The vasodilation theory of migraine, originally proposed in 1938, centers on vasodilatation of extracranial arteries as the predominant causative factor in the development of migraine pain. Graham and Wolff measured the diameter of temporal arteries during migraine attacks and found them to be dilated (9). When these patients were treated with vasoconstrictors such as ergotamines, the arterial dilation decreased and the pain was relieved. They also observed that physical pressure exerted on the extracranial arteries temporarily decreased the pain. This theory is further supported by the observation that mechanical distention of cerebral arteries can induce pain (10). An additional study found the temporal artery dilated ipsilateral to the side of the migraine pain compared to the contralateral vessel (11). Also, organic nitrates, such as nitroglycerin, which dilate cerebral arteries, often causes headache when administered clinically (12).

The vasodilation theory is not uniformly supported in the literature. Studies have shown that pain and cerebral vasodilation are not always temporally correlated during migraine attacks (13). Transcranial Doppler studies have shown that sumatriptan changed cerebral blood flow before the resolution of migraine pain (14). Additional studies with positron emission tomography (15) and functional magnetic resonance imaging (16) have illustrated that changes in cerebral blood flow do not correspond with changes in pain, and in fact oligemia was observed in posterior parts of the brain when imaged early in the headache period of many migraine attacks, with or without aura (15,17).

Admittedly, the available techniques cannot detect blood flow changes in the dura mater of humans at present, and thus the data supporting or refuting the vasodilation theory of migraine is based solely on cerebral blood flow changes. Laser speckle imaging of dural arterial blood flow following cortical spreading depression (CSD) in animals revealed sustained vasodilation mediated by trigeminal activation while the cerebral cortex was oligemic (18). This technique potentially provides an opportunity to better characterize changes in dural and cortical blood flow during migraine attacks.

Some of the pharmacologic agents mentioned in the previous paragraphs that either constrict or dilate cerebral arteries possess additional in vivo effects. For example, organic nitrates induce the release of inflammatory neuropeptides such as CGRP from perivascular nerve endings of animals in addition to causing vasodilation of vessels (19,20). Several of the vasoconstrictive, therapeutic agents like ergotamines and sumatriptan also inhibit the release of inflammatory neuropeptides (21). These issues have spawned alternative theories of migraine.

To fully appreciate the neurogenic inflammation theory of migraine, an introduction to the concept of neurogenic inflammation is warranted. The roots of neurogenic inflammation reach back to 1874 when Goltz found that stimulation of the sciatic nerve induced vasodilation in the tissue innervated by the sciatic nerve (22). In 1901, Bayliss reported that stimulation of the peripheral end of dorsal root ganglion (DRG) neurons elicited vasodilation in the hind legs of dogs that was not affected by removal of the sympathetic efferents (23). However, this effect was lost following degeneration of peripheral sensory fibers following ablation of the DRG. He went on to use the term antidromic to describe the effect he noted as “impulses passing along sensory fibers in a direction contrary to what is regarded as the usual one,” as now the neurons have an efferent function in addition to the customary role in afferent conduction.

Later experiments by Bruce showed that the chemical irritant mustard oil induced vasodilation and inflammation when applied to the conjunctival membrane of the rabbit eye. This effect could be reduced by degeneration of the trigeminal nerve following transection or by direct administration of the local anesthetics cocaine or alypine into the eye (24,25). Subsequent studies showed that local anesthesia of human skin prevented the cutaneous flare owing to injury (26).

Lewis noted that distal stimulation of cutaneous nerves induced hyperalgesia and reddening of skin, signs of inflammation. This led him to propose the concept of “nocifensor nerves.” He thought these nerves were not sensory nerves, but rather a part of the posterior root nervous system involved with the “local defense against injury” (26).

Jancso et al. observed arteriole vasodilation, enhanced vascular permeability, plasma protein extravasation (PPE), and fixation of colloidal silver into the wall of venuoles in skin following antridromic stimulation of the saphenous and trigeminal nerves (27). They also observed a nasal discharge following antidromic stimulation of the trigeminal nerve. Jansco was the first to collectively refer to these symptoms as neurogenic inflammation. Extravasation of plasma protein was also observed in the skin and visceral organs following antidromic stimulation of dorsal roots (28).

