Multiinstitutional European validation of the 2002 TNM staging system in conventional and papillary localized renal cell carcinoma

Authors


Abstract

BACKGROUND

The current study validated the 2002 edition of the TNM staging system in a multicenter, multinational European series of localized renal cell carcinoma (RCC).

METHODS

The authors analyzed the clinical data of 2217 patients who had undergone radical or partial nephrectomy for localized RCC in 7 urologic centers.

RESULTS

In the current study, 1065 patients (48%) were classified as having pT1a disease, 771 (34.8%) were classified as having pT1b disease, and 381 (17.2%) were classified as having pT2 disease. Tumor histotype was conventional RCC in 1886 patients (85%), papillary in 182 (8.2%) patients, chromophobe in 64 (2.9%) patients, and unclassified in 85 (3.8%) patients. The mean follow-up time was 65.36 ± 52.09 months. The 5 and 10-year disease-specific survival probabilities were 95.3% and 91.4% in patients with pT1a disease, 91.4% and 83.4% in patients with pT1b disease, and 81.6% and 75.2% in patients with pT2 disease (log-rank test P value = 0.0000). The disease-specific survival rates of patients with pT1a RCC were significantly higher than those recorded in patients with pT1b and pT2 RCC. Similarly, the disease-specific survival probabilities of patients with pT1b RCCwere significantly better than those of patients with pT2 RCC. Analyzing the seven series individually, the 2002 TNM staging system provided appropriate stratification for only one series. The 2002 TNM staging system allowed significant stratification of the cancer-related outcomes in the subgroup of patients with conventional RCC but not in those with papillary carcinomas.

CONCLUSIONS

The application of the 2002 TNM staging system in the current multicenter series enabled the authors to demonstrate optimal stratification of patients with localized RCC. Stratifying by tumor histotype, the data coming from the whole group analysis were reconfirmed for clear cell RCC only. Cancer 2005. © 2005 American Cancer Society.

Over the years, the codification of staging systems has been a fundamental process in oncology because of the related prognostic relevance. The TNM staging system proposed by the International Union Against Cancer (UICC) is the most widely used classification of local extension of the primary tumor (pT), lymph node involvement, and the presence of distant metastases. The TNM staging system is a powerful prognostic factor of both disease-specific and disease-free survival probabilities in patients with cancer.

The first edition of the TNM staging system proposed by the UICC in 1968 did not include a classification of renal tumors.1 Renal cell carcinoma (RCC) found a place in the second edition of the TNM staging system, published by the UICC in 19742. The TNM staging system of RCC has been modified several times over the years, with the intent to improve outcome predictability and, consequently, to provide more appropriate data to plan therapy. In the most recent TNM editions, the most common modifications involved the stratification of localized RCC (T1–2N0M0) into subgroups with different cancer-related outcomes. The 1987 version of the staging system proposed for the first time a subclassification of localized RCC into 2 groups (T1N0M0 and T2N0M0) according to the tumor size breakpoint of 2.5 cm.3 In 1997, according to the data of the Surveillance, Epidemiology and End Results(SEER) program, the 5th edition of the TNM staging system settled on 7 cm as the new breakpoint to distinguish between patients with T1 and T2 disease.4 The main drawbacks of both the 1987 and 1997 TNM staging systems were the imbalanced distribution of patients in the two subgroups and the presence of disease-specific survival probabilities, which often resulted in overlapping for both stages.5 The latest edition of the TNM system, published in 2002, introduced a further subclassification of patients with T1N0M0 disease according to the tumor size breakpoint of 4 cm: pT1a (with a dimension ≤ 4 cm) and pT1b tumors (dimenstion ranging from 4–7 cm). That modification aimed at providing the clinicians with better criteria to identify patients suitable for elective nephron-sparing surgery.6, 7 Indeed, this update appeared to keep up with the increase noted in elective nephron-sparing procedures, which paralleled the increase in the diagnoses of incidentally detected, small neoplasms.

