What's known on the subject? and What does the study add?
The more that bladder cancer progresses from the urothelium to the outside of the bladder the worse the prognosis. To date, the use of adjuvant chemotherapy has not been completely defined.
The present study clarifies the prognosis and benefits of adjuvant chemotherapy for different stages of bladder cancer that invade perivesical fat.
• To assess the prognosis of pT2b, pT3a and pT3b bladder cancers after radical cystectomy (RC) in order to define potential situations where chemotherapy may be of benefit.
PATIENTS AND METHODS
• Between 1985 and 2009, 903 patients underwent a RC and pelvic bilateral lymphadenectomy in an Institutional Referral Centre.
• In all, 87 patients (9.6%) had a pT2b tumour, 111 patients (12.3%) a pT3a tumour, and 129 patients (14.3%) a pT3b tumour.
• The median (range) overall follow-up was 23 (1–350) months.
• Overall (OS), disease-specific (DSS), metastases-free (MFS) and local recurrence-free survival (LRFS) was estimated and compared using Kaplan–Meier plots and log-rank test.
• The 5-year survivals pT2b and pT3a were similar for LRFS (86% vs 84%), MFS (69% vs 63%), DSS (72% vs 70%) and OS (66% vs 61%), and the prognosis was better than for pT3b stage tumours (69%, 44%, 40%, and 31% respectively).
• In pN0 disease, MFS differences between pT2b–pT3a and pT3b tumours were not significant in patients who had received adjuvant chemotherapy (MSF of 87%, 69% and 56%, respectively) while they were significant in patients without adjuvant chemotherapy (MFS of 70%, 68% and 42%, respectively).
• Bladder cancers invading perivesical tissue macroscopically have a greater propensity to produce lymph node metastases, local recurrence, and have lower MFS, DSS, and OS. In pN0 disease, pT3b tumours may receive more benefit from adjuvant chemotherapy.
• Our results could be a useful for selecting patients for adjuvant chemotherapy.
In patients with non-metastatic bladder cancer, prognosis greatly depends on local extension of the tumour. Depending on the level of infiltration by the tumour tissue, the American Joint Cancer Committee TNM 2002 classification distinguishes stages T2a (superficial invasion of the detrusor), T2b (deep invasion of the detrusor), T3a (microscopic extravesical invasion), pT3b (macroscopic extravesical invasion), and pT4 (invasion of adjacent organs). The prognostic value of the distinction between microscopic or macroscopic perivesical fat invasions remains a subject of debate. Therefore, to compare the prognosis of neighbouring stages of pT3a bladder cancers, nominally pT2b and pT3b stages appears to be of interest. In a cohort of 3217 patients from the Surveillance, Epidemiology and End Results (SEER) registry, Scosyrev et al.  showed a difference of prognosis between pT2b and pT3a bladder cancers but not between pT3a and pT3b bladder cancers. Conversely, Bastian et al.  have shown that the prognosis of pT2 bladder cancers was not significantly different from that of pT3a stage, unlike pT3b stage.
Therefore, use of adjuvant chemotherapy, which is to date not completely defined, has been particularly debated in these situations. Although cisplatin-based chemotherapy is the only effective treatment, its use is limited due to its side-effects and the general condition and renal function of patients (unfit patients).
The aim of the present study was to assess the usefulness of perivesical fat invasion used as a prognosis factor to better identify patients who would definitely benefit by adjuvant chemotherapy to reduce the metastatic recurrence rate.
PATIENTS AND METHODS
Between 1986 and 2009, 903 patients underwent a radical cystectomy (RC) and pelvic bilateral lymphadenectomy at our institution for non-metastatic documented bladder cancer with no neoadjuvant chemotherapy. In this cohort, 87 patients (9.6%) had a pT2b tumour, 111 patients (12.3%) a pT3a tumour, and 129 patients (14.3%) a pT3b tumour. The detailed distribution of the 903 patients according to the TNM 2002 classification and the definition of the studied population are shown in Table 1. For the surgical procedure, homogeneous techniques were used, i.e. cystoprostatectomy for men and anterior exenteration for women. Pelvic lymph node dissection included the external iliac, ilio-obturator and internal iliac territories. The dissection stopped above the common iliac bifurcation.
