Firas Abdollah and Maxine Sun contributed equally to this article.
Firas Abdollah, Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center, 1058, rue St-Denis, Montreal, Quebec H2X 3J4, Canada. e-mail: firstname.lastname@example.org
Study Type – Therapy (case series)
Level of Evidence 4
What's known on the subject? and What does the study add?
To date, there is controversy about the impact of histological subtype of bladder cancer (nonbilharzial squanous cell carcinoma vs. urothelial carcinoma) on cancer control outcomes.
Our study shows that the histological subtype may have an impact on the stage of bladder cancer at presentation. However, after adjusting to stage, the histological subtype has no impact on cancer control outcomes.
• To test the effect of histological subtype (NBSCC vs UC) on cancer-specific mortality (CSM), after adjusting for other-cause mortality (OCM).
• In Western countries, non-bilharzial squamous cell carcinoma (NBSCC) is the second most common histological subtype in bladder cancer (BCa) after urothelial carcinoma (UC).
PATIENTS AND METHODS
• We identified 12 311 patients who were treated with radical cystectomy (RC) between 1988 and 2006, within 17 Surveillance, Epidemiology and End Results (SEER) registries.
• Univariable and multivariable competing-risks analyses tested the relationship between histological subtype and CSM, after accounting for OCM.
• Covariates consisted of age, sex, year of surgery, race, pathological T and N stages, as well as tumour grade.
• Histological subtype was NBSCC in 614 (5%) patients vs UC in 11 697 (95%) patients.
• At RC, the rate of non-organ confined (NOC) BCa was higher in NBSCC patients than in their UC counterparts (71.7% vs 52.2%; P < 0.001).
• After adjustment for OCM, The 5-year cumulative CSM rates were 25.0% vs 19.8% (P= 0.2) for patients with NBSCC vs UC organ confined (OC) BCa, respectively. The same rates were 46.3% vs 49.3% in patients with NOC BCa (P= 0.1).
• In multivariable competing-risks analyses, histological subtype (NBSCC vs UC) failed to achieve independent predictor status of CSM in patients with OC (hazard ratio, 1.2; P= 0.06) or NOC BCa (hazard ratio, 1.1; P= 0.1).
• At RC, the rate of NOC BCa is higher in NBSCC patients than in their UC counterparts.
• Despite a more advanced stage at surgery, NBSCC histological subtype is not associated with a less favourable CSM than UC histological subtype, after accounting for OCM and the extent of the disease (OC vs NOC).
Radical cystectomy (RC) is the treatment of choice in patients with non-metastatic bladder cancer (BCa) [1–4]. At RC, urothelial carcinoma (UC) represents the most common histological subtype, followed by non-bilharzial squamous cell carcinoma (NBSCC) in Western countries [5–7]. The effect of histological subtype on survival after RC in patients with BCa remains unclear.
Two large studies have examined the effect of histological subtype on cancer-specific mortality (CSM). One showed the same CSM in NBSCC as in UC . The other study suggested more aggressive behaviour for patients with non-organ confined (NOC) NBSCC relative to NOC UC . Neither study accounted for the effect of other-causes mortality (OCM) on CSM. OCM is a reflection of comorbidity and represents an important determinant of overall mortality, especially in patients with BCa who are known to have multiple other health problems . Consequently, a lack of consideration of OCM may spuriously inflate the importance of BCa mortality.
The present study aimed to better examine the effect of histological subtype on CSM, after accounting for the effect of OCM. The study hypothesis was that patients with NBSCC have a less favourable CSM rate than their UC counterparts, especially in the presence of NOC BCa.
