Introduction

Calcitonin gene-related peptide (CGRP) is a potent endogenous vasodilator and an important neurotransmitter within the migraine headache-generating trigeminovascular system1. CGRP induces endothelium-independent vasodilation via direct action on vascular smooth muscle cells in the cerebral and coronary vascular beds and helps mediate the autoregulation of cerebral blood flow (CBF)2.

However, CGRP is generally assumed to not be involved in the regulation of steady-state vascular tone3. As a result, anti-CGRP monoclonal antibody (anti-CGRP) therapy is generally considered to not alter CBF under steady-state conditions. Against this background, anti-CGRP therapy is widely used for preventing migraine. However, migraine and insomnia often show a common pathophysiology involving impairment of the glymphatic system (GS) and can aggravate each other. Ultimately, both migraine and insomnia lead to impaired CBF autoregulation and the formation of white matter hyperintensities (WMHs)4,5,6. The cerebral circulation in migraine patients thus cannot be considered to be in a steady state. Given this information, we wondered whether CGRP inhibition by anti-CGRP therapy in migraine patients would result in changes in CBF.

Compared with normal individuals, migraine patients typically present with increased concentrations of CGRP in plasma, tears, and saliva during both migraine attacks and the interictal period7,8,9,10. The trigeminovascular system thus seems to be continually activated both during and between migraine attacks. We found that cortical hyperperfusion is frequently observed on magnetic resonance imaging (MRI) using arterial spin labeling (ASL) during the interictal period, particularly among migraine patients without insomnia11. We speculated that findings of cortical hyperperfusion (CHP), which are caused by high interictal CGRP levels, might also be altered by anti-CGRP therapy.

In this prospective study, we measured interictal CBF with ASL in migraine patients before and after anti-CGRP therapy to determine whether CBF changes after therapy and to identify CBF-related predictors of therapeutic efficacy.

Materials and methods

Subjects

This prospective, longitudinal, single-center study investigated outpatients who underwent anti-CGRP therapy at our institution from October 2022 to July 2023. All migraine patients included in this study met the diagnostic criteria of the International Headache Classification of the International Headache Society, third edition12. None of the included migraine patients showed abnormalities on neurological examination, MRI, or magnetic resonance angiography. Exclusion criteria were as follows: age < 18 years; current smoker; severe obesity (body mass index > 40 kg/m2); women receiving hormonal therapy; and patients with diabetes mellitus, epilepsy, cardiovascular disorders, ongoing treatment with calcium blockers or beta blockers, alcohol abuse, cerebrovascular disease, or known systemic diseases such as anemia and hypertension. In accordance with the relevant Japanese guidelines, indications for anti-CGRP therapy were: (i) mean number of monthly headache days (MHDs) ≥ 4 for ≥ 3 months before starting treatment; (ii) impacts on daily life even with appropriate daily acute treatment; and (iii) inability to undergo treatment with Japan-approved migraine preventive drugs (valproate sodium, etc.) due to low efficacy and/or tolerability.

Treatment protocols

All oral preventive drug therapies received before the start of anti-CGRP therapy were discontinued at the same time anti-CGRP therapy was started. None of the patients in the present study were administered prophylactic medications such as beta-blockers or Ca-antagonists that may affect CBF. The included patients underwent ASL before and after receiving anti-CGRP treatment. We prospectively analyzed the data of all included outpatients treated with one of two anti-CGRP monoclonal antibodies (mAbs), erenumab or galcanezumab, at our institution who underwent ASL both before and after treatment.

All patients started anti-CGRP therapy after providing written, informed consent and received another anti-CGRP mAb if the first proved ineffective. In this study, erenumab was administered at 70 mg/month, and galcanezumab was initially injected subcutaneously at 240 mg, followed by administration at 120 mg/month. The initial mAb to be administered was selected by the patient after the doctor explained the characteristics of both drugs. In general, the effectiveness of each anti-CGRP mAb was determined by review at least 3 months after starting administration of anti-CGRP therapy, in accordance with the consensus statement from the American Headache Society13. Specifically, treatment was considered efficacious only if at least one of the following was achieved: (i) a decrease in average MHDs by ≥ 50% relative to the pretreatment baseline value; or (ii) a decrease of ≥ 5 in the Headache Impact Test (HIT-6)14 score as an index of the degree of interference with daily life due to migraine relative to the pretreatment baseline score. If the first treatment was the anti-CGRP ligand mAb (galcanezumab), the second treatment was the anti-CGRP receptor mAb (erenumab), and vice versa. Switching between erenumab and galcanezumab was not performed if the initial treatment proved effective. If mAbs were switched, a 28-day washout period was provided between first and second treatments prior to assessing the efficacy of the second treatment.

