Proposal for revision of the TNM classification system for renal cell carcinoma

Authors


Abstract

BACKGROUND

The current study defined an optimal tumor size breakpoint to stratify localized renal cell carcinoma (RCC) into groups with significantly different cancer-related outcomes and proposed a revision of the TNM classification system.

METHODS

The authors analyzed the data from 1138 patients who had undergone partial or radical nephrectomy for localized RCC at 7 European urologic centers. The optimal pathologic size breakpoint was calculated using the martingale residuals from a Cox proportional hazards regression model.

RESULTS

The mean follow-up time was 87 months. The scatterplot of tumor size versus expected risk of death per patient suggested that an interval of 5–6 cm was appropriate. A total of 720 (63.3%) and 418 (36.7%) patients had tumors measuring ≤ 5.5-cm and tumors measuring > 5.5-cm, respectively. Significant cancer-specific survival differences between the two groups of patients were reported in the series by all the centers participating in the study. On univariate analysis, the other variables found to be associated with cancer-specific survival were the patient's age, symptomatic tumor presentation, and the Fuhrman nuclear grade. On multivariate analysis, the pathologic stage of the primary tumor defined according to the 5.5-cm breakpoint was found to be an independent predictor of cancer-specific survival, as well as age, mode of presentation, and nuclear grade. According to the multivariate analysis, the authors clustered patients into 3 groups with statistically significant outcome differences: 1) patients with ≤ 5.5-cm incidentally detected RCC; 2) patients with ≤ 5.5-cm symptomatic RCC; and 3) patients with > 5.5-cm RCC. This cancer-related outcome stratification was valid regardless of the patient's age.

CONCLUSIONS

The 5.5-cm breakpoint was found to be the optimal tumor size breakpoint with which to stratify patients with organ-confined RCC. The study supported the upgrade of the TNM classification system according to this breakpoint. Cancer 2005. © 2005 American Cancer Society.

The tumor, nodes, and metastasis (TNM) classification system, proposed by the International Union Against Cancer (UICC) and the American Joint Committee on Cancer (AJCC), is the most widely used to assess the local extension of tumors, as well as the locoregional lymph node status and the presence of distant metastasis. TNM status is a relevant outcome predictor, strictly related to cancer-specific and disease-free survival probabilities in oncology.

The first edition of the TNM, proposed by the UICC in 1968, did not include a classification of renal tumors.1 Renal cell carcinoma (RCC) was included in the third edition of the TNM classification system, which was published by the UICC in 1974.2 The TNM classification system has been modified over the years, mainly with the intent of improving prognostic predictivity and, consequently, the quality of the data useful in planning therapeutic strategy. In the most recent TNM editions, the most common modifications concerned the stratification of organ-confined RCC into subgroups with different prognoses (T1–2N0M0). In the 1987 TNM version, localized tumors were subclassified using a size breakpoint of 2.5 cm.3 According to the data from the Surveillance, Epidemiology, and End Results program, the 1997 TNM version settled on 7 cm as the new breakpoint.4 A recent systematic review of the literature showed that both the 1987 and the 1997 TNM versions failed to distribute the observed cases adequately within Stage I and Stage II and that significantly different cancer-specific survival probabilities were reported in only a few series.5 The last TNM edition, published in 2002, introduced a further subclassification of Stage I tumors, according to a 4-cm breakpoint (T1a vs. T1b), with the intent of providing an easier clinical identification of the patients who would be suitable for elective nephron-sparing surgery (NSS).6, 7 However, Stage I and Stage II RCC continued to be classified according to the 7-cm cutoff point.

The aim of the current study was to define an optimal tumor size breakpoint capable of stratifying localized RCC into groups with significantly different cancer-related outcomes, providing all the information required by the UICC TNM committee to propose a revision of the TNM staging system.8

MATERIALS AND METHODS

We analyzed the clinical data of 2217 patients who had undergone either partial or radical nephrectomy for localized RCC (T1–2N0M0) between 1984 and 2001 at 7 different European urologic centers: Verona, Padua, and Naples (Italy); Rennes, St. Etienne, and Creteil (France); and Graz (Austria). All the patients who were alive and disease free but with a follow-up time of < 60 months were excluded from the analysis. The remaining 1138 patients were investigated in the current study.

