European headache federation guideline on the use of monoclonal antibodies acting on the calcitonin gene related peptide or its receptor for migraine prevention

The search strategy was formulated taking into account the PICO and clinical questions. A systematic review of the literature was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [19]. We identified key papers on the use CGRP mAbs in patients with migraine. An initial literature search included all papers indexed on PubMed and Scopus, from inception to April 2, 2018. The systematic literature search was repeated at the end of the consensus procedure to include all relevant papers published until November 2018. The following search string was used in both databases: “migraine OR headache AND (CGRP OR eptinezumab OR erenumab OR fremanezumab OR galcanezumab)”. Two investigators independently screened the titles and abstracts of the publications identified to verify study eligibility. Literature screening was conducted in two steps. In the first step, studies were excluded after reading the title and the abstract for clear exclusion criteria. For studies that passed the first step, the full text was assessed to decide about inclusion/exclusion. Disagreements were resolved by consensus. The reference lists and Google Scholar citations of the selected articles were also screened. The reasons for exclusion were recorded and summarized. To summarize the search results, a data extraction sheet was developed including the information of interest. Papers retrieved from the literature search as well as summary tables were shared among the panelists.

A general description of the study was extracted for each publication. We extracted first author name and year of publication, full citation, study design and setting, study period, number of included patients, diagnostic criteria for migraine, migraine type, treatment type, duration of observations and treatments, study results. Data extraction was performed by a single researcher (SS) and double checked.

Inclusion and exclusion criteria were selected prior to the literature search.

We included 1) observational (prospective) and intervention studies in which an CGRP mAb was assessed as possible treatment strategy for migraine prevention; 2) studies published in English or in other languages if a reliable translation could be obtained; 3) reliable criteria to diagnose migraine; 4) treatment for migraine prevention with any form of CGRP mAb; 5) reporting any outcome referring to migraine frequency, severity, duration, disability, or use of drugs to treat the acute attacks before and after treatment or in treated and untreated patients. Whenever different studies referring to the same population of patients were available we included the study reporting the outcome of interest for the specific analysis or with the largest population or with the longest follow-up; different studies can be eligible for different questions. We excluded studies 1) with observational designs not reporting outcomes with treatment or not comparing at least two treatment strategies; 2) performed in patients with headache other than migraine; 3) not reporting information on the outcomes selected for the PICO and clinical questions; 4) published only in the form of abstracts or presented at conferences only.

For each of the selected studies one author (SS) addressed the quality of evidence. Quality of evidence was addressed for single studies and for selected outcomes according to the GRADE approach [18]. Randomized trials were considered as high quality of evidence but their quality was downgraded in the case of study limitations such as lack of allocation concealment, lack of blinding, incomplete accounting for patients and outcome events, selective outcome reporting bias, or other limitations such as inadequate sample or lack of sample size calculation [20]. Observational studies were considered as low quality of evidence but their quality was upgraded in the case of large magnitude effects, dose-response gradient, if plausible confounding can increase confidence in the estimate or other considerations [21]. Final quality of evidence was rated as high, medium, low or very low based on study design, study limitations, inconsistency, indirectness, imprecision, publication bias, effect size, dose response and confounding [18]. Summary of findings tables were drafted using the GRADE pro statistical software considering all the outcomes considered important or critical. For the analysis of extracted data we used R statistical software. Data analysis was performed on a random-effects basis and results were summarized as risk ratio (RR) and 95% confidence intervals (CI).

The consensus process was performed according to the Delphi method [22]. Development of the consensus statement was organized into rounds. In each round, panelists were instructed not to discuss among themselves and to send their feedback only to the facilitator (SS). The facilitator collected all the answers and issued for each round an anonymized report with comments. Rounds were repeated until a final consensus was reached among Panel members.

Strength (strong or weak) and direction (for or against) of recommendation were determined on basis of balance between desirable and undesirable effects, quality of evidence, values and preferences and costs [18]. If GRADE was not applicable, an ungraded good practice statement based on experts’ opinions was given, according to the available level of evidence.

