Introduction

Overactive bladder (OAB), defined by the International Continence Society (ICS) as a condition characterized by urinary urgency, usually affects frequency and nocturia with or without urinary incontinence [1]. Overactive bladder (OAB) is a common health disorder of multifactorial origin. It affects quality of life and imposes an economic burden [2, 3], and its prevalence is approximately 12–19% in both men and women [4,5,6].

The primary treatment for OAB is behavior therapy according to ICS guidelines. Although antimuscarinic agents are currently a treatment option for OAB [7, 8], their use in certain patients can be discontinued due to inadequate efficacy or side effects [9, 10]. In such situations, onabotulinumtoxinA (BoNT/A) has been suggested as an alternative treatment. There is evidence of the efficacy and tolerability of BoNT/A injections in patients with OAB [11, 12]. This drug acts to prevent the release of acetylcholine, adenosine triphosphate (ATP), and substance P. Moreover, it down-regulates capsaicin receptors and purinergic receptors on afferent neurons. These mechanisms are related to the pathophysiology of OAB [12, 13].

However, there is no standard method of injecting BoNT/A for treatment of OAB. Accordingly, we performed a systematic review and meta-analysis to evaluate the efficacy and safety of BoNT/A in patients with OAB according to injection site.

Materials and methods

We used a systematic approach to locate publications comparing the efficacy and safety of BoNT/A in patients with OAB according to injection site. The study is based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) and on Cochrane Review Methods [14].

Data and literature sources

We searched MEDLINE In-Process and Other Non-Indexed Citations, Ovid MEDLINE (R) Daily, and Ovid MEDLINE (R) 1946 to the present (OVID platform), EMBASE (from 1974) (OVID platform), the Cochrane Controlled Trials Register (OVID platform), and the Cochrane Database of Systematic Reviews (OVID platform) from inauguration to February 22, 2017. A literature search of the Web of Science and Google Scholar was additionally performed to search all relevant studies. We manually searched the reference lists of the retrieved studies, ClinicalTrials.gov, and the World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) for additional unpublished/published studies. The main keywords were idiopathic overactive bladder, urinary urgency, urinary urgency incontinence, neurogenic bladder, onabotulinumtoxinA, and randomized controlled trial.

Study selection

All selected studies were independently identified by two reviewers (KNK and JYK) based on predefined selection criteria, and disagreements on primary study selection were arbitrated by a third reviewer (JKJ). Studies were included in our meta-analysis if they fulfilled the following criteria: (1) randomized controlled trial in any published international journal without language restriction, (2) adult patients undergoing treatment for idiopathic detrusor overactivity or neurogenic detrusor overactivity with BoNT/A, (3) studies comparing the effects of BoNT/A according to injection site, and (4) primary outcomes of bladder symptoms, such as incontinence-specific quality of life (I-QOL), and overactive bladder symptom score (OABSS). Secondary outcomes were complete dryness rate; change in the number of incontinence episodes; and urodynamic variables of detrusor pressure at maximum flow rate, volume at first desire to void, post-void residual volume, maximum cystometric bladder capacity, and adverse events. The outcome variables were mean differences or incidences of events between the groups at designated times.

Data extraction

After the two reviewers (JKJ and DWK) independently extracted data using a pre-specified data extraction form, the third reviewer (KNK) confirmed the extracted data. The following variables were extracted: (1) number of patients and patient characteristics, (2) means and standard deviations or incidences of events regarding outcome data, (3) protocol for administration and dosage, (4) follow-up time for outcome data, and (5) adverse events in patients. If the above variables were not mentioned in a study, then the data was requested via email.

Assessment of methodological quality

The reviewers (JKJ and DWK) independently estimated the risks of bias in the studies using the Cochrane risk-of-bias tool. This tool evaluates randomized controlled studies by assessing the methods for generating random sequences, concealing allocations, blinding participants, and assessing outcomes, as well as evaluates any incompleteness in outcome data, selective outcome reporting, and other possible sources of risk of bias.

Quality of evidence

We used GRADE assessments to determine the quality of evidence [14]. Two reviewers (JKJ and KNK) assessed the quality of each outcome independently. The five categories, based on GRADE quality assessment, were limitations of design, inconsistency, indirectness, imprecision, and publication bias. “Summary of findings” tables were presented by a GRADE profiler (GRADEpro) and included the following outcomes: (1) changes in patient symptom score, (2) complete dryness rate, (3) change in number of incontinence episodes among patients, (4) detrusor pressure at maximum flow rate, (5) volume at first desire to void, (6) incidence of hematuria, and (7) incidence of large post-void residual.

