Fremanezumab in the prevention of high-frequency episodic and chronic migraine: a 12-week, multicenter, real-life, cohort study (the FRIEND study)

Migraine is a complex neurologic disorder characterized by recurrent disabling headache episodes associated with autonomic symptoms [1]. The conventional oral preventive migraine therapies are non-selective, non-specific, poorly tolerated, and burdened by a high discontinuation rate [2‐4]. Monoclonal antibodies (mAbs) targeting the calcitonin gene-related peptide (CGRP)—the first specific migraine prophylactic agents—are changing the scene of migraine prevention, coupling promising efficacy to an excellent tolerability profile [5]

Fremanezumab is a humanized mAbs targeting both the α and β CGRP isoforms, indicated for the prevention of episodic or chronic migraine in adults. Its peculiarity is the flexible dose regimen which allows to personalize the treatment choosing between the dose of 225 mg on monthly basis or 675 mg quarterly [6]. Fremanezumab has been extensively investigated in randomized, placebo-controlled trials (RCTs) in patients affected by episodic migraine (EM) (HALO-EM study), chronic migraine (CM) (HALO-CM study) and EM or CM with 2 to 4 prior therapeutic failures (FOCUS study), documenting a significant superiority over placebo in reducing migraine frequency, analgesic use and disability, and a good efficacy/tolerability profile also in long-term treatment trials [7‐10].

While pharmacological RCTs test drugs under ideal conditions, real-life studies explore their effectiveness, safety, and tolerability in unselected patients in routine circumstances, detecting rare or late-onset adverse events, assessing adherence and patterns of use, providing higher generalizability results, and testing new hypotheses [11]. In addition to the growing real-life evidence on erenumab [12, 13] and galcanezumab [14, 15], preliminary results from non-peer reviewed publications on retrospective, real-world studies are available for fremanezumab, documenting a reduction ranging from 68.7% to 77% in monthly migraine days (MMDs) and from 65.9% to 74.8% in monthly headache days (MHDs) in EM or CM, lower acute medication use and emergency department, and outpatient physician costs [16].

The present paper is aimed at evaluating the effectiveness, safety, and tolerability of fremanezumab in a prospective real-life, multicenter Italian study in patients affected by high-frequency episodic (HFEM: ≥ 8 MMDs) or CM.

This is a 12-week multicenter, prospective, cohort, real-life study ongoing at 9 headache centers distributed across 4 Italian regions (Lombardy, Latium, Campania, and Calabria) from July 28^th, 2020, with the latest data analysis performed on November 11^th, 2020. We considered all consecutive patients affected by HFEM or CM—according to the criteria of the International Classification of Headache Disorders, 3rd edition [17]—with indication to fremanezumab preventive treatment according to the reimbursement rules of the Italian Medicine Agency (AIFA) [18]. None of them was previously treated with any antiCGRP mAbs.

After signing the informed consent, all patients underwent a careful physical and neurological examination and were interviewed using a shared semi-structured questionnaire by specifically trained, board-certified neurologists who gathered information on socio-demographic characteristic, migraine features, past and current migraine treatments, comorbidities, and concomitant medications [12].

Patients were given subcutaneous fremanezumab at the doses of 225 mg monthly or 675 mg quarterly, according to their preference. During the 28-day run-in baseline period and the entire study duration, patients were asked to fill-out a paper–pencil diary recording MMDs for HFEM, MHDs for CM, monthly analgesic intake, and rating pain intensity of the monthly most painful attack (0–10, Numerical Rating Scale, NRS). A migraine day was defined as a calendar day characterized by > 4 consecutive hours of a migraine with or without aura or a headache of any duration successfully treated with migraine-specific acute medications (triptans). Pain disability was measured monthly using the Headache Impact Test (HIT-6) and quarterly with the Migraine Disability Assessment Scale (MIDAS).

The primary study endpoints were the change in MMDs for HFEM and MHDs for CM at weeks 9–12 compared to baseline. Secondary endpoints encompassed variation in monthly analgesic intake, NRS, HIT-6 and MIDAS scores and ≥ 50%, ≥ 75% and 100% response rates at the same time intervals. All adverse events (AEs) were evaluated.

The study, not preregistered, was approved by the IRCCS San Raffaele Roma Institutional Review Board (RP 19/26) as coordinating center and mutually recognized by the other local Institutional Review Boards.

