The efficacy and safety of botulinum toxin type A in treatment of trigeminal neuralgia and peripheral neuropathic pain: A meta‐analysis of randomized controlled trials

Abstract Background Although recent studies have shown that botulinum toxin‐A (BTX‐A) has a good analgesic effect on trigeminal neuralgia (TN) and peripheral neuropathic pain (PNP), the quality of evidence is low due to limited data. This meta‐analysis is used to synthesize existing evidence for the treatment of these conditions with BTX‐A. Methods Relevant trials were accessed by using an electronic search in databases (Web of Science, PubMed, EMBASE, Cochrane Library, and http://ClinicalTrials.gov). Data from included randomized controlled trials (RCTs) on the efficacy and safety of BTX‐A in treating TN and PNP were extracted for meta‐analysis. Results Finally, 10 RCTs (n = 391) were included in this meta‐analysis. The pooled effect of BTX‐A was superior to placebo based on pain intensity (SMD = −0.48, 95% CI [−0.74, 0.23] at 1 month, SMD = −0.58, 95% CI [−0.91, −0.24] at 2 months, and SMD = −0.55, 95% CI [−0.87, −0.22] at 3 months). Number needed to treat (NNT) for 50% pain intensity reduction showed better effect of BTX‐A on TN and postherpetic neuralgia (PN). Adverse events associated with BTX‐A were similar to placebo (OR = 1.58, 95% CI [0.51, 4.87], p = .424). Conclusion Pooled data from our meta‐analysis suggest that BTX‐A is efficacious and safe in treating TN and PNP. However, due to the limited sample size and heterogeneity, further larger and well‐designed RCTs are imperative to validate these findings.


| INTRODUC TI ON
Neuropathic pain is defined as pain caused by a lesion or disease of the somatosensory system and affects 7%-10% of the general population (Colloca et al., 2017;Finnerup et al., 2016). Peripheral neuropathic pain (PNP) is the most common type of neuropathic pain presenting in the clinical conditions. As the population ages, its incidence will continue to increase (Colloca et al., 2017). Neuropathic pain, as a largely unmet medical need (Finnerup et al., 2015), is one of the most difficult pain syndromes to manage, and outcomes often are unsatisfactory (van Hecke, Austin, Khan, Smith, & Torrance, 2014). It seriously affects the quality of life of patients owing to increased drug prescriptions and visits to healthcare providers, as well as the morbidity from the pain itself and the inciting disease (Colloca et al., 2017). It imposes a huge economic burden on the individual and society (Attal, Lanteri-Minet, Laurent, Fermanian, & Bouhassira, 2011;Doth, Hansson, Jensen, & Taylor, 2010;Langley, Van Litsenburg, Cappelleri, & Carroll, 2013).
Medication and neurosurgery are two major therapies for neuropathic pain. These currently recommended first-line medical treatments may only partially relieve pain in 30%-40% of patients (Hansson, Attal, Baron, & Cruccu, 2009) and are accompanied by systemic adverse events (Freynhagen et al., 2015). Surgical interventions carry a risk of serious and intractable complications and even worsen the initial conditions.
However, it is lack of power to elucidate the efficacy and safety implications of BTX-A for TN and PNP owing to small sample size.
Meta-analysis method can pool the data from small studies to provide evidence with better power. Hence, we performed this metaanalysis to synthesize all this evidence from the published RCTs on the efficacy and safety of BTX-A in treating TN and PNP.

| Study selection
We imported search results from the aforementioned electronic databases into EndNote X6 (Thompson Reuter) for selection. Two authors independently screened the references according to the predetermined inclusion and exclusion criteria. The inclusion criteria were as follows: (a) RCTs evaluating efficacy and safety of BTX-A on TN and PNP; (b) no limitation to age of patients and study area. The exclusion criteria were as follows: (a) unreliably extracted data; (b) datasets from articles unable to extract; and (c) case reports, case series, reviews, Cochrane bias assessment for individual trial commentaries, errata, notes, letters, and abstract articles only. All conflicts in screening step were discussed between two authors to reach an agreement, and supervisor adjudicated unresolved conflicts if necessary. The full texts of all included trials were downloaded for screening to identify available data for extraction.

| Data extraction and quality assessment
The standardized template was developed for extraction. The relevant data were extracted into the template independently by two authors.
Extracted data included: journal, author, publication year, sample size, population, patients' gender and age, injection route, duration of treatment, study design, follow-up time, and main outcomes measures.
Cochrane risk-of-bias tool was used to assess the methodological quality of the included trials based on sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other sources of bias (Higgins & Green, 2008).

| Data synthesis
Statistical analysis was carried out with R3.4.3 (http://www.r-proje ct.org/). Continuous variables were pooled as standard mean difference (SMD) using the generic inverse variance method, while dichotomous data were aggregated as odds ratio (OR) by Mantel-Haenszel (M-H) method. All analyses were conducted using random-effect and fixed-effect model. The sensitivity analysis was used to assess the effect of the hypothetical model on the overall effect size, by comparing the analysis between the random-effect model and fixedeffect model. Heterogeneity analysis was performed with I-square and chi-square tests across the included trials. p-value < .05 was identified as a statistically significant level.

