Prognostic impact of extensive parenchymal invasion pattern in pT3 renal pelvic transitional cell carcinoma




Pathologic T3 renal pelvic transitional cell carcinoma exhibits various patterns of invasion. The authors investigated the prognostic impact of three patterns of invasion of pT3 renal pelvic transitional cell carcinoma.


Of 212 patients who underwent surgery for renal pelvic transitional cell carcinoma, 70 with pT3 disease were eligible for the main analyses. The candidate predictors of prognosis included patient age, gender, lesion laterality, tumor grade, perioperative cisplatin-based systemic chemotherapy, lymph node involvement, vascular involvement, and patterns of invasion. Invasion patterns were classified as fat invasion, ductal involvement, or parenchymal invasion.


Mean postoperative followup was 33.5 months (range, 1–136 months). On univariate analysis, gender, lymph node involvement, vascular involvement, and extensive parenchymal invasion each had a significant impact on the cause specific survival rate. A multivariate analysis using Cox stepwise regression revealed that extensive parenchymal involvement was the strongest prognostic predictor (P = 0.0004, hazard ratio = 5.59). Lymph node involvement (P = 0.0175, hazard ratio = 3.14) and gender (P = 0.0361, hazard ratio = 2.42) were other weaker predictors. Statistically, pT3 disease without extensive parenchymal invasion had a prognosis similar to that of lower stage disease, and pT3 disease with extensive parenchymal invasion had a prognosis similar to that of pT4 disease.


Extensive parenchymal invasion has a strong prognostic impact in renal pelvic transitional cell carcinoma. pT3 disease should be subclassified into two separate entities, that with and that without extensive parenchymal invasion, in view of prognosis. Cancer 2002;94:3150–6. © 2002 American Cancer Society.

DOI 10.1002/cncr.10609

Carcinoma of the upper urinary tract, including renal pelvic tumors and ureteral tumors; accounts for only 2–4% of all urothelial tumors.1 Only a few large series have reported the natural history of these tumors because of their rarity.2–6 Although it has been generally recognized that the prognosis of patients worsens with increased pathologic stage,2–6 T3/ pT3 stage renal pelvic carcinoma, as classified according to the TNM/pTNM system of the International Union Against Cancer,7 includes heterogeneous diseases. Although we previously reported a study on the prognostic value of the patterns of invasion of pT3 renal pelvic transitional cell carcinoma,8 to our knowledge this is the only report regarding this issue.

In the current study, we expanded the study population of pT3 renal pelvic transitional cell carcinoma to investigate the prognostic impact of patterns of invasion of pT3 renal pelvic transitional cell carcinoma and other candidate predictors of prognosis, such as tumor grade, vascular involvement, and lymph node involvement.


A total of 212 patients with renal pelvic carcinoma underwent nephroureterectomy from May 1974 to July 2000 in three Japanese institutes (96 in Kyoto University, 60 in the National Cancer Center Central Hospital, and 56 in Kurashiki Central Hospital). Depth of local invasion was pathologically classified according to the TNM pathologic staging system of the International Union Against Cancer.7 Of the 212 patients, 49 (23%) were diagnosed with pTa or pTis disease, 41 (19%) pT1, 17 (8%) pT2, 89 (42%) pT3, and 16 (8%) pT4.

Of the 89 cases of pT3 disease, pathologic specimens were not available or were incomplete in 19, mainly due to a lack of the lymph node status. Therefore, only the remaining 70 patients were eligible for the main analyses in the current study. The 47 (67%) men and 23 (33%) women ranged from 42 to 90 years old (median, 69 years; mean, 68.5 years ± 11.3 standard deviation). Postoperative followup ranged from 1 to 136 months (median, 18.5 months; mean, 33.5 months ± 33.9). Thirty six (51%) had right-sided and 34 (49%) left-sided lesions. Twenty seven (39%) patients underwent cisplatin-based systemic chemotherapy perioperatively. Thirty eight (54%) were alive at final followup, 25 (36%) had died of cancer, and 7 (10%) had died of other causes.

