OnabotulinumtoxinA for the prophylactic treatment of headaches in adult patients with chronic migraine: a safety evaluation

ABSTRACT

Introduction: Existing oral prophylaxis for chronic migraine (CM) are often ineffective or poorly tolerated. OnabotulinumtoxinA (onabotA) is approved for headache prophylaxis in CM and ameliorates headaches in patients refractory to multiple preventatives.

Areas covered: We appraise evidence regarding action mechanisms, pharmacodynamics, and pharmacokinetics of onabotA in CM prophylaxis. We critically evaluate salient clinical and real-world studies demonstrating its efficacy in improving multiple aspects of CM. We discuss onabotA safety, tolerability, and adverse events (AEs) for CM prophylaxis from clinical trials, post-authorization studies and meta-analyses, including novel pregnancy safety data and comparisons with oral prophylactics. We explore areas of future interest, particularly onabotA safety and efficacy in the context of novel antibody-based prophylaxis.

Expert opinion: Clinical and real-world evidence demonstrate onabotA safety, tolerability and efficacy for CM prophylaxis. Most AEs are mild/moderate and self-limiting, with few serious AEs and no treatment-related deaths. Common AEs include neck pain, ptosis, muscle weakness, and stiffness. Modifying existing responder-criteria enables more patients to benefit from onabotA. OnabotA shows superior safety and efficacy to oral preventatives, and appears safe in pregnancy. Future pregnancy-risk register will clarify pregnancy and lactation safety further. Future research comparing onabotA safety and efficacy with newly emergent antibody-based prophylaxis is keenly awaited.

KEYWORDS:

• Onabotulinumtoxin a
• chronic migraine
• prophylaxis
• safety
• tolerability

1. Introduction

Migraine is a disabling condition with 15% lifetime-prevalence and associated with significant morbidity and socioeconomic burden [1]. Migraines are classified as episodic (EM) or chronic migraine (CM). EM is subdivided into high- and low-frequency EM and, over time, may evolve into CM, defined by the International Classification of Headache Disorders 3^rd-Edition (ICHD3) as ≥15 headache days/month for >3 months, with ≥8 days/month of headaches fulfilling migraine diagnostic criteria [2]. CM affects 1.4–2.2% of the population [3], and prophylaxis to reduce headache frequency/duration/severity, reduce abortive medication use and ameliorate migraine-associated disability is the mainstay management strategy. Efficacious oral anti-migraine preventatives include anticonvulsants (topiramate), antidepressants (amitriptyline), and beta-blockers (propranolol) [4,5]. However, oral prophylaxis do not specifically target the molecular pathophysiology of CM. Patients are often unresponsive to or intolerant of multiple preventatives, with 5–31% of CM patients unresponsive to all existing prophylaxis [6]. Poorly controlled pain additionally predisposes to medication-overuse headache (MOH) [7,8]. Thus, optimal CM management clearly requires more efficacious preventatives. In this context, OnabotulinumtoxinA (onabotA), a therapeutic Botulinum neurotoxin preparation administered to multiple locations in the facial and cranial musculature, demonstrates efficacy in CM prophylaxis in clinical and real-world studies, and was approved for CM prophylaxis by the Medicines and Healthcare Products Regulatory Agency (MHRA) and US Food and Drug Administration (FDA) in 2010, and recommended by the National Institute for Health and Care Excellence (NICE) in 2012 for UK patients unresponsive to, or intolerant of, ≥3 oral prophylaxis classes [9]. However, in determining the clinical utility of a given therapeutic, it is necessary to consider both its efficacy and safety and tolerability. In clinical practice, this frequently necessitates a balance to be struck between the two, whereby a highly efficacious treatment with some adverse effects may nonetheless possess good clinical utility. Therefore, this review will consider and critically evaluate the clinical and real-world evidence pertaining to the high safety and efficacy, and favorable risk-benefit ratio, of onabotA in adult CM prophylaxis.

2. Pharmacological properties and mechanisms of action

The mechanism of action of Botulinum neurotoxins is well-understood. There are seven immunologically distinct Clostridium botulinum exotoxins (serotypes A–G) that inhibit presynaptic acetylcholine release at the skeletal muscle neuromuscular junction and preganglionic autonomic synapses, causing flaccid skeletal muscle paralysis and autonomic dysregulation. Toxin serotypes may share target proteins, although the site of enzymatic cleavage on target proteins may differ [10]. Physiological Botulinum toxins are synthesized as polypeptide pre-toxins and cleaved by tissue proteinases to the active 150 kilodalton (kDa) dimeric toxin, containing one heavy and one light chain [11]. Following physiological inoculation and toxin transport to the neuromuscular junction, the heavy chain binds to presynaptic polysialogangliosides and a protein receptor, including synaptic vesicle 2, synaptotagmin-I, and synaptotagmin-II. This enables endocytosis and internalization of the entire neurotoxin molecule into the neuronal endosome [12]. The light chain then translocates from the endosome into the cytoplasm and cleaves cytoplasmic soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) molecules, with Botulinum toxin A light chain cleaving the C-terminus of synaptosome-associated protein of 25 kDa (SNAP25) [13]. This prevents presynaptic neurotransmitter vesicle docking and fusion with the presynaptic membrane, inhibits presynaptic neurotransmitter release, reduces ability of muscle to contract, and causes paralysis [14–16]. This property led to the development of onabotA (Botox) as a therapeutic Botulinum toxin A preparation to treat blepharospasm and strabismus involving ocular muscle overactivity [17,18]. Subsequent clinical trials demonstrated onabotA reduced muscular contraction and pain in cervical and focal dystonias and limb spasticity, resulting in its approval as first-line therapy for both conditions [19,20].

OnabotA efficacy in migraine prophylaxis was initially suggested by incidental migraine improvement in patients undergoing cosmetic toxin treatment. Subsequently, in a non-randomized open-label study, 106 patients received onabotA into the frontal, temporal, glabellar, and suboccipital musculature for either hyperfunctional facial lines or dystonias with concomitant headaches [21]. In patients diagnosed with migraine, 51% experienced complete headache resolution for 4.1 months (mean), whilst 38% reported >50% headache frequency reduction for 2.7 months, with similar baseline headache characteristics between responders and non-responders. This initial observation of onabotA efficacy in migraine treatment was subsequently confirmed by pivotal Phase 3 trials in CM. However, unlike its mechanism for inducing muscular paralysis, the mechanisms by which intramuscular onabotA prevents migraines are less well-understood.

Current evidence suggests onabotA antagonizes peripheral and central sensitization and nociceptive activation of the trigeminovascular pathway, important for chronic migraine pathogenesis [22]. Peripheral sensitization refers to increased responsiveness to nociceptive migrainous triggers/stimuli in first-order sensory afferent neurones of the trigeminal nerve and its vasculature, responsible for producing the initial pain of migraine. Peripheral sensitization to nociceptive stimuli is thought to occur in meningeal sensory afferents innervating the dura and possibly pia mater, which release vasoactive and pro-inflammatory neuropeptides and neurotransmitters that further irritate and potentiate their prolonged activation [23]. Peripheral sensitization, in turn, is thought to lead to central sensitization, with sequential sensitization of second-order neurones in the spinal trigeminal nucleus and third-order thalamic neurones, accounting for the cutaneous allodynia and extracephalic hypersensitivity of migraine and contributing to its chronification [24,25]. OnabotA inhibits release of pain-mediating neuropeptides and neurotransmitters from first-order trigeminal ganglia neurones, including calcitonin gene-related peptide (CGRP) and glutamate, to antagonize peripheral sensitization [26,27], whilst onabotA may downregulate insertion of nociceptive receptors including transient receptor potential cation channel subfamily-V member-1 (TRPV1) and A member-1 (TRPA1) ion channels into neuronal membranes to inhibit afferent neuronal-firing [28]. Therefore, inhibiting peripheral sensitization indirectly inhibits central sensitization. Previous postulations that onabotA might directly antagonize central sensitization by inhibiting second-order neuronal-firing [29] has been refuted by recent evidence against transsynaptic transfer of onabotA, suggesting onabotA is unable to traverse across peripheral neurones to reach and act directly on second- and third-order neuronal synapses [30]. Therefore, onabotA most likely indirectly modulates central sensitization through diminished peripheral sensory afferent input [31].

