Introduction

Facial rhytides, which are perceived as a sign of aging, are a concern of the beauty industry worldwide. Wrinkles on the face may provide an erroneous impression of negative emotions (i.e., anger or sadness) [1, 2]. Individuals have been concerned with improving their appearance throughout history and across virtually all cultures [3]. Currently, with the progress of medical technology, increasingly minimally invasive surgery has made these appeals available [4].

Botulinum toxin A (BTX-A) has been used as the foundation of minimally invasive surgery to improve aesthetic facial rejuvenation. Injectors began with the use of BTX-A to smooth glabellar frown lines and subsequently expanded to other facial areas [5]. BTX-A has also been the first-choice nonsurgical procedure since 2000 [6]. BTX-A treatment for facial rejuvenation has been widely used with approval in more than 50 countries and is currently the most frequently performed nonsurgical cosmetic product in the United States, with more than 3.5 million procedures performed annually as of 2014 [6]. Treatment primarily promotes the individual’s self-perception of appearing younger and better after treatment [7–9]; however, because it is still a medical procedure, concern over the safety profile of BTX-A for facial rejuvenation has increased among health care providers and patients. The overall benefit:risk profile of BTX-A for facial lines can be analyzed by its synthesis because individual clinical trials and data are available. Thus, we performed this meta-analysis to detect treatment group differences in the incidence rate of adverse events (AEs) that may not have a significant difference in individual studies.

Materials and Methods

Data Collection

We performed a search of all available studies present in PubMed, Embase, and Cochrane Central Register of Systematic Reviews, updated on July 2015, on the safety of BTX-A for the treatment of dynamic facial lines from randomized controlled trials (RCTs). The medical subject headings contained the following terms: botulinum toxin A, onabotulinum toxin A, abobotulinum toxin A, incobotulinum toxin A, Hengli botulinum toxin A, glabellar, glabellar lines, glabellar wrinkles, crow’s feet lines, crow’s feet, and crow’s feet wrinkles. The search strategy was restricted to human studies and to articles written in the English language.

At baseline, participants who had moderate-to-severe facial wrinkles at maximum animation met the inclusion criteria according to a standardized facial wrinkle rating scale [10].

Inclusion and Exclusion Criteria

We reviewed abstracts of the retrieved studies with the following criteria: (1) patients who had received treatment of GL or CF; (2) RCTs comparing BTX-A with placebo (saline or vehicle); (3) outcome measures about AEs; and (4) studies containing sufficient raw data reporting BTX-A treatment of GL or CF for the weighted mean difference (WMD) with 95 % confidence intervals. Exclusion criteria included (1) the use of masticatory muscles, masseter hypertrophy, or animals and (2) no raw data reported.

Data Extraction and Assessment of Study Quality

Raw data were independently extracted by 2 reviewers (Z.-H.J. and H.-B.L.) from each article according to the preset selection criteria. Any divergence was resolved by discussion. The Jadad composite scale was used to assess the methodological quality of the included articles [11]. Studies that scored more than 2 out of a maximum possible score of 5 were considered to be high-quality trials [12].

Statistical Analysis

We performed all statistical analyses using Review Manager 5.0.24 statistical software provided by the Cochrane Collaboration. The result was expressed with WMD for the categorical variable and 95 % CIs. We used relative risk (RR) as a summary estimator and the fixed effect model weighted by the Mantel–Haenszel method [13]. If the outcomes had significant heterogeneity (p value of X 2 test < 0.1 or I 2 > 50 %), then the random effect model was used [14, 15]. A funnel plot test was used to assess for evidence of publication bias. Forest plots were used to obtain a graphical representation of the data. A p value less than 0.05 was considered statistically significant.

Results

Study Characteristics

The procedure of filtering eligible studies is shown in Fig. 1. A total of 605 articles remained for further assessment after the titles and abstracts were filtered. After a full-text review, 206 articles were excluded. The three inaccessible full texts were excluded because we obtained no raw data but descriptive abstracts. Adverse events were reported in 16 RCTs that were available between 2002 and 2015.

Fig. 1
figure 1

Study flow diagram

Full size image

Table 1 shows the characteristics of the included trials. A total number of 42,405 participants (26,177 in the BTX-A group and 16,228 in the placebo group) were enrolled. In glabellar lines (GL) studies, the dose/treatment range was 20–80 U, and the participants received five injections/treatment, two in each corrugator muscle and one in the procerus muscle. In crow’s feet (CF) studies, participants were treated bilaterally with the doses of 3,6, 9, or 12 U/side and received one treatment of 3 injections/side in the orbicularis oculi. Further study details are available in published reports of clinical trials [10, 16–29]. According to the injection site, 3 studies described the safety for the treatment of CF [24, 25, 28] and 13 studies referred to GL [10, 16–23, 26, 27, 29]. BTX-A was related to HBTX-A, InT-A, AbT-A, and OnT-A, whereas the placebo materials included saline solution or vehicle. The trials included in our meta-analysis were scored strictly according to the Jadad scoring system. As a result, all of the trials were scored with a value greater than 3.