We found 14 eligible studies which evaluated whether treatment with CGRP mAbs as compared to placebo is effective and safe [26, 27, 31‐36, 41‐45, 50]. Among the eligible studies one was on eptinezumab [32], five studies on erenumab [35, 36, 44, 45, 50], four studies on fremanezumab [26, 27, 34, 41], and four studies on galcanezumab [31, 33, 42, 43]. One study phase IIIb study on erenumab was not included in the PICO question 1 because it included only patients with previous drug failure [50].

Available studies indicated that erenumab, fremanezumab, and galcanezumab are effective for prevention in patients with EM. They reduce the number of headache or migraine days, reduce the number of days using acute medications, improve disability. Evidence for erenumab, fremanezumab, and galcanezumab is based on phase II and III RCTs. For eptinezumab benefits are not entirely clear and improvement was significant only in the reduction of medications used for acute attacks; additionally, evidence is based on an exploratory phase II RCT. Eptinezumab is administered via intravenous injection while erenumab, fremanezumab, and galcanezumab are administered via subcutaneous injections. Ease of use represents a potential advantage as CGRP mAbs offer the convenience and adherence benefits of monthly or quarterly dosing allowing avoidance of the daily pill burden. Treatment effect was evident after the first injection and patients continued to improve within the fifth month of treatment [41‐43]. The quick onset of action is a potential advantage of CGRP mAbs as compared to conventional treatments. Reduction in migraine days with CGRP mAbs were only modest and ranged from 1 to 2 when compared to placebo. However, the absolute effect of treatment was larger considering also the placebo effect. Perhaps, more clinically significant is the at least 50% responder rate, which was consistently increased with treatment in a clinically meaningful way. A proportion of patients may have a 100% response rate to CGRP mAbs [39]. The open-label extension of the phase II RCT of erenumab reported low discontinuation rates [24] which is in contrast to current migraine prophylactics that are associated with high discontinuation rates [8, 51, 52]. Post-hoc analyses of the RCTs indicated that treatment with fremanezumab is associated with improved normal function performance on headache free days [46] and that treatment with galcanezumab is associated with overall functional improvement [23]. At the moment, it cannot be determined whether unique patient populations will have a response to a specific drug.

Data from RCTs indicated that the CGRP mAbs are safe. No relevant SAEs were registered. One death occurred in the phase III RCT on fremanezumab [34] and one death occurred in the open label extension trial on erenumab [24]. Both deaths were considered unrelated to the study drugs. However, it should be noted that further data from the real-life setting are needed to support safety and to provide information on the long-term use.

We found four eligible studies which evaluated whether treatment with CGRP mAbs as compared to placebo is effective and safe [26, 31, 41, 45]. Among the eligible studies one study was on erenumab [45], two studies on fremanezumab [26, 41], and one on galcanezumab [31].

Available studies indicate that erenumab, fremanezumab, and galcanezumab are effective for prevention in patients with CM. They reduce the number of headache days, reduce the number of days using acute medications, improve disability, and are safe. For erenumab evidence is based on a phase II RCT which however was not a dose finding exploratory study but a RCT to assess safety and efficacy. For fremanezumab evidence is based also on phase II and on a phase III RCT while for galcanezumab it is based on a phase III RCT. Studies included patients with a long history of disease and those who had previously failed two or more preventive medications. The trials did not include patients with more refractory disease such as those who had not had a response to two clusters of preventive medications.

Characteristics of patients according to migraine duration and previous use of preventive drugs is reported in Tables 1 and 2. In all the trials, included patients had a long migraine history less than 15 years. RCTs included patients who had not tried any previous preventive strategy, patients who had failed or not tolerated other preventatives. Patients considered as drug-resistant where on the other hand excluded.

Referring to the RCTs on EM, the phase II RCT on eptinezumab did not exclude patients according to previous failure of preventive drugs [32]. All the others RCTs in EM excluded patients who failed 2 to 4 categories of preventive drugs. For erenumab, patients who had previous medication failure represented 26% to 40% [35, 36, 44], for fremanezumab they were 27–33% [27, 34] and for galcanezumab 18–19% [33‐43]. A phase IIIb study, the LIBERTY trial, evaluated the efficacy of erenumab 140 mg monthly dose as compared to placebo in patients with EM who had been treated unsuccessfully (in terms of either efficacy or tolerability, or both) with between two and four preventive treatments [50]. At 3 months, significantly more patients in the erenumab group than in the placebo group had a 50% or greater reduction from baseline in the mean number of monthly migraine days. Erenumab was also significantly more efficacious than placebo for all secondary endpoints, including improvements in migraine frequency, medication use, and functional outcomes.

