Challenges and Pitfalls of Combining Targeted Agents in Phase I Studies

For many years, oncologists have adhered to the principle that more is better, based on the observation that the dose-response curve of various cytotoxic agents shows a steep increase over a clinically achievable range of drug concentrations. Although this concept is largely derived from in vitro data using cell lines, and even though it has been challenged by well-conducted randomized trials, we still use the maximum-tolerated dose (MTD) as the preferred metric for dose finding in phase I clinical trials of classical cytotoxic agents. Such an approach is not unreasonable, because it recognizes the possibility that lower doses might not be as effective, and even if they were, we generally have not had the courage to find out. However, targeted therapies have forced us to rethink this paradigm. For agents such as bevacizumab, it is difficult to rigorously define a dose-limiting toxicity, because the relationship between toxicity and drug dose is generally quite flat. Other agents, such as the signal transduction inhibitors, generally do have an MTD, as defined by dose-related toxicities such as diarrhea, skin changes, constitutional symptoms, and—in some cases—bone marrow suppression or hypertension, depending on the agent. However, for either monoclonal antibodies like bevacizumab, or oral tyrosine kinase inhibitors such as sorafenib, the need to push the dose to the MTD to obtain an optimal clinical effect is an assumption that is generally unproven. For most targeted agents, we simply do not have a good understanding of the relationship between the MTD and the dose required to achieve the desired therapeutic effect. The difficulty in determining the optimal clinical doses of targeted agents is further compounded when these agents are used in combination. In this issue of Journal of Clinical Oncology, Azad et al report the results of a phase I trial of a combination of bevacizumab and sorafenib in 39 patients with a variety of tumor types. Bevacizumab is a humanized monoclonal antibody that neutralizes vascular endothelial growth factor (VEGF), and sorafenib is an oral tyrosine kinase inhibitor with multiple targets that include VEGF receptor, platelet-derived growth factor receptor, and Raf kinase. Because both agents share the VEGF pathway as targets, and because each agent has single-agent activity in a variety of clinical settings, it was thought that additive or synergistic effects might be observed when they were used in combination. The authors found that a regimen of daily oral sorafenib and every 2 weeks intravenous bevacizumab was difficult to tolerate, even at the lowest dose level of sorafenib 200 mg twice daily and bevacizumab 5 mg/kg every 2 weeks. A significant number of patients experienced fatigue, hand-foot syndrome, hypertension, proteinuria, and thrombocytopenia, with 24 of 39 patients experiencing either grade 3 or 4 toxicity at this dose level (Table 3 in Azad et al). These problems required further dose reduction of sorafenib to 200 mg daily, as opposed to twice daily, in 74% of patients. One patient died of hemoptysis in association with a cavitating pulmonary hilar mass. Two patients with ovarian cancer developed an enterocutaneous fistula at the site of disease regression, and one patient with melanoma experienced an appendiceal perforation. Although none of these side effects represents a new toxicity signal for these agents, it seems that the incidence of adverse events was greater than what would be expected through the use of each agent alone at standard doses. For reference, a standard dose of bevacizumab in combination with paclitaxel and carboplatin for non–small-cell lung cancer is 15 mg/kg every 3 weeks, and a standard dose of single-agent sorafenib in the treatment of renal cell cancer is 400 mg twice daily. The authors show compelling evidence for at least additive toxicity of this combination, and they conclude that a further refinement in the sorafenib dose or schedule will be needed in future studies. The enhanced toxicity of this combination did not seem to be related to a pharmacologic interaction between these two drugs, on the basis of pharmacokinetic analysis comparing single-agent with dualdrug therapy. One may ask why future studies should even be performed, given these results. Here we have a rational combination of two agents that target the VEGF pathway, but the toxicity is significant for unexplained reasons, and the final doses of each agent are subtherapeutic as compared with the doses used in single-agent trials of these same agents. Nonetheless, there is good reason to pursue this regimen, a reason that is fortuitous and not the usual purview of a phase I trial. The authors are pursuing this regimen primarily because of an unexpectedly high response rate in patients with relapsed ovarian cancer. Specifically, partial responses were observed in six (46%) of 13 patients as assessed by Response Evaluation Criteria in Solid Tumors (RECIST), compared with the approximately 16% to 21% response rate expected with bevacizumab alone. It is impossible to know whether this high response rate is related to the combination of bevacizumab and sorafenib or whether it might have been seen with bevacizumab alone in this small patient cohort. Information regarding the number of prior regimens and whether the patients were platinum resistant was not provided for the ovarian cancer cohort. Although no evidence of bowel perforation was observed in these patients with ovarian cancer, this serious complication will require close monitoring in future studies. As with all promising phase I or II data, this response rate will need to be confirmed in a larger trial, and this combination JOURNAL OF CLINICAL ONCOLOGY E D I T O R I A L VOLUME 26 NUMBER 22 AUGUST 1 2008