MRI perfusion protocols

All migraine patients underwent CBF studies with ASL during the interictal period ≥ 48 h after the last attack or most recent migraine treatment (such as with triptans), whichever came later. As a rule, ASL after anti-CGRP therapy was performed simultaneously with the clinical evaluation performed at least 3 months after anti-CGRP therapy. However, if a migraine attack occurred on the day scheduled for ASL, the ASL was postponed until the following month. Consequently, the period for evaluating treatment response varied from patient to patient and could not be standardized. All MRI was performed at our hospital using a 1.5-T scanner (Signa Explorer; GE Healthcare, Milwaukee, WI) with an Express head-neck array coil. Total scan time was 16–18 min and included conventional axial T1-weighted imaging, fluid-attenuated inversion recovery imaging, diffusion-weighted imaging, magnetic resonance angiography, and ASL, with approximately 2 min 30 s involving the use of background-suppressed, three-dimensional, pulse-continuous ASL imaging to quantitatively measure CBF using the parameters described in our previous paper11.

Background-suppressed three-dimensional multi-delay pulse-continuous ASL (also called “pseudo-continuous ASL”) images were acquired using the following parameters: repetition time, 4548 ms; echo time, 10.5 ms; field of view, 24 cm; 512 sampling points on 6 spirals (matrix size, 512 × 6); spatial resolution, 5.0 mm; section thickness, 4 mm; number of sections, 30; excitations, 2; and bandwidth, 62.50 Hz. ASL data at two post-labeling delay times (1525 and 2525 ms) were obtained, with the post-labeling delay of 1525 ms used for the CBF analysis in patients without severe cerebral vascular stenosis. In the ASL labeling scheme, a control image without a labeling pulse was acquired, and a labeled image was acquired by applying a labeling pulse to the carotid arterial blood to invert magnetization. The difference between these two images highlights perfused blood and allows for noninvasive measurement of CBF. Proton-weighted images were then acquired for M0 acquisition, normalized, and a CBF map was created. At our institution, we used a simplified model for CBF measurement that does not require arterial time measurement, and arterial transit time correction was therefore not performed. In addition, vessel crusher settings with a 180° pulse were not used. Instead, flow-driven adiabatic inversion was used. Because the ASL used acquires data in three dimensions, a technique to reduce vessel crusher settings by thinning the reconstruction slice thickness while maintaining a good signal-to-noise ratio was used. The cortical rim was not trimmed during image processing, but the cortical rim was interpreted carefully. We also compared the two types of post-labeling delay images to determine macrovascular contamination and blood flow transit delay.

WMHs were defined as hyperintensities on fluid-attenuated inversion recovery imaging that were not hypointense on T1-weighted imaging.

For quantitative region of interest (ROI) analyses, we used fully automated ROI-based analysis software (3DSRT NEURO; FUJIFILM Toyama Chemical Co., Tokyo, Japan) for ROI positioning and selection on ASL maps, as this provides a more objective delineation than manual procedures and excellent reproducibility15,16. Quantitative CBF images obtained from study subjects were anatomically registered to a standard brain atlas (Supplementary Fig. 1). The three-dimensional stereotaxic ROI template (3DSRT) software provided regional CBF (rCBF) values for bilateral callosomarginal, precentral, central, parietal, angular, temporal, posterior cerebral, pericallosal, and thalamic regions using predefined ROIs on the standard atlas (Supplementary Fig. 1). Referring to Inoue et al.17, we determined: (1) rCBF in the territory of the anterior cerebral artery (ACA) through the use of callosomarginal and pericallosal ROIs; (2) rCBF in the anterior part of the territory of the middle cerebral artery (AM) through the use of precentral and central ROIs; (3) rCBF in the posterior part of the territory of the middle cerebral artery (PM) through the use of parietal, angular, and temporal ROIs; and (4) rCBF in the territory of the posterior cerebral artery (PCA) through the use of the posterior cerebral ROI (Supplementary Fig. 1). The ROI parameters for cortical regions were established as described in our previous report11.

In this study, similar to our previous report11, we confirmed the absence of any significant laterality in rCBF by calculating the asymmetry index for migraine patients. An absolute asymmetry index > 10 indicates asymmetric perfusion18,19,20,21. If no significant lateral difference in rCBF was present, we averaged left and right regional CBFs and calculated mean and standard deviation values.

As in our previous report11, an ROI was defined as exhibiting hyperperfusion if the CBF in that ROI was at least two standard deviations greater than the mean reference value for the CBF in control subjects20,21. In addition, according to the results from our previous paper, this study defined CHP as a CBF ≥ 60 ml/100 g/min in two or more ROIs within each cortical area (ACA, AM, PM, and PCA)11.

