Primary working hypothesis is that NON-responders to mAbs bear a dysfunction of the endocannabidiome system (eCBome), as suggested by pre-clinical and clinical data by our group.

They will be identified as patients showing a reduction of monthly migraine days < 50% after three months of treatment.

The clinical, biochemical and neurofunctional impact of novel therapeutic approaches expected to interfere with eCBome will be evaluated in NON-responder patients

Monoclonal antibodies directed against the calcitonin gene related peptide pathway (mAbs) have recently become available as the first targeted-therapy for migraine prevention. Despite their revolution in migraine therapy, clinical evidence demonstrated that nearly one-third of patients do not respond to mAbs. Even among Responders (namely those with a reduction in monthly migraine days - MMDs, of at least 50%), almost half still suffers from 8 residual MMDs or more.

There is therefore room for improvement, and understanding the pathophysiological mechanisms involved in these processes is mandatory to develop novel therapies and optimize treatment outcome.

Among several non-CGRP pathways, the eCBome may play a major role in migraine as it interacts with multiple pathways to modulate inflammation and pain.

Previous evidence showed that genes expression of eCBome system is altered in migraine patients.

Interestingly, it seems possible to normalize eCBome dysfunction in migraineurs, as demonstrated by the decrease of FAAH levels in chronic migraineurs with medication overuse who underwent a successful detoxification.

From a therapeutic point of view, a potential role for PEA in the termination of experimentally-induced migraine attack in humans and in the acute treatment of spontaneous attacks were published.

Polyunsaturated fatty acids (PUFAs) are other eCBome precursors involved in neuronal processes and anti-inflammatory properties. Data supports an interconnection between dietary n-3 PUFAs intake, also showing a protective action of high PUFAs intake in inflammation and neurodegenerative disease.

Another therapeutic weapon is represented by the ketogenic diet (KD), a "fat" diet that induces an increase in PUFAs. In migraine, KD reduced the frequency of migraine attacks when compared to a standard mediterranean diet in episodic migraine.

It may interfere with the eCBome at multiple levels (gut and brain) and via multiple mechanisms: i) increase in substrate availability; ii) improvement of cortical metabolism and hyperexcitability enhancing glutamate clearing from astrocytes; ii) modulation of neuro-inflammation.

Preliminary evidence also support a role for KD in treatment-resistant migraine, and in a small number of CM patients resistant to approved therapies, with preliminary data suggest a positive response. Its efficacy in patients who failed mAbs treatment is yet to be elucidated, though.

As intriguing data suggests the possibility to interfere with the eCBome via multiple modalities, ranging from PEA administration to dietary adaptations, it seems extremely interesting to identify an eCBome-dependent migraine phenotype and to test its performance in proof-of-concept therapeutic challenges.

The investigators intend to act on the alteration of microbiota through the ketogenic diet (KD) and on the alteration in endocannabinoids related lipids with PEA or PUFA supplementation. These therapeutic modalities are already available for clinical use and have a highly safe profile, easily ready to be used for improving health in a group of highly disabled patients.

The investigators will also try a concomitant therapy (mAbs + PEA / PUFA) in order to provide insights on the effect of a concomitant dual modulation (CGRP and eCBome system) in patients undergoing mAbs treatment.

Furthermore, a comprehensive biochemical set of potential biomarkers will be evaluated, including neuropeptides, microRNAs, inflammatory cytokines, kynurenine metabolites and brain cortical connectivity trough recording of HD- EEG and rs-fMRI. The biochemical and functional phenotyping will allow the identification of a multibiomarkers panel signature of migraine patients resisting to specifically targeted preventive treatments, potentially associated to the clinical response to treatments that targets different pathways.

Study design:

Patients will be enrolled among those receiving treatment with monoclonal antibodies against the calcitonin gene receptor peptide pathway (mAbs) at the clinic of IRCCS Mondino Institute (Pavia) and Neurology Department of the University of L'Aquila (Avezzano).

The following groups will be identified:

• NON-responder group: patients who show a reduction < 50% of MMDs after 3 to 6 months of mAbs treatment compared to baseline
• Partial Responder: patients who show a reduction > / = 50% of MMDs after 3 to 6 months of treatment compared to baseline, but still having at least 5 MMDs

NON-responders will stop mAbs treatment and will receive one of the following:

• PEA supplementation (600mg twice a day for three months)
• ketogenic diet (KD) for three months (characterized by the following percentages of macronutrients: 65% fat, 27% protein and 8% carbohydrates)

Partial responders will be assigned to add-on to mAbs:

- PEA (600mg twice a day for three months)

Clinical data will be collected and biochemical and neurofunctional profiling of migraine patients will be performed before starting novel treatment (PEA or KD) (V0) and after three months of treatment (V1). Clinical data will be collected up to 6 months from treatment starting (V2).