Categorical variables in demographic and clinical data were reported as percentages, and group differences were assessed using the χ² test or the Fisher exact test (2-tailed). Continuous data was summarized as mean and standard deviation (SD) for descriptive purposes. The comparison of post treatment values with baseline was done with the paired t-test or non-parametric Wilcoxon test for paired data, while the comparison of groups of patients with episodic or chronic migraine was done with the independent Student’s t-test or by the Mann Whitney U test if the distribution of the data was non normal. The Shapiro–Wilk test was applied to test the departure from normality of data distribution. A multivariate logistic regression models was fitted to identify factors associated with the response. Potential confounders and variables which were statistically significant in the univariate analysis were included in the models. Risks were expressed as odds ratio (OR) along with its 95% confidence interval (95% CI). To consider the possibility that a single center might have driven statistical analysis, a sensitivity analysis was performed, re-evaluating the effects of treatment on selected endpoints (MMDs, monthly analgesic intake, VAS, HIT-6 and MIDAS scores) after removing one-by-one each clinical center. This approach clearly showed that none of the center was an influential data point, and this variable was no longer considered in the statistical analysis. Results were considered statistically significant when p < 0.05. SPSS (IBM SPSS Statistics for Windows, version 27.0) and GraphPad Prism (GraphPad Software, Inc; v8.00.) statistical software was used for statistical analysis.

Sixty-seven migraine patients (HFEM, n = 21; CM, n = 46; F/M 53/14; mean age 48.4 years) at November 11^th 2020 had received at least one subcutaneous fremanezumab dose (225 mg monthly, n

 = 58; 625 mg quarterly, n = 9) and were considered for safety analysis, while 53 patients completed 12 weeks of treatment and were included in the effectiveness analysis (Fig. 1). Table 1 summarizes their demographic and clinical features. Most patients were females (41/53, 77.3%), affected by CM (36/53, 67.9%), with concomitant medication overuse (MO: 31/36, 86.1%) and showed on average 4.5 prior preventative failures. The only variable which showed significant difference between the two groups was the monthly analgesic intake, which was nearly double in CM as compared to HFEM (22.4 ± 18.5 vs 11.9 ± 7.1; p < 0.029). Only 9 patients (17%) were treated with a quarterly fremanezumab dosing regimen.

Fremanezumab was effective in both HFEM and CM patients (Fig. 2, supplementary table 1). At weeks 4, 8 and 12, fremanezumab induced in HFEM patients a significant reduction in MMDs (-5.6 ± 2.9, p < 0.001; -6.1 ± 3.6, p < 0.001; -4.6 ± 6.5, p < 0.05), monthly analgesic use (-7.2 ± 6.7, p < 0.001; -7.5 ± 5.8, p < 0.001; -5.7 ± 6.6, p < 0.05), NRS score (-2.5 ± 2.5, p < 0.05; -2.9 ± 1.9, p < 0.001; -3.1 ± 2.5, p < 0.001) and HIT-6 score (-4.1 ± 10.2, ns; -12.3 ± 5.5, p < 0.001; -18.1 ± 13.2, p < 0.001). At the same time intervals, fremanezumab significantly (p < 0.001 for all) reduced MHDs (-8.2 ± 6.1; -8.3 ± 6.8; -9.4 ± 6.9), monthly analgesic use (-9.6 ± 13.1; -8.2 ± 9.2; -11.1 ± 14.2) and NRS score (-1.7 ± 1.8; -1.7 ± 1.7; -2.5 ± 2.7) in patients with CM. HIT-6 scores did not significantly vary in CM (-4.5 ± 21.3; -1.2 ± 20.1; 0.3 ± 23.3). MIDAS score at weeks 9–12 was significantly reduced in HFEM (-58.3 ± 57.7; p < 0.05) and CM patients (-43.7 ± 63.4; p < 0.001) (Fig. 2, supplementary table 1).

Sixty-one-point-one percent (22/36) of CM patients remitted to episodic migraine at week 4, 72.2% (26/36) at week 8, and 75% (27/36) at week 12, while 61.2% (19/31) of the patients with MO remitted to no-MO at week 4, 64.5% (20/31) at week 8, and 67.7% (21/31) at week 12. Remission from CM to episodic migraine was sustained across week 8 in 100% (22/22), and week 12 in 95.5% (21/22) of the patients, while remission from MO to no-MO was sustained in 94.7% (18/19), and 94.4% (17/18) of the cases, respectively (Table 2).