Table 1 Basic characteristics of included RCTs
Full size table

AEs of BTX-A Vs Placebo Group in All Enrolled Studies

In all of the enrolled trials, the statistical analysis results indicated that the BTX-A group had significantly more AEs (RR = 1.24; 95 % CI 1.07–1.43; p = 0.003) than the placebo group did. There was no substantial heterogeneity as estimated using I 2 statistics [Q (df = 108) = 103.22; p = 0.61; I 2 = 0 %] (Fig. 2). We performed a subgroup analysis of adverse events, including headache, eyelid edema, gastrointestinal disorders, injection site hematoma, injection site hemorrhage, nasopharyngitis, upper respiratory tract infection, sinusitis, bronchitis, influenza, dizziness, pruritus, dermatitis contact, injection site edema, infection, dry eyes, eye disorder, injection site paresthesia, injection site pain, eyelid ptosis, blepharospasm, facial paresis, pain in face, heavy eyelids, somnolence, swollen face, and erythema. Moreover, headache (RR = 1.51; 95 % CI 1.03–2.23; p = 0.04), eyelid ptosis (RR = 4.69; 95 % CI 1.41–15.63; p = 0.01), heavy eyelids (RR = 6.90; 95 % CI 1.28–37.27; p = 0.02), and eye disorder (RR = 2.82; 95 % CI 1.15–6.92; p = 0.02) showed a significantly higher frequency in patients receiving BTX-A injection. Nevertheless, all occurrences of these events were mild or moderate in severity. In contrast, the incidence of the remaining AEs was not significantly different in the BTX-A treatment. The funnel plot showed that the publication bias was not obvious for studies on AEs (Fig. 3a).

Fig. 2
figure 2

Forest plot of trials of BTX-A examining the relative risk of AEs in all enrolled trials

Full size image
Fig. 3
figure 3

a Funnel plot of AEs in all enrolled studies to assess for evidence of publication bias. b Funnel plot of AEs in crow’s feet lines to assess for evidence of publication bias. c Funnel plot of AEs in glabellar lines to assess for evidence of publication bias

Full size image

Comparison of AEs in CF Treatment

Three trials were combined to compare the AEs of BTX-A and placebo injection in the treatment of CF. We observed that the application of BTX-A injection did not result in a significant increase in AEs (RR = 1.19; 95 % CI 0.96–1.48; p = 0.12). Heterogeneity was also not obvious [Q (df = 23) = 19.20; p = 0.69, I 2 = 0 %] (Fig. 4). A subgroup analysis of AEs was performed, including headache, injection site hematoma, injection site hemorrhage, gastrointestinal disorders, sinusitis, nasopharyngitis, and influenza. Except for injection site hematoma (RR = 2.14; 95 % CI 1.13–4.07; p = 0.02), there was no significant difference in the remaining AEs (p > 0.05) between the two groups. The relatively symmetrical funnel plot for studies on AEs is shown in Fig. 3b, indicating no obvious publication bias.

Fig. 4
figure 4

Forest plot of trials of BTX-A examining the relative risk of AEs in crow’s feet lines

Full size image

Comparison of AEs in GL Treatment

There were thirteen studies reporting on the treatment of GL. The pooled results revealed that the BTX-A group had a significantly higher rate of adverse events (RR = 1.47; 95 % CI 1.23–1.77; p < 0.0001). In addition, heterogeneity was not observed [Q (df = 69) = 72.37, p = 0.37, I 2 = 5 %] (Fig. 5). We performed a subgroup analysis, including headache, blepharospasm, eyelid ptosis, eyelid edema, injection site bruising, injection site hemorrhage, injection site erythema, injection site pain, injection site pruritus, injection site edema, dizziness, paresthesia, gastrointestinal disorders, nasopharyngitis, upper respiratory tract infection, dry eyes, infection, heavy eyelids, facial paresis, and acne. Furthermore, we found that headache (RR = 1.53; 95 % CI 1.15–2.03; p = 0.003), eyelid ptosis (RR = 5.56; 95 % CI 1.68–18.38; p = 0.005), and heavy eyelids (RR = 6.94; 95 % CI 1.27–37.93; p = 0.03) occurred more frequently in patients injected with BTX-A, whereas the remaining AEs did not occur significantly more often in the BTX-A group (p > 0.05). Publication bias was also not observed (Fig. 3c).