The modification proposed in the 2002 TNM edition needs verification of the ability of the staging system to identify groups of patients with significantly different cancer-related outcomes. Recently, Ficarra et al.,8 analyzing a single-center series, reported that the disease-specific survival probabilities of patients classified with pT1b and pT2 RCC were statistically overlapping.

The objective of the current studywas to verify whether the 2002 edition of the TNM staging system stratifies localized RCC into subgroups of patients with significantly different disease-specific survival probabilities using a multicenter, multinational, European dataset.

MATERIALS AND METHODS

We analyzed the clinical data of 2217 patients who had undergone either radical or partial nephrectomy for localized (T1–2N0M0) RCC from 1984–2001 in 7 European urologic centers: Verona, Padua, and Naples (Italy); Rennes, Saint Etienne, and Creteil (France); and Graz (Austria).

Chest X-ray and abdominal computed tomography scans were performed preoperatively in each patient. The selection of the kind of surgical treatment was based on the urologists' choice in each of the institutions involved. In patients undergoing radical nephrectomy, only hilar lymphadenetomy was carried out.

The following clinicopathologic data were extracted from databases of each center: age, gender, Eastern Cooperative Oncology Group (ECOG) performance status, mode of presentation, pathologic TNM stage, pathologic tumor size, tumor histotype, Fuhrman nuclear grade, and follow-up data.

Performance status was assigned according to ECOG criteria, which specify 4 classes: 0, normal activity; 1, restricted in physically strenuous activity but ambulatory; 2, patient bedridden < 50% of the time; and 3, patient completely bedridden.9 With regard to the mode of presentation, patients were clustered in three groups: Group 1 (S1) was comprised of patients with incidentally detected RCC, diagnosed during abdominal imaging studies performed for signs and symptoms unrelated to cancer; Group 2 (S2) was comprised of patients with RCC that was detected because of the presence of local symptoms, such as hematuria, flank pain, or flank mass; Group 3 (S3) was comprised of patients with RCC who were identified after the onset of systemic symptoms such as loss of energy, fatigue, weight loss, fever, and cough.10 With regard to the pathologic stage, all neoplasms were reclassified according to the 2002 TNM staging systems: localized tumors with a pathologic size ≤ 4 cm were classified as pT1a; localized tumors ranging from 4 cm to ≥ 7 cm were identified as pT1b or pT2, respectively.7 The Heidelberg and Fuhrman classifications were used to assign the histologic type and nuclear grade, respectively.11, 12

Data on survival were taken from the clinical files of each center and, when necessary, by contacting the patients' general practitioners or relatives or by searching through death records.

Statistical Analysis

Continuous parametric variables were reported as the mean value ± the standard deviation and range. Continuous nonparametric variables were presented as the median values and interquartile ranges. The analysis of variance and the Kruskal–Wallis tests were used to compare continuous parametric and nonparametric variables, respectively. The Pearson chi-square test was employed to compare categoric variables.

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 (disease-specific survival). The log-rank test was used for comparison of the survival curves and for the univariate analysis. The Cox proportional hazard model was used to identify the clinicopathologic variables, which were capable of independently predicting disease-specific survival. ECOG performance status was not included in the multivariate model because the data were missing for the majority of the patients.

For all statistical analyses, P < 0.05 was statistically significant (a two-sided value). All data were analyzed with the Statistical Package for Social Sciences software, version 12.0 (SPSS Inc., Chicago, IL).

RESULTS

In the current study, 492 patients (22.2%) were treated in Verona, 323 (14.6%) were treated in Padua, 147 (6.6%) were treated in Naples, 374 (16.9%) were treated in Rennes, 266 (12%) were treated in Saint Etienne, 209 (9.4%) were treated in Creteil, and 406 (18.3%) were treated in Graz.

