Nonprimary pT1 Nonmuscle Invasive Bladder Cancer Treated

With Bacillus Calmette-Guerin is Associated With Higher Risk of

Progression Compared to Primary T1 Tumors

Sultan S. Alkhateeb, Bas W. G. Van Rhijn, Antonio Finelli,Theodorus van der Kwast, Andrew Evans, Sally Hanna, Rati Vajpeyi,Neil E. Fleshner, Michael A. S. Jewett and Alexandre R. Zlotta*

From the Division of Urology, Department of Surgical Oncology (SSA, BWGVR, AF, SH, RV, NEF, MASJ, ARZ), Department of Pathology,Princess Margaret Hospital, University Health Network (TVDK, AE), and Mount Sinai Hospital, University of Toronto (ARZ), Toronto, Canada

Purpose: Few studies have examined the prognostic significance of prior tumorresection(s) in cases of T1 nonmuscle invasive bladder cancer treated with intra-vesical bacillus Calmette-Guerin. We examined this issue by comparing theprognosis of primary vs nonprimary T1 nonmuscle invasive bladder cancertreated with bacillus Calmette-Guerin.Materials and Methods: Patients with pT1 nonmuscle invasive bladder cancertreated with bacillus Calmette-Guerin were identified and tumor pathology wasreviewed. Patients were then stratified into primary vs nonprimary tumors, andoutcomes were compared using univariate, multivariate and Kaplan-Meier sur-vival analyses, and the Cox regression model adjusting for various clinical andpathological features including, age, gender, tumor size, multifocality, patholog-ical grade and associated carcinoma in situ.Results: The study included 191 patients, 95 (49.7%) with primary and 96(50.3%) with nonprimary tumors. The clinical and pathological characteristicswere comparable. For the primary vs the nonprimary group progression rateswere 24.2% vs 39.6%, respectively (HR 2.07, 95% CI 0.98 –3.71, multivariatep � 0.03) and the 5-year progression-free survival rates were 71.9% vs 51.5%,respectively (log rank p �0.001). This difference remained significant onmultivariate Cox regression analysis (HR 2.53, 95% CI 1.40 – 4.57, p � 0.002).There was no difference between the groups in recurrence or disease specificmortality.Conclusions: Nonprimary T1 nonmuscle invasive bladder tumors treated withbacillus Calmette-Guerin carry a significantly higher risk of progression to mus-cle invasive disease compared to primary tumors. This information may be usedin combination with other prognostic factors to identify those at high risk forprogression when counseling patients.

Key Words: urinary bladder neoplasms, disease progression, recurrence,

Abbreviations

and Acronyms

BCG � bacillus Calmette-Guerin

CIS � carcinoma in situ

DSS � disease specific survival

NMI-BC � nonmuscle invasivebladder cancer

PFS � progression-free survival

RFS � recurrence-free survival

Submitted for publication November 15, 2009.Study received local research and ethics

board approval.* Correspondence: Division of Urology, Depart-

ment of Surgery, Mount Sinai Hospital and Univer-sity Health Network, University of Toronto, 60 Mur-ray St., 6th Floor, Box 19, Toronto, Ontario M5T 3L9,Canada (telephone: 416-586-4800 ext. 3910; FAX:416-586-4776; e-mail: azlotta@mtsinai.on.ca).

For other articles on a related

topic see pages 336 and 344.

mortality, BCG vaccine

THE management of nonmuscle inva-sive bladder cancer treated with in-travesical BCG is challenging, espe-cially since it is difficult to predict the

response to this therapy on an indi-

0022-5347/10/1841-0081/0THE JOURNAL OF UROLOGY®

© 2010 by AMERICAN UROLOGICAL ASSOCIATION EDUCATION AND RES

vidual basis. This is particularly truein high grade T1 disease as it has ahigher risk of progression to poten-tially lethal bladder cancer. Various

prognostic factors including clinical,

Vol. 184, 81-86, July 2010Printed in U.S.A.

