Tipos de estudios

Se consideraron para la inclusión todos los ensayos clínicos aleatorizados (ECA), los ensayos de grupos paralelos o cuasialeatorizados que comparaban el tratamiento con BCG intravesical versus con MMC intravesical para el cáncer de vejiga urotelial sin invasión muscular. No se excluyeron los estudios sobre la base del estado de la publicación o el idioma de la misma. También se consideraron para su inclusión los estudios que incluían otros agentes intravesicales, pero que tenían grupos de tratamiento que permitían una comparación del BCG y la MMC, cuando los resultados se informaban por separado. Se excluyeron los estudios que comparaban el BCG con placebo/ninguna intervención o MMC versus placebo/ninguna intervención. No se identificaron ensayos cruzados (cross‐over).

Tipos de participantes

Esta revisión consideró los estudios que informaban sobre adultos (de 18 años de edad o más) con cáncer de vejiga urotelial sin invasión muscular de riesgo intermedio y alto (Sobin 2009). También se consideraron los estudios que incluían a participantes con Cis de la vejiga. Cuando los estudios también incluían a participantes con cáncer de vejiga con invasión muscular, solo se consideraron los datos del subconjunto de participantes con cáncer de vejiga sin invasión muscular, cuando dichos estudios presentaban datos estratificados para los pacientes con cáncer de vejiga sin invasión muscular de riesgo intermedio y alto.

Las personas elegibles fueron las que estaban en riesgo intermedio o alto de recidiva o progresión del tumor, o ambos. Cuando los estudios también incluían a participantes con riesgo bajo de recidiva o progresión tumoral, se evaluaron nuevamente los datos del subconjunto de participantes con riesgo intermedio o alto (o ambos) cuando dichos datos se informaban por separado.

El riesgo de recidiva y progresión se definió utilizando las guías de la EAU (Babjuk 2018), que hacen referencia a las tablas de riesgo de la European Organisation for Research and Treatment of Cancer (EORTC) (Sylvester 2006):

• el riesgo bajo se define como: primario, solitario, Ta G1 (neoplasia urotelial papilar de potencial maligno bajo, de grado bajo), menos de 3 cm, sin Cis;

• los tumores de riesgo intermedio se definen como: todos los tumores entre las categorías de riesgo bajo y alto;

• de alto riesgo se refiere a cualquiera de los cuatro requisitos siguientes: Tumores T1; tumor G3 (grado alto) de grado alto; Cis; tumores múltiples, recidivantes y grandes (más de 3 cm) Ta G1G2/tumores de grado bajo (todas estas condiciones deben estar presentes).

De acuerdo con la última guía clínica (Babjuk 2018), también se incluyó a los pacientes en mayor riesgo de recidiva/progresión que se definió como tumores T1 G3 asociados con Cis vesical concurrente o T1 G3 recidivante (o ambos), T1 G3 con Cis en la uretra prostática, histología atípica del carcinoma urotelial o invasión linfovascular.

Tipos de intervenciones

El tratamiento intravesical con agente único con BCG o MMC para la prevención o el tratamiento del cáncer de vejiga urotelial sin invasión muscular de riesgo intermedio y alto después de la RTUB fue elegible para la inclusión. Se consideró apropiado incluir el BCG en cualquier esquema o de cualquier cepa, así como cualquier dosis o esquema de MMC.

Tipos de medida de resultado

Para la inclusión en el estudio, no se utilizó la medición de los resultados evaluados en esta revisión como criterio de elegibilidad.

Búsquedas electrónicas

We applied no date or language restrictions.

We searched the following databases: Cochrane Central Register of Controlled Trials (CENTRAL; included in the Cochrane Library; 2018, Issue 11) latest issue (Appendix 1), MEDLINE and MEDLINE in Process via Ovid from 1946 to 13 November 2017 (Appendix 2), Embase via Ovid from 1974 to 13 November 2017 (Appendix 3), Scopus from 1966 to 16 November 2017 (Appendix 4), Web of Science (Thomson Reuters Web of Knowledge) from 1900 to 16 November 2017 (Appendix 5), and LILACS from 1982 to 16 November 2017 (Appendix 6).