PATIENTS AND METHODS
We identified 12 311 patients who were treated with RC between 1988 and 2006 within 17 registries of the Surveillance, Epidemiology and End Results (SEER) database from the National Cancer Institute program. The quality of the SEER database is highly controlled and continuously improved . The 17 SEER registries include the Atlanta, Detroit, San Francisco–Oakland, Seattle–Puget Sound, Los Angeles, San Jose–Monterey metropolitan areas, as well as the states of Connecticut, Hawaii, Iowa, New Mexico, Utah, Alaska, Georgia, California, Kentucky, Louisiana and New Jersey. The SEER database represents a sample (26%) of the US population and is considered representative of the USA in terms of demographic composition, as well as of cancer incidence and mortality .
The BCa diagnostic code (International Classification of Disease for Oncology, Second edition; C67.0–C67.9) combined with RC treatment codes were used as the main selection criteria. Only non-metastatic patients aged 18 years or older with a histological subtype of NBSCC (diagnostic code 8051–8052 and 8070–8075) or UC (diagnostic codes 8120, 8122 and 8130, 8131) were considered (n= 12 311). Exclusions consisted of patients with unavailable disease stage or grade.
Independent sample t- and chi-squared tests were used to compare the statistical significance of differences in means and proportions, respectively. Cumulative incidence plots were used to graphically depict CSM and OCM rates. The Gray test assessed the statistical significance of differences in CSM and OCM rates .
Finally, univariable and multivariable competing-risks regression analyses were used to test the effect of histological subtype on CSM, after accounting for OCM. This type of analyses was used to avoid overestimation of CSM because censoring as a result of OCM may artificially reduce the pool of individuals at risk of CSM events . Covariates consisted of age, sex, race, pathological T and N stages, tumour grade and year of surgery. All analyses were stratified according to organ confined (OC) disease defined as pT1–2 N0 vs non-organ confined (NOC) disease defined as pT3–4 N0 or N+, irrespective of T stage.
All statistical tests were performed using the S-Plus Professional, version 1 (Mathsoft, Seattle, WA, USA) or SPSS, version 15.0 (SPSS, Chicago, IL, USA). P < 0.05 (two-sided) was considered statistically significant.
Between 1988 and 2006, 12 311 patients underwent RC for BCa within the SEER database (Table 1). Mean (median, range) age was 67.3 (68, 26–100) years. Most patients were men (73.5%) and Caucasian (89.8%). Most tumours were pT2 (38.6%) and grade III (52.6%). Most had no lymph node metastases (N0, 78.4%). Because of the inclusion of new SEER registries in the more contemporary years of the database, most patients were treated between years 2003 and 2006 (36.5%).
Table 1. Descriptive characteristics of 12 311 non-metastatic patients treated with a radical cystectomy for non-bilharzial squamous cell carcinoma (NBSCC) or urothelial carcinoma (UC) of the bladder cancer within 17 Surveillance, Epidemiology and End Results database registries from the National Cancer Institute program between 1988 and 2006
Overall (n= 12 311) (100%)
NBSCC (n= 614) (5%)
UC (n= 11 697) (95%)
9 047 (73.5)
8 717 (74.5)
3 264 (26.5)
2 980 (25.5)
11 053 (89.8)
10 516 (89.9)
1 258 (10.2)
1 181 (10.1)
1 600 (13)
1 578 (13.5)
4 755 (38.6)
4 581 (39.2)
3 547 (28.8)
3 287 (28.1)
2 409 (19.6)
2 251 (19.2)
9 656 (78.4)
9 162 (78.3)
2 655 (21.6)
2 535 (21.7)
1 093 (8.9)
6 480 (52.6)
6 219 (53.2)
4 592 (37.3)
4 520 (38.6)
Extent of disease
5 765 (46.8)
5 591 (47.8)
6 546 (53.2)
6 106 (52.2)
Year of surgery
2 210 (18)
2 090 (17.9)
2 383 (19.4)
2 254 (19.3)
3 227 (26.2)
3 049 (26.1)
4 491 (36.5)
4 304 (36.8)
NBSCC was recorded in 614 (5.0%) patients vs UC in 11 697 (95.0%) others. Pathological T stage distribution was statistically significantly different between the two histological subtypes (P < 0.001). The rate of pT1, pT2, pT3, and pT4 tumours was 3.6%, 28.3%, 42.3% and 25.7%, respectively, in NBSCC patients vs 13.5%, 39.2%, 28.1% and 19.2% in UC patients. The rate of NOC BCa was 71.7% in NBSCC vs 52.2% in UC patients (P < 0.001). Tumour grades III and IV were significantly (P < 0.001) less frequent in NBSCC patients (42.5% and 11.7%) vs UC patients (53.2% and 38.6%). Relative to UC patients, NBSCC patients were less frequently men (53.7% vs 74.5%; P < 0.001). A higher proportion of NBSCC patients were non-Caucasian (12.5% vs 10.1%; P= 0.05). Finally, less NBSCC patients (30.5%) were treated in the most contemporary year category (2003–2006) compared to UC patients (P= 0.02). No significant differences were recorded between NBSCC and UC patients, when age and LN stage were compared.
