- Email: [email protected]
Intravesical chemotherapy prophylaxis in primary superficial bladder cancer
Journal of Clinical Epidemiology 53 (2000) 676–680
Intravesical chemotherapy prophylaxis in primary superficial bladder cancer: a meta-analysis of 3703 patients from 11 randomized trials Michael Hunchareka,b,*, Jean-Francois Geschwindc, Bart Witherspoond, Ronald McGarrye, David Adcockb a Uro-Oncology Project, Meta-Analysis Research Group, Columbia, SC, USA Department of Radiology, University of South Carolina School of Medicine, Columbia, SC, USA c Johns Hopkins Hospital, Johns Hopkins School of Medicine, Baltimore, MD, USA d Baptist Hospital Cancer Institute, Columbia, SC, USA e Department of Radiation Oncology, University of Nebraska Medical Center, Omaha, NE, USA Received 17 March 1999, received in revised form 15 September 1999; accepted 7 October 1999
b
Abstract The impact of intravesical chemotherapy prophylaxis on recurrence of superficial transitional cell carcinoma of the bladder is poorly defined. The objective of this report is to present a meta-analysis of the available clinical trial data to quantify the effect of intravesical chemotherapy on tumor recurrence following complete transurethral resection (TURB) in patients with newly diagnosed superficial bladder cancer. A prospective protocol outlining the above meta-analysis was initially developed followed by a thorough search of the existing published literature using strict eligibility criteria. Eleven randomized trials were found that met protocol specifications. These studies contained data on 3703 patients that were statistically combined using a fixed effects model (Peto). The outcome of interest was the proportion of patients recurring at 1, 2, and 3 years post-TURB. Combining all 11 studies using 1-year recurrence as the outcome measure yielded a Peto odds ratio (ORp) of 0.56, demonstrating a 44% reduction in 1-year recurrence among patients treated with intravesical chemotherapy versus those treated with TURB alone. A statistical test for heterogeneity (Q) showed these data to be heterogenous (the studies are not measuring an effect of the same size). Sensitivity analyses were performed to determine sources of heterogeneity. These tests suggest that chemotherapy treatment schedule may account for the wide variation in tumor recurrence rates across studies. When the available clinical trial data were stratified by duration of treatment, the meta-analysis showed that intravesical chemotherapy decreased tumor recurrence from 30% to 80% depending on the outcome of interest (i.e., recurrence at 1, 2, or 3 years post-TURB). Intravesical chemotherapy appears to have a major impact on decreasing the chance of recurrence of superficial transitional cell carcinoma of the bladder. This is in contrast to prior analyses suggesting only modest efficacy in this clinical setting (i.e., on the order of a 14% reduction in recurrence). © 2000 Elsevier Science Inc. All rights reserved. Keywords: Combined modality; Recurrence; Transitional cell carcinoma
1. Introduction Of over 50,000 cases of transitional cell carcinoma of the bladder diagnosed each year in the United States, approximately 70–80% are superficial tumors (i.e., limited to the mucosa or lamina propria) (stage Ta or T1) [1]. Transurethral resection (TURB) is the initial approach for this disease. Unfortunately, superficial bladder cancer commonly recurs in a large proportion of cases (usually within 12 months of surgery) even with complete resection. Although * Corresponding author. Department of Radiation Oncology, Marshfield Clinic Cancer Center, St. Michael’s Hospital, 900 Illinois Ave., Stevens Point, WI 54481. Tel.: 715-343-3030.
most recurrent tumors are the same grade and stage as the primary, 5–30% progress to more advanced disease [2]. These characteristics of superficial bladder tumors prompted the development of intravesical chemotherapy in an attempt to eradicate residual disease and possibly reduce the likelihood of recurrence (i.e., as prophylaxis). While numerous drugs are now utilized in this setting, the true impact of intravesical therapy on the prophylaxis of superficial bladder cancer remains an open question. Some investigators suggest that this treatment has only a minor impact on tumor recurrence rates and questions the advisability of its routine use [3,4]. Due to the difficulty in interpreting the existing published data on this topic secondary to heterogeneity in study
0895-4356/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S0895-4356(99)00 2 0 3 - 6
M. Huncharek et al. / Journal of Clinical Epidemiology 53 (2000) 676–680
design, patient inclusion/exclusion criteria, study outcome of interest, and length of follow-up, among others, this report presents the results of a meta-analysis examining the impact of intravesical chemotherapy on tumor recurrence following primary TURB of transitional cell carcinoma of the bladder. This analysis will provide a systematic evaluation of the clinical trial data addressing this issue and provide a basis for future clinical studies. 2. Materials and methods The methods used in the design of this meta-analysis have been previously described [5]. A prospective study protocol was developed outlining a meta-analysis to determine the impact of intravesical chemotherapy on recurrence of transitional cell carcinoma of the bladder following primary resection. The outcome of interest was tumor recurrence at 1, 2, and 3 years following resection. Eligibility criteria for study inclusion were determined prospectively as were the data elements to be extracted from each published report. The statistical analysis was also outlined in the study protocol. A data extraction form was developed for recording relevant information from each included study. Data extraction was performed by two physicians with differences in extraction forms resolved by consensus. Additional data collected but not specified as inclusion criteria included: study sample size, types of drugs used, drug dosage, and treatment schedule. 2.1. Literature search Information retrieval was performed by previously described methods [6]. A MEDLARS search was conducted covering the years 1966 to 1997. CancerLit and the CDROM version of Current Contents were also searched. The computer data base searches were supplemented by manual searching of bibliographies of all retrieved papers. Published abstracts were not included in the literature search. If a series of papers was published, all data were retrieved from the most recent report. Abstracts were obtained from any reference that appeared relevant to the search. The search was limited to English language literature and the initial citations (in the form of abstracts) were screened by a physician investigator (oncologist) to exclude those that did not meet protocol inclusion criteria. Studies were excluded if only nonintravesical chemotherapy was given or if the study contained patients treated following recurrence and patients could not be stratified by recurrence status. Studies with chemotherapy and other treatment arms were included only if the data pertaining to the chemotherapy arms could be analyzed separately from the other treatment arms. Copies of full articles for the remaining citations were obtained and screened using the following eligibility criteria; 1. Published randomized controlled trials (RCTs) enrolling adult patients comparing TURB alone (control arm) to TURB with adjuvant intravesical chemother-
2.
