Postmarketing safety surveillance data reveals protective effects of botulinum toxin injections against incident anxiety

Randomized controlled trials (RCTs) have shown an antidepressant effect of glabellar botulinum toxin (BoNT) injections. In the FDA Adverse Event Reporting System (FAERS) database, BoNT injection is associated with reduced incidence rates of depression across various non-psychiatric indications, which confirms the previous findings independently of specific expectations to an antidepressant effect of BoNT. The rationale of using BoNT to treat depression is to interrupt proprioceptive body feedback that may reinforce negative emotions. Negative emotions also occur in other mental disorders, suggesting a transdiagnostic therapeutic potential of BoNT in psychiatry. Here we report an analysis of the FAERS database, in which we found that, compared to alternative treatments, BoNT injections were associated with lower incidence of anxiety symptoms and related disorders. Among seven indications/injection sites, we found this protective effect of BoNT in cosmetic use/facial muscles, migraine/facial and head muscles, spasms and spasticity/upper and lower limbs, torticollis and neck pain/neck muscles, and sialorrhea/parotid and submandibular glands (reporting odds ratios 0.79–0.27). These findings are encouraging for possible future RCTs on the use of BoNT as a treatment for anxiety and related disorders.

A series of randomized placebo-controlled trials (RCTs) and meta-analyses have shown that glabellar injections of botulinum toxin can reduce the symptoms of depression [1][2][3][4][5][6] . However, because the noticeable muscle relaxation induced by the toxin makes it impossible to truly blind the study participants for their group allocation, it is unclear to what extent a bias towards expectations/placebo effects in the treatment groups vs. disappointment/ nocebo effects in the control groups may have inflated the large effect sizes observed in these trials.
To overcome this methodical limitation, we have reassessed the antidepressant action of botulinum toxin in the absence of specific expectations to that effect. For that purpose, we have gone into the FDA Adverse Event Reporting System (FAERS) and have compared the incidence rates of depression and related symptoms after treatment with botulinum toxin to a benchmark of alternative treatments for the same indications. Confirming and extending the results of the previous RCTs, we have found a significant preventive antidepressant effect of botulinum toxin across a broad spectrum of indications and injection sites 7,8 .
The rationale for the assessment of botulinum toxin as an antidepressant is the facial feedback hypothesis. The consequential idea that relaxing facial muscles expressing negative emotions would disrupt the proprioceptive afferences from these muscles and their maintaining and reinforcing effect on the expressed emotions 9 .
Since an excess of negative emotions is not specific for depression, but occurs in and determines the suffering associated with the majority of mental disorders, botulinum toxin therapy may not be specific for depression either, but may rather represent a transdiagnostic, emotion-focused treatment approach 10 .
Among the excessively experienced negative emotions, anxiety is one the most common. www.nature.com/scientificreports/ a = No. of anxiety and related AE reports in exposed group, b = No. in exposed group with no anxiety and related AE reports, c = No. anxiety and related AE reports in control group, d = No. in control group with no anxiety and related AE reports.
Standard Error (SE) of the LnROR value was calculated by the following equation: Error bars were computed using 95% confidence intervals.

Results
Botulinum toxin: anxiety and anxiety related adverse events. Patients who were administered BoNT had a significantly lower incidence of anxiety and anxiety-related AE reports, compared to the control groups. It was observed not only for cosmetic use in facial muscles (reporting odds ratios (ROR) 0.79, 95%  Reporting odds ratios with 95% confidence intervals (CI) as calculated by comparing frequencies of anxiety reports in patients administered botulinum toxin for each indication and respective control sub-cohorts.  (Fig. 3).

