BJU International (2), 85, 824±828 Long-term follow-up of noninvasive bladder tumours (stage Ta): recurrence and progression K. ZIEGER, H. WOLF, P.R. OLSEN* and K. HéJGAARD{ Department of Urology, Aarhus University Hospital, Skejby, *Department of General Surgery, Holbñk Central Hospital and {Department of Pathology, Copenhagen University Hospital, Hvidovre, Denmark Objective To evaluate long-term recurrence-free and progression-free survival of noninvasive bladder tumours (stage Ta), and the signi cance of simple risk factors, including concomitant epithelial dysplasia. Patients and methods The study included 27 patients with primary noninvasive bladder tumour (stage Ta) who were followed routinely for up to 2 years. Voided urine cytology (VUC) and preselected site biopsies (PSB) were obtained prospectively to evaluate the signi cance of concomitant epithelial dysplasia. Results The mean follow-up was 84 months (maximum 238). Of all tumours, 39% did not relapse, a further 2% recurred infrequently (less than once a year) and 4% recurred frequently, amongst which the most frequent were multiple and early recurrent tumours; 42 (9%) tumours progressed to stage T+and 23 (%) progressed further (stage T2+or metastases). No grade tumours became invasive. Positive VUC or PSB, a short recurrence-free period or multiplicity, and size >3 cm were signi cant predictive factors. The treatment and surveillance of epithelium-con ned bladder tumours are discussed. Conclusion Concomitant dysplasia and early recurrence are associated with considerable risk of progression in the long-term follow-up in a group of otherwise lowrisk super cial bladder tumours (stage Ta). Keywords Bladder cancer, risk factors, recurrence, stage Ta, cytology, biopsy Introduction In a recent report we found lamina propria invasion to be the most signi cant prognostic factor for the long-term survival of so-called `super cial' urinary bladder tumours; 85±9% of patients with initial Ta tumours, but only a third of those with lamina propria invasion (stage T) nally survived their disease []. This perception is not new and has been con rmed by many others [2±5]. Nevertheless, only a few studies provided a separate follow-up for this low-risk group of tumours [6,7]. However, a ±5% nal risk of progression, eventually leading to death from cancer, is not negligible. Several prognostic factors have been indicated, but little is known about their long-term signi cance and only a few are applied for treatment and surveillance decisions in routine practice [3]. The grade of dysplasia, size and multiplicity of the tumours, the presence/absence of concomitant epithelial dysplasia/ carcinoma in situ (CIS) and the recurrence rate or time to rst recurrence are signi cant prognostic factors, accessible by simple endoscopic and microscopic examination, and well-known for many years [3]. The present series of patients with bladder cancer were enrolled in 976 to Accepted for publication 4 December 999 investigate the impact of concomitant epithelial changes [8]. To our knowledge, no report has yet con rmed the long-term signi cance of this important prognostic factor. In the present study we followed low-risk patients (stage Ta) for up to 2 years, and assessed recurrenceand progression-free survival. Patients and methods Of a series of 584 consecutive, unselected patients admitted for newly diagnosed urinary bladder tumour to one hospital in Copenhagen, 24 had Ta disease; 27 (54 men, mean age 66 years, range 34±84, and 63 women, mean age 67 years, range 2±87) were followed for at least one year and included for evaluation. Tumours were graded following the Bergkvist classi cation [9]. Voided urine cytology (VUC, 83%) and preselected site biopsies (PSB, 87%) were obtained prospectively according to the procedure described previously [8], the results having no in uence on the management of the patients. The samples were graded using the same method and regarded as positive when at least one positive specimen was obtained. Any suspect cells were regarded as positive. The treatment (including that for super cial recurrence) was TURBT. Ten patients received supplementary 824 # 2 BJU International