From the Kidney Cancer Databases of each center, the following clinical and pathologic data were evaluated: age, gender, mode of presentation, TNM pathologic status, pathologic tumor size (cm), histotype, nuclear grade, and follow-up data.

With regard to the mode of presentation, patients were clustered in three different groups: Group 1 (S1) included incidentally detected RCC, diagnosed during abdominal imaging studies performed for signs and symptoms unrelated to cancer; Group 2 (S2) consisted of the RCC found because of the presence of local symptoms, such as hematuria, flank pain, or flank mass; and Group 3 (S3) comprised RCC identified after the onset of systemic symptoms such as loss of energy, fatigue, weight loss, fever, and cough.9, 10 All the neoplasms were reclassified according to the 2002 TNM classification systems: localized tumors with a pathologic size ≤ 4 cm were classified as pT1a, and localized tumors measuring 4–7 cm and > 7 cm were identified as pT1b and pT2, respectively.7 The Heidelberg and Fuhrman classifications were used to assign the histologic type and nuclear grade, respectively.11, 12

Statistical Analysis

Continuous variables were reported as mean value ± standard deviation and range or as median value and interquartile ranges, as appropriate. The Student t test and the Mann–Whitney U test were used to compare the locations of continuous variables, as appropriate. The Pearson chi-square test was employed to compare categoric variables. The relation between tumor size and death from RCC was investigated using a scatterplot of tumor size versus the martingale residuals from a Cox proportional hazards regression model.13 Patients were divided into two groups according to the identified breakpoint.

The survival intervals were defined as the time elapsed from surgery to the last clinical evaluation or the patient's death. The survival curves were estimated using the Kaplan–Meier method. Patients who were alive or who had died of other causes were censored (cancer-specific survival). The log-rank test was used for comparison of the survival curves and for the univariate analysis.

The Cox proportional hazards model was used to identify the clinical and pathologic variables that were capable of independently predicting cancer-specific survival.

In all the statistical analyses, a two-sided value (P < 0.05) was considered to be statistically significant. Data were analyzed using the Statistical Package for Social Sciences (SPSS, Inc., Chicago, IL) (version 12.0) software and the SAS software package (SAS Institute Inc., Cary, NC).

RESULTS

Three hundred twenty-one patients (28.2%) were treated in Verona, 221 (19.4%) were treated in Padua, 60 (5.3%) were treated in Naples, 178 (15.6%) were treated in Rennes, 133 (11.7%) were treated in St. Etienne, 64 (5.6%) were treated in Creteil, and 161 (14.1%) were treated in Graz.

The mean patient age was 58.47 ± 12.39 years (range, 15–87 yrs). Seven hundred fifty-two patients (66.1%) were male and 386 (33.9%) were female. In 637 patients (56%), neoplasms were incidentally detected (S1), whereas RCCs were diagnosed after the onset of local (S2) or systemic symptoms in 451 (39.6%) and in 50 (4.4%) patients, respectively. Radical or partial nephrectomies were performed in 873 (76.7%) and 265 (23.3%) patients, respectively. The median pathologic tumor size was 5 cm (interquartile range, 3–7 cm). Nine hundred twenty tumors (80.8%) were classified as pathologic Stage I (512 pT1a and 408 pT1b) and 218 tumors (19.2%) were classified as pathologic Stage II. According to the Heidelberg classification, 988 patients (86.7%) had conventional RCC, 89 patients (7.8%) had papillary RCC, 14 patients (1.2%) had chromophobe RCC, and 41 patients (4.1%) had unclassified RCC. The Fuhrman nuclear grade was Grade 1 in 452 patients (39.7%), Grade 2 in 544 patients (47.8%), and Grade 3 in 142 patients (12.5%). Table 1 summarizes the clinical and pathologic characteristics of the 1138 analyzed patients.

Table 1. Clinical and Pathologic Characteristics of 1,138 Analyzed Patients
CharacteristicsNo. of patients (%)
  1. SD: standard deviation; RCC: renal cell carcinoma.

