Plasmacytoid Urothelial Carcinoma: Response to Chemotherapy and Oncologic Outcomes

INTRODUCTION

Plasmacytoid urothelial carcinoma (PUC) is a rare urothelial carcinoma (UC) variant with an aggressive phenotype [1], comprising approximately 3– 5% of patients with muscle-invasive bladder cancer. Histologically, PUC can appear as diffuse sheets of cells or as a discohesive, single-cell and highly infiltrative growth pattern. The cells with their eccentrically located nuclei and a moderate-to-abundant eosinophilic cytoplasm, resemble plasma cells [2], Fig. 1. Molecularly and histologically, PUC is characterized by loss of cytokeratin 20, a high proliferation index, tumor protein 53 accumulation, and loss of membranous E-cadherin expression [3]. In terms of clinical behavior, case series have reported that patients with PUC more commonly present with advanced stage [2], positive margins at radical cystectomy (RC) [4], locoregional spread along fascial planes [5] and propensity for peritoneal carcinomatosis [6].

OPEN IN VIEWER

Fig. 1

Photomicrography of bladder tumors. A – Plasmacytoid urothelial carcinoma (40x). B - Plasmacytoid urothelial carcinoma (200x). C – Plasmacytoid urothelial carcinoma with rhabdoid features (100x). D – Plasmacytoid urothelial carcinoma with signet ring features (200x). E – Conventional urothelial carcinoma (40x). F – Conventional urothelial carcinoma with muscle invasion (100x).

The standard of care for muscle-invasive UC is cisplatin-based neoadjuvant chemotherapy (NAC) with either accelerated (dose dense) methotrexate/vinblastine/adriamycin/cisplatin (MVAC) or gemcitabine/cisplatin (GC) [7, 8]. In general, PUC has been consistently associated with poor prognosis despite the use of standard bladder cancer treatment modalities, based on retrospective data across the literature [3]. In addition, applying the NAC paradigm to PUC is based on scarce and equivocal data to date, mainly extrapolating from conventional UC [9, 10], while the question of appropriate management in those patients remains unanswered. The purpose of this study was to elucidate clinicopathologic factors related to PUC and investigate response to NAC, also in relation to conventional UC. We hypothesized that patients with PUC would present with more advanced stage at diagnosis and at RC (vs. conventional UC), and that NAC would not significantly influence pathologic stage or overall survival (OS), in comparison with upfront RC.

MATERIALS AND METHODS

After obtaining institutional review board approval (IRB# STUDY00005690), we retrospectively reviewed the charts of all consecutive patients with PUC treated at our institution from January 2007 to December 2018, based on pathology review [transurethral resection of bladder tumor (TURBT) and/or definitive surgery] by institutional experts. In compliance with the IRB regulations we obtained waiver of consent, also considering the retrospective and observational nature of the study. No prospective human experimentation was involved.

We included all cases with plasmacytoid features, regardless of the percentage of plasmacytoid histology.

In addition, we retrospectively assessed our institutional RC database, to identify consecutive patients only with conventional (pure) UC in both TURBT and RC specimens (no variant histology present). All cases were reviewed by an expert pathologist with significant experience in genitourinary cancers. Since a small number of PUC cases were treated with definitive surgery other than RC (one partial cystectomy and one segmental ureterectomy), we utilized the umbrella term “extirpative surgery” to refer to all types of definitive surgery. We collected demographic data, baseline clinicopathologic data [clinical stage at diagnosis, pathologic stage at surgery, pathologic variants with percentages, presence of lymphovascular invasion (LVI) and carcinoma in situ (CIS)], preoperative laboratory workup, baseline historical data, medical comorbidities and treatment modalities [TURBT, intravesical bacillus Calmette-Guerin (BCG), surgery, neoadjuvant / adjuvant chemotherapy, chemoradiation, palliative chemotherapy and/or radiation regimens]. In terms of NAC, we captured the regimens, number of cycles received, first and last date of infusion, clinically significant complications [deep venous thrombosis/pulmonary embolism (DVT/PE), neutropenia (including febrile), acute kidney injury (AKI), need for hospital admission during NAC receipt] radiographic and pathologic response. Pathologic complete response to NAC was indicated as ypT0N0.

Statistical analysis was performed with STATA package (College Station, Texas). T-tests and chi-squared tests (Pearson, Fischer’s exact test) were used for group comparison analysis. Kaplan-Meier curves were created for median overall survival (OS) estimation, measured from the date of cancer diagnosis to all-cause mortality in the entire cohort, and from the date of extirpative surgery in the surgical subgroup. Median recurrence free survival (RFS) from the time of extirpative surgery to initial radiographic/pathologic recurrence or death was estimated with the same method. Log-rank test was used for unadjusted comparisons of OS and RFS between the groups. 5-year OS and RFS cumulative rates (with standard error reported in parentheses) were estimated with life-table analysis. Multivariate Cox-regression analysis was performed for OS. The primary predictor of interest was the use of NAC. Covariates tested were age at diagnosis, gender, cT stage, adjuvant therapy, pT, pN stage, positive surgical margins, extravesical disease and the presence of CIS and LVI status at extirpative surgery. Covariates were included in the final multivariate Cox-regression model if they changed the hazard ratio (HR) for NAC by > 10% from the base model. A 5% alpha error was considered as the cutoff of statistical significance.