Statistical analysis

Continuous data were reported as mean differences and 95% confidence intervals (CIs) and were analyzed using weighted mean differences (WMDs) and the generic inverse variance method. For the analysis of bladder symptom scores, we used the standardized mean difference in the reported severity of symptoms between injection methods. Binary outcomes were analyzed to compare the odds ratio with a 95% CI. Heterogeneity between studies was evaluated by the χ 2 test and I 2 statistics [15]. An I 2 statistics > 50% and χ 2 tests with P values < 0.10 were regarded as statistically significant. When significant clinical or statistical heterogeneity was found, random-effects models were applied.

A subgroup analysis was conducted according to the dose of BoNT/A administered in patients. We performed all statistical analyses with RevMan version 5.3. If the number of included studies was less than 10, we did not evaluate publication bias because of the low statistical power.

Results

Identification of studies

Initial searches of the databases identified 863 publications. In addition to removing 578 duplicate articles, 285 publications were eliminated as their titles and abstracts showed that they did not fulfill the selection criteria. For the remaining 14 publications, we obtained full manuscripts for scrutiny, subsequently identifying eight publications with potentially relevant studies. The other six publications were excluded, because they used a different study design (one publication), the study design was not randomized (one publication), the patients were not adults (one publication), or the articles reported the same data (three publications). Thus, eight studies and 419 participants were included in this meta-analysis (Fig. 1) [16,17,18,19,20,21,22,23].

Fig. 1
figure 1

Flow chart of the literature search strategy

Full size image

Study characteristics and patient populations

The included articles were published in five countries: Czech Republic (two), Ireland (one), Saudi Arabia (one), Taiwan (two), and China (two) between 2007 and 2016. Of these, five studies compared the effects of trigone-sparing and trigone-including intradetrusor injection of BoNT/A [16,17,18,19,20], and three compared the effects of inradetrusor and suburothelial BoNT/A injections [21,22,23]. The characteristics of the studies are summarized in Table 1.

Table 1 Characteristics of included randomized controlled trials
Full size table

Quality of included studies

All eight studies used a random allocation method, and four studies described blinding methods in detail [16, 18,19,20]. The risk-of-allocation concealment was high in four studies [17, 21,22,23]. The risks of allocation concealment and blinding were unclear in the other studies, and the risks of selective reporting and incomplete outcome data were low. Risk-of-bias graphs and summaries are presented in Fig. 2a, b.

Fig. 2
figure 2

a Risk-of-bias graph for all included randomized controlled trials. b Risk-of-bias summary for all trials

Full size image

Trigone-sparing versus trigone-including intradetrusor injection

Scores that assess bladder symptoms were incontinence-specific quality of life (I-QOL) [16, 17], overactive bladder symptom score (OABSS) [19], QOL using the International Prostate Symptom Score-QOL subset [20], and urgency severity score (USS) [18]. Our meta-analysis revealed significant improvements in patient symptom scores for the trigone-including intradetrusor injection of BoNT/A (SMD = − 0.53, 95% CI − 1.04 to − 0.02, P = 0.04, I 2 = 78%) (Fig. 3a). A subgroup analysis according to dose of BoNT/A revealed that trigone-including intradetrusor injection significantly improved symptom scores when 200-300 units of BoNT/A were used (SMD = − 0.44, 95% CI − 0.73 to − 0.15, P = 0.003, I 2 = 8%).

Fig. 3
figure 3

Effects of trigone-including and trigone-sparing intradetrusor injection. a Impact on patient symptom score; b impact on complete dryness rate (patient number); c impact on change in number of incontinence episodes (number per day)

Full size image

There was a significantly higher complete dryness rate in patients (OR = 2.19 patients, 95% CI 1.32–3.63, P = 0.002, I 2 = 41%) (Fig. 3b) with trigone-including intradetrusor injection. A subgroup analysis revealed that trigone-including intradetrusor injection had a significantly higher complete dryness rate when 200–300 units of BoNT/A were used (OR = 3.35 patients, 95% CI 1.76–6.37, P = 0.0002, I 2 = 0%).

There was a significantly lower frequency of incontinence episodes in patients (WMD = − 0.85 per day, 95% CI − 1.55 to − 0.16, P = 0.02, I 2 = 87%) with trigone-including intradetrusor injection (Fig. 3c). A subgroup analysis also revealed that trigone-including intradetrusor injection demonstrated a significantly lower frequency of incontinence episodes when 200–300 units of BoNT/A were used (WMD = − 0.88 per day, 95% CI − 1.59 to − 0.18, P = 0.01, I 2 = 91%).