3.
4. 5.
677
apy (treatment arm) for primary superficial transitional cell carcinoma of the bladder. RCTs containing patients with both primary and recurrent disease were included only if data on patients with primary tumors could be analyzed separately. Acceptable tumor stages included Ta, T1 with or without carcinoma in situ. Studies containing patients with incomplete TURB were excluded. Studies must have a minimum of 1 year follow-up. RCTs containing one or more treatment arms with data on the outcome of interest (i.e., proportion of patients with recurrence at 1, 2, or 3 years post-TURB).
2.2. Statistical methods The meta-analytic procedures used were those of Peto et al. [7]. This method is a modification of the Mantel-Haenzel method and is based on a fixed effects model. Study data are arranged in a 2 ⫻ 2 matrix and a summary odds ratio (ORp) and its 95% confidence interval is calculated. The outcome (event) of interest is tumor recurrence at specified times following TURB. The expected number of events in the “experimental arm” of each study is calculated and an estimate of the variance of the observed minus expected number of events in each study was then determined as well as the variances across all studies. Prior to estimation of a summary odds ratio, a statistical test for homogeneity was performed (Q). This procedure tests the hypothesis that the effect sizes are equal in all of the studies [5]. If Q exceeds the upper tail critical value of chi-square (P ⬍ 0.05) at k ⫺ 1 degrees of freedom (where k = the number of studies analyzed), the observed variance in study effect size is significantly greater than what would be expected by chance if all studies shared a common population effect size. If the hypothesis that the studies are homogenous is rejected, the studies are not measuring an effect of the same size and calculation of a pooled estimate of effect may be of questionable validity. In this setting, sources of heterogeneity were sought (e.g., via sensitivity analyses).
3. Results The literature search produced a total of 1672 citations. Initial screening yielded 40 references that appeared to meet protocol specifications. Further review found that 29 of these were not eligible for inclusion for various reasons including: 1) study contained only patients with recurrent tumors, 2) contained patients with primary as well as recurrent disease without stratification of data on this parameter, 3) less than 1 year of follow-up, 4) contained patients with incomplete TURB, and 5) lack of data on recurrence rates or study design was nonrandomized. This left 11 RCTs that form the basis of this report [8–18]. Table 1 provides an overview of the included studies. All studies included patients with stage Ta and T1 tumors.
678
M. Huncharek et al. / Journal of Clinical Epidemiology 53 (2000) 676–680
Table 1 Summary of included studies Author
Ref. #
Drug(s) used
Hirao Hirao Akaza
[8] [9] [12]
Niijima
[18]
MRC
[10]
thio-tepa epirubicin adriamycin (2 arms) mitomycin-C (1 arm) adriamycin (2 arms) mitomycin-C (1 arm) thio-tepa (1 arm) thio-tepa (1 arm)
Tolley
[11]
Togashi
[13]
Hirao
[14]
Krege Flamm Tsushima
[15] [17] [16]
mitomycin-C (1 arm) mitomycin-C (1 arm) adriamycin (1 arm) peplomycin (1 arm) neocarzinostat (1 arm) adriamycin (1 arm) mitomycin-C (1 arm) thio-tepa (1 arm) mitomycin-C mitoxantrone adriamycin (1 arm) mitomycin-C
Duration of Tx
Tumor stage/grade
2 years 2 years 1 year
Ta, T1/G1,2 Ta, T1/G1,2 Ta, T1/G1,2
1 month
Ta, T1/G1,2,3
1 year single instillation 1 year
Ta, T1/G1,2,3
1 year
Ta, T1/G1,2
2 years
Ta, T1/G1,2,3
2 years 6 weeks 2 years
Ta, T1/G1,2,3 Ta, T1/G1,2,3 Ta, T1/G1,2,3
Ta, T1/G1,2,3
There were no patients with carcinoma in situ in these studies. Five studies included grade 1 and 2 tumors only, while seven studied included grade 1, 2, and 3 tumors. The most commonly used drug was adriamycin (8 treatment arms) followed by mitomycin-C (7 arms). Thio-tepa and cytosine arabinoside were used in 4 and 3 arms, respectively. Various other drugs were employed in the remaining 5 arms. Initially, all 11 studies were combined in a meta-analysis with the proportion of patients recurring at 1 year as the outcome of interest (Fig. 1). This analysis included a total of 3703 patients and yielded a summary odds ratio (ORp) of 0.56 with a 95% confidence interval of 0.48–0.65 (P ⬍ 0.00001). This shows a highly statistically significant effect for intravesical chemotherapy in reducing bladder tumor recurrence at 1 year. The magnitude of this reduction is, on average, 44%. Calculation of Q (heterogeneity) using these data showed Q to equal 55.6 (with 20 df ), demonstrating heterogeneity across studies. This indicates that the effect sizes (i.e., influence of intravesical chemotherapy on the outcome measure of interest) vary across studies to a greater degree than would be expected by chance. Sources of heterogeneity were sought because any summary estimate of effect may be of dubious validity without a clear understanding of the observed heterogeneity. The studies were therefore reviewed for this purpose. Based on this review, the most likely source of heterogeneity appeared to be the wide variety in treatment schedules employed across studies. Some investigators used only a single intravesical instillation of drug following TURB while others used protocols that consisted of repeated installations over the course of 2 years. This variability was thought to possibly account for the observed heterogeneity and sensitivity analyses were performed to confirm this.
