**Monocentric Prospective Study into the Sustained Effect of Incobotulinumtoxin A (XEOMIN®) Botulinum Toxin in Chronic Refractory Migraine**

**Ioana Ion 1,\*, Dimitri Renard 1, Anne Le Floch 1, Marie De Verdal 1, Stephane Bouly 1, Anne Wacongne 1, Alessandro Lozza <sup>2</sup> and Giovanni Castelnovo <sup>1</sup>**


Received: 16 April 2018; Accepted: 28 May 2018; Published: 1 June 2018

**Abstract:** Refractory chronic migraine is a disabling disorder impacting quality of life. BOTOX® (Onabotulinumtoxin A) is approved as a prophylactic treatment of chronic migraine in patients unresponsive to at least three prior preventive treatments. The objective of this study was to determine the prophylactic effect of 145 U XEOMIN® (Incobotulinumtoxin A) injected at 31 specific sites in adult patients with refractory chronic migraine. Sixty-one patients (8 men and 53 women, mean age 50) with migraine were recruited, including 20 patients with isolated chronic migraine, 18 patients with chronic migraine associating tension-type headache, 12 patients with migraine associating medication *overuse headache*, and 11 patients with episodic disabling migraine. The mean number of injections and duration of treatment per patient was 3.5 (range 2–13) and 21 (6–68) months, respectively. From baseline to first injection, 44 patients (73%) had >50% reduction in frequency of migraine episodes, 29 patients (48%) showed >50% reduction in number of headache days, and 28 patients (46%) had a >50% reduction in drug intake. Stable response for all three parameters was observed after the last injection. XEOMIN® thus seems to represent an effective and sustained prophylactic treatment of chronic migraine.

**Keywords:** refractory chronic migraine; tension headache; medication *overuse headache*; prophylactic treatment; XEOMIN®

**Key Contribution:** In this study, XEOMIN® proved to be an effective prophylactic treatment with sustained efficacy in patients with a refractory CM and EDM.

#### **1. Introduction**

Between 2% and 15% of the world's population suffers from migraines, with a wide variation of frequency of attacks [1]. Refractory chronic migraine (CM) is a disabling illness causing significant interference with quality of life, despite promising results from trials of acute and prophylactic treatments. CM affects up to a fifth of migraine patients [1].

Despite the development of new pathophysiological hypotheses (and associated recent advancements on drug development), the benefit of the majority of conventional migraine preventive drugs does not exceed 50% over placebo [2]. A recent meta-analysis shows that only high dose topiramate and sodium valproate were more effective than placebo at reducing migraine by more than 50% [3].

The National Institute for Health and Clinical Excellence (NICE) and the US Food and Drug Administration have both recently approved BOTOX® (botulinum toxin type A—BoNT/A)

for the prophylaxis of CM, specifically in refractory (i.e., not responding to at least three prior prophylactic treatments) CM patients [4,5]. In studies analyzing efficacy of BOTOX® in migraine patients, treatment also seemed to be effective in patients with migraine and associated medication *overuse headache* [MOH]) [6]. It was recently suggested that BOTOX® also represents an effective and safe intervention to target psychiatric comorbidities of migraine with improvements in depression and anxiety [7]. Double-blind, placebo-controlled trials of toxin A injections in patients with isolated episodic disabling migraine headaches (EDM) or tension-type headache (TTH) did not show significant effect of BoNT/A onabotulinum, even after controlling for a high placebo effect and after dose stratification [8–10]. However, patient numbers were low in these studies. Available clinical trials analyzing BONT/A efficacy in patients with isolated TTH show conflicting data, with a majority of studies showing no efficacy [11].

Botulinum toxin A efficacy has never been analyzed in patients with migraine and associated TTH.

XEOMIN® inhibits the release of acetylcholine from peripheral cholinergic nerve endings acting like a myorelaxant, but also like an analgesic by suppressing the peripheral and central sensitization [12]. Unlike other neurotoxins, XEOMIN® triggers minimal allergic reactions as it has no binding albumin protein [13].

To the best of our knowledge, XEOMIN**®** efficacy in migraine or other headache types has never been reported. The objective of our study was to assess the effect of XEOMIN® treatment in refractory CM (isolated or associated with TTH or MOH) and EDM patients.

#### **2. Results**

Demographic and clinical features of patients are shown in Table 1. Sixty-one patients (8 men and 53 women; mean age 50, range 22–73) were recruited, including 20 patients with isolated CM, 18 with CM-TTH, 12 with CM-MOH, and 11 with EDM. Before inclusion, nonsteroidal anti-inflammatory drugs, paracetamol, morphine, and triptans were used in 18, 20, 9, and 32 of the 61 patients, respectively. The mean number of injections and mean duration of treatment per patient was 3.5 (range 2–13) and 21 (range 6–68) months, respectively (Table 2).


**Table 1.** Demographic and clinical features.

**Table 2.** Treatment effect of XEOMIN®.


For the entire patient group, between baseline and the episode after the last injection, a >50% reduction in frequency of migraine episodes and headache days was observed in 44 (73%) and 29 patients (48%), respectively, whereas a >50% decrease in drug intake was observed in 28 patients (46%). In total, 44 patients (73%) were considered responders, including 18 CM, 16 CM-TTH, 6 CM-MOH, and 4 EDM patients. The 17 non-responders showed absence of therapeutic response in six patients and <50% response in 11 patients after two injections.

Compared to non-responders, responders showed >50% reduction of consumption of any acute medication after a mean of two years of treatment (*p* < 0.001). The mean number of pain-relief pills per month reduced from baseline to the episode after the last injection: 51 to 18 for morphine, 67 to 14 for paracetamol, 12 to 4 for triptan, and 12 to 5 for nonsteroidal anti-inflammatory drugs.

Median migraine episodes decreased from 8.5 (range 1–30) at baseline to 2 (range 0–16) after the first injection (*p* < 0.001) and to 2 (range 0–15) after the last injection (*p* < 0.001) (without difference in efficacy between the first and last injection, *p* = 0.3), corresponding to a 76% decrease from baseline to both later time points (Figure 1). Mean headache days decreased from 20.8 (SD 9.6) at baseline to 8.5 (SD 8.2) after the first injection (*p* < 0.001) and to 7.3 (SD 7.6) after the last injection (*p* < 0.001) (without difference in efficacy between the first and last injection, *p* = 0.4), corresponding to a 41% and 36% decrease from baseline to the first and last injection, respectively (Figure 1).

**Figure 1.** Responder group: box-and-whisker diagrams, presented as medians and interquartile ranges of (**a**) number of migraine episodes and (**b**) number of headache days. Circles represent outliers.

In the subgroup of patients with isolated CM (Figure 2), mean migraine episodes decreased from 9.83 (SD 6.23) at baseline to 4.06 (SD 4.04) and 4.26 (SD 3.98) after the first (59% reduction, *p* < 0.001) and last injection (57% reduction, *p* < 0.001), respectively (without difference in efficacy between first and last injection, *p* = 0.8); mean number of headache days was reduced from 13.11 (SD 7.71) at baseline to 4.40 (SD 4.27) (69% decrease, *p* < 0.001) after the first injection and to 4.66 (SD 4.21) after the last injection (67% decrease, *p* < 0.001) (without difference in efficacy between the first and last injection, *p* = 0.9).

**Figure 2.** Isolated chronic migraine subgroup: box-and-whisker diagrams, presented as medians and interquartile ranges of (**a**) number of migraine episodes and (**b**) number of headache days. Circles represent outliers.

In the subgroup of patients with headache type other than isolated CM (Figure 3), the median number of migraine episodes decreased from 8 (range 1–30) at baseline to 2 (range 0–14) after the first injection (69% decrease, *p* < 0.001) and to 1 (range 0–9) (93% decrease, *p* < 0.001) after the last injection (without difference in efficacy between the first and last injection, *p* = 0.1), and median number of headache days from 27 (range 8–30) at baseline to 9 (range 0–30) after the first injection (57% decrease, *p* < 0.001), and to 4.5 (0–30) after the last injection (69% decrease, *p* < 0.001) (without difference in efficacy between the first and last injection, *p* = 0.3).

**Figure 3.** CM/TTH, CM-MOH and EDM subgroup: box-and-whisker diagrams, presented as medians and interquartile ranges of (**a**) number of migraine episodes and (**b**) number of headache days. Circles represent outliers.

When comparing both subgroups (isolated CM vs. non-isolated CM), non-isolated CM patients showed more frequent treatment response than isolated CM patients (*p* < 0.001 for both number of migraine episodes between baseline and first injection and between baseline and last injection; *p* < 0.001 for both the number of headache days between baseline and first injection and between baseline and last injection; *p* < 0.001 for both reduction in drug intake between baseline and first injection and between baseline and last injection).

The treatment was generally well tolerated. The most frequent adverse events reported were neck pain (7%) and flu-like syndrome (5%), but all these symptoms were transitory, disappearing after a maximum of 72 h, did not interfere with patient activity, and did not need further management. Overall, no patients discontinued treatment due to adverse events.

#### **3. Discussion**

In this study, XEOMIN® proved to be an effective prophylactic treatment with sustained efficacy in patients with a refractory CM and EDM. Treatment efficacy was observed both in patients with isolated migraine and in patients with migraine and associated TTH or MOH. XEOMIN® was well tolerated and no serious adverse events were observed.

To the best of our knowledge, this is the first prospective study analyzing the prophylactic efficacy of XEOMIN® in patients with refractory migraine and the efficacy of botulinum toxin in patients with migraine and associated TTH.

The PREEMPT 2 study has demonstrated that BOTOX® is effective for prophylaxis of adult CM patients [14]. BoNT/A treatment was also effective (significant reduction of headache days and triptan intake) in the subgroup of patients with CM and associated MOH in the PREEMT study [6].

One retrospective case series reporting on 21 Xeomin®-treated CM patients showed improvement in headache severity and frequency [15].

The treatment response we found in our XEOMIN® study corresponded to that observed in the earlier BOTOX® study [16]. However, in contrast with the PREEMPT 2 study where only the number of headache/migraine days was analyzed, we also assessed the number of migraine episodes and the reduction in drug intake. Our data support evidence that XEOMIN® treatment is effective on all three headache parameters.

In our study, the incobotulinum effect was sustained (lasting from the very first injection to the last) in all responders.

Study limitations included the lack of a control group (because of the known high placebo effect for headache treatment and especially for injectable therapies), the small sample size, and the non-inclusion of specific tools assessing depression, disability and quality of life. The long-term follow-up of the patients included in our study is ongoing.

A double-blind BOTOX® and XEOMIN® treatment study should be conducted in more patients (and compared with a placebo arm) with isolated migraine and with migraine associating other headache types in order to compare the efficacy of both BoNT/A treatments in the different patient groups.

#### **4. Conclusions**

These results suggest that IncobotulinumtoxinA toxin may be an effective and safe prophylactic treatment for a variety of refractory migraines. Its effect is sustained over time, reducing medication overuse, as suggested in the open-label phase of PREEMPT.

#### **5. Materials and Methods**

Between August 2009 and January 2016, we invited all adult patients with CM (isolated or associated with TTH or MOH) to a specialized headache consultation at our center (Nîmes University Hospital, France) or with EDM to be included in our study. The diagnosis of the different headache types analyzed in our study fulfilled the criteria of the International Classification of Headache Disorders ICHD-3 [17]. Pregnancy was an exclusion criterion. Signed written informed consent was obtained in all included patients.

Each patient was injected with a total of 145 UI of XEOMIN® at 31 specific points, topographically similar to the myogenic trigger points associated with referred pain locations (facial, pericranial and cervical muscles) [7]. The injections were scheduled at an interval between 3 and 6 months.

We prospectively evaluated the benefit of XEOMIN® by calculating the number of migraine episodes and headache days, and the drug intake (expressed in number of pills of nonsteroidal anti-inflammatory drugs, paracetamol, morphine, and triptans) during the six months preceding the first injection, during the 3 to 6 months after the first injection, and during the 3 to 6 months after the last injection.

Responders were defined as patients with at least 50% reduction in frequency of migraine episodes and/or headache days. Responder analyses were performed for the entire patient group and for 2 pre-specified patient subgroups (i.e., patients with isolated CM and patients with headache type other than isolated CM). Further injections were stopped in case of non-response after 2 injections.

The software used for statistical analysis was Statistical Package for Social Sciences (SPSS 20). Paired *t*-test was used to study differences between outcome variables (with *p* < 0.05 deemed as significant difference). Median was used for the groups with abnormal distribution, and mean for the rest of the analyses when permitted.

**Funding:** This research received no external funding.

**Author Contributions:** Conceptualization, G.C. and I.I; Methodology, I.I.; Software, I.I.; Validation, A.L., M.D.V. and A.L.F.; Formal Analysis, I.I.; Investigation, G.C.; Resources, A.W.; Data Curation, S.B.; Writing-Original Draft Preparation, I.I.; Writing-Review & Editing, D.R.; Visualization, D.R.; Supervision, G.C.; Project Administration, G.C.

**Acknowledgments:** We would like to thank Sarah Kabani (Department of Biostatistics, Epidemiology, Public Health and Innovation in Methodology (BESPIM), CHU de Nîmes, 4 Rue du Professeur Debré, 30029 Nîmes Cedex 09) for substantive editing of our manuscript.

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **The Expanding Therapeutic Utility of Botulinum Neurotoxins**

#### **Elena Fonfria 1,\*, Jacquie Maignel 2, Stephane Lezmi 2, Vincent Martin 2, Andrew Splevins 1, Saif Shubber 3, Mikhail Kalinichev 2, Keith Foster 1, Philippe Picaut <sup>4</sup> and Johannes Krupp <sup>2</sup>**


Received: 13 April 2018; Accepted: 16 May 2018; Published: 18 May 2018

**Abstract:** Botulinum neurotoxin (BoNT) is a major therapeutic agent that is licensed in neurological indications, such as dystonia and spasticity. The BoNT family, which is produced in nature by clostridial bacteria, comprises several pharmacologically distinct proteins with distinct properties. In this review, we present an overview of the current therapeutic landscape and explore the diversity of BoNT proteins as future therapeutics. In recent years, novel indications have emerged in the fields of pain, migraine, overactive bladder, osteoarthritis, and wound healing. The study of biological effects distal to the injection site could provide future opportunities for disease-tailored BoNT therapies. However, there are some challenges in the pharmaceutical development of BoNTs, such as liquid and slow-release BoNT formulations; and, transdermal, transurothelial, and transepithelial delivery. Innovative approaches in the areas of formulation and delivery, together with highly sensitive analytical tools, will be key for the success of next generation BoNT clinical products.

**Keywords:** new indications; formulation; delivery

**Key Contribution:** Review presenting the current therapeutic landscape of BoNTs, their biological effects distal to the site of injection, and the challenges faced in their pharmaceutical development.

#### **1. Current Therapeutic Landscape**

Based upon the use of neutralising antibodies, there are currently seven different serotypes of botulinum neurotoxin (BoNT) that have been reported, from BoNT A through to G. More recently, an eighth serotype has been identified at the protein sequence level, BoNT/X, although this has not yet been produced or characterised as a protein [1]. Over the last 15 years, it has also been recognised that within each serotype there are multiple sub-types, each with a unique protein sequence and a distinct molecular entity [2]. All BoNT serotypes and their subtypes inhibit the neurotransmitter release from nerve terminals through the prevention of acetylcholine release at the neuromuscular junction, which results in flaccid paralysis. Despite this, the intracellular target proteins, receptors, pharmacodynamic properties, and potencies vary substantially between BoNT serotypes.

Despite the great diversity of natural BoNTs and the fact that there are several manufacturers of BoNT for both aesthetic and therapeutic applications, to date, all of the commercially available BoNT products are serotype BoNT/A1, except for a single serotype BoNT/B product, Myobloc®/Neurobloc®. However, BoNT/B showed to be less potent in the clinic than had been anticipated based on its efficacy during animal studies. The reason for this has recently been identified as a residue difference within human synaptotagmin II, which is the protein receptor for BoNT/B [3,4]. The residue difference is within the binding recognition sequence of synaptotagmin II and renders human synaptotagmin II a lower affinity receptor for BoNT/B than in other species. The low affinity for human synaptotagmin II requires higher doses of toxin to be injected in order to achieve efficacy that is similar to BoNT/A, and this results in a different safety profile and a high rate (up to 18%) of neutralizing antibodies as compared to BoNT/A (0 to 5%) ([5]; FDA label). For this reason, the BoNT/B product is not widely used. The identification of the molecular explanation for BoNT/B having lower efficacy in humans has enabled the design of modified BoNT/B sequences that are able to bind human synaptotagmin II with higher affinity, and so to potentially have improved therapeutic characteristics [6].

The three most widely used and commercially available BoNT/A products are Dysport® (abobotulinumtoxinA, Ipsen, Paris, France), Botox® (onabotulinumtoxinA, Allergan, Dublin, Ireland), and Xeomin® (incobotulinumtoxinA, Merz, Frankfurt am Main, Germany). While the mechanism of action of these three products is the same, there are differences between them, since the BoNT/A that is produced and purified is specific to each toxin manufacturer, and the final formulation is different for each product. For example, the potency units are different for each product, and they are not interchangeable [7]. The unit is defined as the mouse LD50 by the intraperitoneal route, and each company assesses these in its own proprietary assay. This means that a "Dysport® unit" is not the same as a "Botox® unit", or yet again, a "Xeomin® unit". The clinically approved doses, which are measured in the respective units, are again, therefore, different between the products. The amount of human serum albumin also differs (highest in Xeomin® [1 mg], when compared to Botox® [500 μg], and lowest in Dysport® [125 μg]), and most importantly, the amount of neurotoxin is different in each individual product (3.24 ng per 500 U vial for Dysport®; 0.73 ng per 100 U vial for Botox®; 0.44 ng per 100 U vial for Xeomin® [8]), leading to potentially different efficacy profiles in humans.

