Upstaging of urothelial cancer at the time of radical cystectomy: factors associated with upstaging and its effect on outcome

Authors


Polat Turker, Namik Kemal University, Department of Urology, 100. Yil Mah. Tunca Cad. 59100, Tekirdağ, Turkey. e-mail: polatturker@gmail.com.

Abstract

Study Type – Therapy (case series)

Level of Evidence 4

What's known on the subject? and What does the study add?

The reported discordance between staging on transurethral bladder resection and on radical cystectomy pathology in the literature ranges from 20 to 80%.Correct staging in bladder cancer has direct implications for its management. The upstaging from organ-confined (OC) to non-organ-confined (nOC) disease has been reported in 40% of cases. Lymphovascular invasion (LVI) is a factor known to be associated with poor clinical outcome.

Pathological upstaging was observed in our cohort in 40% of cases and most cases (80%) were upstaged from OC to nOC disease. During the study period the frequency of upstaging observed increased. We found LVI (hazard ratio [HR]= 5.07, 95% CI = 3.0–8.3, P < 0.001) and any histological variant variant (HR = 2.77, 95% CI = 1.6–4.8, P < 0.001) to be strong independent predictors of upstaging. Patients with clinical T2 bladder cancer found with upstaging at the time of radical cystectomy had a poorer outcome than patients with no upstaging. Identification of patients at high risk of upstaging at radical cystectomy is key to improving their management and outcome.

OBJECTIVES

  • • To analyse the details of bladder cancer (BC) staging in a large combined radical cystectomy (RC) database from two academic centres.
  • • To study rate and time trends, as well as risk factors for upstaging, especially clinical factors associated with staging errors after RC.

PATIENTS AND METHODS

  • • Characteristics of patients undergoing RC at University Health Network, Toronto, Canada (1992–2010) and University of Turku, Turku, Finland (1986–2005) were analysed.

RESULTS

  • • Among 602 patients undergoing RC, 306 (51%) had a discordance in clinical and pathological stages. Upstaging occurred in 240 (40%) patients and 192 (32%) patients were upstaged from organ-confined (OC) to non-organ-confined (nOC) disease.
  • • During the study period, upstaging became more common in both centres.
  • • In multivariate analyses, T2 disease at initial presentation (P= 0.001, odds ratio [OR]= 2.62, 95% confidence interval [CI]: 1.44–4.77), high grade disease (P= 0.01, OR = 2.85, 95% CI: 1.21–6.7), lymphovascular invasion (LVI) (P < 0.001, OR = 5.17, 95% CI: 3.48–7.68), female gender (P= 0.038, OR = 0.6, 95% CI: 0.38–0.97, and histological variants (P < 0.001, OR = 2.77, 95% CI: 1.6–4.8) were associated with a risk of upstaging from OC to nOC disease.
  • • Upstaged patients had worse survival rates than patients with correct staging. This was especially significant among patients with carcinoma invading bladder muscle before undergoing RC (16% vs 46% 10-year disease-specific mortality, P < 0.001).

CONCLUSIONS

  • • Upstaging is a common problem and unfortunately no improvements have been observed during the last two decades.
  • • LVI and the presence of histological variants are strong predictors of upstaging at the time of RC.
  • • Pathologists should be encouraged to report LVI and any histological variant at the time of TURBT.
Abbreviations
BC

bladder cancer

RC

radical cystectomy

OC

organ-confined

nOC

non-organ-confined

LVI

lymphovascular invasion

TURBT

transurethral resection of bladder tumour

CIS

carcinoma in situ

ISUP

International Society of Urologic Pathology

UC

urothelial cancer

TUR

transurethral resection

PLND

pelvic lymph node dissection

HR

hazard ratio

INTRODUCTION

Bladder cancer (BC) has a high incidence and mortality rate with ≈ 297 300 new cases and ≈112 300 deaths in men in 2008 worldwide [1]. Clinical staging is a fundamental step, both in treatment planning and when counselling patients with BC. Clinical staging is based on the pathological examination of transurethral resection of bladder tumour (TURBT) specimens, bimanual physical examination under anaesthesia and cross-sectional imaging (CT, MRI) [2], and, despite modern techniques, discordance in clinical and pathological staging, which is based on pathology after radical cystectomy (RC), is common; the clinical staging of BC is notoriously incorrect. Reported rates of agreement between clinical and pathological staging vary from 20 to 80% [3–6].

