Characteristics and prognosis of chromophobe non-metastatic renal cell carcinoma: A multicenter study

Authors


Seok-Soo Byun md phd, Department of Urology, Seoul National University Bundang Hospital, 166, Gumi-ro, Bundang-gu, Seongnam, Kyunggi-do 463-707, Republic of Korea. Email: ssbyun@snubh.org

Abstract

Objectives:  To analyze the characteristics and the prognostic significance of chromophobe renal cell carcinomas (chRCC).

Methods:  Data about 2981 patients with non-metastatic renal cell carcinomas (RCC) at the time of surgery were retrospectively collected from 26 institutions between 1998 and 2008. All patients had undergone partial or radical nephrectomies. Of the 2981 patients, 2602 patients with conventional RCC (cRCC) and 148 with chRCC were studied. Clinical and pathological parameters were determined in all patients. Recurrence-free survival (RFS) and cancer-specific survival (CSS) were assessed.

Results:  Patients with chRCC differed significantly from those with cRCC on the following parameters: younger age (P = 0.026), greater female ratio (P < 0.001), and larger tumor diameter (P < 0.001). Both groups were alike with respect to body mass index (P = 0.943), Eastern Cooperative Oncology Group performance status (P = 0.163), T stage (P = 0.375), and Fuhrman's grade (P = 0.134). The 5-year RFS rates in patients with chRCC and cRCC were 82.7% and 83.3%, respectively (P = 0.762). The 5-year CSS rates in patients with chRCC and cRCC were 88.8% and 92.2%, respectively (P = 0.980). Both groups showed equivalent oncological outcomes in terms of RFS and CSS for cases stratified by T stage and Fuhrman's grade. In multivariate analysis, the histological subtype was not retained as an independent prognostic variable (RFS: P = 0.893; CSS: P = 0.729).

Conclusions:  Despite being significantly different from cRCC in terms of several clinical and pathological parameters, chRCC shows equivalent oncological outcomes.

Introduction

Renal cell carcinomas (RCC) account for 3.9% of malignancies in humans, and are one of the most fatal genitourinary malignancies.1 Although surgery remains the mainstay of treatment, about one-third of all patients have metastatic disease at the time of initial presentation and 20–40% of patients demonstrate local recurrence or distant metastasis after nephrectomy.2–4 Therefore, it is important to know the prognostic factors for RCC as an aid for optimal clinical decision-making and patient counseling.

A wide variety of prognostic factors have been shown to have an association with patient outcome. However, by common consent, independent prognostic factors in patients with RCC are TNM stage, Fuhrman's grade, tumor size, and patient performance status.5–8

The histological subtypes of RCC include conventional clear cell, papillary, chromophobe, collecting duct, and unclassified, according to the Heidelberg classification.9 Molecular and genetic studies have shown that RCC are not a single entity and have genetic differences among the histological subtypes.10–12 However, although the prognostic significance of the histological subtypes of RCC has been investigated in several large institutional and international studies, the prognostic value of the histological subtypes remains controversial. Also, far from the prognostic value of histological subtypes, the distribution varies between studies.13

Chromophobe RCC (chRCC) are the third most prevalent form of RCC after conventional (cRCC) and papillary (pRCC) RCC. In general, it is a common belief that chRCC have a more favorable prognosis than cRCC. However, studies for clinical outcomes of chRCC are relatively rare compared with cRCC or pRCC, and conflicting published data exist regarding the independent prognostic value of chRCC.14–17

In the current study, our objective was to analyze the characteristics and prognostic values of chRCC based on a large, multicenter experience.

Methods

Approval of the study was obtained from the institutional review board of Seoul National University Bundang Hospital (IRB No. B-0902-070-103). A total of 2981 patients with sporadic and non-metastatic RCC (TxN0M0) at the time of surgery were collected from 26 institutions between 1998 and 2008 in the Republic of Korea. The patients' clinical and pathological information were retrospectively reviewed. All patients had undergone partial or radical nephrectomies with curative intent.