All included studies were stratified by length of treatment into three distinct groups: (1) short-term protocols (which used either a single instillation of drug or a few instillation delivered within a 2-month period), (2) 1-year protocols using multiple drug instillations over the course of 1 year, and (3) 2-year protocols in which treatment consisted of multiple intravesical treatments over 2 years. Meta-analysis and Q statistics were recalculated using data stratified on the above parameters. Studies using short-term treatment protocols were initially combined, which included a total of 1258 patients. The ORp using 1-year recurrence as the outcome was 0.70 with a 95% confidence interval of 0.55–0.90 (P ⫽ 0.006). Q was found to equal 10.2. This was nonsignificant at the P ⬍ 0.05 level, which is consistent with a lack of heterogeneity. This demonstrated that treatment length probably contributed to the initially observed variance in outcomes. An ORp of 0.70 favors a positive impact of intravesical chemotherapy in reducing 1-year tumor recurrence by approximately 30%. Recalculating an ORp for short-term treatment using 2-year recurrence as the outcome of interest showed a similar effect of therapy on reducing 2-year recurrence [i.e., 32% (95% CI ⫽ 0.54–0.85, P ⫽ 0.0007)]. Therefore, short-term intravesical treatment reduces recurrence at 1 and 2 years by approximately 30%. Studies employing multiple intravesical instillations of chemotherapeutics for 1 year were next combined in a meta-analysis. This analysis included 1721 patients and showed an odds ratio of 0.65 (95% CI ⫽ 0.46–0.80, P ⭓ 0.0001), highly favoring intravesical therapy versus TURB alone with a 35% reduction in recurrence at 1 year. Q again showed a lack of heterogeneity (Q ⫽ 10.53, with 7 df). The data on 2-year recurrence as the outcome were consistent with those seen at 1 year [i.e, 31% reduction in 2-year recurrence (95% CI ⫽ 0.57–0.83, P ⬍ 0.0001)]. Five hundred and seventy-five patients in 5 RCTs were treated on protocols for 2 years. The reduction in recurrence at the completion of therapy (i.e., 2 years) was 73%, corresponding to an ORp of 0.27 (95% CI ⫽ 0.19–0.39, P ⫽ 0.0001), a highly significant result. This reflects a profound reduction in 2-year recurrence associated with this treatment scheme. Analyzing the same studies after patients received 1 year of treatment supported the above results [i.e., odds ratio of 0.20 (95% CI ⫽ 0.13–0.29)]. Five studies contained data on 3-year recurrence [8,9,15,16,18]. Due to the small number of studies, a stratified analysis based on treatment schedule was not possible. Nonetheless, 1371 patients were available for the metaanalysis and yielded on ORp of 050 (95% CI ⫽ 0.40–0.62), strongly favoring the use of intravesical chemotherapy (P ⬍ 0.0001) (i.e., a 50% reduction in recurrence at 3 years versus TURB alone). Only three studies contained 5-year survival data, making a rigorous analysis of this endpoint impossible due to the small numbers of patients represented and short follow-up times. A further analysis was performed to determine whether drug type was an important determinant of outcome. When
M. Huncharek et al. / Journal of Clinical Epidemiology 53 (2000) 676–680
679
Fig. 1. Results of meta-analysis combining data from all 3703 patients. Outcome measure for this analysis is proportion of patients with tumor recurrence at 1 year. (Duplicate citation of studies represents multiple treatment arms as outlined in Table 1.)
the available data were stratified by the most common drug classes (i.e., mitomycin-C, adriamycin, thio-tepa) no differences were seen in effect on recurrence due to the presence of statistical heterogeneity. Q was found to be 16.3 for the adriamycin group (P ⫽ 0.02) and 9.9 for thio-tepa (P ⫽ 0.03). The mitomycin-C group was associated with an ORp of 0.45 and a Q of 7.3 with 5 df (P ⬎ 0.10). Of the three drug classes studies, the mitomycin-C group was the only one showing a statistically significant reduction in tumor recurrence in the absence of heterogeneity. These findings suggests that factors other than drug type may have greater effect on tumor recurrence, although the data must be interpreted cautiously do the small numbers involved in the stratified analysis. Nonetheless, this may provide additional support for an important effect of treatment duration as noted previously. 4. Discussion Despite its availability and use over several decades, intravesical chemotherapy prophylaxis in the management of superficial bladder cancer remains controversial. For instance, in a 1995 article, 22 RCTs of intravesical chemotherapy were reviewed with the authors concluding that such treatment had only a minor impact on the incidence of tumor recurrence, averaging only 14% [3]. This paper also suggested that maintenance chemotherapy is less effective than a single postoperative instillation. Unfortunately, inter-
pretation of these data is difficult for several reasons. These include: 1) the analysis combined studies containing patients with primary and recurrent bladder tumors without stratification on this parameter, 2) failure to calculate recurrence rates as specified endpoints (e.g., 1 or 2 years postTURB), and 3) failure to use statistical methods designed for combining data from multiple randomized clinical trials. In an attempt to clarify the ambiguities in the existing published data base, our report presents a meta-analysis to determine the impact of intravesical chemotherapy on tumor recurrence rates at 1, 2, and 3 years post-primary TURB. Using a fixed effects model, our study demonstrates that intravesical therapy produces statistically significant reduction in tumor recurrences at 1, 2, and 3 years in this patient group, ranging from 30% to 80% depending on the endpoint of interest. This is in contrast to the much more modest 14% cited above. This analysis also demonstrates that treatment schedule may also influence recurrence rates with longer term therapy producing greater reduction in recurrence rates than short-term schedules. Existing data on this issue are equivocal [4]. In our study, patients receiving therapy for 2 years showed the greatest reduction in tumor recurrence. Interestingly, 1-year recurrence rates among those patients on 2-year chemotherapy protocols showed greater reduction in recurrence (approx. 80%) when compared with patients treated on 1-year protocols. This may be due to the fact that the treatment schedules for 2-year protocols generally used fre-
