Low-Grade Papillary Urothelial Carcinoma of the Urinary Bladder: A Clinicopathologic Analysis of a Post–World Health Organization/International Society of Urological Pathology Classification Cohort From a Single Academic Center

Carcinoma of the urinary bladder is the fifth most commonly diagnosed malignant neoplasm in the United States with an estimated 68 810 new cases diagnosed in 2008 and a prevalence of roughly 500 000 cases.1 Outcome of patients with a noninvasive papillary tumor is largely dependent on tumor pathologic grade. Indeed, several classification systems for grading papillary urothelial neoplasms have been proposed in an attempt to more precisely predict recurrence and progression in this group of patients. The earlier 1973 World Health Organization (WHO) grading system suffered from a relative lack of interobserver reproducibility and a tendency to disproportionally classify tumors in the intermediate (grade 2) category.2,3 In 2004, the WHO adopted the 1998 International Society of Urological Pathology (ISUP) consensus classification system,4 and the 2004 WHO/ISUP system emerged. The latter system has been in use in our department since 1998 and has gained wide acceptance from urologic and general pathologists as well as urologists.5 It classifies papillary urothelial neoplasms into 4 categories: papilloma, papillary urothelial neoplasm of low malignant potential (PUNLMP), low-grade papillary urothelial carcinoma (LG-UrCa), and high-grade papillary urothelial carcinoma (HG-UrCa). Low-grade papillary urothelial carcinoma is characterized by an overall orderly appearance, minimal variability in architectures, and lack of significant cytologic atypia and mitotic activity.2,5 

Patients with a noninvasive low-grade bladder tumor are at risk for tumor recurrence following transurethral resection (TUR), with a subset of patients developing grade and/or stage progression.6,7 In the current study, we aimed to evaluate outcome in a cohort of patients with LG-UrCa who were diagnosed and treated at a single tertiary-care, academic facility after the adoption of the WHO/ISUP grading system. We also aimed at evaluating the potential role of clinicopathologic prognostic parameters of recurrence and progression.

The surgical pathology electronic database at The Johns Hopkins Hospital (Baltimore, Maryland) was searched for all in-house cases with a diagnosis of bladder LG-UrCa rendered between 1998 and 2008. Cases containing a synchronous, flat urothelial carcinoma (carcinoma in situ) or a prior positive urine cytology diagnosis consistent with high-grade urothelial carcinoma were not included in this study. Our cohort consisted of cases with a first-time diagnosis of LG-UrCa rendered in our institution, with the exception of 5 cases where initial LG-UrCa diagnoses were made in an outside institution, and the cases were referred to our hospital for confirmation. We excluded cases with an upper urinary tract tumor as the presenting lesion.

A total of 112 cases with LG-UrCa as the first lesion were identified. Hematoxylin-eosin–stained sections and clinical and pathologic records were available and retrieved in all cases. All histologic sections were reevaluated by 2 urologic pathologists on the study (H.M. and G.J.N.). Histologic criteria for LG-UrCa include an overall orderly appearance but with recognizable variation in architectural and cytologic features (eg, uniformly enlarged, scattered hyperchromatic nuclei with fine chromatin and small or inconspicuous nucleoli and infrequent mitotic figures). Tumor size was assessed as the largest dimension indicated in the pathologic gross description of our surgical pathology reports. Multifocality was assessed from the number of specimens in the gross pathology and cystoscopy reports. Outcome, follow-up, and other clinical data were obtained by reviewing our hospital electronic medical records.

Findings were analyzed using the Stata 9.2 (StataCorp, College Station, Texas) software package. Equality of population medians among groups was tested using Kruskal-Wallis test for nonparametric 1-way analysis of variance by ranks. Contingency table χ² test analysis was used for categoric variables. P values less than .05 were considered to be statistically significant.

Eight of 55 cases (14.5%) that had been originally diagnosed by a general surgical pathologist, without urologic pathology subspecialty expertise, were reclassified as HG-UrCa based on the current review. The reason for reclassification was the presence of focal areas of cytologic or architectural features or both meeting the criteria for high-grade carcinoma that was not initially detected. These 8 cases were excluded from further outcome analysis. The original diagnosis of LG-UrCa was confirmed in the remaining 47 cases as well as all of the 57 cases that had been originally reviewed by 1 of 2 senior urologic pathologists on the study (J.I.E., and G.J.N.).