Methods:

BIOCHEMICAL PROFILING

All patients will undergo a biochemical profiling that will include analysis of:

• eCBome system: FAAH, MAGL, DAGL, NAPE, NAAA mRNA in PBMCs;
• plasma levels of AEA, 2-AG, PEA and OEA; CGRP, PACAP, and VIP; IL-1beta, TNF-alpha, IL-4, and IL-10; kynurenic and quinolinic acids;
• miR-382-5p, miR-34a, miR-30a, and miR-155 in PBMCs;
• shotgun analysis of microbiota in patients' faeces. Biochemical sampling will be collected between 9 and 11 a.m. to avoid circadian rhythm influence. All evaluation will be performed in the interictal phase. All samples will be pre-processed within 30 minutes from blood sampling, and immediately stored at -80° C. The final processing will be performed within 3 months.

The following collection methods will be adopted:

• mRNA of eCBome enzymes and microRNA analysis in PBMCs. Blood samples will be collected within ethylenediamine tetra-acetic acid tubes, with isolation of PBMCs and total RNA. Ubiquitin C and U6 will act as housekeeping genes for genes coding for the eCBome enzymes and miRNAs.
• Endocannabinoids and related lipids extraction; kynurenic acid and quinolinic acid, AEA, PEA, 2-AG and OEA will be determined according to the method published by Gao and collaborators with minor modifications(Gao, 2020). Measurement of kinurenine metabolite levels will be performed according to the method described by Fuertig et al. with minor modifications (Fuertig, 2016).
• Plasma samples will be prepared by centrifugation of blood samples. CGRP alpha, PACAP-38 and VIP levels will be measured using a commercial enzyme linked immunosorbent assay; while IL-1beta, TNF-alpha, IL-4, IL-10 cytokine will be measured by the Ella Automated Immunoassay System with a Simple Plex assay panel.
• Microbiome analysis: patients will be instructed for the correct collection and preservation of stool specimens; these will be delivered to the Translational Neurovascular Research Unit (IRCCS Mondino Foundation) for DNA extraction. DNA samples will be stored at -80°C until shotgun microbiome analysis.

NEUROFUNCTIONAL PROFILING All patients will undergo high density-EEG and subset of 40 patients will also be studied in parallel with resting state- functional MRI at T0.

- HD-EEG: the investigators will randomly acquire 4 recordings (6 minutes each) in resting-state condition, 2 with opened eyes, and 2 with closed eyes. The following frequency bands will be considered: delta (1-4 Hz), theta (4-8 Hz), alfa (8-13 Hz), beta (13-30 Hz), gamma (30-80 Hz).

Acquisition parameters will be: High-Pass: 0.5 Hz; Low-Pass: 100 Hz; Notch: 50 Hz. For analysis of HD-EEG data, a previously developed and validated tailored analysis pipe-line will be used to reconstruct neural sources from cortical/subcortical gray matter. EEG signals will be band-pass filtered (1-80 Hz) and down-sampled at 250 Hz. Biological artifacts will be rejected using Independent Component Analysis (ICA). EEG signals will be referenced with a customized version of the Reference Electrode Standardization Technique (REST). A matrix will estimate the relationship between the measured scalp potentials and the dipoles corresponding to brain sources. Sources reconstruction will be performed with the exact low-resolution brain electromagnetic tomography (eLORETA) algorithm

Statistical analysis As a pilot study, sample size is not calculated and our data will provide the fundament for further confirmatory randomized controlled studies.

The investigators will separately evaluate the clinical effects of Ketogenic Diet and PEA supplementation in NON-responders and PEA supplementation in partial responders.

As statistical test, ANOVA for repeated measure test, with a within-subject TIME factor (T0 vs. T3 vs. T6) will be used. The primary outcome will be the reduction in MMDs at T3 compared with baseline. As secondary outcome the investigators will analyze: modification of monthly headache days, monthly days of acute drug intake, monthly doses of acute drugs, percentage of patients with a 30-50% reduction of MMDs, migraine related disability (MIDAS, HIT-6), and quality of life (MSQ).

Inclusion Criteria:

• male and female patients aged 18 to 75 years
• diagnosis of episodic migraine or chronic migraine according to ICHD-3 criteria
• for episodic migraine: 8-14 monthly migraine days in the previous 3 months
• diagnosis of resistant migraine, defined by having failed at least 3 classes of migraine preventatives and suffer from at least 8 debilitating monthly headache days for at least 3 consecutive months
• patients naive to CGRP targeting treatments

Exclusion Criteria:

• history of major psychiatric or other neurological conditions
• diagnosis of other primary or secondary headache disorders (only sporadic tension-type headache is allowed if the patients can clearly differentiate between the 2 types of headaches)
• clinically significant medical conditions
• chronic pain conditions
• alcohol and/or drug abuse
• pregnancy or lactation

Subjects with chronic and high frequency episodic migraine attending the outpatient clinic of the Headache Science & Neurorehabilitation Unit of the IRCCS Mondino Foundation (Pavia, Italy) and the Neurology Department of the University of L'Aquila (Avezzano, Italy).