Aims: Clinical records of patients undergoing surgery for breast cancer were reviewed in order to evaluate the prognostic role of lymph-node level involvement.
Methods: From 1982 to 1991, 1143 patients had radical mastectomy or conservative surgery with total axillary dissection: 461 patients of mean age 57.1 years (range: 25-89 years) were lymph-node positive (pN1); 369 patients (80%) had radical mastectomy; and 92 patients (20%) had conservative treatment plus post-operative radiotherapy, with the same mean number (n = 16) of lymph nodes collected in the surgical specimen. Data were analysed for the number of positive lymph nodes and level of involvement.
Results: Level I, Levels I + II and Levels I + II + III were involved in 44.9, 18 and 21.4% of patients, respectively; 'skip metastases' occurred in 72 of 461 pN1 patients (15.5%). A univariate analysis showed that prognosis was directly related to the number of levels involved (P < 0.001), and skip metastases had the same prognostic role as Level I involvement. The numbers of involved lymph-node levels and metastatic lymph nodes were well correlated; multivariate analysis showed that involvement of Levels I and III was independently correlated with prognosis. After adjustment for age and number of positive lymph nodes, the number of involved lymph-node levels was an independent prognostic factor, with highest predictability when all three lymph-node levels were positive (P = 0.009).
Conclusions: The prognostic value of lymph-node status should be defined not only by the number of metastatic lymph nodes, but also by the number of levels of involvement.