OnabotA neurotoxicity hinders elucidation of its pharmacokinetics. OnabotA has low immunogenicity and co-exists in a molecular complex with neurotoxin accessory proteins (NAPs), potentially altering its binding kinetics and enhancing molecular stability through protection from proteolysis [32,33]. Animal studies demonstrate intraneuronal onabotA is physiologically active for ≤80 days, consistent with its clinically observed efficacy duration of 3 months. Although recovery from paralysis was thought to occur through growth of novel neuronal sprouts, recent evidence suggests novel sprouts are poorly efficient in acetylcholine secretion and are pruned upon recovery of the main synapse [34–37]. Intracellular onabotA light chain undergoes ubiquitination and proteasomal degradation, leading to toxin inactivation [38]. Therefore, an alternative postulation is that neuromuscular junction recovery may relate to diminished ability of the light chain to escape ubiquitination-mediated degradation over time [39].

3. Clinical efficacy in migraine

OnabotA efficacy for CM prophylaxis was demonstrated over 1 year by the seminal Phase III PREEMPT 1 and 2 studies, and over ≥2 years by multiple Phase IV real-world studies, including COMPEL and REPOSE (Table 1). OnabotA has not demonstrated efficacy in EM, or chronic daily or tension-type headaches that may co-exist in migraineurs [40–46].

Table 1. OnabotulinumtoxinA efficacy in key clinical and real-world studies

Study name	Study design and duration	Number of patients	Patient characteristics	Key efficacy outcomes
PREEMPT 1, Aurora et al [2010, 47]	DB, placebo-controlled, randomized trial with OL phase DB phase: 24 weeks	OnabotA: 341 Placebo: 338	CM, age 18–65, 59–64% used ≥1 preventative, 66–70% medication overuse	At 24 weeks (DB phase): No difference in headache episodes (onaBoNTA −5.2, placebo −5.3, p = 0.34) Reduced headache days/28 days (−1.4 vs placebo, p = 0.006), migraine days/28 days (−1.5vs placebo, p = 0.002), number of moderate/severe days/28 days (−1.4 vs placebo, p = 0.004), HIT6 (−2.3 vs placebo, p < 0.001), % patients with HIT6 ≥ 60 (−11%, p = 0.001)
PREEMPT 2, Diener et al [2010, 48]	DB, placebo-controlled, randomized trial with OL phase DB phase: 24 weeks	OnabotA: 347 Placebo: 358	CM, age 18–65, 64–66% used ≥1 preventative, 62–63% medication overuse	At 24 weeks (DB phase): Reduced headache days/28 days (−2.3 vs placebo, p < 0.001), migraine days/28 days (−2.4 vs placebo, p < 0.001), moderate/severe headache days/28 days (−2.5 vs placebo, p < 0.001), headache episode frequency (−0.7, p = 0.003), HIT6 score (−2.5 vs placebo, p < 0.001), % patients with HIT6 ≥ 60 (−10.2% vs placebo, p = 0.003)
Dodick et al [2010, 49]	PREEMPT 1 and 2 pooled DB phase 24 weeks	OnabotA: 688 Placebo: 696	CM, age 18–65, 64–66% medication overuse	At 24 weeks (DB phase): Reduced headache days/28 days (−1.8 vs placebo, p < 0.001), migraine days/28 days (−2.0 vs placebo, p < 0.001), moderate/severe headache days/28 days (−1.9 vs placebo, p < 0.001), headache episode frequency (−0.3, p = 0.009), HIT6 score (−2.4 vs placebo, p < 0.001), % patients with HIT6 ≥ 60 (−10.6% vs placebo, p < 0.001)
Aurora et al [2011, 50]	PREEMPT 1 and 2 pooled DB + OL phase 56 weeks	OnabotA: 688 Placebo: 696	CM, age 18–65, 64–66% medication overuse	At 56 weeks (DB+OL phases): Headache day frequency/28 days vs baseline: O/O − 12.17, P/O − 11.32. Reduced headache days/28 days (−0.9 vs P/O, p = 0.019), migraine days/28 days (−0.9 vs P/O, p = 0.018), moderate/severe headache days/28 days (−0.8 vs P/O, p = 0.027), total headache hours on headache days (−23.4 vs P/O, p = 0.018)
Aurora et al [2014, 51]	PREEMPT 1 and 2 pooled DB + OL phase 56 weeks	Patients completing all 5 treatment cycles: 1005 O/O: 515, P/O: 492	CM, age 18–65,	At 56 weeks (DB+OL phase): Reduced headache days/28 days (−0.9 vs P/O, p = 0.035), migraine days/28 days (−0.9 vs P/O, p = 0.038), moderate/severe headache days/28 days (−0.9 vs P/O, p = 0.042), HIT6 score (−0.6 vs P/O, p = 0.157)
COMPEL, Blumenfeld et al [2018, 55]	OL prospective study, 35 centers, USA, Australia, South Korea 108 weeks	716	CM, age ≥18, ≤1 oral preventative	At 108 weeks (vs baseline): Reduced headache days/28 days (−10.7, p < 0.0001), moderate/severe headache days/28 days (−9.5, p < 0.0001), HIT6 score (−7.1, p < 0.0001). 61.1% patients had ≥50% headache day reduction Reduced headache days/28 days at week 24, 60, 84 and 108 vs baseline (p < 0.0001)
REPOSE, Ahmed et al [2019, 60]	Real-world OL study, 78 centers, Europe 2 years	633 (Mean 5.5 cycles, range 1–13)	CM, 41% with medication-overuse	At 2 years (vs baseline): Reduced MHD at all timepoints (20.6 to 7.4, p < 0.01), improved MSQ scores (restrictive: 17.3 to 33.6, preventive: 15.3 to 28.9, emotional: 17.9 to 34.9, p < 0.001), EQ-5D total score (0.69 to 0.89, p < 0.001) and health state score (50 to 70, p < 0.001)
Khalil et al [2014, 61]	Real-world OL study, single-center, UK ≥1 cycle	254 (455 cycles)	CM, all used ≥1 preventative (94% used ≥3 preventatives), 50% medication-overuse	Vs baseline: improved MHD (−7, p < 0.001), MMD (−6, p < 0.001), crystal clear days (+7, p < 0.001), days requiring triptans (−3. p < 0.001), days requiring painkillers (−3, p < 0.001), HIT6 (−9.7, p < 0.001). Hull criteria (≥50% reduction in MHD, (≥50% reduction in MHD, or doubling of HFD if ≥3 HFD at baseline or HFD >6 if <3 HFD at baseline): 34% met 0/3 criteria 24% met 1/3 criteria 16% met 2/3 criteria 26% met 3/3 criteria
Ahmed et al [2021, 62]	Real-world OL study, single-center, UK ≥2 years	655	CM, 100% used ≥1 preventative (98% used ≥3 preventatives), 53% medication-overuse	Vs baseline: improved MHD (−8, p < 0.001), MMD (−6, p < 0.001), crystal clear days (+7, p < 0.001), HIT6 (−6, p < 0.001). After Cycle 2: 58% (380) responders by Hull criteria (48% by NICE criteria) At 2 years (380 responders): 152 continued treatment, 28 became resistant on treatment, 61 converted to EM then reverted to CM and restarted treatment, 112 converted to EM and stopped treatment
Andreou et al [2018, 63]	Real-world OL study, single-center, UK	200	CM, 100% used ≥3 preventatives, 46% with medication-overuse	After 2 cycles (vs baseline): Reduced MHD (24.0 to 11.3, p < 0.025), MMD (13.0 to 5.0, p < 0.025), crystal clear headache-free day (0 to 12.8, p < 0.025), HIT6 score (70 to 64, p < 0.025) After Cycle 2: 64% responder (≥30% MHD reduction after 2 cycles). In responders, MHD (baseline 23) reduced to 8, 8, 8, 11 after cycles 2, 5, 8, 13. 34% converted to EM (29.4% after 2 cycles)
Negro et al [2015, 64]	Real-world OL study, single-center, Italy 2 years	155	CM, 100% used ≥1 preventative, 100% with MOH	After 2 years (vs baseline): Reduced MHD (22.3 to 7.3, p < 0.001), MMD (21.4 to 6.8, p < 0.001), acute medication days (20.8 to 5.3, p < 0.001), HIT6 score (68.9 to 52.0, p < 0.001)
Negro et al [2015, 65]	Real-world OL study, single-center, Italy 2 years	172 (195 units onabotA per patient)	CM, 100% used ≥1 preventative, 100% with MOH	After 2 years (vs baseline): Reduced MHD (22.2 to 4.1, p ≤ 0.05), MMD (21.6 to 3.8, p ≤ 0.05), acute medication days (21.0 to 3.7, p ≤ 0.05), HIT6 score (67.9 to 49.0, p ≤ 0.05)
Dominguez et al [2018, 66]	Real-world OL study, 13 centers, Spain 1 year	725	CM, 100% used ≥2 preventatives including topiramate, 58% with MOH	After 1 year (vs baseline): Reduced MHD (21.8 to 8.4, p < 0.01), MMD (13.8 to 6.0, p < 0.01), analgesic days (17.0 to 6.3, p < 0.01), MIDAS score (35.9 to 9.1, p < 0.01)
Ahmed et al [2019, 67]	Real-world OL study, single-center, UK. 8 years	851	CM, 100% used ≥1 preventative (98% used ≥3 preventatives), 53% medication-overuse	52.6% responders by Hull criteria

