Therapeutic efficacy and safety of Botulinum Toxin A Therapy in Trigeminal Neuralgia: a systematic review and meta-analysis of randomized controlled trials

The study strictly followed the recommendation of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Additional file 1: Table S1). During September and October, we established the protocol and registered our study with PROSPERO, University of York (CRD42015026861). Ten electronic search engines/libraries were systematically searched for relevant publications, including PubMed, Scopus, Web of Science, Google Scholar, Virtual Health Library (VHL), WHO Global Health Library (GHL), ClinicalTrials, POPLINE, System for Information on Grey Literature in Europe (SIGLE), and the New York Academy of Medicine (NYAM). Except for Google Scholar, the search term for all other libraries was as follows: (botulinum OR onabotulinumtoxinA OR onabotulinum OR botox OR botulinus) AND ((trigeminal neuralgia) OR prosopalgia OR (Fothergill's disease) OR (tic doloureux)). For Google Scholar, we used the advanced setting in which “trigeminal neuralgia” was filled in “with all of the words” and “botulinum botulinus botox onabotulinum toxin A onabotulinum” for “with at least one of the words”. Additionally, we conducted a manual search by reviewing the citations within the included publications and reviewing the related references presented in PubMed.

All patients’ outcomes were considered to analyze the efficacy and safety of botulinum toxin in the treatment of trigeminal neuralgia: (1) proportion of respondents – defined as patients with >50 % reduction in mean pain score from baseline to endpoint; (2) mean paroxysms frequency per day; (3) mean visual analog scale (VAS) score at the end of follow up; and (4) adverse events and complications of BTX A treatment.

The standardized template was developed through a pilot extraction with the two most relevant references. Two researchers then independently extracted the data into the template. Extracted data included: authors, publication year, journal, country of authors, country of patients, source of BTX-A, active drug types and their doses, control types and their doses, number and site of injection, characteristics of patients (gender and age), duration of treatment, duration of follow up, year of neuralgia, affected branch, surgical procedures, efficacy point, proportion of responders, relapse, visual analog scale, number of paroxysms per day, quality of life, anxiety scale, and depression scale .

The methodological quality of each RCT was independently assessed by two reviewers using "The Cochrane Collaboration's tool for assessing risk of bias". It is a two-part tool, addressing seven specific domains, including: sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting, and other sources of bias. The judgment of each reviewer on each domain was categorized as ‘low risk,’ ‘high risk,’ or ‘unclear risk’ of bias. Any disagreement was resolved by discussion between two reviewers and by consultation from a supervisor to reach a consensus.

A total number of 267 articles were retrieved from six search engines/libraries (Fig. 1). SIGLE, POPLINE, NYAM, and ClinicalTrials generated no results. After the initial title and abstract screening of the 267 articles, 28 articles were selected for full-text reading. Two independent reviewers performed the full-text screening after which 24 articles were excluded due to: (1) inappropriate study design; (2) unreliably extracted data; (3) in vitro or animal study; and (4) posters. Finally, a total of four unique RCTs, with a total of 178 TN patients, were included for data extraction and final analysis.

The characteristics of four RCTs included in the final analysis were summarized in Table 1. A total of 178 patients (99 in the BTX-A group and 79 in the placebo control group) were included from four trials. The mean age for both case and control groups in three RCTs [23‐25] was ranging from 58.1 to 64.5 for case group and from 58 to 66 for control group. There was no significant difference between BTX-A and placebo groups in terms of frequency of attacks and pain severity before treatment as reported in each included study. The follow-up period varied from 8 to 12 weeks except for no report from study of Zhang et al. The mean duration of the effect was also ranging between 8 to 12 weeks in two trials [23, 25].

The amount of BTX-A injected, the route of injection, the number of injections, and the injection site varied among studies. The amount of BTX-A injected ranged from a minimum of 25 U in study of Zhang et al. [24] to a maximum of 100 U in study of Shehata et al. [26]. In addition, 50 U and 75 U of BTX-A were used in studies of Zúñiga et al. [25] and Wu et al. [23], respectively. The administration of BTX-A included subcutaneous or intradermal routes. The details of the injection protocol in each trial are summarized in Table 2.

The quality of included studies ranged from medium to high as illustrated in Fig. 2. Authors’ judgments with justifications are shown in (Additional file 2: Table S2).

Three patients’ therapeutic outcomes include: (1) proportion of respondents – defined as patients with >50 % reduction in mean pain score from baseline to endpoint; (2) mean paroxysms frequency per day; and (3) mean visual analog scale (VAS) score at the end of follow up.

The overall effect favored BTX-A compared with placebo in terms of proportion of responders (RR = 2.87, 95 % CI [1.76, 4.69], P < 0.0001) with no significant heterogeneity detected (P = 0.31; I² = 4 %) (Fig. 3a). Mean VAS score was significantly lower for BTX-A group at the end of the first month (MD = -2.89, 95 % CI [-4.66,–1.12], P = 0.001), at the end of the 2^nd month (MD = -2.47, 95 % CI [-3.96,–0.99], P = 0.001), and at the end of the 3^rd month of follow up (MD = -3.43, 95 % CI [-5.21,–1.64], P = 0.0002) with no significant heterogeneity detected for all endpoints (Fig. 3b). Mean paroxysms frequency per day was significantly lower for BTX-A group (MD = -29.79, 95 % CI [-38.50,–21.08], P < 0.00001) with no significant heterogeneity (P = 0.21; I² = 36 %) (Fig. 3c).

Regarding safety of BTX-A, there were two reported injection-related side effects: facial asymmetry and edema/hematoma at the site of injection, and both of which were generally tolerated and transitory in nature. The overall occurrence of facial asymmetry in BTX-A group ranged from 2-5 patients while it ranged between 1–2 for edema/hematoma. Facial asymmetry recovered within 5–7 weeks while edema/hematoma resolved within 5–6 days. None of the controlled patients developed facial asymmetry or edema/hematoma except two patients in Shehata et al. [26] and one patient in Wu et al. [23]. The adverse events were fully summarized in Table 3.

Only one study [23] had both sequence generation and allocation concealment rating of "high risk of bias". Omitting this study from the analysis, slightly changed the effect estimate and CI but did not have a serious impact on the findings (Fig. 4). Moreover, the overall effect estimate did not differ significantly for any of the outcomes on using the random effects model (Fig. 5). This in turn adds to the robustness of the results.