Table 1. Distribution of the 903 patients according to the TNM 2002 classification
For pathological assessment, experienced genitourinary pathologists examined all RC specimens. The pathological staging was systematically reviewed and updated according to the 2002 TNM criteria. If perivesical extension was palpably or visibly obvious on careful systematic assessment of the RC specimen and could be confirmed histologically, stage pT3b was assigned. If perivesical invasion was diagnosed on routine histological examination of perivesical fat sections in the periphery of any tumour-bearing area of the bladder wall (without previous palpable or visible evidence of perivesical fat infiltration) stage pT3a was assigned. The presence of nodal metastases was analysed as a dichotomous variable. All tumours were classified as high grade according to WHO 2004 pathological grading.
In all patients with pT3 tumours and/or lymph node involvement, adjuvant chemotherapy was routinely proposed and patients were offered consultation with a medical onco-urologist. Adjuvant chemotherapy use has always included cisplatin associated with adriamycin (between 1986 and 1990) and to methotrexate, vinblastine, and doxorubicin (from 1990) or gemcitabine (from 2002). The toxicities of these therapies are referenced [3,4]. Finally, 152 patients received adjuvant chemotherapy (47%), 89 patients were considered unfit for adjuvant chemotherapy due to age, renal function, or comorbidity (27%) and 86 refused the treatment (26%).
Follow-up clinical and radiological examinations were performed every 3 months for the first year, semi-annually for the second year, and annually thereafter. The cause of death was either obtained from the medical chart or recorded prospectively, or it was obtained from the death certificate. Bladder cancer-specific deaths were classified as those that were directly attributable to bladder cancer. The median (range) overall follow-up was 23 (1–350) months. The median (range) follow-up for patients alive at last follow-up was 58 (1–350) months. Only 11 patients (3.4%) were lost to follow-up with a median (range) follow-up of 43 (1–219) months.
The database was established with the consent of patients and was approved by the Ethics Committee of our institution.
The Student's t-test was used to compare continuous variables. Fisher's exact test (small size sample) or Pearson's chi-square test were used to compare categorical variables. Binary logistic regression was used as a multivariate model to determine predictive factors for metastatic recurrence. Overall (OS), disease-specific (DSS), metastasis-free (MFS) and local recurrence-free survival (LRFS) was estimated using Kaplan–Meier plots. The log-rank test was used to compare pathological groups for results of adjuvant chemotherapy. A P < 0.05 was considered to indicate statistical significance. If necessary, Bonferroni P value adjustment was applied for multiple comparisons.
The characteristics of the 327 evaluable pT2b–pT3b patients were tabulated according to the pathological stages and are shown in Table 2. Lymph node invasion rates were similar for stage pT2b and pT3a but significantly higher for pT3b (14%, 21% and 47% respectively, P < 0.001).
Table 2. Characteristics of the 327 evaluable patients tabulated according to the pathological stages
Analysed lymph nodes, n
Lymph node metastases, n (%)
Adjuvant chemotherapy, n (%)
Follow-up, months mean (median; range)
65 (48; 1–80)
67 (122; 1–219)
44 (17; 1–226)
For the overall cohort, the 5-year LRFS in pT2b, pT3a, and pT3b cases was 86%, 84% and 69%, respectively. Local recurrence occurred more frequently in pT3b cases than in pT2b and pT3a cases (log-rank P < 0.001 and 0.009, respectively). In contrast, there was no difference between pT2b and pT3a tumours. The 5-year MFS in pT2b, pT3a, and pT3b cases was 69%, 63% and 44%, respectively. Metastases occurred more frequently in pT3b cases than in pT2b and pT3a cases (log-rank P < 0.001 and 0.004, respectively). The 5-year DSS in pT2b, pT3a, and pT3b cases was 72%, 70% and 40%, respectively. Cancer related death occurred more frequently in pT3b cases than in pT2b and pT3a (both log-rank P < 0.001). The 5-year overall mortality in pT2b, pT3a, and pT3b cases was 66%, 61% and 31%, respectively. Overall mortality was higher in pT3b cases than in pT2b and pT3a (both log-rank P < 0.001). In summary, pT2b and pT3a were similar for local recurrence, metastasis occurrence, DSS and OS. However, for all these parameters they were prognostically better than pT3b stage tumours (Fig. 1).