EARCH, INC. DOI:10.1016/j.juro.2010.03.022www.jurology.com 81

PROGRESSION OF NONPRIMARY VERSUS PRIMARY pT1 NONMUSCLE INVASIVE BLADDER CANCER82

pathological, molecular and immunological parame-ters have been described for NMI-BC treated withBCG.1 The prognostic value of some of these factorshas been established while the value remains con-troversial for others. T1 high grade tumors repre-sent a mixed bag comprising tumors presenting forthe first time as such but also tumors which origi-nally presented as Ta or CIS and later progressed tohigh grade lamina propria invading tumors.

While several studies examined the prognosticsignificant of prior tumor resection(s) in NMI-BC,and showed a significant relation to recurrence,2,3

progression,4 and recurrence and progression,5 oth-ers have failed to show such a correlation.6–10 More-over most of these studies included all pT stages ofNMI-BC collectively and did not focus on T1 tumorsonly. However, we believe there are strong argu-ments to analyze T1 tumors as a separate entity dueto their different behavior compared to Ta and Tis.We asked whether the response to BCG therapy andrisk of recurrence and/or progression to muscle in-vasive disease is different in primary vs nonprimaryhigh grade T1 bladder tumors.

MATERIALS AND METHODS

Patients and FollowupAfter obtaining approval from our local research and eth-ics board we retrospectively identified all patients withpT1 disease treated with intravesical BCG at our institu-tion (University Health Network, University of Toronto)between 1990 and 2008. After excluding those patientswho underwent upfront radical cystectomy we divided ourcohort into 2 main groups of primary—those patients di-agnosed with pT1 disease as the first presentation of blad-der cancer, and nonprimary—those with a history of lowerstage NMI-BC who had progression from pTa and/or CISto pT1 disease during followup. Patients with a previousmuscle invasive tumor were not considered in the analy-sis. We then recorded clinical and pathological variablesincluding age at diagnosis of T1 disease, gender, tumorgrade, size, multiplicity and associated CIS.

Repeat transurethral bladder tumor resection for pa-tients with pT1 disease has been the standard of care atour institution since 2000. All patients were followed witha routine history, physical examination, cystoscopy, urinecytology every 3 months and upper tract imaging on anannual basis. Followup was calculated starting from theday of diagnosis of pT1 disease to the date of the lastencounter with the patient or death. Intravesical BCGwas given to all patients at a dose of 81 mg in 50 cc normalsaline once weekly for 6 weeks as induction. MaintenanceBCG was given at the discretion of the treating urologist.

Pathological EvaluationAll pathological specimens were processed and read byexperienced genitourological pathologists. A standard re-

porting template, which includes reporting of secondary

pattern of disease and whether detrusor muscle was sam-pled, is used at our pathology department, and we rou-tinely send all slides from outside hospitals for review byour pathologists as well. We used the WHO 1973 gradingsystem and a pathological review of tumor grade wasperformed for all patients.

Statistical AnalysisThe clinical and pathological features as well as recur-rence (defined as recurrence of at least pT1 disease), pro-gression (defined as progression to muscle invasive dis-ease, ie pT2 or higher) and disease specific mortality rateswere compared between the groups using univariate andmultivariate logistic regression analyses with p � 0.05considered statistically significant. RFS, PFS and DSSwere compared between the groups using Kaplan-Meiersurvival curves. The log rank test was used to comparestrata with p � 0.05 considered statistically significant. Amultivariate Cox regression analysis adjusting for all clin-ical and pathological variables, and primary vs nonpri-mary disease was used to identify factors associated withPFS, RFS and DSS. Statistical analysis was performedusing SPSS® version 16.