Referring to the RCTs on CM, the phase II RCT on erenumab, excluded patients who had no therapeutic response on an adequate trial of > 3 preventive medications [45]. In the overall study population, 74% of patients had previously received preventive treatments [25]; among them, 48 to 50% of patients had failed ≥2 preventive drugs, 66 to 70% had failed > 1 preventive drug and 30 to 34% had no drug failure; 47 to 52% of patients had previous use of topiramate. OnabotulinumtoxinA injections for migraine prevention were prohibited during the study and for at least 4 months before the start of the baseline phase; 23 to 26% of patients had used onabotulinumtoxinA before study entry. MOH was explicitly allowed. Mean monthly acute migraine-specific drug use days was around 9. Treatment differences for erenumab versus placebo were numerically greater in patients with ≥1 or ≥ 2 failed preventive medications than in patients with no prior treatment failure, particularly for the 140 mg dose [25]. The effect was attributable to a lower placebo effect in patients with prior medications failure than in patients without medications failure.

The phase II and III RCTs on fremanezumab in CM excluded patients who had no therapeutic response of ≥2 preventive medications [26, 41]. The phase III study explicitly excluded patients with unremitting headaches (headaches for more than 80% of the time they were awake, and less than 4 days without headache per month); daily headache was acceptable if patients had headaches on less than 80% of the time they were awake on most days [41]. In the study, 28 to 31% of patients had previous use of topiramate. In both trials, onabotulinumtoxinA injections were prohibited during the study and in the 4 [41] or 6 [26] months before study entry. In the phase III study, 13 to 18% of patients had used onabotulinumtoxinA before study entry [41]. Patients with medication overuse headache (MOH) were allowed in the study [26, 41]. Days of acute drug use per month ranged from 15 to 16 in one study [26] whereas mean 28-day days of use of any acute headache medications was around 13 in the other study [41].

The phase III RCT on galcanezumab in CM excluded patients who had no therapeutic response of > 3 preventive medications [31]. OnabotulinumtoxinA injections were prohibited during the study. Patients who had failed 2 or 3 previous preventive treatments represented 24–35% of the study population. Patients with MOH were allowed in the study and represented 63–64% of all the study population. In this study patients without headache free days were excluded.

In EM, CGRP mAbs were evaluated both in patients with and without previous drug failure. So far, in most of the available phase II and phase III RCTs, participants with previous failure of as few as 2 preventive medication classes for migraine were excluded. This implies that efficacy can be different for patients with severe, treatment-resistant migraine. Only in the LIBERTY study on erenumab 140 mg monthly patients treated unsuccessfully with between two and four preventive treatments were included. The study confirmed effectiveness of erenumab in this subgroup of patients. However, no results were provided for patients stratified according to previous preventive failure versus non tolerability.

In CM, erenumab, fremanezumab, and galcanezumab were evaluated both in patients with and without previous drug failure. Data on erenumab indicated that the drug is effective even in patients with failure to previous drugs. Patients who had previous use of onabotulinumtoxinA were included in RCTs but no information referring to previous efficacy of onabotulinumtoxinA and response to study treatment is available. Erenumab, fremanezumab, and galcanezumab were not evaluated in patients with CM refractory to current available medical treatments. However, due to the poor quality of life of patients with refractory CM it is reasonable to treat them in daily clinical practice with erenumab, fremanezumab, or galcanezumab. Post-marketing studies are needed to provide information about efficacy of CGRP mAbs in refractory CM.