For patients with OC disease (Fig. 1A), the CSM rates at 1, 2 and 5 years were 9.6%, 15.3% and 25.0%, respectively, in NBSCC patients vs 6.4%, 12.2% and 19.8% in UC patients (P= 0.2). For patients with NOC disease (Fig. 1B), the CSM rates at 1, 2 and 5 years were 28.8%, 41.8% and 46.3%, respectively, in NBSCC patients vs 21.4%, 37.3% and 49.3% in UC patients (P= 0.8).
In multivariable competing-risks regression analyses that adjusted for the effect of OCM on CSM (Table 2), histological subtype (NBSCC vs UC) failed to achieve an independent predictor status for CSM in patients with OC (hazard ratio, HR, 1.2; P= 0.2) or NOC BCa (HR, 1.1; P= 0.1).
Table 2. Multivariable competing risks regression models for prediction of cancer-specific mortality (after accounting for non-cancer-related mortality) and other-cause mortality (after accounting for cancer-specific mortality) in patients with organ confined disease (T1N0, T2N0) and non-organ confined disease (T3N0, T4N0, Tany N+)
The study hypothesis proposed that, in patients who were treated with RC, those with NBSCC may have less favourable CSM rates than their UC counterparts. We tested this hypothesis in 12 311 patients who were treated with RC, within 17 SEER registries. The statistical analyses relied on competing-risks methodology, which quantifies CSM after accounting for OCM.
The results obtained showed that, at RC, patients with NBSCC harbour NOC disease more frequently than their UC counterparts (71% vs 52%; P < 0.001). However, the 5-year CSM rate was not statistically significantly higher in NBSCC patients relative to UC patients in either OC (25.0% vs 19.8%; P= 0.2) or NOC (46.3% vs 49.3%; P= 0.8) disease. This observation was further confirmed in multivariable competing-risks analyses that adjusted for age, sex, race, T stage, N stage, tumour grade and year of diagnosis. Histological subtype (NBSCC vs UC) failed to achieve statistical significance for the prediction of CSM in either OC (HR, 1.2; P= 0.02) or NOC (HR, 1.1; P= 0.1) disease (Table 2).
Taken together, the results obtained in the present study show that patients with NBSCC harbour a more advanced disease at RC than patients with UC BCa. However, after adjusting for stage, the CSM rates of patients with NBSCC vs UC BCa were almost the same.