An excellent review of commercially available BoNT products and their manufacture is provided in Pickett, 2014 [9]. The key aspects of the major BoNT products that are available in the United States and Europe are given in Table 1. Currently, all of the commercially available BoNT products are manufactured using native *Clostridium botulinum* as the production organism, although some companies are exploring the use of recombinant expression using *Escherichia coli*. This would allow a standardised and safer manufacturing process for BoNT/A and for other serotypes and genetically modified versions of the neurotoxin.


**Table 1.** Characteristics of current major botulinum neurotoxin (BoNT) products.

† Protein (ng/vial) is for entire neurotoxin complex, the total protein load being dominated by albumin. HSA = human serum albumin. <sup>1</sup> AboA = abobotulinumtoxinA (Dysport®). Dysport® PI, Ipsen, 2015. <sup>2</sup> IncoA = incobotulinumtoxinA (Xeomin®). Xeomin® PI, Merz, 2015. <sup>3</sup> OnaA = onabotulinumtoxinA (Botox®). Botox® PI, Allergan, 2015. <sup>4</sup> RimaB = rimabotulinumtoxinB (Myobloc®/Neurobloc®). Myobloc® PI, Worldwide Meds, 2010.

In the United States of America (USA) and Europe, both Dysport® and Botox® are approved for use in adult upper and lower limb spasticity, while Xeomin® is only approved for use in adult upper limb spasticity (the Xeomin adult lower limb Phase 3 pivotal study did not meet its primary endpoint at the 400 U dose [NCT01464307, https://clinicaltrials.gov/]). Dysport® is the only BoNT/A product approved in the USA for the use in children with lower limb spasticity. The Botox® study is still ongoing, (NCT01603628), while the Xeomin® pivotal Phase 3 study did not meet its primary endpoint (NCT01893411).

Over the last two decades, the localized efficacy of BoNT/A, and its well tolerated safety profile, has resulted in significant growth of on-label use across multiple therapeutic and aesthetic indications. BoNT/As are well tolerated, which has also favoured significant growth in its empirical/off-label use in a variety of movement, ophthalmologic, gastrointestinal, urologic, orthopedic, dermatologic, secretory, and pain disorders.

BoNT is injected locally in skeletal muscles (on-label such as cervical dystonia, hemifascial spasm, blepharospasm, spasticity in adult and children; or, off-label such as writer cramp, tremors, spasmodic dysphonia), smooth muscles (on-label neurogenic detrusor overactivity, idiopathic bladder overactivity; or, off label such as bladder pain syndrome, detrusor sphincter dyssynergia) or exocrine gland hyperfunction (on-label sialorrhea, axillary hyperhidrosis; or, off-label, such as Frey's syndrome, plantar/palmar hyperhidrosis). More recently, BoNT has been used in pain related disorders (on-label chronic migraine or off-label, such as osteoarthritis, neuropathic pain, lower back pain). In aesthetic, BoNT is used to treat multiple facial hyperkinetic lines that are related to striated muscles spasm (on-label glabellar lines, lateral canthal lines, front lines; or, off-label such as lateral eyebrow lift, nasal lines, mid and lower face, neck).

Currently available BoNT products have certain limitations. As the injection is local, there is the risk, although very low, for the toxin to diffuse locally in the vicinity of the tissue injected and induce unwanted effects, or more importantly, to spread far from the original injection site. These unwanted side effects are influenced by the injection technique, the dose, and the volume. The injection procedure can also be a limitation, as it necessitates the specific training for injectors and can be painful for patients, hence there is a need to develop novel formulations and delivery techniques, such as transdermal delivery or products with a longer duration of response (BoNT/A products are reinjected every three to four months in most indications, while an injection interval of six to nine months duration would be of clinical interest). Limitations of the current toxin products may also relate to vial size (3 to 10 mL) relative to some therapeutic indications where the injection volume can be 15 to 30 mL. In addition to these limitations, when seeking to expand the approved clinical indications for BoNT products, there is a strong need to assess the risk-benefit profile through well-designed research clinical trials.

#### **2. Alternative Serotypes, Broadening the Therapeutic Landscape**

Although only BoNT/A or BoNT/B serotype products are currently available as licenced clinical products, several other serotypes of BoNT have been explored as potential therapeutic agents in humans. The use of BoNT/F as an alternative serotype in patients who had become resistant to BoNT/A through antibody formation was first reported in 1992 [10]. Subsequent studies confirmed the clinical benefit of BoNT/F in dystonic patients, and revealed that it had a significantly shorter duration of response [11–17].

Another BoNT serotype, with an even shorter duration of response, is BoNT/E. A comparison of BoNT/E and BoNT/A in the *extensor digitorum brevis* muscle of human volunteers using compound muscle action potential (CMAP) to assess the efficacy that was demonstrated a much faster recovery following BoNT/E administration [18]. Surprisingly, in this study, when the *extensor digitorum brevis* was injected with a combination of both BoNT/A and E, recovery was similar to that observed with BoNT/E alone; this is not consistent with the findings in a number of animal experiments, nor with the understanding that BoNT/A light chain survives in the neuronal cytosol and results in ongoing paralysis, even after the BoNT/E has worn off [19–21]. Short acting BoNTs, such as E

and F, have potential clinical applications in therapeutic areas where a significantly shorter duration of response (3–6 weeks) when compared to that of BoNT/A (3–4 months) is required, for example, in orthopedics and rehabilitation medicine [22]. Recently, a biotechnology company, Bonti, announced that it is developing a BoNT/E product, EB-001, in aesthetic and therapeutic indications. The first phase 2 study, with EB-001 in glabellar frown lines, demonstrated the safety and clinical efficacy for this indication. Bonti have also initiated a phase 2 clinical study to evaluate the safety and efficacy of EB-001 by intramuscular injection in reducing musculoskeletal pain in patients undergoing elective augmentation mammoplasty. Ipsen also have a recombinant BoNT/E in phase 1 clinical studies; this is the first recombinant botulinum neurotoxin ever to have entered clinical trials in humans.

In addition to its shorter duration of response, which offers therapeutic differentiation from existing BoNT products, preclinical studies also indicate that BoNT/E has additional properties that may open up novel therapeutic applications. Experimental studies in rats showed that BoNT/E that was injected into the hippocampus inhibited glutamate release and reduced both focal and generalised kainic acid-induced seizures [23]. BoNT/E also prevented neuronal loss and long term cognitive defects that were associated with kainic acid seizures, and reduced sensitivity to electrical stimulation of kindling, indicating an antiepileptogenic activity. The authors suggested that this activity is a result of selective inhibition of excitatory glutaminergic versus GABAergic transmission by BoNT/E attributable to preferential localisation of SNAP-25 in excitatory hippocampal neurons.

A third alternative to BoNT/A, which was explored for its potential therapeutic utility, is serotype BoNT/C. A comparison of the neuromuscular blockade induced by BoNT/C, assessed electrophysiologically in the *extensor digitorum brevis* muscle of human volunteers, showed an efficacy and duration of action that was very similar to BoNT/A [24]. The same authors reported that it was used to treat two patients with idiopathic facial hemispasm and one patient with blepharospasm with long lasting beneficial effects. BoNT/C injections, like BoNT/A injections, did not affect the motor neuron count in human volunteers, showing it to be well tolerated [25]. BoNT/C was successfully tested in a pilot series of BoNT/A non-responsive patients with focal dystonia (four with torticollis and two with blepharospasm) [26]. Patients did not develop secondary resistance to BoNT/C after chronic use. The activity of BoNT/C in BoNT/A-resistant patients was further confirmed in patients with cervical dystonia [27]. Despite this early clinical interest, as of yet, no commercial product based upon BoNT/C has been produced, possibly reflecting that it is not sufficiently differentiated from BoNT/A in its clinical characteristics.

In addition to the BoNT serotypes that are described above, potential differences between BoNT/A subtypes have also been explored. Comparison of BoNT/A subtypes 1 to 5 revealed distinct characteristics both in vitro and in vivo [28,29]. BoNT/A2 was more potent in vitro and in vivo compared to BoNT/A1 [28,30–32]. As BoNT/A2 was reported to enter neuronal cells faster than BoNT/A1, this was proposed to be responsible for the difference in potency [30]. It has also been suggested that a higher occupancy of the cellular receptors by BoNT/A2 as compared to BoNT/A1 is the underlying mechanism [33]. The crystal structure of BoNT/A2 bound to the luminal domain of its cognate protein receptor SV2C, has been recently resolved, showing that the mode of binding of BoNT/A2 to SV2C does not substantially differ from that of BoNT/A1 [34]. The intoxication symptoms in mice that were injected i.v. with 10<sup>4</sup> units/mL of BoNT/A2 have been reported to differ from those in mice injected i.v. with 105 units/mL of BoNT/A1 [29,35]. Using a twitch tension assay in mouse hemi-diaphragm and rat grip strength model it was proposed that BoNT/A2 was a more potent neuromuscular blocker and spread less to the contralateral limb than BoNT/A1 [32]. For further discussion of the differences in spread between BoNT/A1 and BoNT/A2, see Section 4.4. It has also been reported that BoNT/A2 is less immunogenic than BoNT/A1 [36], but the comparison compared a complexed form in the case of BoNT/A1 with purified BoNT/A2, so there is the potential that the differential presence of complexing proteins influenced the measured immunogenicity. The authors also claimed that BoNT/A2 was less susceptible to neutralisation by human antisera raised to BoNT/A1 complex toxoid vaccine. In healthy volunteers, BoNT/A2 caused a reduction in CMAP with a

comparable onset and duration as onabotulinumtoxin A [37]. Subsequently, BoNT/A2 was compared to onabotulinumtoxin A in post-stroke spasticity. BoNT/A2 was reported to show higher efficacy and less spread, as measured by the hand grip of the unaffected side, than the A1 toxin [38]. Animal models have also shown that BoNT/A2 could be a promising therapeutic in Parkinson's disease and inflammatory and neuropathic pain [39–41]. In 2016, Shionogi & Co. Ltd. (Osaka, Japan) announced a licence agreement with Tokoshima University for BoNT/A2, and is undertaking its global development as a novel BoNT therapeutic.

#### **3. Novel Indications**

#### *3.1. Pain and Migraine*

Pain is the most common reason for a patient to seek medical help, and it is seen by physicians as a symptom of an underlying medical condition [42]. While acute pain typically measures in days to weeks, chronic pain can persist from months to years, thus representing a major health-care issue with a serious impact on quality of life and significant socio-economic cost [43,44]. Chronic pain is classified as inflammatory (such as osteoarthritis, rheumatoid arthritis), neuropathic (such as diabetic, post-herpetic neuralgia), or dysfunctional (such as tension type headache, migraine, interstitial cystitis, irritable bowel syndrome) [45]. Current medication for acute and chronic pain includes opioids, cyclooxygenase inhibitors, acetaminophen, as well as several repositioned drugs, such as the antidepressant drug duloxetine and the antiepileptic drug pregabalin. While these drugs offer some pain relief, they are not consistently effective, often result in tolerance when being taken over a prolonged period and cause serious side-effects that hinder their use [46]. Therefore, there is a need for new pain therapeutics that can offer improved efficacy, reduced propensity to cause tolerance, and fewer side-effects. There is rapidly growing evidence that BoNT can offer an effective, long-lasting pain relief, and very few side-effects in a wide range of medical conditions.

The initial evidence of pain relief in response to BoNT treatment is linked to several serendipitous observations in clinical studies involving muscle hyperactivity and muscle pain in patients with spasticity, dystonia, and related conditions. While muscle relaxant effects may play a role in some of these conditions, there is growing evidence that they cannot account for all mechanisms mediating pain relief following BoNT treatment across a growing range of medical conditions. For example, even in conditions involving intramuscular-administered BoNT, effects in pain can precede and/or last longer in comparison to the muscle-relaxant effects [47,48]. Also, it is now well-accepted that, in addition to local uptake in the synaptic terminal, a distinct secondary uptake pathway results in retrograde transport of BoNT and its activity at distal sites [49]. The retrograde transport of BoNT from the site of uptake at the sensory neuron ending into the dorsal root ganglion and the spinal cord is believed to play a pivotal role for the activity of BoNT in pain [50,51]. In addition to neurons, BoNT can impact the functional activity of glial cells, such as Schwann cells and astrocytes, which suggests the presence of yet another mechanism of pain modulation by BoNT [52–54].

Since the initial discovery several decades ago, the field of therapeutic application of BoNT in pain has been growing rapidly, which was predominately led by observations and studies performed in clinical settings. Despite the rapid growth of the field, prophylactic treatment of chronic migraine is the only pain indication currently approved based on the outcome of two multi-center phase 3 studies [55] and their combined *post hoc* analysis [56]. Reflecting the recent grade A recommendation of the American Academy of Neurology, BoNT/A should be used in chronic migraine and should not be used (as ineffective) in episodic migraine [57]. Apart from these two migraine indications with clear recommendations on the use of BoNT/A, there are a number of other types of headaches where the evidence is not as clear, as both positive and negative clinical results have been obtained. For example, the effects of BoNT/A in tension-type headache, which is the most common primary headache and pain condition, have been evaluated in several open-label and randomized, placebo-controlled studies. BoNT/A injected into the pericranial muscles every three months during 18 months (in an open-label

study) or only once (in a double-blind study), reduced the severity of headache, reduced pericranial muscle tenderness, and increased the number of headache-free days [58]. In contrast, BoNT/A failed to improve any measures of tension-type headache in two other placebo-controlled, double-blind clinical studies [59,60]. Interestingly, in another double-blind placebo-controlled study, no difference was seen between the BoNT/A and the placebo group on Day 60, but a significant number of patients reported a 50% decrease in headache days at Day 90 [61]. This suggests that, at least with tension-type headache, longer post-treatment periods need to be evaluated in order to see the effect. Similar discrepancies in the efficacy of BoNT in pain across clinical studies have been seen in other indications, such as osteopathic pain [62].

What makes it difficult for a given pain indication to interpret and to compare the results across clinical studies, is the fact that there is no consistency in the experimental variables used, such as sites and routes of injection, number of injections needed, doses, administration regimens (single vs. repeated), etc. Back-translating the human findings into an appropriate animal model with the objective to perform a systematic evaluation of the experimental variables can greatly benefit the field. Furthermore, performing clinical studies in large animals, which show pain conditions that are closer to humans in etiology, can provide additional value. For example, in a placebo-controlled, randomized, double-blind clinical study, intra-articular injection of BoNT/A reduced the joint pain in osteoarthritic dogs [63]. In another randomized, placebo-controlled study in dogs, sub-cutaneous administration of BoNT into the mammary glands 24 h before bilateral, radical, cancer-related mastectomy significantly decreased the post-operative need for rescue morphine analgesia [64]. These findings are well aligned with those showing BoNT/A-mediated pain relief in cancer patients after surgery and/or radiation [65].

Overall, while the diverse pool of clinical studies assessing the efficacy of BoNT in pain is an important source of potential new indications, the questions of where, when, and how to inject require systematic assessment in pre-clinical studies. In addition, pre-clinical studies have been essential in building our understanding of potential mechanisms of BoNT-induced pain relief [28,66,67]. Testing BoNT in disease-relevant animal models, including those in rodents and in larger species, combined with a better understanding of the mechanisms that are involved, can build a solid case for initiation of clinical studies in a new pain or headache indication.

#### *3.2. Osteoarthritis*

Osteoarthritis (OA) is the most common form of arthritis in humans and OA-related joint pain is a major health concern [68]. Since no disease modifying agents for OA exist, the clinical focus is on pain management and minimizing the functional impairment of the joint.

In OA, changes to the joint articular cartilage surfaces and underlying cartilage matrix lead to the loss of joint space and joint misalignment, as well as inflammation of the joint synovium [69,70]. Fissures and fractures in the cartilage surface appear. Changes at the interface between cartilage and bone, the osteochondral junction, allow for the penetration of sensory and sympathetic nerves from the richly innervated bone marrow [70]. The inflammation eventually causes both the peripheral and central sensitization of neurons, leading to spontaneous joint pain at rest and hyperalgesia. Despite progress on the exact mechanism leading to the pain sensation [71], our detailed mechanistic understanding remains sketchy. However, it seems reasonable to assume that release of sensory neuropeptides such as substance P, calcitonin gene-related peptide (CGRP), and neurokinin A, contributes to the pain sensation in OA [62,72]. Since BoNTs can inhibit the release of these peptides, intra-articular BoNT administration may be able to directly reduce peripheral sensitization, and indirectly reduce central sensitization.

From clinical studies, there is however limited evidence that intra-articular BoNT injections have beneficial effects. Hsieh and colleagues [73] studied 46 patients with symptomatic OA knee who were randomly assigned to a BoNT/A treatment group (100 U Botox® into the affected knee) or a control group. The pain visual analogue scale score in the treatment group significantly decreased from the pretreatment value early after treatment (one week) and was still decreased at the six-month post-treatment follow-up, resulting in a significant difference to the control group at both timepoints. Similar findings were obtained with two other evaluation scales, thus showing that intra-articular injection of BoNT/A provided pain relief and improved functional abilities for the patients with OA knee in this study. However, in a recent double-blind, randomized, placebo-controlled, 12-week trial using a single ultrasound-guided intra-articular injection of BoNT/A in 121 patients no significant differences in clinical efficacy parameters were found between Botox® and placebo in the entire population [74]. In another study recently completed in 176 patients with OA knee with Botox®, no statistically significant difference was observed on any of the parameter assessing pain between the onabotulinum toxin and placebo (NCT02230956). Both of the studies add to earlier clinical studies using intra-articular BoNT injections in various rheumatic conditions that were recently reviewed by Khenioui and colleagues [62]. Although the 16 reviewed studies were heterogeneous and had various shortcomings that prohibited the reaching of a generalized conclusion with a satisfactory level of confidence, the reviewing authors did notice that they provided some trend towards the anti-nociceptive effect of intra-articular BoNT/A. Such results show that more research is necessary to understand the mechanism of action and the behaviour of the botulinum toxin when being injected intra-articular. Furthermore, the potential analgesic effects of intra-articular BoNT injections in clinical studies could be explained based on the results of preclinical studies [63,75–78]. New botulinum toxins that are engineered to specifically target some pain receptors and act on some biomarkers could be developed in future as novel therapeutics.