The most significant implication of inaccurate staging can be observed when muscle invasion is not detected. As neoadjuvant chemotherapy has been shown to improve survival rates in patients with cT2-4 disease, an opportunity to improve outcome in these patients is then lost. This is particularly relevant with respect to upstaging from organ-confined (OC) to non-organ-confined (nOC) disease (T3 or N+), because some studies indicate that patients with nOC disease benefit the most from neoadjuvant chemotherapy [7,8].

The purpose of the present study was to analyse the details of BC staging in a large combined RC database from two academic centres. In addition to rate and time trends, we aimed to study risk factors for upstaging in our analysis.

PATIENTS AND METHODS

A total of 678 consecutive patients undergoing RC and lymphadenectomy for urothelial BC at two institutions (University Health Network, Toronto, Canada, 1992–2008 and University of Turku, Turku, Finland, 1986–2005) were retrospectively analysed. Both datasets were approved by institutional research ethics boards. Patients with non-urothelial BC (n= 46), or salvage RC after failed radiotherapy or neoadjuvant chemotherapy (n= 30) were excluded.

The following clinicopathological data were analysed: age, gender, initial BC presentation (date, grade, stage), intravesical therapies, date of surgery, extent of lymphadenectomy, clinical stage before RC, lymphovascular invasion (LVI), the presence of concomitant carcinoma in situ (CIS) and pathological stage after RC. Tumour grade was reported according to either WHO 1973 or WHO/International Society of Urologic Pathology (ISUP) 2004 classifications and stage was recorded according to 2002 TNM classification [9–11]. In the Turku cohort, all slides were reviewed and only the WHO/ISUP 2004 grading system was used in the analysis. The grade was determined by an experienced pathologist according the WHO/ISUP guidelines. In the Toronto cohort, original pathology reports were used and a systematic review of all slides was not performed. We chose to define all Grade 1–2 tumours as ‘low grade’ and only grade 3 tumours as ‘high grade’ tumours. Only urothelial cancers (UCs) were included in the analysis. In the Toronto cohort, additional urothelial histological variants (any histological subtype noted in the RC specimen in addition to UC) were recorded.

At the time of initial BC treatment, bimanual examination was performed before transurethral resection (TUR) under general anaesthesia. Abdominal and pelvic CT and chest radiography were performed routinely. All TURBT specimens and RC specimens were evaluated by genitourinary pathologists. When the TUR was performed at an outside institution, either TURBT slides were reviewed or re-TUR was performed at the study institutions. For patients with non-muscle-invasive BC, who were treated during the study period, the presence of muscle tissue was confirmed by the pathologist. The use of second-look TUR varied during the study period as this was not standard practice throughout the study.

Indications for RC included carcinoma invading the bladder muscle or non-muscle-invasive BC failing intravesical therapy. In addition to removal of bladder and distal ureters, RC included removal of seminal vesicles and the prostate in males and the uterus, ovaries and anterior vaginal wall in females. There was a significant difference in the institutional policy of pelvic lymph node dissection (PLND). In the Turku University Hospital cohort, PLND was not performed before 1995, and only PLND limited to the obturator nodes was performed from 1995. In the University of Toronto cohort, most of the patients underwent extended node dissection, the cranial boundary of dissection being either aortic bifurcation or crossing of ureters over the common iliac vessels.

Two definitions were used for upstaging: 1. Any upstaging: any increase in T- and/or N-stage when findings before cystectomy, i.e. clinical stage, were compared with those after cystectomy, i.e. pathological stage; 2. Upstaging from OC to nOC tumour: any cT≤2, cN0 tumour being upstaged to pT ≥ 3 or pathologically confirmed node-positive disease.

Postoperative follow-up was performed every 3 months for the first year, bi-annually until year 5 and annually thereafter. Patients were monitored based on the decision of the treating physician.

