Persistent post-traumatic headache: a migrainous loop or not? The preclinical evidence

In the last years, different animal models of mTBI have been used to reproduce the traumatic events preceding PTH and thus allowing the study of this condition and its associated symptoms (Fig. 1). However, it is necessary to specify that, up to date, there are no well-established models of PTH, as all the models are related to TBI.

Models of mTBI can be divided into penetrative and non-penetrative injuries. Among penetrative injuries, two models have been developed: the Controlled Cortical Impact injury (CCI) and the Lateral Fluid Percussion injury (LFP). CCI is realized using a pneumatic impactor that hits the cortex through a unilateral craniotomy [29], whereas LFP induces brain injury by generating a pulse of pressurized fluid to the intact dura mater through a craniotomy [30] (Table 1). Among non-penetrative injuries, in the weight-drop injury model a projectile-shaped weight with a smooth surface falls from a predetermined height and hits the head (fixed or not) of an anesthetized animal [31], while in the blast injury model, the animal is usually exposed to an explosive detonation [32] (Table 1, Fig. 1). It is worth noting that penetrative injury models are the most used in the headache field, although the non-penetrative weight drop model is the most relevant from a translational point of view [1].

From a clinical standpoint, following mTBI, patients may present PTH with clinical features frequently resembling migraine or tension-type headache [33]. Importantly, not all people after mTBI suffer from PTH, and the lack of disease development may also occur in animals used to model it. Considering that animals can not verbally refer pain, in vivo models have focused on pain-related behavioral phenotypes as indicative of headache. The onset of pain-related behaviors not previously present just after mTBI suggests that PTH may be studied in animal models as well. The frequent PTH phenotype similar to migraine and the existence of animal pain-related behaviors aimed to study this primary headache, probably explain why certain behavioral tests have been exploited in the study of PTH [34]. It could be argued that observed features are just modeling TBI without any PTH, and this can not be excluded at all. However, some initial evidence just points to the other direction. For instance, cranial hypersensitivity has been assessed as a marker of cephalic cutaneous allodynia, a common migraine feature [35] that is reported in PTH patients as well [36], reflecting sensitization of the trigeminal system [37]. The behavioral test used to evaluate the mechanical pain hypersensitivity in rodents consists in the application of calibrated or electronic von Frey monofilaments to their head, usually in the periorbital region, assessing head retraction as a response [38]. This test is widely used and has been applied to different animal models, such as CCI [39] and mild-closed head injury (mCHI ) [40], a type of weight-drop injury.

Von Frey filaments can be used to assess also hind paw hypersensitivity, which represents a marker of extra-cephalic allodynia, reflecting central sensitization at a higher level than the trigeminal nucleus caudalis. Nevertheless, this symptom is less commonly described in headache patients, including migraine [37], and in PTH, it has been reported immediately following mTBI in nonfixed head weight-drop injury [41] but not in fixed head weight-drop models [34], making its interpretation difficult in translational terms. If on one side, hind paw hypersensitivity cannot be considered specific for PTH, on the other, it can still be an additional tool to evaluate if PTH involves central sensitization. In this context, provocation studies assessing pericranial and hind paw hypersensitivity after administration of glyceryl trinitrate (GTN) or bright light stress (BLS) have been used to study susceptibility to headache triggers and therefore to investigate the presence of persistent central sensitization, a hallmark of headache chronification, in both migraine [42] and PTH models [34, 41].

However, not all PTH patients present allodynia, and other features have been investigated, using other evoked or spontaneous behavioral models. The multidimensional nature of PTH pain can be studied observing the spontaneous locomotor and exploratory activities in an open field environment, associating headache-behaviors to a reduction in such activities [34]. Other cognitive symptoms are tested by observing the presence of deficits in recognition memory using a Novel Object Recognition test [34], which could reflect the impairment due to the severity of the brain injury. Moreover, the aversive state of pain has been evaluated using the Conditioned Place Preference model and the Condition Place Aversion model. The Conditioned Place Preference model evaluates whether mCHI rodents, compared to sham controls, prefer spending more time in a chamber where a specific treatment is administered [34]. The Condition Place Aversion model assesses if drug-treated rodents no longer avoid chambers where a trigger was previously administered [34]. These behavioral models may not be sensitive enough to distinguish migraine from PTH from a phenotypical standpoint, even though they provide fundamental insight in the evaluation of the efficacy of specific treatments to alleviate headache-like symptoms in PTH. In this context, response to certain treatments could allow the elaboration of hypotheses on specific pathophysiological mechanisms that can be activated in PTH. Moreover, since certain PTH features such as cognitive impairment may also result from the brain injury itself, positive response to migraine treatments with the consequent application of ex iuvantibus criterion may further support their association with migraine rather than the exclusively underlying mTBI sequelae.