The ≥ 50%, ≥ 75% and 100% responder rates at week 12 were 76.5%, 29.4% and 9.9% in HFEM and 58.3%, 25% and 0% in CM patients (Fig. 3). Safety and tolerability data were provided by 60 out of the 67 patients treated with ≥ 1 fremanezumab dose because at the time of data analysis 7 patients had not yet performed the first follow-up visit, scheduled 4 weeks after the first dose administration. Treatment-emergent adverse events, rated as mild and transient, were reported by 1 patient at week 4 (1.7%), 2 at week 8 (3.4%) and 3 (5.7%) at week 12 (Table 3). The most common was injection site erythema. No patient discontinued fremanezumab treatment for any reason.

The univariate analysis of determinant of ≥ 50% response documented that responders were significantly younger (44.3 ± 11.3 vs 53.8 ± 9.4 years; p = 0.03), had lower monthly analgesic intake (15.3 ± 15.0 vs 25.7 ± 17.3; p = 0.026), shorter medication overuse duration (13.8 ± 12.1 vs 40.4 ± 49.5 months; p < 0.041), and more frequent use of the monthly dosing regimen (75% vs 11%; p = 0.001) (Table 4).

When stratifying patients according to migraine frequency, younger age and shorter medication overuse emerged as positive predictors only in CM (p < 0.001 and 0.041, respectively), whereas lower monthly analgesic intake was found only in HFEM patients (p = 0.046). Monthly dosing regimen was associated to higher probability of fremanezumab responsiveness in both HFCM (p = 0.022) and CM patients (p = 0.008), although the small number of responders among those treated quarterly makes these statistics highly unstable. Lastly, CM patients with lower HIT-6 score were more likely to be fremanezumab responders (p = 0.038). To consider the role of confounders, a multivariate logistic regression analysis model was fitted to data. The only variable which survived the backward process was age (OR = 0.91; 95% CI 0.85–0.98, p = 0.013), confirming the results of univariate analysis, which showed a better response rate in younger patients. The effect of fremanezumab dosing regimen could not be properly estimated because of the small number of subjects treated quarterly.

Real-life studies are needed to translate the evidence concerning the efficacy of antiCGRP mAbs into effectiveness, to confirm their safety and tolerability, and to identify response predictors in a multifaceted clinical setting. This approach is particularly valuable to study patients with various comorbidities and inadequate response to diverse classes of prior preventive treatments.

The present prospective, multicenter, real-life Italian study documents that fremanezumab is effective, safe, and well tolerated in the prevention of subjects affected by difficult-to-treat HFEM or CM with multiple therapeutic failures. These results demonstrate an early and progressive improvement in migraine frequency, analgesic use, pain intensity, and MIDAS score in these patients. Of note, the treatment induced the remission from CM to episodic migraine and from MO to no-MO in over two-thirds of the patients, the improvement being persistent over 12 weeks in almost all cases. Adverse events were extremely rare, and no patient discontinued the treatment for any reason.

In published RCTs focusing on patients with < 2 prior therapeutic failures, fremanezumab demonstrated to be significantly superior to placebo in reducing MMDs (HALO-EM: -3.7/-3.4 vs -2.2; p < 0.001) and MHDs (HALO-CM: -4.6/-4.3; p < 0.001) as well as in ≥ 50% responder rate (HALO-EM: 47.7%/44.4% vs 27.9%, p < 0.0001; HALO-CM: 40.8%/37.6% vs 18.1%; p < 0.0001), revealing also a good safety and tolerability profile [7, 8]. Interestingly, fremanezumab proved an even better efficacy/tolerability ratio in patients who had not responded to 2–4 preventive medication clusters (FOCUS study), documenting a higher therapeutic gain over placebo in terms of MMDs/MHDs change (-4.1/-3.7 vs -0.6; p < 0.0001) and ≥ 50% responder rate (34% vs 9%, p < 0.0001) when compared to the HALO-EM and HALO-CM trials, coupled to a lower incidence of adverse events (45%-55% vs 66.2%-66.3% and 70%-71%, respectively) [9].