Follow-up time was defined as the period from date of RC to the patient's last visit, death, or recurrence. The time period between TURBT and RC was measured as the number of days from date of TURBT to date of RC. Statistical analyses were performed using SPSS version 17.0 (SPSS Inc., Chicago, IL, USA). Categorical variables were evaluated and compared using a chi-squared and Fisher exact test. Multivariate analysis was performed using logistic regression analysis. Disease-specific (DSS) and overall survival (OS) analyses were performed using the Kaplan–Meier method and log-rank statistic.

RESULTS

A total of 602 patients from two institutions (382 patients from University of Toronto and 220 patients from University of Turku) were analysed. Patient characteristics are shown in Table 1. The median (range) patient age was 67 (32–88) years and 78% were males. The primary stage at initial BC presentation was <T2 in 50% and ≥T2 in 50% of the patients and one third of patients received intravesical instillations. High grade disease was detected in 88% of the patients.

Table 1.  Basic characteristics of the study population
CharacteristicToronto, n= 382Turku, n= 220Total, n= 602
  • *

    Histology of TUR at the time of initial BC diagnosis.

Gender, n (%)   
 Female78 (20)54 (25)132 (22)
 Male304 (80)166 (75)470 (78)
Median (range) age69 (32–88)64 (35–80)67 (32–88)
Initial TUR histology*, n (%)   
 Ta, Tis84 (22)61 (28)145 (24)
 T1102 (27)54 (24)156 (26)
 T2196 (51)105 (48)301 (50)
Grade, n (%)   
 G1-2/low grade33 (9)37 (17)70 (12)
 G3/high grade349 (91)183 (83)532 (88)
Intravesical therapy, n (%)   
 None231 (60)165 (75)396 (66)
 Chemo / BCG151 (40)55 (25)206 (34)
Year of surgery, n (%)   
 2006–2010152 (40)0152 (25)
 2001–2005138 (36)46 (21)184 (30)
 1996–200060 (16)54 (24)114 (19)
 1991–199531 (8)63 (29)94 (16)
 1985–19901 (1)57 (26)58 (10)
Cystectomy pT-stage, n (%)   
 pT0/Ta/CIS64 (17)68 (31)132 (22)
 pT142 (11)55 (25)97 (16)
 pT286 (22)44 (20)130 (22)
 pT3131 (34)43 (20)174 (29)
 pT4a52 (14)8 (3)60 (10)
 pT4b8 (2)1 (1)9 (1)
Histological variant, n (%)   
 None292 (77)N/A292
 Squamous48 (13)N/A48
 Glandular13 (3)N/A13
 Other26 (7)N/A26
 Missing data3220223
Concomitant CIS, n (%)   
 No202 (53)143 (65)345 (57)
 Yes180 (47)77 (35)257 (43)
Median (range) no. of removed lymph nodes12 (1–62)7 (1–18)11 (1–62)
Lymph node invasion, n (%)   
 Negative275 (72)204 (93)479 (80)
 Positive107 (28)16 (7)123 (20)
LVI, n (%)   
 No243 (64)153 (70)396 (66)
 Yes139 (36)67 (30)206 (34)

Pathological stage distributions after RC were as follows; pT0 (37/602, 6%), CIS (54/602, 9%), pTa (45/602, 7%), pT1 (16%), pT2 (22%), pT3 (29%), and pT4 (11%). Concomitant CIS was detected in 43% and LVI was detected in 34% of the RC specimens. The median (range) number of removed lymph nodes was 11 (1–62) and 20% of patients undergoing lymphadenectomy had positive lymph nodes on pathological examination. The difference in lymph node invasion frequencies between the two centres can be attributed to the extent of lymphadenectomy performed.

In addition to UC, histological variants were routinely reported in the Toronto cohort. 87/379 (23%) of the patients in Toronto had additional histological variants (in addition to UC). The frequency of different variants was as follows: squamous (49/379, 13%), glandular (13/379, 3%), glandular and squamous (8/379, 2%), nested (6/379, 1.5%), sarcomatoid (4/379, 1%), neuroendocrine (4/379, 1%), lymphoepithelium-like variant (3/379, 1%) and micropapillary (1/379, 0.5%).