The following parameters were noted in each patient: pathological stage, tumor grade, histological subtype, recurrence-free survival (RFS), and cancer-specific survival (CSS). Other parameters analyzed included age, sex, body mass index (BMI), general health status, tumor size, tumor laterality, type of surgery, and lymph node dissection.

Pathological staging was performed using the International Union Against Cancer (UICC) and the American Joint Commission on Cancer 2002 TNM classification system.18 Fuhrman's nuclear grade was used for tumor grading.19 Histological subtyping was performed according to the 2004 World Health Organization classification based on the Heidelberg classification.20 Tumor size was determined from the pathological specimen as the greatest diameter in centimeters. To determine pathological features, urological pathologists reviewed all specimens at each institution. The general health status was measured using the Eastern Cooperative Oncology Group performance status (ECOG PS) score. Survival was determined from the date of surgery to the date of last imaging modalities, such as computed tomography. The postoperative follow-up scheme was every 3 months for 6 months, every 6 months for 3 years, and then yearly in most patients.

The correlation between clinical and pathological parameters of subject groups was compared using an independent t-test for quantitative variables and a χ2-test for qualitative variables, respectively.

The RFS rate (RFSR) and CSS rate (CSSR) were determined using the Kaplan–Meier method stratified by cRCC and chRCC, and the log–rank test was used to compare groups. The prognostic values of clinical and pathological parameters were evaluated using Cox proportional hazards models. spss 17.0 (spss, Chicago, IL, USA) was used for all statistical assessments. All P-values were two-sided, and P < 0.05 was considered significant.

Results

Of the 2981 patients, 2602 (87.3%) with cRCC and 148 (5.0%) with chRCC were investigated. The excluded histological subtypes included the following: pRCC, 168 (5.6%); collecting duct RCC, five (0.2%); and unclassified RCC, 58 (1.9%).

The mean age of all patients was 54.9 years (range, 12–90 years). There were 1877 men (68.2%) and 873 women (31.7%). The mean BMI of all patients was 24.1 kg/m2 (range, 12.0–44.8 kg/m2). The mean tumor size of all patients was 4.4 cm (range, 0.6–24.0 cm). The T stage was T1a in 1423 (51.7%), T1b in 682 (24.8%), T2 in 276 (10.0%), T3a in 275 (10.0%), T3b in 81 (2.9%), T3c in seven (0.3%), and T4 in six patients (0.2%). Fuhrman's grade was grade 1 in 277 (10.1%), grade 2 in 1545 (56.2%), grade 3 in 836 (30.4%), and grade 4 in 92 patients (3.3%). All patients were followed up for at least 6 months and the mean duration of follow up for all patients was 34.8 months (median, 28.0 months; range, 6–118 months).

Characteristics of the patients and tumors are shown in Table 1. Patients with chRCC differed significantly from those with cRCC on several parameters. Patients with chRCC tended to be younger (P = 0.026) and had a greater female ratio (P < 0.001). The mean tumor size was larger in chRCC compared to cRCC (P < 0.001). Conversely, there was no association with respect to BMI (P = 0.943), ECOG PS (P = 0.163), T stage (P = 0.375), or Fuhrman's grade (P = 0.134). Also, there was no association with respect to tumor laterality (P = 0.566), type of surgery (P = 0.911), or lymph node dissection (0.273).

Table 1.  Association of different clinical and pathological variables with histological subtype
VariablechRCCcRCCP-value
  1. BMI, body mass index; chRCC, chromophobe renal cell carcinoma; cRCC, conventional renal cell carcinoma; ECOG PS, Eastern Cooperative Oncology Group performance status; SD, standard deviation.

