A systematic review and meta-analysis of clinical trials of bladder-sparing trimodality treatment for muscle-invasive bladder cancer (MIBC)

Fig. 1 shows the flow-chart of identification and inclusion of trials, as recommended by PRISMA [5] . Overall 224 references were identified and screened. One hundred and twenty two studies were excluded because CRT was not employed in the bladder preservation approach. One hundred and two trials were selected and retrieved for full-text analysis. After discarding duplicates and papers that were updated or did not fulfil our inclusion criteria, the search yielded 31 articles with a total of 3315 evaluated patients[3], [4], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], and [41]. The material was largely heterogeneous with regard to tumour stage, prognostic variables, treatment, outcome evaluation and reporting results. The percent of patients with locally advanced or non resected tumours included in the trials ranged from 0% to more than 50%. Several trials in the continuous course schedule included medically-inoperable patients. Intention-to-treat analysis was reported by most trials although, in some retrospective studies, the results were reported only for patients completing the planned treatment. Radiotherapy total dose and fractionation varied among the studies, ranging from 24 to 48 Gy in 8–16 fractions with hypofractionation, 40 to 65 Gy in 20–32 fractions with conventional fractionation, or 66 to 69 Gy in 44–46 fractions with hyper-fractionation. Also concomitant chemotherapy varied among the studies, with the use of several dose schedules and drug combination (cisplatin ± other drugs, fluorouracil, carboplatin, gemcitabine, mitomycin C, vinblastine, metotrexate, paclitaxel, etc.), with or without neoadjuvant courses. The patient characteristics of the two treatment groups are summarized in Table 1 . Details on treatment, number of patients, and outcomes are reported in Table 2 .

Fig. 1 Flow of identification and inclusion of trials as recommended by PRISMA [5] .

Fig. 2 a shows the cumulative results from the evaluated trials. CR was observed in 78% of patients. LF, i.e. tumours not achieving CR plus later muscle-infiltrating recurrences, occurred in 28% of cases. SC was necessary in 21% of patients. At 5 years, OS and BIS rates were 56% and 42%, respectively. The CR rate appeared significantly better in patients treated with the continuous than split schedule with a HR of 0.513, (95% CI of 0.430–0.611), and an absolute benefit of more than 11% ( Fig. 2 b). No significant differences were observed in 5-year OS, BIS and infiltrating LF rates. Of relevance, SC rates were 19% (268/1416) vs. 25% (287/1144) in the continuous and split regimen, respectively (HR = 1.435, 95%CI = 1.188–1.732), with an absolute benefit of 6% ( Fig. 2 b).

Fig. 2 (a) Cumulative results from all evaluated trials; in parentheses is the number of patients evaluated for each outcome. (b) Forest plot of HRs and 95% CIs of outcomes comparing continuous vs. split treatment course in all evaluated patients.

3.1. Subgroup analysis

A subgroup analysis was attempted, testing the effects of the two treatment regimens on outcomes according to prognostic and treatment variables. As expected, CR and 5-year OS rates were in general significantly better in patients with T2 than >T2 tumour stage (79% vs. 62%,p < 0.0001 and 61% vs. 47%,p = 0.0001, respectively, Fig. 3 a). In this latter group, the 5-year OS rate appeared significantly better in patients treated with the continuous rather than the split treatment course (HR = 0.641, 95%CI = 0.424–0.969) with an absolute benefit of 11% ( Fig. 3 b).

Fig. 3 (a) CR and 5-year OS rates in patients with T2 or >T2 tumour stages. (b) Forest plot of HRs and 95%CIs of CR and 5-year OS rates comparing continuous vs. split treatment course in patients with T2 or >T2 tumour stages.

CR rates resulted better also in patients without rather than with hydronephrosis (77% vs. 38%,p = 0.0001), while the difference in OS rates between these two groups resulted not significant ( Fig. 4 ).

Fig. 4 CR and 5-year OS rates in patients with or without hydronephrosis.

Fig. 5 shows CR (panel a) and OS (panel b) rates as a function of the EQD2. While there seem to be no influence of total dose on 5-year OS rates, CR rates appear to increase when increasing the EQD2.

Fig. 5 Diagram showing (a) CR and (b) 5-year OS rates as a function of EQD2 in all trials of continuous and split treatment course. In the split treatment, CR rate was plotted as a function of dose delivered by induction CRT course, while 5-year OS rate was plotted as a function of dose delivered by both induction and consolidation CRT courses also for patients undergone to SC.

Data on the effect of different drug regimes are too sparse to be correctly evaluated. The addition of other drugs to CP or the use of different drugs does not seem to give a benefit for 5-year OS and CR (data not shown). Because of the limited number of patients receiving drugs other than CP, we analyzed only those patients receiving CP and CP containing regimes. In both these groups, while no significant difference in survival could be detected between the two radiation schedules, significantly higher CR and lower SC rates were observed by continuous compared to split course regime (HR = 0.411, 95%CI = 0.305–0.554 and HR = 1.487, 95%CI = 1.161–1.903, respectively ( Fig. 6 ).

Fig. 6 Forest plot of HRs and 95%CIs of CR and 5-year OS rates comparing continuous vs. split treatment course in patients treated with CP alone or CP plus other drugs.

From the limited, currently available data, the use of NACT did not appear to produce any significant benefit in terms of rate of OS and CR ( Fig. 7 a). However, while no difference in OS was detected between the two radiation schedules in the absence of NACT, significantly better CR and OS HRs were observed in the continuous compared to the split regime when NACT was added to CRT ( Fig. 7 b).

Fig. 7 (a) CR and 5-year OS rates in patients receiving neoadjuvant chemotherapy (NADCT+) or not (NADCT−). (b) Forest plot of HRs and 95%CIs of CR, cystectomy and 5-year OS rates comparing continuous vs. split treatment course in patients receiving or not receiving NACT.

Multivariate analysis testing the significance of treatment type (continuous vs. split), NACT and CP-based treatment for the various endpoints, confirmed the continuous course as an independent prognostic variable only for CR rate (p = 0.0036), while CP containing regimes (yes/no) and NACT (yes/no) were not significant at univariate analysis. For SC, only the CP-based regimen resulted a significant prognostic factor (p = 0.0219) ( Table 3 ).

Table 3 Univariate and multivariate analyses for different end-points.

Endpoints	Variables significant to univariate analysis	Variables significant to multivariate analysis
OS	§	§
CR	Continuous course	Continuous course
BIS	§	§
SC	Continuous course CP-based treatment	CP-based treatment
LF	§	§

^§ No significant association including continuous course, neoadjuvant or CP-based treatment.

Bladder preservation strategies are perceived by many urologists to result in inferior survival in comparison to radical cystectomy (RC) and, therefore, RC is considered as the “gold standard” in the treatment of MIBC. Nevertheless, many patients may be not candidates for radical surgery due to advanced age and/or multiple co-morbidities and need a safe and effective treatment that can be an alternative to radical cystectomy. Furthermore, organ preservation and quality of life in bladder cancer, as with other sites (breast, larynx, anus, prostate, etc.), are increasingly being requested by patients, providing the complete eradication of cancer cells remains the primary goal of the treatment. Appropriate randomized trials would give definitive answers on the possibility of choosing the most beneficial treatment for patients with MIBC. In the meanwhile, a systematic review of radical cystectomy vs. organ-sparing TMT trials, reporting a temporary evaluation of the benefit of these two most important treatment modalities, has been just completed by our group and will be separately reported (paper submitted).

The present overview shows that an average 56% of patients with MIBC who underwent TMT survived at 5 years, with approximately 80% of them living with their own bladder. These seem to be excellent results in consideration that, with the exception of the RTOG 8903 [19] and BC 2001 trials [35] , they are not obtained by well conducted randomized studies, but are results emerging from prospective and retrospective studies conducted by individual institutions reporting the results of the routine therapeutic activity without any patient selection.

Bladder cancer constitutes a heterogeneous population of patients with a variety of factors that may influence the outcomes (age, gender, co-morbidities, T-stage, hydronephrosis, completeness of TURBT, etc.). The comparison of outcomes not within but only across studies raises concern about absence of prognostic balance between the two treatment schedules. Furthermore, the intention-to-treat report, specified only in few papers, may represent another confounding variable.

For the aim of this study, the trials were simply grouped into two major treatment approaches, i.e. continuous and split treatment course. Although at univariate analysis the continuous treatment showed better outcomes than split regime, multivariate analysis confirmed the continuous schedule as a significant variable only for CR rate. Of relevance, SC rate is unfortunately not reported in all the investigated studies and, owing to the presence of other confounding factors, a higher number of patients is needed to reach the significance at a multivariate analysis. The higher CR and lower SC rates observed at the univariate analysis by continous treatment, may be due to the fact that patients undergone split course regimen received, by protocol, an early assessment for SC, potentially leading to some unnecessary SC avoidable by a later outcome evaluation. On the other hand, the continuous treatment course allows a sufficient time to exclude SC in patients who either reached CR after the full CRT course or developed distant disease progression during the treatment/follow-up. In this context, although the therapeutic schedule was not shown to significantly influence the 5-year OS rates, continuous treatment seems to offer some advantage for patient quality of life in terms of bladder sparing.

The worst results were observed in patients with stage >T2 or obstructive uropathy. In these cases, however, continuous seems to be more effective than split treatment, with significantly higher CR rates in both with/without hydronephrosis subgroups, and significantly better OS rates in patients with later tumour stages. In these latter, SC does not seem to be sufficient to compensate for the lower CR rate observed after the split course.

As expected, higher CR rates were observed by increasing total RT dose. Owing to the early evaluation after the induction CRT course, when RT dose was only partially delivered, CR rates resulted to be lower in the split than continuous regimen where the evaluation was done after full RT course. Instead no effect of total dose, in general, was observed on 5-year OS rates, possibly because SC was able to compensate for the lower CR rate achieved by the induction CRT course of the split treatment.