Urodynamic variables were extracted from four randomized trials [16, 18,19,20]. We found a lower detrusor pressure at maximum flow rate following trigone-including intradetrusor injection than the measure following trigone-sparing injection (WMD = − 2.55 cm H2O, 95% CI − 4.16 to − 0.95, P = 0.002, I 2 = 0%) (Fig. 4a). In addition, following trigone-including intradetrusor injection, we found a higher volume at the first desire to void (WMD = 17.54 ml, 95% CI 1.00–34.07, P = 0.04, I 2 = 0%) (Fig. 4b). The two methods of trigone-including and trigone-sparing injection did not differ in maximum cystometric capacity (WMD = − 19.54 ml, 95% CI − 44.87–5.80, P = 0.13, I 2 = 0%) (Fig. 4c) or post-void residual volume (WMD = 20.14 ml, 95% CI − 2.25–45.52, P = 0.08, I 2 = 0%) (Fig. 4d).

Fig. 4
figure 4

Effects of trigone-including and trigone-sparing intradetrusor injection. a Impact on detrusor pressure at maximum flow rate (mmHg); b impact on volume at the first desire to void (ml); c impact on maximum cystometric capacity (ml); d impact on post-void residual volume (ml)

Full size image

In five studies [16,17,18,19,20] representing 334 patients, the incidence of vesicoureteral reflux was not reported in either group of patients receiving trigone-including injections or trigone-sparing injections. The rates of hematuria were 11.3 and 9.4% after the trigone-including injections and trigone-sparing injections, respectively, and there was no statistical difference between the groups (OR = 1.21, 95% CI 0.59–2.50, P = 0.60, I 2 = 0%) (Fig. 5a). There were also no differences between the patient groups in bladder discomfort (OR = 1.15, 95% CI 0.36–3.68, P = 0.82, I 2 = 13%) (Fig. 5b), incidence of large post-void residual volume (> 150 ml) (OR = 0.98, 95% CI 0.46–2.09, P = 0.96, I 2 = 0%) (Fig. 5c), general weakness (OR = 1.09, 95% CI 0.19–6.27, P = 0.92) (Fig. 5d), or urinary tract infection (OR = 0.85, 95% CI 0.34–2.13, P = 0.72, I 2 = 0%) (Fig. 5e).

Fig. 5
figure 5

Complications of trigone-including and trigone-sparing intradetrusor injection. a Hematuria; b general weakness; c bladder discomfort; d incidence of large post-void residual; e urinary tract infection

Full size image

Intradetrusor versus suburothelial injection

Our meta-analysis demonstrated that detrusor pressure did not differ between intradetrusor and suburothelial injection (WMD = − 5.21 cmH2O, 95% CI − 12.65–2.23, P = 0.17, I 2 = 3%) (Fig. 6a). In addition, there were no differences in maximum cystometric capacity (WMD = − 35.24 ml, 95% CI − 73.90–3.42, P = 0.07, I 2 = 0%) (Fig. 6b), detrusor compliance (WMD = 4.22 ml/cmH2O, 95% CI − 20.14–28.58, P = 0.73, I 2 = 74%) (Fig. 6c), or reduction in incontinence episodes (WMD = − 1.32 per day, 95% CI − 5.69–3.06, P = 0.56, I 2 = 78%) (Fig. 6d).

Fig. 6
figure 6

Effects of intradetrusor and suburothelial injection. a Impact on detrusor pressure at the maximum flow rate (mmHg); b impact on maximum cystometric capacity (ml); c impact on detrusor compliance (ml/cmH2O); d impact on change in number of incontinence episodes (number per day)

Full size image

Adverse events were reported in all of the included studies [21,22,23]. Of these adverse events, loss of muscle strength was extracted from two studies [21, 23], and there was no difference in loss of muscle strength between groups of patients with intradetrusor and suburothelial injection (OR = 0.26, 95% CI 0.03–2.67, P = 0.26, I 2 = 0%). One study reported the adverse events of dysuria, acute urinary retention, urinary tract infection, and gross hematuria [22], and the events did not differ between patient groups.

Quality of evidence

The quality of each outcome was evaluated by GRADE approach and “summary of findings” data is shown in Table 2. As a result, the quality of evidence in this meta-analysis ranged from very low to moderate. Most studies had problems with risk of bias, inconsistency, and imprecision. The number of included studies was fewer than 10, demonstrating low statistical power, so publication bias was not assessed.

Table 2 GRADE summary of findings. Risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effects of intervention (and its 95% CI)
Full size table

Discussion

To our knowledge, this is the first study of botulinum injection for OAB-based meta-analysis to assess treatment effects according to injection site and depth of injection. We observed significant differential effects in trigone-including injection versus trigone-sparing injection and no different effects according to injection depth. Trigone-including injection provides greater improvements in patient symptom scores, higher complete dryness rates, and lower frequency of incontinence episodes. Trigone-including injection also provides lower detrusor pressure at maximum flow rate and higher volume at first desire to void. In terms of safety, trigone-including injection showed no increase of adverse effects.