Institution 
 Verona321 (28.2)
 Padua221 (19.4)
 Naples60 (5.3)
 Rennes178 (15.6)
 St. Etienne133 (11.7)
 Creteil64 (5.6)
 Graz161 (14.1)
Age in (yrs) 
 Mean ± SD58.47 ± 12.39
 Range15–87
Gender 
 Male752 (66.1)
 Female386 (33.9)
Mode of presentation 
 Incidental (S1)637 (56)
 Local symptoms (S2)451 (39.6)
 Systemic symptoms (S3)50 (4.4)
Type of surgery 
 Radical nephrectomy873 (76.7)
 Partial nephrectomy265 (23.3)
Histotype 
 Conventional RCC988 (86.8)
 Papillary RCC89 (7.8)
 Chromophobe RCC14 (1.2)
 Unclassified RCC41 (4.1)
Pathologic stage using (TNM, 1997–2002) 
 Stage I920 (80.8)
 Stage II218 (19.2)
Fuhrman nuclear grade 
 Grade 1452 (39.7)
 Grade 2544 (47.8)
 Grade 3142 (12.5)

The mean follow-up time was 87 months (IQR, 68–130 cm). The martingale residuals from a Cox proportional hazards regression model suggested that a breakpoint between 5 cm and 6 cm was appropriate (Fig. 1).

Figure 1.

Scatterplot of tumor size versus the expected risk of death from renal cell carcinoma (RCC) for each of the 1138 patients. Patients above the horizontal line are at an increased risk of death compared with the expected risk from a Cox proportional hazards regression model. Patients below the horizontal line are at a decreased risk of death compared with what we would expect. The curved line represents a scatterplot smoother. The point at which this smoother crosses the horizontal line occurs between 5 cm and 6 cm, indicating an optimal cutoff point with which to predict death from RCC occurs between these 2 sizes.

According to the statistical analysis, we identified the 5.5-cm pathologic size as the optimal breakpoint to distinguish 2 groups of organ-confined RCC with significantly different cancer-specific survival rates.

A total of 720 patients (63.3%) were classified with ≤ 5.5-cm tumors and 418 (36.7%) were classified with > 5-cm tumors. The patients with tumors measuring > 5.5 cm had higher rates of local or systemic symptoms (P < 0.001) and higher Fuhrman nuclear grades (P < 0.001). The mean follow-up time was similar for the two groups (P = 0.53). Table 2 reports the characteristics of the patients stratified by the new 5.5-cm breakpoint.

Table 2. Clinical and Pathologic Characteristics of the Examined Patients Stratified by the 5.5-cm Breakpoint
Variables≤ 5.5 cm (%)> 5.5 cm (%)P value
Mean age in (yrs)59.0257.520.05
Gender  0.11
 Male488 (64.9)264 (35.1) 
 Female232 (60.1)154 (39.9) 
Mode of presentation  < 0.001
 Incidental (S1)470 (73.8)167 (26.2) 
 Local symptoms (S2)235 (52.1)216 (47.9) 
 Systemic symptoms (S3)15 (30)35 (70) 
Nuclear grade  < 0.001
 Grade 1350 (77.4)102 (22.6) 
 Grade 2308 (56.6)236 (43.4) 
 Grade 362 (43.7)80 (56.3) 
Mean follow-up time in (mos)101.0899.060.53
Cancer-related deaths78 (40.8)113 (59.2)< 0.001

The 5-year and 10-year cancer-specific survival probabilities were 93.2% and 87.9%, respectively, for patients with RCCs measuring ≤ 5.5 cm. Patients with neoplasms > 5.5 cm had 5-year and 10-year cancer-specific survival rates of 79.9% and 73.5%, respectively (log-rank P < 0.0001) (Fig. 2). Statistically significant cancer survival differences between the two groups of patients were reported by all the centers participating in the study. Conversely, using both the 1997 and 2002 TNM classification systems, statistically significant survival differences between patients with Stage I and Stage II disease were found in only 4 of the 7 centers (Table 3). On univariate analysis, the other variables found to be associated with disease-specific survival were age > 60 years (log-rank P = 0.0002), symptomatic tumor presentation (log-rank P < 0.0001), and Fuhrman nuclear grade (log-rank P < 0.0001). Tumor histotype was not found to be statistically significant (log-rank P = 0.68) (Table 4).

Figure 2.

Kaplan–Meier cancer-specific survival according to the new breakpoint of 5.5 cm. Log-rank P < 0.0001. Solid line: ≤ 5.5-cm renal cell carcinoma [RCC]; dotted line: > 5.5-cm RCC.