Although the association between histological subtype and prognosis has been investigated in previous studies, to the best of our knowledge, the present study is the first analysis of CSM to takeinto account the effect of OCM using competing-risks methodology. Previously, Scosyrev et al. , also relying on the SEER database, reported that, in patients with NOC disease, NBSCC histological subtype was associated with a less favourable CSM than UC histological subtype. The results of the present study do not support these findings. In our analyses, when patients were stratified into OC vs NOC, individuals with NBSCC, BCa did not have a less favourable CSM. The differences between the results obtained in the present study and those of Scosyrev et al.  may stem from the different types of statistical method used. In the study by Scosyrev et al. , CSM rates were estimated without adjusting for OCM. A lack of adjustment for OCM may lead to an artificial overestimation of the number of individuals who died from BCa, as shown in a previous study . This may be the result of a censoring of non-cancer-related deaths. This phenomenon may in turn reduce the number of individuals at risk of CSM and may overestimate the importance of cancer-related deaths. The present study obviates the confounding effect related to the lack of accounting for comorbidity. The use of competing-risks modeling adjusts for OCM and provides the least unbiased estimates of CSM. This consideration is of vital importance, especially when OCM effect is very important, as in the case of individuals with BCa, who are known to have multiple comorbidities .
It is noteworthy that other studies focussing on the effect of histological subtype on CSM after RC failed to reach a consensus. Some studies report no association between histological subtype and CSM . Conversely, others report that NBSCC histological subtype increased the CSM rate by 5.2-fold (P= 0.002) . However, only one of these studies had a large sample size . All other studies had a small number of NBSCC patients (range 8–26). Such small populations may undermine the generalizability of the results.
In the present study cohort, the rate of UC was significantly lower in women vs men (26% vs 75%), corroborating previous studies [14,15]. Excessive environmental exposure to carcinogens such as tobacco and industrial chemicals in men has been proposed as an explanation for this . Similarly, other carcinogenic exposures, routes of entry, enzymatic processing of environmental substances, and cellular and physiological responses were proposed as other possible causes . However, there is no uniform theory to explain these observations. Conversely, in the present study cohort, the difference in NBSCC rate was not so evident between women and men (46% vs 54%). This implies that the mechanisms referred to may not attribute to the oncogenesis of NBSCC. However, being a women was associated with a less favourable CSM rate for both histological subtypes. Previous investigators observed similar results . Differences in treatment and hormonally-mediated differences in BCa biology may be contributing factors to the worse outcome of women than men [17–22].
It is noteworthy that more than half (53%) of the examined population had a NOC disease at RC. This percentage was even higher in patients with NBSCC (72%). These findings are worrisome and represent an important practical concern because the efficacy of RC substantially decreases for NOC disease . The higher rate of NOC BCa in both histological subtypes may be indicative of diagnostic and/or therapeutic delays. Alternatively, it may be indicative of a more aggressive natural history of NBSCC at presentation. Regardless of the underlying reasons for the presence of NOC BCa in a large proportion of patients with NBSCC, it is important to consider how to improve the current CSM rates in those individuals. Expedited work-ups and minimal delays before RC may partially improve the recorded patterns. Receipt of adjuvant or even neoadjuvant therapy may be of benefit in some patients .
The present study has limitations. For example, central pathology was unavailable. Thus, the assignment of NBSCC vs UC histological diagnosis may differ between centres. The same consideration may also apply to grade. These limitations are shared in numerous bladder cancer studies [25,26] and the lack of a central review represents a reality for most studies and almost all institutions that do not participate in clinical trials, where central review is required. The assignment of a cause of death may represent another source of bias. However, it is unlikely that the bias in the ascertainment of the cause of death is distributed differently between NBSCC and UC patients. Moreover, previous studies have shown that the cause of death assignment in the SEER database is valid . Finally, similar to other studies based on the SEER-limited database [25,26], we obtained no data with respect to neo-adjuvant, adjuvant or salvage chemotherapy.
In conclusion, at RC, patients with NBSCC harbour a NOC disease more frequently than UC patients. Despite a more advanced stage at surgery, the NBSCC histological subtype is not associated with a less favourable CSM than the UC histological subtype, after accounting for OCM and the extent of the disease (OC vs NOC).
Pierre I. Karakiewicz is partially supported by the University of Montreal Health Centre Urology Specialists, Fonds de la Recherche en Sante du Quebec, the University of Montreal Department of Surgery and the University of Montreal Health Centre (CHUM) Foundation