We defined appropriate changes as changes in CBF after treatment that tended toward the normal range, and inappropriate changes as changes that tended away from the normal range.

Variables and data extraction

MHDs for each patient were obtained from patient-maintained headache diaries. We also obtained the following patient characteristics from electronic medical records, self-reports, etc.: sex; age; incidents of episodic migraine (EM), chronic migraine (CM), aura, morning migraine, menstrual migraine, weather-related migraine, and medication-overuse headache (MOH); HIT-6 score14; Athens Insomnia Scale (AIS) score22; score for the Beck Depression Inventory, second edition (BDI-II)23; and use of a combination of ≥ 3 types of preventative treatment. “Morning migraine” was defined as waking up in the morning due to migraine pain. An AIS score ≥ 4 was defined as insomnia, and a BDI-II score ≥ 11 was defined as depression. Finally, we quantified the degree of interference with daily life during the interictal period before and after anti-CGRP therapy using the Migraine Interictal Burden Scale (MIBS-4) score24.

Subgroups based on CBF changes after anti-CGRP therapy

We classified migraine patients in this study into the following four clinical groups based on the presence or absence of CHP before treatment and the change in CBF after treatment, and examined their clinical characteristics: Group CHP and CBF- (CHP before anti-CGRP therapy and decreased CBF after anti-CGRP therapy); Group NoCHP and CBF- (no CHP before anti-CGRP therapy and decreased CBF after anti-CGRP therapy); Group NoCHP and CBF+ (no CHP before anti-CGRP therapy and increased CBF after anti-CGRP therapy); and Group ConstantCBF (no change in CBF between before and after anti-CGRP therapy).

Endpoints and analysis

The primary endpoint was determination of whether interictal CBF from ASL changed between before and after anti-CGRP therapy. The secondary endpoint was whether interictal CBF measurements before and after treatment could be used as a CBF-related predictor of treatment efficacy.

Statistical analysis

All statistical analyses were performed using commercially available software (IBM SPSS Statistics for Windows, version 27.0; IBM Corp., Armonk, NY). Distributions of each variable were checked for normality using the Shapiro–Wilk test, and homogeneity of variance was determined using the Levene test. Clinical factors potentially associated with the effectiveness of anti-CGRP therapy were compared between groups using Fisher’s exact test. Continuous variables were compared between groups using the independent sample Student’s t-test. Values of p < 0.05 were considered to indicate statistical significance. All data are presented as mean ± standard deviation unless otherwise specified. Clinical factors with a significance level of p < 0.10 were subjected to multivariable logistic regression analysis with prediction of treatment efficacy from anti-CGRP therapy as the dependent variable. To confirm the robustness of the results, the multivariate analysis for predicting treatment effect with anti-CGRP therapy was also analyzed for all cases (73 cases), as well as for only those who underwent drug switching (46 cases), as a sensitivity analysis. When CBF data showed normal distributions, one-way analysis of variance (ANOVA) with post-hoc multiple comparison tests was used to assess differences in CBF data among the four clinical groups.

Institutional review board approval.

All study protocols were approved by the Institutional Review Board for Clinical Research (approval no. 22R-078) and the Conflict of Interest Management Committee (approval no. 22–168) at our university. The study protocol was conducted in accordance with the Declaration of Helsinki, and all migraine patients included in the study provided written, informed consent before participating in the study. Our report complies with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement for cohort studies.

Results

Patient characteristics

Among 77 patients who underwent anti-CGRP therapy during the study period, 3 patients who were unable to undergo MRI due to claustrophobia and 1 patient for whom satisfactory images could not be obtained due to image artifacts caused by a metal orthodontic appliance were excluded. Ultimately, this prospective cross-sectional study enrolled 73 outpatients who received anti-CGRP therapy from October 2022 to July 2023 at our hospital.

Clinical factors such as age distribution obtained before anti-CGRP therapy are presented in Supplementary Table 1. Among the 73 patients treated with anti-CGRP therapy, 66 were women (90%), 54 (74%) had CM, and 19 (26%) had EM. Only two patients had been given valproate sodium for migraine prophylaxis. Both were over 50 years old and confirmed they did not wish to become pregnant. Thirty patients (41%) experienced aura, and 46 patients (63%) experienced MOH. Fifty-nine patients (81%) had insomnia with an AIS score ≥ 4, and 43 patients (59%) had depression with a BDI-II score ≥ 11. Thirty-three patients (45%) had CHP findings on ASL, 24 of whom also had insomnia. Among the 28 patients (38%) with WMH findings on MRI, 25 patients also experienced insomnia. All patients with WMHs had a Fazekas scale25 grade of 1, except for one 75-year-old patient (grade 2).