CM chronic migraine, DB double-blind, EM episodic migraine, EQ-5D EuroQoL 5 Dimension Questionnaire, HIT6 score Headache Impact Test-6 score, MHD monthly headache days, MIDAS The migraine disability assessment test, MOH medication-overuse headache, MSQ Migraine specific quality of life questionnaire, OL open-label, OO onabotA/onabotA group, PO placebo/onabotA group.

3.1. Randomized-controlled trials: the PREEMPT studies

PREEMPT 1 was a multi-center randomized-control trial of 679 CM patients, with a 24-week double-blind parallel-group phase containing two injection cycles of 12 weekly onabotA or placebo (at 0 and 12-weeks), preceding a 32-week open-label phase with three injection cycles (at 24, 36 and 48-weeks) [47]. 59–64% used ≥1 preventative and 66–70% overused analgesia. Patients received 155 units of onabotA in 31 fixed sites across head/neck musculature (frontalis, procerus, corrugator, occipitalis, temporalis, cervical paraspinal muscles, and trapezius), with additional 40 units at clinician discretion using a follow-the-pain strategy. However, the primary outcome, mean change from baseline headache-episode frequency at 24-weeks, was not achieved, with no significant difference between onabotA and placebo groups (onabotA −5.2 vs placebo −5.3, p = 0.344), although in a post-hoc subanalysis, post-treatment headache-episode frequency was significantly lower in the onabotA group. There was statistically significant reduction in triptan use from baseline in onabotA group compared to placebo (onabotA −3.3 vs placebo −2.5, p = 0.023), whilst onabotA also improved headache and migraine-day frequency and Headache Impact Test-6 (HIT-6) score indicating migraine-related disability.

PREEMPT 2 was identical in design to PREEMPT 1, although primary outcome was switched to baseline headache-day frequency reduction [48]. At 24 weeks, onabotA-treated patients had significantly greater improvement in headache-day frequency (onabotA −9.0 vs placebo −6.7, p < 0.001), significantly greater triptan use reduction (onabotA −3.0 vs placebo −1.7, p < 0.001), migraine-day frequency, moderate/severe headache-day frequency and proportion of patients with severe HIT6 scores than placebo, thus achieving primary outcome. Unlike PREEMPT 1, PREEMPT 2 demonstrated significant difference in headache-episode frequency reduction from baseline between onabotA and placebo-arms, likely due to lower placebo-effect in PREEMPT 2 creating comparatively greater differences between the two groups. PREEMPT 2 results at 24 weeks were confirmed by a pooled analysis of 1384 PREEMPT patients, which also demonstrated statistically significant greater reduction in headache-episode frequency in the onabotA group compared to placebo (onabotA −8.4 vs placebo −6.6, p < 0.001) [49], while 56-week pooled analysis of all PREEMPT patients, and subanalysis of patients completing all five treatment cycles, further demonstrated onabotA/onabotA treatment statistically significantly reduced headache-day frequency, migraine-day frequency, and moderate/severe headache days compared to placebo-onabotA [50,51]. Significantly improved headache-day frequency and headache hour reduction began from Cycle 2 of onabotA onwards [52]. Pooled analyses confirmed greater reduction in triptan use in onabotA-treated patients than placebo-group [49,50]. Overall, these confirm durable onabotA efficacy in reducing headache and migraine days and headache severity at 1 year.

3.2. Critical appraisal of PREEMPT studies

PREEMPT were the first and largest randomized placebo-controlled studies to evaluate onabotA for CM prophylaxis, comprehensively investigating multiple clinical facets of CM and establishing the standard injection paradigm with 155-units in 31 designated pericranial locations, plus 40 units possible via a follow-the-pain strategy. However, one important criticism was switching of the primary outcome for PREEMPT 2 before study completion, from baseline headache episode frequency reduction used in PREEMPT 1 to mean change in baseline headache-day frequency, given PREEMPT 1 failure to meet its primary outcome. The authors also noted the 2008 International Headache Society clinical trial guidelines for evaluating CM prophylaxis, and expressed preference of the FDA, as reasons for switching [48]. It can be argued that using headache-episode frequency as primary outcome lacked meaningfulness in patients with near-constant daily headaches, contributing to statistically insignificant difference with placebo, whilst achieving primary outcome in PREEMPT 1 was further complicated by baseline headache-episode frequency differences between the two groups (onabotA 12.3 vs placebo 13.4, p = 0.023). Nevertheless, the authors identified a statistically significantly greater reduction in baseline headache-episode frequency in the onabotA-group compared to placebo during pooled analysis of PREEMPT 1 and 2 [49], suggesting insufficient sample size in PREEMPT 1 may also contribute to inability to achieve primary outcome. However, pooled analysis is complicated by the fact that PREEMPT 1 and 2 comprised different patient-populations, with critics suggesting higher placebo rates amongst US patients than European patients [53]. Secondly, unrealistically for real-world clinical practice, >30% recruited patients used no previous preventatives. High placebo-effect (in both trials) meant absolute gain of only 2.3 headache-free days over placebo at 24-weeks in PREEMPT 2, impairing its clinical significance. 50% baseline headache-day reduction responses may better evaluate outcome, since averaging to calculate mean headache-day reductions inevitably underestimates response magnitude in those with high baseline headache-burden, particularly with short study durations. Another caveat concerns effective blinding, given patients may recognize onabotA administration by observing for diminished forehead/facial lines, although strong placebo-effect argues against this. Finally, whilst there was statistically significantly reduced triptan use frequency from baseline in the onabotA group compared to placebo, reduced acute medication use during study may contribute to improvements in both treatment and placebo arms by alleviating MOH, which PREEMPT did not address. Modest improvement over placebo in PREEMPT led NICE to recommend onabotA as an option for headache prophylaxis in UK adults with CM and appropriately managed medication overuse, after failing ≥3 other preventatives, with treatment cessation if <30% monthly headache-day reduction after two cycles, or upon reversion to EM with <15 headache days/month for three consecutive months [9].

