Factors contributing to cure in T2 pancreatic cancer.
234 Background: It is crucial to determine whether cancer therapy achieves cure or merely prolongs time to death from cancer (failure time). Unfortunately, conventional survival tests such as log-rank test and Cox regression fail to distinguish between cure and prolonged failure time, both making a great difference in survival benefit. In 1994, Gamel extended the Boag model to three regressions, allowing us to study whether prognostic factors have significant effect on cure rate or failure time. Using the Gamel-Boag regressions, we study curative effects of clinicopathological factors in pancreatic cancer. Methods: Follow-up data from 452 patients who underwent pancreatectomy for T2 ductal cell cancer were analyzed. The prognostic factors to be studied were converted to binary variables, which include sex, age (<=60 or >60 years), tumor size (<=2 or >2 cm) lymphnode metastasis, and six types of invasions, i.e., serosal, retroperitoneal, portal vain, common bile duct and duodenal invasions. By calculating coefficients of the three regressions and their confidence intervals, we evaluated whether each (or combinations) of these variables are associated with increased cure rate or prolonged failure time. Results: Tthere were 11 relapse-free 5-year survivors. Except for tumor size (TS) and lymphnode metastasis (N), no factors were significantly associated with increased cure rate or prolonged failure time. The only factor found to significantly affect cure was TS; the estimated cure rate being 44% for patients with TS <= 2 cm and 4% otherwise (p=0.0001). This effect was not influenced by N and other factors. Only N significantly accelerated failure time; the median failure time being 11.9 months when N was negative and 6.7 months otherwise (p<0.0001). Thus, both TS and N affected the prognosis significantly but in different ways. By contrast the Cox regression showed that both TS and N were significantly associated with hazard reduction (p=0.003 and p<0.001, respectively), which, however, did no reveal a qualitative difference in survival outcome between TS and N. Conclusions: In T2 pancreatic cancer, a highly significant increase in cure rate is found when TS≤2 cm. This increase is not affected by N or tumor extension to nonpancreatic tissues. No significant financial relationships to disclose.