No. subjects1482602 
Follow up duration, mean (SD, range)35.2 (23.7, 6–101) months34.8 (22.6, 6–118) months0.849
Age   
 (1) Continuous, mean (SD, range)52.7 (11.2, 29–76) years55.0 (12.4, 12–90) years0.026
 (2) >55 years61 (41.2%)1291 (49.6%)0.047
Sex  <0.001
 Male71 (48.0%)1806 (69.4%)
 Female77 (52.0%)796 (30.6%)
BMI   
 (1) Continuous, mean (SD, range)24.1 (3.7, 16.9–44.8) kg/m224.1 (3.5, 12.0–41.8) kg/m20.943
 (2) ≥30 kg/m28 (5.7%)112 (4.3%)0.424
ECOG PS ≥133 (22.3%)739 (28.4%)0.163
Tumor size   
 (1) Continuous, mean (SD, range)5.6 ± 3.8 (1.2–24.0) cm4.4 ± 2.7 (0.6–20.0) cm<0.001
 (2) Category  <0.001
  <4 cm58 (39.2%)1408 (54.1%)
  4–7 cm49 (33.1%)801 (30.8%)
  7 cm ≤41 (27.7%)393 (15.1%)
Side  0.566
 Left72 (48.9%)1179 (45.3%)
 Right76 (51.1%)1413 (54.3%)
 Bilateral0 (0%)10 (0.4%)
Type of surgery  0.911
 Radical nephrectomy120 (81.1%)2100 (80.7%)
 Partial nephrectomy28 (18.9%)502 (19.3%)
T stage (TNM 2002)  0.375
 pT1102 (68.9%)2003 (77.0%)
 pT230 (20.3%)246 (9.5%)
 pT3 and 416 (10.8%)353 (13.6%)
Fuhrman's grade  0.134
 G111 (7.4%)266 (10.2%)
 G280 (54.1%)1465 (56.3%)
 G3 and 457 (38.5%)871 (33.5%)

During the follow-up period, 14 patients (9.5%) with chRCC and 242 patients (9.3%) with cRCC had recurrences. The 5-year RFSR were 82.7% in patients with chRCC and 83.3% in patients with cRCC, and there was no statistical difference between the two groups (P = 0.762, Fig. 1a). When patients were stratified according to T stage, the 5-year RFSR for chRCC and cRCC with T1 were 91.1% and 86.7% (P = 0.744), those with T2 were 80.3% and 69.6% (P = 0.532), and those with T3–4 were 53.2% and 75.1% (P = 0.998), respectively. When patients were stratified according to Fuhrman's grade, the 5-year RFSR for chRCC and cRCC with grade 1 were 100% and 89.3% (P = 0.433), those with grade 2 were 88.1% and 85.3% (0.669), and those with grade 3–4 were 71.9% and 78.3% (P = 0.869), respectively. No significant RFS difference remained among the two groups stratified according to T stage or Fuhrman's grade.

Figure 1.

Kaplan–Meier curve for (a) recurrence-free and (b) cancer-specific survival for patients with non-metastatic renal cell carcinoma according to histological subtype. (inline image) cRCC, conventional renal cell carcinoma; (inline image) chRCC, chromophobe renal cell carcinoma.

During the follow-up period, six patients (4.1%) with chRCC and 102 patients (3.9%) with cRCC died of tumor-related causes. The 5-year CSSR were 88.8% in patients with chRCC and 92.2% in patients with cRCC, and there was no statistical difference between the two groups, as with RFSR (P = 0.980, Fig. 1b). When patients were stratified according to T stage, the 5-year CSSR for chRCC and cRCC with T1 were 93.3% and 94.3% (P = 0.434), those with T2 were 81.8% and 85.6% (P = 0.578), and those with T3–4 were 85.9% and 85.8% (P = 0.513), respectively. When patients were stratified according to Fuhrman's grade, the 5-year CSSR for chRCC and cRCC with grade 1 were 100% and 94.5% (P = 0.623), those with grade 2 were 90.7% and 92.9% (P = 0.869), and those with grade 3–4 were 84.1% and 88.0% (P = 0.954), respectively. Similar to RFSR, no significant CSS difference remained among the two groups stratified according to T stage or Fuhrman's grade.