3.3. Real-world studies

Multiple real-world studies confirmed onabotA efficacy. COMPEL was a multinational, multicentre, open-label prospective study, designed to reflect real-world practice, of 715 CM patients receiving nine cycles of 12 weekly onabotA (PREEMPT paradigm) over 108 weeks [54–58]. Concomitant preventative use was permitted. OnabotA significantly reduced headache days from baseline at all timepoints including 108-weeks (108 weeks: −10.7 days, p < 0.0001), besides improving moderate-severe headache-day frequency, HIT-6 score and quality-of-life measures. 24-week data were similar to PREEMPT [59]. However, COMPEL’s high drop-out rate of 47.9%, primarily due to withdrawn consent and being lost-to-follow-up, may leave a more-responsive cohort that positively skews outcome measures. The number of preventatives tried per patient before onabotA was unclear, whilst efficacy in those with or without MOH was not compared. Moreover, COMPEL did not address positive or negative stopping rules, likely reflecting stylistic differences in medical-practice between countries.

REPOSE was a pan-European multicentre open-label prospective study with 633 CM patients receiving eight onabotA cycles over 2 years [60]. Each patient received a mean of 5.5 cycles. Headache-day frequency was reduced from 20.6 to 7.4 days at endpoint (p < 0.01) with significantly improved quality-of-life. Recruited patients reflected real-world practice, with 70% each having previously used anticonvulsants, beta-blockers or antidepressants and 41% with medication-overuse. In this observational study, investigators were free to determine dosage and inter-injection intervals, with adherence to licensed administration regimen recommended but not obligatory. 79% and 46% patients received onabotA at >13 weeks and > 16 weeks intervals at least once, although reasons for delays were unspecified. It is possible that lack of obligation to follow PREEMPT dosages and administration regimen results in overtreatment and overestimation of true onabotA efficacy in REPOSE. Nonetheless, overall, both COMPEL and REPOSE demonstrate sustained onabotA efficacy in CM prophylaxis to 2 years.

These results were corroborated by other real-world studies. 254 CM patients receiving 455 onabotA treatment cycles at the Hull Migraine Clinic, a large tertiary UK headache-center, reported statistically significantly improved monthly headache-days, migraine-days, gain in crystal-clear days, triptan and non-triptan days and HIT-6 score (all p < 0.001), while 47% and 32% patients achieved ≥30% and ≥50% headache-day frequency reduction, respectively [61]. 94% patients failed ≥3 prior preventatives, while half misused analgesia/triptans, illustrating real-world onabotA efficacy in refractory patients with medication-overuse. The PREEMPT protocol was followed, although no patients received any additional injections permitted by it, beyond the standard 155 units in 31 sites. However, the average number of onabotA-cycles each patient received was unspecified. Noting that some experienced improved migraine-days without headache-day improvement, and NICE criteria deficiencies that do not account for improvements in headache severity and recommend treatment discontinuation upon reversion to EM, meaning those with 16 headache-days/month can continue onabotA whilst those with <15 headache-days/month cannot, the authors proposed the Hull responder criteria, defining response as ≥50% reduction in either headache or migraine-days, or doubling of baseline crystal-clear headache-free days (if ≥3/month pre-treatment) or achieving ≥6 headache-free days (if <3/month pre-treatment) to enable continuation. 66% met ≥1 criterion and 26% achieved all three criteria. In 655 onabotA-treated Hull CM patients followed-up for 24–70 months, 58.0% and 47.8% fulfilled Hull and NICE responder criteria after Cycle 2. 23%, 39%, and 39% achieved ≥50% reduction in headache days, migraine days and ≥2-fold increase in crystal-clear days at latest follow-up. Of those remaining in study at 2 years, 60% discontinued treatment, of which 29% reverted to EM, 16% transiently reverted to EM for ≤9-months, and 7% became onabotA-refractory [62]. In CM patients who failed ≥3 prophylactics, Andreou et al. demonstrated 64% met NICE continuation criteria by achieving ≥30% baseline monthly headache-day reduction after two cycles. 34% reverted to EM, similar to Hull group data. Those treated for 36 months experienced sustained monthly headache-day, migraine-day, and headache-free day improvements [63]. Other real-world studies corroborated these results at 1–2 years [64–66]. Moreover, in a 8-year follow-up of 972 refractory Hull CM patients treated with onabotA, 52.6% were responders according to Hull-criteria with significantly improved HIT-6 scores [67]. Overall, real-world evidence conclusively demonstrates onabotA produces sustained clinically meaningful efficacy up to 8-years.

One important question concerns onabotA efficacy in concomitant MOH. Hull data with 434 CM patients suggested those with and without medication-overuse experienced similar headache and migraine-day reductions and headache-freedom gain [68]. Two open-label Italian studies reported significantly reduced headache and migraine-days, increased headache-freedom and reduced medication-use at 2 years in >200 CM patients with MOH treated off-label with 155 or 195 units of onabotA, suggesting onabotA efficacy in patients with MOH [64,65]. However, off-label use without specific stopping-criteria warrants caution in extrapolating results to contexts where positive and negative stopping criteria govern onabotA administration, although currently the UK is the only country with positive stopping criteria.

4. Safety and tolerability in chronic migraine

4.1. Safety and tolerability in clinical trials

Salient studies regarding onabotA safety are summarized (Table 2). For onabotA-treated patients in the 24-week double-blind phase of PREEMPT 1, 59.7% experienced adverse events (AEs) (placebo: 46.7%), 25.3% experienced treatment-related AEs (TRAEs) (placebo: 11.7%), 5.3% experienced serious AEs (placebo: 2.4%), and 13.2% discontinued treatment (placebo: 12.7%), with 4.1% discontinuing during double-blind or open-label phases due to adverse effects with onset during the double-blind phase (placebo: 0.9%) [47]. Similarly, in the first 24-weeks, for onabotA-treated PREEMPT 2 patients, 65.1% experienced AEs (placebo: 56.4%), 33.4% experienced TRAEs (placebo: 13.7%), 4.3% experienced serious AEs (placebo: 2.2%) with 1 (0.3%) serious TRAE (migraine requiring hospitalization), and 3.5% discontinued treatment during double-blind or open-label phases due to AEs which onset during double-blind phase (placebo: 1.4%) [48]. Pooled PREEMPT 24-week analysis confirmed 62.4% of onabotA-treated patients experienced AEs (placebo: 51.7%), 29.4% experienced TRAEs (placebo: 12.7%), 4.8% experienced serious AEs (placebo: 2.3%), and 3.8% discontinued treatment due to AEs (placebo: 1.2%). Most AEs in PREEMPT were mild/moderate and self-resolving. Consistent with data from cervical dystonia treatment, commonest TRAEs of onabotA were neck pain (6.7%) and muscle weakness (5.5%). Other TRAEs included eyelid ptosis (3.3%), musculoskeletal pain (2.2%), injection-site pain (3.2%), headache (2.9%), myalgia (2.6%) and muscle stiffness (2.3%). AEs resulting in discontinuation included neck pain (0.6%), muscular weakness (0.4%), headache (0.4%), and migraine (0.4%) [49]. Overall, these indicate that, despite more AEs than placebo, two cycles of onabotA were safe and generally well tolerated, with no new emergent safety/tolerability issues.