#### *3.3. Overactive Bladder and Neurogenic Detrusor Overactivity*

The bladder is a densely innervated organ, and its functions (storing and discarding urine) are under the synergistic control of the sensory, parasympathetic, and somatic nervous system [79]. Overactive bladder (OAB) is a syndrome where urinary urgency is often accompanied by frequency and nocturia. The mechanisms underlying OAB are still a matter of debate. A 'myogenic' hypothesis proposes that OAB results primarily from detrusor myocyte overexcitability, whereas a more recent, 'urothelium-based', hypothesis highlights the action of the mucosa on afferences sensitization and overactivity. Interestingly, specific cell types, such as urothelial cells [80], and, more recently, telocytes [81], have emerged as pivotal in the development of pathological conditions. Neurogenic detrusor overactivity (NDO) is described in patients with neurologic lesions (e.g., multiple sclerosis, spinal cord injury) where a sacral spinal micturition reflex develops through the activation and the remodelling of C fibres.

Oral antimuscarinic agents are the first line pharmacological therapy for urine storage dysfunction. BoNT started being used off-label in the early 2000s as a second line therapy in OAB [82]. Soon, it became clear that BoNTs action was not limited to its action on parasympathetic structures and acetylcholine release controlling detrusor contractility, but that it was also targeting sensory afferents. Animal models showed that BoNT/A was able to decrease the release of ATP and increase the level of NO [83,84], thus reducing the bladder afferences sensitization, and also decrease CGRP and substance P release, acting on the inflammatory component that may accumulate in bladder dysfunction [85]. Furthermore, clinical evidence highlighted the effect of BoNT/A on nerve growth factor (NGF) release, as well as TRPV1 expression, which are involved in the development of detrusor overactivity [86,87].

Moreover, clinical data suggest that BoNT/B may be efficacious at lower doses in the autonomic nervous system of the human bladder when compared to the somatic nervous system, and this was also shown by experiments in rodents [88,89], see also Figure 1. Therefore, there may be a therapeutic opportunity for treating patients with different serotypes [3], within the limits of species sensitivity, as discussed earlier, in Section 2. SV2/SNAP25 are more abundant in cholinergic as compared to sensory fibres in the human bladder [90], which may explain some cases of urinary retention in patients being treated by intradetrusor injections of BoNT/A. Intrathecal administration of BoNT/A in spinal cord-injured rats, the prototypical animal model of neurogenic detrusor overactivity, resulted in normalization of pathological bladder contractions and bladder basal pressure [91]. This may open

research opportunities for natural or engineered neurotoxins that are targeting the afferent and pivotal component of the disease with a different therapeutic profile compared to BoNT/A.

**Figure 1.** Whatever the mouse strain, BoNT/B1 is significantly more potent than BoNT/A1 in a model of detrusor contractions (**A**,**C**), while the potencies of BoNT/A1 and BoNT/B1 were equal in the phrenic nerve-hemidiaphragm preparation (**B**,**D**).\* *p* < 0.05 (unpaired Student's *t* test) [89].

#### *3.4. Wound Healing*

The healing of a wound comprises four overlapping phases: haemostasis, inflammation, tissue proliferation, and remodeling (for review see [92]). Increased metabolic activity and inflammation during healing can induce muscle contractions around the edges of the wound, which in turn, creates repetitive tension on the wound [93]. This has the potential to not only delay the healing of the wound, but also to increase fibrosis and induce or aggravate hypertrophic scarring. Chemoimmobilisation of the musculature adjacent and under a surgical wound through injection of BoNT could thus be beneficial for the overall wound healing process and the cosmetic appearance of the healed wound.

Indeed, there is clinical evidence that BoNT/A may have beneficial effects in wound healing. For example, using a visual analogue scale in patients who had undergone surgery for facial wounds, Ziade and colleagues [94] found a statistically significant improvement in scarring of the tissue in those patients who had been injected with BoNT within three days post-surgery, compared to patients not injected with BoNT. However, no statistically significant differences were found between the two groups in this study when using other assessment methods. Similarly, a systematic review of the literature for the use of BoNTs for the prevention of hypertrophic scars that included ten clinical studies, found improved cosmetic outcomes among certain studies. However, due to the heterogeneity of the studies, as well as other factors, the authors concluded that the data were not supportive of the clinical usage of BoNT for this indication at this time, but recommended randomized controlled trials to reach a firm conclusion [95]. At the time of writing, April 2018, two randomized controlled clinical trials are currently ongoing (NCT02623829 and NCT02886988).

While the clinical data are thus suggestive, but presently are not yet conclusive, preclinical experimental work has provided evidence that BoNTs do have beneficial effects in wound healing. For example, using a study design that allowed for each animal to be its own control, Lee and

colleagues [96] showed significant differences in wound size between BoNT-treated and untreated control wounds. The treated wounds also showed less infiltration of inflammatory cells, a smaller number of fibroblasts, less fibrosis, and a lower expression of transforming growth factor (TGF)-β1, as compared to the control wounds. TGF-β1 is a fibrotic cytokine that has pleiotropic actions in wound healing [97], and it is involved in the formation of hypertrophic scars. The finding by Lee and colleagues [96] that BoNT-treated wounds show a lower expression of TGF-β1 may be the result of the chemoimmobilisation of the muscle. However, BoNTs may also directly interfere with the expression of TGF-β1 in fibroblasts and fibroblast proliferation [98,99].

#### **4. BoNT Effects Remote from the Site of Injection**

#### *4.1. Neuronal Retrograde Transport and Central Effects*

The vast majority of clinical effects that are exhibited by BoNTs are attributed to their wellestablished actions at the site of injection [100–105]. However, in some instance, the in situ activity of BoNT seems insufficient to fully explain its clinical efficacy, thus raising the idea that other "non-classical" actions on the central nervous system might also be involved [51,106,107]. Indeed, some data have shown that peripherally injected BoNT triggers distant changes at various central levels.

For example, the involvement of cortical networks after BoNT/A administration has been reported. Patients suffering from cervical dystonia were shown to display a higher neuronal excitability (measurement of P22/N30 cortical component of median nerve somatosensory evoked potentials) and reduced gray matter volume that was assessed by magnetic resonance imaging (MRI) in discrete cortex areas [108,109]. Local injections of BoNT/A in cervical muscles had long-lasting beneficial effects on these central alterations.

Some studies also assessed the effect of BoNT/A peripheral muscular injections on brainstem activity in patients with blepharospasm [110] or dysphonia [111]. In the first study, unilateral BoNT/A injections in the *orbicularis oculi* decreased in a strong and similar fashion the blink reflex after the stimulation of supra-orbital nerve in both ipsilateral and contralateral muscles. On the other hand, the activity of injected muscle measured by electromyography was not inhibited in the same proportion as the latter parameter. According to the authors, the beneficial action of BoNT/A in blepharospasm is thus more likely due to changes in brainstem interneuronal pathways than local muscular action [110]. The same conclusions were reached by Bielamowicz and Ludlow [111], who observed after unilateral BoNT/A administration in *thyroaritenoid* muscles decreased activation levels and spasmodic bursts that were measured by electromyography in both injected and contralateral muscles.

Central changes at the spinal cord level were also demonstrated. Marchand-Pauvert, Aymard [112] investigated, in patients exhibiting lower limb spasticity, the recurrent inhibition from a BoNT/ A-injected muscle (*soleus*) to a distant untreated muscle (*quadriceps femoris*). After stimulation of the tibialis nerve that contains motoneurons innervating the *soleus*, the recurrent inhibition was found to be decreased in comparison with measurements that were recorded before treatment. The authors hypothesized that this effect was unlikely due to the local action of the neurotoxin in the injected muscle or on spindle afferent input, but is more likely due to a modification of spinal synaptic transmission. Indeed, they suggested that BoNT/A underwent retrograde neuronal transport from the injected muscle to the spinal cord and locally reduced the stimulation of motoneuron terminals on Renshaw cells, inhibitory interneurons projecting to motoneurons innervating the *quadriceps* muscle. This probably led to the decreased recurrent inhibition of the latter muscle [112]. A similar study obtained comparable findings [113], providing more evidence for a central action of BoNT in humans as a result of a putative neuronal retrograde transport.

Cellular and electrophysiological alterations in Renshaw cells were also found in rats that were injected peripherally with neurotoxins [114,115], indicating that such a phenomenon could also exist in animals. Antonucci, Rossi [21] demonstrated that BoNT/A injections in the hippocampus, whisker pads, or optic tectum led to the appearance of cleaved SNAP25 in the contralateral hippocampus, in the neuropil surrounding facial neurons soma, and in the retina, respectively. Interestingly, in the latter pathway, pretreatment with colchicine (which inhibits microtubule-dependent retrograde transport) abolished the appearance of cleaved SNAP25 in the retina. These first sets of data suggest that either an active form of BoNT/A or cleaved SNAP25 itself could have been transported from nerve terminals to neuron soma. The authors tried to answer that question by injecting BoNT/A in the optic tectum of rats, and, three days later, concomitantly sectioning the optic nerve and injecting BoNT/E in the eye vitreous. As expected, BoNT/E rapidly eliminated SNAP25 cleaved by BoNT/A from the eye. However, 25 days later, the appearance of SNAP25 cleaved by BoNT/A was observed, suggesting the presence of active BoNT/A in the same region [21].

Similar results were found in various in vivo studies, with the appearance of cleaved SNAP25 at the spinal cord level following peripheral BoNT/A injections in rat or mice hindlimb (Figure 2) [52,116–118], and in the brainstem after BoNT/A administration in the whisker pad [119–121] or in the optic tectum [122]. However, direct evidence for a retrograde transport of BoNT would be to detect an active form (either holotoxin or light chain) in central regions after peripheral administration. To our knowledge, this has never been demonstrated, although a work from Wang, Martin [123] showed the presence of fluorescently tagged BoNT/A heavy chain (thus not active) in mice spinal cord after hindlimb injection.

**Figure 2.** Immunohistochemistry staining of cleaved SNAP25 in rat lumbar spinal cord after BoNT/A injection in the left gastrocnemius muscle. Intense staining was found in the ipsilateral ventral horn neuropil, and only traces were found in the contralateral side (SL, VM, unpublished data).

Data from in vitro experiments shed some light on this question. Indeed, numerous works were able to demonstrate, in motoneuron or hippocampal neuron cultures, retrograde transport of either BoNT/A holotoxin or its heavy chain domain [49,120,123–125]. Interesting to note, heavy chains of both BoNT/A and tetanus toxin were found to share the same retrograde carriers when being retrogradely transported [120]. More precisely, BoNT/A heavy chain is thought to be internalized in nerve terminals within a specific pool of non-recycling synaptic vesicles [124], and undergo axonal retrograde transport in autophagosomes that finally fuse to lysosomes at the soma level [123]. In an elegant study, Bomba-Warczak, Vevea [49] used microfluidic chambers to demonstrate that application at the soma side of anti-BoNT/A antibodies or heavy chain competing with holotoxin prevents the appearance of cleaved SNAP25 in this compartment following BoNT/A application at the axon side. These data are thus in favor of a retrograde transport of BoNT/A holotoxin to the soma of the primary neuron, which is then exposed to the extracellular medium before entering the secondary neuron, leading to distal effects [49].

In vitro data are thus suggestive of retrograde transport of BoNT/A besides its classical mechanism of action, underlining some clinical observations. However, the axonal transport of a functional form of the toxin still needs to be clearly demonstrated in vivo. Apart from BoNT/A, little is known about the potential retrograde transport of other BoNT serotypes. Some data support the existence of such phenomenon for BoNT/E, but apparently, to a lesser extent than for BoNT/A [21,120,125]. BoNT/B [126] and BoNT/D [49] were also shown to undergo retrograde transport.

#### *4.2. Diffusion within the Injected Muscle, Local Spread and Contributing Factors*

When injected in a tissue, BoNTs present with a remarkable safety profile, mostly remaining within the injected organ, allowing for their use for various medical applications, as indicated in various reviews and meta-analyses [127–129]. While the diffusion of the injected toxin is somehow expected within the injected muscle or in very close neighboring muscles, local spread can be observed beyond the site of injection leading to different unexpected effects, depending of the targeted area. Local adverse effects that are associated with the intramuscular use of BoNT are thus due to excessive local muscle weakness and leakage to nearby muscles, and are reported as leading potentially to severe adverse events [130]. When used for the treatment of blepharospasm the most frequently reported adverse reactions were eyelid ptosis (21%), superficial punctate keratitis (6%), and dry eye (6%) for Botox® and eyelid ptosis (19%), dry eye (16%), and dry mouth (16%) for Xeomin® (for further details see prescribing information for both products, FDA). When used for the treatment of cervical dystonia, the most frequently reported adverse reactions were dysphagia (19%), upper respiratory infection (12%), neck pain (11%), and headache (11%) for Botox®; muscular weakness (16%), dysphagia (15%), dry mouth (13%), and injection site discomfort (13%) for Dysport®; and, dysphagia (18%), neck pain (15%), and muscle weakness (11%) for Xeomin® (for further details see prescribing information, FDA). Theses adverse effects suggest that unbound toxin moves away from the muscle, most likely through the extracellular space and determined by a concentration gradient. The dose, volume, and injection techniques that are used to target the different muscles are important to limit the occurrence of such adverse reactions.

The evaluation of local spread has been conducted in humans and animal models using electrophysiology and/or tissue markers, such as cleaved SNAP25 or other surrogate markers (for example, see Figure 3). Using electrophysiology and periodic acid-Schiff stain to identify glycogen content in muscle fibers, it was estimated that BoNT/A can pass through the *tibialis* muscle fascia (thick fibrous tissue covering the muscle), and the presence of the *tibialis* fascia can reduce the local spread by 20% in the rat [131]. This local spread was also confirmed in larger species, such as in the cat [132], and primates [133]. In rabbits that were injected with 2.5 U/kg of BoNT/A (unknown origin) by i.m. route in the *longissimus dorsalis* muscle, the intra-muscular diffusion of the toxin from the injection site was reported over a distance of 3 to 4.5 cm, and local spread across facial planes identified over a distance of 1.5 to 2.5 cm with the same intensity as within the injected muscle [134]. In the mouse, N-CAM (a marker that is expressed in denervated muscles) was strongly expressed in the myofiber of the injected *tibialis* muscle, but limited expression was detected in the ipsilateral *gastrocnemius* and *quadriceps femoris* muscles of the injected leg, suggesting a limited local spread in this model using three different BoNT/A formulations (Dysport®, Botox®, and Xeomin®) [135].

Regarding the effect of injection volume, little is known and the literature does not support a substantial influence on local spread. In humans treated for dynamic forehead lines at relatively low doses of BoNT/A (5 U Botox® /injection site) the area affected was slightly greater with an injected volume of 250 μL when compared to a volume of 50 μL; the increase in local diffusion was estimated to be around 1.5 times higher (+50%) for a five times larger injection volume [136]. Similarly, in humans, doubling the concentration (same dose in half the injection volume) resulted in significantly greater effects on the injected muscle, as assessed by CMAP amplitude evaluation (effect on the intra-muscular diffusion). However, effects in nearby muscles were only slightly dependent on the injected volume, with limited differences between the groups [137].

**Figure 3.** Immunohistochemistry staining of cleaved SNAP25 in rat muscle injected with BoNT/A (SL, VM, unpublished data).

#### *4.3. Systemic Spread and Contributing Factors*

Beyond local spread, leakage from the injection site may occur, resulting in a hematogenous spread and systemic exposure, leading to symptoms that are consistent with the mechanism of action of botulinum toxin [138]. Hematogenous spread may result in signs of asthenia, generalized muscle weakness, diplopia, blurred vision, ptosis, dysphagia, dysphonia, dysarthria, urinary incontinence, and breathing difficulties. These symptoms have been reported hours to weeks after injection. Extremely rare cases of swallowing and breathing difficulties have been reported [138–140], as indicated in all the commercially available BoNT labels. These observations are extremely rare and BoNTs present with a very good benefit/risk balance [128,129].

Various factors may affect the fate the injected toxin and its effect in the body, such as volume/concentration, amount of toxin, formulation, needle size, number of injections, precision of injection, and the biologic properties of the injected toxin (mechanism of action, duration of action, half-life in tissues), as well as the anatomy of the target area and tissue type [138,141].

BoNT serotype is also an important contributing factor affecting systemic spread. There was potentially greater spread to nearby and remote non-injected muscles that are associated with BoNT/A when compared to BoNT/B at equivalent efficacious doses in the mouse and primate using the CMAP test [134]. Similar observations were reported in the mouse, in which i.m. LD50 was 13.9 U/kg for BoNT/A (Botox®) and 104.6 U/kg for BoNT/B (Myobloc®), while safety margins (LD50/ED50) were lower for BoNT/B (5.4) than BoNT/A (13.9) [142], suggesting that BoNT/B is less safe. Other factors probably influence spread and its clinical consequences (e.g., toxin half-life in tissues and blood, distribution of receptors, co-receptors, and SNARE proteins in the targeted physiological system [muscular or autonomic], etc.). This was confirmed by a Japanese group using the CMAP test in the injected and contralateral limb muscle in the rat to compare toxins at equivalent efficacious doses [143]. They determined that BoNT/CD is the most prone to spread remotely, and that BoNT/D is least

prone to spread. The rank order of BoNT serotypes based upon spread in this study in rats was CD > A > E > C > B > F > D. Interestingly, when ranked based upon safety index a different order was obtained: F > C > D > E > A > CD > B, with BoNT/F being the safest. Recent data obtained using CMAP amplitude and/or cleaved SNAP25 immunohistochemistry in remote muscles, also support the dose dependent systemic spread effect in rats that were injected with BoNT/A [116,117].