RESULTS

Survival analysis and prognostic factors

In the entire PUC cohort, including surgical and non-surgical patients (n = 64), median follow-up was 34.1 months [95% Confidence Interval (CI): 10.1 – 58.0], while 32/64 (50%) deaths were noted (27 bladder cancer - related). Median OS was 17.6 months (95% CI: 4.6 – 30.6) from PUC diagnosis. Patients who underwent extirpative surgery had superior OS (from diagnosis) compared with the non-surgical group (27.9 vs. 9.2 months, p < 0.01), Fig. 2.

OPEN IN VIEWER

Fig. 2

Overall survival (OS) from the time of diagnosis of plasmacytoid urothelial carcinoma (PUC), between surgical and non-surgical group.

Median OS (from the time of extirpative surgery) in the surgical PUC subgroup was significantly inferior to conventional (pure) UC (23.6 vs. 154.0 months, p < 0.01, Fig. 3), and the same was true for median RFS (13.7 vs. 99.4 months, p < 0.01). Five-year cumulative OS was 35% (±10) in patients with PUC and 71% (±3) in those with conventional UC, while five-year cumulative RFS was 33% (±9) and 60% (±3), respectively. In univariate Cox regression for OS in the entire surgical cohort (UC and PUC, n = 461), presence of plasmacytoid variant was significantly associated with worse OS (HR 2.78, 95% CI: 1.65 – 4.64). However, in the multivariate model, plasmacytoid histology was no longer significant, with pT stage, pN+ and advanced age at surgery being the primary determinants of OS, Table 3. Additionally, when we stratified OS analysis by pT stage at RC, no significant differences between the PUC and the UC subgroup were noted for non-muscle invasive disease (<pT1, p = 0.48) and extravesical disease (pT3/4, p = 0.28), while there was a marginal difference in OS for those with muscle invasive tumors confined to bladder (pT2, p = 0.04), in favor of conventional UC. Similar results were yielded when we compared the plasmacytoid surgical subgroup with conventional UC patients that presented with cT2-4 disease only (n = 289). In overall unadjusted comparison, PUC had a significantly inferior median OS (p < 0.001) but in stratified analysis, this effect was primarily attributed to the pT2 subgroup (p < 0.01), while no significant differences were noted for pT1 (p = 0.58) and pT3/4 (p = 0.19) stages.

OPEN IN VIEWER

Fig. 3

Overall survival (OS) from the time of extirpative surgery for conventional (pure) urothelial carcinoma (UC) and plasmacytoid UC (PUC).

OPEN IN VIEWER

Table 3

Multivariate Cox Regression for prognostic factors of overall survival (OS) from the time of extirpative surgery

UC and PUC	HR (95% Confidence Interval)	p-value
(N = 461)
Plasmacytoid component	1.20 (0.62–2.32)	0.58
Gender (female)	1.27 (0.81–2.01)	0.29
Age at Surgery	1.02 (1.01–1.04)	0.03
pT3/4	2.18 (1.18–4.04)	0.01
pN+	2.14 (1.33–3.43)	< 0.01
Positive surgical margins	0.77 (0.37–1.6)	0.49
LVI at surgery	1.56 (0.94–2.59)	0.08
Neoadjuvant Chemotherapy	1.32 (0.89 –1.94)	0.17

Abbreviations: UC – urothelial carcinoma, PUC – plasmacytoid urothelial carcinoma, HR - Hazard Ratio, pT – pathologic tumor stage, pN – pathologic nodal stage, LVI - lymphovascular invasion.

DISCUSSION

In this study, we present outcomes from a single institutional cohort of 64 patients with PUC. A significant finding was the low ypT0N0 rate (10%) after cisplatin-based NAC, which was inferior to conventional UC patients (33%) over the same time period. In addition, NAC use in PUC patients was not associated with improvements in the presence of extravesical disease, positive surgical margins and pN+ status at the time of surgery, nor with improved OS, compared to upfront RC. These results suggest that NAC use may be less effective in the PUC setting, thus raising the question whether it adds benefit vs. RC alone in those patients.

