In general, we have come to think of mismatch repair–deficient colon cancer as having a more favorable prognosis, being less likely to metastasize to regional nodes or distant sites, and being resistant to fluoropyrimidines. Much of our data, however, come from trials combining stage II and III patients, and treating them predominantly with fluoropyrimidines alone.

In their recent manuscript, reviewed in this issue of The ASCO Post, Sinicrope et al evaluated the impact of mismatch repair deficiency and other potential prognostic indicators in the North Central Cancer Treatment Group (NCCTG) N0147 trial, which treated stage III colon cancer patients with FOLFOX (leucovorin, fluorouracil, and oxaliplatin) with or without cetuximab (Erbitux). Since cetuximab was not active in this situation, both arms were combined for these molecular analyses. The findings, while instructive, are more complex than expected.

Location and Mutation

In the overall analysis, mismatch repair status did not correlate with prognosis. In a deeper analysis, however, mismatch repair status became prognostic when looked at by the side of the colon in which the tumor arose.

Within the right, or proximal colon, mismatch repair–deficient tumors had a more favorable disease-free survival, while in the left-sided, or distal portions of the colon, mismatch repair–deficient tumors not only did not have a better prognosis, but rather, trended strongly toward a worse prognosis than mismatch repair–proficient tumors from the same region. Further complicating matters, the site of tumor was itself independently prognostic, with distal, or left-sided tumors, having a better prognosis than proximal, or right-sided, tumors regardless of mismatch repair status.

In addition, mutation in either KRAS or BRAF independently conferred a worse disease-free survival prognosis in mismatch repair–proficient tumors. This effect was not seen in mismatch repair–deficient tumors; however, the numbers begin to get small at this point, and the possibility of an interaction cannot be reliably excluded on the basis of these data. The degree of nodal involvement was also independently prognostic in both mismatch repair–proficient and mismatch repair–deficient tumors, with N2 tumors having a worse prognosis than N1 tumors.

Underlying Physiology

Why should location within the colon carry such strong prognostic implications? The answer is not clear, but it may be instructive to recall that the distal portion of the large intestine, from the cloaca through the distal third of the transverse colon, is embryologically derived from the hindgut, which is drained by the inferior mesenteric vein, whereas the proximal colon, from the cecum through the proximal two-thirds of the transverse colon, is derived from the midgut, and is drained by the superior mesenteric vein.

To what degree these differences in embryologic origin and/or vascular drainage contribute to the prognostic implications of the tumor location is not known at this point. However, it is reasonable to speculate that these issues may contribute to the differences observed.

Practical Implications

The more immediate question, however, is how can we utilize this information today? From a clinical trials perspective, these prognostic data can be used to either stratify patients prospectively in new trials, or to evaluate for potential imbalances that may either explain or invalidate apparent differences in already-accrued trials.

From a current patient care perspective, however, it is not yet possible to reliably bring these data into the decision-making process, since these data are prognostic and not predictive. Prognostic data tell us who has a better or worse chance of a successful outcome, but they do not tell us whether we are able to influence that outcome. In fact, by definition, we cannot influence the outcome, or the information would be predictive, rather than prognostic.

We cannot choose which side of the colon our patient’s tumor will arise in, nor can we influence the mismatch repair or other mutational status of our patient’s tumor. Predictive data, which would tell us who is or is not likely to benefit from a treatment or intervention, are not derivable from the current report, since all patients effectively received the same treatment.

Actionable Markers Needed

At present, oxaliplatin plus a fluoropyrimidine remains the recommended adjuvant therapy for all patients with stage III colon cancer in the absence of medical contraindications. While it is reasonable to conjecture that some of the prognostic subgroups may either benefit more or fail to benefit from a particular maneuver, such as chemotherapy or oxaliplatin-based chemotherapy, we simply do not know yet if this is the case. It is therefore important to recognize that these interesting data, while informing us about who is more likely to have a favorable outcome, do not yet help us figure out who is or is not likely to have an improved outcome as the result of one intervention or another.

It would be highly desirable to have data that define subgroups of stage III patients whose prognoses do not improve with oxaliplatin, so that these patients could be spared the unnecessary toxicity, or to define subgroups who would likely benefit from irinotecan, cetuximab, or other agents. We are not there yet, and these current data should not be misconstrued as such, but analyses of this type in other trials will hopefully provide us with actionable predictive markers in the near future. ■

Dr. Saltz is Chief of the Gastrointestinal Oncology Service in the Department of Medicine at Memorial Sloan-Kettering Cancer Center, New York. 

Disclosure: Dr. Saltz is a consultant for Roche, Genentech, Bristol-Myers Squibb, Sanofi, and Pfizer.