The 104 patient cohort with confirmed LG-UrCa diagnoses included 76 men and 28 woman. The average patient age at diagnosis was 65.3 years with a range of 31 to 89 years. Forty-eight patients (46.1%) either were smokers at diagnosis or had a history of prior smoking. Ninety patients had a single initial lesion, whereas 14 had multiple lesions at initial TUR. Mean initial tumor size was 1.73 cm with a range of 0.1 to 6.8 cm. Mean follow-up period was 40.1 months (range, 2–113). All patients were initially treated with TUR. Follow-up included repeat cystoscopy and urine cytology at standard intervals. In a subset of patients, intravesical bacillus Calmette-Guérin (13 patients; 12.5%) or mitomycin intravesical therapy (28 patients; 26.9%) was administered. Seven patients (6.7%) ultimately received radical cystectomy/cystoprostatectomy.

Of the 104 patients, 56 (53.8%) developed one or more episodes of recurrence. Among these, 37 of 104 cases (35.6%) had recurrence as LG-UrCa or a lower-grade urothelial neoplasm (ie, urothelial papilloma or PUNLMP) according to the 2004 WHO/ISUP classification and 19 of 104 cases (18.3%) demonstrated grade progression, developing HG-UrCa in one or more recurrence episode. Seven of 19 patients (36.8%) with recurrent HG-UrCa also developed muscle-invasive (5 cases) or metastatic (2 cases) urothelial carcinoma (stage progression). Seven patients who developed HG-UrCa underwent radical cystectomy with residual tumor (2 pTa, 2 pT1, and 3 ≥pT2 diseases).

None of the 104 patients died of bladder cancer during follow-up (100% disease-specific survival). Among the 56 patients with tumor recurrence, the mean number of recurrence episodes was 3.11 (range, 1–16). The mean durations of time to first recurrence (in 56 cases) and time to grade progression (in 19 cases) were 13.9 months (range, 2–72) and 25.1 months (range, 2–88), respectively.

Our analyses of prognostic factors of tumor recurrence and tumor progression are summarized in Tables 1 and 2, respectively. Patients with multiple tumors at initial diagnosis (n  =  14; 13.5%) had a significantly higher risk of recurrence than those with single tumor (n  =  90, 86.5%; P  =  .04). However, the presence of multiple tumors on initial TUR was not associated with subsequent grade progression (P  =  .08). No statistically significant difference in tumor size was present between cases with and without recurrences (mean size, 1.60 cm versus 1.89 cm, respectively; P  =  .67). Similarly, initial tumor size was not associated with the likelihood of grade progression (mean size, 1.67 cm versus 1.46 cm in those with and without grade progression, respectively; P  =  .49). We found no significant correlation between patient age, sex, smoking history, or intravesical therapy and the likelihood of tumor recurrence, grade progression, or stage progression (see Tables 1 and 2).

Among patients with tumor recurrence, no statistically significant difference was found in the number of recurrence episodes between patients with and without progression (P  =  .86) or those with stage progression (P  =  .30).

Awaiting well-characterized, prospectively validated, prognostic and predictive molecular markers that can be integrated with current clinicopathologic parameters,6–11 achieving accurate histopathologic grading, remains a crucial step in the management of noninvasive urothelial neoplasms. Because grading is a paramount parameter for both prognostication and guidance of therapy in this group of neoplasms,12 pathologists and urologists have endured several attempts at achieving an optimal, effective, and reproducible grading system for papillary urothelial neoplasms.2,4,5 The current study on LG-UrCa was undertaken as part of our attempt to retrospectively characterize and compare outcome in a relatively homogeneously managed cohort of malignant, superficial, papillary urothelial neoplasms that were prospectively classified according to the 2004 WHO/ISUP grading system during a 10-year period that started with our department adoption of that grading system (1998–2008). Our findings on patient groups with HG-UrCa and PUNLMP will be separately reported.