FOLLOW-UP OF NONINVASIVE BLADDER TUMOURS 825 intravesical chemotherapy (thiotepa), retinoid was given in nine patients (with no effect) and radiotherapy in two. One tumour was detected incidentally during transvesical prostatectomy and resected. Four patients were treated with intravesical BCG (after 987). Patients were followed by cystoscopy every fourth month. The time to rst recurrence (observed recurrence-free period, RFP) and the recurrence rate (RR) were determined as distinct end-points. For the RR, the number of relapses was counted, so that multiple relapses corresponded to single occurrences at reduced control intervals. `Flat papillomatosis' was counted as one recurrence. The RR was de ned as (no relapses observed), (< relapse per year) or 2 (o relapses per year) and encompassed the total follow-up up to 2 years. `Conversion' was de ned as progression in grade with no invasion; `progression' was de ned as progression in stage, i.e. invasion of the lamina propria or deeper (T+), or distant metastases. T tumours were treated by TURBT, but one open bladder resection and two cystectomies were performed, and ve patients received full-course radiotherapy when muscle invasion (T2+) occurred. The routine follow-up was continued to a RFP of 5 years. Some patients continued voluntarily or were readmitted with relapses later. The hospital les were reviewed for all those patients who ceased follow-up at some time. If the patient moved from the area, the hospitals taking the patients were contacted for further data. The mean follow-up was 84 months (maximum 238). The nal status was obtained by autopsy or laparotomy in 7 patients (33%) and by cystoscopy or relevant radiological imaging (metastases) in 46 (66%); 59 patients (26%) were alive at the end of the study ( January 997) and were censored at the time of their last control cystoscopy. In the statistical evaluation the log-rank test and Cox regression analysis model were applied where indicated [] and differences regarded as signi cant when P<.5. Results The initial tumour characteristics are shown in Table, given as the Bergkvist grade of dysplasia, which is recognized as one of the major determinants of malignancy [3,9]. There was a close positive correlation between grade, multiplicity, size of tumour, and the results of VUC and PSB. Of all tumours, 6% recurred (Fig. a). After a RFP of one year, the cumulative probability was 46%, and after a RFP of 2 years it was 34%. In 85 patients (39%) there was no recurrence (RR ) and another 43 patients (2%) had less than one recurrence per year (RR ); 4% of the tumours recurred frequently (RR 2). After a RFP of one year the probability of frequent recurrence was 6% and after 2 years it was 6%. Of 89 tumours with a RR of 2, 8% had recurred within one year and 9% within 2 years. If the tumour had recurred within one year, it continued to recur (RR 2) in 84%. Five patients were readmitted after a RFP of >5 years; two of these had a RR of 2, corresponding to 5% of 38 who survived 5 years with no recurrence. In the multiple Cox regression analysis of risk factors (Table 2), tumour multiplicity, size and urine cytology, in descending order, were independent signi cant variables predicting recurrence, but the grade of dysplasia was not, because it was correlated with the other factors (Table ). Forty-two tumours (9%) had invasive potential; 9 of these were limited to stage T, but 23 (%) showed muscle-invasion (T2+) or distant metastases. Twentythree patients died from bladder cancer-related causes, including three with no apparent muscle invasion. The cause-speci c survival rate was equal to that reported for the initial T tumours []. None of the grade tumours progressed but six converted to grade 2; Ta grade 2 tumours converted Table Characteristics of 27 consecutive, unselected noninvasive bladder tumours (Ta category) graded using the Bergkvist method. Differences in totals (last column) were caused by the item not being evaluable or not determined. Percentages are given in brackets Characteristic n (%) Grade / 2 3 Total Single 29 (88) 35 (76) 2 66 Multiple 4 43 3 5 <3 cm 3 (9) 37 (77) 3 7 >3 cm 3 4 2 46 VUC negative 25 (86) 95 (66) 2 positive 4 5 4 58 PSB negative 26 () 33 (84) 3 62 positive 26 27 Total 33 79 5 27 # 2 BJU International 85, 824±828