RESULTS

A total of 191 patients were identified from ourdatabase, of which 148 (77.5%) were male and 43(22.5%) were female. Mean age at diagnosis was68.5 years and median followup was 48 months.Based on previous history of bladder tumor 95(49.7%) patients were classified as having primarytumors and 96 (50.3%) as having nonprimary tu-mors. In the nonprimary group and before progres-sion to pT1 disease 50 patients (52.1%) received atleast 1 induction course of intravesical BCG, and ofthese 22 (22.9%) received at least 1 maintenanceintravesical treatment for various indications in-cluding concomitant CIS, large tumors or multiplerecurrences. Median number of BCG courses was 2.The time between receiving the last BCG therapyand progression to pT1 ranged from 2 to 225 months(median 19).

On univariate analysis both groups were compa-rable in terms of age, gender, median followup, tu-mor grade, multifocality and associated CIS, whilethe only difference was in the higher percentage oflarger tumors (greater than 3 cm) in the primarygroup (53.7%) vs the nonprimary group (33.3%,p � 0.006, table 1). The rate of detrusor musclesampling at pT1 diagnosis was 59% for nonprimarytumors vs 63% for primary tumors (p � 0.7).

There were 7 patients (7.3%) in the primary groupvs 9 (9.4%) in the nonprimary group who underwentdeferred cystectomy for BCG refractory pT1 disease(p � 0.51) with a mean time to cystectomy of 11.5months (median 8). The percentage of patients lost

to followup was 4.2% in the primary group vs 9.4%

PROGRESSION OF NONPRIMARY VERSUS PRIMARY pT1 NONMUSCLE INVASIVE BLADDER CANCER 83

in the nonprimary group with no statistically signif-icant difference.

Relationship to Progression

The progression rate for the primary group was24.2% vs 39.6% for the nonprimary group. This dif-ference was statistically significant on univariateanalysis (HR 2.05, 95% CI 1.10–3.82, p � 0.02), andon multivariate logistic regression analysis adjust-ing for age, gender, grade, tumor size, multifocalityand associated CIS (HR 2.07, 95% CI 0.98–3.71,p � 0.03). The 5 and 10-year PFS in the primarygroup was 71.9% and 66.7%, respectively, vs 51.5%and 36.4% for the nonprimary group. Kaplan-Meiersurvival curves showed a statistically significant dif-ference in PFS between the groups (log rank p �0.001,fig. 1). This difference remained significant in a mul-tivariate Cox regression analysis adjusting for otherclinicopathological features (HR 2.53, 95% CI 1.40–4.57, p � 0.002, table 2).

Relationship to Recurrence

The percentage of patients who experienced 1 ormore recurrence(s) in the primary vs the nonpri-mary group was 67.7% vs 65.3%, respectively. Thisdifference was not statistically significant on uni-variate (HR 1.12, 95% CI 0.61–2.03, p � 0.72) andmultivariate logistic regression analysis adjustingfor other clinicopathological features of the dis-ease (HR 1.26, 95% CI 0.65–2.42, p � 0.49). The 5and 10-year RFS for the primary group was 32.5%and 20.7%, respectively, vs 27.5% and 18.3%, re-spectively, for the nonprimary group, and onKaplan-Meier survival curves there was no differ-

Table 1. Comparison of clinical and pathological features

Primary Nonprimary p Value

No. gender (%):M 78 (82.1) 70 (72.9)F 17 (17.9) 26 (27.1) 0.16

Mean age (�SD) 69.9 (10) 67.1 (12.4) 0.095Median mos followup (�SD) 61.9 (41.4) 55.3 (44.5) 0.29No. cm size (%):

Less than 3 44 (46.3) 64 (66.7)Greater than 3 51 (53.7) 32 (33.3) 0.006

No. multiplicity (%):Solitary 53 (55.8) 54 (56.2)Multiple 42 (44.2) 42 (43.8) 1.0

No. grade (%):1 0 (0) 2 (2.1)2 31 (32.6) 25 (26.0)3 64 (67.4) 69 (71.9) 0.054

No. CIS (%):Absent 61 (64.2) 74 (77.1)Present 34 (35.8) 22 (22.9) 0.58

No. lost to followup (%) 4 (4.2) 9 (9.4) 0.25No. deferred cystectomy for pT1 7 (7.3) 9 (9.4) 0.51

disease (%)

ence in RFS between the groups (log rank p � 0.56,fig. 2).