680
M. Huncharek et al. / Journal of Clinical Epidemiology 53 (2000) 676–680
quent intravesical instillations of drug compared to 1-year protocols. For instance, Krege et al. [15] instilled mitomycin-C every 2 weeks for the first year and once a month during the second year. This contrasts with many of the 1-year protocols, which delivered treatment every 3 months [11]. This suggests that clinical effect may be dose related. The major strength of the presented analysis is its restriction to a single, well-defined question (i.e., the impact of intravesical chemotherapy on tumor recurrence rates following primary complete TURB at specified endpoints). The study also followed a rigorously defined, prospective protocol using accepted meta-analytic techniques. These characteristics should help provide a clearer interpretation of the existing published information. One of the major limitations of the existing data is the lack of studies with long-term follow-up. Using specified inclusion criteria, data were unavailable on recurrence rates beyond 3 years. This limits assessments of long-term influence of such treatment, which represents an important clinical and economic question. Lack of long-term follow-up data also limits examination of the impact of intravesical chemotherapy on tumor progression and overall survival. As noted earlier, our analysis included only patients undergoing primary TURB of superficial transitional cell carcinoma of the bladder. The effect of intravesical chemotherapy on recurrence rates in patients with recurrent tumors may differ from those treated for primary disease due to differences in tumor biology. We are therefore conducting an additional meta-analysis including those containing patients with recurrent tumors in order to address this issue. It is hoped that this latter analysis will provide further information to clarify the role of such treatment in this disease setting.
References [1] Wingo PA, Tong T, Bolden S. Cancer statistics. CA 1995;45:8. [2] Hall RR, Parmar MKB, Richards AB, et al. Proposal for changes in cystoscopic follow up of patients with bladder cancer and adjuvant intravesical chemotherapy. Br Med J 1994;23:308–11. [3] Lamm DL, Riggs DR, Traynelis C, Nseyo UO. Apparent failure of current intravesical chemotherapy prophylaxis to influence the long term course of superficial transitional cell carcinoma of the bladder. J Urol 1995;153:1444–50.
[4] Herr HW, Laudone VP, Whitmore WF. An overview of intravesical therapy for superficial bladder tumors. J Urol 1987;138:1363–8. [5] Cooper H, Hedges LV. The Handbook of Research Synthesis. New York: Russel Sage Foundation, 1994. [6] Chalmers TC, Lau J. Meta-analysis stimulus for changes in clinical trials. Stat Methods Med Res 1993;2:161–72. [7] Yusuf S, Peto R, Lewis J, et al. Beta-blockade during and after myocardial infarction: an overview of the randomized trials. Prog Cardiovasc Dis 1985;27:335–71. [8] Hirao Y, Okajima E, Ozono S, et al. A prospective randomized study of prophylaxis of tumor recurrence following transurethral resection of superficial bladder cancer—intravesical thio-TEPA versus oral UFT. Cancer Chemother Pharmacol 1992;30:S26–S30. [9] Hirao Y, Ozono S, Momose H, et al. Prospective randomized study of prophylaxis of superficial bladder cancer with epirubicin: the role of a central pathology laboratory. Cancer Chemother Pharmacol 1994;35: S36–S40. [10] Medical Research Council Working Party on Urological Cancer. The effect of intravesical thiotepa on tumor recurrence after endoscopic treatment of newly diagnosed superficial bladder cancer. A further report with long term follow-up of a Medical Research Council randomized trial. Br J Urol 1994;73:632–8. [11] Tolley DA, Parmar MKB, Grigor KM, et al. The effect of intravesical mitomycin-C on recurrence of newly diagnosed superficial bladder cancer: A further report with 7 years follow-up. J Urol 1996;155: 1233–8. [12] Akaza H, Isaka S, Koiso K, et al. Comparative analysis of short term and long term prophylactic intravesical chemotherapy of superficial bladder cancer. Cancer Chemother Pharmacol 1987;20:S91–S96. [13] Togashi M, Shinohara N, Toyota K, et al. Prophylactic chemotherapy for primary and recurrent superficial bladder cancer: preliminary results. Cancer Chemother Pharmacol 1992;30:S21–S25. [14] Hirao Y, Okajima E, Ohara S, et al. Prophylactic treatment for superficial bladder cancer following transurethral resection. Cancer Chemother Pharmacol 1987;20:S85–S90. [15] Krege S, Giani G, Meyer R, et al. A randomized multicenter trial of adjuvant therapy in superficial bladder cancer: transurethral resection only versus transurethral resection plus mitomycin-C versus transurethral resection plus bacillus Calmette–Guerin. J Urol 1996;156: 962–6. [16] Tsushima T, Matsumura Y, Ozaka Y, et al. Prophylactic intravesical instillation therapy with adriamycin and mitomycin-C in patients with superficial bladder cancer. Cancer Chemother Pharmacol 1987;20: S72–S76. [17] Flamm J, Donner G, Oberleitner S, et al. Adjuvant intravesical mitoxantrone after transurethral resection of primary superficial transitional cell carcinoma of the bladder. A prospective randomized study. Eur J Cancer 1995;31A:143–6. [18] Niijima T, Koiso K, Akaza H, et al. Randomized clinical trial on chemoprophylaxis of recurrence in cases of superficial bladder cancer. Cancer Chemother Pharmacol 1983;11:S79–S82.
