Back to 2011 Program

BACKGROUND: Lymph node ratio (LNR) has been shown to predict survival in patients with pancreatic cancer. However, its role has not been evaluated in the context of the number of lymph nodes (LNs) examined. METHODS: A national population-based dataset (SEER) of 10,254 patients and a prospective database of 827 patients who underwent surgery at a single institution were reviewed for tumor and patient characteristics. In each database, patients were divided into subsets based on the number of examined LNs, with cut points putting approximately a third of the patients into each subset. The SEER database was divided into: ≤ 5, 6- 12, and ≥13 examined LNs. The number of examined LNs at the single institution was generally higher and correspondingly the single institution’s patients were divided into: ≤ 9, 10 - 16, and ≥ 17 lymph nodes. Univariate and multivariate analyses were done using Kaplan-Meier curves and Cox regression modeling. RESULTS: The different subgroups were uniform in terms of patient’s age at presentation, sex, tumor size, stage and primary site (chi2, P < 0.05). In SEER, the mean LNR decreased in a stepwise manner as the number of examined lymph nodes increased (0.38 for ≤ 5LNs vs. 0.19 for ≥13LNs, P < 0.05). A similar trend was also seen in the single institution’s dataset (0.29 for ≤9LNs vs. 0.15 for ≥17LNs, P < 0.05). Overall, age at diagnosis (> 65 yrs), absolute numbers of positive LNs, and a higher LNR (> 0.2) were associated with a worse survival. On multivariate analysis in SEER, a Cox regression model showed that LNR > 0.2 has an adverse impact on survival in each sub-group and the magnitude of this impact increased with the number of examined LNs: HR (±95% CI); ≤5LNs:1.52(±0.20); 6-12LNs: 2.95(±0.55), and ≥13LNs: 3.25(±0.76). For the single institution, the Cox regression showed that a lymph node ratio > 0.2 had an adverse effect on survival, except in the subset where ≤9LNs were examined; HR, (±95% CI): 1.37(±0.69), P = 0.126. In the subgroups where LNR was a significant predictor, the strength of this prediction correlated positively with the numbers of examined LNs: HR, (± 95% CI); 10-16LNs: 1.61(±0.49), P = 0.010; ≥17LNs: 2.17(±0.86), P = 0.003. CONCLUSION: Although an LNR of >0.2 has a strong negative predictive value for survival, the accuracy of this prediction and the relative risk of death is higher when more lymph nodes are examined. This suggests that LNR must always be interpreted in the context of the number of examined lymph nodes.