Although other drugs in combination or not with CP have shown to be very effective[21], [23], [32], [34], [36], and [38], no conclusion could be made whether their addition to CP or their use in place of CP is better than using CP alone, since CP alone was used in most trials. However, using CP either alone or in combination with other drugs, higher CR and lower SC rates were observed with the continuous rather than the split regime.

Data on the benefit of using NACT are sparse. Although some trials showed improved outcomes with NACT in patients treated by radical cystectomy or RT alone, a recent report from Cochrane Database [42] concluded that there was insufficient information to obtain a definitive answer to the question of whether NACT improves survival in patients with MIBC. The RTOG 8903 randomized trial [18] concluded that two cycles of NACT with the methotrexate, cisplatin, and vinblastine (MCV) regime was not shown to have an impact on outcomes with respect to CRT alone in patients with MIBC, although the number of patients was too small to detect small significant differences. The results of this overview on more than 900 patients confirm the lack of benefit of the addition of NACT to CRT. However, by NACT addition, CR, OS and SC rates were significantly improved in the continuous with respect to the split treatment. Appropriate randomized trials should be addressed to test this suggestion.

In the last two decades several retrospective and prospective trials have shown that the trimodality therapy (TMT), based on the concurrent delivery of chemotherapy and radiotherapy after muscle-infiltrating bladder tumours (TURBT) can be a safe and effective treatment in the management of muscle-invasive bladder cancer (MIBC). In fact, this strategy has the advantage, with respect to radical cystectomy, of preserving, in most of the patients, a normally functional bladder, reserving cystectomy as a salvage option only in cases with an infiltrating local failure (LF) without distant metastases. The rationale for the combination of chemotherapy and radiotherapy (CRT) is based on two main premises. First, clinical and autopsy data have shown that micrometastases may occur concurrently with MIBC, and their frequency increases with tumour stage [1] . Therefore, it is expected that the addition of chemotherapy to radiotherapy may reduce the probability of distant failure by eradicating occult metastases. Second, some drugs such as fluorouracil, cisplatin, gemcitabine, paclitaxel, etc. act as radio-sensitizers, rendering cancer cells more sensitive to radiation [2] . To avoid a salvage cystectomy (SC), the immediate most important goal of TMT is that of achieving and maintaining the highest rate of CR with the lowest rate of infiltrating LF. Two main different CRT approaches have been employed following TURBT. In protocols developed at the Massachusetts General Hospital (MGH) and adopted in the Radiation Therapy Oncology Group (RTOG) trials, candidates for bladder preservation are selected according to their response to induction CRT, with a cystoscopic assessment of response after a preliminary course. In these protocols, consolidation CRT (split treatment) is given only to completely responding patients, whereas all others undergo SC with a curative intent [3] . The second bladder preservation approach, mainly used in the University of Erlangen [4] and adopted by many European centres, consists of the delivery of an upfront full-dose CRT course (continuous treatment). In this case, patients with an incomplete response and continuing pelvis confined disease at post-treatment evaluation undergo SC. These two therapeutic approaches have been planned with a different intent: to reduce the risk of both uncontrolled loco-regional disease and SC-associated complications, the former approach gives priority to cancer control by carrying out SC in patients with an incomplete response after induction CRT, with a minimal delay after low dose preoperative CRT. The aim of the continuous treatment schedule is that of increasing the chances of bladder preservation by giving a more intensive CRT course and more time for tumours to respond, reserving SC only for patients with an incomplete response and continuing pelvis confined disease.

With the aim of identifying the optimal strategy for bladder sparing in the treatment of MIBC, we performed a systematic review and meta-analysis of the published trials employing TMT.

Fig. 1 shows the flow-chart of identification and inclusion of trials, as recommended by PRISMA [5] . Overall 224 references were identified and screened. One hundred and twenty two studies were excluded because CRT was not employed in the bladder preservation approach. One hundred and two trials were selected and retrieved for full-text analysis. After discarding duplicates and papers that were updated or did not fulfil our inclusion criteria, the search yielded 31 articles with a total of 3315 evaluated patients[3], [4], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], and [41]. The material was largely heterogeneous with regard to tumour stage, prognostic variables, treatment, outcome evaluation and reporting results. The percent of patients with locally advanced or non resected tumours included in the trials ranged from 0% to more than 50%. Several trials in the continuous course schedule included medically-inoperable patients. Intention-to-treat analysis was reported by most trials although, in some retrospective studies, the results were reported only for patients completing the planned treatment. Radiotherapy total dose and fractionation varied among the studies, ranging from 24 to 48 Gy in 8–16 fractions with hypofractionation, 40 to 65 Gy in 20–32 fractions with conventional fractionation, or 66 to 69 Gy in 44–46 fractions with hyper-fractionation. Also concomitant chemotherapy varied among the studies, with the use of several dose schedules and drug combination (cisplatin ± other drugs, fluorouracil, carboplatin, gemcitabine, mitomycin C, vinblastine, metotrexate, paclitaxel, etc.), with or without neoadjuvant courses. The patient characteristics of the two treatment groups are summarized in Table 1 . Details on treatment, number of patients, and outcomes are reported in Table 2 .

Fig. 1 Flow of identification and inclusion of trials as recommended by PRISMA [5] .

Fig. 2 a shows the cumulative results from the evaluated trials. CR was observed in 78% of patients. LF, i.e. tumours not achieving CR plus later muscle-infiltrating recurrences, occurred in 28% of cases. SC was necessary in 21% of patients. At 5 years, OS and BIS rates were 56% and 42%, respectively. The CR rate appeared significantly better in patients treated with the continuous than split schedule with a HR of 0.513, (95% CI of 0.430–0.611), and an absolute benefit of more than 11% ( Fig. 2 b). No significant differences were observed in 5-year OS, BIS and infiltrating LF rates. Of relevance, SC rates were 19% (268/1416) vs. 25% (287/1144) in the continuous and split regimen, respectively (HR = 1.435, 95%CI = 1.188–1.732), with an absolute benefit of 6% ( Fig. 2 b).

Fig. 2 (a) Cumulative results from all evaluated trials; in parentheses is the number of patients evaluated for each outcome. (b) Forest plot of HRs and 95% CIs of outcomes comparing continuous vs. split treatment course in all evaluated patients.

3.1. Subgroup analysis

A subgroup analysis was attempted, testing the effects of the two treatment regimens on outcomes according to prognostic and treatment variables. As expected, CR and 5-year OS rates were in general significantly better in patients with T2 than >T2 tumour stage (79% vs. 62%,p < 0.0001 and 61% vs. 47%,p = 0.0001, respectively, Fig. 3 a). In this latter group, the 5-year OS rate appeared significantly better in patients treated with the continuous rather than the split treatment course (HR = 0.641, 95%CI = 0.424–0.969) with an absolute benefit of 11% ( Fig. 3 b).

Fig. 3 (a) CR and 5-year OS rates in patients with T2 or >T2 tumour stages. (b) Forest plot of HRs and 95%CIs of CR and 5-year OS rates comparing continuous vs. split treatment course in patients with T2 or >T2 tumour stages.

CR rates resulted better also in patients without rather than with hydronephrosis (77% vs. 38%,p = 0.0001), while the difference in OS rates between these two groups resulted not significant ( Fig. 4 ).

Fig. 4 CR and 5-year OS rates in patients with or without hydronephrosis.

Fig. 5 shows CR (panel a) and OS (panel b) rates as a function of the EQD2. While there seem to be no influence of total dose on 5-year OS rates, CR rates appear to increase when increasing the EQD2.

Fig. 5 Diagram showing (a) CR and (b) 5-year OS rates as a function of EQD2 in all trials of continuous and split treatment course. In the split treatment, CR rate was plotted as a function of dose delivered by induction CRT course, while 5-year OS rate was plotted as a function of dose delivered by both induction and consolidation CRT courses also for patients undergone to SC.

Data on the effect of different drug regimes are too sparse to be correctly evaluated. The addition of other drugs to CP or the use of different drugs does not seem to give a benefit for 5-year OS and CR (data not shown). Because of the limited number of patients receiving drugs other than CP, we analyzed only those patients receiving CP and CP containing regimes. In both these groups, while no significant difference in survival could be detected between the two radiation schedules, significantly higher CR and lower SC rates were observed by continuous compared to split course regime (HR = 0.411, 95%CI = 0.305–0.554 and HR = 1.487, 95%CI = 1.161–1.903, respectively ( Fig. 6 ).

Fig. 6 Forest plot of HRs and 95%CIs of CR and 5-year OS rates comparing continuous vs. split treatment course in patients treated with CP alone or CP plus other drugs.

From the limited, currently available data, the use of NACT did not appear to produce any significant benefit in terms of rate of OS and CR ( Fig. 7 a). However, while no difference in OS was detected between the two radiation schedules in the absence of NACT, significantly better CR and OS HRs were observed in the continuous compared to the split regime when NACT was added to CRT ( Fig. 7 b).

Fig. 7 (a) CR and 5-year OS rates in patients receiving neoadjuvant chemotherapy (NADCT+) or not (NADCT−). (b) Forest plot of HRs and 95%CIs of CR, cystectomy and 5-year OS rates comparing continuous vs. split treatment course in patients receiving or not receiving NACT.

Multivariate analysis testing the significance of treatment type (continuous vs. split), NACT and CP-based treatment for the various endpoints, confirmed the continuous course as an independent prognostic variable only for CR rate (p = 0.0036), while CP containing regimes (yes/no) and NACT (yes/no) were not significant at univariate analysis. For SC, only the CP-based regimen resulted a significant prognostic factor (p = 0.0219) ( Table 3 ).