Table 3. 1997 and 2002 TNM Classification and 5.5-cm Breakpoint Stratification: Application within the Series of Each Participating Center
Center (no. of cases)1997–2002 TNM classification systems5.5-cm tumor size break-point stratification
5-yr cancer-specific survival probability, %Log-rank P value5-yr cancer-specific survival probability, %Log-rank P value
Verona (n = 321)Stage I = 94.60.10≤ 5.5 cm = 96.10.001
 Stage II = 86.4 > 5.5 cm = 87.7 
Padua (n = 221)Stage I = 91.90.08≤ 5.5 cm = 94.40.006
 Stage II = 79.7 > 5.5 cm = 81.2 
Naples (n = 60)Stage I = 950.0006≤ 5.5 cm = 96.60.005
 Stage II = 65 > 5.5 cm = 73.3 
Rennes (n = 178)Stage I = 92.20.0004≤ 5.5 cm = 92.60.02
 Stage II = 74.1 > 5.5 cm = 80.2 
St. Etienne (n = 133)Stage I = 96.40.01≤ 5.5 cm = 98.80.006
 Stage II = 85 > 5.5 cm = 87.2 
Creteil (n = 64)Stage I = 920.09≤ 5.5 cm = 94.60.04
 Stage II = 64 > 5.5 cm = 73.9 
Graz (n = 161)Stage I = 91.30.001≤ 5.5 cm = 94.40.0002
 Stage II = 60 > 5.5 cm = 71.4 
Overall (n = 1138)Stage I = 91.3< 0.0001≤ 5.5 cm = 93.2< 0.0001
 Stage II = 75.6 > 5.5 cm = 79.9 
Table 4. Results of the Univariate Analysis (Log-Rank Test)
Variables5-yr cancer- specific survival probability, %10-yr cancer- specific survival probability, %Log-rank P value
  1. RCC: renal cell carcinoma.

Age in (yrs)  0.0002
 ≤ 6091.586 
 > 6084.678.5 
Gender  0.14
 Male87.680.7 
 Female89.686.2 
Mode of presentation  < 0.0001
 Incidental (S1)93.488.9 
 Local symptoms; (S2)/systemic; symptoms (S3)81.875 
Fuhrman nuclear grade  < 0.0001
 Grade 193.190.2 
 Grade 287.680.7 
 Grade 375.363.4 
Histotype  0.68
 Conventional RCC88.682.8 
 Papillary RCC84.281.2 
 Chromophobe RCC85.785.7 
Pathologic stage  < 0.0001
 Stage I91.385.5 
 Stage II75.669.7 
Tumor size breakpoint  < 0.0001
 ≤ 5.5 cm93.287.9 
 > 5.5 cm79.973.5 

On multivariate analysis, the pathologic stage of the primary tumor defined according to the 5.5-cm breakpoint was found to be a variable capable of independently predicting cancer-specific survival, as were age, mode of presentation, and Fuhrman nuclear grade (Table 5).

Table 5. Variables Capable of Independently Predicting Cancer-Specific Survival in Patients with Localized RCC (T1–2-N0-M0): Cox Proportional Hazards Model
VariablesCategoriesHazards ratio95% CIP value
  • RCC: renal cell carcinoma; 95% CI: 95% confidence interval.

  • a

    Mode of presentation included incidental (S1), local symptoms (S2), and systemic symptoms (S3).

Age in (yrs)≤ 60 vs. > 601.9501.459–2.606< 0.001
Mode of presentationaS1 vs. S2/S3a2.1851.589–3.005< 0.001
Nuclear gradeGrade 1 vs. Grade 2 vs. Grade 31.5871.281–1.967< 0.001
Pathologic size≤ 5.5 vs. > 5.5 cm1.8931.392–2.575< 0.001

According to the multivariate analysis data, we clustered patients into 3 groups: 1) patients with ≤ 5.5-cm incidentally detected RCC (T1aN0M0); 2) patients with ≤ 5.5-cm symptomatic RCC (T1bN0M0); and 3) patients with > 5.5-cm RCC (T2N0M0). The 5-year and 10-year cancer-specific survival probabilities were 95.7% and 90.7%, respectively in the first group; 88.4% and 82.9%, respectively in the second group; and 79.9% and 73.5%, respectively in the third group. Statistically significant differences in disease-specific survival probabilities were observed between the first and the second group (log-rank P = 0.0008) and between the first and the third group (log-rank P < 0.0001). Similarly, patients in the second group had statistically significant higher cancer-specific survival rates than those in the third group (log-rank P = 0.002) (Fig. 3). This cancer-related outcome stratification was found to be valid regardless of the patient's age (log-rank P < 0.0001) (Figs. 4 and 5).

