Advanced renal cell carcinoma (RCC) is relatively resistant to systemic therapy. Response rates (RRs) to the chemotherapy agents studied most extensively, primarily floxuridine (FUDR) or fluorouracil(Drug information on fluorouracil) (5-FU), have ranged from 0% to 14% in the majority of trials. Current standard treatment with cytokines including interferon (IFN) and interleukin-2 (IL-2, Proleukin) benefit only selected patients, with few surviving long term, and is associated with significant toxicities. Based on better understanding of the genetics of RCC, several targeted therapies have been investigated in this disease, and results from recent randomized phase II and III trials of multitargeted agents have been encouraging.[3-6]
In the largest phase III trial in advanced RCC completed to date, the Treatment Approaches in Renal Cancer Global Evaluation Trial (TARGETs), progression-free survival (PFS) doubled from 12 weeks to 24 weeks (P < .00001) vs placebo among patients who received sorafenib(Drug information on sorafenib) (Nexavar; formerly known as BAY 43-9006), an inhibitor of serine-threonine kinases c-raf and b-raf, as well as multiple tyrosine kinases, mainly vascular endothelial growth factor receptor 2 (VEGFR2).[3,5] However, RR among sorafenib-treated patients, based on investigator assessment, was 10%, supporting the increasing evidence that RRs are rarely the best surrogates for anticancer treatment-induced clinical benefit, particularly for targeted therapies. Patients treated with targeted agents may achieve stabilization of disease, decreased tumor growth rate, or tumor regression, rather than shrinkage in tumor size qualifying as objective response based on traditional criteria, such as Response Evaluation Criteria in Solid Tumors (RECIST) or World Health Organization (WHO) criteria.
Furthermore, the high variability of tumor growth rates as well as rare spontaneous regressions known to exist in RCC present additional caveats in evaluating efficacy of targeted cytostatic, or even cytotoxic agents in this tumor.[1,8-11] For example, it may be difficult to differentiate treatment effect from the natural history of the disease in patients with slow-growing tumors. Furthermore, if tumor growth rates vary substantially in a study population, the statistical power of the findings is diluted, requiring a larger sample size for valid results.
These issues and recent findings of discordance between RR and clinical benefit in TARGETs and other trials of molecular targeted therapies have raised new questions regarding optimal methodologies to evaluate and monitor activity of these agents.[13,14] This article will review some of those questions, focusing on recent studies in RCC.
An Endpoint Hierarchy
The quantity of novel targeted agents becoming available (as well as the high proportion of "negative" phase III trials despite promising phase II findings) underscores the importance of defining endpoints that reliably predict efficacy of these agents. A hierarchy of endpoints for outcome measures was proposed by Fleming, who described a true clinical efficacy measure as being at the top of the hierarchy. While improving survival is the clear primary goal, this endpoint requires large sample sizes and a long duration, and therefore conducting such a trial with this primary endpoint would be costly. Surrogate endpoints follow on this hierarchy, and those that can be validated are of most value. Notably, one important example of a validated surrogate endpoint accepted by the US Food and Drug Administration (FDA) for regulatory drug approval is the use of 3-year disease-free survival (DFS) as a surrogate for the previously accepted 5-year overall survival (OS) for adjuvant therapy in colorectal cancer (CRC). (Specific details are described in detail elsewhere in this supplement by Sargent.) Validation of this surrogate was based on meta-analysis of data from 20,898 patients who participated in 18 trials worldwide.[17,18]
PFS is increasingly accepted as a surrogate endpoint for survival in oncology, including RCC. However, improved PFS (and implied clinical benefit) in the absence of a placebo control may be difficult to validate and to interpret, especially in particular tumor types (eg, advanced RCC). Advantages of using PFS as the primary outcome measure in clinical trial design are that data can be obtained earlier than survival results and outcome is less affected by subsequent-line therapies. Although large-volume data sets validating PFS as a reliable surrogate are lacking, especially from phase III studies, results of several prospective randomized comparisons of cytokine regimens in patients with advanced RCC suggest that increases in PFS may correlate with OS benefit in this disease.[19-21]
Biologic Surrogate Markers
Additionally, pharmacodynamic markers or measures of mechanism-based effects are also needed to confirm whether the treatment modulates the intended target protein, and whether patient outcome improves as a result. The majority of genetic alterations in clear-cell RCC, the most common histologic subtype, derive primarily from changes in the von Hippel-Lindau (VHL) tumor suppressor gene.[22,23] Changes in VHL contribute to RCC tumorigenesis through pathways including VEGF, platelet-derived growth factor-beta (PDGF-ß), and transforming growth factor-alpha (TGF-α), as well as other downstream target proteins (Figure 1).[2,22,24] Thus, multiple molecular pathways (and perhaps redundant pathways) play active roles in this tumor, and most of the novel agents that have demonstrated some efficacy in RCC are targeted to multiple proteins, which complicates identification of useful pharmacodynamic markers of efficacy. In patients with RCC, there currently is no biomarker that can be obtained from blood; tumor biopsies are also difficult to obtain routinely for biomarker analysis.