In Reply

Abstract

We appreciate the interest of the authors in our review, and commend them and their collaborators on the completion of two very important studies that have challenged current dogma regarding combination chemotherapy. We believe that questioning dogma is always a useful exercise. In our review and in clinical practice, cross-study comparisons are inevitably used to make judgments on how to manage patients. The comparison of data across studies is both a necessary endeavor to exploit the body of evidence on which to make clinical decisions and a hazardous one because of subtle and sometimes unrecognized differences in study populations and, in the case of the CAIRO [1] and FOCUS [2] trials, in study regimens. Currently, comparisons across studies with disparate outcomes are based on subjective rather than objective interpretation and that is likely the root of our different conclusion from Koopman and colleagues when we view the same data. However, we believe that there is some additional objective information relevant to our conclusions that should be considered.

Koopman and colleagues conjecture that lack of access to second-line therapy is the reason that IFL was not as good as FOLFOX in N9741. There are solid data to refute that assertion. In the report by Goldberg et al. [3] of the reduced-dose IFL (rIFL) comparison with FOLFOX, there was equal access to second-line therapy, and 58% of FOLFOX patients went on to get an irinotecan-based second-line therapy, and 58% of rIFL patients went on to get an oxaliplatin-based second-line therapy. The dose intensity of rIFL was nearly identical to that of IFL in the original report but with significantly better toxicity. The outcomes were similar to the original report of N9741 favoring FOLFOX for both activity and toxicity, including a better overall survival (OS) result. We have also shown that patients who were not judged to have a response by World Health Organization criteria still benefited from FOLFOX over IFL, obscuring the value of using response rate as a primary endpoint [4]. On the other hand, complete response (CR) does seem to matter in that patients with CRs on N9741 had a very long median survival time, as compared with those with less or no response [5]. The preponderance of data across trials suggests that combination chemotherapy is associated with higher response and higher CR rates than single-agent initial therapy.

With regard to the issue of outcomes in the CAIRO and FOCUS trials, we stand by our statement that the relatively poor outcomes in the combination arms of both the FOCUS and CAIRO studies, compared with several other trials [6–12], make the results difficult to generalize. Whether these outcomes were a result of patient selection factors, as suggested by the authors of this letter, or other factors, including imbalances across populations in predictive biomarkers such as microsatellite instability or others, is not entirely clear based on the reported results of the studies. Specifically, the assertion that less positive outcomes arose from the inclusion of differing patient populations in CAIRO and FOCUS as compared with other trials that show better results may or may not be valid. For example, the inclusion of patients with unresectable metastatic colorectal cancer was apparently equivalent in the N9741 and CAIRO studies, yet outcomes were significantly better in N9741. Use of the combination of capecitabine and irinotecan as the preferred first-line therapy (in lieu of infusional 5-fluorouracil [FU]) in the CAIRO study could possibly have altered the OS result. To our knowledge, there was no definitive evidence at the time the CAIRO study was developed that the capecitabine and irinotectan (CapeIri) regimen was equivalent in its efficacy or toxicity profile to the IFL, Douillard, or irinotecan combined with bolus and continuous infusion 5-FU plus leucovorin (FOLFIRI) regimens. In fact, the results of the randomized BICC-C [9] trial strongly suggest that CapeIri is inferior to FOLFIRI both in terms of activity measures and toxicity patterns. The FOCUS trial used a variation of FOLFIRI and a variation of FOLFOX, both of which prescribe less dose-intensive regimens than the regimens described by de Gramont et al. [8]. It appears that optimizing regimens matters, and the combination regimens employed in the CAIRO and FOCUS studies may not have been the optimal regimens to use as comparators to single-agent therapy; an assertion supported by the comparatively short OS time reported for the combination regimens of both studies.

Our statement about differences in survival between arms should also be clarified. The schema for the complicated FOCUS study is provided for reference (Fig. 1). In the FOCUS trial, the OS time for the reference or “A” strategy (5-FU followed by irinotecan) versus the “C” strategy (upfront FOLFIRI-like regimen) was actually different (hazard ratio, 0.84), and even the B strategy (5-FU followed by FOLFIRI-like regimen) versus the C strategy using irinotecan produced a full 1.7-month differential in the median OS time. This difference is likely not significantly different only because of sample size. Interestingly, the patients randomized to the oxaliplatin-based regimens in FOCUS did not seem to fare as well as those on the irinotecan-based regimens. This finding is inconsistent with the results of many other contemporary studies and suggests that important characteristics of the enrolled population in the FOCUS trial may differ in an important way from those enrolled in the other studies.

Figure 1.

FOCUS study arms.

Abbreviations: 5-FU, 5-fluorouracil; FOCUS, Fluorouracil, Oxaliplatin, and CPT11—Use and Sequencing; Iri, irinotecan; LV, leucovorin; MCRC, metastatic colorectal cancer; Ox, oxaliplatin.

Open in new tabDownload slide

Lastly, with regard to the comment about the appropriateness of looking at nonsignificant differences between arms, both FOCUS and CAIRO were designed as superiority trials, but analyzed and interpreted as noninferiority studies (in the case of FOCUS, a noninferiority analysis was performed post hoc, but not so for the CAIRO study). The fact remains that several large, randomized trials demonstrate benefits in terms of OS for superior combination regimens, where these two studies do not. We assert that it is incorrect to interpret a negative superiority trial as demonstrating equivalency just as it is to interpret a nonstatistically significant difference in outcome as superiority. Thus, our feeling remains that although some subpopulation of patients with advanced colorectal cancer may do as well with 5-FU–based single-agent therapy as they would with combination therapy, to date this subgroup remains undefined. Until we can define that subpopulation, combination chemotherapy remains our preferred approach.

References

Author notes