In pN0 subgroups, there were statistically differences in local recurrence, metastases occurrence, DSS and OS between the pT3b and pT2b–pT3a cases. In contrast, there was no difference between pN ≥1 subgroups according to the pathological stage (Table 3).
Table 3. The 5-year LRFS, MFS, DSS and OS according to lymph node invasion in pT2b, pT3a and pT3b cases.
When not stratified for pathological stage, MFS in pN0 patients who received adjuvant chemotherapy was not significantly different from those who did not receive adjuvant chemotherapy. However, in pN ≥1 diseases, the MFS of patients treated with adjuvant chemotherapy was significantly better than their counterparts without treatment (log-rank P < 0.001; Fig. 2).
For pN0 stage, the 5-year MFS in patients with pT2b, pT3a and pT3b tumours who did not received adjuvant chemotherapy, was 70%, 68% and 42%, respectively. The outcome of patients with pT3b tumours was significantly worse than pT2b (log-rank P= 0.002) and pT3a (log-rank P= 0.02). Comparatively, in patients with pT2b, pT3a and pT3b tumours who received adjuvant chemotherapy, the 5-year MFS was 87%, 69% and 56%, respectively. The outcome of patients with pT3b tumours was not significantly different from those with pT2b and pT3a stage tumours (Fig. 3).
Patients with metastatic recurrence more often had pT3b stage disease (51% vs 33%, P < 0.001) and lymph node invasion (41% vs 24%, P < 0.001) but less often pT2b tumours (20% vs 30%, P= 0.02) than their counterparts without metastasis. In multivariate analysis, predictive factors for metastatic recurrence were lymph node invasion (hazard ratio 1.97, 95% CI 1.15–3.38), and pT3b stages (hazard ratio 1.70, 95% CI 1.03–2.81).
The current TNM classification distinguishes pT3a and pT3b stages according to the microscopic or macroscopic aspect of tumour invasion on the RC specimen. Moreover, these stages are also clearly distinguished from pT2b stage that corresponds to an invasive bladder cancer infiltrating the deep bladder muscle without extravesical extension. However, these morphological distinctions are not clearly correlated with specific prognostic profiles. The present study attempts to present a new perspective for the evaluation of pT2b, pT3a and pT3b bladder cancer stages. Firstly, the risk of lymph node invasion was similar for stage pT2b and pT3a but significantly higher for pT3b. Secondly, irrespective of the lymph node status, prognostic differences have been shown between pT3b and pT2b–pT3a stages, but not between pT2b and pT3a, considering the proportion of patients receiving adjuvant chemotherapy was not significantly different in each subgroup. Thirdly, when stratified according to the lymph node status, prognostic differences between pT2b, pT3a and pT3b bladder cancer stages persisted in pN0 but not in pN ≥1 situations.
For prognostic differences between these above mentioned stages, various authors have reported contradictory results. Based on the SEER registry, Scosyrev et al.  compared lymph node invasion and outcome features of 866 pT2b, 1030 pT3a, and 446 pT3b bladder cancers. These authors showed that the risk of nodal metastases increased according to the sub-stage (pT2b 20%, pT3a 36%, pT3b 48%, trend P= 0.001). Although, the present results are in agreement with these specific findings, concerning the threshold pT2/3, nevertheless they, in fact, tend to show that when adjusted according to the lymph node status the prognosis of pT3a and pT3b stages does not differ but was significantly worse than that of pT2b stages.
Quek et al.  and Boudreaux et al.  retrospectively compared outcomes of RC for 69 and 46 pT3a pN0 and 167 and 29 pT3b pN0 bladder cancers respectively, but no prognostic differences between the groups were found in these two studies which involved small cohorts. Moreover, Bastian et al.  have shown that, where adjustment was made for age, grade, lymphovascular invasion and nodal status, for patients with pT3b tumours the rate of recurrence increased by 1.8 and death rate increased by 2.0 compared with their counterparts with pT2 tumour invasion alone. Conversely, patients with pT3a bladder tumour did not have a statistically significant greater risk of recurrence or death than patients with pT2 tumours. Very recently, Sonpavde et al.  compared 356 pT3a pN0 and 222 pT3b pN0 bladder cancers and reported a rate of recurrence increased by 2.1 between the groups. Unfortunately, their cases originated from nine centres worldwide and were not reviewed by a pathological central review board. The present study provides further evidence for a prognostic difference between pT3b and pT2b–pT3a stage tumours and confirms the predominant prognostic value of lymph node invasion, which is more frequent in pT3b than in pT2b–pT3a.