Intravesical chemotherapy prophylaxis in primary superficial bladder cancer: a meta-analysis of 3703 patients from 11 randomized trials Michael Hunchareka,b,*, Jean-Francois Geschwindc, Bart Witherspoond, Ronald McGarrye, David Adcockb a Uro-Oncology Project, Meta-Analysis Research Group, Columbia, SC, USA Department of Radiology, University of South Carolina School of Medicine, Columbia, SC, USA c Johns Hopkins Hospital, Johns Hopkins School of Medicine, Baltimore, MD, USA d Baptist Hospital Cancer Institute, Columbia, SC, USA e Department of Radiation Oncology, University of Nebraska Medical Center, Omaha, NE, USA Received 17 March 1999, received in revised form 15 September 1999; accepted 7 October 1999
b
Abstract The impact of intravesical chemotherapy prophylaxis on recurrence of superficial transitional cell carcinoma of the bladder is poorly defined. The objective of this report is to present a meta-analysis of the available clinical trial data to quantify the effect of intravesical chemotherapy on tumor recurrence following complete transurethral resection (TURB) in patients with newly diagnosed superficial bladder cancer. A prospective protocol outlining the above meta-analysis was initially developed followed by a thorough search of the existing published literature using strict eligibility criteria. Eleven randomized trials were found that met protocol specifications. These studies contained data on 3703 patients that were statistically combined using a fixed effects model (Peto). The outcome of interest was the proportion of patients recurring at 1, 2, and 3 years post-TURB. Combining all 11 studies using 1-year recurrence as the outcome measure yielded a Peto odds ratio (ORp) of 0.56, demonstrating a 44% reduction in 1-year recurrence among patients treated with intravesical chemotherapy versus those treated with TURB alone. A statistical test for heterogeneity (Q) showed these data to be heterogenous (the studies are not measuring an effect of the same size). Sensitivity analyses were performed to determine sources of heterogeneity. These tests suggest that chemotherapy treatment schedule may account for the wide variation in tumor recurrence rates across studies. When the available clinical trial data were stratified by duration of treatment, the meta-analysis showed that intravesical chemotherapy decreased tumor recurrence from 30% to 80% depending on the outcome of interest (i.e., recurrence at 1, 2, or 3 years post-TURB). Intravesical chemotherapy appears to have a major impact on decreasing the chance of recurrence of superficial transitional cell carcinoma of the bladder. This is in contrast to prior analyses suggesting only modest efficacy in this clinical setting (i.e., on the order of a 14% reduction in recurrence). © 2000 Elsevier Science Inc. All rights reserved. Keywords: Combined modality; Recurrence; Transitional cell carcinoma
1. Introduction Of over 50,000 cases of transitional cell carcinoma of the bladder diagnosed each year in the United States, approximately 70–80% are superficial tumors (i.e., limited to the mucosa or lamina propria) (stage Ta or T1) [1]. Transurethral resection (TURB) is the initial approach for this disease. Unfortunately, superficial bladder cancer commonly recurs in a large proportion of cases (usually within 12 months of surgery) even with complete resection. Although * Corresponding author. Department of Radiation Oncology, Marshfield Clinic Cancer Center, St. Michael’s Hospital, 900 Illinois Ave., Stevens Point, WI 54481. Tel.: 715-343-3030.
most recurrent tumors are the same grade and stage as the primary, 5–30% progress to more advanced disease [2]. These characteristics of superficial bladder tumors prompted the development of intravesical chemotherapy in an attempt to eradicate residual disease and possibly reduce the likelihood of recurrence (i.e., as prophylaxis). While numerous drugs are now utilized in this setting, the true impact of intravesical therapy on the prophylaxis of superficial bladder cancer remains an open question. Some investigators suggest that this treatment has only a minor impact on tumor recurrence rates and questions the advisability of its routine use [3,4]. Due to the difficulty in interpreting the existing published data on this topic secondary to heterogeneity in study
0895-4356/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S0895-4356(99)00 2 0 3 - 6
M. Huncharek et al. / Journal of Clinical Epidemiology 53 (2000) 676–680
design, patient inclusion/exclusion criteria, study outcome of interest, and length of follow-up, among others, this report presents the results of a meta-analysis examining the impact of intravesical chemotherapy on tumor recurrence following primary TURB of transitional cell carcinoma of the bladder. This analysis will provide a systematic evaluation of the clinical trial data addressing this issue and provide a basis for future clinical studies. 2. Materials and methods The methods used in the design of this meta-analysis have been previously described [5]. A prospective study protocol was developed outlining a meta-analysis to determine the impact of intravesical chemotherapy on recurrence of transitional cell carcinoma of the bladder following primary resection. The outcome of interest was tumor recurrence at 1, 2, and 3 years following resection. Eligibility criteria for study inclusion were determined prospectively as were the data elements to be extracted from each published report. The statistical analysis was also outlined in the study protocol. A data extraction form was developed for recording relevant information from each included study. Data extraction was performed by two physicians with differences in extraction forms resolved by consensus. Additional data collected but not specified as inclusion criteria included: study sample size, types of drugs used, drug dosage, and treatment schedule. 2.1. Literature search Information retrieval was performed by previously described methods [6]. A MEDLARS search was conducted covering the years 1966 to 1997. CancerLit and the CDROM version of Current Contents were also searched. The computer data base searches were supplemented by manual searching of bibliographies of all retrieved papers. Published abstracts were not included in the literature search. If a series of papers was published, all data were retrieved from the most recent report. Abstracts were obtained from any reference that appeared relevant to the search. The search was limited to English language literature and the initial citations (in the form of abstracts) were screened by a physician investigator (oncologist) to exclude those that did not meet protocol inclusion criteria. Studies were excluded if only nonintravesical chemotherapy was given or if the study contained patients treated following recurrence and patients could not be stratified by recurrence status. Studies with chemotherapy and other treatment arms were included only if the data pertaining to the chemotherapy arms could be analyzed separately from the other treatment arms. Copies of full articles for the remaining citations were obtained and screened using the following eligibility criteria; 1. Published randomized controlled trials (RCTs) enrolling adult patients comparing TURB alone (control arm) to TURB with adjuvant intravesical chemother-
2.
