Clinicopathological parameters of the included patients are shown in Table 1. Although a limited number of patients were included from the MPC and PUC tumor types, they appear to be more frequent in male patients than in female patients (UC: 3.3:1, PUC: 8:1, MPC 9:0). The mean age of the PUC patients seemed to be somewhat lower than that of patients suffering from UC or MPC (57.9, 61.0, and 62.3 years, respectively, p= 0.057). A significant difference for chemotherapy treatment between the different histotypes (P=0.043; Fisher’s exact test; Table 1) is detected. This difference is reasoned by the fact that only UC patients but not MPC or PUC patients have been treated with gemcitabine and cisplatin. However, when we consider only M-vec and Cm treatment regimens no significant difference between the different histotypes is detected (P=0.451; Fisher’s exact test; data not shown). Overall survival of PUC patients was significantly worse than that of patients suffering from UC or MPC (27.4 months with 95% CI: 16.8-37.9 months, 62.6 months with 95% CI: 54.8-70.4 months, and 64.2 months with 95% CI: 41.9-86.4, respectively; p=0.013 using Kaplan Meier analysis; Figure 3). A backward multivariate Cox´ s proportional hazards regression analysis including clinicopathological parameters (age, sex, tumor grade, tumor stage, lymph node and metastases status, type of chemotherapy) showed a hazard ratio of 3.2 (95% CI: 1.0-9.9; p=0.045) for PUC in contrast to patients suffering from MPC (Table 2 and Figure 4). When we analyzed the hazard ratio of the PUC patients compared with the UC patients with the same backward model we still see a 2.4-fold (95% CI: 1.3-4.4; p=0.006) increased risk of death for the PUC patients but no significant difference in risk of death between UC and MPC patients was seen (data not shown). The lymph node status was only in the first step significantly associated with the risk of death but at further backward steps only a trend towards significance was observed (Table 2).

	Total	UC	PUC	MPC	P-Value*
Age					0.057**
Range	29-75	29-75	46-70	56-70
Mean	60.9	61.0	57.9	62.3
Median	61.7	61.9	59.3	60.8
Gender	205	178	18	9	0.116
Female	45 (22%)	43 (24%)	2 (11%)	0 (0%)
Male	160 (78%)	135 (76%)	16 (89%)	9 (100%)
pT	205	178	18	9	0.168
pT1	5 (2,5%)	5 (3%)	0 (%)	0 (0%)
pT2	22 (11%)	19 (11%)	3 (17%)	0 (0%)
pT3	142 (69%)	125 (70%)	13 (72%)	4 (45%)
pT4	36 (17.5%)	29 (16%)	2 (11%)	5 (55%)
Grade (1973)	205	178	18	9	0.263
Grade 2	22 (11%)	22 (12%)	0 (0%)	0 (0%)
Grade 3	183 (89%)	156 (88%)	18 (100%)	9 (100%)
Grade (2004)
Low-grade	0%	0%	0%	0%
High-grade	100%	100%	100%	100%
pN	205	178	18	9	0.743
pN0	90 (44%)	80 (45%)	6 (33%)	4 (44.4%)
pN1	45 (22%)	40 (22.5%)	3 (17%)	2 (22.2%)
pN2	69 (33.5%)	57 (32%)	9 (50%)	3 (33.3%)
pN3	1 (0.5%)	1 (0.5%)	0 (0%)	0 (0%)
pM	205	178	18	9	1.000
pM0	204 (99.5%)	177	18 (100%)	9 (100%)
pM1	1 (0.5%)	1	0 (0%)	0 (0%)
Chemotherapy	205	178	18	9	0.043
Gc	39	39 (%)	0 (0%)	0 (0%)
M-vec	80	64 (%)	11 (61%)	5 (55%)
Cm	86	75 (%)	7 (39%)	4 (45%)
Survival	205	178	18	9	0.061
Alive	120	109	6 (33.3%)	5 (55%)
Dead	85	69	12 (66.6%)	4 (45%)

Abbreviations: Gc - Gemcitabine, cisplatin, M-vec - Methotrexate, vinblastine, epirubicin and cisplatin, Cm – Cisplatin, methotrexate.

* statistical test: Fisher’s exact test.

** statistical test: t-test.

At considering that we do not have for all patients the information for cancer-associated death we calculated their cancer-associated risk of death. In Kaplan Meier analysis we detected mean cancer-associated survival rates for PUC patients of 35.5 months, for UC patients of 67.8 months and for MPC patients of 64.2 months but this is not significant (P=0.22, log rank test). In the multivariate Cox’s regression hazard analysis a 2.0-fold increased risk of tumor-related death for PUC patients compared to MPC patients was found but it was not significant (P=0.26) and an 1.8-fold increased risk of tumor-related death for PUC patients compared to UC patients what was again not significant (P=0.15).