The potential real benefit of adjuvant chemotherapy for this heterogeneous group of patients with invasive bladder cancer is directly linked to the prognostic value in order to identify the targeted sub-population. During the past decades severla phase II clinical trials have investigated the role of chemotherapy in patients with locally advanced bladder cancer . These studies have identified a limited number of active single-agent cytotoxic drugs, e.g. cisplatin . Although promising results have also been achieved with various combinations of drugs, there is no conclusive evidence that chemotherapy improves survival, irrespective of whether it is given before (neoadjuvant or pre-emptive), with (concomitant) or after (adjuvant) local treatment . Very recently Wosnitzer et al.  have shown that there are neither DSS nor OS differences when cisplatin- or carboplatin-based chemotherapy was administered before vs after RC. In the context of pN0 staging, the present study has shown that MFS differences between pT2b–pT3a and pT3b disease were not significant in patients who received adjuvant chemotherapy, while they were significant in patients without adjuvant chemotherapy. The present study therefore provides an argument for a benefit of adjuvant chemotherapy in the subgroup of pT3b tumours but not in less invasive stage disease. Chemotherapy sequence relative to surgery appears less important than whether or not a patient received the chemotherapy perioperatively. Therefore, RC (with pelvic lymphadenectomy) allows accurate pathological staging to select patients who may benefit or not from this adjuvant chemotherapy.
The prognostic differences between pT2b, pT3a and pT3b tumours, namely the higher frequency of lymph node invasion and, independently, the MFS in pT3b tumours, could be subtended by anatomical and biological factors. Anatomically, large lymphatic vessels are located in the perivesical fat away from the serosa . Marchetti et al.  have shown that the proximity of invasive bladder cancer with lymphatic vessels promotes the spread of cancer to lymph nodes. Consequently, the lymphatic vessel geometric topography could explain the higher rate of lymph node invasion associated with pT3b tumours and hence the real prognosis differences in the present study. Secondly, the bladder serosa is a very thin layer that does not constitute a barrier for bladder cancer growth. There is no reported modification of tumour aggressiveness before serosal invasion. This could explain the similarity of prognosis for pT2b and pT3a tumours. Biologically, there is no evidence for phenotypical transformation between pT3a and pT3b tumours. However, the epithelial–mesenchymal transition is not easy to show because it seems that only the tumour cells at the periphery of the tumour trigger this transition . Consequently, larger tumours have a greater peripheral volume where cells can acquire the phenotypical modifications necessary for metastasogenesis .
We acknowledge that there are some limitations in the present study, inherent to any retrospective study. Nevertheless only 11 patients were lost to follow-up (3.4%) after 43 months. As our institution is a tertiary reference centre for bladder cancer surgery, the high surgical level could lead to a overestimation of the oncological results compared with the practice of general centres [16,17]. Adjuvant chemotherapy was not randomly prescribed but was proposed to all pT3 and/or pN ≥1 disease. Consequently, the level of evidence concerning optimal chemotherapy results in pT3b pN0 diseases was limited. There was a significant percentage of ‘unfit’ patients (27%) and refusals for adjuvant chemotherapy (20%). However, this does describe the current pragmatic situation for patients who undergo adjuvant chemotherapy after RC.
In conclusion, the prognoses of pT2b pN0 and pT3a pN0 bladder cancers are comparable and significantly better than those of pT3b pN0 diseases. Bladder cancers invading perivesical tissue macroscopically have a greater propensity to produce lymph node metastases and local recurrence, and have lower MFS, DSS, and OS. When lymph nodes are invaded the prognostic differences between pT2b, pT3a, and pT3b stages does not persist. In this retrospective study, there were MFS differences between pT2b–pT3a pN0 and pT3b pN0 diseases in patients without adjuvant chemotherapy but not in those with adjuvant chemotherapy. These results could be a useful for selecting patients for adjuvant chemotherapy.