Table 3 Univariate and multivariate analyses for different end-points.

Endpoints	Variables significant to univariate analysis	Variables significant to multivariate analysis
OS	§	§
CR	Continuous course	Continuous course
BIS	§	§
SC	Continuous course CP-based treatment	CP-based treatment
LF	§	§

^§ No significant association including continuous course, neoadjuvant or CP-based treatment.

Bladder preservation strategies are perceived by many urologists to result in inferior survival in comparison to radical cystectomy (RC) and, therefore, RC is considered as the “gold standard” in the treatment of MIBC. Nevertheless, many patients may be not candidates for radical surgery due to advanced age and/or multiple co-morbidities and need a safe and effective treatment that can be an alternative to radical cystectomy. Furthermore, organ preservation and quality of life in bladder cancer, as with other sites (breast, larynx, anus, prostate, etc.), are increasingly being requested by patients, providing the complete eradication of cancer cells remains the primary goal of the treatment. Appropriate randomized trials would give definitive answers on the possibility of choosing the most beneficial treatment for patients with MIBC. In the meanwhile, a systematic review of radical cystectomy vs. organ-sparing TMT trials, reporting a temporary evaluation of the benefit of these two most important treatment modalities, has been just completed by our group and will be separately reported (paper submitted).

The present overview shows that an average 56% of patients with MIBC who underwent TMT survived at 5 years, with approximately 80% of them living with their own bladder. These seem to be excellent results in consideration that, with the exception of the RTOG 8903 [19] and BC 2001 trials [35] , they are not obtained by well conducted randomized studies, but are results emerging from prospective and retrospective studies conducted by individual institutions reporting the results of the routine therapeutic activity without any patient selection.

Bladder cancer constitutes a heterogeneous population of patients with a variety of factors that may influence the outcomes (age, gender, co-morbidities, T-stage, hydronephrosis, completeness of TURBT, etc.). The comparison of outcomes not within but only across studies raises concern about absence of prognostic balance between the two treatment schedules. Furthermore, the intention-to-treat report, specified only in few papers, may represent another confounding variable.

For the aim of this study, the trials were simply grouped into two major treatment approaches, i.e. continuous and split treatment course. Although at univariate analysis the continuous treatment showed better outcomes than split regime, multivariate analysis confirmed the continuous schedule as a significant variable only for CR rate. Of relevance, SC rate is unfortunately not reported in all the investigated studies and, owing to the presence of other confounding factors, a higher number of patients is needed to reach the significance at a multivariate analysis. The higher CR and lower SC rates observed at the univariate analysis by continous treatment, may be due to the fact that patients undergone split course regimen received, by protocol, an early assessment for SC, potentially leading to some unnecessary SC avoidable by a later outcome evaluation. On the other hand, the continuous treatment course allows a sufficient time to exclude SC in patients who either reached CR after the full CRT course or developed distant disease progression during the treatment/follow-up. In this context, although the therapeutic schedule was not shown to significantly influence the 5-year OS rates, continuous treatment seems to offer some advantage for patient quality of life in terms of bladder sparing.

The worst results were observed in patients with stage >T2 or obstructive uropathy. In these cases, however, continuous seems to be more effective than split treatment, with significantly higher CR rates in both with/without hydronephrosis subgroups, and significantly better OS rates in patients with later tumour stages. In these latter, SC does not seem to be sufficient to compensate for the lower CR rate observed after the split course.

As expected, higher CR rates were observed by increasing total RT dose. Owing to the early evaluation after the induction CRT course, when RT dose was only partially delivered, CR rates resulted to be lower in the split than continuous regimen where the evaluation was done after full RT course. Instead no effect of total dose, in general, was observed on 5-year OS rates, possibly because SC was able to compensate for the lower CR rate achieved by the induction CRT course of the split treatment.

Although other drugs in combination or not with CP have shown to be very effective[21], [23], [32], [34], [36], and [38], no conclusion could be made whether their addition to CP or their use in place of CP is better than using CP alone, since CP alone was used in most trials. However, using CP either alone or in combination with other drugs, higher CR and lower SC rates were observed with the continuous rather than the split regime.

Data on the benefit of using NACT are sparse. Although some trials showed improved outcomes with NACT in patients treated by radical cystectomy or RT alone, a recent report from Cochrane Database [42] concluded that there was insufficient information to obtain a definitive answer to the question of whether NACT improves survival in patients with MIBC. The RTOG 8903 randomized trial [18] concluded that two cycles of NACT with the methotrexate, cisplatin, and vinblastine (MCV) regime was not shown to have an impact on outcomes with respect to CRT alone in patients with MIBC, although the number of patients was too small to detect small significant differences. The results of this overview on more than 900 patients confirm the lack of benefit of the addition of NACT to CRT. However, by NACT addition, CR, OS and SC rates were significantly improved in the continuous with respect to the split treatment. Appropriate randomized trials should be addressed to test this suggestion.

In the last two decades several retrospective and prospective trials have shown that the trimodality therapy (TMT), based on the concurrent delivery of chemotherapy and radiotherapy after muscle-infiltrating bladder tumours (TURBT) can be a safe and effective treatment in the management of muscle-invasive bladder cancer (MIBC). In fact, this strategy has the advantage, with respect to radical cystectomy, of preserving, in most of the patients, a normally functional bladder, reserving cystectomy as a salvage option only in cases with an infiltrating local failure (LF) without distant metastases. The rationale for the combination of chemotherapy and radiotherapy (CRT) is based on two main premises. First, clinical and autopsy data have shown that micrometastases may occur concurrently with MIBC, and their frequency increases with tumour stage [1] . Therefore, it is expected that the addition of chemotherapy to radiotherapy may reduce the probability of distant failure by eradicating occult metastases. Second, some drugs such as fluorouracil, cisplatin, gemcitabine, paclitaxel, etc. act as radio-sensitizers, rendering cancer cells more sensitive to radiation [2] . To avoid a salvage cystectomy (SC), the immediate most important goal of TMT is that of achieving and maintaining the highest rate of CR with the lowest rate of infiltrating LF. Two main different CRT approaches have been employed following TURBT. In protocols developed at the Massachusetts General Hospital (MGH) and adopted in the Radiation Therapy Oncology Group (RTOG) trials, candidates for bladder preservation are selected according to their response to induction CRT, with a cystoscopic assessment of response after a preliminary course. In these protocols, consolidation CRT (split treatment) is given only to completely responding patients, whereas all others undergo SC with a curative intent [3] . The second bladder preservation approach, mainly used in the University of Erlangen [4] and adopted by many European centres, consists of the delivery of an upfront full-dose CRT course (continuous treatment). In this case, patients with an incomplete response and continuing pelvis confined disease at post-treatment evaluation undergo SC. These two therapeutic approaches have been planned with a different intent: to reduce the risk of both uncontrolled loco-regional disease and SC-associated complications, the former approach gives priority to cancer control by carrying out SC in patients with an incomplete response after induction CRT, with a minimal delay after low dose preoperative CRT. The aim of the continuous treatment schedule is that of increasing the chances of bladder preservation by giving a more intensive CRT course and more time for tumours to respond, reserving SC only for patients with an incomplete response and continuing pelvis confined disease.

With the aim of identifying the optimal strategy for bladder sparing in the treatment of MIBC, we performed a systematic review and meta-analysis of the published trials employing TMT.

Fig. 1 shows the flow-chart of identification and inclusion of trials, as recommended by PRISMA [5] . Overall 224 references were identified and screened. One hundred and twenty two studies were excluded because CRT was not employed in the bladder preservation approach. One hundred and two trials were selected and retrieved for full-text analysis. After discarding duplicates and papers that were updated or did not fulfil our inclusion criteria, the search yielded 31 articles with a total of 3315 evaluated patients[3], [4], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], and [41]. The material was largely heterogeneous with regard to tumour stage, prognostic variables, treatment, outcome evaluation and reporting results. The percent of patients with locally advanced or non resected tumours included in the trials ranged from 0% to more than 50%. Several trials in the continuous course schedule included medically-inoperable patients. Intention-to-treat analysis was reported by most trials although, in some retrospective studies, the results were reported only for patients completing the planned treatment. Radiotherapy total dose and fractionation varied among the studies, ranging from 24 to 48 Gy in 8–16 fractions with hypofractionation, 40 to 65 Gy in 20–32 fractions with conventional fractionation, or 66 to 69 Gy in 44–46 fractions with hyper-fractionation. Also concomitant chemotherapy varied among the studies, with the use of several dose schedules and drug combination (cisplatin ± other drugs, fluorouracil, carboplatin, gemcitabine, mitomycin C, vinblastine, metotrexate, paclitaxel, etc.), with or without neoadjuvant courses. The patient characteristics of the two treatment groups are summarized in Table 1 . Details on treatment, number of patients, and outcomes are reported in Table 2 .

Fig. 1 Flow of identification and inclusion of trials as recommended by PRISMA [5] .

Fig. 2 a shows the cumulative results from the evaluated trials. CR was observed in 78% of patients. LF, i.e. tumours not achieving CR plus later muscle-infiltrating recurrences, occurred in 28% of cases. SC was necessary in 21% of patients. At 5 years, OS and BIS rates were 56% and 42%, respectively. The CR rate appeared significantly better in patients treated with the continuous than split schedule with a HR of 0.513, (95% CI of 0.430–0.611), and an absolute benefit of more than 11% ( Fig. 2 b). No significant differences were observed in 5-year OS, BIS and infiltrating LF rates. Of relevance, SC rates were 19% (268/1416) vs. 25% (287/1144) in the continuous and split regimen, respectively (HR = 1.435, 95%CI = 1.188–1.732), with an absolute benefit of 6% ( Fig. 2 b).

