (mean, 3.46 v 2.77 cycles) and a greater proportion of patients with prior systemic therapy in the doxorubicin arm compared with the FOLFOX arm (30% v 21%). More patients from the doxorubicin study arm withdrew from the study after randomization but before treatment (n 13 v n 1) but were included in the intention-totreat efficacy analyses. Toxicity was generally similar between treatment arms, though grade 3 or higher all-cause adverse events were more common on the FOLFOX arm (55.74% v 45.40%; P .05). At the prespecified final analysis, there was a nonsignificant trend toward improved median overall survival in the FOLFOX arm compared with the doxorubicin arm (6.40 v 4.97 months; hazard ratio [HR], 0.80; P .07). A subsequent post hoc analysis of survival 7 months later found an overall survival improvement in the FOLFOX arm (6.47 v 4.90 months; P .04), although this unplanned look does not make this study positive. The median progression-free survival rate, response rate, and disease control rate were significantly higher in the FOLFOX arm compared with the doxorubicin arm at both the prespecified final analysis and the post hoc analysis. The proportion of patients in each arm who received subsequent sorafenib or other therapies was not reported. How should we interpret the significant improvement in response-based end points and the post hoc survival analysis in this study while acknowledging the negative primary survival end point? An unavoidable limitation of EACH was the lack of blinding. The open-label design is likely to have contributed to the imbalance in withdrawal rates observed between treatment arms and had the potential to confound local response assessment. The intention-to-treat efficacy analysis (owing to the imbalance in withdrawals) and the numerically higher rates of prior therapies in the control arm may have favored the treatment arm, as well. Another fundamental limitation was the use of doxorubicin as the control, a consequence of the study’s inception in the presorafenib era. The evidence for doxorubicin’s benefit in HCC is marginal, arising from a 106-patient trial in which doxorubicin was superior to no antitumor therapy but not in a clinically meaningful context, with dismal overall survival rates in both arms (2.7 v 1.9 months, respectively; P .036). Doxorubicin also was superior compared with nolatrexed in a randomized phase III trial of 445 patients. Should we infer from these data and EACH that systemic doxorubicin is incrementally better than best supportive care and presume transitively that FOLFOX would be superior? JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 31 NUMBER 28 OCTOBER 1 2013