In the Cox proportional hazards model analyses, which included age, sex, BMI, ECOG PS, tumor size, T stage, Fuhrman's grade, and histological subtype, independent prognostic factors of RFS were age, sex, BMI, tumor size, T stage, and Fuhrman's grade. However, the histological subtype did not remain an independent prognostic factor of RFS in both univariate and multivariate analyses (chRCC vs cRCC, P = 0.762 and 0.893 in univariate and multivariate analyses, respectively; Table 2).

Table 2.  Univariate and multivariate analyses predicting probability of recurrence-free survival
VariablesUnivariate analysisMultivariate analysis
HR95%CIP-valueHR95%CIP-value
  1. BMI, body mass index; chRCC, chromophobe renal cell carcinoma; CI, confidence interval; cRCC, conventional renal cell carcinoma; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio.

Age (>55 years)1.8891.468–2.430<0.0011.8041.374–2.369<0.001
Sex (female vs male)0.7760.589–1.0220.0720.7390.547–0.9970.048
BMI (≥30 kg/m2)0.3030.097–0.9460.0400.2150.053–0.8680.031
ECOG PS (≥1)1.4261.091–1.8650.0091.1550.870–1.5330.318
Tumor size  <0.001  <0.001
 4–7 cm vs <4 cm2.6011.923–3.518<0.0012.5441.833–3.531<0.001
 ≥7 cm vs <4 cm3.7312.674–5.204<0.0012.6941.641–4.422<0.001
T stage (TNM 2002)  <0.001  <0.001
 T2 vs T11.8261.265–2.6340.0011.0740.597–1.9290.812
 T3–4 vs T12.8532.127–3.827<0.0012.0331.392–2.969<0.001
Fuhrman's grade  <0.001  0.007
 G2 vs G11.3230.803–2.1800.2721.0880.629–1.8810.764
 G3,4 vs G12.4441.477–4.0450.0011.6610.952–2.8970.074
Histological subtype      
 chRCC vs cRCC1.0870.634–1.8630.7621.0400.587–1.8410.893

Independent prognostic factors of CSS in Cox proportional hazards model analyses were age, sex, T stage, and Fuhrman's grade. However, the histological subtype was not an independent prognostic factor of CSS, as with the analysis for RFS (chRCC vs cRCC, P = 0.979 and 0.729 in univariate and multivariate analyses, respectively; Table 3).

Table 3.  Univariate and multivariate analyses predicting probability of cancer-specific survival
VariablesUnivariate analysisMultivariate analysis
HR95%CIP-valueHR95%CIP-value
  1. BMI, body mass index; chRCC, chromophobe renal cell carcinoma; CI, confidence interval; cRCC, conventional renal cell carcinoma; ECOG PS, Eastern Cooperative Oncology Group performance status; HR, hazard ratio.

Age (>55 years)2.5231.687–3.772<0.0012.6381.694–4.107<0.001
Sex (female vs male)0.5160.320–0.8310.0060.4870.290–0.8180.007
BMI (≥30 kg/m2)0.5500.136–2.2330.4030.5990.146–2.4560.477
ECOG PS (≥1)1.4700.978–2.2090.0641.1590.751–1.7890.505
Tumor size  <0.001  0.098
 4–7 cm vs <4 cm1.8531.153–2.9760.0111.7481.050–2.9100.032
 ≥7 cm vs <4 cm3.5412.171–5.777<0.0011.4380.711–2.9110.312
T stage (TNM 2002)  <0.001  <0.001
 T2 vs T12.8801.717–4.831<0.0012.2980.992–5.3230.052
 T3,4 vs T14.8803.192–7.459<0.0013.8442.288–6.456<0.001
Fuhrman's grade  <0.001  0.007
 G2 vs G11.0350.487–2.2010.9290.8720.367–2.0700.756
 G3,4 vs G12.6871.277–5.6550.0091.7290.729–4.1030.214
Histological subtype      
 chRCC vs cRCC1.0110.444–2.3040.9791.1610.499–2.6980.729