In the literature, data on PUC “chemo-sensitivity” is scarce, with variable results reported across studies, Table 4. In one cohort of 31 patients with PUC, NAC was administered to five (stage I-III) patients and it was associated with 80% pathologic downstaging and 60% ypT0N0. However, in the same study, NAC was also administered to seven patients with stage IV, followed by salvage cystectomy, but no pathologic response was noted [9]. Hashemi-Shadraei et al. reported poor “chemo-sensitivity” in a subset of 14 patients with PUC, treated with cisplatin-based NAC and RC; one patient had ypT0N0 and 12/14 (86%) had persistent extravesical disease at RC [10]. Another study with 33 patients treated with NAC and RC demonstrated 22% pathologic response (<pT2N0), with 12% ypT0N0 [11]. Similar results were yielded by Ericson et al. [12]; NAC was administered to 12 patients, but in multivariate analysis, it failed to show a significant effect on pT0 incidence or organ confined disease. Overall, these results are in concordance with our findings, in part validating the observation of low likelihood of PUC pathologic response to NAC. The complete response rate (ypT0N0) in our conventional UC patients over the same time frame is consistent with prior studies (25–30%) [7, 8].

In terms of clinicopathologic behavior and outcomes, our observations confirmed the aggressive nature of plasmacytoid variant reported previously [4, 9–13]. Patients in our surgical PUC cohort had significantly higher propensity towards advanced stage at diagnosis (47%) and extravesical disease at surgery (65%), accompanied by frequently positive nodal status (37%) and surgical margins (23%), as well as inferior median OS and RFS, when compared to the conventional UC group. Additionally, there was a significantly higher proportion of patients with postoperative peritoneal recurrence in the PUC group. On the other hand, presence of plasmacytoid histology was not independently associated with worse survival outcomes in multivariate analysis, with advanced age, pT3/4 stage and pN+ disease being the primary negative prognostic factors. This result could be partially attributed to the small proportion of patients with PUC in the entire cohort. In a similar fashion, there were no significant differences in OS when the analysis was adjusted for pT stage at RC, with the exception of pT2 stage. In general, those findings are suggesting that the worse outcomes in patients with PUC are primarily mediated by well-established factors (age, stage). In the literature, historical data for conventional UC reported lower rates of cT3/4 stage (11%) at diagnosis and pT3/4 (43%) stage at RC [14]. Positive surgical margins have a reported incidence of 1–10% in conventional UC across stages [15], and up to 12% in cases of extravesical disease extension [16]. Nodal-positive disease at RC can be found in approximately 25% in general, and up to 34% in patients with advanced pT stage [16, 17].

The highly infiltrative potential of PUC has been linked to PUC-specific somatic loss-of-function mutations of CDH1, leading to loss of E-cadherin expression, likely promoting cellular detachment and aggressive localized invasion [18]. The discohesive, single-cell growth pattern in histopathology (Fig. 1), along with the increased frequency of localized and peritoneal recurrence (Table 4) are consistent with this finding. E-cadherin loss in 57% of PUC specimens has also been reported by others, including Perrino et al. [19]. These authors described stratification of PUC into desmoplastic, classic and pleomorphic morphology, with loss of E-cadherin being more prominent in the last two categories, while 9p21 deletion and FGFR3 mutations were frequently identified, providing insight on molecular pathogenesis and potential therapeutic targets. Taking these findings into consideration, expanding our knowledge in the genomic level of metastatic UC and its variants (including PUC) is of paramount importance, with rapid autopsy programs being an interesting initiative that allows tumor tissue acquisition from multiple sites of metastasis, in the immediate postmortem period [20].

Our PUC cohort is one of the largest reported in the literature, with the additional inclusion of a comparative analysis with conventional UC. However, due to its retrospective nature, our study has a number of limitations, including data acquired from a single institution, small sample size with low number of events, lack of randomization, presence of potential selection and confounding biases, variability in treatment modalities and monitoring/surveillance schedules, lack of biomarker evaluation and of specific % of plasmacytoid component, as well as loss to follow-up in a number of patients. Patients with advanced - and potentially more aggressive - disease, were more likely to receive NAC, which could also influence outcomes independent of NAC. Associations can be deduced but direct causality cannot be inferred. Despite the above, our results are hypothesis-generating, supporting further studies in PUC treatment strategies and evaluation of biomarkers.

CONCLUSIONS

This institutional study confirms that PUC is an aggressive variant, demonstrating increased propensity for early extravesical spread and lymph node infiltration, when compared with conventional UC. In addition, we observed suboptimal response to NAC, with a 10% ypT0N0 rate. The low likelihood of complete response or downstaging raises a therapeutic dilemma between NAC and upfront RC. Considering the inconclusive literature on the “chemo-sensitivity” for PUC, our findings underscore the importance of larger dedicated studies on PUC, investigating the rationale of chemotherapy utilization in the peri-operative setting, with additional focus on molecular profiling, further genomic interrogation for potential “druggable” targets, and assessment of immunotherapy.