Responder rates with eptinezumab over 24 weeks in patients with prior preventive migraine treatment failures: post hoc analysis of the DELIVER randomized clinical trial

Abstract Background and purpose Eptinezumab reduced monthly migraine days more than placebo in the DELIVER study, a clinical trial with patients with difficult‐to‐treat migraine and prior preventive treatment failures. This post hoc analysis assesses the sustained response to eptinezumab at the population and patient level and evaluates the potential for response in initial non‐responders. Methods Adults with chronic or episodic migraine and two to four prior preventive treatment failures were randomized to eptinezumab 100 mg, 300 mg or placebo every 12 weeks. Primary outcomes in this post hoc analysis are the proportion of patients with ≥30%, ≥50% or ≥75% reduction in monthly migraine days (i.e., migraine responder rates [MRRs]) during weeks 1–12 and weeks 13–24 and across 4‐week intervals. Secondary outcomes are maintenance and shifts in MRRs from weeks 1–12 to weeks 13–24. Results Between weeks 1–12 and 13–24, ≥30% MRRs increased from 65.9% to 70.4% (100 mg) and from 71.0% to 74.5% (300 mg), versus 36.9% to 43.1% (placebo). The ≥50% and ≥75% MRRs were sustained or increased over the 24‐week period. The largest increase in ≥30% MRRs occurred after the second infusion with eptinezumab. The percentage of initial non‐responders (<30% MRRs during weeks 1–12) who experienced response (≥30% MRRs during weeks 13–24) to the second dose was 34.7% (100 mg) and 30.4% (300 mg) with eptinezumab versus 21.1% with placebo. Conclusion Across MRR thresholds, most patients who responded to eptinezumab during weeks 1–12 maintained or improved response during weeks 13–24. More than one‐third of initial non‐responders became responders after their second infusion.


INTRODUC TI ON
Migraine is a burdensome, prevalent neurological disorder [1] that reduces an individual's ability to participate in society [2].
Discontinuation of traditional oral preventive medications is common due to insufficient efficacy or poor tolerability [3].There remains an unmet need for preventive treatment that provides robust, well-tolerated and sustained migraine prevention that reduces monthly headache day frequency and headache-related disability as well as improving health-related quality of life.Several treatments targeting calcitonin gene-related peptide (CGRP) or its receptor have been approved in recent years for the preventive treatment of migraine with the goal of reducing migraine frequency and associated burden, increasing health-related quality of life and improving patient adherence to treatment [4][5][6][7][8].
In many countries, people suffering from migraine must fail multiple preventive treatments before they are prescribed an anti-CGRP monoclonal antibody (i.e., eptinezumab, erenumab, fremanezumab and galcanezumab) [9][10][11][12].The European Headache Federation and the European Academy of Neurology endorse these anti-CGRP monoclonal antibodies as third-line preventive medications [13].Guidance from the National Institute for Health and Care Excellence (NICE) in the United Kingdom recommends failure of at least three preventive migraine treatments prior to prescribing erenumab [10], fremanezumab [11] and galcanezumab [12], with eptinezumab guidance in development [9].The French National Authority for Health (Haute Autorité de Santé), Canadian Agency for Drugs and Technologies in Health (CADTH) and the US Institute for Clinical and Economic Review recommend at least two prior preventive treatment failures [14][15][16][17][18][19][20].
Continued preventive treatment is contingent on the level of response experienced over 3-6 months, with assessments made after 3 months of treatment and response typically meaning the reduction in monthly migraine days (MMDs) since starting treatment [21].