826 K. ZIEGER et al. to Ta grade 3 (three) or CIS (seven), with later progression in ve of the latter. Forty-one Ta grade 2 tumours (23%) and one of ve Ta grade 3 tumours progressed in stage (T+). Progression occurred after a mean (range) of 55 (3±29) months. Four tumours progressed with no previous benign recurrences. The progression-free survival (T+and T2+) for Ta tumours is shown in Fig. b. The progression rate was approximately constant over the observation period, i.e. there were many late progressions. a..8.6.4.2 b..8.6.4.2 c..8.6.4.2 2 3 4 5 6 7 8 9 Follow-up period (years) 2 3 4 5 6 7 8 9 2 3 4 5 Follow-up period (years) 2 3 4 5 6 7 8 9 2 3 4 5 Observation time (years) Fig.. Kaplan±Meier plot of the cumulative probability of: a, recurrence for Ta tumours; b, lamina propria invasion (green) and muscle-invasion or metastases (red) for Ta tumours; and c, progression (in T+) for Ta tumours in a combined analysis of major risk factors (84 patients at low-risk, PSB + VUC ±ve, RFP > year, green; 92 patients PSB or VUC +ve, or RFP < year, light green; 4 patients at high-risk, PSB or VUC +ve and time to recurrence < year, red). In all, the 95% CI are indicated as stippled lines. Positive VUC, positive PSB, a RFP of < year, tumour size and multiplicity (all P<.) were predictive of progression in the log-rank test. Multiple regression analysis showed that RFP, VUC and size, in descending order of signi cance, were independent factors predicting future progression (Table 2). There was a close correlation between the effects of VUC and PSB, and between RFP and multiplicity on the progression rate. For progression to muscle invasion (T2+) only VUC and the RFP were independent signi cant factors, because there were fewer events (23). Only two of 84 patients with initially negative VUC/ PSB and a RFP of > year progressed; none of the tumours became muscle invasive; 92 patients with one factor positive and 4 with both factors positive had progression rates to muscle invasion of 2% and 5% after 5 years, respectively (Fig. c). For comparison, all patients with CIS (dysplasia grade 3), including primary CIS with symptoms (seven), concomitant CIS with Ta grade 2 or 3 (three), and CIS with previous Ta grade 2 tumour (conversion, seven) were also reviewed. In all the patients with CIS, tumours developed later; of 7 progressed to invasive TCC, seven of whom progressed to advanced disease (T2+). As the need for routine IVU in the follow-up of super cial bladder tumours is contentious [4,,2], the incidence of upper urinary tract TCC (UTT) was examined retrospectively in the present series of patients at low-risk after the long-term follow-up. Fifteen patients (7%) had UTT (seven in the renal pelvis, three in the ureter and ve with multiple tumours); nine tumours were invasive (T+). The tumours were found after a mean of 52 months. Frequently recurrent bladder tumours accompanied all but two UTT; these two tumours were found incidentally and were the only tumours detected > years (6 and 28 months) after the diagnosis of the primary bladder tumour; neither was treated. Discussion This study presents the natural history of 27 unselected, newly diagnosed, noninvasive bladder tumours with up to 2 years of follow-up under conservative treatment. The conditions for long-term follow-up in Scandinavian countries are excellent, because there is a stable society, a central population register and a public health system covering nearly all the population. If contact is lost, the patients are easy to track in the register and are likely to be identi ed from the les of their local hospital if they have any health problems. This enabled the present review to be based on objective data with little `loss to follow-up'. Knowledge of the long-term natural history of super- cial bladder tumours is increasing [2±5]. Unfortunately, # 2 BJU International 85, 824±828