Relationship to Disease Specific Mortality

The disease specific mortality rates for the pri-mary vs the nonprimary group were 14.6% vs11.6%, respectively, which also did not show astatistically significant difference on univariate(HR 1.30, 95% CI 0.56 –3.04, p � 0.54) and multi-variate analyses (HR 1.45, 95% CI 0.58 –3.60). The5 and 10-year DSS for the primary group was83.3% and 80.2%, respectively, vs 79.4% and77.2%, respectively, for the nonprimary group, and onKaplan-Meier curves there was no difference in DSS

Figure 1. Kaplan-Meier survival curves for difference in PFSbetween nonprimary and primary (prim) tumors.

Table 2. Cox regression analysis for factors associatedwith PFS

HR 95% CI p Value

Primary 1 (reference)Nonprimary 2.53 1.40–4.57 0.002Size (cm):

Less than 3 1 (reference)Greater than 3 1.01 0.59–1.73 0.95

Multiplicity:Solitary 1 (reference)Multiple 1.33 0.78–2.26 0.28

Grade:1 1 (reference)2 0.95 0.52–1.75 0.893 1.72 0.21–14.1 0.61

CIS:Absent 1 (reference)Present 1.21 0.64–2.26 0.56

Age at diagnosis 1.01 (for IQR) 0.98–1.03 0.51Gender:

M 1 (reference)

F 1.13 0.60–2.11 0.71

PROGRESSION OF NONPRIMARY VERSUS PRIMARY pT1 NONMUSCLE INVASIVE BLADDER CANCER84

between the groups (log rank p � 0.36, fig. 3). Therewere 25 disease related mortalities in total, 6 fromT1 disease while the remaining 19 died of muscleinvasive disease.

DISCUSSION

Our study demonstrates that the prognosis andprogression rate of primary T1 tumors treatedwith BCG is different from that of nonprimarytumors. Indeed patients with a history of tumorresection for Ta disease or CIS before presentingwith a T1 tumor had higher progression ratesafter intravesical BCG on univariate and multi-variate analysis, and this finding supports theprognostic significance of prior tumor resection(s)in this subset of patients with high risk nonmuscleinvasive disease. Recurrence and disease specificmortality as well as RFS and DSS were not differ-ent between the 2 groups of primary and nonpri-mary tumors. Our observation has direct clinicalimplications because as urologists dealing with T1tumors that previously presented at a lower stageor as CIS, we should be careful because the risk ofprogression after BCG is high in these cases,higher than in cases of primary T1.

Fernandez-Gomez et al demonstrated the prog-nostic significance of prior tumor resection(s) inrelation to recurrence and progression.5 However,in their large cohort of patients with NMI-BC,including 848 with pT1 disease, the analysis wasperformed for all patients including those withpTa disease. We believe that pT1 tumors are adifferent disease in terms of biology and prognosis,and should be examined as a separate entity.

Few authors have analyzed T1 tumors sepa-

Figure 2. Kaplan-Meier survival curves for difference in RFSbetween nonprimary and primary tumors.

rately. Hurle et al showed a relation between prior

tumor resection(s) and progression on univariateanalysis in 51 patients.4 Cookson and Sarosdy (86patients)8 and Pansadoro et al (81 patients)9 didnot reveal any prognostic significance for priortumor resection(s). Our study has the largestnumber of patients of those examining this issuein T1 disease as a separate entity on multivariateanalyses adjusting for other important prognosticfeatures of the disease and, thus, we believe ourfindings are important.

Fernandez-Gomez5 and Hurle4 et al reported ona significant correlation between progression andprior tumor resection(s), and our findings confirmthis correlation. The mechanism behind the ob-served difference in progression is not well under-stood. A possible explanation was given by Schrieret al with the hypothesis that in high risk super-ficial disease therapy sensitive and insensitivecells coexist, and intravesical therapy may selectsensitive clones leaving more aggressive tumorcells to continue growing and progressing.11 In ourcohort 52.1% of patients in the nonprimary grouphave been exposed to at least 1 BCG treatment,putting them at a higher risk for progressionbased on this hypothesis.