The mean patient age was 59.49 ± 12.44 years (range, 15–89 years). Of the patient sample studied, 1464 patients (66%) were male and 753 (34%) were female. ECOG performance status class was 0 for 1222 patients (55%) and ≥ 1 for 265 patients (12%). ECOG performance status data were not available for 730 patients (33%). For 1340 patients (60.4%), neoplasms were incidentally detected (S1), whereas RCCs were diagnosed after the onset of local (S2) and systemic symptoms in 729 (32.9%) and 99 (4.5%) patients, respectively.

Radical or partial nephrectomies were performed in 1738 (78,4%) and 479 patients (21.6%), respectively.

The median pathologic tumor size was 4.5 cm (interquartile range, 3–6 cm). Of the 2217 patients, 1065 (48%) were classified as having pT1a tumors, 771 (34.8%) were classified as having pT1b tumors, and 381 (17.2%) were classified as having pT2 tumors.

According to the Heidelberg classification, tumor histotype was clear cell RCC in 1886 patients (85%), papillary in 182 patients (8.2%), chromophobe in 64 (2.9%) patients, and unclassified in 85 (3.8%) patients. Fuhrman nuclear grades were assigned by the pathologists of each participating center for 2204 patients (99.6%): Grade 1 was reported in 775 patients (35.2%), Grade 2 in 1097 patients (49.8%), Grade 3 in 308 patients (13.9%), and Grade 4 in 24 patients (1.1%).

Table 1 summarizes the clinicopathologic data of the patients stratified by the 2002 TNM pathologic stage (Table 1). The mean follow-up time was 65.36 ± 52.09 months (range, 2–286 months). The median follow-up time was 54 months (interquartile range, 25–88.5 months).

Table 1. Clinical and Pathologic Characteristics of the 2,217 Patients, Stratified by Pathologic Status
VariablespT1a (n = 1065) (%)pT1b (n = 771) (%)pT2 (n = 381) (%)P value
  1. ECOG: Eastern Cooperative Oncology Group; PS: performance status.

Mean age (yrs)59.61 ± 12.2459.79 ± 12.2858.59 ± 13.230.28
Gender   0.43
 Male718 (67.4)500 (64.8)246 (64.5) 
 Female347 (32.6)271 (35.2)135 (35.5) 
ECOG PS   0.000
 0580 (90)437 (81.3)205 (66.9) 
 ≥164 (10)100 (18.7)101 (33.1) 
Mode of presentation   0.000
 Incidental (S1)769 (74.5)440 (58)131 (34.5) 
 Local symptoms (S2)245 (23.7)277 (36.5)207 (54.6) 
 Systemic symptoms (S3)17 (1.8)41 (5.7)41 (10.9) 
Median pathological size (cm)3 ± 0.845.6 ± 0.829.8 ± 2.250.000
Histotype   0.01
 Conventional919 (89)653 (88.3)314 (84) 
 Papillary80 (7.7)68 (9.1)34 (9.2) 
 Chromophobe29 (2.8)13 (1.7)22 (6) 
 Unclassified5 (0.5)7 (0.9)3 (0.8) 
Fuhrman nuclear grade   0.000
 1468 (44.3)273 (35.7)34 (9) 
 2496 (46.7)361 (47.3)240 (63) 
 393 (8.7)122 (16)93 (24) 
 43 (0.3)7 (1)14 (3.6) 

The 5-year and 10-year disease-specific survival probabilities were, respectively, 95.3% and 91.4%, respectively, for patients with pT1a disease; 91.4% and 83.4%, respectively, for patients with pT1b disease; and 81.6% and 75.2%, respectively, for patients with pT2 disease (log-rank P = 0.0000). The disease-specific survival rates of patients with pT1a RCC were significantly higher than those recorded for patients with pT1b RCC (log-rank P = 0.0000) and pT2 RCC (log-rank P = 0.0000), respectively. Similarly, the disease-specific survival probabilities of patients with pT1b RCC were significantly better than those of patients with pT2 RCC (log-rank P = 0.0002) (Fig. 1).

