Response

We thank Wen-zhuo et al. for their comments on our article, “A Randomized Phase II/III Study of Dalotuzumab in Combination With Cetuximab and Irinotecan in Chemorefractory, KRAS Wild-Type, Metastatic Colorectal Cancer,” which has been recently published in this journal (1). The points raised by Wen-zhuo et al. reiterate the challenges encountered when conducting biomarker analyses in clinical trials.

Quantitative real-time polymerase chain reaction has been widely used for the assessment of members of the IGF pathway (2). We acknowledge that the accuracy of this detection method is potentially limited by post-translational modifications and degradation of nucleic acids secondary to tissue fixation (3,4). However, one of the main advantages of mRNA analysis over immunohistochemistry is to provide a quantitative measure of the expression of the biomarker, simplifying data interpretation and minimizing the risk of inter- and intra-observer variability. This is especially relevant for ligands such as IGF-1 and IGF-2 that are produced by both cancer and stromal cells and are involved in autocrine and paracrine signaling (5).

As shown in the supplementary files, the cutoff value for IGF-1 was chosen based on data from an independent gene expression profiling dataset, where IGF-1 overexpression was found in 12 of 43 (27.9%) patients with KRAS wild-type colorectal tumors. We are not aware of previously validated or widely accepted criteria for the definition of IGF-1 positivity.

We agree with Wen-zhuo et al. that knowledge of the source of tumor tissues that were used to assess biomarker expression would allow a better interpretation of study findings. IGF-1 expression may vary depending on the site of the primary tumor (colon vs rectum) or the nature of the sampled lesion (primary tumor vs metastasis) (6). Moreover, contamination by normal surrounding tissue is possible, and the relevance of this phenomenon is likely to be higher for liver lesions compared with extrahepatic lesions (5). Unfortunately, as acknowledged in the manuscript, data on the source of tumor tissues are not available. Therefore, we would like to take this opportunity to recommend once again caution when interpreting the results of our biomarker study.

The reason why subgroup analyses by IGF-1 and IGF-2 expression were reported only for arms A and C is that the former appeared to be the more “biologically active” of the investigational arms. Indeed, the dose intensity of dalotuzumab in arm A was higher compared with arm B. Moreover, an unexpected detrimental effect with the addition of this agent to standard therapy was observed only in arm A, a phenomenon that might also suggest the potential of IGF-1R inhibition to accelerate tumor growth via aberrant feedback loops in intrinsically resistant patients. No difference in progression-free survival (PFS) or overall survival (OS) was observed between arms B and C according to the expression of IGF-1 (P_interaction = .32 and .17, respectively). When the outcome of all dalotuzumab-treated patients (arm A + arm B) was compared with that of the placebo group, a treatment by IGF-1 interaction was confirmed and approached statistical significance for both PFS (P = .06) and OS (P = .06).

David Cunningham received research funding from: Roche, Amgen, Celgene, Sanofi, Merck Serono, Novartis, AstraZeneca, Bayer, Merrimack, and MedImmune. All the other authors do not have conflicts of interest to disclose.