Concerning pathophysiological mechanisms, it is worth noting that the current models of PTH have disclosed specific pathways that can be present in primary headaches, such as migraine. Bree et al. showed persistent hypersensitivity to headache triggers in concussed animals [34], a finding that may explain persistent PTH and which is present as well in chronic migraine. In their study, 14 days after mTBI, when the mCHI-evoked cephalic hypersensitivity had disappeared, a systemic administration of low doses of GTN resulted in renewed cephalic tactile hypersensitivity, which could be attenuated by administration of sumatriptan or prevented by chronic treatment with murine anti-calcitonin gene-related peptide (CGRP) antibodies, started immediately after mTBI. Moreover, they showed a sumatriptan-induced conditioned place preference in mCHI animals, but not in sham controls. Considering also that triptans, migraine-specific acute medications, have presynaptic receptors that inhibit CGRP release, overall, these findings suggest that pain-related behaviors in the current mTBI model depend on peripheral CGRP.

In another study [41], cutaneous allodynia (both at periorbital and hind paw level) after mTBI was also attenuated by the administration of murine anti-CGRP antibodies, a fact that once again underlines a CGRP-dependent mechanism in acute PTH. Furthermore, they observed that early treatment after mTBI with anti-CGRP antibodies prevented the re-establishment of cutaneous allodynia after provocation with BLS, but a single administration right before BLS was not able to avoid its onset. These findings support the hypothesis that once central sensitization is established, this headache-related feature becomes independent from CGRP and may involve other mechanisms.

It seems that all the studies mentioned above strongly corroborate the hypothesis that common pathways, especially those involving CGRP, are present in both migraine and PTH, raising the question whether the two conditions could represent a continuum within a spectrum of headache disorders, especially considering that animal models of PTH have achieved reproducing migraine-like features, as seen in humans. On the one side, although being careful as data come from not specific PTH models, the existence of a link between PTH and migraine may be hypothesized; on the other, the same lack of specificity makes it difficult to identify the pathophysiological differences between the two disorders.

However, it is more likely that from distinct initial pathophysiology, some shared mechanisms are activated and converge in a similar subset of characteristics while other features may require the involvement of completely different pathways. For example, Navratilova et al. [41] showed that CGRP plays a major role in acute PTH and in promoting the transition to a persistent form of PTH, while it is probably less determining the symptoms once central sensitization is established, therefore once PTH has already become persistent. This fact seems to be partially different from chronic migraine patients, in whom, first, central sensitization is also present but together with elevated CGRP levels [43], and second, treatment with anti-CGRP antibodies is effective in chronic migraine [42, 44, 45]. Although chronic migraine animal models evaluating specific response to murine anti-CGRP drugs are lacking and human, and animal models are not directly comparable, the coexistence of other different mechanisms in chronic migraine and persistent PTH could still be supposed and should be further investigated.

Considering that no well-established animal models of PTH exist at present, the animal models mentioned above of mTBI displaying headache behaviors have helped the description of some of the pathophysiological pathways considered to be involved in PTH. Further studies need to be conducted to better investigate in PTH the role of certain mechanisms elicited in the study of mTBI and to assess their differential expression in migraine. For example, neurometabolic changes, where neuronal damage may be produced by glutamate release and an increase of extracellular potassium, have been shown in models of LFP concussive brain injury [46]. Other relevant mechanisms include neuroinflammation, as a consequence of glial cell activation after mTBI, as demonstrated in mCHI models that especially highlighted the role of mast cells [47]. At the same time, neuroinflammation seems to promote central nervous system excitability and therefore facilitates cortical spreading depression [48, 49], another pathway that has been observed in animal models of mTBI [39, 50] and implicated in the activation of the trigeminal sensory system.

Hereupon, it is fundamental to dispose of both PTH and migraine models and, although intrinsically different, applying similar measures and comparing them may represent a good strategy to detect similarities and differences in underlying mechanisms, therefore enabling further understanding of these disorders. For example, mechanisms related to cortical spreading depression could be studied, or features such as photophobia and its mechanisms can be investigated. Besides, considering as well that one of the most critical risk factors for developing PTH after mTBI is migraine [19], the study of both PTH and migraine, for example by reproducing mTBI in genetic migraine models, may provide further information, allowing comparison to exclusively migraine or PTH models.

At present, the preclinical study of PTH has to face other significant limitations. For example, current PTH animal models have shown impaired cognitive activities as well as altered responses to BLS, suggestive of migraine-like features as observed in human PTH studies [51]. PTH without migraine features is therefore not well studied, probably due to the lack of representative animal models that reproduce conditions similar to TTH, reflecting the fact that the underlying mechanisms of this disorder are still poorly understood. However, a recent study [51] on human PTH has shown different testing profiles in migraine-like PTH compared to TTH-like PTH, observing more cephalic and extracephalic thermal hypoalgesia that was accompanied by cephalic mechanical hyperalgesia in TTH-like patients. These findings should be further investigated in animal models of PTH with the objective of better characterizing these two subgroups and defining whether certain features, commonly tested, such as mechanical cephalic hyperalgesia are specific of one type or not.

Other questions, such as the multiple concussions and sub-concussive hits and their relation to PTH and its chronification have not been sufficiently addressed. Another issue is that the predisposition towards PTH in humans is not only associated with previous migraines but also with a history of psychiatric illnesses and comorbid psychiatric disorders [52], making it difficult to reproduce these aspects in animal models. All limitations considered, in the absence of better and well-established PTH models, at present mTBI animal models should still be considered useful in studying PTH when pain-related behavioral phenotypes indicative of headache are present.