Introduction: Surgery remains the only potentially curative treatment for pancreatic cancer. The presence of positive lymph nodes (LN) has consistently been found to be associated with poor prognosis. Recent studies in gastric and colon cancer have demonstrated that the number of harvested LN holds prognostic significance. In this study, we sought to determine the relevance of LN harvest during resection for pancreatic cancer in the VA patient population.
Methods: A VA Central Cancer Registry query for pancreatic adenocarcinoma yielded 884 individuals for whom adequate data was available. 506 individuals had positive LN on pathologic examination. The number of LN examined, number of positive LN, and lymph node ratio (LNR) (the ratio of positive to examined LN) were analyzed using Kaplan-Meier survival analysis and its log-rank test.
Results: Overall survival (OS) decreases markedly with fewer examined LN. (p < 0.0001) Five-year OS in the highest quartile (11-46 LN) of nodes examined was 9%, while only 0% for the lowest quartile (1 LN). As expected, the detection of even one positive LN in the resection significantly decreases OS (p < 0.0001) with 5 year survival decreasing from 13% to 0%. Among individuals with positive LN, 5 year OS increased between the first (1 LN) and fourth quartile (11-46 LN) of LN examined. (p<0.001). An increasing number of positive LN, however, does not appear to have an effect on OS. (p = NS) For patients with negative LN, survival curves were similar to positive LN when only one LN was examined. LNR is an important prognostic factor for OS in individuals with positive LN.(p<0.0001) Five year OS for individuals with the lowest ratio was 2% while 0% for the quartile with the highest ratio. In univariate analysis, LNR had a hazard rate of 2.070 and odds ratio of 2.265.
Conclusions: For patients undergoing resection for pancreatic cancer, LN positivity affects OS, but, LNR appears to have greater prognostic significance for OS than the number of LN examined. This suggests that the number of positive LN present needs to be evaluated in the context of the total number of LN evaluated.