BIOmarkers of MIGraine: a Proof of Concept Study Based on the Stratification of Responders to CGRP Monoclonal Antibodies

Almost 15% of European citizens suffer from migraine and its comorbidities. Migraine is the second most disabling chronic neurological disease, with devastating repercussions on the life of those who are affected, their relatives and society. Migraine manifests in recurring attacks lasting 4-72 hours characterized by severe head pain, often throbbing, associated with nausea, photo and phonophobia, vomiting, cognitive deficits, severe emotional distress and complete disability. If not diagnosed and appropriately treated, migraine can transform into a chronic form (15 or more days of headache/ month) often associated with acute medication overuse and severe comorbidities such as depression, anxiety and panic disorders.

The disability associated to migraine is closely related to the frequency of the attacks. Epidemiological studies show that in Europe 34$ of migraineurs suffer from more than 5 headache days per month and 1-3% of the general population has migraine on more than 15 days/month, thus qualifying for the diagnosis of Chronic Migraine (CM).

Migraine is a genetically-driven, chronic "neurosensorial" disorder, where the unpredictable recurrence of the attacks is modulated by the variable efficiency of the fine-tuned interplay of the brain with dynamic sensory stimuli originating from the internal or external environment.

Migraine can be viewed as a cycling functional disorder that affects several areas of the brain, involving multiple pathways ultimately leading to the excessive activation of trigeminovascular afferents in the meninges with the local release of vasoactive substances (calcitonin gene-related peptide - CGRP ) and the activation of the neuroinflammatory cascade.

Unlike several other chronic neurological diseases, migraine can be successfully treated and prevented. Unfortunately, this is true only for subgroups of patients and for a limited period, due to the nonspecific therapeutic armamentarium available: calcium antagonists, beta-blockers, antidepressants, antiepileptics and, only for chronic migraine, onabotulinumtoxin-A. This is the direct consequence of the complex pathophysiology of the disease, but mostly and, more importantly, by the total absence of biomarkers of disease and of responsiveness to available drugs.

In the last years, enormous efforts have indeed been undertaken to identify biomarker of migraine, especially for the chronic form. Functional imaging has provided important clues on the machinery underlying attack generation and imaging biomarkers for chronic migraine, whereas it has become evident that epigenetic processes - DNA methylation in particular - may play an important role in the phenotypical expression of migraine. Attack recurrence and the associated changes in neuronal activity may indeed cause epigenetic changes that alter synaptic plasticity, thus leading to more frequent migraine attacks and ultimately promotes stable epigenetic changes in a feed-forward loop. Circulating levels of CGRP have also been evaluated as a possible peripheral biomarker of migraine in consideration of the seminal observations that CGRP levels increase in the external jugular vein during a migraine attack.

The second half of 2018 marked a revolution in the treatment of migraine: the advent of the monoclonal antibodies (mABs) directed against CGRP or its receptor. They proved safe and effective in reducing migraine frequency and represent the first class of preventive drugs targeting a specific step in the cascade of events leading to the migraine attack. Results from clinical trials have shown that approximately 1/3 of patients are excellent responders (>75% reduction in headache days) whilst 1/3 patients are non- or poor responders (<25% reduction in headache days) to CGRP-targeted mABs.

CGRP-targeted mABs do not interfere with other systems or medications. Consequently, the fact that patients can be stratified as responders and non-responders creates a unique opportunity to better understand the difference of response in patients clinically alike and, through this, provide unprecedented insights into the pathophysiology of migraine allowing the migraine field to move towards a pathophysiological driven classification.

For decades the migraine scientific community has been searching for biomarkers associated with the disease or with its response to treatment. Unfortunately, to date none has been validated. This failure is likely related to multiple reasons: the complex and dynamic nature of migraine-associated biological processes (attack vs. pre-post interictal phase), the multiplicity of neurotransmitters (serotonin dopamine, neuropeptides, etc.), the neuroanatomical structures (peripheral, central) and the interindividual variability of disease expression. Moreover, the reasons for the failure of previous studies are also associated with the variability between studies, in terms of methodology, population and endpoints, which has led to inconsistent findings and low estimate precision.

Now, a selective class of preventive drugs that specifically target a biological step in migraine disease has become available, prompting the possibility to precisely stratify migraine subjects on the basis of their pharmacological response. The availability of high-throughput technologies is providing unprecedented opportunities to identify molecular markers of complex disease by extracting meaningful signatures from complex and heterogeneous sets of data. Molecular biomarkers of thoroughly stratified populations can be combined with deep clinical phenotyping and also with epigenetic, pharmacogenetic and neuroimaging variables, that are relevant for migraine. The availability of migraine-specific animal models also contributes to identifying and firmly placing important pieces to the migraine puzzle.

Against this background, a hybrid data-driven and knowledge-based approach building on clinical and preclinical inputs and including phenotype, genetic, neuroimaging and animal data will provide a large informative dataset that can be effectively analyzed with machine and deep learning algorithms for feature selection, modeling and for subsequent cross-validation to ultimately yield a precise, robust and reproducible model for predicting the response of migraine to a specifically targeted preventive treatment. This will be a crucial step in the therapeutic decision process, especially in the most severe and disabling form of migraine, chronic migraine. It will also provide additional positive implications in the field, such as prompting the ground for a pathophysiology-driven diagnosis and, equally important, robust scientific evidence to European decision makers and health-care providers for optimizing the distribution of resources and rationalizing the access to care of migraine sufferers. Engagement of patients is envisaged both in the planning and conducting of the study to ensure that project methodology and results are in line with their needs and expectations, thus favoring the translation into clinical practice.