Systemic cisplatin-based chemotherapy is regarded as the therapy of choice in metastatic UC. However, the role of adjuvant or neoadjuvant chemotherapy remains under debate [10, 11]. Most urological and oncological guidelines recommend neoadjuvant cisplatin-based chemotherapy as the therapy of choice in locally advanced bladder cancer [12]. Because the description of several rare variants of UC of the bladder in the WHO classification of 2004, descriptions of their clinical course and the molecular features have become more prevalent. However, despite improved knowledge regarding the molecular characteristics of the histologic subtypes of bladder cancer, the impact of these differences on systemic therapies is lacking. Most patients with locally advanced UC are treated without regard to the underlying histology, although it has been reported that tumors with variant histology are associated with a higher risk of progression than conventional high grade UC [13]. As for PUC we could show in the largest series described to date, that on the one hand this subtype of UC accumulates prognostic unfavourable molecular features, like such as the loss of CK20, a high proliferation index and p53 accumulation, and as well as on the other hand exhibits characteristic molecular features, like such as a complete loss of membranous E-cadherin expression [4]. Moreover, our results from this study demonstrate that patients suffering from PUC are of younger age than those suffering from UC or MPC. However, therapeutic strategies with radical cystectomy and cisplatin-based chemotherapy does not seem to be as effective for PUC as it is described for locally advanced UC or MPC, even though the number of patients is limited and may serve as a bias. As it was reported recently, a complete response to adjuvant chemotherapy administering M-VEC and neoadjuvant chemotherapy using gemcitabine and cisplatin may occur at least in a subgroup of PUC -patients [14]. Therefore, chemotherapy in a neoadjuvant or adjuvant setting should be part of the therapeutic considerations. In the present study, we could not confirm these results in our analysis of adjuvant cisplatin-based chemotherapy with a median overall survival of 27.4 months, which is approximately half of that observed in the patients suffering from conventional locally advanced UC (Kaplan-Meier analysis). Thus, PUC tumor biology represents a negative prognostic factor for patients suffering from this histologic variant. Loss of E-cadherin as a sign of epithelial-mesenchymal transition (EMT) and upregulation of transcriptional repressors of E-cadherin may contribute to the aggressiveness of these tumors and a possibly reduced sensitivity to chemotherapeutic agents [4, 15, 16].

Micropapillary carcinomas have been described in different tumor entities, such as colorectal, breast, lung and others [17–19]. Thus, immunohistochemical panels for discriminating micropapillary tumors of different origins have been reported previously. Within this broad molecular panel, CK20 and uroplakin are the best molecular markers to determine urothelial origin of MPC [19]. In MPC, the initial molecular data explaining their unfavourable clinical course were recently identified, with HER2/neu gene amplification, amongst others, as a frequent molecular alteration in this variant [20]. Clinical reports suggest these are markers of biologically aggressive carcinoma with frequent lymphatic vessel invasion in TURB specimens and lymph node metastasis [6, 8, 21]. Moreover, clinical upstaging to locally advanced diseases occurs in the vast majority of the cases and represents a problem in planning therapeutic strategies [22]. In response, Compérat et al. highlighted the importance of adequate tumor sampling, including analysis of the detrusor muscle, to avoid possible upstaging. Moreover, they state that due to the associated aggressive behaviour, the proportion of micropapillary differentiation should be reported in all cases, even if it represents less than 10% of the specimen, as it has prognostic relevance [22]. Additionally, inter-observer reproducibility of the diagnosis of MPC is low [23], which may lead to treatment delays or the use of inappropriate therapeutic strategies adversely affecting patients’ survival. Therefore, pathologists should be aware of the histologic subtypes on diagnosis. Furthermore, urologists or oncologists should take this information into account when planning surgical or chemotherapeutic treatment options. Supporting the recommendations of Compérat and coworkers Kamat et al. postulated that even papillary and non-invasive MPC should be treated by radical cystectomy to prevent progression and systemic disease [22, 24]. In their analysis they demonstrated that neoadjuvant cisplatin-based chemotherapy did not result in an improved 5-year overall survival and that intravesical immunotherapy using BCG was not effective in this histologic variant [24]. Most studies on MPC describe poor disease-specific survival following adjuvant chemotherapy [8, 20, 25]. However, our data are in contrast to the experiences reported previously. We demonstrated that the survival rates were comparable for MPC and UC if treated with radical cystectomy and adjuvant chemotherapy. These contradictory results may be explained by the prospective randomized nature in which the patients were recruited and included only upon the ability to compare UC, MPC and PUC within a single trial. Although the relatively low number of patients suffering from MPC or PUC may limit the value of our study, it provides important information regarding their clinical course and the aggressive biology of the tumor subtypes. A possible limitation of our study might be the interobserver variability in defining histological subtypes as there is still no consensus on the optimal cut off value of variant histology in the specimen to define PUC or MPC. Another limitation of our results is the measurement of overall survival in our series as this could be affected by several variables besides tumor characteristics. However, on the other hand chemotherapy can have effects on comorbidity and therefore finally affect overall survival what is of relevance for the patients. Awareness of these different bladder cancer variants appears to be crucial when analyzing the molecular characteristics of advanced bladder cancers and when tailoring personalized therapeutic procedures in the future.

The specific tumor histology gives important prognostic information of patients suffering from locally advanced bladder cancer treated by radical cystectomy and adjuvant chemotherapy. Our results implicate that determining the exact pathological diagnosis, including the description of histologic subtypes of bladder cancers according to the WHO classification of 2004, are important. As UC, PUC and MPC are associated with a different clinical course if treated with cystectomy and adjuvant cisplatin-based chemotherapy prospective multicenter studies, comparing the different histologic variants of bladder cancer and their molecular features are necessary to tailor therapeutic strategies in the future. Furthermore, a joint study including the collection of rare bladder cancers is strongly warranted with a goal of enforcing additional molecular studies for the identification of potential prognostic and therapeutic targets.

During the revision process Dayyani et al. published in accordance with our findings that PUC is a very aggressive variant of bladder cancer with overall poor outcomes. In addition they report that despite pathologic downstaging of patients treated with neo-adjuvant chemotherapy, relapses commonly occurred and no difference in survival between patients treated with neo-adjuvant chemotherapy compared to initial surgery was observed.