In Response: Dr. Watanabe and colleagues argue that we might have underestimated the actual prognostic benefit conferred by tumor microsatellite instability (MSI; ref. 1), because the vast majority of MSI cancers were in the proximal colon, a tumor location possibly associated with poorer prognosis (2, 3). The hypothesis is theoretically intriguing but inconsistent.
The prognostic role of tumor site in patients with colon cancer remains controversial. A small but significant disadvantage for patients with colon cancer proximal to the splenic flexure (5-year disease-specific survival rate: 63.9% versus 68.6%) can be recognized among patients with any-stage cancer included in the Surveillance, Epidemiology, and End Results (SEER) U.S. registry (4). Conversely, the proximal site has been recently associated with a better 5-year disease-free survival rate (66.8% versus 60.3%) in a large series of patients with stage II and stage III colon cancer (5).
In any event, tumor site was not a significant predictor of survival in our series. At univariate analysis (Supplemental Data 4 of ref. 1), the risk of death for patients with distal colon cancer (hazard ratio, 1.0; reference) was similar to that of patients with proximal colon cancer (hazard ratio, 0.91; 95% confidence interval, 0.68-1.23, P = 0.55) or rectal cancer (0.87; 95% confidence interval, 0.65-1.19, P = 0.39). At step-down variable selection (Table 4 of ref. 1), tumor site was rejected as a nonsignificant predictor of death both in model A (including tumor stage, P = 0.31) and in model B (not including stage, P = 0.41). As both models included microsatellite status, the implication is that the tumor location had no influence on the assessment of MSI as a prognostic predictor. Beyond models, MSI can be confirmed to be a stage-dependent predictor of survival if analysis is limited to patients with proximal colon cancer (hazard ratio, 0.47; 95% confidence interval, 0.24-0.90, P = 0.02, at univariate; hazard ratio, 1.04; 95% confidence interval, 0.52-2.08, P = 0.91, after stage-adjustment), although no conclusion can be drawn about patients with distal cancer because of the small number of patients with MSI tumor. Finally, we must reject any hypothesis of selection bias possibly accounting for the high prevalence (81%) of proximal cancers among the MSI tumors. In fact, all cancers consecutively resected at our institution were analyzed for microsatellite instability. Rather, the prevalence of proximal MSI tumors in our series is in full accordance with that of nearly all published studies (6).
RSS Feeds