Switching anti-CGRP therapy is effective

To investigate changes in CBF among migraine patients who underwent anti-CGRP therapy, we first assessed the proportions of patients who achieved a response, as shown in the treatment flowchart for anti-CGRP therapy, including any drug switching (Fig. 1). Among the 73 patients treated with anti-CGRP therapy, 46 switched anti-CGRP mAbs. Overall, anti-CGRP appeared efficacious, with ≥ 50%, ≥ 75%, and 100% reductions in MHDs obtained for 54 (74%), 47 (64%), and 23 patients (32%), respectively (Supplementary Table 2).

Fig. 1
figure 1

Treatment flowchart for anti-CGRP therapy. ASL, magnetic resonance imaging-arterial spin labeling; CGRP, calcitonin gene-related peptide; mAb, monoclonal antibody; HIT-6, headache impact test.

Full size image

Changes in CBF after anti-CGRP therapy

The mean (± standard deviation) and median times from initial anti-CGRP therapy to the date of ASL assessment were 91 ± 15 days and 91 (84–140) days, respectively, in the 27 patients who did not switch medications, and 85 ± 12 days and 86 (28–112) days, respectively, in the first period of the 46 patients who switched medications (t-test, p = 0.075). The mean (± standard deviation) and median times from anti-CGRP therapy to the date of ASL assessment in patients who switched to a second agent were 88 ± 18 days and 87 (84–196) days, respectively. No significant difference was observed in the time from administration to ASL between “ligand → receptor” (Phase 1: 88 ± 10 days, Phase 2: 86 ± 7 days) and “receptor → ligand” (Phase 1: 84 ± 10 days, Phase 2: 79 ± 17 days). In one patient, the medication was switched after only 28 days due to side effects. In the remaining patients, ASL assessment was performed more than 3 months after starting medication.

In this study, the asymmetry index demonstrated a lack of significant laterality in rCBF and an absolute asymmetry index < 10 both before and after anti-CGRP therapy20,21. We therefore averaged the left and right rCBFs in the ROIs of migraine patients, calculated mean and standard deviation values for each clinical group, and used these values to determine whether patients showed CHP.

No trends useful for predicting the efficacy of anti-CGRP therapy based on ASL baseline imaging and CBF data alone were found. Therefore, the focus was on changes in CBF before and after anti-CGRP therapy.

To evaluate changes in CBF before and after anti-CGRP therapy, we first divided patients into those for whom CBF increased after treatment and those for whom CBF decreased after treatment. We then used the Shapiro–Wilk test to confirm that CBF increase rates in each group followed a normal distribution. Next, we calculated the quartiles of the increase rate for each group of patients for whom CBF increased and decreased, defining a decrease in CBF as a decrease of more than 5%, and an increase in CBF as an increase of more than 5%, by using the median values of -5% and + 5%.

The number of cases in each subgroup was 27 in Group CHP and CBF-, 19 in Group NoCHP and CBF-, 18 in Group NoCHP and CBF+, and 9 in Group ConstantCBF (6 patients had CHP before anti-CGRP therapy). Typical imaging findings for each group are shown in Fig. 2A–D.

Fig. 2
figure 2

Example patients from the four groups defined by changes in CBF after anti-CGRP therapy. (A) Representative patient from Group CHP and CBF-: A 75-year-old woman with episodic migraine (MHD: 10, HIT-6: 66) without insomnia presented with CHP on ASL before treatment (a). After administration of galcanezumab, no changes are observed on ASL images (b). After switching to erenumab, CHP has disappeared and treatment is considered to have been 100% efficacious (MHD: 0, HIT-6: 36) (c). Please see Supplementary Figure 1 for a depiction of the ROIs in the cortical region. (B) Representative patient from Group NoCHP and CBF-: A 64-year-old man with chronic migraine (MHD: 30, HIT-6: 78), insomnia, morning migraine, WMHs on MRI, and no CHP on ASL (a). After administration of erenumab, the patient achieves a 50% reduction in MHDs (MHD: 8, HIT-6: 50), but insomnia and morning migraine do not improve, and CBF is lower than normal in all ROIs (b). Please see Supplementary Figure 1 for a depiction of the ROIs in the cortical region. (C) Representative patient from Group NoCHP and CBF+: A 44-year-old woman with episodic migraine (MHD: 10, HIT-6: 60), insomnia and morning migraine but no CHP on ASL (a). After administration of galcanezumab, insomnia and morning migraine improve, CBF increases, and 100% treatment efficacy is achieved (MHD: 0, HIT-6: 36) (b). Please see Supplementary Figure 1 for a depiction of ROIs in the cortical region. (D) Representative patient from Group ConstantCBF: A 40-year-old woman with episodic migraine (MHD: 5, HIT-6: 60) and CHP on ASL despite insomnia (a). Neither galcanezumab (b) nor erenumab administration (c) produce changes on ASL, and no therapeutic response is achieved (MHD: 5, HIT-6: 56). Please see Supplementary Figure 1 for a depiction of ROIs in the cortical region.