The 2004 WHO/ISUP classification system describes, in detail, the specific architectural (ie, papillae and overall organization of the cells) and cytologic (ie, nuclear size, nuclear shape, chromatin content, nucleoli, mitoses, and umbrella cells) characteristics in each category of noninvasive papillary urothelial neoplasm.5 It defines LG-UrCa as a urothelial neoplasm composed of papillary fronds, which exhibit an overall orderly appearance but have mild variability in architectural and/or cytologic features. In contrast, HG-UrCa is characterized by a disorderly appearance with marked architectural and cytologic atypia, which are recognizable at low magnification. Our retrospective review identified a relatively high rate (14.5%; 8 of 55 cases) of misclassification by general surgical pathologists in our tertiary care, academic center. Given the recent trend for more aggressive management of HG-UrCa, especially in residual disease or recurrence,12 such misclassification rate could have significant implications on overall patient management. Additional studies that review cohorts originally diagnosed in a community hospital setting (referral material) are needed to evaluate whether our observed 14.5% rate of misclassifying (undergrading) of HG-UrCa is applicable to pathology practices that have a lower volume of bladder biopsy specimens. One potential source of misclassification errors could be the not-infrequent variability in histologic feature displayed in a given TUR specimen. Grading based on dominant pattern, rather than the highest-grade area, would lead to undergrading. In this regard, it remains to be seen whether a minuscule (eg, <5%) area of an HG-UrCa pattern affects the prognosis of an otherwise low-grade tumor.

Few prior, large cohort studies have assessed outcome in patients with noninvasive LG-UrCa since implementation of the updated WHO/ISUP consensus classification.2,13–16 Most recently, Herr et al17 evaluated 172 patients with LG-UrCa of the bladder (1998 WHO/ISUP classification) and detected a 72% recurrence rate, with median time to first recurrence of 18 months and mean number of recurrence episodes of 6.6 (range, 1–19). The authors found a 10% progression rate from either grade (4%) or stage (6 %) progression. In all prior studies based on the WHO/ISUP classification, low-grade, papillary urothelial carcinoma has consistently demonstrated a recurrence rate (34%–72%) that is higher than papilloma (0%–31%) and PUNLMP (32%–51%) but lower than HG-UrCa (43%–74%).13–17 Likewise, these studies have demonstrated a progression rate in LG-UrCa (4%–10.5%) that is higher than that in papilloma (0%) or PUNLMP (0%–3%).7–9,11 Our 53.8% recurrence rate in LG-UrCa is comparable to prior studies. However, we observed a relatively higher overall progression rate in our cohort of LG-UrCa (18.3%) but observed a similar rate of stage progression to that of Herr et al.17 Likewise, our cohort endured a shorter mean time to first recurrence (13.9 months) compared with prior studies. Similar to prior studies, our current cohort demonstrated a recurrence rate (53.8%) that is intermediate to those of our PUNLMP (42%) and HG-UrCa (56%) and a stage progression rate (6.7%) that falls between that of our separately reported PUNLMP (0%) and HG-UrCa (31%) cases.18 

The largest study to date on survival in patients with pTa tumors showed a 4% rate of death by disease in the 160 patients with LG-UrCa included in the study.13 None of our patients with LG-UrCa died of urothelial carcinoma; however, 2 patients developed metastatic disease and are alive with disease at their last follow-up.

Most prior studies evaluating clinicopathologic factors that predict recurrence in low-grade pTa tumors are based on the earlier 1973 WHO classification. Tumor multiplicity at initial diagnosis has been found in all prior studies to be one of the most predictive factors of recurrence.11–14 In the most recent study by Herr et al,17 a recurrence rate of 39% was found in patients with single initial lesion versus a rate of 85% in those with multiple tumors on initial diagnosis (hazard ratio, 3.2; P < .001). Our finding of a higher rate of recurrence in patients with multiple lesions on initial TUR is in agreement with prior studies by Herr et al.12 Tumor multiplicity did not predict progression in our cohort. Tumor size has been found to correlate with risk of recurrence in occasional studies,12,19 whereas it lacked prognostic significance in other studies20,21 including our current study. Finally, we found no statistically significant difference in number of recurrence episodes between patients with and without progression.

Our study has inherent limitations, including the relatively short, mean follow-up of our cohort that was dictated by our focus on the cohort after the WHO/ISUP consensus classification. Another limitation is the retrospective nature of the review.

In summary, more than half of our patients with noninvasive LG-UrCa developed recurrent tumor with almost one-fifth undergoing grade progression and occasional stage progression. The risk of recurrence was higher among patients with multiple tumors at initial diagnosis. We found a tendency to undergrade HG-UrCa (as LG-UrCa) in our academic center using the 2004 WHO/ISUP consensus grading classification.