3.
4. 5.
677
apy (treatment arm) for primary superficial transitional cell carcinoma of the bladder. RCTs containing patients with both primary and recurrent disease were included only if data on patients with primary tumors could be analyzed separately. Acceptable tumor stages included Ta, T1 with or without carcinoma in situ. Studies containing patients with incomplete TURB were excluded. Studies must have a minimum of 1 year follow-up. RCTs containing one or more treatment arms with data on the outcome of interest (i.e., proportion of patients with recurrence at 1, 2, or 3 years post-TURB).
2.2. Statistical methods The meta-analytic procedures used were those of Peto et al. [7]. This method is a modification of the Mantel-Haenzel method and is based on a fixed effects model. Study data are arranged in a 2 ⫻ 2 matrix and a summary odds ratio (ORp) and its 95% confidence interval is calculated. The outcome (event) of interest is tumor recurrence at specified times following TURB. The expected number of events in the “experimental arm” of each study is calculated and an estimate of the variance of the observed minus expected number of events in each study was then determined as well as the variances across all studies. Prior to estimation of a summary odds ratio, a statistical test for homogeneity was performed (Q). This procedure tests the hypothesis that the effect sizes are equal in all of the studies [5]. If Q exceeds the upper tail critical value of chi-square (P ⬍ 0.05) at k ⫺ 1 degrees of freedom (where k = the number of studies analyzed), the observed variance in study effect size is significantly greater than what would be expected by chance if all studies shared a common population effect size. If the hypothesis that the studies are homogenous is rejected, the studies are not measuring an effect of the same size and calculation of a pooled estimate of effect may be of questionable validity. In this setting, sources of heterogeneity were sought (e.g., via sensitivity analyses).
3. Results The literature search produced a total of 1672 citations. Initial screening yielded 40 references that appeared to meet protocol specifications. Further review found that 29 of these were not eligible for inclusion for various reasons including: 1) study contained only patients with recurrent tumors, 2) contained patients with primary as well as recurrent disease without stratification of data on this parameter, 3) less than 1 year of follow-up, 4) contained patients with incomplete TURB, and 5) lack of data on recurrence rates or study design was nonrandomized. This left 11 RCTs that form the basis of this report [8–18]. Table 1 provides an overview of the included studies. All studies included patients with stage Ta and T1 tumors.
678
M. Huncharek et al. / Journal of Clinical Epidemiology 53 (2000) 676–680
Table 1 Summary of included studies Author
Ref. #
Drug(s) used
Hirao Hirao Akaza
[8] [9] [12]
Niijima
[18]
MRC
[10]
thio-tepa epirubicin adriamycin (2 arms) mitomycin-C (1 arm) adriamycin (2 arms) mitomycin-C (1 arm) thio-tepa (1 arm) thio-tepa (1 arm)
Tolley
[11]
Togashi
[13]
Hirao
[14]
Krege Flamm Tsushima
[15] [17] [16]
mitomycin-C (1 arm) mitomycin-C (1 arm) adriamycin (1 arm) peplomycin (1 arm) neocarzinostat (1 arm) adriamycin (1 arm) mitomycin-C (1 arm) thio-tepa (1 arm) mitomycin-C mitoxantrone adriamycin (1 arm) mitomycin-C
Duration of Tx
Tumor stage/grade
2 years 2 years 1 year
Ta, T1/G1,2 Ta, T1/G1,2 Ta, T1/G1,2
1 month
Ta, T1/G1,2,3
1 year single instillation 1 year
Ta, T1/G1,2,3
1 year
Ta, T1/G1,2
2 years
Ta, T1/G1,2,3
2 years 6 weeks 2 years
Ta, T1/G1,2,3 Ta, T1/G1,2,3 Ta, T1/G1,2,3
Ta, T1/G1,2,3
There were no patients with carcinoma in situ in these studies. Five studies included grade 1 and 2 tumors only, while seven studied included grade 1, 2, and 3 tumors. The most commonly used drug was adriamycin (8 treatment arms) followed by mitomycin-C (7 arms). Thio-tepa and cytosine arabinoside were used in 4 and 3 arms, respectively. Various other drugs were employed in the remaining 5 arms. Initially, all 11 studies were combined in a meta-analysis with the proportion of patients recurring at 1 year as the outcome of interest (Fig. 1). This analysis included a total of 3703 patients and yielded a summary odds ratio (ORp) of 0.56 with a 95% confidence interval of 0.48–0.65 (P ⬍ 0.00001). This shows a highly statistically significant effect for intravesical chemotherapy in reducing bladder tumor recurrence at 1 year. The magnitude of this reduction is, on average, 44%. Calculation of Q (heterogeneity) using these data showed Q to equal 55.6 (with 20 df ), demonstrating heterogeneity across studies. This indicates that the effect sizes (i.e., influence of intravesical chemotherapy on the outcome measure of interest) vary across studies to a greater degree than would be expected by chance. Sources of heterogeneity were sought because any summary estimate of effect may be of dubious validity without a clear understanding of the observed heterogeneity. The studies were therefore reviewed for this purpose. Based on this review, the most likely source of heterogeneity appeared to be the wide variety in treatment schedules employed across studies. Some investigators used only a single intravesical instillation of drug following TURB while others used protocols that consisted of repeated installations over the course of 2 years. This variability was thought to possibly account for the observed heterogeneity and sensitivity analyses were performed to confirm this.