Patients considered in the FRIEND study were harder to treat than those included in the FOCUS trial because almost all of them had failed ≥ 3 preventive treatments (93% vs 53%), had higher migraine frequency (18.9 days/month vs 14.3 days/month), were more frequently affected by CM (67.9% vs 60.1%) and MO (86.1 vs 52%), and showed higher disability (MIDAS score 89.4 vs 61.8). Further, among episodic migraine patients, we considered only HFEM, i.e., those with ≥ 8 MMDs.

A direct comparison of real-life data with the results of RCTs is not scientifically accurate, especially when the number of patients studied is heterogeneous. In addition, differences in placebo and nocebo effects may affect anti-CGRP mAbs treatment success in different study settings. This notwithstanding, it’s worth mentioning that at weeks 9–12 the reduction in MMDs/MHDs and the proportion of ≥ 50% responders in our patients were greater than those reported in the FOCUS study (-7.9 vs -3.7/-4.1; 64.2% vs 34%, respectively), suggesting that the effectiveness of fremanezumab could be better than its efficacy. We reported similar findings also for erenumab and galcanezumab in the EARLY and GARLIT real-world studies [12‐15], where we documented a nearly twofold increase in the proportion of ≥ 50% responders at 12 weeks (~ 60% and ~ 70%, respectively) compared to the LIBERTY and CONQUER studies [20, 21]. Why antiCGRP mAbs seem to work better in migraine patients with more complex clinical picture (as the real-life ones) compared to those enrolled in the RCTs, is matter of speculation. This could depend—at least in part—on their higher baseline MMDs/MHDs. As a matter of fact, migraine frequency positively correlates with CGRP plasma levels, which are higher in CM than in episodic migraine patients and healthy controls [22]. Notably, even among real-life patients, we documented a better antiCGRP mAbs responsiveness in those with higher baseline migraine frequency (OR: 1.12; 95%CI: 1.05–1.20, p < 0.001) [13]. Another possible reason is the high proportion of patients with allodynia in real-life (59.2% in our study). Allodynia is seen as a CGRP-related symptom and prevails in patients with long-lasting attacks, long disease history, severe pain intensity and disability, MO, and psychiatric comorbidities, features typically characterizing the real-world setting [23‐25]. For the above reasons, it can be argued that real-life patients could be characterized by a greater CGRP pathogenetic involvement, thus being more sensitive to CGRP-targeting treatments.

In the FRIEND study, fremanezumab significantly improved MIDAS score in all the patients, but reduced HIT-6 only in those with HFEM. We have no reliable explanation for this controversial finding, probably biased by the small number of patients studied. MIDAS and HIT-6 are complimentary in measuring migraine-related disability but also show some substantial differences. MIDAS is basically a function of migraine frequency and relies on concrete variables (i.e., days of work or school missed), whilst HIT-6 depends more on headache severity and explores patients’ impressions on how migraine is affecting them [26].

Consequently, we cannot exclude that the above discrepancy is also somehow related to the different benefits produced by fremanezumab on migraine frequency and intensity in our patients’ groups. Indeed, at week 12, the reduction in MMDs/MHDs was comparable in HFEM (-47%,) and CM (-43.8%), whereas the decrease in intensity (NRS) was more evident in episodic (-36.5%) than in chronic migraine patients (-29%).

The results of the present study should be interpreted with caution due to some limitations. The patient sample is small, due to limited pre-reimbursement access to fremanezumab in our Country (in Italy the reimbursement is now limited to adult patients with > 8 MMDs over the last 3 months, MIDAS score > 11, and documented failure, contraindications, or low tolerability to > 3 pharmacological classes of migraine preventive medications among beta-blocker, anticonvulsants and tryciclics, or onabotulinum toxinA for CM). Some comparisons between responders and non-responders were based on very small samples, such in the case of treatment regimen. The limited size of these analyses recommends the use of p-values to achieve information about the credibility of the result rather than for hypothesis testing. Lastly, patients reported migraine data using paper–pencil diaries, a less reliable tool compared to modern electronic diaries. Strengths are the multicenter prospective design, the involvement of 9 headache centers representative of northern, central, and southern Italy and a careful clinical characterization of the patients through a shared detailed semi-structured questionnaire.

In conclusion, our multicenter, prospective study indicates that fremanezumab is effective and extremely tolerated also in real-life subjects affected by HFEM or CM with multiple therapeutic failures. Our results deserve confirmation in larger real-life studies.