#### *4.4. New BoNT Therapeutics with Decreased Spread*

Spread could be an important differentiating factor in the BoNT therapeutic landscape. BoNT/A3 [144] and BoNT/A2 [38] are two subtypes that may offer reduced spread. BoNT/A2 has been reported as showing less spread to neighbouring muscles than BoNT/A1 [38]. In a study with healthy volunteers, the effect of 6.5 U BoNT/A2 and 10 U BoNT/A1 that is injected into the *extensor digitorum brevis* muscle on either side of the same subject were compared [37]. Although a similar onset, duration, and magnitude of the effect was seen for the two toxins, there was less spread of the effect to the neighbouring *abductor hallucis* muscle for BoNT/A2 compared to BoNT/A1. In a randomized double-blinded controlled phase 2/3 study in post-stroke spasticity, hand grip strength on the unaffected side was significantly reduced after BoNT/A1 treatment, but not after BoNT/A2, which is again consistent with less spread of A2 when compared to A1. In a later study, BoNT/A2 was reported to have reduced systemic toxicity, which is in agreement with the reduced spread that was reported in the clinical studies above [145], and also reduced anterograde transport compared to BoNT/A1, as measured by muscle contraction and immunocytochemical analysis [31,117]. However, despite these encouraging results, there is much lower number of safety data in humans for BoNT/A2 as compared to BoNT/A1, and as such, more human exposure to BoNT/A2 will be needed prior to establishing BoNT/A2 as a differentiated therapeutic. The biological properties and potential clinical utility of BoNT/A2 are discussed more fully in Section 2 above.

#### **5. Novel Formulations**

#### *5.1. Current Products and Challenges Associated with Formulation of BoNTs*

The BoNT formulations for the main commercially available products have existed for over 20 years. The formulation of BoNTs is challenging, owing to their structural complexities, the analytical tools that are available, and the low product concentrations used. The extremely high potency and selectivity of BoNT for cholinergic neurons, with a human lethal dose of parenteral injection estimated as between 0.1–1 ng/kg [146], requires that pharmaceutical preparations of BoNT comprise vanishingly small quantities of the toxin itself. Drug products have been shown to contain between 0.44 and 3.24 ng of the 150 kDa toxin per vial [8], and as such, present unique challenges for the formulation and the analytical strategy employed to assess stability. BoNTs in this respect differ from conventional biologics where much of the focus of formulation is on controlling protein aggregation [147]. One of the primary challenges in the formulation of BoNTs is minimising loss during manufacture and storage. BoNTs are also very susceptible to denaturation due to surface denaturation, heat, and alkaline conditions. Due to these contributing factors, the lyophilisation or freeze-drying of the toxin complex or the neurotoxin is the favoured method of distributing the product in a form that is stable and readily used by the clinician [9]. All BoNT/A clinical products currently available, apart from in Japan, Korea and Taipei, are dried powder products that need to be reconstituted for use. The BoNT/B product has been provided as a liquid, which is ready for use, since its license approval in December 2000. This product may be painful when injected, however, due to the acidic pH of the solution. Medytox was the first BoNT/A liquid product launched, Innotox®, but limited to Korea, Taipei, and Japan, as of today. Allergan licenced worldwide rights to Innotox outside Korea from Medytox. Most of the major clinical BoNT/A manufacturers are working to develop liquid formulations. Liquid formulation is considered the next generation of toxin products because of its convenience and ease of administration. Recently, Ipsen has reported a phase 2 clinical trial of a ready-to-use liquid formulation

of abobotulinumtoxinA (abobotulinumtoxinA solution for injection, ASI). This formulation was shown to be efficacious, with comparable results to reconstituted abobotulinumtoxinA, and to have a favorable safety profile in subjects with severe to moderate glabellar lines [148]. Another feature of all the existing product formulations is the presence of human serum albumin (HSA) as a stabilising agent. This is not desirable due to the theoretical risk of contamination with pathogens. As part of the development of next generation liquid formulations, companies are exploring formulations that have different stabilisers and components, including the omission of excipients from animal origin and HSA.

#### *5.2. Slow-Release Formulations*

Advanced formulations for BoNT include the slow or sustained release formulations, which aim to release BoNTs over a defined period in a controlled manner. This may be through colloidal carrier technology and/or exploiting the thermo-responsive nature of polymeric solutions. Depending on the clinical application, such formulations may provide a benefit to the existing delivery methods that are used. The clinical use of a slow-release formulation or depot requires careful consideration. In many indications, the accuracy of injection is a critical factor during treatment, with skilled injectors depositing the BoNT at the neuromuscular junctions to minimise the spread to adjacent muscles. In many treatments, this is performed in multiple locations within a muscle to achieve the desired outcome. A successful slow-release depot would, therefore, have to remain in situ without migration for the period of release. The duration of BoNT therapeutic benefit in movement disorders is typically 3–4 months [149]. Currently, the sustained release profile for most clinically applicable depot formulations is no more than approximately seven days, which would not produce a significant improvement in duration for BoNTs. For a significant improvement in the duration of BoNT effects, release over 3–4 weeks is likely to be necessary. The challenge of formulating to achieve the desired release profile is therefore much greater and further compounded by the additional complexity of needing to stabilise a large and complex protein at 37 ◦C for this duration. Also, any depot would need to be sufficiently small that the lasting presence is not uncomfortable for the patient.

#### *5.3. Analytical Challenges*

If a slow-release formulation were to be developed, it would be desirable to demonstrate the release profile of the drug over time, both in vitro and in vivo. The concentration at the point of administration, however, is already in the picomolar range, a level at which most pharmacokinetic studies are performed. It seems likely that the released material from a slow-release depot would be at even lower concentration, and over a greater time. Accurate determination of this profile is important for a product possessing such high potency as BoNT [146]. This increases the analytical challenge in preparing such a formulation, ensuring the accuracy of the final dose, and developing a consistent release profile. The absence of circulating biomarkers for the intoxication of neurons by the toxin further increases the complexity, as the pharmacological measure is only obtained from measurement of the local paralysis, electrophysiology, or other functional responses. With recent advances in analytical sensitivities [150,151], the accurate detection of BoNT at the required concentrations is beginning to become achievable, but the technologies are still not in routine use. This analytical improvement will be vital for development of slow-release formulations of BoNTs.

#### **6. Novel Delivery Methods**

Currently clinical BoNT products are delivered by local injection into the target tissue. This can be a painful process and requires clinical expertise on the part of the administrating physician to ensure the correct local delivery and avoid the application to adjacent tissues resulting in a lack of therapeutic efficacy and/or off-target effects. For conditions such as hyperhidrosis, application can require multiple subcutaneous injections in sensitive locations, such as the palm or axilla, resulting in discomfort or even pain for the patient. Injection in such situations can require local analgesia or pain block. In bladder applications, where multiple injections via cystoscopy are required, this can require local or even general anaesthesia. Other limitations to the injection of BoNT products include erythema, swelling, and potential for infection. In aesthetic use in facial areas, there is also the risk of bruising and/or bleeding. Bleeding can also be an issue for injections of BoNT into the bladder urothelium. Development of needle-free delivery would therefore be of great benefit in such circumstances, requiring less specialised clinician input, increased convenience, and reduced discomfort/pain for the patient. Non-invasive delivery of BoNTs is challenging owing to the biological barriers limiting successful access to the intended tissue and the large size of the protein. Approaches under development involve either a physical or chemical approach to transiently permeabilise the barrier and enable entry of the BoNT. To date, all of the needle-free approaches in development are aimed at the relatively superficial delivery and not replacing intramuscular injection to deeper muscles involved in many of the movement disorders where BoNT is used therapeutically.

#### *6.1. Transdermal Delivery*

Development of transdermal drug delivery methods has a long history and has already made a significant contribution to medical practice [152], although largely for small molecule drugs to date. A number of investigators have studied the ability of enhancers to increase skin permeability and enable the transdermal entry of BoNT. One approach that has received a lot of attention is the use of Cell Penetrating Peptides (CPP), and particularly ones that are derived from the transactivator of transcription (TAT) protein of human immunodeficiency virus. Revance Therapeutics Inc. have progressed this approach for transdermal delivery of BoNT/A to the clinic for both lateral canthal lines and axillary hyperhidrosis. The neurotoxin construct that was progressed by Revance, RT001, consists of a purified 150 kDa BoNT/A protein formulated in a poloxamer gel containing a CPP, RTP004, a single chain peptide consisting of 35 L-amino acids. RTP004 has two domains, a core sequence of 15 lysines that confer positive charge to the peptide, and a protein transduction domain that is derived from residues 49–57 of the TAT protein [153]. The CPP associates with the neurotoxin protein non-covalently via ionic interactions between cationic regions within the CPP and the negatively charged surface of the neurotoxin protein. The resulting complex enables transcutaneous flux of the neurotoxin via cell to cell micropinocytosis. Varying the CPP was found to vary the delivery characteristics of the payload, enabling the delivery to different depths of penetration; the longer the peptide the deeper the penetration. In RT001, the CPP is formulated to deliver the BoNT/A to the mid-dermis appropriate for treatment of lateral canthal lines and hyperhidrosis. Phase 1 studies with RT001 showed it to be safe and tolerable [154]. Subsequently, a number of phase 2 studies demonstrated RT001 to be safe and effective in the treatment of moderate to severe lateral canthal lines or hyperhidrosis [155–157]. While in 2015, Revance announced that it had commenced dosing patients in a Phase 3 pivotal study to evaluate the safety and efficacy of RT001 for the treatment of lateral canthal lines the study did not achieve its co-primary endpoints, and was discontinued. Revance is now focused on developing an injectable form of BoNT/A combined with its proprietary peptide technology, RT002, which is claimed to spread less from the site of injection and to give a greater safety margin and enable the enhanced duration of therapeutic benefit.

Another approach to transdermal delivery of BoNT/A, which was developed by Anterios Inc., was the use of liposomes of a controlled size range to deliver BoNT/A for hyperhidrosis. Anterios was acquired by Allergan plc in 2016. In 2015, the lead Anterios formulation, ANT-1207, entered a phase 2 trial (NCT02479139) for axillary hyperhidrosis. Allergan have yet to report on the outcome of this study.

Chajchir and colleagues reported on the use of a microcrystal and nanoparticle formulation, InParT (Transdermal Corp), reconstituted with Botox® at 2 U/mL for transdermal delivery in the treatment of upper facial lines [158]. The formulation was found to be effective versus placebo based on patient reported Facial Line Outcome scores. The formulation was applied daily for a period of 4 to 7 weeks, with a 12 week follow up period. The improved outcome scores were maintained during the

whole study period. Given the very low dose of BoNT/A that was employed, this result is surprising, but may reflect the frequent initial dosing regimen.

Transdermal delivery of BoNT/A using a short synthetic peptide, TD-1, as an enhancing agent has been reported to reduce plasma extravasation and blood flow changes that are evoked by electrical stimulation of the hind-paw in a rat model of neurogenic inflammation [159]. Peptide-mediated transdermal delivery of BoNT/A was proposed to offer an easy and non-invasive route of administration for the treatment of neurogenic inflammation.

Perhaps the most unusual and potentially risky uses of CPP to achieve transdermal delivery of BoNT was the direct fusion of the TAT peptide to the light chain of BoNT/A [160,161]. This was reported to enable the direct application and delivery of the light chain into cells without the need for any of the other domains of BoNT that mediate neuronal specific binding and the translocation of the light chain into the cytosol [160]. Transcutaneous delivery through mouse skin was also evaluated by immunohistochemistry. The authors discuss the therapeutic potential of such an approach, but it did not consider the lack of selective targeting or the potential for non-specific entry into any cell that was exposed to the fusion protein. Such a lack of cell selectivity must significantly increase the risk of non-specific toxicity and represent a significant safety barrier to therapeutic application of such a fusion protein.

The requirement for an enhancer to enable the transdermal delivery of BoNT is demonstrated by the inability of BoNT/A applied directly to the skin to impact palmar hyperhidrosis [162]. Without any enhancer to allow for the entry of the neurotoxin across the dermal barrier no effect above control was observed, clearly showing the inability of a large multi-domain protein, like BoNT, to cross the dermal layer unaided.

In addition to chemical enhancing agents, the use of physical approaches to disrupt the permeability barrier of the skin and enable entry of BoNT has been explored. Kavanagh and Shams reported the use of iontophoresis to deliver BoNT/A for the treatment of palmar hyperhidrosis, with an improvement in the sweat rate of 66%, which is a similar result to that obtained with intradermal BoNT/A [163]. Subsequently, Pacini and colleagues confirmed in rats that pulsed current iontophoresis delivered BoNT/A through living skin [164]. Following iontophoresis, BoNT/A was clearly detected in association with cutaneous striated skeletal muscle fibres in the deep dermis of the rat skin. Transdermal delivery by jet nebulization of BoNT/A in combination with lidocaine has been reported to be an effective and painless method for the successful treatment of palmar, plantar, and axillary hyperhidrosis [165]. An emerging technology for transdermal drug delivery is the use of microneedles [166], and the use of microneedles for the delivery of BoNT/A into the skin has been reported [167]. As yet, however, there are surprisingly few publications on this route of transdermal delivery for neurotoxin.

#### *6.2. Transurothelial Delivery*

The other major application of BoNT to have received considerable attention in regards to needle-free delivery is the use in the bladder for treatment of OAB and NDO. The current procedure is invasive, requiring cystoscopic guidance and multiple injections into the bladder, and often necessitating intravenous sedation or anaesthesia. If a less invasive, needle-free method of administration could be demonstrated to be effective, this would be preferable. There is also evidence that whilst injection of BoNT into the bladder wall impacts both the afferent and efferent nerves, approaches that avoid injection impact only afferent nerves, which may be therapeutically advantageous [168].

Intraluminal instillation of BoNT/A without injection failed to deliver toxin into the bladder wall due to the impermeability of the bladder urothelium to large macromolecules and degradation of the toxin by urine proteases [169,170]. In order to enable the entry of BoNT into the bladder wall, it is necessary to use chemical agents to denude the urothelium. One such approach is the use of the organic solvent dimethyl sulphoxide (DMSO) [170]. The safety and the efficacy of co-administration of BoNT/A with DMSO has been demonstrated in a clinical trial in women with refractory OAB [171]. Protamine

sulphate has also been employed to denude the bladder urothelium and aid access of BoNT/A to the bladder wall [169,172]. Protamine sulphate pre-treatment of bladder prior to the instillation of BoNT/A has been reported to be safe and potentially effective in patients with refractory OAB [173]. Instillation of BoNT/A was reported to reduce the inflammatory effects that are caused by the prior injection of cyclophosphamide in a rat model of cystitis [174]. In this case, it was proposed that the permeability barrier of the bladder urothelium had been impaired by the cyclophosphamide-induced inflammation enabling access of the instilled BoNT/A. Hyaluronan-phosphatidylethanolamine (HA-PE) has been reported to be another potential chemical agent allowing for bladder instillation of BoNT/A to be effective [175]. Based upon SNAP25 cleavage, the instillation of BoNT/A with HA-PE was as effective as intra-detrusor injection in enabling BoNT/A delivery into the bladder.

Inert heat sensitive hydrogel preparations that increase the residence time of neurotoxin adjacent to the urothelium and allow for slow release exposure has been used to enhance the effectiveness of instilled BoNT/A. OnabotulinumtoxinA embedded in the hydrogel TC-3 has shown efficacy in the treatment of OAB [176]. TC-3 is produced by an Israeli company UroGen Pharma Ltd. (Ra'anana, Israel) as RTGelTM. In 2016, UroGen licensed the worldwide rights to develop RTGelTM in combination with BoNT/A to Allergan Inc. (Dublin, Ireland), who are currently conducting phase 2 studies in OAB. Liposomes are being developed as a delivery vehicle enabling intravesical BoNT/A delivery in the bladder by the US biotech Lipella Pharmaceuticals [177]. Bladder instillation of liposome encapsulated BoNT/A was effective at treating the symptoms of OAB in patients [178].

As with transdermal delivery, in the bladder, physical methods to enhance the permeability of the urothelium have been employed as well as chemical. Low energy shock waves have been reported to enable delivery of intravesical BoNT/A into the bladder wall and block acetic acid-induced hyperactivity in a rat model of bladder hyperactivity [179]. The use of intravesical electromotive BoNT/A delivery in the bladder has been widely studied, both in animal models [180] and clinical application [180–183]. Intravesical electromotive BoNT/A has been reported to be effective in women with OAB [181] and in NDO in children [182,184]. In children, a long-term follow-up study has shown it to be a safe and effective method of treating NDO resulting from myelomeningocele [183].

#### *6.3. Transepithelial Delivery*

One other area where needle-free delivery of BoNT/A has been explored is topical intranasal delivery. Successful intranasal delivery of BoNT/A and the relief of symptoms of allergic rhinitis in rats was achieved using the proprietary CPP RT001 formulation [185]. Effective relief of idiopathic rhinitis in some patients has been reported following simple intranasal application of a sponge soaked in BoNT/A solution [186].

#### **7. Conclusions**

BoNTs have proved to be highly effective therapeutic proteins offering symptomatic relief across a wide spectrum of neurological and muscular conditions that are characterized by neuronal hyperactivity. The increasing understanding of the biology of the neurotoxins and the availability of highly differentiated toxin serotypes and subtypes offers the prospect in coming years of expanding this therapeutic benefit and extending it to a greater range of clinical conditions, offering the benefit of being more specific and the efficacy of these unique therapeutic proteins to a much greater number of patients. Whilst botulinum toxins have the advantage of being extremely potent with an extended duration of biological activity, properties that are key to their clinical success, these very characteristics present significant challenges to the scientific field of pharmaceutical development. Furthermore, in the coming years, modified and recombinant toxins would expand even more the future therapeutic utility of BoNTs.