Discordance between clinical and pathological staging was noted in 306 (51%) cases. Of these patients, 240 (40%) were upstaged and 192 (32%) were upstaged from OC to nOC disease. The downstaging rate was 11% (n= 66) and 49% (n= 296) of patients were correctly staged. The pathological stages after RC for each clinical stage group are presented in Table 2. The upstaging rate of clinically non-muscle-invasive tumours (cCIS, cTa, cT1) to cT2 disease was 15% (38/255), most of which were initially cT1 tumours (87%, 33/38). Overall upstaging rates and OC to nOC upstaging rates of CIS and Ta tumours were relatively low (12% and 5%; 21% and 2% for CIS and Ta tumours respectively). Upstaging was most common in cT2 tumours for both overall upstaging and OC to nOC upstaging (74% and 57%, respectively).

Table 2.  Distribution of pathological stages within each clinical stage group Thumbnail image of

We also evaluated the rate of upstaging within different 5-year time periods (Table 3, Fig. 1). In both study centres, upstaging became more common during the study period but the difference seemed to be more prominent in the Toronto cohort.

Table 3.  Rate of upstaging by 5-year time period by study centre
Year of surgeryTorontoTurkuTotal upstaging*, n (%)
N Upstaging*, n (%) N Upstaging*, n (%)
  • *

    Defined as a clinically OC tumour being upstaged to a nOC tumour.

2006–201015279 (52)00NA
2001–200513865 (47)4611 (24)76 (41)
1996–20006019 (32)5413 (24)32 (28)
1991–19953111 (35)6314 (22)25 (27)
1985–199010576 (11)NA
Figure 1.

Trends in upstaging during the past 25 years.

Clinical factors affecting the risk of upstaging were studied in uni- and multivariate analysis. In the univariate analysis, the rate of upstaging was higher in patients with a primary T2 disease (P < 0.001), high grade tumours (P < 0.001), patients with no history of intravesical therapy before RC (P= 0.047), LVI (P < 0.001) or any histological variant (P < 0.001 [Table 4]). The median (range) time from decision to undergoing RC (last TURBT before RC) to the day of surgery was 69 (6–735) days. Interestingly, the time interval between TURBT and RC did not affect the rate of upstaging when the threshold was set at 60 days. Patients with lymphadenectomy had a higher rate of upstaging when compared with those who did not undergo PLND but the extent of lymphadenectomy had no effect on the rate of upstaging.

Table 4.  Factors associated with upstaging from OC to nOC tumour: univariate analysis
VariableOdds ratio (95% CI) P
  • Data of variant pathology was available only in University of Toronto cohort.

Gender  
 Female (Reference)1.0 
 Male1.53 (1–2.33)0.04
Grade  
 G1-2/low grade (Reference)1.0 
 G3/High grade4.14 (2.07–8.28)<0.001
Presence of CIS  
 No1.0 
 Yes0.7 (0.51–1.01)0.06
Time to RC (336 patients)  
 ≤60 Days1.0 
 >60 Days0.84 (0.54–1.29)0.43
LVI  
 No (Reference)1.0 
 Yes5.77 (4–8.34)<0.001
Histological variant  
 No (Reference)1.0 
 Yes2.83 (1.72–4.66)<0.001
Number of lymph nodes removed during cystectomy  
 01.0 
 1–101.41 (0.92–2.15)0.1
 >101.61 (0.99–2.63)0.054

In multivariate analyses, T2 disease at initial presentation (P= 0.001, hazard ratio [HR]= 2.62, 95% CI: 1.44–4.77), high grade disease (P= 0.01, HR = 2.85, 95% CI: 1.21–6.7), LVI (P < 0.001, HR = 5.17, 95% CI: 3.48–7.68), female gender (P= 0.038, HR = 0.6, 95% CI: 0.38–0.97) and having received intravesical therapy before RC (P= 0.04, HR = 1.7, 95% CI: 1.02–2.84) were independent predictors of upstaging from OC to nOC disease (Table 5). The multivariate analysis was repeated for the Toronto cohort including variant histology as a variable. After LVI (HR = 5.07, 95% CI = 3.0–8.3, P < 0.001), variant histology was the second most important risk factor for upstaging (HR = 2.77, 95% CI: 1.6–4.8, P < 0.001).