Pathologic specimens, which were fixed in 20% buffered formalin, embedded in paraffin, and stained with hematoxylin and eosin, were microscopically reviewed by one clinical pathologist (T.O.), who was blind to the clinical outcome of the patients. All tumors were transitional cell carcinoma, although three patients had minor squamous cell carcinoma components, one had a minor adenocarcinoma component, and one had a minor sarcomatoid component as well. Tumor grade was determined by the histologic classification of bladder carcinoma in the World Health Organization classification system.9 Along with tumor involvement of the vasculature and/or lymph nodes, the patterns of local tumor invasion were carefully examined. We modified the classification system reported previously8 and categorized the patterns of tumor involvement as follows: 1) spread into peripelvic fat (fat invasion); 2) intraductal involvement within the renal ducts (ductal involvement); or 3) direct invasion into the renal parenchyma with degradation of basement membrane (parenchymal invasion; Fig. 1A, B, and C). Parenchymal invasion was further subclassified into two categories: microscopic involvement, which indicated parenchymal invasion to a depth of less than 5 mm from the basement membrane, and extensive involvement, which indicated deeper parenchymal invasion. Generally, microscopic parenchymal involvement could be detected only by microscopic examination, while extensive parenchymal involvement was easily discerned from the gross appearance of the specimen. The distribution of pathologic variables is shown in Table 1. The pathology reviewer could not distinguish vessels from lymphatic ducts without Victoria blue or elastic van Gieson staining. Therefore, lymphatic involvement and vessel involvement were combined as vascular involvement.

Figure 1.

Microscopic view of patterns of involvement. Reduced from ×40 (upper column) and ×100 (lower column). A) Fat invasion. Cancer cells invade the perirenal sinus fat tissue (arrows). Arrowheads indicate carcinoma cells on the mucosa surface. B) Ductal involvement. Arrows indicate cancer cells spreading in the collecting ducts. C) Parenchymal invasion. Cancer cells directly invade the renal parenchyma (arrows). Arrowheads indicate glomuruli.

Table 1. Distribution of Pathologic Variables
VariablesPatient no.%
Tumor grade  
Vascular involvement  
Ductal involvement  
Fat invasion  
Parenchymal invasion  

Disease-specific survival was calculated with the Kaplan-Meier method, and statistical analyses were performed using the log-rank test. The Cox proportional hazards model was used for multivariate analysis of parameters, and a stepwise model was used to select independent prognostic factors. P values less than 0.05 were considered statistically significant.


Univariate Analysis

Table 2 shows the results of univariate analyses. Patient age, laterality of disease, tumor grade, fat invasion, ductal involvement, and perioperative systemic chemotherapy had no impact on the cause specific survival rate. On the other hand, gender, vascular involvement, lymph node involvement, and parenchymal invasion were each significantly associated with cause specific survival (Figs. 2 A–D). Patients with extensive parenchymal invasion had a significantly poorer survival rate than those without parenchymal invasion and those with microscopic parenchymal invasion. However, no significant difference in this parameter was observed between those without parenchymal invasion and those with microscopic invasion (data not shown).

Table 2. Univariate Analyses Using Log-rank Test
  Cause specific survival rate (%)P value
1 year3 year
Perioperative chemotherapyYes75.760.40.6714
Tumor grade1 and 284.661.50.1700
Vascular involvementYes92.079.50.0041
Lymph node involvementYes91.272.7<0.0001
Ductal involvementYes79.167.00.0968
Fat invasionYes75.555.60.2463
Parenchymal invasionNo and microscopic Extensive90.776.4<0.0001
Figure 2.

Cause specific survival. A) By gender. B) By vascular involvement status. C) By lymph node involvement status. D) By pattern of parenchymal invasion.

Multivariate Analysis

To determine the most informative combination of independent prognostic factors, variables identified as statistically significant in predicting prognosis, including gender, vascular involvement, lymph node involvement, and patterns of parenchymal invasion, were subjected to multivariate analysis using the stepwise Cox proportional hazards model (Table 3). On this analysis, extensive parenchymal involvement (P = 0.0004), lymph node involvement (P = 0.0175), and gender (P = 0.0361) remained independently significant as prognostic predictors. The hazard ratios of extensive parenchymal invasion, lymph node involvement, and gender were 5.59, 3.14, and 2.42, respectively. Based on these results, female patients with extensive parenchymal invasion and pathologic lymph node metastasis had a risk of cancer death increased by a factor of 42.5 compared with male patients without lymph node involvement or extensive parenchymal invasion.