**Author Contributions:** All authors contributed to the writing of one or more sections of the review. E.F. and J.K. co-design the review. E.F. and K.F. coordinated the production of the review.

**Conflicts of Interest:** All authors are Ipsen employees.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Analgesic Effects of Botulinum Toxin in Children with CP**

#### **Josephine Sandahl Michelsen, Gitte Normann and Christian Wong \***

Department of Orthopedics, Hvidovre Hospital, Hvidovre 2650, Denmark;

Josephine.sandahl.michelsen@regionh.dk (J.S.M.); gittenormann@gmail.com (G.N.) **\*** Correspondence: Christian.nai.en.tierp-wong@regionh.dk; Tel.: +45-38626966

Received: 30 March 2018; Accepted: 13 April 2018; Published: 19 April 2018

**Abstract:** Experiencing pain is the greatest contributor to a reduced quality of life in children with cerebral palsy (CP). The presence of pain is quite common (~60%) and increases with age. This leads to missed school days, less participation, and reduced ambulation. Despite these alarming consequences, strategies to relieve the pain are absent and poorly studied. Moreover, it is difficult to evaluate pain in this group of children, especially in cases of children with cognitive deficits, and tools for pain evaluation are often inadequate. Botulinum toxin has been shown to alleviate pain in a variety of disorders and could potentially have an analgesic effect in children with CP as well. Even though most of the studies presented here show promising results, many also have limitations in their methodology as it is unlikely to capture all dimensions of pain in this heterogeneous group using only one assessment tool. In this review, we present a new way of examining the analgesic effect of botulinum toxin in children with CP using a variety of pain scores.

**Keywords:** pain; cerebral palsy; botulinum toxin A

**Key Contribution:** This review describes the causes of pain in children with cerebral palsy; how to evaluate this and the analgesic effect of botulinum toxin; including a description of a prospective study protocol examining the effect of botulinum toxin on muscle-related pain.

#### **1. Pain in Children with Cerebral Palsy**

#### *1.1. Introduction*

Cerebral palsy (CP) is a heterogeneous group of non-progressive neurological disorders caused by damage to either the fetal or the infant brain, affecting the development of posture and movement. CP is the most common cause of physical disability in children, affecting 2.5 children for every 1000 children being born. CP causes variances in the level of motor function, from ambulatory children to nonambulatory children that depend on full-time assistance. Motor function is classified using the Gross Motor Function Classification System (GMFCS), where level 1 describes the most functional group and level 5 describes the least function group. Secondary musculoskeletal problems such as spasticity, muscle deformities, hip dislocation, or scoliosis often occur and can contribute to the pain experienced by this population. Cognitive, perceptional, or communicative disturbances are often present, which can make pain scoring difficult [1].

The pain these children experience includes both chronic pain and pain related to procedures such as physiotherapy (assisted stretching), passive joint movement, botulinum toxin injections, and surgery, with physiotherapy being reported as the cause of the most intense and most frequent pain [2,3]. Recurrent nonprocedural musculoskeletal pain is also a serious and disabling problem and has been reported in 62% of children with CP [4].

Pain is reported to be the most important factor affecting the quality of life and the participation of children with CP, and it is reported to be an even stronger contributor than GMFCS levels [4–11].

In this review, we describe the characteristics of pain in children with CP and the problems with recognizing and grading pain. We also discuss if botulinum toxin can be used to treat pain in this group of patients. As this review focuses on botulinum toxin as a treatment for general musculoskeletal pain, studies on procedural pain will not be discussed in detail.

#### *1.2. Prevalence and Intensifiers*

A European multicenter project (SPARCLE) involving seven European countries began in 2004 to study the prevalence of pain in 818 eight to twelve year old children with CP (SPACLE 1) [4–11]. Of these children, 60% had experienced pain in the last week according to self-reports and 73% according to parental reports. A follow-up study was conducted in 2009 (SPARCLE2) including 594 of the children who had participated in the first study and were now considered adolescents. During the five years of the study, the frequency of pain had increased to 74% and 77%, respectively. The prevalence reported in this study seems consistent with other studies reporting 48–67% [2,3,12–19]. In contrast, Alriksson-Schmidt and Hägglund (2016) included 2777 children and assessed pain among all GMFCS groups, but reported only a prevalence of 32.4% [20]. In contrast to many of the other studies, only children aged 0–14 were included, which easily could have lowered the prevalence as increasing age has been shown to be a contributing factor [2,4,14]. The SPACLE studies actually found that an age of more than 14 years was the only significant predictor for recurrent musculoskeletal pain [4]. Findlay et al. (2015) also showed that an increasing age, together with a presence of pain, negatively affected the quality of life of children [14]. Compared to the general population where approximately every fifth child experiences chronic pain [21,22], there is no doubt that children with CP experience pain more often.

The presence of pain in children with CP has been correlated with missed school days, less participation in activities, and reduced ambulation [23,24], which, combined with the negative impact on their quality of life, make the high prevalence of pain even more alarming.

Despite the high prevalence of pain, strategies to reduce it are inadequate or absent [12,13,25–27]. This indicates that there is a need for more knowledge and awareness, and strategies to reduce the pain need to be optimized.

#### *1.3. Pain Characteristic, Location, and Frequency*

The location, intensity, frequency, and origins (neuropathic, nociceptive) of the pain seem to vary across this heterogeneous population, which contributes to its complexity. Even the factors triggering the pain can vary. Inactivity, physical activity, stress, weather, and sleep have all been described as intensifiers [10,15,28].

The pain that children with CP experience can be both neuropathic and nociceptive in origin. A recent study [27] used quantitative sensory testing (QST) to determine the sensory detection and the pain thresholds in 30 children and adolescents with CP. When compared with the reference values from healthy controls, the children with CP were less sensitive to mechanical and thermal stimuli, but were more sensitive to mechanic pain. This might explain the high prevalence of musculoskeletal pain and pain during assisted stretching. The study proposed that the alternation in sensory detection is caused by a dysfunction in the sensory tract neurons, which makes neuropathic pain possible. This study proposes that an increased sensitivity to mechanic pain could be a contributing factor to the general experience of pain. As a consequence, the treatment of pain must involve strategies aimed to reduce the causes or effects of the (mechanic) pain.

Several secondary problems related to the diagnosis of cerebral palsy have been described by physicians as contributing factors to the pain. This includes hip dislocation, dystonia, muscle spasms, spasticity, deformities, constipation, and abnormal joint compression due to an abnormal gait pattern [12]. Pain seems to be most commonly located in the lower extremities [12,14,15]. The cause and the location of the pain seem to vary across GMFCS levels. The most functional children (GMFCS 1) complain of pain in the lower extremities, especially in the feet as a result of spasticity and muscular deformities, whereas the least functional children (GMFCS V) seem to have more pain in the hip and the abdominal region as a result of hip dislocation and constipation [12,20].

The pain intensity ranges from mild to severe, with most children reporting mild–moderate pain that occurs once or twice a week and does not affect their ability to perform activities [2,10–14]. Approximately one-quarter experience moderate–severe pain on a daily basis and report that the pain affects their ability to perform activities [2,12,14].

Pain intensity and frequency have been shown to correlate with increasing age [2,4,14] and GMFCS level [10,13,16,22] However, the correlation with the GFMCS levels is not consistent throughout all studies [2,14,17,19]. A gender difference has also been found as a trend, where girls experience more pain than boys [12,17], but only one study has shown significant results [20].

#### *1.4. Pain is Unrecognized and Undertreated*

A study from Penner et al. in 2013, where caregivers and a physician independently evaluated the presence of pain, suggested that musculoskeletal pain in particular is missed by physicians, and as a consequence, might be undertreated [13]. The presence of cognitive deficits can make it difficult for these children to report the pain and for the physician to recognize it. This observation is supported by a study from Stallard et al. (2001), where parental pain diaries were used to follow 34 non-communicative and cognitively impaired children during a two-day observation [29]. Even though 25 of the 34 children had pain on at least one of the two days, no-one received active pain management.

Literature describing the degree of pain management in children with CP is sparse, which could reflect an unrecognized and undertreated problem. A recently published study showed that only half of children with CP used specific pain strategies (medication, massage, activity, rest, and coping strategies) to relieve their pain [27], and Russo et al. (2008) have shown that only 16% used analgesia to relieve their pain [15]. The same is seen in the adult population [25,26]. Engel et al. showed in 2002 that even though many pain interventions such as ibuprofen, stretching, exercise, acetaminophen, and massage were perceived as moderately helpful, only a small group of patients were using them [25]. Moreover, a large variation in the efficacy of the interventions has been demonstrated [26], which seems plausible due to the heterogeneity of this patient group. As a consequence, an individual and multilevel pain management seems of great importance.

#### *1.5. Pain Scoring*

Recommendations in pain scoring have pointed out the use of self-reports to score the pain [30,31], and the most common tool used at the hospitals seems to be numeric intensity scales, which are usually the visual analogue scale (VAS) or the faces pain rate scales for smaller children (e.g., Wong–Baker FACES scale, faces pain scale). However, children with neurological deficits might also show atypical pain behavior such as drooling and laughing [3,32,33], which can make it difficult for physicians, caregivers, and parents to recognize the pain. Moreover, some of these children are not able to communicate due to cognitive deficits, which make pain scoring even more difficult. In recent years, several behavioral-based tools for pain scoring (NCCPC, pediatric pain profile and FLACC) have been developed for children with neurological and communitive deficits [30,31,33,34]. Most of these behavioral scoring tools have been modified in a revised edition, making it possible to add individual pain behavior to the scale and thus ensure that the level of pain is captured and correctly scored. Russo et al. (2008) proposed another problem in that some children and parents might see the pain as part of the underlying condition and as something that they have to live with. They might be unaware that the pain can be relieved, which may prevent them from seeking help [15].

Even if proper pain-scoring tools are available in the clinic, physicians have to be aware of the problem, be able to recognize pain behaviors, and to have knowledge of possible strategies to relieve the pain.

One reason for this issue of undertreated pain could be the difficulties in assessing the pain due the heterogeneity of the cognitive impairments related to CP, which makes it impossible to use a single standardized test [30,31].

In conclusion, there is no doubt that pain is an important and neglected issue in children with CP, which, when undertreated, has significant consequences such as missed school days, reduced participation, and a lower quality of life. Part of the problem seems to be the heterogeneity of the pain in its location, intensity, and cause. Pain is perhaps one of the most subjective feelings in the world, and as a consequence, there is a need for sufficient pain-scoring tools that are able to capture all dimensions of the pain independent of any potential cognitive deficits. Sufficient strategies in pain management might be difficult to find due to the complexity of the pain, and therefore, pain warrants individual assessment and treatment.

#### **2. Existing Literature on the Analgesic Effect of Botulinum Toxin**

#### *2.1. The Mechanism of Botulinum Toxin*

The mechanism of botulinum toxin on a cellular level is believed to be the blockage of the release of acetylcholine to the motor endplates, thereby preventing muscular contraction. Acetylcholine is transported in small vesicles and is released from the neuron through exocytosis. Large proteins called snares are connected to the outside of the vesicle and to the membrane of the neuron. Upon exocytosis, the snares perform a complex which enables the fusion and the release of the neurotransmitter. Botulinum toxin prevents the release of acetylcholine by cleaving one or more of the three snares (SNAP-25, Syntaxin, and VAMP) that form the complex. The snares involved in the docking of the vesicle are not specific to the neurotransmitter being transported [35]. Nonclinical studies have shown that botulinum toxin also blocks the release of neurotransmitters (e.g., glutamate, CGRP and substance P) that are involved in pain and inflammatory pathways [35,36]. The analgesic effect of botulinum toxin seems to be multifactorial. Botulinum toxin seems to have the ability to reduce pain indirectly as a muscle relaxant or by blocking local nociceptive neuropeptides involved in the peripheral sensitization caused by inflammation or injury. Besides its local effect, nonclinical studies have also suggested that botulinum toxin can be transported by retrograde axonal transport to the dorsal root, where it can exert a more central analgesic effect [35,36]. This is supported by studies showing bilateral analgesic effects following an ipsilateral injection of botulinum toxin [35]. By preventing the release of these neurotransmitters in the central nervous system, botulinum toxin may relieve neuropathic pain.

This analgesic effect of botulinum toxin is also supported by clinical studies where a painrelieving effect has been observed in various pain disorders [36,37]. An excessive review from Safarpour et al. (2018) found level A evidence (effective) for post-herpetic neuralgia, trigeminal neuralgia, and post-traumatic neuralgia. Level B evidence (probably effective) was found for diabetic neuropathy, plantar fasciitis, piriformis syndrome, pain caused by knee arthroplasty, male pelvic pain syndrome, chronic back pain, and neuropathic pain secondary to traumatic spinal cord injury. Only level C evidence (possibly effective) was observed for female pelvic pain, knee osteoarthritis, and postoperative pain in children with CP [37]. In conclusion, the various ways by which botulinum toxin can exert its effects may support the theory that it has a pain-relieving effect for all causes of pain and not only those related to muscular hyperactivity.

#### *2.2. Botulinum Toxin A for Spasticity-Related Pain*

Studies on other populations with spasticity have shown promising results for botulinum toxin A as a treatment for spasticity-related pain. This has been described thoroughly by B. Jabbari et al. (2015), who examined several studies [38]. Five out of nine studies on upper limb spasticity showed positive results, and in four of the studies, the pain-relieving effect was shown in approximately 90% of the patients. Jabbari et al. proposed the use of invalidated pain scores as the reason for the lack of a measurable effect in the other studies. Lower limb spasticity was also examined by Jabbari, but the

literature was sparse. Only three studies were examined but a reduction in limb pain or the number of painful spasms was seen in each of the studies as a result of the treatment.

Similar positive results were shown by Wissel et al. (2000). They studied the management of spasticity-associated pain with botulinum toxin A injections in a heterogeneous population of 60 patients with spasticity (including nine children with CP), and they reported a pain reduction in 90% of the patients [39]. Recently, they published an even larger double-blinded, randomized study, including 273 stroke patients who confirmed the result, and the reduction was sustained through to week 52 [40]. This suggests a long-lasting analgesic effect on spasticity-related pain, and the same effect may be seen in children with CP.

#### *2.3. Botulinum Toxin A as a Treatment for Pain in Children with CP*

The aforementioned mechanisms for botulinum toxin suggest that botulinum toxin could be a possible treatment for all kinds of pain that children with CP experience. However, the existing literature is sparse.

Botulinum toxin injections have, since the early 1990s, been used to reduce spasticity in children with cerebral palsy, to improve motor function, and to delay the need for surgery [41]. A Norwegian registry-based study (2012) examined the characteristics of children treated with botulinum toxin [42] and found that approximately two-thirds of all children with spastic CP were treated with botulinum toxin to improve their motor function. Pain relief and ease of care were less common indications for treatment. Interestingly, the study also showed that children with severe cognitive deficits were less commonly treated.

Pin et al. (2013) reviewed the evidence for the efficacy of botulinum toxin in children with CP (GMFCS IV and V) [43]. The study examined different parameters (pain reduction, motor function, ease of care, and comfort) and found that the level of methodological quality was weak to moderate in many of the studies. Out of 19 studies, only six studies included pain as an endpoint, with one measuring postoperative pain and five measuring general pain. Two studies [44,45] showed that the treatment had a significant effect, while one study showed a trend towards an analgesic effect, but the result was not significant [46]. These studies will be described in detail later in this review. The last three studies [47–49] were case studies and therefore, no statistical analyses were conducted. A total of six patients with pain were examined in the three case studies, and a positive effect was reported in all of them. The type of pain was painful shoulder luxation, hip pain, and pain during nursing and physiotherapy. A long-lasting effect of 6 and 7 months was reported in two patients [47,49]. Pin et al. concluded that only weak to moderate evidence exists for the effect of botulinum toxin in relieving spasticity-related (general) pain.

The analgesic effect of botulinum toxin in children with cerebral palsy has often only been investigated as a secondary or exploratory endpoint, sometimes only as part of a quality-of-life questionnaire or as an individual treatment goal [46,50,51]. Vles et al. (2008) used the VAS score to study the effect of botulinum toxin injection in regards to individual therapeutic goals [46]. The VAS score ranges from 0–10, where a score of 0 reflects a very satisfactory treatment, while a score of 10 reflects a very dissatisfactory treatment. Pain reduction was set as a treatment goal for four out of 55 children, and three of them reported individual scores of 9.4 to 0, 5.5 to 2.7, and 4.0 to 2.0 pre- and post-treatment. The last one did not find any change after the treatment and had the same score of 7.4 before and after the injection. Despite the positive effect seen in three of the children, no significant effect was found.

The effect on pain after botulinum toxin injections in children with cerebral palsy has, to our knowledge, only been measured as a primary endpoint in three studies [44,45,52]. Barwood et al. (2000) used an observational pain score and the level of analgesic requirements to study the effect of botulinum toxin on postoperative pain after hip adductor release [44]. A 74% reduction in the mean pain score was observed in children receiving botulinum toxin injections as compared to placebo. The group receiving treatment also needed 50% less analgesics than the children receiving a placebo treatment.

Rivard et al. (2009) used parent-proxy ratings to evaluate the analgesic effect of botulinum toxin injections on general pain in children with cerebral palsy [52]. Of the children with CP, 62% did not experience any pain one month post-injection. The assessment was done solely as a telephone interview, and therefore, the results have to be seen with caution. Lundy et al. (2009) used a validated behavioral pain score (the pediatric pain profile) to measure the effect of botulinum toxin in relieving hip pain in nonambulant children with CP [45]. The study showed promising results, with a reduction in hip pain in all participants from baseline to three months post-injection. Out of a maximum score of 60, the mean pain score fell from 42.2 (SD 8.6, range 20–59) at the baseline measurements to 9.5 (SD 5.2, range 1–23) at three months post-injection. The response to the treatment varied with individual reductions in the range of 12–58 points.

