How many lymph nodes is enough? Defining the optimal lymph node dissection in stage I-III gastric cancer using the National Cancer Database.

Abstract

4034 Background: Gastric cancer is one of the most causes of cancer-related death worldwide. Surgical resection with lymph node dissection is the primary therapeutic modality. However, the appropriate extent of lymph node dissection remains controversial. Herein, the National Cancer Database (NCDB) was used to determine the optimal number of lymph nodes (LNs) to be dissected for resectable gastric cancer. Methods: The NCDB was queried from 2004-2013 for patients with invasive gastric cancer who underwent surgical resection with negative margins. The optimal number of LNs dissected was determined using a univariate χ2 cut-point analysis. Actuarial survival was determined using the Kaplan Meier method, and comparisons of survival estimates were completed with log-rank tests. Multiple sensitivity analyses were utilized to decrease bias. Results: 17,851 patients were included. The mean (±SD) number of LNs examined was 16 ± 11. For all patients, the optimal number of LNs needed to be examined was 20+ nodes. When correcting for stage migration ( < 7 LNs removed), the optimal cut-off value was 20+ LNs. When stratifying by pathologic nodal stage, the cutpoint was 10+ LNs for pN1 and pN2. The 5-year survival was 30.6 ± 1.6% for 0-9 removed LNs compared to 48.2 ± 1.2% for 10+ removed LNs (p < 0.001) in pN1 disease and 18.3 ± 1.7% for 0-9 removed LNs compared to 32.6 ± 1.2% for 10+ removed LNs (p < 0.001) in pN2 disease. For pN3 disease, the optimal cut-off point was 20+ LNs; the 5-year survival was 17.2 ± 1.3% for 0-19 removed LNs compared to 28.5 ± 1.7% for 20+ removed LNs (p < 0.001). Moreover, the outcome was inferior among patients who had > 10% positive dissected LNs (p < 0.05). Conclusions: The optimal number of dissected LNs of 20+ LNs was associated with superior survival. Extended LN dissection is to be considered especially in patients with > 10% positive dissected LNs.