Discussion
In this survey of the FAERS database we found that treatment with BoNT has a protective effect against incident anxiety disorders or symptoms. This effect was significant for the indications/injection sites cosmetic use/facial muscles, migraine/facial and head muscles, spasms and spasticity/upper and lower limbs, torticollis and neck pain/neck muscles, and sialorrhea/parotid and submandibular glands. There was no effect for hyperhidrosis/ axilla and palm neurological and bladder disorders/detrusor muscle. With no reports in the control group, we found a numerically increased incidence of anxiety after BoNT injection in the blepharospasm/eyelid muscles indication. Although a bit less pronounced and consistent, these findings are largely in line with those from an analogous study on depression (ROR ranging from 0.13 to 0.60), supporting the potential of BoNT injections in the management of mental disorders 8 .
The evaluation of BoNT as a therapeutic for depression and other mental disorders associated with an excess of negative emotions was motivated by the facial feedback hypothesis 9 . However, the cumulating evidence of the efficacy of BoNT in such indications is not per se evidence of the accuracy of this rationale. Our previous study, which showed an antidepressant effect of BoNT across a broad range of indications and injection sites, opened up a broad spectrum of possible explanations for this effect 8,40 . Some of theses explanations are compatible with the facial feedback hypothesis while others challenge it. We have discussed these possibilities at length in the corresponding paper. In principle, they may also apply for our findings on anxiety. In the following, we will discuss them shortly in this regard.
As for modulation of facial feedback, as a mechanism of action, behind the observed effects on anxiety, it may explain the findings for cosmetic use and migraine. The corrugator muscles, which represent the key effectors in the facial expression of any emotions with negative valence, are the main site of BoNT injections in the cosmetic indication and are targeted in the migraine injection scheme, too. Raising the eyebrows belongs to the expression of anxiety and is accomplished by the frontalis muscle, which is also covered by the migraine scheme and is frequently injected for cosmetic reasons, too 41,42 . Hence, interruption of the corresponding proprioceptive feedback may explain the reduced incidence of anxiety. In blepharospasm, the numerically higher incidence of anxiety after BoNT treatment also fits into a similar concept. The main target in this indication is the orbicularis oculi muscle, which is involved in the expression of happiness (Duchenne's smile) and narrows the palpebral fissure 43 . Its relaxation widens the eyes and may confer a negative shift in emotional expression and experience which, in turn, may promote anxiety. Of note, we observed a strong antidepressant effect of BoNT in the blepharospasm indication in our previous study with an overlapping population and an identical analytic approach 8 . Thus, BoNT injections around the eyes may have a differential effect on different psychiatric symptomatology. However, in the present study it is impossible to make a sharp distinction between the glabellar and orbital injections and their possibly opposite emotional effects, because the former is sometimes included in the treatment of blepharospasm and the latter may be injected in the cosmetic treatment of crow's feet.
The reverberating interrelation between muscle activity and emotions is effective beyond the face 44 . Increased muscle tone in various body regions is a common phenomenon in anxiety disorders and may be both cause and effect of anxiety. In the treatment of anxiety disorders progressive muscle relaxation (PMR) is used to induce mental relaxation via tension and subsequent relaxation of skeletal muscles 12,15 . Proprioceptive afferences from the hypertonic musculature may account for the high prevalence of comorbid anxiety disorders or symptoms in patients suffering from dystonia or spasticity 14,19,23,25 . Accordingly, the anti-anxiety effect of BoNT injections in spasms and spasticity/upper and lower limbs as well as torticollis and neck pain/neck muscles may be explained by the interruption of these afferences 45 .
The body feedback concept may be extended to vegetative feedback mechanisms: hyperhidrosis is strongly associated with anxiety, and it is conceivable that increased sweating is not only a vegetative manifestation of anxiety but may also have an anxiety-enhancing feedback effect [46][47][48] . Botulinum toxin treatment has been successfully used as a treatment of anxiety disorders associated with hyperhidrosis 49 . However, we did not find a significant effect in our analyses. Bladder hyperactivity is also a vegetative correlate of anxiety, but we did not find association between BoNT treatment for this indication and decreased incidence of anxiety either 50 . As for saliva production, xerostomia is rather associated with anxiety than sialorrhea 51 . However, we found an association of BoNT treatment of sialorrhea with absence of anxiety. In summary, these findings do not support a role of interoceptive/vegetative feedback mechanisms in the observed anti-anxiety effect of BoNT.