All included studies were stratified by length of treatment into three distinct groups: (1) short-term protocols (which used either a single instillation of drug or a few instillation delivered within a 2-month period), (2) 1-year protocols using multiple drug instillations over the course of 1 year, and (3) 2-year protocols in which treatment consisted of multiple intravesical treatments over 2 years. Meta-analysis and Q statistics were recalculated using data stratified on the above parameters. Studies using short-term treatment protocols were initially combined, which included a total of 1258 patients. The ORp using 1-year recurrence as the outcome was 0.70 with a 95% confidence interval of 0.55–0.90 (P ⫽ 0.006). Q was found to equal 10.2. This was nonsignificant at the P ⬍ 0.05 level, which is consistent with a lack of heterogeneity. This demonstrated that treatment length probably contributed to the initially observed variance in outcomes. An ORp of 0.70 favors a positive impact of intravesical chemotherapy in reducing 1-year tumor recurrence by approximately 30%. Recalculating an ORp for short-term treatment using 2-year recurrence as the outcome of interest showed a similar effect of therapy on reducing 2-year recurrence [i.e., 32% (95% CI ⫽ 0.54–0.85, P ⫽ 0.0007)]. Therefore, short-term intravesical treatment reduces recurrence at 1 and 2 years by approximately 30%. Studies employing multiple intravesical instillations of chemotherapeutics for 1 year were next combined in a meta-analysis. This analysis included 1721 patients and showed an odds ratio of 0.65 (95% CI ⫽ 0.46–0.80, P ⭓ 0.0001), highly favoring intravesical therapy versus TURB alone with a 35% reduction in recurrence at 1 year. Q again showed a lack of heterogeneity (Q ⫽ 10.53, with 7 df). The data on 2-year recurrence as the outcome were consistent with those seen at 1 year [i.e, 31% reduction in 2-year recurrence (95% CI ⫽ 0.57–0.83, P ⬍ 0.0001)]. Five hundred and seventy-five patients in 5 RCTs were treated on protocols for 2 years. The reduction in recurrence at the completion of therapy (i.e., 2 years) was 73%, corresponding to an ORp of 0.27 (95% CI ⫽ 0.19–0.39, P ⫽ 0.0001), a highly significant result. This reflects a profound reduction in 2-year recurrence associated with this treatment scheme. Analyzing the same studies after patients received 1 year of treatment supported the above results [i.e., odds ratio of 0.20 (95% CI ⫽ 0.13–0.29)]. Five studies contained data on 3-year recurrence [8,9,15,16,18]. Due to the small number of studies, a stratified analysis based on treatment schedule was not possible. Nonetheless, 1371 patients were available for the metaanalysis and yielded on ORp of 050 (95% CI ⫽ 0.40–0.62), strongly favoring the use of intravesical chemotherapy (P ⬍ 0.0001) (i.e., a 50% reduction in recurrence at 3 years versus TURB alone). Only three studies contained 5-year survival data, making a rigorous analysis of this endpoint impossible due to the small numbers of patients represented and short follow-up times. A further analysis was performed to determine whether drug type was an important determinant of outcome. When
M. Huncharek et al. / Journal of Clinical Epidemiology 53 (2000) 676–680
679
Fig. 1. Results of meta-analysis combining data from all 3703 patients. Outcome measure for this analysis is proportion of patients with tumor recurrence at 1 year. (Duplicate citation of studies represents multiple treatment arms as outlined in Table 1.)
the available data were stratified by the most common drug classes (i.e., mitomycin-C, adriamycin, thio-tepa) no differences were seen in effect on recurrence due to the presence of statistical heterogeneity. Q was found to be 16.3 for the adriamycin group (P ⫽ 0.02) and 9.9 for thio-tepa (P ⫽ 0.03). The mitomycin-C group was associated with an ORp of 0.45 and a Q of 7.3 with 5 df (P ⬎ 0.10). Of the three drug classes studies, the mitomycin-C group was the only one showing a statistically significant reduction in tumor recurrence in the absence of heterogeneity. These findings suggests that factors other than drug type may have greater effect on tumor recurrence, although the data must be interpreted cautiously do the small numbers involved in the stratified analysis. Nonetheless, this may provide additional support for an important effect of treatment duration as noted previously. 4. Discussion Despite its availability and use over several decades, intravesical chemotherapy prophylaxis in the management of superficial bladder cancer remains controversial. For instance, in a 1995 article, 22 RCTs of intravesical chemotherapy were reviewed with the authors concluding that such treatment had only a minor impact on the incidence of tumor recurrence, averaging only 14% [3]. This paper also suggested that maintenance chemotherapy is less effective than a single postoperative instillation. Unfortunately, inter-
pretation of these data is difficult for several reasons. These include: 1) the analysis combined studies containing patients with primary and recurrent bladder tumors without stratification on this parameter, 2) failure to calculate recurrence rates as specified endpoints (e.g., 1 or 2 years postTURB), and 3) failure to use statistical methods designed for combining data from multiple randomized clinical trials. In an attempt to clarify the ambiguities in the existing published data base, our report presents a meta-analysis to determine the impact of intravesical chemotherapy on tumor recurrence rates at 1, 2, and 3 years post-primary TURB. Using a fixed effects model, our study demonstrates that intravesical therapy produces statistically significant reduction in tumor recurrences at 1, 2, and 3 years in this patient group, ranging from 30% to 80% depending on the endpoint of interest. This is in contrast to the much more modest 14% cited above. This analysis also demonstrates that treatment schedule may also influence recurrence rates with longer term therapy producing greater reduction in recurrence rates than short-term schedules. Existing data on this issue are equivocal [4]. In our study, patients receiving therapy for 2 years showed the greatest reduction in tumor recurrence. Interestingly, 1-year recurrence rates among those patients on 2-year chemotherapy protocols showed greater reduction in recurrence (approx. 80%) when compared with patients treated on 1-year protocols. This may be due to the fact that the treatment schedules for 2-year protocols generally used fre-