The electronic search were complemented by a search of the World Health Organization International Clinical Trials Registry Platform Search Portal (WHO ICTRP Search Portal; www.who.int/ictrp/en/, no restricted time period) (Appendix 7) and ClinicalTrials.gov (clinicaltrials.gov/, no restricted time period) (Appendix 8) to identify further completed or ongoing trials.

We updated the searches for all relevant databases shortly before publication of the review (23th September 2019) and screened the results for further potentially eligible studies. We documented and reported the search process in detail.

Búsqueda de otros recursos

We manually screened the reference lists of included articles to identify potentially relevant citations. We searched the American Society of Clinical Oncology (ASCO) database for grey literature (2011 to 2018; meetinglibrary.asco.org/). We contacted authors to request missing information.

Selección de los estudios

Two review authors (SS and RD or DD) independently reviewed titles and abstracts of identified references according to the predefined inclusion criteria. Two review authors (SS and RD or DD) independently assessed the full texts of all potentially relevant studies. We resolved disagreements by discussion or, if necessary, with the help of a third review author (JJM or FK). We recorded the reasons for study exclusion in the Characteristics of excluded studies table. We identified duplicate publication of studies by checking potentially relevant references for author names, locations and settings, details of interventions, numbers of participants, baseline data, study date and duration of the study. We used EndNote software to manage the references (endnote.com/).

Extracción y manejo de los datos

Two review authors (SS and DD) independently extracted relevant data on study characteristics, participant population and study setting, follow‐up time, tumour characteristics and relevant comorbidities, intervention characteristics on agent and administration, study methodology, study results and author conclusion using a data extraction form. A third review author (KJ) checked the extracted outcome data relevant to this review as needed for calculation of summary statistics and measures of variance. The data extraction form was based on the recommendations of the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a), and was pilot tested before routine use. The review authors resolved any potential disagreement by consensus or through discussion with a third review author (JJM or FK). In addition, when necessary, we contacted the original investigators. We collected and used the most detailed numerical data in order to facilitate similar analyses of included studies. We displayed the information in the Characteristics of included studies table.

Evaluación del riesgo de sesgo de los estudios incluidos

For the 'Risk of bias' assessment, we used the Cochrane 'Risk of bias' tool for RCTs (Higgins 2011b). Two review authors (SS and DD or LMK) independently assessed all included studies for potential risk of bias. We resolved discrepancies through discussion or by contacting a third review author (JJM or FK). We assessed the following domains.

• Random sequence generation (selection bias).

• Allocation concealment (selection bias).

• Blinding of participants and personnel (performance bias).

• Blinding of outcome assessment (detection bias).

• Incomplete outcome data (attrition bias).

• Selective reporting (reporting bias).

• Other sources of bias.

We judged risk of bias domains as 'low risk', 'high risk' or 'unclear risk' and evaluated individual bias items as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b). We presented a 'Risk of bias' summary figure to illustrate these findings.

We further summarised the risk of bias across domains for each outcome in each included study, as well as across studies and domains for each outcome.

For performance bias (blinding of participants and personnel), we considered all outcomes similarly susceptible to performance bias and assessed them in one group.

For detection bias (blinding of outcome assessment), we grouped outcomes as susceptible to detection bias (subjective) or not susceptible to detection bias (objective). Objective outcomes: time to death from any cause. Subjective outcomes: serious adverse effects, time to recurrence, time to progression, adverse effects and quality of life.

We assessed attrition bias (incomplete outcome data) on a per‐outcome basis and created groups of outcomes based on similar reporting characteristics. Time‐to‐event outcomes: time to death from any cause, time to recurrence, time to progression; adverse effects outcomes: serious adverse effects, adverse effects; quality‐of‐life outcomes.

Medidas del efecto del tratamiento

We extracted hazard ratios (HRs) with 95% confidence intervals (CIs) for time to event outcomes (time to recurrence, time to progression and time to death from any cause). Adjusted HRs based on multivariate analysis were preferred to univariate HRs. An indirect estimation method was used to calculate HRs and their variances if they were not reported (Parmar 1998; Tierney 2007; Williamson 2002). We expressed results of dichotomous outcomes (e.g. serious adverse effects, adverse effects) as risk ratios (RRs) with 95% CIs, results of continuous outcomes (e.g. quality of life) as mean difference (MD) with corresponding 95% CI, unless different studies use different measures to assess the same outcome, in which case we expressed data as standardised mean differences (SMDs) with 95% CIs.