It is possible, yet improbable that direct pharmacological BoNT effects within the CNS may explain its psychotropic action. BoNT may undergo targeted, transneuronal transport into the CNS where it may theoretically reach structures involved in the regulation of emotions 52,53 . In theory, BoNT may also reach the CNS and accomplish its anti-anxiety effect via systemic distribution. However, the amount of circulating BoNT may be very low, and the anti-anxiety effect shows no dose-dependence across the investigated indications with large vs. small muscles/muscle groups 54 . More likely, the peripheral action of BoNT may initiate a chain of neurochemical and neuroplastic changes that may be propagated to remote sites within the CNS 55  www.nature.com/scientificreports/ has been observed in patients treated for dystonia and spasticity 45,[56][57][58] . It may also explain anxiolytic effects of BoNT applied at various injection sites in rats or mice [30][31][32] .
In the investigated indications, BoNT may have higher efficacy and better tolerability than the treatment options that were taken as comparators. Unfortunately, the FAERS database does not include efficacy data. As some of these conditions are chronic and burdensome, they may lead to secondary, reactive psychiatric comorbidities including anxiety disorders and related symptoms 59 . Hence, the more a treatment improves the primary condition for which it is given, the more it may also protect against the sequel of this condition. Thus, differential relief from the burden of disease between the BoNT and the control group may lead to overestimation of a possible specific anti-anxiety effect of BoNT. This may include relief from pain, which is a symptom of some of the investigated indications, especially migraine. However, superior efficacy and tolerability is not a unifying explanation of our findings either, because in the blepharospasm indication, in which it is the most effective treatment, there is no protective effect of BoNT against anxiety, but rather an anxiogenic tendency 60 . This also applies for the other indications in which BoNT did not show a protective effect against anxiety.
A neuronal structure that may mediate effects of BoNT on emotional experience and anxiety, is the amygdala [61][62][63] . Experimental studies have shown that facial injections of BoNT can modify its activity in response to emotional stimuli 64,65 .
There are some general limitations of this study. FAERS/AERS reporting is voluntary and often incomplete. Thus, the investigated data sets represent only a fraction of actual cases and the frequencies do not represent population incidences. Moreover, legal and scientific variables as well as newsworthiness may influence reporting to FAERS/AERS 66,67 . To address these limitations and to assess the significance of the difference between the sub-cohorts, we used disproportionality analysis with reporting odds ratios and 95% CI. Other limitations to consider include occasionally missing demographic variables, treatment doses and durations, and comprehensive medical records as well as bias associated with the comparator (differential efficacy, undetected differences between patients treated with the substance of interest and the comparator). Moreover, unreported life events and situations may have an imponderable impact on the incidence of anxiety. We excluded all the reports with comorbid anxiety disorders or anxiolytic medications (both labelled and off-label use); however, both may be underreported, which may affect the results. Exclusion of these reports may lead to underestimation of the efficacy of BoNT against anxiety, because we capture only preventive effects on incident anxiety. Therapeutic effects on prevalent anxiety may be more pronounced, but are not accessible to our analytic approach. Across all indications, there are differences in the concurrent medications between the BoNT and the reference groups, which may have confounding effects.
In conclusion, our findings show that BoNT administered for various indications and injection sites may have a protective effect against incident anxiety. The anti-anxiety effect represents an advantage over the alternative treatment options, because anxiety disorders and related symptoms are a frequent comorbidity in the respective indications. Even though afflicted with several limitations, our findings are encouraging to pursue the anxiolytic potential of BoNT in RCTs with patients suffering from anxiety disorders. Though there are effective pharmacological and psychotherapeutic treatments for these disorders, there is a need for further therapeutic options and BoNT may be one of them.

Data availability
The data sets are de-identified and made available to the public online by the United States Food and Drug Administration. Institutional Review Board requirements do not apply under 45 CFR 46.102. https:// www. fda. gov/ drugs/ quest ions-and-answe rs-fdas-adver se-event-repor ting-system-faers/ fda-adver se-event-repor ting-system-faers-latest-quart erly-data-files. Both FAERS and AERS datasets are de-identified and are made available online at: http:// www. fda. gov/ Drugs/ Guida nceCo mplia nceRe gulat oryIn forma tion/ Surve illan ce/ Adver seDru gEffe cts/ ucm08 2193. htm. Institutional Review Board Requirements do not apply under 45 CFR 46.102. There was no direct human participation in the study. Thus, all experiments were performed in accordance with relevant guidelines and regulations.