CADTH recommends stopping anti-CGRP treatment if the reduction
in MMDs is <50%, although it notes that, for renewal, "some jurisdictions may want to include a reduction of at least 30% in the number of headache days per month" [16][17][18][19].In patients with chronic migraine, NICE recommends stopping treatment if response is <30% after multiple treatment cycles (vs.<50% in episodic migraine), in alignment with the position of the International Headache Society Clinical Trials Subcommittee and the Danish Medical Council, which deemed a ≥30% migraine responder rate (MRR) in chronic migraine to be clinically meaningful [22][23][24].These reductions in MMDs characterized as MRRs are known as MRR thresholds.
Eptinezumab is a humanized monoclonal antibody that inhibits CGRP and is indicated for the preventive treatment of migraine in adults in the United States, Europe and other regions [25].The efficacy and safety of eptinezumab was established in multiple large-scale phase 3 clinical trials, which demonstrated rapid and sustained preventive effects in patients with episodic and chronic migraine [26][27][28].One of these studies, DELIVER, was a phase 3b clinical study to evaluate the efficacy and safety of eptinezumab for migraine prevention in patients with two to four prior preventive treatment failures [28].The primary data from the placebo-controlled portion of DELIVER showed reductions in migraine frequency and severity with acceptable safety and tolerability in patients with difficult-to-treat migraine [28].
Given differences in prescribing and renewal guidelines for anti-CGRP monoclonal antibodies across countries, it is important to understand the response for patients treated with these medications across a variety of thresholds.The aim here is to illuminate the expectations for the maintenance and shift in migraine responder status (i.e., 30%-49%, 50%-74% or ≥75% MRR) in patients with migraine and two to four prior treatment failures treated with eptinezumab versus placebo in the DELIVER study.The objective of these analyses was to evaluate the sustained response at both the population level and patient level during the 24-week, placebo-controlled portion of DELIVER.In addition, these analyses assessed the proportion of response in patients who were initially non-responders.

Study design
DELIVER was a multinational, multicenter, parallel-group, doubleblind, randomized, placebo-controlled phase 3b clinical study.The protocol and statistical analysis plan have been previously published, including full details of the study design, inclusion and exclusion criteria, efficacy and safety outcomes, and statistical considerations [28].Some methodological details are briefly summarized here.

Patients
Eligible patients included adults aged 18-75 years with onset of migraine at ≤50 years of age and with a history of migraine [29] for ≥12 months before screening.Patients had to have documented evidence of two to four prior preventive migraine treatment failures of different classes (i.e., propranolol/metoprolol, topiramate, amitriptyline, flunarizine, candesartan, valproate/divalproex and botulinum toxin A/B) in the past 10 years.Documented evidence included medical record or physician's confirmation, and at least one prior treatment failure must have been due to inadequate efficacy, tolerability reasons or contraindications.Patients with prior treatment failure with a CGRP inhibitor history or diagnosis of other headache types (e.g., chronic tension-type headache, cluster headache etc.) or history of clinically significant cardiovascular disease were excluded.

Treatments
Patients were randomized equally to eptinezumab 100 mg, eptinezumab 300 mg or placebo, administered by intravenous infusion at baseline (day 0) and week 12. Randomization was stratified by monthly headache days at baseline (≤14 or >14) and by country.The intravenous infusion was administered over 30 min.
To capture headache occurrence, patients completed a daily electronic diary (eDiary) with an evening report (completed daily even in the absence of headache) and a headache report (completed for each headache).Patients had four scheduled in-person visits (screening, baseline, end of week 12 and end of week 24) and four telephone contact visits (end of weeks 4, 8, 16 and 20).

Statistical analysis
Full details of sample size calculations from the DELIVER trial have been published [28].The power was determined by simulating the testing strategy, assuming normal distributions with means and standard deviations for continuous end-points and success rates.
Briefly, with 280 patients per treatment group, the power was calculated to be at least 90% for the primary end-point and at least 68% for individual key secondary end-points (i.e., ≥50% and ≥75% were calculated by prorating if the eDiary was completed on at least 14 of the 28 days of each 4-week period.A missed day was defined as one on which the patient did not complete the evening report or the headache report.Missing data were imputed over 4-week periods using MMDs divided by the number of days with observations × 28 as a 1-month score for that 4-week period.For analyses of the shifts in response threshold, patients with a non-missing migraine responder status for weeks 1-12 and weeks 13-24 were evaluated.

MRRs
Comparisons between active treatment and placebo were based on the odds ratios of response using the likelihood ratio test.p values for key secondary end-points (i.e., ≥50% and ≥75% MRRs during weeks 1-12) were based on two-sided tests controlled for multiplicity.The DELIVER study was not designed to investigate differences between different doses of active treatment but rather between active treatment and placebo.p values for all other end-points are presented with nominal p values and 95% confidence intervals, with no control for multiplicity; multiplicity was not controlled for beyond the end-points included in the predefined statistical hierarchy.All statistical analyses were conducted using SAS (version 9.4 or later).