Costs of the CGRP mAbs are not yet entirely known but they will be higher as compared to costs of the other available drugs. Pharmacogenomics studies should provide analyses to consider the economic impact of those drugs taking into account the overall direct and indirect costs related to untreated migraine or to migraine treated with the available drugs. Differences in reimbursement and regulations among countries will probably be present. Efficacy, safety, good tolerability profile and ease of use may represent advantages of CGRP mAbs drugs which may lead patients to prefer those drugs as first-line options. Rather than only efficacy, CGRP mAbs have advantages referring to side effects and treatment administration. Poor response in patients with migraine may also be attributed to lack of compliance to available medical treatments because of the need of taking multiple doses of the drugs or side effects. CGRP mAbs may represent suitable options for patients who have contraindications to other preventive treatments because of comorbidities or side effects and in patients who have poor compliance to other treatments where strategies to improve compliance have failed. However, due to high costs it will not be possible to offer those drugs to all patients with migraine requiring preventive treatment. At the moment, limiting prescription to patients with prior drug failure may represent a reasonable option until pharmaeconomics studies will provide more data. It is important to point out that patients with multiple drug failures were mostly excluded by RCTs. It is important to note that early treatment of patients with high frequency EM may prevent CM with important impact on individuals and society.

Final recommendations based on experts’ opinions are reported in Table 19.

Summary about concomitant preventive treatments in the RCTs is available in Tables 1 and 2. In the RCT with eptinezumab, preventive drugs were not allowed [32]. In the phase II RCTs on erenumab in EM and CM preventive treatments were not allowed [44, 45] whereas in the phase III RCTs preventatives were allowed but a low proportion of patients (2–7%) had concomitant use [35, 36]. All RCTs on fremanezumab allowed the inclusion of those patients [26, 27, 34, 41]. Concomitant users of preventive drugs ranged from 20 to 34% for EM [27, 34] and from 20 to 43% for CM [26, 41]. Notably, a sub-analysis of patients using fremanezumab as an add-on treatment [30] and pooling together data of patients with EM and CM showed that among patients who received fremanezumab as add-on to their preventive treatment there was a significant decrease in the number of migraine days relative to placebo (− 4.1 versus − 2.5), an increase in the number of patients who had improvement by 50% or more of migraine days (40% versus 24%; P = 0.0505) and a reduction in the mean number of days using acute medications (33% vs 26%). Patients who were taking oral preventive drugs were not included in the RCTs on galcanezumab in EM [33, 42, 43] but were allowed in the RCT in CM [31].

We have scarce information on how to manage other oral preventive treatments in association with anti-CGRP mAb in patients with migraine. No interaction is supposed by CGRP mAbs and available preventive treatments. Data on erenumab and fremanezumab suggest that the two drugs are beneficial also when added to ongoing oral preventive treatment. Combined use of other prophylactics and CGRP mAbs may be considered in patients with insufficient response to a single type prophylactics. If patients are on preventive drugs that do have some but not sufficient effect, anti-CGRP antibodies can be added because no interaction is expected. When a possible efficacy of anti-CGRP mAb is established in a given patient it should be discussed with the patient whether withdrawal from the oral prophylactic drug should be tried.

In patients with CM, it is reasonable not to stop current ongoing migraine preventive drugs in patients before initiating the use of erenumab, fremanezumab, or galcanezumab in order to avoid possible rebound effects. Withdrawal of other preventive drugs may be done later in patients showing favorable clinical response after starting anti-CGRP mAb. A further point is to clarify, in patients with CM who had favorable response to anti-CGRP mAb but who may continue to experience a significant burden of migraine attacks if adding-on any preventive strategy may further improve attacks frequency, attacks severity, use of preventive drugs and quality of life. At the moment, no such information is available but it is reasonable to allow the use of additional preventive drugs where prevention with anti-CGRP mAb is still considered not optimal.

No information on current use of erenumab, fremanezumab, and galcanezumab with onabotulinumtoxinA is available and this association is not supported at the moment. For those patients who are on botulinum toxin and who show an inadequate response, withdrawal of onabotulinumtoxinA with start of the anti-CGRP mAb may be considered. While in the trials there were time restriction referring to onabotulinumtoxinA withdrawal and start of the anti-CGRP mAb, they represented procedures to avoid confounders and are not reasonable in daily clinical practice. At the moment, we do not know whether it is reasonable to consider combining onabotulinumtoxinA with anti-CGRP mAb in patients who have a suboptimal response to each of those drugs.