Table 2. OnabotulinumtoxinA adverse events and tolerability in key clinical and real-world studies

	PREEMPT 1, Aurora et al [2010, 47]	PREEMPT 2, Diener et al (2010 48)	Dodick et al (2010 49)	Aurora et al (2011 50)	Aurora et al (2014 51)	Diener et al [2014, 69]	COMPEL, Winner et al [2019, 71]	Khalil et al (2014 61)	Andreou et al [2019, 63]	Dominguez et al [2018, 66]	Negro et al (2015 64)	REPOSE, Ahmed et al [2019, 60]	CM-PASS, Matharu et al [2017, 70]	Ahmed et al [2019, 67]	Total numbers of unique study patients
Study design	DB, placebo-controlled, randomized trial with OL phase	DB, placebo-controlled, randomized trial with OL phase	PREEMPT 1 and 2 pooled 24 week DB phase	PREEMPT 1 and 2 pooled DB plus 32 week OL phase	PREEMPT 1 and 2- all Pt treated with 3–5 cycles	4-trial pooled data: PREEMPT 1 and 2 plus 2 Phase II trials	OL prospective study, 35 centers, USA, Australia, South Korea	Real- world OL study, single-center, UK	Real-world OL study, single-center, UK	Real-world OL study, 13 centers, Spain	Real- world single-center OL study, Italy, CM + MOH	Real- world OL study, 78 centers, Europe	Real- world multi-center OL study, Europe	Real- world OL study, single-center, UK
Study duration	DB phase- 24 weeks	DB phase- 24 weeks	DB phase- 24 weeks	OL phase- 32 weeks	56 weeks	≥ 1 cycle (16,926 patient-months)	108 weeks	≥ 1 cycle	6 –36 months	1 year	2 years	2 years	64 weeks	8 years
Number of onabotA-treated patients (N)	340	347	687	1205 at start of OL phase	1005	1997	716	254	200	725	155	633	1160	972	6558
Any AEs, n (%)	203 (59.7%)	226 (65.1%)	429 (62.4%)	703 (58.3%)	775 (77.1%)	1455 (72.9%)	436 (60.9%)	n/a	n/a	89 (12.3%) after 1 cycle, 37 (5.1%) after 12 months	n/a	n/a	478 (41.2%)	88 (12.6%)	2546
TRAEs	86 (25.3%)	116 (33.4%)	202 (29.4%)	245 (20.3%)	332 (33.0%)	n/a	131 (18.3%)	n/a	n/a	n/a	23 (17.5%)^c	116 (18.3%)	291 (25.1%)	n/a	893
Serious AEs	18 (5.3%)	15 (4.3%)	33 (4.8%)	46 (3.8%)	62 (6.2%)	107 (5.4%)	75 (10.5%)	n/a	0 (0%)	n/a	0 (0%)	n/a	61 (5.3%)	n/a	243
Serious TRAEs	0 (0%)	1 (0.3%)	1 (0.1%)	1 (0.1%)	1 (0.1%)	n/a	1 (0.1%)	n/a	n/a	n/a	0 (0%)	8 (1.3%)	1 (<0.1%)	n/a	11
Mild, n (%)	Majority	Majority	Majority	Majority	n/a	318 (15.9%)^a	n/a	n/a	n/a	76 (10.2%)	All	45 (7.1%)	n/a	n/a	439
Moderate, n (%)					n/a	711 (35.6%)	n/a	n/a	n/a	n/a		47 (7.4%)	n/a	n/a	758
Severe, n (%)	n/a	n/a	n/a	n/a	n/a	n/a	n/a	n/a	0 (0%)	n/a	0 (0%)	24 (3.8%)	n/a	n/a	24
Neck pain	20 (5.9%)	26 (7.5%)	46 (8.7%)	55 (4.6%)	22 (4.3%)^b	252 (12.6%)	29 (4.1%)	n/a	2 (1.0%)	n/a	5 (3.8%)^c	18 (2.8%)	51 (4.4%)	n/a	357
Muscle weakness	20 (5.9%)	18 (5.2%)	38 (5.5%)	47 (3.9%)	8 (1.6%)^b	160 (8.0%)	10 (1.4%)	n/a	2 (1.0%)	n/a	5 (3.8%)^c	N/A	31 (2.7%)	n/a	208
Eyelid ptosis	n/a	<5%	23 (3.3%)	30 (2.5%)	10 (1.9%)^b	91 (4.6%)	18 (2.5%)	28 (11.0%)	14 (7.0%)	n/a	4 (2.9%)^c	34 (5.4%)	47 (4.1%)	58 (6.8%)	294
Injection-site pain	n/a	n/a	22 (3.2%)	24 (2.0%)	11 (2.1%)^b	89 (4.5%)	14 (2.0%)	38 (14.9%)	n/a	n/a	4 (3.3)^c	n/a	n/a	n/a	145
Headache	n/a	n/a	20 (2.9%)	17 (1.4%)	n/a	142 (7.1%)	12 (1.7%)	n/a	n/a	n/a	5 (3.7%)^c	n/a	n/a	n/a	159
Migraine	n/a	n/a	n/a	n/a	n/a	88 (4.4%)	7 (1.0%)	11 (4.3%)	n/a	n/a	n/a	n/a	34 (2.9%)	n/a	140
Myalgia	n/a	<5%	18 (2.6%)	15 (1.2%)	n/a	58 (2.9%)	n/a	n/a	n/a	n/a	n/a	n/a	11 (0.9%)	n/a	69
Muscle stiffness	n/a	<5%	16 (2.3%)	20 (1.7%)	n/a	122 (6.1%)	17 (2.4%)	37 (14.6%)	n/a	n/a	n/a	17 (2.7%)	23 (2.0%)	Present	216
Musculoskeletal pain	n/a	n/a	15 (2.2%)	13 (1.1%)	n/a	89 (4.5%)	n/a	n/a	n/a	n/a	n/a	n/a	10 (0.9%)	n/a	99
Facial paresis	n/a	n/a	n/a	n/a	n/a	122 (6.1%)	9 (1.3%)	15 (5.9%)	n/a	n/a	n/a	n/a	15 (1.3%)	n/a	160
Facial spasm	n/a	n/a	n/a	n/a	n/a	n/a	n/a	n/a	n/a	n/a	n/a	n/a	11 (0.9%)	n/a	11
Dysphagia	n/a	n/a	n/a	n/a	n/a	n/a	n/a	5 (1.96%)	1 (0.5%)	n/a	n/a	n/a	3 (0.3%)	n/a	9
Hypersensitivity	n/a	n/a	n/a	n/a	n/a	n/a	n/a	n/a	1 (0.5%)	n/a	n/a	n/a	10 (0.9%)	n/a	11
Fainting during injection	n/a	n/a	n/a	n/a	n/a	n/a	n/a	3 (1.2%)	1 (0.5%)	n/a	n/a	n/a	n/a	n/a	4
Skin tightness	n/a	n/a	n/a	n/a	n/a	n/a	7 (1.0%)	n/a	n/a	n/a	n/a	n/a	n/a	n/a	7
Discontinuation due to AEs, n (%)	14 (4.1%)	12 (3.5%)	26 (3.8%)	31 (2.6%)	n/a	67 (3.4%)	32 (4.5%), due to serious AE- 13 (1.8%)	n/a	0 (0%)	n/a	n/a	10 (1.6%)	51 (4.4%)	n/a	160
Deaths, n (%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	n/a	0 (0%)	n/a	n/a	0 (0%)	2 (0.2%)- not TRAE	n/a	2 (not TRAE)
Pregnancy outcomes	n/a	n/a	n/a	n/a	n/a	n/a	n/a	n/a	n/a	n/a	n/a	1 spontaneous abortion- unclear if treatment-related	n/a	n/a	1

AE adverse events, CM chronic migraine, DB double-blind phase, MOH medication-overuse headache, OL open label, TRAE treatment-related adverse events.