Fig. 2 (a) Cumulative results from all evaluated trials; in parentheses is the number of patients evaluated for each outcome. (b) Forest plot of HRs and 95% CIs of outcomes comparing continuous vs. split treatment course in all evaluated patients.

3.1. Subgroup analysis

A subgroup analysis was attempted, testing the effects of the two treatment regimens on outcomes according to prognostic and treatment variables. As expected, CR and 5-year OS rates were in general significantly better in patients with T2 than >T2 tumour stage (79% vs. 62%,p < 0.0001 and 61% vs. 47%,p = 0.0001, respectively, Fig. 3 a). In this latter group, the 5-year OS rate appeared significantly better in patients treated with the continuous rather than the split treatment course (HR = 0.641, 95%CI = 0.424–0.969) with an absolute benefit of 11% ( Fig. 3 b).

Fig. 3 (a) CR and 5-year OS rates in patients with T2 or >T2 tumour stages. (b) Forest plot of HRs and 95%CIs of CR and 5-year OS rates comparing continuous vs. split treatment course in patients with T2 or >T2 tumour stages.

CR rates resulted better also in patients without rather than with hydronephrosis (77% vs. 38%,p = 0.0001), while the difference in OS rates between these two groups resulted not significant ( Fig. 4 ).

Fig. 4 CR and 5-year OS rates in patients with or without hydronephrosis.

Fig. 5 shows CR (panel a) and OS (panel b) rates as a function of the EQD2. While there seem to be no influence of total dose on 5-year OS rates, CR rates appear to increase when increasing the EQD2.

Fig. 5 Diagram showing (a) CR and (b) 5-year OS rates as a function of EQD2 in all trials of continuous and split treatment course. In the split treatment, CR rate was plotted as a function of dose delivered by induction CRT course, while 5-year OS rate was plotted as a function of dose delivered by both induction and consolidation CRT courses also for patients undergone to SC.

Data on the effect of different drug regimes are too sparse to be correctly evaluated. The addition of other drugs to CP or the use of different drugs does not seem to give a benefit for 5-year OS and CR (data not shown). Because of the limited number of patients receiving drugs other than CP, we analyzed only those patients receiving CP and CP containing regimes. In both these groups, while no significant difference in survival could be detected between the two radiation schedules, significantly higher CR and lower SC rates were observed by continuous compared to split course regime (HR = 0.411, 95%CI = 0.305–0.554 and HR = 1.487, 95%CI = 1.161–1.903, respectively ( Fig. 6 ).

Fig. 6 Forest plot of HRs and 95%CIs of CR and 5-year OS rates comparing continuous vs. split treatment course in patients treated with CP alone or CP plus other drugs.

From the limited, currently available data, the use of NACT did not appear to produce any significant benefit in terms of rate of OS and CR ( Fig. 7 a). However, while no difference in OS was detected between the two radiation schedules in the absence of NACT, significantly better CR and OS HRs were observed in the continuous compared to the split regime when NACT was added to CRT ( Fig. 7 b).

Fig. 7 (a) CR and 5-year OS rates in patients receiving neoadjuvant chemotherapy (NADCT+) or not (NADCT−). (b) Forest plot of HRs and 95%CIs of CR, cystectomy and 5-year OS rates comparing continuous vs. split treatment course in patients receiving or not receiving NACT.

Multivariate analysis testing the significance of treatment type (continuous vs. split), NACT and CP-based treatment for the various endpoints, confirmed the continuous course as an independent prognostic variable only for CR rate (p = 0.0036), while CP containing regimes (yes/no) and NACT (yes/no) were not significant at univariate analysis. For SC, only the CP-based regimen resulted a significant prognostic factor (p = 0.0219) ( Table 3 ).

Table 3 Univariate and multivariate analyses for different end-points.

Endpoints	Variables significant to univariate analysis	Variables significant to multivariate analysis
OS	§	§
CR	Continuous course	Continuous course
BIS	§	§
SC	Continuous course CP-based treatment	CP-based treatment
LF	§	§

^§ No significant association including continuous course, neoadjuvant or CP-based treatment.

Bladder preservation strategies are perceived by many urologists to result in inferior survival in comparison to radical cystectomy (RC) and, therefore, RC is considered as the “gold standard” in the treatment of MIBC. Nevertheless, many patients may be not candidates for radical surgery due to advanced age and/or multiple co-morbidities and need a safe and effective treatment that can be an alternative to radical cystectomy. Furthermore, organ preservation and quality of life in bladder cancer, as with other sites (breast, larynx, anus, prostate, etc.), are increasingly being requested by patients, providing the complete eradication of cancer cells remains the primary goal of the treatment. Appropriate randomized trials would give definitive answers on the possibility of choosing the most beneficial treatment for patients with MIBC. In the meanwhile, a systematic review of radical cystectomy vs. organ-sparing TMT trials, reporting a temporary evaluation of the benefit of these two most important treatment modalities, has been just completed by our group and will be separately reported (paper submitted).

The present overview shows that an average 56% of patients with MIBC who underwent TMT survived at 5 years, with approximately 80% of them living with their own bladder. These seem to be excellent results in consideration that, with the exception of the RTOG 8903 [19] and BC 2001 trials [35] , they are not obtained by well conducted randomized studies, but are results emerging from prospective and retrospective studies conducted by individual institutions reporting the results of the routine therapeutic activity without any patient selection.

Bladder cancer constitutes a heterogeneous population of patients with a variety of factors that may influence the outcomes (age, gender, co-morbidities, T-stage, hydronephrosis, completeness of TURBT, etc.). The comparison of outcomes not within but only across studies raises concern about absence of prognostic balance between the two treatment schedules. Furthermore, the intention-to-treat report, specified only in few papers, may represent another confounding variable.

For the aim of this study, the trials were simply grouped into two major treatment approaches, i.e. continuous and split treatment course. Although at univariate analysis the continuous treatment showed better outcomes than split regime, multivariate analysis confirmed the continuous schedule as a significant variable only for CR rate. Of relevance, SC rate is unfortunately not reported in all the investigated studies and, owing to the presence of other confounding factors, a higher number of patients is needed to reach the significance at a multivariate analysis. The higher CR and lower SC rates observed at the univariate analysis by continous treatment, may be due to the fact that patients undergone split course regimen received, by protocol, an early assessment for SC, potentially leading to some unnecessary SC avoidable by a later outcome evaluation. On the other hand, the continuous treatment course allows a sufficient time to exclude SC in patients who either reached CR after the full CRT course or developed distant disease progression during the treatment/follow-up. In this context, although the therapeutic schedule was not shown to significantly influence the 5-year OS rates, continuous treatment seems to offer some advantage for patient quality of life in terms of bladder sparing.

The worst results were observed in patients with stage >T2 or obstructive uropathy. In these cases, however, continuous seems to be more effective than split treatment, with significantly higher CR rates in both with/without hydronephrosis subgroups, and significantly better OS rates in patients with later tumour stages. In these latter, SC does not seem to be sufficient to compensate for the lower CR rate observed after the split course.

As expected, higher CR rates were observed by increasing total RT dose. Owing to the early evaluation after the induction CRT course, when RT dose was only partially delivered, CR rates resulted to be lower in the split than continuous regimen where the evaluation was done after full RT course. Instead no effect of total dose, in general, was observed on 5-year OS rates, possibly because SC was able to compensate for the lower CR rate achieved by the induction CRT course of the split treatment.

Although other drugs in combination or not with CP have shown to be very effective[21], [23], [32], [34], [36], and [38], no conclusion could be made whether their addition to CP or their use in place of CP is better than using CP alone, since CP alone was used in most trials. However, using CP either alone or in combination with other drugs, higher CR and lower SC rates were observed with the continuous rather than the split regime.

Data on the benefit of using NACT are sparse. Although some trials showed improved outcomes with NACT in patients treated by radical cystectomy or RT alone, a recent report from Cochrane Database [42] concluded that there was insufficient information to obtain a definitive answer to the question of whether NACT improves survival in patients with MIBC. The RTOG 8903 randomized trial [18] concluded that two cycles of NACT with the methotrexate, cisplatin, and vinblastine (MCV) regime was not shown to have an impact on outcomes with respect to CRT alone in patients with MIBC, although the number of patients was too small to detect small significant differences. The results of this overview on more than 900 patients confirm the lack of benefit of the addition of NACT to CRT. However, by NACT addition, CR, OS and SC rates were significantly improved in the continuous with respect to the split treatment. Appropriate randomized trials should be addressed to test this suggestion.

In the last two decades several retrospective and prospective trials have shown that the trimodality therapy (TMT), based on the concurrent delivery of chemotherapy and radiotherapy after muscle-infiltrating bladder tumours (TURBT) can be a safe and effective treatment in the management of muscle-invasive bladder cancer (MIBC). In fact, this strategy has the advantage, with respect to radical cystectomy, of preserving, in most of the patients, a normally functional bladder, reserving cystectomy as a salvage option only in cases with an infiltrating local failure (LF) without distant metastases. The rationale for the combination of chemotherapy and radiotherapy (CRT) is based on two main premises. First, clinical and autopsy data have shown that micrometastases may occur concurrently with MIBC, and their frequency increases with tumour stage [1] . Therefore, it is expected that the addition of chemotherapy to radiotherapy may reduce the probability of distant failure by eradicating occult metastases. Second, some drugs such as fluorouracil, cisplatin, gemcitabine, paclitaxel, etc. act as radio-sensitizers, rendering cancer cells more sensitive to radiation [2] . To avoid a salvage cystectomy (SC), the immediate most important goal of TMT is that of achieving and maintaining the highest rate of CR with the lowest rate of infiltrating LF. Two main different CRT approaches have been employed following TURBT. In protocols developed at the Massachusetts General Hospital (MGH) and adopted in the Radiation Therapy Oncology Group (RTOG) trials, candidates for bladder preservation are selected according to their response to induction CRT, with a cystoscopic assessment of response after a preliminary course. In these protocols, consolidation CRT (split treatment) is given only to completely responding patients, whereas all others undergo SC with a curative intent [3] . The second bladder preservation approach, mainly used in the University of Erlangen [4] and adopted by many European centres, consists of the delivery of an upfront full-dose CRT course (continuous treatment). In this case, patients with an incomplete response and continuing pelvis confined disease at post-treatment evaluation undergo SC. These two therapeutic approaches have been planned with a different intent: to reduce the risk of both uncontrolled loco-regional disease and SC-associated complications, the former approach gives priority to cancer control by carrying out SC in patients with an incomplete response after induction CRT, with a minimal delay after low dose preoperative CRT. The aim of the continuous treatment schedule is that of increasing the chances of bladder preservation by giving a more intensive CRT course and more time for tumours to respond, reserving SC only for patients with an incomplete response and continuing pelvis confined disease.