Full size image

Clinical features associated with the four groups based on CBF changes are shown in Supplementary Table 3.

This study did not identify any differences in the potential clinical impact on treatment efficacy, including CBF changes, between erenumab and galcanezumab.

Pretreatment clinical factors

The proportions of patients who experienced failure with ≥ 3 types of preventive drugs (14 of 19 patients, 74%) and who presented with WMHs (16 of 19 patients, 84%) and insomnia with WMHs (15 of 19 patients, 79%) were significantly greater in Group NoCHP and CBF- than in the other three groups (p < 0.001). Significantly fewer patients in Group CHP and CBF- (14 of 27, 52%, p < 0.001) had a pretreatment AIS score ≥ 4 than in Groups NoCHP and CBF- (18 of 19, 95%), NoCHP and CBF+ (18 of 18, 100%), and D (9 of 9, 100%). Significantly more patients in Group ConstantCBF (9 of 9, 100%, p = 0.028) showed a pretreatment BDI-II score > 11 compared to Groups CHP and CBF- (13 of 27, 48%), NoCHP and CBF- (9 of 19, 48%), and NoCHP and CBF+ (12 of 18, 67%). No significant differences in the proportions of patients with MIBS-4 scores > 1 or in mean MIBS-4 score before anti-CGRP therapy were observed between groups. Finally, significantly more patients presented with CHP findings despite insomnia in Group CHP and CBF- (18 of 27, 67%, p < 0.001) and Group ConstantCBF (6 of 9, 67%, p < 0.001) than in Group NoCHP and CBF- (0 of 19) and Group NoCHP and CBF+ (0 of 18).

Posttreatment clinical factors

Proportions of patients with responses of ≥ 50% and ≥ 75% fewer MHDs were significantly greater in Group CHP and CBF-, Group NoCHP and CBF-, and Group NoCHP and CBF + than in Group ConstantCBF (Supplementary Table 3). Similarly, proportions of patients with 100% fewer MHDs were significantly lower in Group CHP and CBF- and Group ConstantCBF than in Group NoCHP and CBF- or Group NoCHP and CBF+ (p = 0.046) (Supplementary Table 3). Further, Group NoCHP and CBF + showed the highest percentages of patients with ≥ 50% (94%), ≥ 75% (89%), and 100% fewer MHDs (50%) after treatment, but these differences were not significant. In addition, the proportion of patients for whom the HIT-6 score was lower after treatment was significantly greater in Group CHP and CBF- (p = 0.027), Group NoCHP and CBF- (p = 0.023), and Group NoCHP and CBF+ (p = 0.002) than in Group ConstantCBF (Supplementary Table 3). These results show that Group ConstantCBF did not achieve either a decrease in MHDs or an improvement in HIT-6. Group CHP and CBF- had a significantly lower frequency of 100% responders, due to the high frequency of CHP despite insomnia (Supplementary Table 3).

The proportion of patients who demonstrated improvements in posttreatment AIS score was significantly lower in Group NoCHP and CBF- (3 of 18, 17%) and Group ConstantCBF (1 of 9, 11%) than in Group CHP and CBF- (7 of 14, 50%) or Group NoCHP and CBF+ (14 of 18, 78%, p < 0.001). The proportion of patients with a posttreatment BDI-II score ≥ 11 was significantly greater in Group ConstantCBF (8 of 9, 89%, p = 0.003) than in Group CHP and CBF- (8 of 27, 30%), Group NoCHP and CBF- (4 of 19, 21%), or Group NoCHP and CBF+ (5 of 18, 28%). Finally, the proportion of patients for whom MIBS-4 score improved after anti-CGRP therapy was significantly greater in Group CHP and CBF- (15 of 17, 88%, p = 0.001) and Group NoCHP and CBF+ (20 of 27, 74%, p = 0.001) than in Group NoCHP and CBF- (8 of 19, 42%) or Group ConstantCBF (1 of 9, 11%).