RE SULTS Patients
Comprehensive details of study design, patient disposition, demographics, baseline characteristics, primary and key secondary efficacy end-points and safety/tolerability outcomes have been published previously [28].The full analysis set comprised 890 patients (100 mg, n = 299; 300 mg, n = 293; placebo, n = 298), and 97.2% (865/890) completed the 24-week placebo-controlled treatment period.Withdrawal due to lack of efficacy comprised 0.4% (4/891) of patients.Patients with a non-missing responder status for both weeks 1-12 and 13-24 included 287 patients who received eptinezumab 100 mg, 286 patients who received eptinezumab 300 mg and 295 patients who received placebo.
The mean (standard deviation) age of the full analysis set was 43.8 (10.6) years.A total of 800 patients (89.9%; 800/890) were female and 854 (96.0%; 854/890) were White.All but one patient had experienced at least one prior treatment failure due to lack of efficacy, 494/890 (55.5%) had at least one prior treatment failure due to safety/tolerability concerns and 28/890 (3.1%) had at least one prior treatment failure due to contraindication.
During the 28-day baseline period, mean (standard deviation) MMDs were 13.8 (5.6) in the eptinezumab 100 mg group, 13.7 (5.4) in the eptinezumab 300 mg group and 13.9 (5.7) in the placebo group.Based on the number of migraine and headache days during the baseline period, 484/890 (54.4%) patients had episodic migraine and 405/890 (45.5%) had chronic migraine; one patient did not fit into either group after re-calculation of baseline values.
During the first dosing interval (weeks [1][2][3][4][5][6][7][8][9][10][11][12] a second infusion of active treatment.An exploratory analysis identified that this increase in effect was observed in the episodic migraine population, and maintenance of the increased effect was observed in the chronic migraine population (Table S2).Although the recommended threshold for clinical response is ≥50% MRR for episodic migraine, ≥30% MRR was analyzed here for comparison to chronic migraine.
The ≥50% MRRs (Figure 2; Table S3) and ≥75% MRRs (Figure 3; Table S4) followed a similar pattern to ≥30% MRRs, wherein eptinezumab treatment resulted in higher rates of response than placebo in each 4-and 12-week interval, with the largest increase in MRR occurring in the 4 weeks following the second infusion of eptinezumab (i.e., between weeks 9-12 and 13-16) for both the 100 mg and 300 mg doses.

DISCUSS ION
Given the differences in prescribing guidelines and criteria for treatment initiation and discontinuation for anti-CGRP monoclonal antibodies across countries, it is important to understand the potential trajectory of response for patients treated with these medications across a variety of thresholds.The data presented here may guide clinicians to continue treatment with eptinezumab for at least two infusions before making dosage change evaluations or discontinuing treatment.The current analysis illuminates the potential expectations for the maintenance and shift in migraine responder status in patients with migraine and multiple prior treatment failures treated with eptinezumab versus placebo in the DELIVER study.The results highlight that significantly more patients treated with eptinezumab than with placebo will experience migraine response and that a substantial proportion of eptinezumab-treated patients will experience a maintained or improved migraine response over 24 weeks of treatment.Additionally, approximately one-third of eptinezumab-treated patients with insufficient migraine response during the first dosing interval (weeks 1-12) responded after a second infusion (weeks 13-24) in both the 100 mg and 300 mg treatment groups.
The ≥50% and ≥75% MRRs during weeks 1-12 and 13-24 in the eptinezumab groups were consistent with data from the pivotal PROMISE studies in episodic [30] and chronic migraine [31], but placebo MRRs in the PROMISE studies were higher than observed in this study where patients have experienced two to four prior preventive treatment failures.A post hoc analysis of the PROMISE studies found that 37.6% (100 mg), 36.3% (300 mg) and 33.9% (placebo) of patients with <50% migraine response during weeks 1-12 in PROMISE-1 experienced ≥50% response during weeks 13-24; in PROMISE-2, these percentages were 28.7% (100 mg), 29.0% (300 mg) and 18.5% (placebo) [32].Thus, across three trials of eptinezumab in different migraine populations, approximately a third of eptinezumab-treated patients who did not experience response during the first infusion experienced migraine response after a second infusion, highlighting the importance of a series of at least two treatments.Whilst eptinezumab does has a rapid onset of action [33,34], related to its intravenous route of administration, some patients, due to the natural fluctuation of disease [35], may require additional time with preventive treatment to gain the full benefits.
DELIVER comprised patients with episodic or chronic migraine with documented evidence of two to four prior preventive treatment failures.Patients with multiple treatment failures could be considered to have resistant or refractory migraine [21].Resistant migraine is defined as at least three failures of different preventive migraine medication classes with at least eight debilitating headache days per month for at least three consecutive months; preventive treatment refractory migraine is defined as failure of all available preventive migraine medication classes with at least eight debilitating headache days per month for at least six consecutive months [21].Patients with resistant or refractory migraine experience a negative impact on quality of life and personal, professional and social activities, as well as increased healthcare costs [2], and patients can feel hopeless about their disease state and remission.Resistant or refractory migraine may be more difficult to treat, and although the demographic was not captured in this study a proportion of the DELIVER population may have met resistant or refractory migraine criteria.The ≥30% MRR threshold has been deemed clinically meaningful by the International Headache Society Clinical Trials Subcommittee [22,23] and is a treatment-renewal guideline per NICE and CADTH [10][11][12][16][17][18].Beyond guidelines, whether treatment is providing meaningful benefit to the patient should be amongst the items discussed between clinicians and their patients when determining next steps with treatment.
Although the observed placebo response over 24 weeks in the DELIVER study was lower than that observed in the PROMISE studies [30,31] ).With eptinezumab, the largest percentage-point differences always occurred between weeks 9-12 and 13-16 and ranged from 10.9 to 16.6 percentage-points.The placebo response is unlikely to be biased by dropout rate, which was less than 3% of patients overall and less than 0.5% due to lack of efficacy [28].The percentage-point differences with eptinezumab may be related to a slight end-of-dose deterioration effect that is not observed with placebo; however, further analysis of longer-term data is needed to assess if the end-of-dose effect remains beyond the two infusions in the placebo-controlled period.
These differences also underscore the benefit of a second infusion with eptinezumab that is not observed with placebo.

