Purpose: In tumor-node-metastasis (TNM) staging, no recommendations are provided on the number of lymph nodes to examine per N level. We investigated the relationship between number of nodes examined and probability of detecting metastases, and ultimately staging accuracy.
Patients and methods: Data from 473 curatively resected gastric cancer patients, from a randomized trial that compared the Western resection with limited and the Japanese resection with extended lymphadenectomy, were used.
Results: Yields varied between zero to 67 in N1 and zero to 43 in N2. Yields correlated significantly with classification; the more nodes examined, the more likely the classification was N+. Calculating the probability of detecting metastases using either a likelihood-ratio estimate or an exponential model, this probability increased steeply in the lower range and more gradually in the higher range yield.
Conclusion: To standardize treatment results, we emphasize histologic examination of a fixed number of nodes per anatomically defined N level (20 in N1 and 15 in N2), corresponding with a probability to detect metastases of 60% to 65% and 40% to 45%, respectively. These yields are feasible in all resection specimens, while the extra increase in the chance of detecting metastases by examining more nodes does not compensate the effort to increase yields further. Thus, standardization of histologic TNM staging can improve and render comparison of surgical and adjuvant treatment modalities feasible.