Table 3. Multivariate Analysis Using Stepwise Cox Hazard Model
VariableP valueChi squareHazard ratioCoefficientSD
  1. SD: standard deviation, NS: not significant.

Parenchymal invasion0.000412.485.591.7180.486
Lymph node involvement0.01755.653.141.1460.482
Vascular involvementNS    

Comparison with Patients with Other Stages of Disease

Based on the 212 patients studied, the actuarial one year cause-specific survival rates in the current series were 100%, 100%, 100%, 93%, 54%; and 48%; three year survival rates were 84%, 93%, 92%, 81%, 25%, and 8%; and five year survival rates were 80%, 88%, 73%, 77%, 25%, and 8%, respectively, in pTa or pTis, pT1, pT2, pT3 without extensive parenchymal invasion, pT3 with extensive parenchymal invasion, and pT4 patients (Fig. 3). Survival rates of the two subgroups of pT3 disease were thus positioned between pT2 or lower stage and pT4. However, pT3 disease without extensive parenchymal invasion was similar in prognosis to pT2 or lower stage disease, and there was no difference in cause specific survival between patients with pT2 or lower stage disease and those with pT3 disease without extensive parenchymal invasion (Table 4).

Figure 3.

Cause specific survival by pathologic stage.

Table 4. Differences in Prognosis Among Stages. P values Using Log-rank Test
 pT4pT3 (+)pT3 (−)pT2pT1
  1. pT3 (−): pT3 disease without extensive parenchymal invasion.

  2. pT3 (+): pT3 disease with extensive parenchymal invasion.

pTa (pTis)<0.0001<0.00010.42240.83570.3444
pT3 (−)<0.0001<0.0001   
pT3 (+)0.5001    

Furthermore, pT3 disease with extensive parenchymal invasion was similar in prognosis to pT4 disease, and there was no prognostic difference between the patients with pT3 disease with extensive parenchymal invasion and those with pT4 disease (Table 4).


According to the pTNM classification system of renal pelvic malignancies, pT3 disease is defined as “invasion beyond muscularis into peripelvic fat or renal parenchyma.”7 There are thus two potential sites of invasion, peripelvic fat and renal parenchyma. Renal parenchymal involvement can be further divided into two types, such as prostatic tumor involvement by bladder carcinoma. If a tumor does not destroy the basement membrane, it spreads along the collecting ducts but does not actually invade the parenchyma. Conversely, with degradation of the basement membrane, a tumor can directly invade the renal parenchyma.

With respect to prostatic involvement by bladder carcinoma, several investigators have reported that stromal invasion is an independent predictor for poor prognosis.10–13 Although a similar finding for renal pelvic carcinoma was obtained in our previous study,8 that series of 21 patients was too small to conclude that extensive parenchymal invasion with or without ductal involvement had a negative impact on prognosis. The current study was based on a relatively large series of 70 patients with pT3 disease among a total of 212 patients with surgically-resected renal pelvic urothelial carcinoma, despite the rarity of this disease. The current study reconfirmed the finding that extensive parenchymal invasion had a negative impact on prognosis. Furthermore, the study showed that extensive parenchymal invasion was the strongest predictor of prognosis among several candidates examined. Other patterns of local tumor involvement, i.e., fat invasion and ductal involvement, had no impact on prognosis.

One of the important findings of the current study is that the prognostic impact of extensive parenchymal invasion surpassed even that of lymph node involvement (pN status). While lymph node involvement is considered to present an advanced state of metastatic potential via lymphatic channel, extensive parenchymal invasion might inevitably involve both routes of metastasis, i.e., lymphatic and venous. Conversely, microscopic parenchymal invasion, which exhibited no difference in prognostic impact from lack of parenchymal invasion, might involve these vasculatures only minimally.

