How many nodes must be examined to accurately stage gastric carcinomas?

BACKGROUND There are few population based data about lymphadenectomy practices in resections for stomach carcinoma. The aim of the current study was to describe these practices and to determine how many nodes must be examined in order to accurately stage gastric carcinoma. METHODS The current study included all patients (749 patients total) with TNM Stage I, II, or III resected gastric carcinoma diagnosed over a 21 year period (1976–1996) in a well-defined French population. A model of the relationship between the proportion of positive nodes and the number of examined nodes was performed. The relationship between the proportion of positive nodes and the number of nodes examined was modeled to determine the number of nodes beyond which the proportion of N+ tumors no longer changed. RESULTS The average number of examined lymph nodes was 8.4 per patient. More than 15 lymph nodes were examined in 17.6% of patients. Under 10 examined nodes, the proportion of N+ classified tumors significantly decreased with the number of examined nodes, whereas it remained stable beyond this value. The type of gastrectomy and patient age were the two factors independently associated with the resection of at least 10 nodes, whereas the diagnosis period was not associated. After adjustment for the type of surgical resection, three variables independently influenced the prognosis of TNM Stage I or II tumors: extension within the gastric wall (extension T3/T4 vs. T1/T2; odds ratio [OR] = 2.05, P < 0.001), age at diagnosis (age ≥ 70 years vs. < 70 years; OR = 4.06, P < 0.001), and the number of examined nodes (10 resected lymph nodes vs. 0–9; OR = 0.57, P < 0.001). CONCLUSIONS The current study strongly suggests that staging is not reliable when fewer than 10 lymph nodes are examined. The number of examined lymph nodes should be used as a stratification criterion in clinical trials and as an adjustment variable in survival studies. Cancer 2002;94:2862–6. © 2002 American Cancer Society. DOI 10.1002/cncr.10550