Cuestiones relativas a la unidad de análisis

The unit of analysis was the individual participant. In the event we identified trials with more than two intervention groups for inclusion in the review, we handled these in accordance with guidance provided in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a).

Manejo de los datos faltantes

We contacted the corresponding author of the original publication to request any missing data. We did not impute missing data and considered only the available data in the analyses. We did not conduct best‐case and worst‐case scenarios.

Evaluación de la heterogeneidad

We examined statistical heterogeneity using the I² statistic. The thresholds for interpretation of the I² statistic are in accordance with the definitions presented in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2011):

• 0% to 40% might not be important;

• 30% to 60% may represent moderate heterogeneity;

• 60% to 90% may represent substantial heterogeneity;

• 75% to 100% considerable heterogeneity.

Subgroup analyses was done for the examination of clinical heterogeneity. For details, see Subgroup analysis and investigation of heterogeneity.

Evaluación de los sesgos de notificación

To account for possible publication bias, we conducted a combination of electronic and manual searches of multiple databases without language restrictions. In case of sufficient data, we created funnel plots to assess the likelihood of publication bias. Several explanations can be offered for the asymmetry of a funnel plot, including true heterogeneity of effect with respect to trial size, poor methodological design (and hence bias of small trials) and publication bias. Therefore, we interpreted results with caution (Sterne 2011).

Síntesis de los datos

We performed data synthesis using Review Manager 5 software provided by Cochrane (Review Manager 2014).

In the meta‐analyses, we used the random‐effects model that assumes that the treatment effect among studies varies and, therefore, incorporates the heterogeneity among studies in the synthesis of primary study results. We combined the estimated log HRs using the generic inverse‐variance method, the result of which is presented as pooled HR with 95% CI on a logarithmic scale. HRs were given for BCG compared to MMC, therefore, an HR less than 1 indicates a benefit of BCG. We calculated summary statistics with respect to the RR and its 95% CI using the Mantel‐Haenszel method (Lane 2013).

Three‐arm trials comparing two BCG arms with one MMC arm but without clinical relevant difference in the BCG treatment approaches were included in the meta‐analysis with both treatment arms of BCG versus MMC (Ojea 2007a; Ojea 2007b; Witjes 1996a; Witjes 1996b). The standard error of the HRs were adjusted according to Woods 2010 in order to avoid a unit‐of‐analysis error (i.e. using the participants of the MMC group twice). In Friedrich 2007, we included one MMC arm (six weeks) in the primary meta‐analysis to give attention to the comparable duration of medication in the BCG and the MMC arm. The second MMC arm (three years) was used for a sensitivity analysis.

For adverse effect outcomes, we did not pool study data to give an overall result on adverse effects, as in all studies (except Ojea 2007b) adverse effects were not reported on a per‐patient basis, but as the number of the different adverse effects that had occurred. We chose to present cystitis as a patient‐relevant outcome in Summary of findings table 1.

Análisis de subgrupos e investigación de la heterogeneidad

We explored the following potential sources of clinical heterogeneity using the following subgroup analyses:

• different doses of BCG installations;

• different doses of MMC installations;

• different strains of BCG;

• different BCG maintenance therapies (posthoc subgroup analyses).

We used the fixed‐effect models for the subgroup analyses due to the limited number of available studies (Bender 2018).

Análisis de sensibilidad

We aimed at examining the methodological quality according to risk of bias, by conducting separate meta‐analyses for low risk of bias studies, excluding studies judged as high or unclear (or both) risk of bias. As there were no studies with low risk of bias, this analysis was not performed.

Instead we tested the robustness of results using sensitivity analysis. The fixed‐effect model was used to explore visually if results of the meta‐analysis varied substantially when using a model that does assume homogeneity of effects among studies and gives greater weight to larger studies. Furthermore, the second MMC arm in Friedrich 2007 (three years) was used instead of the six weeks MMC arm for a sensitivity analysis.