With the aim of identifying the optimal strategy for bladder sparing in the treatment of MIBC, we performed a systematic review and meta-analysis of the published trials employing TMT.

Fig. 1 shows the flow-chart of identification and inclusion of trials, as recommended by PRISMA [5] . Overall 224 references were identified and screened. One hundred and twenty two studies were excluded because CRT was not employed in the bladder preservation approach. One hundred and two trials were selected and retrieved for full-text analysis. After discarding duplicates and papers that were updated or did not fulfil our inclusion criteria, the search yielded 31 articles with a total of 3315 evaluated patients[3], [4], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], and [41]. The material was largely heterogeneous with regard to tumour stage, prognostic variables, treatment, outcome evaluation and reporting results. The percent of patients with locally advanced or non resected tumours included in the trials ranged from 0% to more than 50%. Several trials in the continuous course schedule included medically-inoperable patients. Intention-to-treat analysis was reported by most trials although, in some retrospective studies, the results were reported only for patients completing the planned treatment. Radiotherapy total dose and fractionation varied among the studies, ranging from 24 to 48 Gy in 8–16 fractions with hypofractionation, 40 to 65 Gy in 20–32 fractions with conventional fractionation, or 66 to 69 Gy in 44–46 fractions with hyper-fractionation. Also concomitant chemotherapy varied among the studies, with the use of several dose schedules and drug combination (cisplatin ± other drugs, fluorouracil, carboplatin, gemcitabine, mitomycin C, vinblastine, metotrexate, paclitaxel, etc.), with or without neoadjuvant courses. The patient characteristics of the two treatment groups are summarized in Table 1 . Details on treatment, number of patients, and outcomes are reported in Table 2 .

Fig. 1 Flow of identification and inclusion of trials as recommended by PRISMA [5] .

Fig. 2 a shows the cumulative results from the evaluated trials. CR was observed in 78% of patients. LF, i.e. tumours not achieving CR plus later muscle-infiltrating recurrences, occurred in 28% of cases. SC was necessary in 21% of patients. At 5 years, OS and BIS rates were 56% and 42%, respectively. The CR rate appeared significantly better in patients treated with the continuous than split schedule with a HR of 0.513, (95% CI of 0.430–0.611), and an absolute benefit of more than 11% ( Fig. 2 b). No significant differences were observed in 5-year OS, BIS and infiltrating LF rates. Of relevance, SC rates were 19% (268/1416) vs. 25% (287/1144) in the continuous and split regimen, respectively (HR = 1.435, 95%CI = 1.188–1.732), with an absolute benefit of 6% ( Fig. 2 b).

Fig. 2 (a) Cumulative results from all evaluated trials; in parentheses is the number of patients evaluated for each outcome. (b) Forest plot of HRs and 95% CIs of outcomes comparing continuous vs. split treatment course in all evaluated patients.

3.1. Subgroup analysis

A subgroup analysis was attempted, testing the effects of the two treatment regimens on outcomes according to prognostic and treatment variables. As expected, CR and 5-year OS rates were in general significantly better in patients with T2 than >T2 tumour stage (79% vs. 62%,p < 0.0001 and 61% vs. 47%,p = 0.0001, respectively, Fig. 3 a). In this latter group, the 5-year OS rate appeared significantly better in patients treated with the continuous rather than the split treatment course (HR = 0.641, 95%CI = 0.424–0.969) with an absolute benefit of 11% ( Fig. 3 b).

Fig. 3 (a) CR and 5-year OS rates in patients with T2 or >T2 tumour stages. (b) Forest plot of HRs and 95%CIs of CR and 5-year OS rates comparing continuous vs. split treatment course in patients with T2 or >T2 tumour stages.

CR rates resulted better also in patients without rather than with hydronephrosis (77% vs. 38%,p = 0.0001), while the difference in OS rates between these two groups resulted not significant ( Fig. 4 ).

Fig. 4 CR and 5-year OS rates in patients with or without hydronephrosis.

Fig. 5 shows CR (panel a) and OS (panel b) rates as a function of the EQD2. While there seem to be no influence of total dose on 5-year OS rates, CR rates appear to increase when increasing the EQD2.

Fig. 5 Diagram showing (a) CR and (b) 5-year OS rates as a function of EQD2 in all trials of continuous and split treatment course. In the split treatment, CR rate was plotted as a function of dose delivered by induction CRT course, while 5-year OS rate was plotted as a function of dose delivered by both induction and consolidation CRT courses also for patients undergone to SC.

Data on the effect of different drug regimes are too sparse to be correctly evaluated. The addition of other drugs to CP or the use of different drugs does not seem to give a benefit for 5-year OS and CR (data not shown). Because of the limited number of patients receiving drugs other than CP, we analyzed only those patients receiving CP and CP containing regimes. In both these groups, while no significant difference in survival could be detected between the two radiation schedules, significantly higher CR and lower SC rates were observed by continuous compared to split course regime (HR = 0.411, 95%CI = 0.305–0.554 and HR = 1.487, 95%CI = 1.161–1.903, respectively ( Fig. 6 ).

Fig. 6 Forest plot of HRs and 95%CIs of CR and 5-year OS rates comparing continuous vs. split treatment course in patients treated with CP alone or CP plus other drugs.

From the limited, currently available data, the use of NACT did not appear to produce any significant benefit in terms of rate of OS and CR ( Fig. 7 a). However, while no difference in OS was detected between the two radiation schedules in the absence of NACT, significantly better CR and OS HRs were observed in the continuous compared to the split regime when NACT was added to CRT ( Fig. 7 b).

Fig. 7 (a) CR and 5-year OS rates in patients receiving neoadjuvant chemotherapy (NADCT+) or not (NADCT−). (b) Forest plot of HRs and 95%CIs of CR, cystectomy and 5-year OS rates comparing continuous vs. split treatment course in patients receiving or not receiving NACT.

Multivariate analysis testing the significance of treatment type (continuous vs. split), NACT and CP-based treatment for the various endpoints, confirmed the continuous course as an independent prognostic variable only for CR rate (p = 0.0036), while CP containing regimes (yes/no) and NACT (yes/no) were not significant at univariate analysis. For SC, only the CP-based regimen resulted a significant prognostic factor (p = 0.0219) ( Table 3 ).

Table 3 Univariate and multivariate analyses for different end-points.

Endpoints	Variables significant to univariate analysis	Variables significant to multivariate analysis
OS	§	§
CR	Continuous course	Continuous course
BIS	§	§
SC	Continuous course CP-based treatment	CP-based treatment
LF	§	§

^§ No significant association including continuous course, neoadjuvant or CP-based treatment.

Bladder preservation strategies are perceived by many urologists to result in inferior survival in comparison to radical cystectomy (RC) and, therefore, RC is considered as the “gold standard” in the treatment of MIBC. Nevertheless, many patients may be not candidates for radical surgery due to advanced age and/or multiple co-morbidities and need a safe and effective treatment that can be an alternative to radical cystectomy. Furthermore, organ preservation and quality of life in bladder cancer, as with other sites (breast, larynx, anus, prostate, etc.), are increasingly being requested by patients, providing the complete eradication of cancer cells remains the primary goal of the treatment. Appropriate randomized trials would give definitive answers on the possibility of choosing the most beneficial treatment for patients with MIBC. In the meanwhile, a systematic review of radical cystectomy vs. organ-sparing TMT trials, reporting a temporary evaluation of the benefit of these two most important treatment modalities, has been just completed by our group and will be separately reported (paper submitted).

The present overview shows that an average 56% of patients with MIBC who underwent TMT survived at 5 years, with approximately 80% of them living with their own bladder. These seem to be excellent results in consideration that, with the exception of the RTOG 8903 [19] and BC 2001 trials [35] , they are not obtained by well conducted randomized studies, but are results emerging from prospective and retrospective studies conducted by individual institutions reporting the results of the routine therapeutic activity without any patient selection.

Bladder cancer constitutes a heterogeneous population of patients with a variety of factors that may influence the outcomes (age, gender, co-morbidities, T-stage, hydronephrosis, completeness of TURBT, etc.). The comparison of outcomes not within but only across studies raises concern about absence of prognostic balance between the two treatment schedules. Furthermore, the intention-to-treat report, specified only in few papers, may represent another confounding variable.