Discussion

It is known that the different histological subtypes of RCC have different morphological, genetic, and pathological characteristics.10–12 However, the prognostic value of the histological subtypes remains controversial, and the distribution of histological subtypes varies between studies, with the conventional subtype ranging from 63% to 88.7%, the papillary type from 7.3% to 18.5%, and the chromophobe type from 2.1% to 6.2%.13 In our study, the distributions of each histological subtype were the conventional type in 2602 (87.3%) and the chromophobe type in 148 (5.0%) patients. These proportions were similar to those reported in previous studies.2,13

Previous studies have primarily analyzed the prognostic value using three main subtypes (cRCC, pRCC, and chRCC), including lymph node and/or distant metastasis. Recently, two distinct molecular sporadic classes of pRCC have been identified,21 and are associated with different clinical outcomes.13,22 Unfortunately, as our study was a retrospective design, it was difficult to distinguish one class from the other. Accordingly, our study included only two subtypes, including conventional and chromophobe types. It is also known that chRCC are a heterogeneous group, including classic type (containing >80% pale cells), eosinophilic type (containing >80% eosinophilic cells), and mixed type (containing a variable admixture of pale and eosinophilic cells).23 However, as chRCC are relatively rare, the number of patients with chRCC was not sufficient to analyze prognostic value of distinct classes. Similarly, as a very low prevalence of patients with chRCC had lymph node and/or distant metastasis, these patients were excluded.

Generally, chRCC are suggested to have better clinical behavior than cRCC, although the clinical behavior of chRCC has not been adequately defined.2 From the viewpoint of tumor characteristics, most previous studies have shown that chRCC present at a lower stage and grade than cRCC.2,15,16 However, some recent studies did not confirm this tendency. Kim et al.24 reported that chRCC are diagnosed at an advanced stage and Fuhrman's grade compared with cRCC in 795 cases of RCC. Capitanio et al.25 reported that there was a tendency for chRCC to present at a higher grade than cRCC in 11 618 patients treated with nephrectomies for RCC. Another study (Ku et al., unpubl. data, 2009) could not confirm the tendency for chRCC to present at a lower grade compared with cRCC in 657 patients treated with surgery. Our study also showed that there was no significant difference in the distribution of both stage and grade between chRCC and cRCC, although the chRCC tended to present at a lower stage and higher grade than cRCC. In addition, patients with chRCC were a younger age (P = 0.026), had a greater female ratio (P < 0.001), and larger tumor size (P < 0.001) than those with cRCC. The survival of patients with T3 and T4 seemed to be better compared to previous studies,2,15 and was probably due to the relatively short duration of follow up and extremely small number of patients with T4.

There are few studies analyzing over 100 patients with chRCC, including multivariate analyses. Patard et al.15 published a large study analyzing 4063 patients with RCC, including 103 with chRCC. They found that patients with chRCC had a better outcome compared with cRCC in cases of localized or high-grade disease. However, based on multivariate analysis, the histological subtype was not an independent prognostic variable. Klatte et al.17 studied 124 patients with chRCC who underwent nephrectomies. They found that chRCC led to a better survival rate compared with cRCC. However, when adjusted for 20 patients with metastatic disease, this survival difference was lost. Moreover, based on multivariate analysis, the histological subtype was not an independent prognostic factor of CSS. Similarly, in our study the CSSR of non-metastatic chRCC did not have a significant difference compared with non-metastatic cRCC. When stratified by T stage and Fuhrman's grade, both groups had comparable CSSR. And, based on univariate and multivariate analyses, histological subtypes were not an independent prognostic factor. In addition, we analyzed RFSR using the same methods, and found that the RFSR of chRCC did not have a significant difference compared with cRCC. More recently, Capitanio et al.25 published a large multicenter study using nine Surveillance Epidemiology and End Results (SEER) registries involving 11 618 patients, including 195 patients with chRCC. Based on a multivariate analysis of CSS, histological subtype was an independent predictor of CSS. However, in this study, the proportion of chRCC (1.6%) was smaller than most series and the mean duration of follow up of chRCC (17.2 months) was significantly shorter than cRCC (54.5 months). Therefore, it is reasonable to assume that statistical errors existed in the results. Moreover, on subgroup analysis, there was no statistical difference for the prognostic value between chRCC and cRCC.

