Calcitonin gene-related peptide (CGRP): Role in migraine pathophysiology and therapeutic targeting

2.1. Where is CGRP acting in migraine?

CGRP has been shown to act both in the peripheral and CNS in preclinical studies and both sites are likely important in migraine pathophysiology (Figure 1). In the periphery, CGRP targets mast cells, blood vessels, glial cells, trigeminal afferents in the meninges, and neural cell bodies and satellite glia in the trigeminal ganglia (for review see Messlinger and Russo 2018) [33]. In the meninges, CGRP likely contributes to neurogenic inflammation by triggering the release of neuron sensitizing agents from mast cells, which in turn leads to increased vasodilation in the dura [34]. The modulation of neural activity in the meninges could trigger a feedback loop that ultimately results in peripheral sensitization of nociceptors [33]. The efficacy of systemically administered CGRP targeting monoclonal antibodies strongly supports a peripheral role of CGRP in migraine since the antibodies have very poor BBB permeability [35]. Overall, peripheral sensitization is critical for CGRP’s actions and likely sets the stage for CGRP actions in the CNS.

Within the CNS, CGRP and its receptor are present in multiple pathways believed to play a role in migraine pathophysiology [8]. The trigeminal ganglion (located outside of the BBB) projects to the trigeminal nucleus caudalis (TNC) where second order neurons carry the signals to the posterior thalamic area (PTA). We use PTA as a term to encompass all nuclei in the posterior thalamic area. The PTA appears to be a sensory integration center that is abnormal during migraine. Neurons in the thalamus receive input from the TNC and retinal ganglion cells [36], and several key studies have demonstrated in rodents the importance of the PTA in the development of photophobia and highlight the PTA as a possible center for the integration of light and pain [36–38]. CGRP likely contributes to this pathway as both CGRP and its receptors are present in discrete nuclei of the PTA. This is further supported by a study showing that injection of CGRP into the PTA facilitates neuronal firing [38]. Moreover, somatosensory and nociceptive stimuli from ascending pathways converge on the CGRP producing neurons of the subparafasicular and intralaminar nuclei [39]. In humans, the posterior thalamus is known to be activated during migraine attacks and has altered functional connectivity with multiple brain regions [40, 41]. Taken together, these data suggest that pre- and post-synaptic neuromodulatory actions of CGRP that have been reported in other neural circuits [42], could contribute to a state where the PTA is hypersensitive to sensory stimuli.

Pathways contributing to this sensory sensitization are likely to involve the “general alarm” system of the parabrachial nucleus (PBN) [43, 44]. The PBN is a relay station for pain and other sensory input as it travels to the forebrain [43]. The PBN receives direct projections from the trigeminal nucleus, leading to the affective component of pain [45]. CGRP is abundant in the PBN and the PBN projects to multiple brain areas that are thought to be involved in migraine pathophysiology [43]. It is possible that altered signaling in this pathway could underly some of the hypersensitivity experienced during migraine. Peripheral and central actions of CGRP probably work in concert to induce migraine, and the complex nature of the disorder makes it improbable that just one of CGRP’s action is solely responsible for inducing migraine.

2.2. How are the migraine actions of CGRP assessed?

Most of the mechanistic evidence concerning the involvement of CGRP in migraine comes from preclinical studies. CGRP is one of the most well-defined migraine triggers and can induce multiple migraine-like symptoms in animals that mimic CGRP in infusion in humans, including pain-like symptoms. Mechanical hypersensitivity is a common symptom of migraine [46] and can be reproduced in rodents after CGRP administration [47]. Dural administration of CGRP in mice induced periorbital touch hypersensitivity [48]. Intrathecal injection of CGRP induced hyperalgesia in the hindpaw in response to pinch in rats [49] and increased mechanical allodynia in mice [50]. While evoked pain phenotypes are well described preclinically, for many years there was no assay to measure spontaneous pain in animals. In 2010 Mogil and colleagues demonstrated that some types of pain could be assessed through facial grimace without having to evoke a response [51]. We used that assay to demonstrate that peripheral CGRP injection in mice could produce spontaneous pain that could be partially blocked with sumatriptan, suggesting the pain was migraine-like [52]. The eye-opening action unit proved to carry most of the weight for the grimace assay, and we were able to use a continuous objective measurement of eye closure to measure CGRP-induced grimace [52]. With this set of experiments, we also demonstrated that CGRP induced pain was light independent, which suggests that pain and light-aversion (another CGRP induced phenotype) are independent of each other.