Another promising and long-lasting result was reported in a multicentered observational study by Chaléat-Valayer et al. (2011), which examined 282 children. Each child received injections with botulinum toxin based on self-selected therapeutic objectives. Those having an objective related to comfort and pain showed improvements of 80 percent at the end of the study, one year post-injection. Overall, the study also found that significantly fewer children experienced daily and intermittent pain as a result of the treatment [50].

Copeland et al. (2014) found in a double-blinded, randomized control study (RCT) that a significant pain reduction occurred between baseline and follow-up measurements, which was only evident in the group receiving the treatment [51]. However, the pain reduction was not significantly different from the group receiving the placebo treatment. Pain is, as described, an important factor for quality for life. An improvement in quality of life was also evident between the baseline and the follow-up measurements in the treatment group, which was significantly different from the placebo group at 16 weeks post-treatment.

Other RCT studies investigating pain have shown the same conflicting results, where patients respond positively to both the placebo and the treatment with botulinum toxin [38]. This may reflect a high placebo effect, which plausibly could be interrelated to the use of only one, highly subjective pain tool. This warrants a more strenuous methodological approach to evaluating pain in future studies.

Even though most of the studies presented here show promising results for botulinum toxin as a treatment for pain, many of the studies has limitations in regards to their methodology. This includes the absence of a control group, the exclusion of some GMFCS levels, the use of proxy reports, and the measurement of the effect of pain through the use of only one subjective measurement or as an individual therapeutic goal.

The absence of sufficient evidence for the efficacy of botulinum toxin as a treatment for pain in children with CP could also be caused by a lack of a proper evaluation of the pain as a result of the use of different pain-scoring tools. For such a heterogeneous group as these children, it is unlikely that all dimensions of pain will be captured using only one assessment tool.

In conclusion, further investigation into this matter is needed to establish a pain-relieving effect of botulinum toxin. This should include a detailed description of the type, the location, the intensity, and the frequency of pain, as well as the degree of cognitive deficits in this heterogeneous population, a more thorough examination of the pain, and the inclusion of a broader spectrum of this patient population.

#### **3. Botulinum Toxin A as a Treatment for Spasticity-Related Pain**

#### *3.1. A Novel Approach—The Use of a Variety of Pain Scores*

We are commencing a prospective multicenter study investigating the effect of a single injection of abobotulinumtoxinA (Dysport) on the pain and the quality of life in children with CP. Based on the parents' and patients' feedback after botulinum toxin treatment, we are hopeful that botulinum toxin could be a possible pain treatment for children with cerebral palsy. The process by which botulinum toxin alleviates pain and whether botulinum toxin injected locally in the muscles could have more direct effect on pain and inflammatory pathways is still not fully understood. Only children suffering from spasticity-related pain in the lower extremities will therefore be included in this study. This will be measured by the presence of a pain response (r-FLACC) and increased muscle tone during passive joint movement. If a painful reaction in a muscle is being detected, the muscle will be a target for treatment. However, the children must experience at least moderate pain (4 on the r-FLACC) to be considered for this study. In contrast to the previous studies already mentioned in this review, we will focus on a more thorough methodological approach to investigate the effect of the botulinum toxin on pain. The level of pain will be measured pre-injection and 4, 12, and 28 weeks post-injection, which will enable us to measure a possible long-lasting effect. To capture as many dimensions of the children's pain experiences, we will use a variety of pain scores. The revised FLACC during passive range of motion will also be used to measure localized pain in the muscles. The pain scoring will be done blinded using a video recording of the session. The pediatric pain profile previous used by Lundy et al. (2009) [45] will be used to investigate both the characteristics of the pain and to evaluate the level of daily pain. All communicative children will be asked to evaluate the pain on an intensity scale using VAS or Wong-Baker, since a self-reported rating is the gold standard for pain evaluation, but as mentioned, it not always applicable in this patient population. For the non-communicative children, proxy reports by the parents will therefore be utilized in the pediatric pain profile. In general, one observer will perform all pain scoring measurements to ensure a uniform and standardized evaluation. In this study, we will include children with CP belonging to all GMFCS levels for a comprehensive analysis of this heterogeneous group of disabled children.

Pain has a negative impact on quality of life as previously described. We will therefore monitor the quality of life with questionnaires (CPchild and CPQOL) throughout the study period. Recommendations state that it is essential to define a realistic, therapeutic, and measurable goal for the treatment [41]. A goal will therefore be set for each child describing a desirable pain-related effect of the treatment. It could be fewer awakenings during the night or to be able to perform activities previously described as painful. The goal attainment scale will be used to monitor the progress at each follow-up (4, 12, 28 weeks post-treatment).

It is still not clear whether a decrease in spasticity will relieve the pain or if other mechanisms are involved. Spasticity measurements (Modified Ashworth Scale and Tardieu scale) will therefore be performed throughout the study period by the same rater. This might enable us to correlate the pain with the presence of spasticity even though some criticism has been reported regarding the reliability of these measurements [53].

In conclusion, we hope to be able to perform a comprehensive study where we examine the possible effects of muscular botulinum toxin injections in children with cerebral palsy, with a focus on spasticity-related pain. It would have been preferable but ultimately not feasible to include a placebo treatment due to ethical considerations which prevent the use of sham injections, which (in principle) are without effect and painful to perform.

#### *3.2. Botulinum Toxin Injection—Limitations and Safety*

Botulinum toxin injections seem to have a pain-relieving effect but have limitations, as with all other treatments. Botulinum toxin injections have been considered as a safe treatment for spasticity in children with CP, but minor side effects like pain at the injection site, urinary incontinence, and influenza-like symptoms have been observed [54]. The injections are considered painful for many of the children and some children will thus be offered general or local anesthesia before the injection. Even though general and local anesthetics are generally considered safe, they are associated with an elevated risk of adverse events. In addition, the development of antibodies against the toxin can occur in some children, thereby preventing an effective response to the treatment [55], and there is always a chance that the toxin will spread and cause a partial, unintended paralysis and weakness in the muscles. Another limitation is the duration of the treatment. Nonclinical studies have shown that the motor endplate recovers after approximately three months [56], and even though

the therapeutic effect has been observed to persist for up to several months [40,47,49,50], a strategy involving repeated injections is necessary for a sustained effect. Delgado et al. examined in 2017 the safety and efficacy of repeated injections in 216 children with cerebral palsy, and they did not find any reason to suspect a high risk for adverse events, and the clinical improvements were also sustained through repeated treatment cycles [57]. The study confirmed a long-lasting therapeutic effect as one out of five patients showed an effect duration of at least 28 weeks. The main indication for treating with botulinum toxin has been the presence of spasticity, but as spasticity can develop into structural contractures over time, some children will be unfit for the treatment with increasing age. Botulinum toxin has been considered useful for other pain conditions, and hypotheses have been made regarding the direct effects of botulinum toxin on pain and the inflammatory effect, as mentioned previously. Pain due to contractures, deformities, simple muscle overload or other causes unrelated to spasticity in children with CP might therefore be alleviated with botulinum toxin, but it should only be considered in moderate and severe pain if less invasive pharmacological and nonpharmacological strategies are ineffective [58].As injection treatment requires accurate skills and is time-consuming, expensive, invasive, and has possible side effects, it is important to weigh these limitations and concerns against the possible beneficial effects of the treatment prior to commencing the treatment.

**Acknowledgments:** Funds to cover the costs to publish in this open access comes from a pending study sponsored by Ipsen A/B.

**Conflicts of Interest:** First author is conducting the above described study and last author is supervising the study. Second author has no conflict of interest. Ipsen A/B is sponsoring this study, but has not been taking part in the creation of this article. The founding sponsors commented on the above described pending study in the design of the study, and will be allowed to comment on manuscripts coming from this study before submission. The founding sponsors will have no role in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Lumbar Sympathetic Block with Botulinum Toxin Type A and Type B for the Complex Regional Pain Syndrome**

**Yongki Lee 1, Chul Joong Lee 2, Eunjoo Choi 1, Pyung Bok Lee 1, Ho-Jin Lee <sup>1</sup> and Francis Sahngun Nahm 1,\***


Received: 8 March 2018; Accepted: 16 April 2018; Published: 19 April 2018

**Abstract:** A lumbar sympathetic ganglion block (LSB) is a therapeutic method for complex regional pain syndrome (CRPS) affecting the lower limbs. Recently, LSB with botulinum toxin type A and B was introduced as a novel method to achieve longer duration of analgesia. In this study, we compared the botulinum toxin type A (BTA) with botulinum toxin type B (BTB) in performing LSB on patients with CRPS. LSB was performed with either BTA or BTB on patients with CRPS in their lower extremities. The length of time taken for patients to return to the pre-LSB pain score and the adverse effect of LSB with BTA/BTB were investigated. The median length of time taken for the patients to return to the pre-LSB pain score was 15 days for the BTA group and 69 days for the BTB group (*P* = 0.002). Scores on a visual analogue scale decreased in the patients of both groups, and no significant adverse effects were experienced. In conclusion, the administration of either BTA or BTB for LSB is a safe method to prolong the sympathetic blocking effect in patients with CRPS. BTB is more effective than BTA to prolong the sympathetic blocking effect in CRPS patients.

**Keywords:** botulinum toxin; complex regional pain syndrome; lumbar sympathetic ganglion block; pain

**Key Contribution:** The administration of either BTA or BTB for LSB is a safe method to prolong the sympathetic blocking effect in patients with CRPS. BTB is more effective than BTA to prolong the sympathetic blocking effect in CRPS patients.

#### **1. Introduction**

Complex regional pain syndrome (CRPS) is a rare chronic pain syndrome that causes sensory symptoms such as spontaneous pain and allodynia, as well as motor and autonomic nervous system symptoms [1]. It is an intractable pain disorder that requires a multimodal treatment approach, and there is currently no single treatment method that is specific to CRPS [2]. Of the existing methods, sympathetic blocks are known to reduce pain as well as improve motor and autonomic nervous system functions [3]. Despite the small body of research on the effectiveness of sympathetic blocks, the procedure is widely used for the treatment of CRPS [3,4].

A nerve block using local anesthetics is generally administered to the lumbar sympathetic chain in patients with CRPS in their lower extremities. The blocking effect is temporary in most patients, and therefore, achieving a more long-term effect necessitates one or more of the following additional procedures: repeated sympathetic blocks with local anesthetics, radiofrequency thermocoagulation

or neurodestructive procedures that use alcohol or phenol. The merits of a chemical neurolysis of the sympathetic nerve using alcohol or phenol include that it is a simple, economical procedure that can elicit a relatively long-term treatment effect. However, complications can occur, such as genitofemoral neuralgia, which has a prevalence of about 4–10% [5,6]. Moreover, post-sympathectomy neuralgia can occur after chemical sympathetic neurolysis with alcohol or phenol [7]. Radiofrequency thermocoagulation is reported to have similar long-term effects to those of neurodestructive procedures that use phenol [6]. Due to the anatomical location of the lumbar sympathetic chain, however, difficulties have been associated with the procedural method in terms of needle positioning during radiofrequency lesioning [8]. Additionally, radiofrequency lumbar sympatholysis is related to post-sympathectomy neuralgia [9], similar to the chemical sympatholysis.

Botulinum toxin (BT) inhibits the release of acetylcholine from cholinergic nerve endings and is typically used in the treatment of dystonia [10]. The sympathetic blocking effect can be obtained by injecting BT at the sympathetic ganglion, because presynaptic fibers of sympathetic ganglia are also cholinergic. There have been several reports on the prolonged sympathetic blocking effect of BT. When botulinum toxin type A (BTA) was injected into the superior cervical ganglion of a rabbit, the sympathetic blocking effect lasted for a minimum of 1 month and no pathological changes occurred [11]. Another study reported on BT use for a ganglion impar block in patients with chronic perineal pain; a ganglion impar block that used BTA was performed and resulted in pain relief for 3 months and longer [12]. Moreover, Carroll et al. [13] reported that a sympathetic block that used BTA in patients with CRPS in their lower extremities yielded a blocking effect (without any serious complications) for approximately 2 months longer than a block that only used local anesthetics.

Differences exist between BT type A and type B (BTB) with regard to their mechanism of action and target proteins, and moreover BTB can block the release of acetylcholine in the cholinergic nerve fiber [10,14]. Based on this mechanism of action, BTB has been used in clinical practice for treating cervical dystonia [15], adductor muscle spasms [16], piriformis syndrome [17], subacromial bursitis [18], overactive bladder [19] and hyperhidrosis [20]. However, only a few studies have compared the effects of BTA and BTB. When BT was used in patients with cervical dystonia, type A had a longer duration of action than type B [21], but there is no clear conclusion regarding the clinical effect of this difference. Additionally, BTB has been reported to spread into the surrounding tissue less than BTA [14]. With these background studies in mind, we searched the literature for the use of BTB for a lumbar sympathetic block (LSB), but only one case report was found [22]. Furthermore, there has been no published research comparing the effect of the two types of BT when used for LSB. Therefore, the purpose of this retrospective observational study was to compare the effect of BTA and BTB when performing LSB in patients with CRPS in the lower extremities.

#### **2. Results**

A total of 18 patients were included in this study. The demographic data of the patients are shown in Table 1.



Values represent the number of patients. M: male, F: female, BTA: botulinum toxin A, BTB: botulinum toxin B, VAS: visual analogue scale, LSB: lumbar sympathetic block.

One patient in group A and one patient in group B complained of dizziness after the procedure, but the symptoms improved within a week without any particular treatment.

There were no significant differences among the demographic characteristics between the groups. The median duration of observation was 56 days (range 14–570 days). The time taken to return to the pre-procedure VAS score was significantly longer in group B than group A (Figure 1). The median time to return to baseline pain was 69 days in group B (95% CI, 45.5–92.5 days) compared to 15 days for group A (95% CI, 12.9–17.2 days) (*P* = 0.002). The VAS scores that were recorded 1 week after the procedure were found to be significantly lower in both groups compared to those which were recorded prior to the procedure. The median VAS score change that occurred between these time points was 2.5 (range 1.0–6.0) (*P* = 0.043) in group A and 3.0 (range 0.5–5.0) (*P* = 0.001) in group B. However, there was no significant difference in the delta-VAS score before and after treatment in both groups 1 week after the procedure (*P* = 0.633).

**Figure 1.** Log-rank analysis of the duration of analgesia. The median duration of analgesia was 15 (95% CI, 12.9–17.1) days for the BTA group, and 69 (95% CI, 45.5–92.5) days for the BTB group. There was a significant difference in the duration of analgesia between the two groups (*P* = 0.002).

#### **3. Discussion**

In our study, the sympathetic blocking effect of BTB was longer than that of BTA in the patients with CRPS in the lower limbs. Chemical sympathectomy is used widely: it is economical, easy to perform and anticipated long-term effects. In patients with neuropathic pain or CRPS, however, there is no high-quality evidence for the use of chemical sympathectomy, and it is only recommended for use with caution only in clinically selected patients [4,23]. When sympathetic nerves are destroyed through chemical, surgical, or radiofrequency methods, further neuropathic pain can occur due to nerve destruction [24]. Post-sympathectomy neuralgia is a syndrome of pain with a burning nature, which can last 2–3 months [25]. This syndrome can be more severe and debilitating than the initial pain complaint [24], which is thought to be a result of the sensitization of the viscerosomatic nociceptive spinal neuron caused by the sympathectomy. An additional contributing factor may be differentiation hyperactivity in the sensitized neurons as a result of the axotomy [4,26]. Due to the anatomical location, genitofemoral neuralgia can also occur after chemical lumbar sympathectomy [5,6]. Additionally, surgical sympathectomy has been found to produce postsympathectomy neuralgia in up to 39% of patients [27].

BT has seven serotypes, and among these, types A and B are used for the treatment of various conditions. The target site of the toxin is the neuromuscular junction, and the toxin works by dissolving the SNARE [soluble N-ethylmaleimide sensitive factor attachment protein (SNAP) receptor] complex which functions in exocytosis. Type A works on SNAP25, and type B works on synaptobrevin. As a result, BT blocks the release of acetylcholine and exocytosis, causing flaccid paralysis [10]. Therefore, BT is used in the treatment of pain that occurs from dystonia and excessive muscle contraction [28].

BT is also reported to be effective in neuropathic pain syndromes, such as trigeminal neuralgia, postherpetic neuralgia, and CRPS, as well as pain caused by dystonia [13,29–31]. The effectiveness of BT is due not only to the repression of the secretion of acetylcholine, but also to the blocking of the exocytosis of various pain-related neurotransmitters such as substance P and calcitonin gene-related peptide [32–34]. Therefore, there can be additional benefit in using BT for sympathetic block, not only prolonging the duration of sympathetic block but also interrupting the cascade of pain-related neurotransmitters in CRPS.

Few studies have directly injected BT near the nerve, especially near the sympathetic ganglion [35]. In the first study to report on the action and effect of BT on the sympathetic ganglion, BTA was injected into the surgically exposed superior cervical ganglion of a rabbit [11]. The effect of BTA was compared through miosis to a control group that received normal saline, and there was a dose-dependent sympathetic blocking effect of BTA. However, there was no significant change in the number or shape of the nerve cells as compared to the control group. In rabbits in which miosis was observed, the mean time of duration was 5.3 weeks. These findings suggest that BT causes a sympathetic block without causing any histological changes, and this blocking effect reflects the largest difference when compared to chemical sympathectomy using alcohol or phenol.

There was a crossover study that compared the duration of pain relief when using only local anesthetics versus using both local anesthetics and BTA together in performing a lumbar sympathetic ganglion block in nine patients with CRPS in the lower extremities [13]. When BTA was used, the median time taken to return to the level of pain prior to the procedure was 69 days, whereas only 8 days were required when only local anesthetics were used. In addition, with the exception of one patient who complained of nausea and vomiting, which improved without any particular treatment, there were no incidents of adverse effects. This shows that a long-term blocking effect can be safely attained by injecting BTA near the sympathetic ganglion in humans.