Table 5.  Multivariate analysis of pre-RC factors for prediction of extravesical disease at the time of cystectomy
VariableOdds ratio (95% CI) P
  1. *Initially non-muscle-invasive tumours treated with intravesical instillations undergoing RC for recurrent and/or progressive disease. †Multivariable analysis repeated in the cohort (Toronto) using histological variants data.

Primary stage  
 Ta, CIS (Reference)1.0 
 T11.71 (0.95–3.06)0.07
 T22.62 (1.44–4.77)0.001
IV therapy before RC*  
 No (Reference)1.0 
 Yes1.7 (1.02–2.84)0.04
Presence of CIS  
 No1.0 
 Yes0.8 (0.54–1.18)0.26
Grade  
 G1-2/low grade (Reference)1.0 
 G3/High grade2.51 (1.21–5.21)0.01
LVI  
 No (Reference)1.0 
 Yes5.17 (3.48–7.68)<0.0001
Gender  
 Female (Reference)1.00.038
 Male1.6 (1.02–2.62) 
Histological variant  
 No (Reference)1.0<0.0001
 Yes2.77 (1.6–4.8) 

The median (mean [sd]; range) follow-up after RC was 27 (46 [50]; 0–264) months. DSS and OS rates for patients with and without upstaging are shown in Fig. 2. Although statistically significant, the effect of upstaging in patients with <T2 disease was relatively modest (Fig. 2A,B). By contrast, patients with clinical T2 disease with upstaging had a dramatically poorer outcome when compared with patients with no upstaging (Fig. 2C,D).

Figure 2.

OS and DSS rates in patients with clinical <cT2 (A and B) and cT2 diseases (C and D) with and without upstaging from OC to nOC disease.

DISCUSSION

Clinical staging based on physical examination, transurethral resection pathology and imaging are the most important factors for predicting patient outcome and treatment planning. Unfortunately, the ability to predict pathological stage from clinical stage in BC remains limited. In the present paper, we analysed the results of clinical staging and factors associated with upstaging in two academic centres. Our results show that incorrect clinical staging is common and furthermore, there has been no improvement in the past couple of decades. Thus, methods for improving staging accuracy should remain high on the agenda of physicians treating BC. Clinical cTa and interestingly, cCIS tumours had the lowest upstaging ratio compared with other clinical stages. Upstaging was most common among patients with cT2 tumours (by contrast to <T2 cases) and among these cases, upstaging also had the greatest effect on survival. Our analyses recognized several additional factors associated with upstaging.

The reported discordance between TUR and RC pathology in the literature is ≈20 to 80% [3–6,12]. Svatek et al. [5] recently reported the ratio of clinical understaging as 50% in their cohort, which was the largest in the literature. In accordance with previous studies, we found that the agreement between clinical staging from a TUR and pathological staging after RC was only 49%. The pathological upstaging ratio in our cohort was 40% and most of these patients (80%) were upstaged from OC to nOC disease. In other series, the upstaging ratio from OC to nOC disease was reported as 36–43% [3,5].

Unfortunately, there has been no evidence of improved staging accuracy during the last 25 years [3,5]. Even worse, in accordance with previous results, we found a gradual increase in the rate of upstaging over time. It is difficult to estimate the reason for this negative trend, but possibly, this reflects a stage migration to more advanced cases (i.e. less early cystectomies) recently, as we found that clinically muscle-invasive cases were at the highest risk of being upstaged. On the other hand, restaging TUR, a recent strongly advocated manoeuvre, should have improved our staging results [13,14]. During the study period, restaging TUR was not frequently practised and therefore could not be included in our analyses. Despite technological improvements, imaging studies are unfortunately still inaccurate, both in staging of primary tumour as well as in nodal staging [2].