Photophobia is one of the diagnostic criteria of migraine (Table 1). Photophobia occurs when ordinarily non-painful light becomes uncomfortable or painful. Patients with migraine are bothered by even dim light and want to escape bright environments [53]. This idea of using increased light sensitivity has translated well into a mouse model. Light aversion can be induced by both central and peripheral CGRP injection in wild-type mice, and it can be treated by triptans suggesting that light-aversive behavior in mice is migraine-like. This behavior requires a bright light stimulus (~25,000 lux) in wild-type mice. In a CGRP-sensitized mouse model that overexpresses human RAMP1 (the rate limiting component of the CGRP receptor) in the nervous system, only a low light stimulus of 55 lux is needed to induce light aversion after central administration of CGRP [54]. However, peripheral administration of CGRP in those mice didn’t induce any light aversion with dim light, suggesting that neural CGRP receptors are not the rate-limiting site of action outside the CNS. Of importance, this light aversion behavior is not due to anxiety as mice show no difference in a light independent anxiety test (open field assay) [54–57]. One interesting feature of the light aversion assay is that mice do not move as much when injected with CGRP, but this is only seen in the dark side of the box [54–57]. This suggests that the mice are not sedated or that their motor activity is not impaired, since they move just as much as vehicle-injected animals in the light. The desire to go into a dark area and rest is consistent with human behavior.

2.3. CGRP-based treatments

In the past years, different molecules have been developed to block CGRP signaling in order to treat migraine symptoms. The first molecules to show potential were the CGRP receptor antagonists called “gepants”. These molecules have a high affinity for the canonical CGRP receptor and prevent CGRP binding and signal transduction. Gepants do not cause direct vasoconstriction [58–61], which is an advantage compared to triptans since the migraine population is known to have an increased prevalence of cardiovascular diseases [62]. Multiple clinical trials showed that intravenous and oral gepants alleviated migraine symptoms acutely (Table 2). In contrast, the prophylactic efficacy of gepants are still debatable as some clinical trials had to be interrupted because of adverse effects, and others are still ongoing (for review see Negro and Martelletti, 2019 [63] and Table 2). The development of several gepants was stopped for different reasons: olcegepant has a low oral bioavailability and telcagepant and MK-3207 were discontinued because of liver toxicity after frequent use [63].

Despite those initial safety concerns, the efficacy of gepants led to further efforts to develop safe CGRP blocking molecules. Three gepants remain in clinical development: rimegepant, ubrogepant and atogepant. In phase 2b clinical trials, efficacy of rimegepant to treat acute migraine was assessed with different endpoints such as pain free, migraine free, photophobia and phonophobia free, and nausea remission [64]. Intermediate doses of rimegepant (75, 150 and 300 mg) were found to be significantly more efficient than placebo. In addition, rimegepant did not show any effect on liver function, suggesting that it is safer than previously terminated gepants. Intriguingly, a higher dose of rimegepant (600 mg) did not significantly differ from placebo [64], and authors hypothesized that it was due to inherent variability present in the patients randomized to this dose group. Following this study, three phase 3, double-blind, randomized, placebo-controlled studies (NCT03235479, NCT03237845, NCT03461757), and a safety study (NCT03266588) [65–67] were started, for which the results are still awaited. Similarly, ubrogepant demonstrated a positive dose-response outcome in the treatment of acute migraine in a phase 2b, double-blind, randomized, placebo-controlled study [68], with minimum adverse effects. However, results from this study are dampened by the high response rate of the placebo group, and the limited number of patients included. Two phase 3, double-blind, randomized, placebo-controlled studies (NCT02867709, NCT02828020) [69, 70] were completed in December 2017 and February 2018. Preliminary results confirm the findings of the phase 2b study. Atogepant has a different chemical structure than other gepants and is currently tested for migraine prophylaxis [71]. The very first results from a phase 2b/3 study (NCT02848326 [72]) show that adults treated with atogepant had greater reduction from baseline in monthly migraine days on average, compared with those treated with placebo [73]. No serious adverse events related to treatment occurred. As we are writing this review, the first gepant (ubrogepant) was approved for acute treatment of migraine with and without aura by the FDA.