Study limitations
DELIVER may have limited generalizability to the general migraine population, given that the study population was mainly female and White and MRRs were not analyzed by sex or race subgroups.
Individuals with previous anti-CGRP therapy failures were excluded from participation, as were those with cardiovascular disease or certain pain syndromes; thus, the findings may have limited generalizability in patients with these or other excluded conditions.
Sustained dosing of eptinezumab 100 mg was associated with incremental improvements; however, the DELIVER study did not assess dose escalation so comparative effectiveness remains to be explored.It may be of interest to clinicians and payers to determine when dosage should increase from 100 mg to 300 mg, although no data are currently available to predict which patients will respond better to the higher dose of eptinezumab [36].Additionally, intravenous infusion with CGRP antibodies may lead to a higher placebo response [37].The placebo response here increased over the course of the 6-month treatment period, which could be due to the nature and context of treatment administration and fluctuation of disease.

CON CLUS ION
In the DELIVER study, ≥30%, ≥50% and ≥75% MRRs were higher with eptinezumab than with placebo, and the MRRs over time supported a beneficial effect of a second infusion of eptinezumab.
While many healthcare providers traditionally use ≥50% migraine response as the threshold for defining clinically meaningful benefit, these data inform clinicians on rates of patients fulfilling the threshold for additional dosing in patients with migraine and prior preventive treatment failures.Further, more than one in three patients who did not achieve migraine response with eptinezumab during the first dosing interval did achieve response following the second infusion; thus, these data support that two infusions of eptinezumab may be needed before evaluating treatment outcome.

ACK N O WLE D G E M ENTS
The authors thank Emily Bolen and Nicole Coolbaugh, CMPP, of the Medicine Group (New Hope, PA, USA), for providing medical writing support in accordance with Good Publication Practice guidelines, which was funded by H. Lundbeck A/S.

FU N D I N G I N FO R M ATI O N
The clinical trial and publication were funded by H. Lundbeck A/S, Copenhagen, Denmark.
during weeks 1-12) based on simulations.The full analysis set comprised all randomized patients who received at least one dose of study medication and had at least one valid post-baseline 4-week assessment of MMDs in weeks 1-12.In this post hoc analysis, MRRs for each 4-week interval were calculated as the change from baseline in MMDs during the given interval.MRRs across three 4-week intervals were calculated as the average percentage change in MMDs, based on available monthly values.The treatment comparison was based on a logistic regression model, including baseline MMDs as a continuous covariate with treatment and stratification (monthly headache days at baseline ≤14 or >14) as factors.If the MMD value was missing for a given month, the responder status was derived based on the available values.MMDs