Monoclonal antibodies targeting CGRP (CGRP-mABs) have been shown to be effective in the preventive treatment of migraine. They are large molecules that do not cross the blood brain barrier, have a long half-life (20-45 days) and do not interact pharmacologically. They are administered parenterally, which warrants less intra-individual variations and a better adherence to treatment.

CGRP-mABs have proven safe and effective in multiple phase II and III randomized controlled clinical trials and were approved by FDA and European Medicines Agency (EMA) as a preventive treatment for migraine. Four CGRP-mABs have been studied and 3 of them have been approved by EMA in late 2018 and 2019: erenumab, galcanezumab and fremanezumab.

A strong body of evidence suggests that CGRP-mABs induce a 50% response (50% reduction in monthly headache days) in about 50% of subjects, with a 75% response in 30-40% of patients. The latter group, from now on defined as 'excellent responders', represents an interesting subgroup of migraine subjects whose disease is likely particularly dependent on CGRP-related pathways. Hence, the availability of mABs targeting CGRP together with the demonstration of their efficacy in relevant subgroups of subjects prompt a formidable tool to tackle precisely a subset of mechanisms and pathways underlying the disease.

CGRP-targeted mABs do not interfere with other systems or medications. Consequently, the fact that patients can be precisely stratified as excellent responders and non-responders creates a unique novel opportunity to unravel hitherto concealed differences between patients with the same diagnosis and, through this, provides unprecedented insights into the pathophysiology of migraine. Ultimately, this will allow the migraine field to move towards a neurobiological definition of responders and a pathophysiological driven classification. Defining predictors to this response will in turn allow obtaining a more cost-effective approach to migraine healthcare. Engagement of patients is envisaged both in the planning and conducting of the study based on the long-standing collaboration with the European Migraine and Headache Alliance (EMHA), which has been aligned and involved in the preparation of the present proposal.

Epigenetics is acquiring an increasing importance in the panorama of biological events that leads to the transformation of episodic migraine (EM) into chronic migraine (CM). Frequent recurrence of migraine attacks may indeed promote stable epigenetic changes in a feed-forward loop thus altering synaptic plasticity. So far, the most significant associated CpG sites related to migraine chronification are SH2D5, NPTX2 and RAMP1, three brain-expressed genes involved in the regulation of synaptic plasticity and CGRP modulation. However, it is still not known whether a specific profile of DNA methylation, which is a stable epigenetic chemical marker crucial for many cellular activities, can predict the clinical response to preventive therapy in migraine patients.

Neuroimaging Functional imaging studies have revolutionized this area and provided unique neuroanatomical and functional insights into migraine machinery. Functional imaging has revealed that the pathophysiology of migraine crucially involves hypothalamo-thalamo-brainstem networks. Pharmacological imaging has begun to shed light on how medications may enfold their effect in migraine.

The combination of structural and functional imaging parameters with and without external input will give us the unique opportunity to characterize patients who respond to CGRP_targeting antibodies against those who do not. These data will then be an additional powerful source for informing animal projects and also machine learning algorithms.

The overarching and main objective of this proposal is the identification of predictive biomarkers of response to treatment with CGRP-targeted mAbs in chronic migraine, the most severe and disabling form of migraine.

Screening will be performed at visit 1 (month 0). Patients will sign the informed consent and will be instructed to fill in a headache diary and will return after 1 month (+2 days) for visit 2 (V2). At V2, after verification of inclusion/exclusion criteria, enrolled subjects will fill in the protocol scales (Assessment of disability - monthly MIDAS, Headache Impact Test 6 (HIT-6) -, depression - Beck Depression Inventory, BDI-II -, anxiety - Beck Anxiety Inventory, Beck Anxiety Inventory (BAI) -, quality of life - MSQ- , subjective cognitive impairment scale for migraine attacks - Mig-SCog -, comorbidities, acute medications intake, anxiety and depression - Beck scales - and Emergency Room Use), they will undergo blood sampling, neuroimaging and they will receive the first treatment administration.

Healthy controls will be screened at V1 and, following the verification of the satisfaction of inclusion/exclusion criteria and the signature of the informed consent, will fill in the protocol scale, they will undergo blood sampling and neuroimaging.

The first dose (defined according to the indication approved by the Local Health Authorities) of the CGRP-targeting monoclonal antibody will be administered during the hospital visit.

The patient will receive 2 additional doses of the drug and a training on the modality of self-injection and specific indication of the day they will perform the self-injection. They will also receive the headache diary and associated instructions for filling it in regularly during the following 3 months.

No drug treatment is foreseen for healthy controls.

No concomitant migraine preventive treatment will be allowed during the screening, baseline and the 3-month treatment period.

Abortive medications will be allowed, but the type and formulation should be kept stable during the study. Data about their doses and frequencies will be recorded.

A remote visit will be scheduled 30 (week 4) and 60 (week 8) days after day 1 to monitor patient's health status, remind the treatment schedule and collect the data from the headache diaries (visit 3 and 4).

At visit 5 (week 12), all migraine subjects will undergo:

• Assessment of disability (MIDAS, HIT-6), work impairment (WPAI scale), evaluation of acute medications intake, anxiety and depression, quality of life (MSQ), Emergency Room Use.
• Assessment of treatment efficacy through the headache diaries (migraine days/month)
• Assessment of treatment satisfaction with validated patient reported outcomes (PROs).

Excellent responders and non-responders will also undergo neuroimaging evaluation: 3Tesla Magnetic resonance imaging (MRI) involving Structural (sMRI) resting state MRI (rsMRI), Arterial Spin Labelling (ASL) and Diffusion Tensor Imaging (DTI).