For the aim of this study, the trials were simply grouped into two major treatment approaches, i.e. continuous and split treatment course. Although at univariate analysis the continuous treatment showed better outcomes than split regime, multivariate analysis confirmed the continuous schedule as a significant variable only for CR rate. Of relevance, SC rate is unfortunately not reported in all the investigated studies and, owing to the presence of other confounding factors, a higher number of patients is needed to reach the significance at a multivariate analysis. The higher CR and lower SC rates observed at the univariate analysis by continous treatment, may be due to the fact that patients undergone split course regimen received, by protocol, an early assessment for SC, potentially leading to some unnecessary SC avoidable by a later outcome evaluation. On the other hand, the continuous treatment course allows a sufficient time to exclude SC in patients who either reached CR after the full CRT course or developed distant disease progression during the treatment/follow-up. In this context, although the therapeutic schedule was not shown to significantly influence the 5-year OS rates, continuous treatment seems to offer some advantage for patient quality of life in terms of bladder sparing.

The worst results were observed in patients with stage >T2 or obstructive uropathy. In these cases, however, continuous seems to be more effective than split treatment, with significantly higher CR rates in both with/without hydronephrosis subgroups, and significantly better OS rates in patients with later tumour stages. In these latter, SC does not seem to be sufficient to compensate for the lower CR rate observed after the split course.

As expected, higher CR rates were observed by increasing total RT dose. Owing to the early evaluation after the induction CRT course, when RT dose was only partially delivered, CR rates resulted to be lower in the split than continuous regimen where the evaluation was done after full RT course. Instead no effect of total dose, in general, was observed on 5-year OS rates, possibly because SC was able to compensate for the lower CR rate achieved by the induction CRT course of the split treatment.

Although other drugs in combination or not with CP have shown to be very effective[21], [23], [32], [34], [36], and [38], no conclusion could be made whether their addition to CP or their use in place of CP is better than using CP alone, since CP alone was used in most trials. However, using CP either alone or in combination with other drugs, higher CR and lower SC rates were observed with the continuous rather than the split regime.

Data on the benefit of using NACT are sparse. Although some trials showed improved outcomes with NACT in patients treated by radical cystectomy or RT alone, a recent report from Cochrane Database [42] concluded that there was insufficient information to obtain a definitive answer to the question of whether NACT improves survival in patients with MIBC. The RTOG 8903 randomized trial [18] concluded that two cycles of NACT with the methotrexate, cisplatin, and vinblastine (MCV) regime was not shown to have an impact on outcomes with respect to CRT alone in patients with MIBC, although the number of patients was too small to detect small significant differences. The results of this overview on more than 900 patients confirm the lack of benefit of the addition of NACT to CRT. However, by NACT addition, CR, OS and SC rates were significantly improved in the continuous with respect to the split treatment. Appropriate randomized trials should be addressed to test this suggestion.

In the last two decades several retrospective and prospective trials have shown that the trimodality therapy (TMT), based on the concurrent delivery of chemotherapy and radiotherapy after muscle-infiltrating bladder tumours (TURBT) can be a safe and effective treatment in the management of muscle-invasive bladder cancer (MIBC). In fact, this strategy has the advantage, with respect to radical cystectomy, of preserving, in most of the patients, a normally functional bladder, reserving cystectomy as a salvage option only in cases with an infiltrating local failure (LF) without distant metastases. The rationale for the combination of chemotherapy and radiotherapy (CRT) is based on two main premises. First, clinical and autopsy data have shown that micrometastases may occur concurrently with MIBC, and their frequency increases with tumour stage [1] . Therefore, it is expected that the addition of chemotherapy to radiotherapy may reduce the probability of distant failure by eradicating occult metastases. Second, some drugs such as fluorouracil, cisplatin, gemcitabine, paclitaxel, etc. act as radio-sensitizers, rendering cancer cells more sensitive to radiation [2] . To avoid a salvage cystectomy (SC), the immediate most important goal of TMT is that of achieving and maintaining the highest rate of CR with the lowest rate of infiltrating LF. Two main different CRT approaches have been employed following TURBT. In protocols developed at the Massachusetts General Hospital (MGH) and adopted in the Radiation Therapy Oncology Group (RTOG) trials, candidates for bladder preservation are selected according to their response to induction CRT, with a cystoscopic assessment of response after a preliminary course. In these protocols, consolidation CRT (split treatment) is given only to completely responding patients, whereas all others undergo SC with a curative intent [3] . The second bladder preservation approach, mainly used in the University of Erlangen [4] and adopted by many European centres, consists of the delivery of an upfront full-dose CRT course (continuous treatment). In this case, patients with an incomplete response and continuing pelvis confined disease at post-treatment evaluation undergo SC. These two therapeutic approaches have been planned with a different intent: to reduce the risk of both uncontrolled loco-regional disease and SC-associated complications, the former approach gives priority to cancer control by carrying out SC in patients with an incomplete response after induction CRT, with a minimal delay after low dose preoperative CRT. The aim of the continuous treatment schedule is that of increasing the chances of bladder preservation by giving a more intensive CRT course and more time for tumours to respond, reserving SC only for patients with an incomplete response and continuing pelvis confined disease.

With the aim of identifying the optimal strategy for bladder sparing in the treatment of MIBC, we performed a systematic review and meta-analysis of the published trials employing TMT.

Fig. 1 shows the flow-chart of identification and inclusion of trials, as recommended by PRISMA [5] . Overall 224 references were identified and screened. One hundred and twenty two studies were excluded because CRT was not employed in the bladder preservation approach. One hundred and two trials were selected and retrieved for full-text analysis. After discarding duplicates and papers that were updated or did not fulfil our inclusion criteria, the search yielded 31 articles with a total of 3315 evaluated patients[3], [4], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], and [41]. The material was largely heterogeneous with regard to tumour stage, prognostic variables, treatment, outcome evaluation and reporting results. The percent of patients with locally advanced or non resected tumours included in the trials ranged from 0% to more than 50%. Several trials in the continuous course schedule included medically-inoperable patients. Intention-to-treat analysis was reported by most trials although, in some retrospective studies, the results were reported only for patients completing the planned treatment. Radiotherapy total dose and fractionation varied among the studies, ranging from 24 to 48 Gy in 8–16 fractions with hypofractionation, 40 to 65 Gy in 20–32 fractions with conventional fractionation, or 66 to 69 Gy in 44–46 fractions with hyper-fractionation. Also concomitant chemotherapy varied among the studies, with the use of several dose schedules and drug combination (cisplatin ± other drugs, fluorouracil, carboplatin, gemcitabine, mitomycin C, vinblastine, metotrexate, paclitaxel, etc.), with or without neoadjuvant courses. The patient characteristics of the two treatment groups are summarized in Table 1 . Details on treatment, number of patients, and outcomes are reported in Table 2 .

Fig. 1 Flow of identification and inclusion of trials as recommended by PRISMA [5] .

Fig. 2 a shows the cumulative results from the evaluated trials. CR was observed in 78% of patients. LF, i.e. tumours not achieving CR plus later muscle-infiltrating recurrences, occurred in 28% of cases. SC was necessary in 21% of patients. At 5 years, OS and BIS rates were 56% and 42%, respectively. The CR rate appeared significantly better in patients treated with the continuous than split schedule with a HR of 0.513, (95% CI of 0.430–0.611), and an absolute benefit of more than 11% ( Fig. 2 b). No significant differences were observed in 5-year OS, BIS and infiltrating LF rates. Of relevance, SC rates were 19% (268/1416) vs. 25% (287/1144) in the continuous and split regimen, respectively (HR = 1.435, 95%CI = 1.188–1.732), with an absolute benefit of 6% ( Fig. 2 b).

Fig. 2 (a) Cumulative results from all evaluated trials; in parentheses is the number of patients evaluated for each outcome. (b) Forest plot of HRs and 95% CIs of outcomes comparing continuous vs. split treatment course in all evaluated patients.

3.1. Subgroup analysis

A subgroup analysis was attempted, testing the effects of the two treatment regimens on outcomes according to prognostic and treatment variables. As expected, CR and 5-year OS rates were in general significantly better in patients with T2 than >T2 tumour stage (79% vs. 62%,p < 0.0001 and 61% vs. 47%,p = 0.0001, respectively, Fig. 3 a). In this latter group, the 5-year OS rate appeared significantly better in patients treated with the continuous rather than the split treatment course (HR = 0.641, 95%CI = 0.424–0.969) with an absolute benefit of 11% ( Fig. 3 b).

Fig. 3 (a) CR and 5-year OS rates in patients with T2 or >T2 tumour stages. (b) Forest plot of HRs and 95%CIs of CR and 5-year OS rates comparing continuous vs. split treatment course in patients with T2 or >T2 tumour stages.

CR rates resulted better also in patients without rather than with hydronephrosis (77% vs. 38%,p = 0.0001), while the difference in OS rates between these two groups resulted not significant ( Fig. 4 ).

Fig. 4 CR and 5-year OS rates in patients with or without hydronephrosis.

Fig. 5 shows CR (panel a) and OS (panel b) rates as a function of the EQD2. While there seem to be no influence of total dose on 5-year OS rates, CR rates appear to increase when increasing the EQD2.

Fig. 5 Diagram showing (a) CR and (b) 5-year OS rates as a function of EQD2 in all trials of continuous and split treatment course. In the split treatment, CR rate was plotted as a function of dose delivered by induction CRT course, while 5-year OS rate was plotted as a function of dose delivered by both induction and consolidation CRT courses also for patients undergone to SC.

Data on the effect of different drug regimes are too sparse to be correctly evaluated. The addition of other drugs to CP or the use of different drugs does not seem to give a benefit for 5-year OS and CR (data not shown). Because of the limited number of patients receiving drugs other than CP, we analyzed only those patients receiving CP and CP containing regimes. In both these groups, while no significant difference in survival could be detected between the two radiation schedules, significantly higher CR and lower SC rates were observed by continuous compared to split course regime (HR = 0.411, 95%CI = 0.305–0.554 and HR = 1.487, 95%CI = 1.161–1.903, respectively ( Fig. 6 ).