Another possible explanation could be the findingof El-Abbady et al that after meticulous histopatho-logical examination patients who underwent previ-ous transurethral resections had significantly morelocal spread of malignant cells into the bladder mus-cle vs those with primary tumors.12 Alternativelysome patients with prior resection(s) may have haddisease under staged at first which is not uncommonin T1 disease.13,14 Therefore, they were not treatedadequately and had a higher chance for progres-

Figure 3. Kaplan-Meier survival curves for difference in DSS

between nonprimary and primary tumors.

PROGRESSION OF NONPRIMARY VERSUS PRIMARY pT1 NONMUSCLE INVASIVE BLADDER CANCER 85

sion. A correlation between prior tumor resec-tion(s) and recurrence was reported by Losa,2

Takashi3 and Fernandez-Gomez5 et al. However,the results in the present study as well as those of5 others do not support this correlation.6 –10 Therewere only 2 previous studies that reported on thecorrelation between disease specific mortality andprior tumor resection(s). Takashi et al found asignificant correlation with prior tumor resec-tion(s)3 while Pansadoro et al as well as thepresent study did not find such a correlation.9

There are different clinical, pathological and mo-lecular prognostic parameters described in the liter-ature for each domain of the disease outcome (ierecurrence, progression and mortality), and someparameters were significantly associated with 1 spe-cific outcome and not with the others. For example,prior tumor resection(s) was associated with recur-rence but not with progression by Losa et al,2 whileit was associated with progression and not recur-rence by Hurle et al.4 Our findings are in keepingwith this phenomenon, possibly because each out-come has a different pathogenesis associated withcertain prognostic parameters. It is also evident thatthere is discrepancy among the results of these stud-ies. However, the prognostic importance of variousfactors is not always the same across publications.15

This may be due to differences in the choice of vari-ables analyzed or their coding, the multivariateanalysis used, the correlation between factors or theend point assessed.16

We believe that predicting the outcome of T1 tu-mors treated with BCG is still challenging and thatit makes sense to combine several prognostic factorstogether rather than use a single one. Prior tumorresection(s) has an added prognostic value to theother established factors in terms of predicting pro-gression. This may help in identifying those tumors

with a higher probability of progression, eventually

REFERENCES

Calmette-Guerin (Tokyo 172 strain) therapy: sig- Calmette-Guerin treatment

treating them with more aggressive therapy such asearly cystectomy.

A potential bias of our findings in a series whereBCG is widely used even in low risk patients isthat the nonprimary T1 tumors are in a selectedpopulation in which by definition previous BCGtherapy has failed and, therefore, was at higherrisk for the subsequent failure of BCG. However,our series represents real-life data and is close tousual practice. In addition, because age, gender,median followup, tumor grade, multifocality andassociated CIS were evenly distributed in primaryand nonprimary groups, this further supports theimportance of prior resections as a risk factor forprogression.

There are a few other limitations to this study. Itis a retrospective study with inherent limitationssuch as selection bias. In addition, the median fol-lowup was 48 months, which is relatively short, andmedian time to progression in the primary grouphas not yet been reached. However, most of thepatients are still being followed and we are planningto update our results with a longer followup in thefuture. Another potential limitation is that we are atertiary referral center and our cohort in generalmay represent a selected type of patient due to re-ferral patterns.

CONCLUSIONS

Patients who had 1 or more prior tumor resection(s)with pTa and/or CIS preceding a pT1 tumor have ahigher risk of progression after BCG therapy com-pared to those with primary T1 tumors. This infor-mation may be used in combination with other es-tablished clinical, pathological, immunological andmolecular prognostic factors to predict individualoutcomes, and to recommend more aggressive treat-ment for those with a higher likelihood of progres-

sion.