^aas proportion of all AEs

^bof 515 patients completing all 5 cycles

^cof 132 patients who completed study

During the 32-week PREEMPT open-label phase, when all patients received onabotA, 58.3% developed AEs, 20.3% developed TRAEs (mostly mild/moderate), and 3.8% developed serious AEs with one treatment-related serious AEs (0.1%). During the open-label phase, 25.4% and 29.3% of patients in the onabotA-onabotA and placebo-onabotA groups discontinued onabotA, with 2.6% discontinuing due to AEs. TRAEs during the open-label phase included neck pain (4.6%), muscle weakness (3.9%), eyelid ptosis (2.5%), muscle tightness (2.2%), injection-site pain (2.0%), musculoskeletal stiffness (1.7%), headache (1.4%), myalgia (1.2%), and musculoskeletal pain (1.1%) [50]. Overall, over 56 weeks, 78.3% and 75.9% of patients receiving 5 or 3 onabotA cycles developed AEs, 34.8% and 31.2% developed TRAEs, 7.4% and 4.9% developed serious AEs, 0.2% and 0% developed treatment-related serious AEs, and no deaths occurred. The commonest overall AEs in patients completing all five cycles were neck pain (4.3%), muscle weakness (1.6%), injection-site pain (2.1%), and eyelid ptosis (1.9%). Rate of TRAEs decreased with each onabotA treatment cycle [51]. These indicate 12 weekly onabotA according to PREEMPT paradigm is safe and well-tolerated at 56 weeks, with neck pain, muscle weakness, eyelid ptosis and muscle tightness the commonest side effects.

These results were further confirmed in a 4-trial pooled analysis of 1997 patients who received ≥1 onabotA treatment cycles from both PREEMPT studies and two Phase II studies [69]. Patients received 163 units onabotA per cycle on average. 72.9% reported AEs, mostly mild/moderate. Commonest AEs were neck pain (12.6%), muscle weakness (8.0%), musculoskeletal stiffness (6.1%), facial weakness (6.1%), and eyelid ptosis (4.6%). Double-blind placebo-controlled phase data indicate neck pain, muscular weakness, facial paresis, and musculoskeletal stiffness developed within 5–7 days of most recent injection, lasting 20–60 days. 5.4% of onabotA-treated patients developed serious AEs, including migraine (0.5%), pneumonia (0.3%), uterine leiomyoma (0.3%), and headache (0.2%). Pneumonia and uterine leiomyoma were not considered treatment-related. Possible cases of distal muscle weakness were medically determined as unrelated to distal toxin spread. Patients receiving 150–200 units per cycle had lower AE frequency than those receiving >200 units. Overall, these confirm general onabotA safety for migraine prophylaxis at doses used by PREEMPT. Most AEs were mild, typically onsetting within one week and resolving within 8–9 weeks.

4.2. Safety and tolerability in post-marketing real-world studies

Extensive prospective data attest to real-world onabotA safety and tolerability for CM prophylaxis. Shortly after NICE-guidance publication, the Hull Migraine Clinic identified AEs including injection-site pain (14.9%), neck stiffness (14.6%), ptosis (11%), inability to frown (5.9%), headache exacerbation (4.3%), dysphagia (1.96%), and fainting during injection (1.2%) in 254 patients treated with ≥1 onabotA cycles [61]. The higher AE incidence than other studies could relate to proactive patient education regarding potential side effects resulting in greater patient awareness, rather than reliance on spontaneous reporting. Likewise, whole-cohort AEs in 200 CM patients in Andreou et al. who received 2–12 onabotA-cycles included mild/moderate ptosis (7.0%), neck weakness/pain (1.0%), dysphagia (0.5%), allergic reaction (0.5%), and fainting during injection (0.5). All observed AEs were mild/moderate and transient, with no treatment discontinuations due to AEs [63]. Of note, dysphagia had not previously been published as an AE during onabotA treatment for CM.

In a 1-year Spanish prospective study of 725 onabotA-treated CM patients, 12.3% developed AEs after first dose, of which 82.3% were mild. After 12 months, 94.9% reported no AEs. 0.7% discontinued onabotA due to treatment intolerance. However, individual AEs were unspecified [66]. In the largest post-authorization safety study of onabotA for migraine, CM-PASS, 1160 onabotA-treated migraine patients from 58 European centers were prospectively followed up for 64 weeks [70]. At baseline, 86% were chronic migraineurs, 44% received ≥1 acute and ≥1 preventative therapy, and 25% had co-existing MOH. Median number of injection sites [31] and total dose (155-units) were consistent with PREEMPT paradigm, although median inter-cycle interval was 13.7 weeks. 41.2% patients reported AEs, 25.1% reported TRAEs, and 5.3% reported serious AEs. 4.4% discontinued treatment due to AEs. 1 patient (0.1%) reported a serious TRAE (migraine exacerbation). 2 patients (0.2%) reported fatal AEs not considered treatment-related (myocardial infarction and metastatic lung cancer). Special-interest TRAEs included migraine worsening (4.0%), intractable migraine (0.4%), hypersensitivity (0.9%) and dysphagia (0.3%). Percentage of patients reporting AEs decreased with treatment duration, from 26.8% after 1 cycle to 18.6% after 5 cycles. These confirm real-world onabotA safety for CM prophylaxis at 1 year.

One Italian open-label study of 155 refractory CM patients with concomitant MOH treated with 155 units/cycle of onabotA found that at 2 years, 17.5% patients developed TRAEs, including neck pains (3.8%), musculoskeletal weakness (3.8%), headache (3.7%), injection-site pain (3.3%), and ptosis (2.9%). All AEs were mild to moderate, with no serious AEs and no disruption to administration regime. All AEs lasted between <1 week and 2 months, the longest being eyelid ptosis and muscle weakness. 75% of those developing neck pain and musculoskeletal stiffness did so during the first three cycles. 85.2% completed the study, with no discontinuations due to AEs [64]. Similar results were obtained in those receiving 195 units/cycle of onabotA [65]. These suggest onabotA is safe and well tolerated at 2 years.

These findings were confirmed at 2 years by COMPEL and REPOSE. In COMPEL, at 108 weeks (after nine cycles), 436 of 716 total patients (60.9%) receiving ≥1 treatment cycles developed AEs. 18.3% developed TRAEs. 10.5% developed serious AEs. There were no deaths. TRAEs included neck pain (4.1%), eyelid ptosis (2.5%), musculoskeletal stiffness (2.4%), injection-site pain (2.0%), headache (1.7%), muscular weakness (1.4%), facial paresis (1.3%), migraine (1.0%), and skin tightness (1.0%). Similar to PREEMPT, one patient developed one treatment-related serious AE causing discontinuation (rash). 4.5% patients discontinued treatment due to AEs, including eyelid ptosis (0.4%), headache (0.4%), rash (0.4%), and suicidal ideation (0.4%). TRAE incidence decreased with treatment cycles, from 24.2% at Cycle-1 to 12.2% at Cycle-9 [71].