Figure 3.

Cancer-specific survival stratified by the 5.5-cm tumor size and symptom classification. The 5-year and 10-year cancer-specific survival probabilities were 95.7% and 90.7%, respectively, in the first group (black curve: ≤ 5.5-cm asymptomatic renal cell carcinoma [RCC]); 88.4% and 82.9%, respectively, in the second group (red curve: ≤ 5.5-cm symptomatic RCC); and 79.9% and 73.5%, respectively, in the third group (green curve: > 5.5-cm RCC). Log-rank P < 0.0001.

Figure 4.

Cancer-specific survival in patients age ≤ 60 years, stratified by the 5.5-cm tumor size and symptom classification. Log-rank P < 0.0001. Black curve: ≤ 5.5-cm asymptomatic renal cell carcinoma [RCC]; red curve: ≤ 5.5-cm symptomatic RCC; green curve: > 5.5-cm RCC.

Figure 5.

Cancer-specific survival in patients age > 60 years, stratified by the 5.5-cm tumor size and symptom classification. Log-rank P < 0.0001. Black curve: ≤ 5.5-cm asymptomatic renal cell carcinoma [RCC]; red curve: ≤ 5.5-cm symptomatic RCC; green curve: > 5.5-cm RCC.

DISCUSSION

The data from the current European multicenter study show that 5.5 cm is the optimal tumor size breakpoint with which to stratify patients with organ-confined RCC into groups with different cancer-specific survival probabilities. The study supports the upgrading of the TNM classification system according to the new breakpoint.

The use of the 2.5-cm breakpoint as a prognostic variable in RCC was introduced for the first time in the 1987 edition of the TNM.3 According to this cutoff point, statistically significant differences in cancer-specific survival probabilities between Stage I and Stage II were not shown in any study. A further drawback of the 1987 TNM edition was the unbalanced allocation of the patients into the 2 subgroups, with only 4–27% of patients assigned to Stage I.5

In 1997, the UICC and the AJCC settled on 7 cm as the new breakpoint with which to distinguish T1 and T2 cases with the objective of better stratifying patients.4 That new criterion caused a sharp increase in the percentage of patients assigned to Stage I, paralleled by a clear decrease in Stage II cases, resulting in a different but still unbalanced distribution.5 Moreover, to our knowledge, only a few published articles to date have reported statistically significant survival differences between Stage I and Stage II patients.14–18

The latest TNM edition, published in 2002, introduced a further subclassification of Stage I patients according to the tumor size breakpoint of 4 cm.7 That modification aimed at providing the clinicians with better criteria to identify patients who were suitable for elective NSS. Hafez et al.,6 analyzing a cohort of patients who had undergone elective partial nephrectomy, demonstrated higher cancer-specific survival rates in patients with ≤ 4 cm RCC compared with patients with neoplasms ranging from 4 cm to 7 cm or > 7 cm. However, Stage I and Stage II RCC were still separated at the 7-cm breakpoint in the 2002 TNM edition.

Over the last decades, several tumor size cutoff points have been proposed to adequately distinguish localized RCC. The most frequently suggested cutoff ranged from 4.5 cm to 5.5 cm. Zisman et al.,19 analyzing the UCLA Kidney Cancer Registry, recommended a breakpoint of 4.5 cm. In 1993, Gelb et al.20 suggested stratifying organ-confined RCC according to a 5-cm breakpoint. The same data were reconfirmed by Targonski et al.,21 who evaluated 93 patients for whom radical nephrectomy was performed for localized RCC Similarly, data from the Mayo Clinic Kidney Cancer Registry confirmed the use of the 5-cm breakpoint.22, 23 Moreover, this tumor size breakpoint was included in the SSIGN (stage, size, grade, and secrosis) score by Frank et al.24 Other recent articles have proposed the same data.25–27 A 5.5-cm breakpoint was indicated in 1999 by Kinouchi et al.,28 who analyzed a series of radical nephrectomies, and more recently by Ficarra et al.29 Ficarra et al. investigated a group of 813 patients who had undergone either radical or partial nephrectomies, using the receiver operator characteristic curve analysis to pinpoint the optimal breakpoint, to subclassify organ-confined RCC. The data in this last article were reconfirmed in the current multicenter European study, although different statistical methods were used.