Fig. 5
figure 5

Forest plot of trials of BTX-A examining the relative risk of AEs in glabellar lines

Full size image

Discussion

To the best of our knowledge, there are few meta-analyses examining the AEs of currently available BTX-A in the removal of facial wrinkles. In clinical practice, we usually observe that the side effects associated with BTX-A injections are uncommon and fleeting. However, improper usage or incorrect layering of filling materials may result in severe or perpetual AEs. As a neuromuscular toxic drug, the potency of the toxin may spread from the point of injection to the surrounding areas and produce symptoms that are similar to BTX-A effects. Thus, it is necessary to systematically evaluate the safety of BTX-A in cosmetic injections.

Although some studies have confirmed the safety of BTX-A in facial aesthetic treatment [30–32], randomized, controlled trials have also shown no significant differences in the incidence of treatment-related AEs between the BTX-A and placebo groups [20, 24, 25]. For example, Bonaparte [33] reviewed three formulations of BTX-A and concluded that the unintended effects included hyperesthesia, flu-like malaise, ectropion, and entropion. Rubi [18] reported that there were many AEs in the BTX-A group than in the controlled group, primarily involving nasopharyngitis (10.0 vs. 4.0 %), sinusitis (4.2 vs. 0.6 %), injection site pain (3.5 vs. 0.6 %), and hypertension (3.2 vs. 0.6 %). In this study, we found that BTX-A injection had a significantly higher rate of AEs in the treatment of CF and GL, particularly involving eye disorders, eyelid ptosis, and heavy eyelids, which is not the same as Zagui [34] who reported that headache (R = 1.07), eyelid ptosis (R = 3.25), local reaction (R = 0.99), and infection(R = 0.94) were significantly different (p > 0.05) in facial application of BTX-A. Thus, eye disorder reflects the sensation of lessened muscle movement, most of which appeared by day 14 [18]. Pharmacological analysis indicated that injected toxins can diffuse through the orbital septum and paralyze the levator palpebrae superioris, thereby causing ptosis [35]. Previous studies have reported that the incidence of eyelid ptosis was 0.8–3.4 % [18, 19, 25, 26, 34]. Most ptosis or heavy eyelid events occurred within 2 weeks of injection and lasted at least 2 weeks [26].

In the current analysis, we observed that headache occurred more often in the BTX-A treatment group than in the placebo group (2.6:1). A review by Aguggia [36] of headache caused by aesthetic injections concluded that prolonged headache may be correlated to the process or pattern of needle insertion rather than a pharmacological effect of BTX-A. In addition, Monheit [25] and Brandt [19] observed that nasopharyngitis was the most commonly reported side effect. In contrast, there was no significant difference in this analysis (p = 0.08). Thus, when performing neurotoxin injections in the upper face, aesthetic providers should perform the operation canonically to lessen the above treatment-related AEs.

To examine the specific AEs in GL and CF, respectively, we performed two subgroup analyses. A previous meta-analysis performed by Brin [37] indicated that there were no reports of significant AEs in CF studies. Interestingly, in this study, we found that there was a statistically significant difference regarding injection site hematoma (p = 0.02), which was consistent with the reports of Ascher [38] and Rzany [39]. However, the specific reason remains to be elucidated. Conversely, in the treatment of GL categories, no significant injection site hematoma (p = 0.94) appeared when applying BTX-A. Guo [32] reported that there were no significant differences in the frequency of the most frequently reported adverse events (headache, blepharoptosis, facial paresis, and injection site disorders including pain, swelling, and hematoma) in the treatment of GL. Nevertheless, in this study, we concluded that there were still statistically more AEs concerning headache, eyelid ptosis, and heavy eyelids in the BTX-A group.

Fortunately, the vast majority of AEs were well tolerated. All of the AEs were mild and recovered without sequelae, except one serious event reported by Carruthers [28], in which a nonviable twin pregnancy was terminated early in the first trimester of pregnancy in the OnT-A group. This result was thought to be treatment related. In addition, Kane [26] reported that the incidence of treatment-emergent AEs was not dose dependent within a specific range.

Limitations

As with most meta-analyses, these results should be interpreted with caution. Several limitations exist. First, some studies lack definite allocation concealment, and there may be selection bias. In addition, reporting biases are likely to exist because the follow-up procedure was performed by telephone recall or self-report at different times after the injection. Second, the outcomes may be affected by heterogeneities between studies due to different BTX-A products from different countries and the different effects on different ethnic groups. Furthermore, the units of biological activity of one formulation cannot be converted or compared with units of any other BTX-A [40, 41]. Finally, these results are based mainly on the doses analyzed.

Conclusion

In conclusion, this meta-analysis confirms the safety profile of BTX-A for GL and CF, at the doses studied, and BTX-A used for removing upper facial wrinkles has some significant mild-to-moderate adverse profiles, including headache, eye disorder, eyelid ptosis, and heavy eyelids. Facial injectors should abide by the technical standards of neurotoxic drugs and be familiar with the local pharmacological effects to lessen severe side effects. Moreover, large-scale, randomized, controlled trials of BTX-A injection for facial rejuvenation are needed, which will provide additional convincing data regarding safety.