Fig. 6 Forest plot of HRs and 95%CIs of CR and 5-year OS rates comparing continuous vs. split treatment course in patients treated with CP alone or CP plus other drugs.

From the limited, currently available data, the use of NACT did not appear to produce any significant benefit in terms of rate of OS and CR ( Fig. 7 a). However, while no difference in OS was detected between the two radiation schedules in the absence of NACT, significantly better CR and OS HRs were observed in the continuous compared to the split regime when NACT was added to CRT ( Fig. 7 b).

Fig. 7 (a) CR and 5-year OS rates in patients receiving neoadjuvant chemotherapy (NADCT+) or not (NADCT−). (b) Forest plot of HRs and 95%CIs of CR, cystectomy and 5-year OS rates comparing continuous vs. split treatment course in patients receiving or not receiving NACT.

Multivariate analysis testing the significance of treatment type (continuous vs. split), NACT and CP-based treatment for the various endpoints, confirmed the continuous course as an independent prognostic variable only for CR rate (p = 0.0036), while CP containing regimes (yes/no) and NACT (yes/no) were not significant at univariate analysis. For SC, only the CP-based regimen resulted a significant prognostic factor (p = 0.0219) ( Table 3 ).

Table 3 Univariate and multivariate analyses for different end-points.

Endpoints	Variables significant to univariate analysis	Variables significant to multivariate analysis
OS	§	§
CR	Continuous course	Continuous course
BIS	§	§
SC	Continuous course CP-based treatment	CP-based treatment
LF	§	§

^§ No significant association including continuous course, neoadjuvant or CP-based treatment.

Bladder preservation strategies are perceived by many urologists to result in inferior survival in comparison to radical cystectomy (RC) and, therefore, RC is considered as the “gold standard” in the treatment of MIBC. Nevertheless, many patients may be not candidates for radical surgery due to advanced age and/or multiple co-morbidities and need a safe and effective treatment that can be an alternative to radical cystectomy. Furthermore, organ preservation and quality of life in bladder cancer, as with other sites (breast, larynx, anus, prostate, etc.), are increasingly being requested by patients, providing the complete eradication of cancer cells remains the primary goal of the treatment. Appropriate randomized trials would give definitive answers on the possibility of choosing the most beneficial treatment for patients with MIBC. In the meanwhile, a systematic review of radical cystectomy vs. organ-sparing TMT trials, reporting a temporary evaluation of the benefit of these two most important treatment modalities, has been just completed by our group and will be separately reported (paper submitted).

The present overview shows that an average 56% of patients with MIBC who underwent TMT survived at 5 years, with approximately 80% of them living with their own bladder. These seem to be excellent results in consideration that, with the exception of the RTOG 8903 [19] and BC 2001 trials [35] , they are not obtained by well conducted randomized studies, but are results emerging from prospective and retrospective studies conducted by individual institutions reporting the results of the routine therapeutic activity without any patient selection.

Bladder cancer constitutes a heterogeneous population of patients with a variety of factors that may influence the outcomes (age, gender, co-morbidities, T-stage, hydronephrosis, completeness of TURBT, etc.). The comparison of outcomes not within but only across studies raises concern about absence of prognostic balance between the two treatment schedules. Furthermore, the intention-to-treat report, specified only in few papers, may represent another confounding variable.

For the aim of this study, the trials were simply grouped into two major treatment approaches, i.e. continuous and split treatment course. Although at univariate analysis the continuous treatment showed better outcomes than split regime, multivariate analysis confirmed the continuous schedule as a significant variable only for CR rate. Of relevance, SC rate is unfortunately not reported in all the investigated studies and, owing to the presence of other confounding factors, a higher number of patients is needed to reach the significance at a multivariate analysis. The higher CR and lower SC rates observed at the univariate analysis by continous treatment, may be due to the fact that patients undergone split course regimen received, by protocol, an early assessment for SC, potentially leading to some unnecessary SC avoidable by a later outcome evaluation. On the other hand, the continuous treatment course allows a sufficient time to exclude SC in patients who either reached CR after the full CRT course or developed distant disease progression during the treatment/follow-up. In this context, although the therapeutic schedule was not shown to significantly influence the 5-year OS rates, continuous treatment seems to offer some advantage for patient quality of life in terms of bladder sparing.

The worst results were observed in patients with stage >T2 or obstructive uropathy. In these cases, however, continuous seems to be more effective than split treatment, with significantly higher CR rates in both with/without hydronephrosis subgroups, and significantly better OS rates in patients with later tumour stages. In these latter, SC does not seem to be sufficient to compensate for the lower CR rate observed after the split course.

As expected, higher CR rates were observed by increasing total RT dose. Owing to the early evaluation after the induction CRT course, when RT dose was only partially delivered, CR rates resulted to be lower in the split than continuous regimen where the evaluation was done after full RT course. Instead no effect of total dose, in general, was observed on 5-year OS rates, possibly because SC was able to compensate for the lower CR rate achieved by the induction CRT course of the split treatment.

Although other drugs in combination or not with CP have shown to be very effective[21], [23], [32], [34], [36], and [38], no conclusion could be made whether their addition to CP or their use in place of CP is better than using CP alone, since CP alone was used in most trials. However, using CP either alone or in combination with other drugs, higher CR and lower SC rates were observed with the continuous rather than the split regime.

Data on the benefit of using NACT are sparse. Although some trials showed improved outcomes with NACT in patients treated by radical cystectomy or RT alone, a recent report from Cochrane Database [42] concluded that there was insufficient information to obtain a definitive answer to the question of whether NACT improves survival in patients with MIBC. The RTOG 8903 randomized trial [18] concluded that two cycles of NACT with the methotrexate, cisplatin, and vinblastine (MCV) regime was not shown to have an impact on outcomes with respect to CRT alone in patients with MIBC, although the number of patients was too small to detect small significant differences. The results of this overview on more than 900 patients confirm the lack of benefit of the addition of NACT to CRT. However, by NACT addition, CR, OS and SC rates were significantly improved in the continuous with respect to the split treatment. Appropriate randomized trials should be addressed to test this suggestion.

In the last two decades several retrospective and prospective trials have shown that the trimodality therapy (TMT), based on the concurrent delivery of chemotherapy and radiotherapy after muscle-infiltrating bladder tumours (TURBT) can be a safe and effective treatment in the management of muscle-invasive bladder cancer (MIBC). In fact, this strategy has the advantage, with respect to radical cystectomy, of preserving, in most of the patients, a normally functional bladder, reserving cystectomy as a salvage option only in cases with an infiltrating local failure (LF) without distant metastases. The rationale for the combination of chemotherapy and radiotherapy (CRT) is based on two main premises. First, clinical and autopsy data have shown that micrometastases may occur concurrently with MIBC, and their frequency increases with tumour stage [1] . Therefore, it is expected that the addition of chemotherapy to radiotherapy may reduce the probability of distant failure by eradicating occult metastases. Second, some drugs such as fluorouracil, cisplatin, gemcitabine, paclitaxel, etc. act as radio-sensitizers, rendering cancer cells more sensitive to radiation [2] . To avoid a salvage cystectomy (SC), the immediate most important goal of TMT is that of achieving and maintaining the highest rate of CR with the lowest rate of infiltrating LF. Two main different CRT approaches have been employed following TURBT. In protocols developed at the Massachusetts General Hospital (MGH) and adopted in the Radiation Therapy Oncology Group (RTOG) trials, candidates for bladder preservation are selected according to their response to induction CRT, with a cystoscopic assessment of response after a preliminary course. In these protocols, consolidation CRT (split treatment) is given only to completely responding patients, whereas all others undergo SC with a curative intent [3] . The second bladder preservation approach, mainly used in the University of Erlangen [4] and adopted by many European centres, consists of the delivery of an upfront full-dose CRT course (continuous treatment). In this case, patients with an incomplete response and continuing pelvis confined disease at post-treatment evaluation undergo SC. These two therapeutic approaches have been planned with a different intent: to reduce the risk of both uncontrolled loco-regional disease and SC-associated complications, the former approach gives priority to cancer control by carrying out SC in patients with an incomplete response after induction CRT, with a minimal delay after low dose preoperative CRT. The aim of the continuous treatment schedule is that of increasing the chances of bladder preservation by giving a more intensive CRT course and more time for tumours to respond, reserving SC only for patients with an incomplete response and continuing pelvis confined disease.

With the aim of identifying the optimal strategy for bladder sparing in the treatment of MIBC, we performed a systematic review and meta-analysis of the published trials employing TMT.

Fig. 1 shows the flow-chart of identification and inclusion of trials, as recommended by PRISMA [5] . Overall 224 references were identified and screened. One hundred and twenty two studies were excluded because CRT was not employed in the bladder preservation approach. One hundred and two trials were selected and retrieved for full-text analysis. After discarding duplicates and papers that were updated or did not fulfil our inclusion criteria, the search yielded 31 articles with a total of 3315 evaluated patients[3], [4], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], and [41]. The material was largely heterogeneous with regard to tumour stage, prognostic variables, treatment, outcome evaluation and reporting results. The percent of patients with locally advanced or non resected tumours included in the trials ranged from 0% to more than 50%. Several trials in the continuous course schedule included medically-inoperable patients. Intention-to-treat analysis was reported by most trials although, in some retrospective studies, the results were reported only for patients completing the planned treatment. Radiotherapy total dose and fractionation varied among the studies, ranging from 24 to 48 Gy in 8–16 fractions with hypofractionation, 40 to 65 Gy in 20–32 fractions with conventional fractionation, or 66 to 69 Gy in 44–46 fractions with hyper-fractionation. Also concomitant chemotherapy varied among the studies, with the use of several dose schedules and drug combination (cisplatin ± other drugs, fluorouracil, carboplatin, gemcitabine, mitomycin C, vinblastine, metotrexate, paclitaxel, etc.), with or without neoadjuvant courses. The patient characteristics of the two treatment groups are summarized in Table 1 . Details on treatment, number of patients, and outcomes are reported in Table 2 .