| Search results and characteristics of eligible RCTs
The detailed characteristics of eligible RCTs are summarized in Table 1 Zhang et al. (2014) to 300 U in study of Attal et al. (2016). Routes for BTX-A injection include subcutaneous, submucosal, or intradermal.

| Risk of bias
The quality of included trials is illustrated in Figure 2. Authors' evaluations on the risk of each biased item presented as percentage are shown in Figure 3. Assessment of publication bias was conducted based on funnel plot. There was no obvious publication bias on the grounds of almost symmetric funnel plot for the effects of BTX-A on pain scale ( Figure 4). However, the risk of bias across all included trials was medium due to unclear quality and potential publication bias.

| D ISCUSS I ON
In the present meta-analysis, our pooled data showed that BTX-A provides significant benefit in the treatment of patients with TN and PNP compared with placebo, improving the proportion of respondents and pain scores at follow-up. The reported adverse reactions were mild, transient, and nonsystemic. No serious adverse reactions were presented in the treatment than placebo arms in these studies.
All these evidence suggested that BTX-A is efficacious and safe in treating TN and PNP compared with placebo. However, the evidence was moderate owing to limited studies and small sample size.

As early as 2002, Micheli et al. reported that a 70-year-old man
with hemifacial spasm associated with trigeminal neuralgia secondary to an ectatic basilar artery was treated with BTX-A, and relief was gained not only from twitching but also from pain (Micheli, Scorticati, & Raina, 2002 BTX-A, VAS score, surface area of pain, and therapeutic coefficient F I G U R E 7 Forest plots of standard mean difference in pain score for BTX-A versus placebo at 1 month (a), 2 months (b), and 3 months (c), excluding the single-blind study (Shehata et al., 2013) F I G U R E 8 Forest plot of odds ratio (OR) in adverse events comparing BTX-A with placebo were reduced in all patients and for all branch trigeminal nerves studied (Piovesan et al., 2005). These findings indicated that BTX-A has a beneficial effect in the treatment of TN.
Recently, three reviews have demonstrated that BTX-A may provide a clinically significant benefit in treating TN (Morra et al., 2016), PN (Shackleton et al., 2016), and DNP (Lakhan et al., 2015). However, thus far, few investigations have explored the evidence for the overall effect of BTX-A on TN and PNP. Therefore, we conducted this metaanalysis to fill this gap. Our review showed that the incidence of adverse events from BTX-A was similar to placebo and was consistent with other studies (Lakhan et al., 2015;Morra et al., 2016). Nonetheless, our effect size is different from other reviews in this field. We are not bioequivalent (Ranoux, Gury, Fondarai, Mas, & Zuber, 2002).
Non-neuropathic pain and other neurological conditions were also involved in the complex mechanisms of PNP (Cohen & Mao, 2014). The mechanism of BTX acts on neuropathic pain involves deactivating the sodium channel and inhibiting the release of inflammatory mediators and peripheral neurotransmitters from sensory nerves (Park & Park, 2017). BTX-A changes the Na current of a neuronal excitable membrane and controls the Na current with a non-concentration-dependent manner, which is different from tetrodotoxin, antiepileptic drugs, and local anesthetics (Cui, Khanijou, Rubino, & Aoki, 2004). The results of subgroup analysis also show good agreement with the findings from previous studies. However, there are sources of variability among studies including injection sites, recurrent injections, dosing regimens, baseline characteristics of subjects, and time of follow-up. The small number of studies available for review, the diversity of treatment options, limited our ability to more accurately assess true treatment outcomes. So, the conclusion should be interpreted with cautious. This raises the need for future trials, with a focus on which dose of BTX-A is optimal for pain relief, how long the duration of treatment lasts, and which route of administration is optimal.
To our knowledge, this is the first meta-analysis to assess the overall impact of BTX-A on TN and PNP compared with placebo.
However, several limitations associated with this meta-analysis should be noted. First, because of the limited number of studies included in this meta-analysis, type-II errors owing to chance cannot be completely excluded (Thorlund et al., 2011). In theory, randomization can eliminate the problem of unknown confounders, but the randomization of a small number of patients may lead to an imbalance in the arms. Therefore, our meta-analysis may produce biased results, especially considering that meta-analysis is based on the normality hypothesis of the significance test (Jackson, Kuriyama, & Hayashino, 2012;Lakhan et al., 2015). Second, meta-analysis has been often criticized for the inclusion of poor-quality trials and the potential of publication bias (Rosenthal & DiMatteo, 2001). Most studies characterized by small samples and poor methodological quality were included in this meta-analysis. In addition, the risk of bias in most trials is unclear. Therefore, the possibility of biased results could not be ruled out. To avoid this debate, more well-designed RCTs were required. In addition, an objective assessment of heterogeneity and publication bias also helps to maintain reliable conclusions. Third, our search strategy only included trials in English database and excluded trials of other languages, which may lead to, some extent, selective bias. The treatment protocols of those trials have some differences. We cannot conclude from the current limited trials which strategy should be the best approach.

| CON CLUS IONS
In summary, our pooled results support the injection of BTX-A as a promising alternative treatment for TN and PNP. Further larger and well-designed RCTs are still needed to provide more in-depth insight into current issues.

ACK N OWLED G M ENTS
We thank all the study participants for their contributions.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request. Botulinum toxin A in postherpetic neuralgia: A parallel, randomized, double-blind, single-dose, placebo-controlled trial. The Clinical