A few studies have compared the duration of the blocking effect when using BTA or BTB in a sympathetic block. In a randomized double-blind study that compared the duration of the effect using BTA or BTB in patients with cervical dystonia, similar durations of 14.0 weeks and 12.1 weeks were found, respectively, and this small difference was statistically significant [21]. Similar results were observed in preclinical research [36]. In another study that compared the effect of BTA and BTB in toxin-naïve cervical dystonia patients, no significant difference in the duration of action was found, and the durations were 13.1 weeks and 13.7 weeks, respectively [37].

In our study, BTB resulted in a 4-fold longer duration compared to BTA. These results are contrary to preceding studies; however, the disease and the treatment areas are different. Previous researches that compared the two types of BT have been limited to cervical dystonia and aesthetic treatments. Both of these procedures can involve injecting BT directly into the muscle, and the results may differ when injecting BT closer to the sympathetic ganglia. When BT was injected into the paws of the monkeys, the spreading of toxin was smaller with BTB than with BTA [14]. This result implies that the BTB can be more focused than BTA at the targeted tissue. However, this result conflicts with clinical data that has reported more side effects, such as dry mouth and dysphagia, when BTB was used [38]. Nevertheless, these adverse effects were mostly light to moderate and tended to decrease when injections were repeated, which suggest that rather than occurring from the toxin spreading from the injected area, the side effects likely occurred because of the high sensitivity of BTB to cholinergic autonomic neurons, particularly postganglionic neurons, e.g., the M3 receptor [39]. The results of the present study, including longer duration of action in BTB can be explained in light of the fact that BTB

spreads less to surrounding tissues than BTA and that BTB works more specifically on cholinergic autonomic neurons in the sympathetic ganglia.

Our result that LSB with BTA showed only 15 days (95% CI, 12.9–17.2 days) of median time to return of baseline pain was quite short compared to the previous study result. The previous study by Carroll et al. reported that the 95% confidence interval of the median duration of LSB with BTA was 12–253 days [13], which was very wide. Only three of the eight patients in the study of Carroll et al. showed significantly prolonged effect of BTA. Another possible reason is that the study design of the previous manuscript was 'cross-over design'. According to the article by Nordmann et al. "the successive LSB provided increasing duration of completely pain free period" [40]. Therefore, the repetitive LSB with local bupivacaine only and bupivacaine with BTA in the study by Carroll et al. resulted in the prolonged effect of LSB.

There are limitations to our study. First, this was a retrospective study. Due to the lack of information on BTB use for LSB, there were difficulties in designing and performing a prospective double-blinded study. We used BTA 100 IU and BTB 5000 IU for LSB. In the previous studies which compared the BTA with BTB in the patients with cervical dystonia, the dose ratio of 1 U BTA was comparable to 40–67 U BTB [21,37]. From this, we set the equivalent dose ratio of 1 U BTA to 50 U BTB (BTA 100 = BTB 5000 U) in our study. We think our study can provide a foundation for future well-designed studies on BT use for LSB. Second, because of the retrospective design, we could not measure various changes in symptoms and signs (e.g., oedema, cold sensation and mechanical allodynia), but could measure only the VAS scores, in each patient. Third, the sample size was quite small; however, CRPS is a rare, intractable disorder; therefore, it was not easy to recruit many CRPS patients for a 3-year study period. A multi-center study is necessary to recruit many patients.

#### **4. Conclusions**

In this study, the authors confirmed that both BTA and BTB can be used with no serious side effects when used for an LSB in patients with CRPS type 1 in their lower extremities. In terms of the duration of pain relief, BTB was found to be more effective than BTA. However, this research is limited in that it was a retrospective study and did not include many patients in the BTA group. A well-designed, randomized prospective study is needed to compare the effects of the two types of BT in LSB.

#### **5. Materials and Methods**

#### *5.1. Study Participants*

This retrospective observational study was approved by the institutional review board of our hospital (IRB No. B-1311/228-102), and written consent was obtained from the patients and their guardians after they received an explanation of the methods and the purpose of the procedure as well as recent research results.

The inclusion criteria for patients consisted of the following: (1) patients who visited the pain center at our university hospital and were diagnosed with CRPS type 1 in the unilateral lower extremities from March 2010 to June 2013 and (2) patients who received LSB with either BTA or BTB. The diagnosis of CRPS was made according to the criteria known as the 'Budapest research criteria' [41], which is a revised version of the criteria that were originally proposed by the International Association for the Study of Pain (IASP) [42]. According to the criteria, the patients should meet at least one symptom in all four categories (sensory, vasomotor, sudomotor/edema, and motor/trophic change) and at least one sign in two or more categories. Furthermore, LSB with BT was performed on patients who had previously shown significant pain reduction (50% or more) after LSB with 5 mL of 0.25% ropivacaine. The exclusion criteria consist of the following: (1) patients who had received any neuro-destructive procedure on the lumbar sympathetic ganglia using radiofrequency thermocoagulation or chemicals (alcohol or phenol) or (2) patients who had received a sympathetic block within 6 months.

#### *5.2. LSB Procedure*

Percutaneous LSB was performed in the operating room with fluoroscopic guidance (Figure 2). The patient was placed in a prone position on a radiographic table with a 15-cm-high pillow underneath the patient's abdomen to reduce lumbar lordosis. The patient was given an intravenous infusion of lactated Ringer's solution and was monitored by pulse oximetry, electrocardiogram and blood pressure readings during each procedure. For the LSB procedure, a 21-gauge 15 cm Chiba needle (Cook Inc., Bloomington, IN, USA) was advanced under the oblique projection of a C-arm fluoroscopy. When the Chiba needle was inserted properly at the anterolateral side of the L3 vertebral body, 1–2 mL of contrast agent (Omnipaque®, Nycomed Ireland Ltd., Cork, Ireland) was injected to confirm adequate spread around the target site (anterolateral border of the vertebral body) and to identify any intramuscular or intravascular spreading. When the contrast agent did not spread adequately, the same procedure was repeated by inserting another needle at the L2 vertebral level. In group A, BTA 100 IU (Botox®, Allergan Inc., Irvine, CA, USA) combined with 0.25% ropivacaine (2.5 cc) was injected; in group B, 5000 IU BTB (Myobloc™, Solstice Neurosciences Inc., Louisville, KY, USA) combined with 0.25% ropivacaine (2.5 cc) was injected. When the needle was inserted only at the L3 level, the entire dosage of the medication was injected. Furthermore, when the needle was inserted at the L2 and L3 level, the dosage of BTA or BTB was divided in half and combined with 0.25% ropivacaine (2.5 cc) for injection at each level. After LSB, Skin temperature was measured 30 min after the LSB. The sympathetic block was considered successful when the skin temperature of the affected side increased by ≥2 ◦C.

**Figure 2.** Spread of the contrast agent during the lumbar sympathetic block. The contrast agent was confined to the anterolateral border of the vertebral body without any psoas spread laterally or intravascularly. (**1**) Antero-posterior view. (**2**) lateral view of the fluoroscopy.

#### *5.3. Data Collection and Statistical Analyses*

From the patients' medical records and demographic data, the VAS scores prior to and 1 week after the procedure were examined, and the number of elapsed days until the VAS score returned to the pre-LSB state was recorded. The occurrence of any adverse effects or complications was also examined. The primary endpoint—the time taken to return to the baseline VAS score—was analyzed using Kaplan–Meier analysis. The difference between the two groups was analyzed by the log-rank test. The secondary endpoint—the VAS score change from before to after the procedure—was expressed as the median with a range and was tested using the Wilcoxon signed rank test. PASW® software version 18.0 (Chicago, IL, USA, 2009) was used for the statistical analysis, and *P*-values < 0.05 were considered statistically significant.

**Author Contributions:** Y.L. drafted this article. C.J.L. contributed the idea of this study. E.C. analyzed the data. P.B.L. provided the critical revision of this article. H.-J.L. contributed the data collection. F.S.N. Conceived and designed this study.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Review* **Botulinum Toxin for Central Neuropathic Pain**

#### **Jihye Park <sup>1</sup> and Myung Eun Chung 2,\***


Received: 30 April 2018; Accepted: 28 May 2018; Published: 1 June 2018

**Abstract:** Botulinum toxin (BTX) is widely used to treat muscle spasticity by acting on motor neurons. Recently, studies of the effects of BTX on sensory nerves have been reported and several studies have been conducted to evaluate its effects on peripheral and central neuropathic pain. Central neuropathic pain includes spinal cord injury-related neuropathic pain, post-stroke shoulder pain, multiple sclerosis-related pain, and complex regional pain syndrome. This article reviews the mechanism of central neuropathic pain and assesses the effect of BTX on central neuropathic pain.

**Keywords:** botulinum toxin; BTX; central neuropathic pain; spinal cord injury; post-stroke shoulder pain; complex regional pain syndrome

**Key Contribution:** This review summarizes the mechanism of central neuropathic pain and botulinum toxin action against it based on preclinical and clinical studies.

#### **1. Introduction**

Botulinum toxins (BTXs) are neurotoxic substances produced by *Clostridium botulinum*, a gram-positive anaerobic bacterium. In botulism poisoning, flaccid paralysis occurs by inhibiting the release of neurotransmitters from the peripheral cholinergic nerve terminals of the skeletal and autonomic nervous system. Paralysis begins at the ocular muscles and then spreads to the muscles of the face, before reaching the respiratory muscles and causing respiratory failure.

BTX has traditionally been found in seven serotypes: A, B, C1, D, E, F, and G [1]. They have similar molecular weights and common subunit structures, but differ in their reaction mechanisms, durations of effect, and side effects.

In recent years, using molecular genetic analysis, many genes have been discovered that encode new BTXs. Thus, subtypes, such as BTX/A1, BTX/A2, BTX/B1, and BTX/B2, and chimeric types, such as BTX/DC, BTX/CD and BTX/FA, have been found.

Clinical use of BTX began in 1973, when Scott demonstrated that injecting the toxin into orbicularis oculi muscles was effective in treating strabismus. Over the next several decades, its application expanded to a variety of diseases.

BTX has a molecular weight of 150 kDa, consists of an inactive single-chain polypeptide, and folds into a 3-domain structure. The light chain (50 kDa) is a zinc-dependent protease that constitutes an active toxin and cleaves the soluble *N*-ethyl-maleimide-sensitive factor attachment receptor (SNARE) complex [2,3]. The heavy chain (100 kDa) consists of an N-terminal translocation domain and a C-terminal receptor binding domain, and it acts in neuron-specific binding. The light and heavy chains are linked by disulfide bonds, which partially obscure the active sites of the toxin. When the single-chain disulfide bond is reduced, the light chain metalloprotease can be released to act as a toxin.

The main functional effect of BTX occurs in the neuromuscular junction, where it inhibits the release of acetylcholine from the presynaptic nerve ending, resulting in muscular and autonomic paralysis [4]. The toxin-mediated muscle relaxation process proceeds in three phases: Binding, internalization, and the inhibition of neurotransmitter release. Specific binding to neurons is mediated by heavy chains [5] and internalization is mediated by receptor-mediated endocytosis [6,7]. Once BTX is internalized, the light chains within the vesicles are translocated across the vesicle membrane and released into the neuronal cytoplasm. SNARE is involved in the exocytosis of acetylcholine vesicles, located at the nerve endings, by attaching acetylcholine vesicles to the cell membrane, thus, allowing acetylcholine exocytosis to occur [8]. BTX causes degradation of the SNARE protein, resulting in paralysis. Based on these mechanisms, BTX is clinically used to treat muscle spasticity associated with central nervous system (CNS) disorders, such as stroke, brain injury, spinal cord injury (SCI), cerebral palsy, and multiple sclerosis (MS).

However, two other functional effects of BTXs exist: The effects on the afferent limb of the motor nervous system and the analgesic effect on the sensory nerve system. Several preclinical studies have shown that BTX inhibits neuromodulator and transmitter secretion, which is important for neurotransmission in the sensory pathway, and, thus, BTX may reduce neuropathic pain.

Preclinical and clinical studies have reported the effects of BTX on peripheral neuropathic pain and, generally, demonstrated a high level of evidence for some diseases [9]. However, few studies have reported its therapeutic effects on central neuropathic pain, and its effects have not been proven. The aim of this article is to review the mechanism of central neuropathic pain and to investigate the effect of BTX on central neuropathic pain.

A PubMed and EMBASE search (1980~March 2018) was performed as follows: 'Botulinum toxins', 'neuropathic pain', 'neuropathy', 'pain', 'allodynia', 'hyperalgesia' and 'spinal cord injury', 'post-stroke pain', 'multiple sclerosis', and 'complex regional pain syndrome'. The results included animal studies, randomized controlled trials (RCTs), observational studies, case reports, and reviews. Editorials, guidelines, and trial protocols were excluded. Two reviewers individually assessed the abstracts to determine the eligibility of the studies. Articles not available in English and studies conducted in children (≤18 years of age) were also excluded.

#### **2. Mechanism of Central Neuropathic Pain**

The International Association for the Study of Pain (IASP) defines neuropathic pain as pain caused by a lesion or disease of the somatosensory nervous system [10]. Neuropathic pain is a clinical description that requires a demonstrable lesion or a disease that satisfies the established neurological diagnostic criteria. It has two typical symptoms, allodynia and hyperalgesia. Allodynia describes a pain due to a stimulus that does not normally provoke pain and hyperalgesia refers to increased pain from a stimulus that normally provokes pain [10].

Several molecular mechanisms are involved in the development of allodynia and hyperalgesia. After nerve injury, changes in the expression of sodium and calcium channels cause spontaneous activity in nerve endings, resulting in spontaneous pain. This is an important factor that causes sensitization. In addition, various cytokines, including glutamate, substance P, and proinflammatory cytokines, are involved in sensitization. Inflamed or ischemic tissues become acidified and this cellular environment causes pain by stimulating the release of neuropeptides from the primary afferent nerve tissue [11]. When neuropeptides, such as calcitonin gene-related protein (CGRP) and substance P, are secreted into the endoneurium, they cause local blood flow and blood vessel leakage, leading to edema and pain.

Because central neuropathic pain is defined by IASP as a pain caused by a lesion or disease of the central somatosensory nervous system [10], central neuropathic pain is a heterogenous group of neuropathic pain conditions. Major diagnostic conditions include: (1) Central pain associated with SCI; (2) central post-stroke pain; and (3) central pain associated with MS.

The mechanisms of neuropathic pain, following SCI, in various animal models have been published (Figure 1). Cellular and molecular responses to SCI occur at various levels, from the distal terminals of primary afferent neurons to the cortical area along the nervous system, leading to neuropathic pain.

Several changes in the primary afferent neuron have been suggested. A definite factor that mediates SCI-related pain is an increased excitability of dorsal root ganglion (DRG) neurons. The spontaneous activity of nociceptive DRG neurons has been shown to increase after SCI [12], and some authors have demonstrated that the expression of the capsaicin-sensitive cation channel transient receptor potential vanilloid type 1 (TRPV1) in DRG neurons increases after SCI, which is enhanced by capsaicin-evoked ion currents and calcium responses in DRG neurons [13].

Neuronal hyperexcitability in the spinal dorsal horn is also associated with SCI-related neuropathic pain [14,15]. This phenomenon might occur through chronic glial cell activation [16,17], dendritic spine remodeling [18,19], dysregulation of glutamate homeostasis [16], glutamate receptor activation [20], loss of GABAergic inhibitory interneurons [21,22], interruption of descending inhibitory modulation by serotonin [23,24], or upregulation of voltage-gated calcium channel alpha-2-delta-1 subunit proteins [25].

In patients with central pain following SCI, neurons in the somatosensory thalamus fire in bursts of action potentials more frequently than do similar neurons in patients without pain [26]. In rats with contusive SCI, thalamic ventralis postero-lateralis neurons exhibited a dysrhythmia in that a significantly higher proportion fired spontaneously when compared with neurons in uninjured rats [27]. Based on these results, abnormal thalamic processes following SCI may mediate neuropathic pain. Additionally, in rats with SCI, neurons in the primary somatosensory cortex had significantly higher spontaneous firing rates, greater evoked responses to noxious mechanical stimulation, and a greater tendency to fire bursts of action potentials [28]. Another study also revealed that phosphorylation of AMPA-type glutamate receptors in the primary somatosensory cortex play an important role in the development of hypersensitivity after SCI [29].

**Figure 1.** Illustrated mechanism of central neuropathic pain associated with spinal cord injury (SCI). These mechanisms include transient receptor potential vanilloid type 1 (TRPV1) overexpression in dorsal root ganglion (DRG) neurons, glial cell activation, dendritic spine remodeling, glutamate receptor activation and loss of GABAergic interneuron in the spinal dorsal horn, and spontaneous firing of neurons in the thalamus and primary somatosensory cortex. It has been suggested that the antinociceptive mechanism of botulinum toxins (BTXs) applied to the nerve endings not only affects the primary afferent neurons but also acts on the DRG and spinal dorsal horn through retrograde axonal transport.

The pathophysiology of central post-stroke pain, another important disease constituting central neuropathic pain, remains uncertain. Recent advances in brain imaging technology have increased the understanding of the role of specific anatomic locations. Several reports have suggested that central post-stroke pain commonly occurs in lesions affecting the thalamus, parietal cortex, dorsal putamen, posterior internal capsule, dorsal basal ganglia, brainstem, and lateral medulla [30–33]. Particularly, spinothalamic tracts terminated with the ventral posterolateral thalamus and lesions of the ventral posteromedial thalamus and medial lemniscal thalamocortical pathway were found to be the major

factors causing central post-stroke pain [34,35]. These results have also been demonstrated in animal model experiments. Thermal hypersensitivity was observed in a rat model of the ventral posterior thalamic infarction [36], with thermal and mechanical hyperalgesia developing in rats with a thalamic hemorrhagic lesion [37].