Changes in CBF before and after anti-CGRP therapy in each group

Since our CBF data were normally distributed, we used ANOVA and post-hoc multiple comparisons tests to assess differences in CBF data among the four clinical groups. The Levene test indicated that CBF data from each of the four clinical groups demonstrated homogeneity of variance. ANOVA results were therefore interpreted using Tukey’s honestly significant difference test to identify which pairs of groups demonstrated significant differences in CBF data.

Changes in CBF in the different ROIs of each group after anti-CGRP therapy are shown in Fig. 3. Compared with the other groups, Group CHP and CBF- showed significantly greater pretreatment CBF in all ROIs (i.e., ACA, AM, PM, PCA, and thalamus), whereas CBF after anti-CGRP therapy was significantly decreased in all ROIs, approaching normal CBF (Supplementary Table 4). CBF after anti-CGRP therapy was significantly lower in Group NoCHP and CBF- than in the other groups for all ROIs and was below the normal range in all regions (Supplementary Table 4). Only Group NoCHP and CBF + showed an increase in CBF after treatment (Supplementary Table 4). As a result, CBF in both Group CHP and CBF- and Group NoCHP and CBF + became approximately equivalent for each ROI (Supplementary Table 4). Group ConstantCBF before and after anti-CGRP therapy showed no significant changes in any ROI.

Fig. 3
figure 3

Changes in CBF measured on ASL before and after anti-CGRP therapy in each group. White bars indicate CBF before treatment and black bars indicate CBF after treatment. (A) Group CHP and CBF-. (B) Group NoCHP and CBF-. (C) Group NoCHP and CBF+. (D) Group ConstantCBF. ACA, anterior cerebral artery; AM, anterior part of the middle cerebral artery territory; PM, posterior part of the middle cerebral artery territory; PCA, posterior cerebral artery; THA, thalamus. Compared with the other groups, Group CHP and CBF- (A) shows significantly greater pretreatment CBF in all ROIs, whereas CBF after anti-CGRP therapy is significantly decreased in all ROIs. CBF of post-CGRP therapy is significantly lower in Group NoCHP and CBF- (B) than in the other groups for all ROIs and is below normal value in all regions. Only Group NoCHP and CBF+ (C) show increased CBF after treatment. As a result, CBF in both Group CHP and CBF- (A) and Group NoCHP and CBF+ (C) becomes approximately equivalent for each ROI. Group ConstantCBF (D) before and after anti-CGRP therapy shows no significant changes in any ROI.

Full size image

Predictors of response to anti-CGRP therapy

Supplementary Table 2 also shows clinical factors for patients in the responder (≥ 50% reduction in MHDs) and non-responder groups. Univariate analysis revealed that factors significantly associated with the responder group were fewer MHDs (p = 0.012), no failure of preventive drugs (≥ 3 types) (p = 0.050), AIS score < 4 (p = 0.015), BDI-II score < 11 (p = 0.014), no CHP findings despite insomnia (p = 0.047), and no insomnia with WMHs (p = 0.050).

The results of multivariate logistic regression analysis of all 73 patients revealed BDI-II score < 11 (p = 0.007) and non-failure of preventive drugs (≥ 3 types) (p = 0.008) as clinical factors, while no CHP findings despite insomnia (p = 0.006) and no insomnia with WMHs (p = 0.010) as neuroradiological factors predicted a ≥ 50% response to anti-CGRP therapy in migraine patients. However, on multivariate analysis conducted as a sensitivity analysis for only those patients who underwent drug switching (46 cases), of these four significant factors, only non-failure of prophylactic medications (≥ 3 types) was not a significant predictor of the efficacy of anti-CGRP therapy (p = 0.335). Based on the results of this sensitivity analysis, the results of the final multivariate analysis predicting a ≥ 50% response to anti-CGRP therapy in migraine patients are shown in Table 1. Neuroradiologically, patients with insomnia accompanied by CHP or WMH on MRI were considered unlikely to benefit from anti-CGRP therapy.

Table 1 Results of multivariable analysis for associations between pretherapy clinical factors and positive treatment response (≥ 50% reduction in monthly headache days) rate in migraine patients.
Full size table

In this study, the switching rate between antibodies was strikingly high (46 of 73 patients, 63%). We compared clinical factors, including changes in CBF before and after treatment between those with and without switching, but found no significant differences, and the cause was unknown.

Discussion

Predictors of therapeutic efficacy from Anti-CGRP

This is the first study to identify factors that predict the effectiveness of anti-CGRP therapy via MRI.