FOLLOW-UP OF NONINVASIVE BLADDER TUMOURS 827 Table 2 The test for equality of distribution of recurrence and progression for Ta tumours in a multiple regression analysis model containing the signi cant predictive variables multiplicity, size, histological grade, VUC, PSB and RFP. b is the estimate of the regression coef cient and P the probability Variable Wald value b (SE) P Recurrence Multiplicity.8 x.7 (.2).9 Size 7.4 x.62 (.23).7 Grade.66 x.24 (.3).4 VUC 5.5 x.46 (.2).2 Progression VUC 7.54 x.7 (.43).6 PSB.5 x.42 (.4).3 RFP 9.8.58 (.5).2 Mult..23 x.42 (.37).27 Size 5.5 x.88 (.37).2 not many authors consequently distinguish between Ta and T categories. The probability of recurrence for Ta tumours is concordantly reported to be < 6% [3,7]. There is broad agreement that multiple tumours and tumours with a history of recurrences are most likely to relapse. The disease-free interval after primary diagnosis is clearly signi cant for the frequency of future recurrences [3,5,3]. In the present study there was no bene t of prolonged surveillance after a RFP of 2 years for the detection of frequently recurrent or progressive disease. The estimation of the probability of progression requires a long-term follow-up; there are few reports assessing lowrisk (stage Ta) super cial bladder tumours. Prout et al. [7] followed 78 patients with Ta grade (WHO) tumours for up to years. Progression rates were 3% (T+) and 2% (T2+), respectively; the authors emphasized that these tumours were not distinguishable initially from others, but there were few progression events. A combined follow-up study of 576 Ta/T tumours from EORTC randomized trials reported a progression rate of 9% (T2+) for 3 Ta tumours after a median follow-up of 4 years [5]. The authors used a multivariate analysis of risk factors, including size, multiplicity, grade, stage, site and recurrence. Except for grade, which is closely correlated with stage, the RFP/previous RR and tumour size were independently signi cant. Cytology and/or PSB results were not assessed. However, as T and nonprimary tumours were included, the results are not directly comparable with the present, study. The Swedish study by Holmang et al. [4] is more comparable, as it was also Scandinavian with a long follow-up based on objective data and tumours were graded using the same system [9], which is based on cellular pattern. There were fewer patients but with a longer and more intense follow-up. The high progression rate of 4% (crude) is thus comparable with the present, but the exclusion of concomitant CIS (two) and those undergoing more aggressive treatment (six cystectomies and three radiotherapy in 77 patients; present study, two radiotherapy in 27) reduced the number of high-risk patients in the follow-up. Interestingly, the progression of three Bergkvist grade tumours was reported. There was no progression in this group in the present patients, despite ve having frequent recurrences. It is also remarkable that the (few) grade 3 tumours had a progression rate that was no higher than in grade ±2 tumours, which is consistent with the present results. Concomitant dysplasia was not addressed by Holmang et al. Cookson et al. [2] reported the long-term follow-up of 86 high-risk patients with Ta/T tumours. The progression rate was 53%, similar to that for the highrisk group of present Ta tumours. As in the EORTC study, randomized patients were followed, biasing the follow-up to younger patients and high survival rates. Although the authors recognized lamina propria invasion, concomitant CIS and multiple recurrences as high-risk factors, they did not separate these factors in their analysis. Indeed, for the present Ta category, positive VUC was a very signi cant prognostic factor and it is likely that the neoplastic cells derive from CIS rather than from the benign Ta tumours [4]. Positive VUC correctly and signi cantly predicted the malignant nature of disease which lead to progression 5±5 years later. Urine cytology is not speci c; half of the present 58 patients with initial positive VUC neither converted nor progressed. Thus, random or preselected site biopsies have been evaluated. Positive biopsies predict recurrence [3], but most tumours with negative samples also recur. PSB were no more speci c than VUC in predicting progression and the sensitivity was low (85%). Concomitant CIS with Ta tumour is rarely found in PSB; intraepithelial dysplasia grade 2 is more common (< 5%), but its signi cance is dubious. Thus, PSB are dispensable as a routine procedure. However, the combined analysis of # 2 BJU International 85, 824±828