Ductal involvement, which appears to invade the renal parenchyma, does not actually invade through the basement membrane. Since preteolysis, cancer cell locomotion, angiogenesis, and other factors are necessary for cancer cells to penetrate the basement membrane,14–17 mere spread along the collecting ducts seems to imply low invading potential of cancer cells. Fat invasion, i.e., the scattering of cancer cells in the perirenal sinus fat tissue, appears not to be identical to extravesical extension of bladder carcinoma in prognostic impact. Bladder carcinoma must penetrate a thick muscle layer with abundant vasculature to extend to the perivesical adipose tissue. Namely, it acquires sufficient opportunity for metastasis even until reaching the extravesical adipose tissue. Moreover, immediately after a tumor penetrates the bladder wall, it encounters adjacent organs or the peritoneum. Renal pelvic tumors, on the other hand, require only penetration of a thin wall with relatively scarce vasculature to invade perirenal fat tissue, and, even after penetration, tumor cells involved in the fat tissue are surrounded by a barrier of the renal parenchyma. Thus there might be no difference in prognosis between tumors with fat invasion and those without fat invasion, if standard nephroureterectomy is performed.

Another important finding of the current study is the strong association of gender with prognosis found on multivariate analysis. Akaza et al. reported that female patients with upper urinary tract carcinoma had a poorer prognosis than male patients.18 However, their female patients tended to have higher histologic grades and higher pN status and to be older than their male patients. Munoz et al. reported that disease-specific annual mortality of upper urinary tract carcinoma was greater in women than in men.19 Urothelial transitional cell carcinoma, including bladder carcinoma20 and upper urinary tract carcinoma,19 is more common in men than in women, and there are several reports of better prognosis in men than in women for both cancers.19, 21–23 These epidemiologic and biologic differences may be due to genetic,24–26 hormonal,27 or other factors. Confirmation by a study with a much larger patient population, especially one using external data sets, is warranted regarding this prognostic difference between the genders; this is an interesting area for future investigation.

Tumor grade did not exhibit statistical significance as a prognostic predictor in the current study. Several previous studies on upper urinary tract carcinoma revealed that tumor stage and grade were well correlated with each other and that both of these parameters had a strong impact on patient prognosis.2–6, 18 However, none of these studies included multivariate analyse. It is possible that tumor grade does not remain as a prognostic indicator after multivariate analysis. Recently, Cheng et al. reached a similar conclusion in their study with bladder carcinoma patients.28 The current analysis using only patients with pT3 renal pelvic carcinoma suggests this possibility as well.

There are some limitations to the current study due to the heterogeneity of patients, who were treated in three separate institutes over a 26 year period. Surgical approach, technique, and extent of lymph node dissection were not strictly identical in all cases. Therefore, underestimation of the number of cases with positive lymph nodes in particular could have influenced the current analysis, although the positive lymph node rate of 27 % was sufficiently high. One further problem with the current study is clinical staging, since that used in the 1970s and early 1980s is not as sophisticated as that used now. The current series included four patients treated in the 1970s and five in the early 1980s. However, even with omission of these nine patients from the analysis, our conclusions are unchanged (data not shown).

The current study yielded one more important finding that pT3 disease without extensive parenchymal invasion was similar in prognostic impact to lower stage disease and that pT3 disease with extensive parenchymal invasion was similar to pT4. Das et al. found statistically significant prognostic differences between Stage A (pTa and pT1) and Stage B (pT2) and between Stage C (pT3) and Stage D (pT4) in their study of upper urinary tract carcinoma.5 Other investigators have also insisted that stage is a strong predictor of prognostic outcome.2, 4, 6 The current study, which focused only on renal pelvic transitional cell carcinoma, supported these findings in part. However, more importantly, the current results suggested that all surgically resected renal pelvic carcinomas could be divided into two large groups by prognostic impact, with one group consisting of pTa (and pTis), pT1, pT2, and pT3 disease without extensive parenchymal invasion and the other of pT4 and pT3 with extensive parenchymal invasion. We therefore conclude that the distinction of pT3 disease without extensive parenchymal invasion from that with extensive parenchymal invasion presents a line of demarcation within all stages of renal pelvic carcinoma in view of prognosis, and that pT3 disease should be subclassified into two separate entities according to the presence or absence of extensive parenchymal invasion.