The work of Purves et al. presents a three-dimensional nerve map of the human bladder including the trigone [24]. The authors found traversing myelinated neural pathways in the trigone of the bladder, showing striking neural density. As a key component in vesical deactivation, the trigone plays a crucial role in detrusor overactivity. Other previous study also reported that trigone has rich sensory neural fibers [25]. In addition, trigonal muscles are sensitive to small amount of pressure changes; it has a crucial role of initiation of involuntary contraction of bladder [26]. Therefore, trigonal injection for denervation can reduce involuntary detrusor contractions. These findings suggested the key role of trigonal denervation using trigonal injection with considering myogenic and neurogenic effect. Our meta-analysis demonstrated the importance of trigonal deactivation. Trigone-including intradetrusor injection was found to yield lower detrusor pressure than trigone-sparing injection, and it led to fewer episodes of incontinence. In our meta-analysis, superior quality of life (QOL) was observed in patients with trigone-including injection than in patients with trigone-sparing injection. There was significantly higher complete dryness rate and lower frequency of incontinence episodes in patients with trigone-including intradetrusor injection, with these factors leading to increased QOL due to trigonal injection. Moreover, the two methods did not differ in post-void residual volume. Consequently, trigone-including intradetrusor injection is more effective than trigone-sparing intradetrusor injection for treatment of OAB.

A subgroup analysis according to dose of BoNT/A revealed that trigone-including intradetrusor injection demonstrated significantly improvement in symptom scores, higher complete dryness rate, and lower frequency of incontinence episodes when 200–300 units of BoNT/A were used. In spite of dose-dependent response in efficacy, a dose-dependent increase in adverse effects precluded using higher dose of BoNT/A, and 100 units of BoNT/A is recommended for the treatment of OAB [27, 28]. Therefore, more extensive, well-controlled, randomized studies using different dose according to site of injection are needed to assess the incidence of adverse effects.

Adverse reactions can be observed according to injection or injected materials. Complications due to injection such as hematuria and UTI were observed with similar rates between trigone-including intradetrusor injection and trigone-sparing injection. There was no difference in injected material-related reactions between the two groups. Functional complications such as bladder discomfort or incidence of large post-void residual were not different between the two groups.

Because of variability in bladder wall thickness due to systematic administration according to the vessels of the musculature, depth of injection is a considerable variable in the treatment of OAB. Although BoNT/A was injected under direct visual cystoscopic guidance, depth of injection can only be determined by the surgeon. In this situation, Mehnert et al. assessed the distribution of BoNT/A in the detrusor of the bladder [29]. Using magnetic resonance imaging (MRI), the authors found that about 20% of administrated BoNT/A injections were localized to outside the detrusor. Because submucosal BoNT/A injection does not require deep injection of the needle, it has some advantages, including visible control to target distribution and reduction of the risk of unintended administration to blood vessels in the detrusor. Our findings demonstrate the effects and side effects of submucosal BoNT/A injection. Comparison of submucosal versus detrusor injection indicates that, among the samples herein, the depth of injection had no significant impact on efficacy and safety. Because submucosal injection was equally effective and safe and was more easily administered; submucosal injection stands to be the recommended method of BoNT/A injection for the treatment of OAB.

Our meta-analysis has some limitations. First, a relatively small number of patients were enrolled in this analysis. Intervention effects can be significantly overstated in small trials where there is incomplete double blinding, allocation sequence generation, and allocation concealment [30]. Second, there was a significant heterogeneity among studies. Clinical heterogeneity was identified in terms of BoNT/A dosage, causes of OAB, and age ranges of patients, and various bladder symptom scores and measurements precluded further synthesis of the data. Finally, we did not separately analyze idiopathic overactive bladder (iOAB) or neurogenic detrusor overactivity (NDO). We also recognized the possible confounding effect due to cause of overactivity. These specific types of OAB differ according to cause and pathologic course. However, OAB is considered as a symptom complex including neurogenic components and myogenic components. Although the causes of NDO and iOAB were different, they are overlapped symptoms and treatments. In other words, they have similar patterns of symptoms in part, and both involve overactivity of the detrusor in bladder; thus, BoNT/A is also used for the control of detrusor overactivity.

Despite the limitations, this study is the first meta-analysis to clarify the effects of site of injection and depth of injection of BoNT/A in patients with OAB. Our results demonstrate the importance of controlling the trigonal area and indicate that depth of injection is not a matter of concern.

Conclusion

Our meta-analysis indicates that the use of different BoNT/A injection sites to treat OAB leads to different outcomes. Trigone-including injection has greater efficacy in terms of improvement in patient symptom score, higher complete dryness rate, and lower frequency of incontinence episodes. Trigone-including injection also provides patients with lower detrusor pressure and higher volume at the first desire to void following treatment, without an increase in adverse effects. In contrast, our findings show no difference in efficacy or safety according to depth of injection.