828 K. ZIEGER et al. the VUC and PSB results had more signi cance; the risk of progression was equivalent to native CIS. Nevertheless, as not all tumours with invasive potential could have been diagnosed at the initial evaluation, repeated samples and control investigations are necessary. The results indicate that there should be more weight placed on the results of VUC and PSB sampling when deciding the therapy for Ta disease. VUC samples should be obtained initially and at every follow-up assessment. PSB should be taken if the VUC is positive. Any dysplasia indicates a treatment equivalent to CIS, i.e. instillation of BCG/mitomycin C, and cystectomy if the therapy fails. Treatment may also be considered if VUC is repeatedly positive. Control assessments should continue until a RFP of 2 years. Frequent recurrences demand intense surveillance, but may safely be treated by TURBT alone, unless positive VUC/PSB samples indicate a more aggressive approach. IVU may be justi ed in these patients as part of the follow-up, but the validity of routine exposures to detect UTT is doubted. Patients with Ta bladder tumours, especially when single and small at diagnosis, may be discharged after 2 years with no recurrence or positive VUC. Their risk of frequent recurrence, CIS, invasive bladder cancer or UTT is little higher than in the general population. Recurrence will occur occasionally in some cases but the informed patient may safely be followed by their GP with haematuria dipstick and VUC evaluation annually, with minimum discomfort [5,6]. This will be valid for < 4% of all newly diagnosed Ta tumours. Acknowledgements Thanks to Vestdansk Sundhedsvidenskabeligt Forskningsforums Konsulenttjeneste for help with the statistical work, and to Fru Agnes Niebuhr Anderssons Cancer forskningsfond for nancial support. References Zieger K, Wolf H, Olsen PR, Hùjgaard K. Long term followup of bladder tumours: the signi cance of risk factors. Br J Urol 998; 82: 667±72 2 Abel PD. Follow-up of patients with `super cial' transitional cell carcinoma of the bladder: the case for a change in policy. Br J Urol 993; 72: 35±42 3 Heney NM. Natural history of bladder cancer. Prognostic features and long-term disease course. Urol Clin North Am 992; 9: 429±33 4 Holmang S, Hedelin H, AnderstroÈm C, Johansson SL. The relationship among multiple recurrences, progression and prognosis of patients with stage Ta and T transitional cell cancer of the bladder followed for at least 2 years. J Urol 995; 53: 823±7 5 Kurth KH, Denis L, Bouf oux C et al. Factors affecting recurrence and progression in super cial bladder tumours. Eur J Cancer 995; 3: 84±6 6 AnderstroÈm C, Johansson S, Nilsson S. The signi cance of lamina propria invasion on the prognosis of patients with bladder tumours. J Urol 98; 24: 23±6 7 Prout GR, Barton BA, Grif n PP, Friedell GH. Treated history of noninvasive grade transitional cell carcinoma. J Urol 992; 48: 43±9 8 Wolf H, Hùjgaard K. Urothelial dysplasia in random mucosal biopsies from patients with bladder tumours. Scand J Urol Nephrol 98; 4: 37±4 9 Bergkvist A, Ljungkvist A, Moberger G. Classi cation of bladder tumours based on a cellular pattern. Acta Chir Scand 965; 3: 37±8 Cox DR. Regression models and life tables. J Roy Stat Soc B 972; 34: 87±22 Hurle R, Losa A, Manzetti A, Lembo A. Upper urinary tract tumours developing after treatment of super cial bladder cancer: 7 year follow-up of 59 consecutive patients. Urology 999; 53: 45±8 2 Cookson MS, Herr HW, Zhang Z, Soloway S, Sogani PC, Fair WR. The treated natural history of high risk super cial bladder cancer: 5-year outcome. J Urol 997; 58: 62±7 3 Morris SB, Gordon EM, Shearer RJ, Woodhouse CRJ. Super cial bladder cancer: for how long should a tumourfree patient have check-cystoscopies? Br J Urol 995; 75: 93±6 4 Harving N, Wolf H, Melsen F. Positive urinary cytology after tumour resection: an indicator for concomitant carcinoma in situ. J Urol 988; 4: 495±7 5 Meuleman EJH, Delaere KPJ. Diagnostic ef cacy of the combination of cytology, urine analysis and history in the follow-up of bladder carcinoma. Br J Urol 988; 62: 5±3 6 Soloway MS, Murphy WM, Johnson DE, Farrow GM, Paulson DF, Garnick MB. Initial evaluation and response criteria for patients with super cial bladder cancer. Report of a workshop. Br J Urol 99; 66: 38±5 Authors K. Zieger, MD. H. Wolf, MD, PhD. P.R. Olsen, MD, PhD. K. Hùjgaard, MD. Correspondence: Lñge Karsten Zieger, Skovvej 27, Gram DK 866 Skanderborg, Denmark. # 2 BJU International 85, 824±828