Figure 1.

Comparison of disease-specific survival probabilities stratified by pathologic status (2002 TNM staging system) for the complete multicenter series. The 5-year and 10-year disease-specific survival probabilities were 95.3% (33 deaths, 479 censored) and 91.4% (44 deaths, 158 censored), respectively, for patients with pT1a tumors; 91.4% (47 deaths, 361 censored) and 83.4% (65 deaths, 137 censored), respectively, for patients with pT1b tumors; and 81.6% (53 deaths, 165 censored) and 75.2% (62 deaths, 86 censored), respectively, for patients with pT2 tumors (log-rank P = 0.0000) (pT1a vs. pT1b, log-rank P = 0.0001; pT1a vs. pT2, log-rank P = 0.0000; and pT1b vs. pT2, log-rank P = 0.0002).

Table 2 reports the data concerning the validation of the 2002 TNM staging system in each participating center. The application of the 2002 TNM staging system allowed statistically significant stratification of the cancer-related outcome in the subgroup of patients with clear cell RCC (log-rank P = 0.0000). In that subset of patients, the 5-year and 10-year disease-specific survival rates were, 95.8% and 91.3%, respectively, for patients with pT1a RCC; 91.6% and 84.1%, respectively, for patients with pT1b RCC; and 80.6% and 74.1%, respectively, for patients with pT2 RCC.Patients with pT1a conventional RCC had significantly higher disease-specific survival rates than patients with pT1b (log-rank P = 0.0001) and pT2 (log-rank P < 0.0001) RCC. Patients with conventional pT1b RCCs demonstrated better disease-specific survival compared with patients with pT2 tumors (log-rank P = 0.0002) (Fig. 2). Conversely, the 2002 TNM staging system did not appropriately stratify the cancer-related outcome of the patients with papillary RCC (log-rank P = 0.08). In that subgroup of patients, the disease-specific survival rates of patients with pT1a tumors overlapped those reported for patients with pT1b tumors (log-rank P = 0.80). A trend close to statistical significance was found when comparing the outcome both of patients with pT1a and pT2 (log-rank P = 0.05) and patients with pT1b and pT2 (log-rank P = 0.06) papillary RCCs (Fig. 3).

Table 2. Five-Year Disease Specific Survival Probability Stratified by the 2002 TNM Pathologic Status in Each Center Participating in the Current Study
Center (no. of patients) 5-yr disease specific survival probability (%)Log- rank P-value
Verona (n = 492) 
pT1a = 97%pT1a vs. pT1b = 0.01
pT1b = 93%pT1a vs. pT2 = 0.003
pT2 = 88%pT1b vs. pT2 = 0.63
Padua (n = 323) 
pT1a = 97%pT1a vs. pT1b = 0.03
pT1b = 91%pT1a vs. pT2 = 0.007
pT2 = 81%pT1b vs. = pT2 = 0.43
Naples (n = 147) 
pT1a = 100%pT1a vs. pT1b = 0.26
pT1b = 96%pT1a vs. pT2 = 0.006
pT2 = 79%pT1b vs. pT2 = 0.006
Rennes (n = 374) 
pT1a = 98%pT1a vs. pT1b = 0.008
pT1b = 91%pT1a vs. pT2 = 0.0000
pT2 = 80%pT1b vs. pT2 = 0.033
Saint Etienne (n = 266) 
pT1a = 98.5%pT1a vs. pT1b = 0.09
pT1b = 95.1%pT1a vs. pT2 = 0.005
pT2 = 88%pT1b vs. pT2 = 0.21
Creteil (n = 209) 
pT1a = 96.6%pT1a vs. pT1b = 0.18
pT1b = 93.4%pT1a vs. pT2 = 0.01
pT2 = 78.4%pT1b vs. pT2 = 0.19
Graz (n = 406) 
pT1a = 95%pT1a vs. pT1b = 0.39
pT1b = 94%pT1a vs. pT2 = 0.001
pT2 = 76%pT1b vs. pT2 = 0.06
Figure 2.