In REPOSE, where 633 CM patients received 3499 total onabotA cycles over 2 years, 116 patients (18.3%) reported 267 adverse drug reactions (ADRs). 7.1%, 7.1%, and 3.8% of total patients developed mild, moderate, and severe ADRs. 1.3% reported ≥1 serious ADR. There were no deaths. Similarly to previous studies, commonest ADRs included ptosis (5.4%), neck pain (2.8%), and musculoskeletal stiffness (2.7%). Serious ADRs included depression, psychosomatic disease, headache, migraine, vomiting, spinal disorder, spontaneous abortion, and asthma, and tended to occur in patients taking concomitant headache medications, thereby making attribution to one specific therapy challenging. 1.6% patients discontinued treatment due to non-serious ADRs including migraines, ptosis, musculoskeletal pains, and dysphagia. One patient (0.2%) discontinued due to a serious ADR (spinal disorder) not considered treatment-related [60]. COMPEL and REPOSE permitted use of concomitant oral preventatives, demonstrating onabotA safety and tolerability over 2 years and when combined with oral preventatives, whilst confirming the frequency and range of AEs in PREEMPT. Moreover, in the 8-year follow-up study by the Hull Migraine Clinic, 12.6% of 972 refractory CM patients reported AEs, particularly neck stiffness, whilst 6.8% reported ptosis [67].

Overall, based on 6558 unique patients across key studies (Table 2), 2546 (38.8%) developed any AEs, 893 (13.6%) developed TRAEs, 243 (3.7%) developed serious AEs, 11 (0.2%) developed serious TRAEs, and 160 (2.4%) discontinued onabotA due to AEs. Most AEs were mild or moderate and self-resolving over weeks, with low probability of serious or severe AEs. Common AEs, affecting >100 total patients, include neck pain (357, 5.4%), eyelid ptosis (294, 4.5%), muscle stiffness (216, 3.3%), muscle weakness (208, 3.2%), facial paresis (160, 2.4%), headaches (159, 2.4%), migraine worsening (140, 2.1%), and injection-site pain (145, 2.2%). Rare but significant AEs included hypersensitivity (11, 0.2%) and dysphagia (9, 0.1%). In individual studies, incidence of serious TRAEs did not exceed 1.3%. AEs diminished with treatment cycles. Whilst inconsistent methods of AE monitoring and reporting across studies may underestimate true overall AE incidences, strong overall evidence exists from both clinical and real-world studies supporting the safety and tolerability of onabotA in CM prophylaxis up to 8-years follow-up.

4.3. Safety in pregnancy

Murine studies suggested fetal body-weight reduction and diminished skeletal ossification with two onabotA doses at >8 units/kg during pregnancy [72]. However, this far exceeds the human dose-equivalent for migraine prophylaxis. One case report described a 26-year-old chronic migraineur who initially stopped onabotA treatment after becoming pregnant, but recommenced onabotA at 18 gestational weeks due to worsening migraine [73]. She gave birth to a healthy full-term girl with normal development at 6.5-year follow-up, suggesting onabotA safety during pregnancy. Brin et al analyzed the Allergan safety database for pregnancies with maternal onabotA exposure, identifying 232 pregnancies with known outcomes [74]. 96% of fetal exposure occurred before or during first trimester. 83% patients received <200 units. In 137 prospective cases, 111 (79%) were live births with 29 (21%) fetal losses. Congenital defect prevalence was 2.7% of live births, comparable to background rate. Fetal defects included ventricular septal defect, laryngomalacia, metatarsus adductus and benign cardiac murmur. However, this data derived from onabotA administration for movement disorders, esthetics and pain, not migraine prophylaxis. Wong et al reported the first prospective real-world study of pregnancy outcomes for onabotA in CM prophylaxis, in the Hull Migraine Clinic [75]. Over 9 years, 45 women received onabotA within 3 months prior to conception. 32 elected to continue treatment through pregnancy. One miscarriage was reported, whilst all other patients had healthy full-term babies with normal birthweight and no congenital malformations. The solitary miscarriage occurred at nine gestational weeks in a woman without previous medical history. She became pregnant again 4 months later, continued onabotA and delivered a healthy full-term baby. Overall, although this study involved small sample sizes, available real-world evidence suggests onabotA administration does not adversely impact pregnancy outcomes. No data are available on onabotA safety during lactation.

4.4. Comparison with the safety of other drugs

Few studies have directly compared onabotA with oral CM prophylaxis. Blumenfeld et al compared onabotA ≤100 units/cycle and sodium valproate 250 mg twice daily in a randomized, double-blind, placebo-controlled study of 59 EM and CM patients over 9.5 months [76]. Patients either received onabotA plus placebo tablet, or saline injection plus valproate. Two injections were administered, 3 months apart. OnabotA-treated patients reported significantly fewer AEs than valproate-treated patients (50% vs 75.8%, p = 0.04) and significantly fewer discontinuations due to AEs (3.3% vs 27.6%. p = 0.01). Commonest AEs were eyelid ptosis (26.7%) and eyebrow droop (16.7%) in onabotA-treated patients, and gastrointestinal discomfort (41.2%), alopecia (17.2%), and fatigue/drowsiness (31.0%) in valproate-treated patients. Commonest onabotA AEs causing discontinuation were nausea and irregular breathing.

Two randomized, placebo-controlled trials compared onabotA with topiramate, the oral prophylactic with best efficacy evidence for CM prophylaxis [77,78]. These demonstrated fewer AEs and discontinuation due to AEs in onabotA-treated patients [79,80]. However, small sample sizes limited interpretation. Consequently, the FORWARD study was conducted as a multicentre, randomized, parallel-group, post-authorization open-label study to compare effectiveness and safety of onabotA and topiramate over 36 weeks [81]. FORWARD employed effectiveness as a composite outcome measure consisting of efficacy and tolerability to compare real-world clinical utility of treatment modalities. 282 CM patients were randomized to 12 weekly onabotA 155-units for three cycles, or topiramate 50–100 mg/day for 36 weeks, with crossover permitted for patients randomized to topiramate after 12 weeks. Greater proportion of onabotA-treated patients achieved ≥50% headache-day reduction, and experienced greater baseline headache-day reduction than the topiramate-group. Overall, AEs occurred in 48% of 220 patients initially randomized or later crossing-over to onabotA, versus 79% of 142 patients initially randomized to topiramate. 17% onabotA-treated patients versus 70% topiramate-treated patients developed TRAEs. 1% of those initially randomized to onabotA and 42% of those randomized to topiramate discontinued treatment due to AEs. No crossover patients discontinued onabotA due to AEs. The commonest onabotA TRAEs were neck pain (4%), musculoskeletal pain (2%), migraine (1%) and visual blurring (1%), whilst those of topiramate were paresthesia (29%), cognitive disorder (12%), fatigue (12%), nausea (12%), decreased appetite (11%), dizziness (11%), and attention disturbance (8%). The only serious TRAE was nephrolithiasis in one topiramate-treated patient, with no deaths. This demonstrates superior safety and tolerability of onabotA compared to topiramate over 36 weeks for CM prophylaxis.

Herd et al compared onabotA to placebo and oral preventatives in adults with CM or EM, with or without MOH, in a Cochrane review and meta-analysis of 28 randomized control trials incorporating 4190 total participants [82]. 23 trials compared onabotA against placebo, while 3 trials compared onabotA against established oral prophylaxis. In summative results of six trials (n = 2839), onabotA had a relative risk (RR) of 2.2 for TRAEs compared to placebo (95% CI: 1.7–2.8). OnabotA demonstrated relative risk reduction (RRR) of 24% for TRAEs and RRR of 72% for AE-induced treatment withdrawal compared to topiramate and valproate in three trials. Treatment-withdrawal rate for onabotA was 3%, based on data of all trials whose patients received ≥1 treatment cycles. Despite small sample sizes and insufficient information in multiple trials that diminish evidence quality, these results suggest onabotA safety and tolerability profile is inferior to placebo, but superior to available oral preventatives.