To the best of our knowledge, there are only three studies in which chRCC were an independent prognostic factor compared with cRCC in multivariate analysis among series analyzing over 100 patients with chRCC. Beck et al.14 analyzed 1057 patients with non-metastatic RCC undergoing surgery, including 106 patients with chRCC. They reported that chRCC were associated with an improved RFS and multivariate analysis revealed chRCC as a significant variable for disease progression. Recently, Teloken et al.16 published a large, single-center study, analyzing 1863 patients with localized RCC, including 220 with chRCC. They evaluated the relationship between tumor histology and outcome, defined as metastasis or death from disease, and chRCC were significantly associated with better outcome in univariate and multivariate analyses. In both studies, there was a limitation that nuclear grade, such as Fuhrman's grade, was not included as a variable. Certainly, the role of the Fuhrman's grade for non-cRCC, such as chRCC, is controversial.26 However, because most studies regarding the role of Fuhrman's grade for chRCC have a limit due to the small size of data and Fuhrman's grade for cRCC has been validated widely, we believe that the role of Fuhrman's grade for chRCC will be comparable with that for cRCC if further evaluations are carried out in the future. It is noted that neither of the studies included Fuhrman's grade as a variable, although histological subtype was an independent prognostic factor in multivariate analysis. Therefore, in those aforementioned studies, it was possible that exclusion of Fuhrman's grade influenced the results of multivariate analyses. In this regard, our analyses included Fuhrman's grade as a variable that was an independent prognostic factor of both RFS and CSS in multivariate Cox models. More recently, in a series of 3062 patients from the Mayo Clinic, Leibovich et al.27 reported that histological subtype was an independent predictor of progression to distant metastasis and CSS in patients with RCC. However, because these authors only analyzed prognostic difference between cRCC and non-cRCC, the prognostic value of chRCC is unclear. Also, ECOG PS was not reported in more than 800 patients.

The other prognostic factors of RFS were age, sex, BMI, tumor size, and T stage; and the other prognostic factors of CSS were age, sex, and T stage. These have been prognostic factors of RCC in previous studies.15,25,27,28 However, the chRCC compared with cRCC was not an independent prognostic factor for RFS and CSS in multivariate analyses.

Recently, hybrid RCC, containing histopathological features of chRCC and oncocytomas, have been described, though the true frequency and clinical courses of these remain unclear.29 Also, although molecular prognostic markers, such as Ki-67, are not currently recommended for routine clinical use,30 molecular markers may better allow the identification of patients at high risk and affect the clinical outcomes. RCC with sarcomatoid features or necrosis are regarded as a potential highly aggressive variant.31 However, most previous studies, including our study, did not mention these as a variable, which may affect the prognostic value of chRCC. Nevertheless, our study had one of the largest scales of all studies to date regarding the prognostic value of non-metastatic chRCC, and analyzed the most widely accepted independent factors of prognosis of non-metastatic RCC, including TNM stage, Fuhrman's grade, tumor size, and patient performance status,5–8 unlike most previous series.

There were a few limitations to this study. First, our study was a retrospective design. Second, this study lacked a central pathological review. However, in general practice, the vast majority of specimens are not subjected to central review. Therefore, our study is reflective of a real-life phenomenon at each individual institution. Also, at each institution, urological pathologists reviewed all specimens. Nevertheless, it is likely that the lack of central review resulted in misclassifications or misdiagnoses due to interobserver variability. Finally, the duration of follow up was relatively shorter than previous studies.

In summary, our results suggest that the chromophobe histological subtype of non-metastatic RCC was significantly different from the conventional subtype in several clinical and pathological aspects, such as age, sex, and tumor diameter distribution. However, the prognosis of chRCC was comparable to that of conventional clear cell carcinomas.

Acknowledgments

The authors wish to thank Yunhee Choi, PhD, and Sue K. Park, MD, PhD (Division of Epidemiology, Medical Research Collaborating Center Seoul National University Hospital, Seoul, Republic of Korea) for their statistical advice.

Ancillary