Comparison of disease-specific survival probabilities stratified by pathologic status (2002 TNM staging system) in patients with conventional renal cell carcinoma only. The 5-year and 10-year disease-specific survival probabilities were 95.8% (25 deaths, 419 censored) and 91.3% (36 deaths, 137 censored), respectively, for patients with pT1a tumors; 91.6% (40 deaths, 312 censored) and 84.1% (54 deaths, 115 censored), respectively, for patients with pT1b tumors; and 80.6% (47 deaths, 145 censored) and 74.1% (55 deaths, 73 censored), respectively, for patients with pT2 tumors (log-rank P = 0.0000) (pT1a vs. pT1b, log-rank P = 0.0001; pT1a vs. pT2, log-rank P = 0.0000; and pT1b vs. pT2, log-rank P = 0.0002).

Figure 3.

Comparison of disease-specific survival probabilities stratified by pathologic status (2002 TNM staging system) in patients with papillary renal cell carcinoma only. The 5-year disease-specific survival probabilities were 89.4% (5 deaths, 33 censored) for patients with pT1a tumors; 91% (4 deaths, 36 censored) for patients with pT1b tumors; and 75% (5 deaths, 9 censored) for patients with pT2 tumors (log-rank P = 0.08) (pT1a vs. pT1b, log-rank P = 0.82; pT1a vs. pT2, log-rank P = 0.0501; and pT1b vs. pT2, log-rank P = 0.0569).

Age > 60 years (log-rank P = 0.0005), an ECOG performance status score > 0 (log-rank P < 0.0001), symptomatic tumor presentation (log-rank P < 0.0001), and Fuhrman nuclear grading system (log-rank P < 0.0001) were the variables found to be predictive of disease-specific survival using univariate analysis (Table 3). Using multivariate analysis, the pathologic status of the pT was found to be an independent predictor of disease-specific survival, as were mode of presentation, patient's age, and Fuhrman nuclear grading system (Table 4).

Table 3. 5-Year and 10-Year Disease Specific Survival Univariate Analysis
Variables5-yr disease specific survival (%)10-yr disease specific survival (%)Log-rank P value
  1. ECOG: Eastern Cooperative Oncology Group; PS: performance status.

Age (yrs)  0.0005
 ≤ 6093.788 
 > 6089.182.7 
Gender  0.12
 Male89.490.7 
 Female92.983.5 
ECOG PS  < 0.0001
 095.579.6 
 ≥ 192.470.9 
Mode of presentation  < 0.0001
 Incidental (S1)95.390.9 
 Local symptoms (S2)87.680 
 Systemic symptoms (S3)69.8 0.68
Histotype91.785.7 
 Conventional87.684.4 
 Papillary96.6 < 0.0001
 Chromophobe   
Fuhrman nuclear grade   
 194.391.4 
 291.584.2 
 385.274.4 
 47147.3 
Table 4. Multivariate Analysis Disease-Specific Survival
VariablesCategoriesHR95% CIP value
  1. HR: hazard ratio; 95% CI: 95% confidence interval.

Mode of presentationS1 vs. S2 vs. S32.1231.673–2.693<0.001
Age (yrs)≤ 60 vs. > 601.8111.350–2.429<0.001
TNM, 2002pT1a vs. pT1b vs. pT21.5161.240–1.854<0.001
Nuclear gradeGrade 1 vs. Grade 2 vs. Grades 3–41.3681.093–1.7120.006

DISCUSSION

The data in the current multicentric European study confirmed that the application of the criteria of 2002 allowed stratification of localized RCC in subgroups of patients with different disease-specific survival probabilities. Moreover, the 2002 version of the TNM staging system was able to predict different disease-specific survival outcomes in the subgroup of patients who underwent partial or radical nephrectomy for localized clear cell RCC, but not for localized papillary neoplasms. However, these data were strongly limited by the number of patients with papillary RCC available for the analysis.