Final recommendations based on experts’ opinions are reported in Table 19.

For eptinezumab duration of treatment in the available RCT was 3 months [32].

For erenumab in available RCTs duration of treatment ranged from 3 months to 6 months for EM and from 3 months to 3 months for CM [35, 36, 44, 45]. In the open-label extension of the phase II RCT in EM in patients who had completed the 1-year open-label follow-up, persistent benefits were reported for patients who continued treatment up to 1 year [24]. There was a fatal event related to atherosclerosis and a non-fatal myocardial ischemia in patients treated with erenumab 70 mg [24]. The fatal event was considered not related to erenumab while for the other event conclusion was uncertain. A post-hoc analysis of data form the STRIVE and the phase II study in CM, showed evidence of onset of efficacy of erenumab during the first week of treatment [40]. At week 1, 43% of EM patients and 26% of CM patients in the erenumab 140 mg group experienced a ≥ 50% reduction in weekly migraine days (15% increase vs placebo for EM and 10% increase vs placebo for CM). For fremanezumab duration of treatment was of 3 months in all the trials in EM and CM [26, 27, 34, 41]. A post-hoc analysis of patients treated with fremanezumab in the phase II studies indicated that fremanezumab may be associated with sustained efficacy in a substantial percentage of those who show an initial response; sustained response is less obvious in patients with CM than in patients with EM [37]. For galcanezumab duration of treatment in available RCTs in EM was from 3 up to 6 months [31, 33, 42, 43]. A pooled analysis of data on galcanezumab in patients with EM (EVOLVE-1 and EVOLVE-2 parallel studies) or CM (REGAIN study) examined the likelihood of response with continued galcanezumab treatment in patients with EM or CM without initial clinical improvement [49]. In patients with EM having “modest” early improvement at month 1, 62% achieved “good” and 20% achieved “better” responses with continued treatment. A percentage of patients with “limited” (43%) or “minimal/no” (34%) early improvement, or “worsening” (20%;) achieved a “good” response after continued treatment. In patients with CM, having “modest” early improvement, 38% achieved “good” and 13% “better” responses with continued treatment. A “good” response was achieved for a percentage of patients with “minimal/no” early improvement (17%). Similar patterns were observed for those without clinical response at month 2, though percentages were lower.

An open label study which evaluated safety and tolerability of galcanezumab 120 and 240 mg in patients with episodic and chronic migraine provided information on treatment at 1-year. In the study authors found high study completion rate (77.8%) supporting the tolerability of the study drug through all 12 months of treatment [47]. In patients who completed the study, treatment compliance was > 95%. Furthermore, the percentage of discontinuations due to adverse events was low (< 5% combined doses), and few SAEs occurred (< 4% combined doses, and none considered related to treatment). A further analysis from this same study indicated that treatment with galcanezumab led to high levels of satisfaction and high levels of preference and less side effects versus previous treatments [48].

As a general rule, treatment can be stopped if migraine is considered too infrequent to justify preventive treatment or if treatment is considered not effective.

Data from the available trials suggest that the effective reduction of monthly headache or migraine days due to treatment with CGRP mAbs may be observed very early, after less than one month from the first dose. Data from RCTs suggest that patients may have additional benefits with continuation of treatment and that some patients who have worsening with treatment or who are considered non-responders may have improvement with continuation of treatment. For those reason it is reasonable not to stop treatment before 3 months even in the absence of a clinical response. Further studies are needed to better assess whether some patients might have even a more delayed response to CGRP mAbs, and to provide information about the durability of the response to treatment with CGRP mAbs. Further data are also needed to clarify whether the response may be sustained even after withdrawal of the CGRP mAbs. For the moment it is reasonable to manage the duration of treatment with CGRP mAbs not differently to other available preventive strategies and to continue it for at least 6–12 in patients who have beneficial effects with those drugs.

Factors contributing to response/nonresponse have yet to be elucidated and clinical judgment should be exercised when deciding whether to discontinue treatment.

