Purpose: We analyzed the effects of the change in TNM classification from the 1987 to the 1997 version and suggest a modified tumor size cutoff point between T stages 1 and 2 for renal cell carcinoma.
Materials and methods: We evaluated a database containing the records of 661 patients who underwent nephrectomy between 1989 and 1999. The effect of the change in TNM classification on the distribution of patients between stages, the rates of M+ and N+ disease, and the local and distant recurrence rates were outlined for 280 patients with T stages 1 and 2 disease. The Cox model was used to identify the optimal cutoff point between T1 and T2 disease, and the resulting effect of adopting this cutoff was outlined.
Results: A total of 174 and 128 cases were down staged from 1987 version stage T2 to 1997 version stage T1 and from 1987 TNM stage II to 1997 TNM stage I, respectively. Survival was not significantly different in patients with 1997 TNM stages I and II disease due to a lack of survival difference during the first 2 years of followup. Stage shift also caused an increase in average tumor size, the proportion of patients with high grade cancer, and M+ and N+ disease at diagnosis in 1997 stages T1 and T2 as well as an increase in the proportion of 1997 stage T2N0M0 cases at diagnosis with systemic failure. Analysis of 11 potential cutoff points between 1 and 10 cm. revealed that 4.5 cm. was most predictive of patients survival (hazards ratio 4.99, p = 0.0001). Using this cutoff resulted in improved discriminatory power of the TNM classification and a moderating effect on the distribution of patients, average tumor size, high grade disease, M+ and N+ disease at diagnosis, and systemic failure between T(14.5) and T(24.5) compared with 1997 T1 and T2.
Conclusions: Our data imply that the current cutoff point of 7 cm. between stages T1 and T2 tumors is too high. Lowering the cutoff to 4.5 cm. resulted in better discriminatory power of the TNM classification in our dataset. This observation should be further validated by external data.