Back to 2008 Program and Abstracts
Staging Error Does Not Explain the Relationship Between the Number of Nodes in a Colon Cancer Specimen and Survival.
Jesse Moore*1, Neil H. Hyman1, Peter Callas2, Benjamin Littenberg3
1Surgery, University of Vermont College of Medicine, Burlington, VT; 2Mathematics & Biostatistics, University of Vermont College of Medicine, Burlington, VT; 3Med-Gen Internal Medicine, University of Vermont College of Medicine, Burlington, VT
Purpose: Survival in colon cancer is greater in those patients who have more lymph nodes identified at the time of resection. It is not clear if this is due to under-staging, confounding by treatment, social or clinical characteristics, or factors intrinsic to the tumor or host response. We studied whether the number of nodes analyzed per specimen remains predictive of survival while controlling for patient factors, surgeon, and hospital.
Methods: A retrospective cohort study of 11,399 patients in the SEER-Medicare database who were diagnosed with Stage I, II, or III colon cancer between 1994 and 1998 was performed. The primary predictor was the number of lymph nodes identified. Cox proportional hazards models were constructed to analyze overall and cancer-specific survival. Analyses were adjusted for age, sex, race, income, use of chemotherapy, surgeon’s specialty, and the usual tendency of the surgeon and hospital to identify lymph nodes.
Results: Overall and cancer specific survival was significantly higher as the number of nodes examined increased. Patients with more than 12 nodes examined had a 20% lower hazard (HR 0.80; 95% confidence interval 0.75-0.85) than those with fewer than 7 nodes examined. Controlling for patient factors did not change the hazard ratio associated with an increasing number of nodes. Further controlling for surgeon type, the surgeon’s and the hospital’s mean number of nodes examined per case in the models did not change the hazard ratios associated with an increasing number of nodes. This indicates that something other than patient demographics, treatment, operating surgeon or the hospital's process for retrieving lymph nodes accounts for the observed effect.
Conclusion: The number of analyzed nodes in a colon cancer specimen is predictive of survival, but is independent of patient demographics, surgeon and hospital. This argues strongly against under-staging or confounding as the mechanism for the improved survival with higher node counts. Other factors, such as those intrinsic to the tumor or the host immune response may be responsible for the differences in prognosis.