Fig. 1 Flow of identification and inclusion of trials as recommended by PRISMA [5] .

Fig. 2 a shows the cumulative results from the evaluated trials. CR was observed in 78% of patients. LF, i.e. tumours not achieving CR plus later muscle-infiltrating recurrences, occurred in 28% of cases. SC was necessary in 21% of patients. At 5 years, OS and BIS rates were 56% and 42%, respectively. The CR rate appeared significantly better in patients treated with the continuous than split schedule with a HR of 0.513, (95% CI of 0.430–0.611), and an absolute benefit of more than 11% ( Fig. 2 b). No significant differences were observed in 5-year OS, BIS and infiltrating LF rates. Of relevance, SC rates were 19% (268/1416) vs. 25% (287/1144) in the continuous and split regimen, respectively (HR = 1.435, 95%CI = 1.188–1.732), with an absolute benefit of 6% ( Fig. 2 b).

Fig. 2 (a) Cumulative results from all evaluated trials; in parentheses is the number of patients evaluated for each outcome. (b) Forest plot of HRs and 95% CIs of outcomes comparing continuous vs. split treatment course in all evaluated patients.

3.1. Subgroup analysis

A subgroup analysis was attempted, testing the effects of the two treatment regimens on outcomes according to prognostic and treatment variables. As expected, CR and 5-year OS rates were in general significantly better in patients with T2 than >T2 tumour stage (79% vs. 62%,p < 0.0001 and 61% vs. 47%,p = 0.0001, respectively, Fig. 3 a). In this latter group, the 5-year OS rate appeared significantly better in patients treated with the continuous rather than the split treatment course (HR = 0.641, 95%CI = 0.424–0.969) with an absolute benefit of 11% ( Fig. 3 b).

Fig. 3 (a) CR and 5-year OS rates in patients with T2 or >T2 tumour stages. (b) Forest plot of HRs and 95%CIs of CR and 5-year OS rates comparing continuous vs. split treatment course in patients with T2 or >T2 tumour stages.

CR rates resulted better also in patients without rather than with hydronephrosis (77% vs. 38%,p = 0.0001), while the difference in OS rates between these two groups resulted not significant ( Fig. 4 ).

Fig. 4 CR and 5-year OS rates in patients with or without hydronephrosis.

Fig. 5 shows CR (panel a) and OS (panel b) rates as a function of the EQD2. While there seem to be no influence of total dose on 5-year OS rates, CR rates appear to increase when increasing the EQD2.

Fig. 5 Diagram showing (a) CR and (b) 5-year OS rates as a function of EQD2 in all trials of continuous and split treatment course. In the split treatment, CR rate was plotted as a function of dose delivered by induction CRT course, while 5-year OS rate was plotted as a function of dose delivered by both induction and consolidation CRT courses also for patients undergone to SC.

Data on the effect of different drug regimes are too sparse to be correctly evaluated. The addition of other drugs to CP or the use of different drugs does not seem to give a benefit for 5-year OS and CR (data not shown). Because of the limited number of patients receiving drugs other than CP, we analyzed only those patients receiving CP and CP containing regimes. In both these groups, while no significant difference in survival could be detected between the two radiation schedules, significantly higher CR and lower SC rates were observed by continuous compared to split course regime (HR = 0.411, 95%CI = 0.305–0.554 and HR = 1.487, 95%CI = 1.161–1.903, respectively ( Fig. 6 ).

Fig. 6 Forest plot of HRs and 95%CIs of CR and 5-year OS rates comparing continuous vs. split treatment course in patients treated with CP alone or CP plus other drugs.

From the limited, currently available data, the use of NACT did not appear to produce any significant benefit in terms of rate of OS and CR ( Fig. 7 a). However, while no difference in OS was detected between the two radiation schedules in the absence of NACT, significantly better CR and OS HRs were observed in the continuous compared to the split regime when NACT was added to CRT ( Fig. 7 b).

Fig. 7 (a) CR and 5-year OS rates in patients receiving neoadjuvant chemotherapy (NADCT+) or not (NADCT−). (b) Forest plot of HRs and 95%CIs of CR, cystectomy and 5-year OS rates comparing continuous vs. split treatment course in patients receiving or not receiving NACT.

Multivariate analysis testing the significance of treatment type (continuous vs. split), NACT and CP-based treatment for the various endpoints, confirmed the continuous course as an independent prognostic variable only for CR rate (p = 0.0036), while CP containing regimes (yes/no) and NACT (yes/no) were not significant at univariate analysis. For SC, only the CP-based regimen resulted a significant prognostic factor (p = 0.0219) ( Table 3 ).

Table 3 Univariate and multivariate analyses for different end-points.

Endpoints	Variables significant to univariate analysis	Variables significant to multivariate analysis
OS	§	§
CR	Continuous course	Continuous course
BIS	§	§
SC	Continuous course CP-based treatment	CP-based treatment
LF	§	§

^§ No significant association including continuous course, neoadjuvant or CP-based treatment.

Bladder preservation strategies are perceived by many urologists to result in inferior survival in comparison to radical cystectomy (RC) and, therefore, RC is considered as the “gold standard” in the treatment of MIBC. Nevertheless, many patients may be not candidates for radical surgery due to advanced age and/or multiple co-morbidities and need a safe and effective treatment that can be an alternative to radical cystectomy. Furthermore, organ preservation and quality of life in bladder cancer, as with other sites (breast, larynx, anus, prostate, etc.), are increasingly being requested by patients, providing the complete eradication of cancer cells remains the primary goal of the treatment. Appropriate randomized trials would give definitive answers on the possibility of choosing the most beneficial treatment for patients with MIBC. In the meanwhile, a systematic review of radical cystectomy vs. organ-sparing TMT trials, reporting a temporary evaluation of the benefit of these two most important treatment modalities, has been just completed by our group and will be separately reported (paper submitted).

The present overview shows that an average 56% of patients with MIBC who underwent TMT survived at 5 years, with approximately 80% of them living with their own bladder. These seem to be excellent results in consideration that, with the exception of the RTOG 8903 [19] and BC 2001 trials [35] , they are not obtained by well conducted randomized studies, but are results emerging from prospective and retrospective studies conducted by individual institutions reporting the results of the routine therapeutic activity without any patient selection.

Bladder cancer constitutes a heterogeneous population of patients with a variety of factors that may influence the outcomes (age, gender, co-morbidities, T-stage, hydronephrosis, completeness of TURBT, etc.). The comparison of outcomes not within but only across studies raises concern about absence of prognostic balance between the two treatment schedules. Furthermore, the intention-to-treat report, specified only in few papers, may represent another confounding variable.

For the aim of this study, the trials were simply grouped into two major treatment approaches, i.e. continuous and split treatment course. Although at univariate analysis the continuous treatment showed better outcomes than split regime, multivariate analysis confirmed the continuous schedule as a significant variable only for CR rate. Of relevance, SC rate is unfortunately not reported in all the investigated studies and, owing to the presence of other confounding factors, a higher number of patients is needed to reach the significance at a multivariate analysis. The higher CR and lower SC rates observed at the univariate analysis by continous treatment, may be due to the fact that patients undergone split course regimen received, by protocol, an early assessment for SC, potentially leading to some unnecessary SC avoidable by a later outcome evaluation. On the other hand, the continuous treatment course allows a sufficient time to exclude SC in patients who either reached CR after the full CRT course or developed distant disease progression during the treatment/follow-up. In this context, although the therapeutic schedule was not shown to significantly influence the 5-year OS rates, continuous treatment seems to offer some advantage for patient quality of life in terms of bladder sparing.

The worst results were observed in patients with stage >T2 or obstructive uropathy. In these cases, however, continuous seems to be more effective than split treatment, with significantly higher CR rates in both with/without hydronephrosis subgroups, and significantly better OS rates in patients with later tumour stages. In these latter, SC does not seem to be sufficient to compensate for the lower CR rate observed after the split course.

As expected, higher CR rates were observed by increasing total RT dose. Owing to the early evaluation after the induction CRT course, when RT dose was only partially delivered, CR rates resulted to be lower in the split than continuous regimen where the evaluation was done after full RT course. Instead no effect of total dose, in general, was observed on 5-year OS rates, possibly because SC was able to compensate for the lower CR rate achieved by the induction CRT course of the split treatment.

Although other drugs in combination or not with CP have shown to be very effective[21], [23], [32], [34], [36], and [38], no conclusion could be made whether their addition to CP or their use in place of CP is better than using CP alone, since CP alone was used in most trials. However, using CP either alone or in combination with other drugs, higher CR and lower SC rates were observed with the continuous rather than the split regime.

Data on the benefit of using NACT are sparse. Although some trials showed improved outcomes with NACT in patients treated by radical cystectomy or RT alone, a recent report from Cochrane Database [42] concluded that there was insufficient information to obtain a definitive answer to the question of whether NACT improves survival in patients with MIBC. The RTOG 8903 randomized trial [18] concluded that two cycles of NACT with the methotrexate, cisplatin, and vinblastine (MCV) regime was not shown to have an impact on outcomes with respect to CRT alone in patients with MIBC, although the number of patients was too small to detect small significant differences. The results of this overview on more than 900 patients confirm the lack of benefit of the addition of NACT to CRT. However, by NACT addition, CR, OS and SC rates were significantly improved in the continuous with respect to the split treatment. Appropriate randomized trials should be addressed to test this suggestion.