The application of a breakpoint ranging between 5 cm and 6 cm could allow the identification of patients who are suitable for elective NSS as well as a better patient distribution into Stage I and Stage II within the analyzed single-center series. A few recently published articles underscored that NSS could be safely undertaken in patients with renal neoplasms ≤ 7 cm.14, 30, 31

Moreover, the data from the current multivariate analysis highlight the relevance of the mode of presentation as a variable capable of independently predicting the cancer-related outcome of the patients with localized RCC. Another multicenter study recently indicated the inclusion of the mode of presentation in the TNM classification system.32 The data in the current analysis show that the patients with incidentally detected neoplasms and those measuring < 5.5 cm have the highest 5-year and 10-year cancer-specific survival probabilities. This issue might justify the stratification of patients with localized ≤ 5.5-cm RCC (Stage I) into 2 subgroups according to the absence (T1aN0M0) or the presence (T1bN0M0) of local and/or systemic symptoms at the moment of the diagnosis. This differentiation might allow the staging system to be able to identify patients suitable for elective NSS.

With the objective of keeping up with the modern demand for evidence-based medical practice, the UICC TNM Committee recently codified a structured process for introducing changes in the TNM classification system. Several factors constitute the cornerstone to such a structured process: the well defined process for annual review of relevant literature, the development of unambiguous criteria for the information and documentation required to consider changes in the classification, the formation of expert panels, and the participation of experts from all over the world in the TNM review process.8

Table 6 summarizes the data concerning the current multicenter study required by the UICC TNM Committee to support a proposal for the TNM revision.

Table 6. Check of Information Required by the UICC TNM Committee8
Information requiredData of the study
  1. UICC: International Union Against Cancer; ICD-O: International Classification of Disease for Oncology; RCC: renal cell carcinoma.

Name of the proposing individual, group, or countrySee the author's list and affiliations
Tumor site and tumor morphologyLocalized RCC (ICD-O = 185)
Rationale for suggesting the change and the clinical context surrounding the rationaleBest stratification of confined RCC according to cancer-specific survival probabilities
Years of diagnosis of the patients cohort under consideration in the proposal1984–2001
Description of the cohortSee Table 1
Description of the referral pattern of patients from the communities to the institution(s) represented in the proposal's cohortAcademic European centers
Description of the selection process for the population under studyDatabases of the participating centers. Cancer-related deaths and patients alive at follow-up > 60 mos
The TNM classification used in the proposal's analysis1997 and 2002 TNM (Stage I–II)
Description of additional essential prognostic factors relevant to the proposalPathologic tumor size of RCC
Description of profile of therapeutic interventions received by cohort in the proposalEither radical or partial nephrectomy
Description of end points used in the proposal of the studyCancer-specific survival and mortality rates
Analyses addressing the stated goal(s) of the studyMartingale residuals from Cox hazards regression model
Analyses contrasting the predictive ability of the current TNM system with that of the proposed systemApplication within single-center analysis
Analyses including inferential statistical comparison of the current system and the proposed systemMartingale residuals from Cox hazards regression model
Description and/or citation of unconventional statistical methodsSee the sections “Statistical Analyses” in the current article
Specific recommendations for revision of the TNM classificationDistinguish Stage I and Stage II RCC according to the tumor size breakpoint of 5.5 cm
Results and interpretation of statistical analysesSee the “Results” and “Discussion” sections in the current article
Limitations of the data presented in the proposalRetrospective analysis. Potential impact of other independent prognostic factors on prognosis
Relation of the proposal to published literature and current practice patternsSee the Discussion section in the current article

Acknowledgements

The authors thank Christine M. Lohse from the Division of Biostatistics, Mayo Clinic (Rochester, MN) for her excellent and kind collaboration with martingale residuals analysis. Receiving this kind of collaboration was a great honor for all the authors.

Ancillary