Tachyphylaxis of preventive treatments for migraine is a frequent problem in the clinical setting. A post-hoc analysis of patients treated with fremanezumab in the phase II study supported a sustained efficacy, over the 3-month trial period, in a substantial percentage of those who show an initial response [37]. One-year interim analysis of a phase II study of erenumab 70 mg suggest that benefits persist over time [24].

Final recommendations based on experts’ opinions are reported in Table 19.

All the available RCTs on CM included patients with MOH [26, 31, 41, 45]. No subgroup analysis of efficacy and safety was performed for patients with MOH.

We have no direct data about the impact of MOH on the treatment of CM with CGRP mAbs. However, the available RCTs of erenumab, fremanezumab, and galcanezumab all enrolled consistent proportions of patients with untreated MOH. Therefore, it might be reasonable to offer treatment with CGRP mAbs to patients with MOH. We have, at this moment, no evidence to indicate that the effect of CGRP mAbs is increased if preceded by detoxification and further research is needed on this issue. Some adopt withdrawal strategies before offering preventive medications to patients with CM and MOH and some of the available evidence indicate that detoxification is feasible and effective [53]. However, detoxification is not easy and feasible with all patients and dedicated resources, which are not always available, are needed. We have no data which indicate if the use of CGRP mAbs may favor detoxification in patients with CM and MOH. Of note, the frequent use of butalbital-containing medications was an exclusion criterion from the trials; therefore, current evidence suggests avoiding the overuse of butalbital before starting treatment with CGRP mAbs.

Final recommendations based on experts’ opinions are reported in Table 19.

Criteria varied across trial but pregnant or nursing women, alcohol or drug abuse, cardio and cerebrovascular diseases, and severe mental disorders were the most relevant conditions.

CGRP mAbs are unlikely to produce drug interactions or affect the course of ongoing disease which may be particularly relevant in patients with comorbidities. CGRP is the most potent vasodilator peptide known [54] and has been theoretically considered as dangerous in patients with diseases of the vascular system. In the cardiovascular system, CGRP is present in nerve fibres that innervate blood vessels and the heart and participates in the regulation of blood pressure [55]. For this reason, patients with cardio and cerebrovascular disease were excluded from available clinical trials. In available studies, there is no evidence of increased cardiovascular events or any other serious concerns. However, the duration of available studies is much shorter than the duration in the clinical settings and registries should record any SAEs to see the long-term effects of continuous blockade of the CGRP pathway.

Additionally, there was no effect on treadmill exercise time in patients with angina who received telcagepant, a small-molecule CGRP antagonist [56]. These results supplement those from a placebo-controlled study of erenumab in a high-risk population of patients with stable angina with a median age of 65 years, in which inhibition of the canonical CGRP receptor with erenumab did not adversely affect total exercise time in a treadmill test, among other safety endpoints [57].

Long-term safety studies with CGRP mAbs are needed to further characterize potential cardiovascular effects. More data from migraine patients with comorbid cardiovascular conditions in a real-world setting may help further assess the theoretical cardiovascular risk of blocking the CGRP pathway.

Final recommendations based on experts’ opinions are reported in Table 19.

The occurrence of binding or neutralizing antibodies against the anti-CGRP mAb is reported in Table 20. It is important to note that some of the patients in the RCTs were positive for those antibodies before initiation of the study drug and that there were antibody positive patients even in those taking placebo.

Data from individual studies indicate that binding and/or neutralizing antibodies occur infrequently and may have a variable course over time. At the moment, the presence of binding and/or neutralizing antibodies has not been associated with poor response to treatment or adverse events. Consequently, there is no evidence which may support the need of antibodies testing in routine clinical practice. However, this issue should be further studied. In fact, duration of treatment in available studies is limited in time and it cannot be excluded that the rate of occurrence of binding and/or neutralizing antibodies in available clinical studies was too low to establish firm conclusions about their possible implications. Pooled data from available RCTs or data from real life studies may add better evidence and further research should clarify the role of binding and/or neutralizing antibodies in patients with poor clinical response and side effects.

Final recommendations based on experts’ opinions are reported in Table 19.