Direct safety comparisons between onabotA and novel anti-CGRP antibodies for CM prophylaxis (erenumab, fremanezumab, and galcanezumab) is currently lacking. Indirect comparisons, using odds ratios (OR) of discontinuation due to AEs/serious AEs for different preventatives compared to placebo, suggest OR for discontinuation due to AEs of 2.6 for onabotA versus 1.0–1.9 for anti-CGRP antibodies, and OR for serious AEs of 2.1 for onabotA versus 0.5–2.6 for anti-CGRP antibodies [83]. A network meta-analysis is currently underway to compare efficacy outcomes and adverse-events rate for CGRP antagonists and onabotA using randomized control trial data, which will further clarify this issue [84].

Altogether, randomized trial and meta-analysis data indicate onabotA has greater safety and tolerability than oral preventatives. Much-needed future comparisons with anti-CGRP antibodies are awaited with interest.

5. Conclusion

OnabotA is efficacious, safe and well-tolerated for CM prophylaxis. Evidence from both randomized trials and real-life studies support this conclusion. Most reported side-effects are mild or moderate, transient and self-limiting, with very few serious AEs and no treatment-related deaths. Incidence of adverse events decreases with treatment duration. OnabotA demonstrates good safety and efficacy when used with other oral preventatives and with medication-overuse. Limited available data indicates onabotA demonstrates no adverse outcomes in pregnancy. Available studies indicate onabotA holds greater safety and tolerability than oral preventatives. Further investigation and monitoring will be required to assess onabotA safety during lactation and compare its safety profile with those of newly emergent anti-CGRP prophylactic therapies.

6. Expert opinion

OnabotA is an integral component of current CM prophylaxis and an effective, safe and well-tolerated treatment option, including with medication-overuse and alongside other oral preventatives. It has superior efficacy and safety to topiramate, the oral preventative with the best evidence for CM prophylaxis. Studies in treatment-naïve and multiple treatment-refractory patients validate its efficacy in both cohorts in meaningfully improving headache and migraine-days, headache-episodes, acute treatment use, headache-freedom and quality-of-life, and its high safety and tolerability.

However, due to high cost, many healthcare systems recommend onabotA for CM prophylaxis following failure of multiple first-line prophylactics. The European Headache Federation (EHF) recommends onabotA in patients who have preferably failed 2–3 prophylactics, with treatment discontinuation if <30% monthly headache-day reduction after 2–3 cycles [85]; whilst NICE recommends it for those with adequately-managed medication-overuse who failed ≥3 previous prophylactics, with treatment cessation if <30% monthly headache-day reduction after 2 cycles, or upon reversion to EM with <15 headache-days/month for three consecutive months [9]. However, these recommendations do not cater for headache severity improvements, or the observation that some onabotA-treated patients experience migraine-day reduction without headache-day reduction. Moreover, evidence suggests high-frequency EM (8–14 headache days/month) and CM patients have similar disability burden [86]. The Hull criteria were proposed to address these shortcomings, defining responders as experiencing >50% reduction in monthly headache or migraine-days, or doubling of monthly headache-free days (if ≥3/month pre-treatment) or achieving ≥6 monthly headache-free days/month (if <3/month pre-treatment). This enabled an extra 10% patients deemed non-responders by NICE criteria to continue treatment and demonstrate reasonable response [87]. Given similar disability burdens between high-frequency EM and CM, a modified positive stopping-rule was applied, with treatment cessation only if patients experience <10 headache days/month for three consecutive months [62]. Overall, the Hull criteria permit a broader patient cohort to benefit from onabotA for longer than NICE guidance. Its positive stopping-rule was similarly adopted by the EHF, who recommends treatment cessation in those with <10 headache days/month for three consecutive months, with reevaluation 4–5 months after stopping to ensure they remain episodic [85].

An important unmet need concerns pregnancy and lactation safety. The first prospective study into onabotA for migraine prophylaxis in pregnancy demonstrated no adverse effects on pregnancy outcomes []. This may encourage greater future onabotA use during pregnancy and, potentially, lactation. Therefore, a future onabotA pregnancy risk-register will facilitate gathering of comprehensive long-term pregnancy safety data, whilst additionally studies should address safety during lactation.

An important future consideration is the relationship between onabotA and novel anti-CGRP antibodies, both efficacious and approved for use in CM refractory to multiple prophylactics. A novel study reported erenumab for 9 months reduced function-limiting migraine days and improved quality-of-life in onabotA-refractory CM [88]. Indirect comparisons suggested onabotA and anti-CGRP therapies have similar odds ratios for discontinuation due to AEs [83]. Moreover, recent studies suggest combination of onabotA with erenumab, fremanezumab or galcanezumab further reduces monthly headache days in CM patients, compared to onabotA-treatment alone, by a further 3.5–5.7 days, whilst conversely, treatment-refractory CM patients treated with onabotA and erenumab also benefited more than those receiving erenumab alone [89–91]. Therefore, one may envisage a complementary and synergistic future relationship, whereby onabotA and anti-CGRP therapies may each be useful for treating patients refractory to the other, whilst combination therapy may be reserved for highly-refractory patients. However, one must note that no direct head-to-head comparisons exist between onabotA and anti-CGRP therapies under trial conditions. Therefore, future direct trial or meta-analytical comparisons of the safety and efficacy of onabotA and anti-CGRP therapies will be critical to delineate and define their precise future relationship.

Drug summary box

Box 1. Drug summary

Drug name	OnabotulinumtoxinA, Botox
Phase	Post-authorization
Indications	UK: Chronic migraine prophylaxis in patients unresponsive to ≥3 previous pharmacological prophylactics with appropriately managed medication-overuse Europe: Symptom relief in adults fulfilling criteria for chronic migraine (headaches on ≥15 days per month, of which at least 8 days with migraine) in patients who have responded inadequately or are intolerant of prophylactic migraine medications US: Prophylaxis of headaches in adult patients with chronic migraine (≥15 days per month with headache lasting 4 hours a day or longer)
Pharmacology description/mechanism of action	Direct inhibition of peripheral sensitisation and indirect inhibition of central sensitisation of the trigeminovascular system, by intraneuronal cleavage of presynaptic synaptosomal-associated fusion attachment protein of 25 kDa (SNAP25), leading to reduced proinflammatory neurotransmitter and neuropeptide release and nociceptive ion channel downregulation
Route of administration	Licensed administration regimen: 155–195 units via 31–39 intramuscular injections into head/neck muscle groups for all countries (excluding USA) PREEMPT paradigm: 12-weekly intramuscular administration into 7 head/neck muscle groups (155 total units in 31 sites)
Chemical structure	Purified neurotoxin complex (protein) derived from Clostridium botulinum
Pivotal studies	PREEMPT 1, PREEMPT 2, COMPEL, REPOSE

Declaration of interest

F Ahmed has been on the advisory board of Allergan, Teva, Lundbeck, Novartis, and Eli Lilly for which he has been paid honorarium donated to registered charities i.e. British Association for the Study of Headache, Migraine Trust and the Anglo-Dutch Migraine Association. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

AbbVie provided a scientific accuracy review at the request of the journal editor.

Reviewer disclosures

A reviewer on this manuscript has disclosed that they consult for Allergan. All other peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This paper was not funded.