The GS is a potential brain waste removal system that includes the perivascular space. This system and the neurovascular unit (NVU), which play important roles in regulating CBF and neurovascular coupling, form a structural and functional continuum to maintain CBF and cerebrospinal fluid homeostasis6. The GS and NVU are involved in the pathophysiology of cerebral small vessel disease (CSVD)6. As the GS functions primarily during sleep, insomnia leads to impairment of the GS and, consequently, to reduced CBF. Impaired GS also leads to the accumulation of neuroexcitatory and proinflammatory chemicals, such as CGRP, which are involved in the process of migraine becoming chronic4. This may predispose individuals to the development of MOH and CM, and the migraine itself may directly or indirectly (through GS dysfunction) exacerbate sleep disorders4. Furthermore, WMHs in migraine patients are thought to be the result of GS dysfunction4. All three factors—migraine, insomnia, and WMHs (reflecting CSVD)—are thus related from the perspective of GS impairment4. Given this background, presentation with WMHs and insomnia was not surprising as another significant, negative predictive factor of treatment efficacy, suggesting that, for patients with migraine who also suffer from insomnia and present with WMHs (reflecting CSVD) associated with GS impairment, anti-CGRP therapy may prove ineffective.

We have previously reported that CHP was observed in 92% of the interictal ASL images from migraine patients without insomnia, whereas only 18% of migraine patients with insomnia demonstrated CHP11. We speculate that this is because CBF is reduced in migraine patients with insomnia due to GS impairment caused by insomnia. In this study, CHP with insomnia was a negative predictor of the efficacy of anti-CGRP therapy in migraine patients. This may be because vasodilator peptides other than CGRP (e.g., vasoactive intestinal polypeptide, adrenal cortical medullary protein, and amylin)26 markedly increase CBF, offsetting the decrease in CBF caused by insomnia. In other words, migraine patients who exhibited CHP despite insomnia were presumed to have CGRP-independent migraine, similar to Group ConstantCBF patients, who showed no change in CBF before or after treatment. Therefore, for migraine patients who exhibit CHP despite insomnia or who show no change in CBF between before and after anti-CGRP therapy, we believe that switching the brand of anti-CGRP agent or monoclonal antibody therapy targeting pituitary adenylate cyclase-activating polypeptide should be considered27.

In this study, multivariable analysis revealed a BDI-II score ≥ 11 and failure of ≥ 3 types of prophylactic drugs (results before sensitivity analysis) as significant independent predictors of non-response (where non-response was defined as a < 50% decrease in MHDs) to anti-CGRP therapy. These results were consistent with findings from previous reports28,29. These results suggest that, in clinical practice, treating migraine with effective preventive treatment early after the onset of migraine is crucial, before migraine becomes chronic due to worsening of migraine-related depression or repeated prophylactic treatment8.

Appropriate changes in CBF after Anti-CGRP therapy

The anti-CGRP monoclonal antibodies used in this study (erenumab and galcanezumab) have different mechanisms of action. Erenumab targets the CGRP receptor, whereas galcanezumab binds directly to the CGRP ligand. However, this study did not find any differences in the potential effects of this pharmacological difference on the therapeutic effect. Future investigations will need to examine CBF, including changes in CGRP concentration, before and after anti-CGRP therapy.

Because CHP reflects a state of high endogenous CGRP levels caused by ongoing activation of the trigeminovascular system, including during the interictal period, the disappearance of CHP and normalization of CBF during the interictal period in Group CHP and CBF- reflects appropriate CBF changes following the initiation of anti-CGRP therapy.

Group NoCHP and CBF + displayed no CHP findings because all patients had insomnia, and CBF in these patients increased after anti-CGRP therapy, but did not reach the level of CHP. As with Group CHP and CBF-, the change in CBF after treatment in Group NoCHP and CBF + was also appropriate, and 94% of patients showed ≥ 50% fewer MHDs, the highest among all groups. Furthermore, both HIT-6 and MIBS-4 scores improved after treatment in both groups, and the interference with daily life caused by migraine improved both during and between attacks. On the other hand, insomnia improved at a high rate in Group NoCHP and CBF + migraine patients after treatment (14 of 18 cases, 78%). This suggests that the insomnia in this group was reversible and attributable to the migraine itself. In other words, we speculate that the improvement in insomnia resulted from a decrease in MHDs and improvements in interference with daily life from migraine. We believe that these improvements in insomnia resolved GS impairment and led to increases in CBF that exceeded the decrease in CBF caused by anti-CGRP therapy.

