Until the 1990s, Robson's modification of Flocks and Kadesky's system, which emphasized outcomes in patients with locally advanced RCC by focusing on the extent of tumor spread, was accepted as the standard staging system. With the rising incidence of localized RCC over the past two decades [4
], the TNM classification has become more widely accepted in clinical practice, and many investigators have assessed the predictive ability of the system, resulting in its continuous refinement [5
Following the introduction of the 2002 TNM classification, large retrospective cohort studies have suggested subdivision of T2 disease on the basis of tumor size [6
], and the National Cancer Data Base findings were confirmed regarding impact of size of T2 disease on CSS [16
]. In addition, several independent studies suggested that T3 RCC should be subdivided, because ipsilateral adrenal involvement was associated with poorer survival than was T3a RCC with fat invasion [8
]. Other studies showed that the survival rates of patients with renal vein thrombosis were better than those of patients with subdiaphragmatic vena caval thrombosis [11
]. These led to further subclassification of T2 RCC in 2009 into T2a and T2b with a 10-cm cutoff, and the criteria for T3 and T4 classifications were modified [16
The 2002 TNM classification of RCC has been validated by various institutions [2
], but the 2009 TNM classification has not yet been widely explored. A multi-institutional Italian study found that the 2009 TNM classification was a powerful independent predictor of CSS, but that some primary tumor classifications afforded overlapping prognoses, and some included patients with heterogeneous outcomes [17
]. However, the cited study had the limitations of heterogeneity of multi-institutional data and a lack of comparison with the 2002 classification. This motivated us to perform our current work, which compares the 2002 and 2009 TNM classification of RCC with the use of a large, single institutional cohort.
Our results indicate that the 2009 primary tumor classification performed well in stratification of patients with locally advanced RCC on the basis of CSS. The 2009 primary tumor classification was statistically significantly associated with death from RCC with a stepwise increase in risk ratios from T3a to T4, which were approximately 19 for T3a, 37 for T3b, 45 for T3c, and 61 for T4 in the univariate analysis. Although the pairwise survival differences between T3c and the adjacent primary tumor classification did not show statistical significance owing to the small sample size, the difference between T3b and T4 approached statistical significance (pairwise p=0.095). In a multi-institutional Italian study, Novara et al also reported similar CSS between patients with T3c and T4 RCC [17
]. Similarly, from a large, single institutional cohort, Kim et al reported no significant survival difference between T3b and T3c diseases [20
]. However, in a multi-institutional study including 585 patients with inferior vena cava involvement or higher, Martínez-Salamanca et al found tumor thrombus level was an independent predictor of CSS even after adjustment for tumor size, grade, regional lymph node involvement, distant metastases, and fat invasion [19
However, we found no significant difference in the CSS of patients with T2a and T2b disease, and we observed that the subclassification of T2 disease by using the 2009 primary tumor classification did not afford superior predictive ability than that offered by the 2002 classification (0.808 vs. 0.810). Our observation that the CSS values of T2a and T2b RCC patients did not differ may be partly attributable to our relatively small sample size. Our study cohort included only 211 patients with T2 RCC and was thus statistically underpowered to detect survival differences between patients with T2a and T2b disease. In a single-institutional study of 544 patients with T2 RCC, Frank et al reported that patients with tumors 10 cm or greater in diameter were more likely to die of RCC than were patients with tumors less than 10 cm in diameter, after adjustment for regional lymph node involvement and distant metastasis [6
]. Recently, from the same institutional cohort including more patients, Kim et al reconfirmed the adverse prognosis for T2b tumors compared with T2a and found modestly superior predictive ability of the new TNM classification compared with the 2002 TNM classification [20
]. In an international multicenter study of 706 patients with T2 RCC, Klatte et al found a significant association of tumor size with CSS (HR 1.11, p<0.001) and suggested that a tumor diameter size cutoff of 11 cm was appropriate [7
]. The results from a multi-institutional Italian study supported a subdivision of T2 tumors by 10 cm [17
]. However, a recent multi-institutional German study of 579 patients with T2 RCC revealed no significant difference in CSS between T2a and T2b patients (p=0.38). Even so, previous reports found that tumor size was significantly associated with CSS when size was modeled as a continuous variable in patients with localized RCC [22
] and also in patients with locally advanced RCC [23
]. Thus, it may be that any cutoff is significantly associated with CSS if the sample size is sufficiently large.
The 2002 lymph node classification considers the absolute number of metastatic lymph nodes (Nx, no regional lymph nodes resected; N0, no nodes with metastasis; N1, metastasis in a single regional lymph node; N2, metastasis in more than one regional lymph node), assuming that at least eight lymph nodes are examined. However, several studies found no significant difference in the survival of N1 and N2 patients [14
], and the 2009 lymph node classification thus integrated N1 and N2 into a single stage, designated N1 [16
]. In the present study, we found no significant difference in the CSS of patients with N1 or N2 disease. Thus, amalgamation of node-positive RCC may improve the predictive ability of the 2009 TNM classification when adjusted for regional lymph node involvement.
The present study had several limitations. The major problem is that we used a retrospective design and did not include pathologic review, and thus our results may be vulnerable to confounding errors and bias. Second, with a tremendous increase in stage I tumors, the proportions of T2b, T3b, T3c, and T4 disease were rather low. Thus, our study may not have had sufficient statistical power to detect survival differences among groups of patients with these classifications of disease. Finally, most of our patients who experienced recurrence did not have access to targeted therapies for RCC because they were treated before the development of these relatively new agents.
Primary tumor size has long been a key component of the primary tumor classification for localized RCC. In the development of the 1997 classification, the only modifications of the 1987 classification were the selection of optimal tumor size cutoff points to distinguish T1 and T2 disease. However, the tumor size was still an independent prognostic factor in integrated prognostic systems, including tumor stage [25
]. The 2002 and 2009 TNM classifications were modified to select optimal tumor size cutoffs for classification of T1 and T2 disease.
In general, TNM classifications must be modified over time to improve prognostic ability and to keep pace with the demands of clinical practice. The frequencies of small and incidental RCCs are increasing, and the cutoff for PN is currently a diameter of 4 cm [27
]. The subclassification of T1 RCC was thus designed to keep pace with the results from and requirements of clinical practice. An RCC patient is rarely assigned to T2 classification, because the 2002 TNM classification included renal sinus fat invasion and sinus vein invasion in T3a and T3b, respectively [28
]. Moreover, both T2a and T2b RCCs are treated by RN. In future updates of the TNM classification of RCC, we suggest that clinical demands and treatment options be considered in addition to patient prognosis.