680
M. Huncharek et al. / Journal of Clinical Epidemiology 53 (2000) 676–680
quent intravesical instillations of drug compared to 1-year protocols. For instance, Krege et al. [15] instilled mitomycin-C every 2 weeks for the first year and once a month during the second year. This contrasts with many of the 1-year protocols, which delivered treatment every 3 months [11]. This suggests that clinical effect may be dose related. The major strength of the presented analysis is its restriction to a single, well-defined question (i.e., the impact of intravesical chemotherapy on tumor recurrence rates following primary complete TURB at specified endpoints). The study also followed a rigorously defined, prospective protocol using accepted meta-analytic techniques. These characteristics should help provide a clearer interpretation of the existing published information. One of the major limitations of the existing data is the lack of studies with long-term follow-up. Using specified inclusion criteria, data were unavailable on recurrence rates beyond 3 years. This limits assessments of long-term influence of such treatment, which represents an important clinical and economic question. Lack of long-term follow-up data also limits examination of the impact of intravesical chemotherapy on tumor progression and overall survival. As noted earlier, our analysis included only patients undergoing primary TURB of superficial transitional cell carcinoma of the bladder. The effect of intravesical chemotherapy on recurrence rates in patients with recurrent tumors may differ from those treated for primary disease due to differences in tumor biology. We are therefore conducting an additional meta-analysis including those containing patients with recurrent tumors in order to address this issue. It is hoped that this latter analysis will provide further information to clarify the role of such treatment in this disease setting.
References [1] Wingo PA, Tong T, Bolden S. Cancer statistics. CA 1995;45:8. [2] Hall RR, Parmar MKB, Richards AB, et al. Proposal for changes in cystoscopic follow up of patients with bladder cancer and adjuvant intravesical chemotherapy. Br Med J 1994;23:308–11. [3] Lamm DL, Riggs DR, Traynelis C, Nseyo UO. Apparent failure of current intravesical chemotherapy prophylaxis to influence the long term course of superficial transitional cell carcinoma of the bladder. J Urol 1995;153:1444–50.
[4] Herr HW, Laudone VP, Whitmore WF. An overview of intravesical therapy for superficial bladder tumors. J Urol 1987;138:1363–8. [5] Cooper H, Hedges LV. The Handbook of Research Synthesis. New York: Russel Sage Foundation, 1994. [6] Chalmers TC, Lau J. Meta-analysis stimulus for changes in clinical trials. Stat Methods Med Res 1993;2:161–72. [7] Yusuf S, Peto R, Lewis J, et al. Beta-blockade during and after myocardial infarction: an overview of the randomized trials. Prog Cardiovasc Dis 1985;27:335–71. [8] Hirao Y, Okajima E, Ozono S, et al. A prospective randomized study of prophylaxis of tumor recurrence following transurethral resection of superficial bladder cancer—intravesical thio-TEPA versus oral UFT. Cancer Chemother Pharmacol 1992;30:S26–S30. [9] Hirao Y, Ozono S, Momose H, et al. Prospective randomized study of prophylaxis of superficial bladder cancer with epirubicin: the role of a central pathology laboratory. Cancer Chemother Pharmacol 1994;35: S36–S40. [10] Medical Research Council Working Party on Urological Cancer. The effect of intravesical thiotepa on tumor recurrence after endoscopic treatment of newly diagnosed superficial bladder cancer. A further report with long term follow-up of a Medical Research Council randomized trial. Br J Urol 1994;73:632–8. [11] Tolley DA, Parmar MKB, Grigor KM, et al. The effect of intravesical mitomycin-C on recurrence of newly diagnosed superficial bladder cancer: A further report with 7 years follow-up. J Urol 1996;155: 1233–8. [12] Akaza H, Isaka S, Koiso K, et al. Comparative analysis of short term and long term prophylactic intravesical chemotherapy of superficial bladder cancer. Cancer Chemother Pharmacol 1987;20:S91–S96. [13] Togashi M, Shinohara N, Toyota K, et al. Prophylactic chemotherapy for primary and recurrent superficial bladder cancer: preliminary results. Cancer Chemother Pharmacol 1992;30:S21–S25. [14] Hirao Y, Okajima E, Ohara S, et al. Prophylactic treatment for superficial bladder cancer following transurethral resection. Cancer Chemother Pharmacol 1987;20:S85–S90. [15] Krege S, Giani G, Meyer R, et al. A randomized multicenter trial of adjuvant therapy in superficial bladder cancer: transurethral resection only versus transurethral resection plus mitomycin-C versus transurethral resection plus bacillus Calmette–Guerin. J Urol 1996;156: 962–6. [16] Tsushima T, Matsumura Y, Ozaka Y, et al. Prophylactic intravesical instillation therapy with adriamycin and mitomycin-C in patients with superficial bladder cancer. Cancer Chemother Pharmacol 1987;20: S72–S76. [17] Flamm J, Donner G, Oberleitner S, et al. Adjuvant intravesical mitoxantrone after transurethral resection of primary superficial transitional cell carcinoma of the bladder. A prospective randomized study. Eur J Cancer 1995;31A:143–6. [18] Niijima T, Koiso K, Akaza H, et al. Randomized clinical trial on chemoprophylaxis of recurrence in cases of superficial bladder cancer. Cancer Chemother Pharmacol 1983;11:S79–S82.