It is well documented that patients who undergo early RC for high-risk superficial disease and progress to muscle invasion have a worse prognosis than patients who present with muscle invasion initially [15,16]. In our study, the upstaging ratio of clinically non-muscle-invasive tumours (<cT2) to muscle-invasive tumours was 15% (38/255), most of which were initially T1 (87% [33/38]). This indicates that a significant proportion of patients were spared early RC because of clinical staging error. More accurate staging can be accomplished by performing re-TUR for patients with clinical cT1 disease, as indicated before [17–19]. When we compared patients who had clinically no evidence of muscle invasion but had muscle invasion in the final cystectomy pathology and patients who had no muscle invasion in the final pathology, the first group had worse DSS and OS rates (P= 0.01 and P= 0.02, respectively). By contrast, tumours that were clinically OC and muscle-invasive and turned out to be nOC after RC had much worse DSS and OS results than the patients whose tumours remained OC. Because of understaging, these patients did not receive the neoadjuvant chemotherapy that is associated with a potential benefit for this group of patients. Available data shows an absolute survival benefit from neoadjuvant chemotherapy of ≈ 5% for patients undergoing RC [20,21]. Only 1.2% of patients with stage III BC actually receive neoadjuvant chemotherapy [22].A meta-analysis of >3000 patients from neoadjuvant studies showed a statistically superior survival benefit from neoadjuvant cisplatin-based combination chemotherapy compared with surgery alone [23].

In general, pathological staging has the strongest predictive value for patient outcome. Unfortunately these data are only available after RC; therefore, we sought to identify factors that may be helpful in identifying patients at a higher risk of upstaging. LVI at TURBT and in RC specimens was reported to be strongly associated with poor clinical outcome [24–28]. In accordance with previous results, we found LVI to be a strong independent predictor of upstaging (P < 0.001) [28,29]. To our knowledge, this is the first study demonstrating the effect of histological variants on pathological upstaging. The most frequent variants are squamous and/or glandular differentiation, which were observed in 10–60% of the patients [30]. It is well documented that variant histology portends a worse prognosis than pure UC [31]. Honma et al. [32] reported squamous differentiation within UC as an independent predictor of local recurrence. In the present cohort, the frequency of histological variants was as high as 70%. This is probably a reflection of dedicated uro-pathologists examining all surgical specimens at our institution. Squamous variant was the most frequent, accounting for 57% of the cases with histological variants. On multivariate analysis, variant histology was an independent predictor of upstaging (HR = 2.77, 95% CI = 1.6–4.8, P < 0.001).

One of the most important factors for upstaging should be the period of time between the last TUR to cystectomy, but we did not find any significant effect of this factor in univariate analyses. One of the main limitations of the present study was that data were available for 336 patients only.

The present study is also limited by the fact that it is retrospective in nature and has a long study time period. During this period, some of the clinical practices have changed. Patients with non-muscle-invasive BC who were treated during the early period of the study did not necessarily benefit from a re-TUR because this was not the standard practice. Also in the earlier period, imaging was not as accurate as it is today, although even modern imaging is far from perfect, especially for T-stage determination.

Another limitation was the lack of systemic pathology review in the Toronto cohort, although all the pathological slides in the Turku cohort were reviewed by an experienced pathologist. During the course of the study, the BC grading system was changed and the 2004 WHO/ISUP classification was defined after the 1973 WHO classification. In the present study, the definitions ‘low grade’ for all grade 1–2 and ‘high grade’ for all grade 3 tumours were used although these definitions can be imperfect as some grade 2 tumours are actually high grade tumours when the 2004 system is used.

Based on the results of the present study, female patients, cT2 and/or high grade disease reported on TUR pathology, LVI and histological variants should alert the clinician to the risk of upstaging. Treatment decisions may further be refined by considering the upstaging risk.

The present study reports a high rate of inaccuracies in clinical staging of BC before RC. Patients with cT2 disease were most likely to experience upstaging and, in this group of patients, upstaging had the most dramatic effect on survival. Tumours with LVI or histological variants were noted to be at a very increased risk of upstaging. Pathologists should be encouraged to report LVI and any histological variant in the TURBT pathological reports as it has a direct impact on patients staging and prognosis.

CONFLICT OF INTEREST

None declared.

Ancillary