Monoclonal antibodies against CGRP (fremanezumab galcanezumab, eptinezumab) and against CGRP receptor (erenumab) are another class of molecules able to block CGRP signaling. Three of these antibodies (fremanezumab, galcanezumab, and erenumab) have recently been approved by the FDA for the prophylactic treatment of migraine and a decision on a fourth (eptinezumab) is expected in 2020. About half of the patients receiving these antibodies experience a 50% reduction in migraine days (Table 2) [74–78]. Interestingly there has been no difference in efficacy between the antibodies, whether they sequester CGRP or bind to the receptor. Moreover, the antibodies still show efficacy for over a month after administration and therefore can be used as prophylactic drugs administered monthly or even quarterly to patients, which presents an advantage compared to daily oral administration for the gepants.

It appears that the newly designed CGRP-blocking drugs are quite safe so far. These safety concerns are serious since the vasodilatory action of CGRP could be critical during certain pathological states like stroke [79, 80]. From the many clinical trials and being on the market for almost a year, it appears that CGRP and CGRP receptor antibodies are safe and well tolerated. What is not known are the long-term effects of CGRP blockade, although it is encouraging that Amgen/Novartis found their antibody to be safe so far at the three-plus year point from an ongoing five-year open label study [81]. In addition, a study with patients with angina found no detrimental effect of the antibody [82]. This study however presents several limitations such as (1) the inclusion of 78% of male subjects in a predominantly female disorder, (2) the inclusion of patients suffering from stable angina pectoris, when patients suffering from a microvascular disease would better represent the population at risk, and (3) assessing the effects of the drug at an early time-point at which it is unsure if the receptor antibody has had time to affix the receptor [79]. Cardiovascular concerns are paramount since migraine patients are known to be at a high risk of stroke and cardiovascular disease (for review see Sacco and Kurth, 2014 [62]). Could blockade of CGRP increase the severity of stroke? In one recent case report, one patient developed an ischemic event following treatment with a CGRP antagonist drug [83]. However, no conclusion should be made from a single patient report. Long term studies on cardiovascular health are warranted, beginning with animal studies investigating the role of CGRP blockade on ischemia.

2.4. CGRP receptors, signaling, and regulation

CGRP belongs to a family of structurally related peptides among which two other major members are amylin and adrenomedullin. Amylin and CGRP have a close molecular composition which explains their overlapping activity at different receptors [84]. Structure, activity and trafficking of those peptides at their receptors have been extensively reviewed by the Hay and Walker labs [84–87]. Briefly, all receptors for this family are composed of a G-protein coupled receptor (calcitonin receptor-like receptor (CLR) or calcitonin receptor (CTR)) associated to an accessory protein (receptor activity-modifying protein (RAMP) 1, 2, or 3). A recent finding is that CGRP can bind and activate both CLR/RAMP1 (canonical CGRP receptor) and CTR/RAMP1 (amylin receptor 1 (AMY₁)) equally. As a result of the late discovery of CGRP binding AMY₁, the majority of migraine research has focused on the canonical CGRP receptor. This problem was exacerbated by the lack of properly validated antibodies for the detection of the different components involved in these pathways. It is therefore a common assumption that CGRP-blocking drugs only block signaling through the canonical CGRP receptor. Even antagonists designed to selectively bind to CLR/RAMP1 are at best “relatively selective”, and this selectivity depends on which agonist is used, and some molecules have not yet been properly investigated [86].

The signaling of CGRP receptors at the plasma membrane generally involves cAMP pathways, but MAP kinase and calcium pathways can also be recruited [88]. However, signaling does not end at the cell surface. The complex CLR/RAMP1-CGRP is internalized into endosomes in response to CGRP binding [89, 90]. A recent study demonstrated that the CGRP receptor can signal from endosomes and that blocking this signaling can have antinociceptive effects [91]. It is unknown whether this signaling is involved in migraine pathogenesis. While the complex CTR/RAMP1-CGRP is also internalized, this phenomenon does not always depend on activation by an agonist [92] and the possible endosomal signaling for this receptor has not yet been investigated.

2.5. Conclusion

Even after the approval of the first molecules targeting CGRP or its receptor, research is still ongoing to elucidate where and how CGRP is involved in migraine pathophysiology. Deepening our knowledge of CGRP signaling in both physiological and pathophysiological conditions is crucial to optimize current and future treatments for migraine.