1. Saint F, Salomon L, Quintela R et al: Do prognos-tic parameters of remission versus relapse afterbacillus Calmette-Guerin (BCG) immunotherapyexist? Analysis of a quarter century of literature.Eur Urol 2003; 43: 351.

2. Losa A, Hurle R and Lembo A: Low dose ba-cillus Calmette-Guerin for carcinoma in situ ofthe bladder: long-term results. J Urol 2000;163: 68.

3. Takashi M, Wakai K, Hattori T et al: Multivariateevaluation of factors affecting recurrence, pro-gression, and survival in patients with superficialbladder cancer treated with intravesical bacillus

nificance of concomitant carcinoma in situ. IntUrol Nephrol 2002; 33: 41.

4. Hurle R, Losa A, Manzetti A et al: Intravesicalbacille Calmette-Guerin in stage T1 grade 3 blad-der cancer therapy: a 7-year follow-up. Urology1999; 54: 258.

5. Fernandez-Gomez J, Solsona E, Unda M et al:Prognostic factors in patients with non-muscle-invasive bladder cancer treated with bacillusCalmette-Guerin: multivariate analysis of datafrom four randomized CUETO trials. Eur Urol2008; 53: 992.

6. Akaza H, Hinotsu S, Aso Y et al: Bacillus

of existing papillary

bladder cancer and carcinoma in situ of thebladder. Four-year results. The Bladder CancerBCG Study Group. Cancer 1995; 75: 552.

7. Torrence RJ, Kavoussi LR, Catalona WJ et al:Prognostic factors in patients treated with intra-vesical bacillus Calmette-Guerin for superficialbladder cancer. J Urol 1988; 139: 941.

8. Cookson MS and Sarosdy MF: Management ofstage T1 superficial bladder cancer with intraves-ical bacillus Calmette-Guerin therapy. J Urol1992; 148: 797.

9. Pansadoro V, Emiliozzi P, de Paula F et al: Long-term follow-up of G3T1 transitional cell carci-

noma of the bladder treated with intravesi-

PROGRESSION OF NONPRIMARY VERSUS PRIMARY pT1 NONMUSCLE INVASIVE BLADDER CANCER86

cal bacille Calmette-Guerin: 18-year experience.Urology 2002; 59: 227.

10. Okamura T, Tozawa K, Yamada Y et al: Clini-copathological evaluation of repeated coursesof intravesical bacillus Calmette-Guerin instil-lation for preventing recurrence of initially re-sistant superficial bladder cancer. J Urol 1996;156: 967.

11. Schrier BP, Hollander MP, van Rhijn BW et al:Prognosis of muscle-invasive bladder cancer: dif-ference between primary and progressive tu-mours and implications for therapy. Eur Urol

2004; 45: 292.

12. El-Abbady AA, Shoukry MS, Hanno AG et al:Repeated transurethral resection of recurrent su-perficial bladder tumors– does it affect the spreadand stage of the tumor? Scand J Urol Nephrol2002; 36: 60.

13. Cheng L, Neumann RM, Weaver AL et al:Grading and staging of bladder carcinoma intransurethral resection specimens. Correlationwith 105 matched cystectomy specimens. Am JClin Pathol 2000; 113: 275.

14. Dutta SC, Smith JA Jr, Shappell SB et al: Clinical

under staging of high risk nonmuscle invasive

urothelial carcinoma treated with radical cystec-tomy. J Urol 2001; 166: 490.

15. Millan-Rodriguez F, Chechile-Toniolo G, Salvador-Bayarri J et al: Multivariate analysis of the prog-nostic factors of primary superficial bladder can-cer. J Urol 2000; 163: 73.

16. Sylvester RJ, van der Meijden AP, Oosterlinck Wet al: Predicting recurrence and progression inindividual patients with stage Ta T1 bladder can-cer using EORTC risk tables: a combined analysisof 2596 patients from seven EORTC trials. Eur

Urol 2006; 49: 466.