In a recently published article, Ficarra et al.8 had highlighted that the 2002 TNM staging system could not stratify the cancer-related outcome of patients with pT1b and pT2 RCC. A possible drawback of that study might have been the limited number of patients and events observed in the pT2 group.

Data from the literature have shown that the percentage of pT2 RCCs range from 6% to 28% in what to our knowledge are all the series published to date.5 In the current multicenter study, this rate was as high as 17.2%. The limited number of patients with pT2N0M0 RCC can be explained by the increasing number of incidentally detected RCC, which was paralleled by an increasing rate of tumors measuring ≤ 4 cm.13

The substratification of localized T1 (T1N0M0) RCC according to the tumor size breakpoint of 4 cm met with the clinical need to identify patients who might be potentially suitable for elective nephron-sparing surgery in the context of the TNM staging system. Over the last decades, the percentage of incidentally detected RCC has increased from 5–13% in the early 1980s to 60% in the most recent series. That issue has been paralleled by an increase in the percentage of localized tumors measuring < 4 cm, with low metastatic potential, suitable for elective nephron-sparing surgery.13, 14 The data published during the last 10 years demonstrated the oncologic efficacy and safety of elective nephron-sparing surgery in patients with a tumor measuring ≤ 4-cm in mean size. The local disease recurrence rates varied from 0% to 7% and the disease-specific survival probabilities ranged from 89% to 100%.15 Moreover, in 1999, Hafez et al.,6 analyzing 485 patients who had undergone elective partial nephrectomy, underscored that the disease-specific survival probabilities were 90% for patients with RCC tumors measuring ≤ 4 cm, 71% for patients with RCC tumors measuring 4–7 cm, and 62% for patients with neoplasms measuring > 7 cm. Nonrandomized comparative studies showed that the 5-year and 10-year disease-specific survival probabilities of patients who had undergone elective partial or radical nephrectomy were overlapping in patients with tumors measuring ≤ 4 cm.16–18 Those figures were the scientific background to support the 2002 TNM staging system update.

Despite all the modifications of the TNM staging system over the last 30 years, the optimal classification of localized RCC remains controversial. The identification of ideal patients suitable for elective nephron-sparing surgery could be biased by tumor localization within the kidney, as well as by mode of presentation, variables that were not included in the TNM staging system. In addition, to our knowledge, few articles published to date have confirmed that elective nephron-sparing surgery might be success in patients whose tumors measure 4–7 cm as well.16, 19 Specifically, Patard et al.20 recently demonstrated that both disease-specific survival and disease recurrence rates were overlapping after either radical or partial nephrectomy in patients with RCCs measuring ≤ 7 cm (T1N0M0). The data presented in the multicentric study of Patard et al., involving 1454 patients with pT1N0M0 RCC who were treated in 7 academic institutions, could make the subclassification of pT1N0M0 RCC unnecessary according to a tumor size breakpoint appropriate for elective nephron-sparing surgery.

Another critical issue in the subclassification of localized RCC according to a tumor size breakpoint is the identification of an ideal cutoff value to discriminate groups of patients with different cancer-related outcomes. Recent articles analyzing series of patients who had undergone only radical nephrectomy identified ideal breakpoints ranging from 5 cm to 5.5 cm.21–25 Similar figures were proposed by analyses of mixed series of radical and partial nephrectomies.26–28 The currently available literature data would support a further TNM redefinition according to such breakpoints.29

The application of the 2002 TNM staging system in a multicenter series of 2217 patients from 7 European academic institutions enabled us to demonstrate the optimal stratification of patients with localized RCC with regard to disease-specific survival probabilities. Stratifying by tumor histotype, the data coming from the entire group analysis were reconfirmed for clear cell RCC only. The subclassification of localized RCCs into three subgroups makes the outcome stratification of patients within single-center series troublesome, most likely because of the numerousness of the cohorts.

Ancillary