Inappropriate CBF changes after Anti-CGRP therapy

CGRP mediates CBF autoregulation in part by acting on cerebral vascular smooth muscle cells and inducing endothelium-independent vasodilation2. As a result, CGRP can rescue vasoconstriction and hypoperfusion of the cerebral vascular bed in patients with cerebral ischemia2,30,31,32. Normally, endogenous CGRP does not change CBF under steady-state conditions, but migraine and insomnia do not represent steady states because of the presence of CBF autoregulation disorders. As anti-CGRP therapy in migraine patients inhibits the potent vasodilator CGRP, interictal CBF may be decreased. In the present study, approximately 90% of patients in Group NoCHP and CBF- displayed ROIs in which the CBF decreased to < 95% of the normal value after anti-CGRP therapy; this change was thus considered inappropriate.

The proportion of patients who experienced improvement in insomnia was significantly lower in the NoCHP and CBF- group (3 of 18, 17%) than in the NoCHP and CBF + group, suggesting that insomnia in the NoCHP and CBF- group was irreversible. Furthermore, compared with Group NoCHP and CBF + patients, Group NoCHP and CBF- patients were significantly more likely to have CM, insomnia with WMHs, and experience of preventive medication failure, and even though HIT-6 scores improved, the degree of improvement in MIBS-4 was significantly lower. In other words, among patients in Group NoCHP and CBF-, frequent cortical spreading of depression due to migraine that has become chronic leads to GS disorder. In addition, the presence of intractable insomnia may have further exacerbated the GS disorder. Given the present results, why CBF decreased below normal after anti-CGRP therapy in this group remains unclear. However, in the presence of decreased CBF due to severe GS disorder, the suppression of CGRP, which has CBF-relieving effects during ischemia, by anti-CGRP therapy may have triggered the decrease in CBF after treatment. Chronic migraine-associated endothelial dysfunction or CGRP-independent pathways may also play a role. Further detailed research targeting the relationships of migraine, insomnia, GS disorder, and CSVD, including biochemical and sleep disorder data, is needed.

Headache clinicians need to be aware that in NoCHP and CBF- groups, many patients may display CBF below the normal range after treatment, which may lead to cerebral ischemia. Regular follow-up with ASL is therefore warranted.

Limitations

This study has several limitations that need to be considered when interpreting the results. First, we did not measure CGRP levels before and after anti-CGRP therapy and therefore could only speculate as to whether patients demonstrated CGRP-dependent or -independent migraine. More comprehensive analysis could be provided in future studies by correlating ASL findings with biochemical markers. Second, this study did not provide a sufficient washout period when switching between CGRP mAbs, as required for a crossover study. We were therefore unable to conduct precise comparisons of treatment efficacy between galcanezumab and erenumab. In addition, we cannot entirely exclude the possibility that some, or even all, patients who appeared to respond after switching from erenumab to galcanezumab (or vice versa) might actually have been properly classified as late responders to the initial therapy. Another problem was that the models were not stratified for “ligand to receptor” or “receptor to ligand” sequences, and exposure-weighted analyses were not performed. Third, if a migraine attack occurred on the day of follow-up, ASL was postponed until the following month, so the period during which treatment response was evaluated varied among patients and could not be standardized. Fourth, this study was a single-center study of Japanese patients, and clarification of how these results apply to other ethnic groups and medical settings would be useful in the future. Fifth, the sample size of this study was insufficient, and the statistical power would thus have been limited by the stratification of patient subgroups. Sixth, the effect of menstrual phase on interictal CBF in female migraine patients was not considered. Seventh, in this study, fully automated ROI-based analysis software was used to place and select ROIs on the ASL map for quantitative ROI analysis. As a result, it was not possible to perform statistical approaches such as estimating treatment-related changes using a mixed-effects model (subject random intercept, ROI nested within subjects), reporting fixed-effects, or controlling for multiplicity using a false discovery rate. Finally, the samples examined in this study were highly variable, since they included migraine patients with and without aura, episodic and chronic migraine patients, migraine patients with and without insomnia, and migraine patients with and without MOH in the one group. This variability is likely to have affected the perfusion parameters observed at baseline in the patients under examination and would thus have limited the generalizability of the results. Despite these limitations, and noting that the number of cases, case selection, study design, and heterogeneity of the study content do not allow for definitive scientific conclusions to be made, the present study is the first to incorporate neuroradiological findings (i.e., MRI findings, including ASL) into predicting the efficacy of anti-CGRP therapy.

Conclusion

Interictal CBF measurements using ASL before and after anti-CGRP therapy revealed four patterns of CBF change. Of them, migraine therapists should keep in mind that cases may be encountered in which “CBF decreases below normal” rather than “CBF changes appropriately to approach normal values.” For these findings to have a further impact on daily clinical practice, studies measuring CGRP levels and evaluating GS disease using imaging are likely to be needed.