In the middle cerebral artery occlusion rat model, increased *N*-methyl-D-aspartate (NMDA) and AMPA receptor mediated excitatory transmission of the dorsal horn, decreases in GABA and glycine receptor mediated inhibitory transmission, and an increase in descending facilitation was proposed to be involved in the development of central post-stroke pain [38]. In the ventral posterior thalamic lesion rat model, P2X7 expression in the medial thalamus was directly involved in nociceptive transmission, and short-term P2X7 inhibition led to a reduction of glutamatergic facilitation and neuronal hyperexcitability [39].

#### **3. Mechanism of BTX for Central Neuropathic Pain**

The initial analgesic effect of BTX is caused by a decrease in muscle spasms. However, many preclinical and clinical studies suggest that a different mechanism underlies the analgesic effect of BTX. The hypothesis is that BTX inhibits the secretion of neuropeptides and suppresses inflammation and pain.

Several preclinical studies have shown that BTX-A inhibits the release of neurotransmitters that regulate pain and inflammation. McMahon et al. showed that BTX preferentially attenuates the slow phase of KCl-evoked glutamate release, which may be associated with synaptic vesicle mobilization according to a study that utilized a guinea pig formalin-induced pain model [40]. Welch et al. reported that BTX inhibits potassium-evoked substance P secretion from cultured embryonic rat DRG neurons [41], and Durham et al. demonstrated that BTX-A can directly decrease the release of CGRP from cultured rat trigeminal ganglion neurons [42].

Xiao et al. demonstrated that BTX significantly reduces TRPV1 expression [43]. A neuropathic pain model was induced by transection of the lumbar 5 ventral root in male rats. BTX-A or normal saline was administered to the plantar surface by subcutaneous injection. TRPV1 expression increased significantly in the lumbar 4–5 DRG after the transection of the lumbar 5 ventral root, and this increase persisted for at least 21 days. Subcutaneous injection of BTX-A significantly, and dose-dependently, reduced the expression of TRPV1 in the DRG neuron and significantly reduced hyperalgesia. A similar effect occurred on the expression of P2X3, one of the purinergic receptors associated with nociceptors, in a study that evaluated the effect of BTX on P2X3 expression, with the same method. Subcutaneous administration of BTX-A significantly, and bilaterally, reduced mechanical allodynia and inhibited the P2X3 overexpression induced by the transection of the lumbar 5 ventral root [44].

One possible interpretation of these findings is that BTX reduces peripheral sensitization and afferent input to the spinal cord by inhibiting the release of neurotransmitters from peripheral nerve endings, thereby, indirectly a decreasing central sensitization. However, it has been hypothesized that the central effect may be direct by retrograde axonal transport of BTX along the branches of nociceptive neurons.

Immunohistochemical experiments have revealed that cleaved SNAP-25, a product of BTX-A action, is found not only in the peripheral region but also in the facial nucleus in the brain stem [45], superior colliculus [46], and motor region of the spinal cord [47,48]. Antonucci et al. found cleaved SNAP-25 in the ipsilateral facial nucleus after a BTX-A injection into rat whisker muscles [45]. Matak et al. found that cleaved SNAP-25 fragments were present in the ventral horn and dorsal horn of the spinal cord after low-dose toxin injections into the gastrocnemius muscle and sciatic nerve [47]. These authors also identified that cleaved SNAP-25 was not detected in the spinal cord when they injected BTX-A into a sciatic nerve pretreated with colchicine, an axonal transport blocker. Therefore, they suggested that BTX-A showed a central effect by microtubule-dependent retrograde axonal transport [49]. Wang et al., however, cited the possibility that disassociated cleaved SNAP-25 might

have migrated from the terminal to the cell body, suggesting that the discovery of cleaved SNAP-25 in the CNS does not necessarily reflect the activity of BTX-A in the CNS [50].

Retrograde axonal transport is well known as a transport pathway for various substances, such as tetanus toxin, and this fact suggests that BTX may also use the same pathway. Several studies have reported that the heavy chain, or the entire toxin, undergoes retrograde transport after a peripheral injection of BTX-A. Restani et al. directly monitored the endocytosis and axonal transport of BTX-A, and they showed that BTX-A was internalized by spinal cord motor neurons and underwent fast axonal retrograde transport [51]. Wang et al. reported that fluorescently labeled BTX heavy chains were detected in spinal cord motor neurons after injection into the mouse hindlimb, which demonstrated retrograde transport of BTX [50].

Several studies of the behavioral effects of BTX-A have demonstrated the central antinociceptive effect of BTX-A. Bilateral pain associated with experimental diabetes [52], carrageenan-induced hyperalgesia, and paclitaxel-induced peripheral neuropathy [53] or acidic saline-induced mirror pain, [54] can be bilaterally reduced by the unilateral injection of BTX-A in rats. Bach-Rojecky et al. reported that mechanical and thermal hypersensitivity of the ipsilateral side, as well as the contralateral side, were decreased after subcutaneous unilateral BTX injection into the plantar surface of the hindlimb [52]. In addition, Favre-Guilmard et al. reported a significant anti-hyperalgesic effect in the uninjected contralateral hindpaw after subcutaneous administration of BTX-A to the plantar surface in carrageenan-induced hyperalgesia and paclitaxel-induced peripheral neuropathy models. These results suggest that the antinociceptive effect of BTX-A cannot be explained by the peripheral action and it is possible that BTX-A has a central action through the retrograde axonal transport [53]. This process is also expected to be a major mechanism in the BTX-A action on central neuropathic pain.

#### **4. Clinical Studies of BTX for Central Neuropathic Pain**

#### *4.1. Neuropathic Pain after Spinal Cord Injury*

Two case series of clinical reports with very small sample sizes have evaluated the effect of BTX-A on neuropathic pain in patients with SCI. Jabbari et al. [55] reported cases of two patients with burning pain in a dermatome due to spinal cord lesion at the cervical level (tumor or stroke). BTX-A (OnabotulinumtoxinA) was injected subcutaneously at multiple points in the area of the burning pain and allodynia. The effect was assessed by the visual analogue scale (VAS) and clinical changes. One patient received 100 units of BTX-A. One week after the injection, the VAS score decreased from 8–10 to 2–3 points, and the frequency of severe spontaneous pain was reduced by 80%. The second patient received 80 units; skin sensitivity and spontaneous burning pain were significantly reduced after approximately 10 days, and this effect lasted approximately three months. Han et al. [56] reported a case of a patient with allodynia and dysesthesia of the lower limb. BTX-A was injected subcutaneously at ~10 units into the painful foot area and the effect was evaluated by the change in VAS score. After four weeks, the pain severity and burst frequency were reduced.

A recent study has been reported on the effect of BTX-A on SCI-associated neuropathic pain. Han et al. [57] reported the effects of BTX-A in a randomized, double-blind, and placebo-controlled study in 40 patients with SCI-associated neuropathic pain. Patients were treated with subcutaneous injections of BTX-A (200 units) or normal saline and the VAS score, the Korean version of the short-form McGill Pain Questionnaire, and the WHOQoL-BREF quality of life assessment were assessed at four and eight weeks. Thus, BTX-A has been shown to be effective in treating intractable chronic neuropathic pain in patients with SCI. The above studies are summarized in Table 1.


**1.** Summary of studies of botulinum toxin (BTX) for central neuropathic pain.

**Table** 


**Table 1.** *Cont.*

#### *4.2. Post-Stroke Shoulder Pain*

Central post-stroke pain occurs after a cerebrovascular event, including lesion of the brainstem, thalamus, and cerebral cortex, and may affect half of the body [73]. Several authors have described central post-stroke pain as a central neuropathic pain syndrome that can occur after a stroke in the body part that corresponds to the cerebrovascular lesion and is characterized by pain and sensory abnormalities, where other causes of obvious nociceptive, psychogenic, or peripheral neuropathic origin have been ruled out [74,75].

Post-stroke shoulder pain is a common disease with an incidence rate ranging from 21–72% [76,77]. Many studies have examined the effects of BTX on the treatment of post-stroke shoulder pain, but the results are conflicting, and, thus, drawing conclusions remains difficult.

Yelnik et al. [58] conducted a double-blind RCT of the effect of BTX on post-stroke shoulder pain in 20 patients. Ten patients were injected with 500 units of BTX-A (AbobotulinumtoxinA) in the subscapularis muscle, and 10 patients in the control group underwent a placebo injection in the same muscle. All participants underwent rehabilitation, including stretching exercises. Pain was improved in the BTX injection group at one week, and pain scores using a 10-point verbal scale at four weeks showed a significant difference between the two groups. Marco et al. [59] reported a double-blind RCT for evaluating the effect of BTX. In 14 patients, 500 units of BTX-A (AbobotulinumtoxinA) was injected into the pectoralis major muscles, and 15 patients in the control group were injected with a placebo. Transcutaneous electrical nerve stimulation was applied for six weeks. After one week, pain during shoulder movement decreased in both groups, but the VAS score in the BTX injection group decreased more significantly, and this trend continued until six months. However, no significant difference in the shoulder range of motion or spasticity was found between the two groups. Kong et al. [60] conducted a double-blind RCT of 17 patients. Five hundred units of BTX-A (AbobotulinumtoxinA) was injected into the pectoralis major and biceps brachii muscles in the experimental group and normal saline was injected into the same region in the control group. The VAS scores at 4, 8, and 12 weeks after injection were not significantly different between the two groups. The median baseline VAS score of the patients was 6, and the scores decreased by 2–3 points in both groups.

Lim et al. [61] reported a double-blind RCT of 29 patients. In the experimental group, 100 units of BTX-A (OnabotulinumtoxinA) was injected into the infraspinatus, pectoralis, and subscapularis muscles, along with intra-articular saline. In contrast, the control group received an intra-articular triamcinolone (40 mg) injection and saline was injected into the muscles described above. The numeric rating scale at 12 weeks was reduced by 4.2 ± 0.4 points in the BTX-A intramuscular injection group and by 2.5 ± 0.8 points in the intra-articular triamcinolone injection group. Intramuscular injection of BTX-A was superior to intra-articular injection of triamcinolone (*p* = 0.051). Boer et al. [62] conducted a double-blind RCT of 22 patients. They injected 100 units of BTX-A (OnabotulinumtoxinA) into the subscapularis muscle in the experimental group and injected saline in the control group. Vertical VAS scores were not significantly different between the two groups at 6 weeks and 12 weeks. Shaw et al. [63] reported the effects of BTX-A (AbobotulinumtoxinA) on spasticity, function, and pain in patients with spasticity of the upper limb after stroke. This study was a multicenter RCT called BoTULS. The pain rating and pain scale evaluated at one and three months showed no significant difference between the two groups, but the pain rating evaluated at 12 months showed a significant decrease in the injection group. Marciniak et al. [64] evaluated the effects of BTX by injecting 140–200 units of BTX-A (OnabotulinumtoxinA) or saline into a pectoralis major muscle, with or without a teres major muscle. At four weeks, worst pain ratings decreased in both groups, but no significant difference was found between the two groups. The above studies are summarized in Table 1.

Post-stroke shoulder pain is thought to be caused by multiple factors, including both the nervous system and mechanical factors. Post-stroke shoulder pain may be associated with spasticity, but it is difficult to determine whether spasticity acts as a mechanism to cause post-stroke shoulder pain, whether it is increased by the shoulder pain, or both. It is well known that the improvement of

spasticity may be associated with an improvement of pain, but the correlation between spasticity and pain is not linear, and multiple factors may be involved [78]. Several studies have suggested a musculoskeletal origin for post-stroke shoulder pain. Musculoskeletal conditions, such as subluxation, tendinitis, adhesive capsulitis, rotator cuff tear, and subacromial bursitis, may contribute to post-stroke shoulder pain. Whether these diseases can cause post-stroke shoulder pain is controversial, because these conditions may result from stroke.

In a study by Zeilig [79], those with post-stroke shoulder pain had higher heat-pain thresholds and exhibited higher rates of hyperpathia, allodynia, and dysesthesia in the affected shoulder and leg than those without post-stroke shoulder pain. The authors suggested that the finding of altered thermal sensitivity, which indicates damage to the spinothalamic-thalamocortical tract, was not restricted to the shoulder, but rather characterized the affected side. An additional support to this central neuropathic pain proposition was the higher rate of damage to the parietal cortex in the post-stroke shoulder pain group. From these results, various factors appear to cause post-stroke shoulder pain, but it seems to be attributable to central neuropathic pain.

Two systematic reviews of the effects of BTX on shoulder pain, including post-stroke shoulder pain, have been reported. In a Cochrane report that focused on BTX-A for shoulder pain, the authors included six RCTs comparing BTX with a placebo or active treatment. Five RCTs in participants with post-stroke shoulder pain indicated that, compared with placebo, a single intramuscular injection of BTX-A significantly reduced pain at three to six months postinjection, but not at one month [80]. Another systematic review included nine RCTs of BTX injections in patients with shoulder pain. The shoulder pain was due to hemiplegia in six studies, adhesive capsulitis in one study, subacromial bursitis or shoulder impingement syndrome in one study, and arthritis in one study [81]. They concluded that BTX injection resulted in minor to moderate pain relief and an increase in shoulder abduction in patients with chronic shoulder pain. Based on these two reviews, BTX injection in post-stroke shoulder pain is expected to be effective in reducing pain.

#### *4.3. Multiple Sclerosis*

MS is a chronic disease in which focal demyelinating lesions of the CNS occur at multiple sites due to autoimmune inflammatory processes. The plaques, located in the subcortical, brainstem, or spinal cord, cause neurological symptoms and signs, including abnormal coordination, motor, sensory, and cognitive function. According to one report, approximately 65% of MS patients with spasticity are known to suffer from pain [82]. Pain appears in the form of central dysesthetic pain, trigeminal neuralgia, Lhermitte's phenomenon, and tonic spasm. In a review published in 2013, the authors classified MS-related pain into nine categories and described each possible mechanism [82]. According to the authors, ongoing extremity pain is caused by thalamic or cortical deafferentation by multiple lesions along the spinothalamocortical pathways. In addition, Lhermitte's phenomenon and painful tonic spasm are caused by demyelination of the dorsal column primary afferents and the corticospinal pathway, respectively, providing a possible mechanism for central neuropathic pain.

Various double-blind RCTs of BTX effects on MS-associated detrusor overactivity and spasticity are available. A preliminary report on the effects of BTX on spastic dysphagia, myokymia, tonic spasm, and internuclear ophthalmoplegia also exists. However, no RCT has been performed to evaluate whether BTX is effective for MS-related pain. In a recent prospective, open-label study of 131 patients with spasticity, 19% were patients with MS, and 60% reported a significant reduction in spasticity-related pain after BTX-A treatment [83]. MS-related pain has the characteristics of central neuropathic pain and conducting well-designed research on whether BTX is effective for MS-related pain will be necessary in the future.

#### *4.4. Complex Regional Pain Syndrome*

Complex regional pain syndrome (CRPS) is a painful disease that can result from an imbalance due to trauma. Unlike other neuropathic pain syndromes, CRPS is accompanied by additional signs, such as abnormal blood flow control, sweating, and active and passive motor impairment. Not all CRPS signs can be fully explained by the peripheral mechanism and several studies suggest there might be a central mechanism [84]. The CNS undergoes functional and structural changes in people with chronic pain and these changes are thought to be particularly important in CRPS because they cause central sensitization [85,86]. CRPS is often spread beyond the original injury and in many cases, it has been reported to spread to the contralateral extremity in a mirror pattern [87,88]. Regarding the bilateral spreading of CRPS signs, it was suggested that trans-synaptic changes in the spinal cord dorsal horn, contralateral to the affected side, may underlie this spreading [89]. It is unclear whether CNS alterations are a primary abnormality in the disease or a secondary change due to pain, but changes in the CNS play an important role in CRPS, thus, this review addresses the disease.

Carroll et al. prospectively investigated 18 subjects with CRPS [66]. Lumbar sympathetic blocks were accomplished with bupivacaine alone or an additional 75 units of BTX-A. The rate of pain return was significantly lower and the duration of pain reduction was longer in the group with BTX-A injection compared to that with local anesthetic alone. Safarpour et al. conducted a double-blind, randomized, controlled, and open-label extension study [67]. BTX-A was injected intradermally and subcutaneously into the allodynic area, but no significant response occurred after treatment. Kharkar et al. injected 10–20 units of BTX-A per muscle in 37 patients with CRPS-related spasm or dystonia in the neck and/or upper limb girdle [68] and found a statistically significant decrease in local pain scores compared with baseline. Several additional case reports and case series with very small sample sizes are listed in Table 1.

Although no systematic review of the effect of BTX on CRPS is available, most studies have shown BTX to be effective in reducing pain. CRPS also has a multifactorial mechanism, but as central sensitization is reported as a major mechanism, BTX is expected to effectively reduce the pain.

#### **5. Conclusions**

The mechanism of central neuropathic pain has been examined according to various hypotheses, including neuronal hyperexcitability and dysfunction of the spinothalamic tract. To date, various preclinical and clinical studies have been published on whether BTX may be effective for central neuropathic pain. BTX inhibits the secretion of substance P and CGRP from DRG, inhibits the expression of TRPV1 and P2X3, and induces a central effect through retrograde axonal transport. In addition, several studies have demonstrated its effect on central neuropathic pain associated with SCI, stroke, MS, and CRPS. Effects of BTX on neuropathic pain after SCI, post-stroke shoulder pain, and CRPS has been shown; therefore, it can be considered as one of the treatment options. In the future, well-designed studies will be necessary to assess the effects of BTX on central neuropathic pain, and, furthermore, effective injection sites, injection techniques, and adequate doses should be considered.

**Acknowledgments:** This study was funded by Ministry of Science, ICT and Future Planning, Republic of Korea (NRF-2016R1C1B1016489).

**Conflicts of Interest:** The authors declare no conflicts of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