Neurogenic vasodilatation occurs by dilation of arterioles, and PPE (leakage) is caused by enhanced permeability of postcapillary venules. The increase in protein extravasation is apparently not caused by more or larger endothelial gaps, but rather an increased number of endothelial pinocytotic vesicles (29). Constituents released from activated mast cells such as histamine, serotonin or vasoactive intestinal peptide may further increase vascular permeability (30).

One of the tools used to activate primary sensory neurons and induce neurogenic inflammation is the vanilloid capsaicin, the pungent component of hot peppers. Topical application of capsaicin causes neurogenic inflammation, burning pain, and hyperalgesia to heat and mechanical manipulation (31). Capsaicin is unique in that it produces an initial excitation of nerves that is followed by long-lasting desensitization. In fact, the topical effects of capsaicin can be prevented by denervation or prior capsaicin treatment (32,33). These unique characteristics have allowed a special subset of sensory neurons, the capsaicin-sensitive neurons, to be described (28,34). These neurons comprise a major portion of the sensory neuron population and most likely play a major role in neurogenic inflammation as loss of these fibers decreases the neurogenic inflammation owing to activation of sensory fibers by antidromic stimulation (28,35). A majority of capsaicin-sensitive neurons can be destroyed by capsaicin administered to neonatal rats (32,36). The afferent fibers involved in neurogenic inflammation are small diameter, nonmyelinated C-fibers and thinly myelinated Aδ-fibers that synapse in the brainstem and spinal cord (33). The capsaicin-sensitive C-fibers
capable of inducing neurogenic inflammation when stimulated exist in the dorsal, but not ventral roots of rat spinal cord (37).

Capsaicin was shown to evoke depolarization of a subset of neurons and cause release of the neuropeptides SP and CGRP from isolated DRG neurons in vitro (38). Capsaicin also evoked the concentration-dependent release of immunoreactive CGRP (iCGRP) from freshly prepared slices of rat trigeminal ganglia. The capsaicin-induced release of iCGRP was significantly inhibited by pretreatment with the competitive capsaicin-receptor antagonist, capsazepine, and was dependent on the presence of extracellular calcium (39). Arterial infusion of capsaicin into rats excited both the peripheral and central terminals of sensory nerves, in addition to causing SP release at both ends of the nerve (33). Neurokinin A (NKA) is also synthesized and released from capsaicin-sensitive neurons (40).

Following their release, neuropeptides such as SP, NKA and CGRP can bind to specific receptors. The tachykinins SP and NKA have distinct affinities for three neurokinin receptors (NK-1, NK-2, and NK-3) (33). CGRP binds at two distinct receptor subtypes (CGRP₁ and CGRP₂) (41). NK-1 and CGRP receptors are present on vascular smooth muscle and the endothelium of blood vessels (42, 43, 44).

Baraniuk et al. found that antidromic electrical stimulation of the saphenous nerve induced neurogenic inflammation as assessed by monitoring PPE in the skin of rat paws several minutes following stimulation (45). Utilizing immunohistochemical techniques, they showed that nerves containing CGRP-, NKA-, and SP-immunoreactive materials were histologically associated with the venules where PPE took place. These studies also suggested that mast cell activation was not necessary for the extravasation component of neurogenic inflammation (45,46).

SP, NKA, and neurokinin B were individually found to induce PPE, one component of neurogenic inflammation, following injection into the abdominal skin of rats. Injections of CGRP into the skin also caused PPE, but at doses approximately 12 times higher than that required for SP. Co-injection of an ineffective dose of CGRP and SP shifted the dose-response curve for SP-induced protein extravasation to the left by approximately 2 orders of magnitude and increased the maximal effect (47). When added to human epidermis in vitro, SP also induced plasma protein extravastion and vasodilatation in a dose-dependent manner (48). PPE in rat paw skin, induced by either topical mustard oil or capsaicin applied directly on the saphenous nerve, was significantly decreased by systemic administration of either SP or CGRP antibodies used to immunoneutralize released SP or CGRP (49). The increased vasodilatation observed in the skin following mustard oil treatment was also significantly decreased by pretreatment with SP or CGRP antibodies. Also, SP and the NK₁ receptor agonist [Sar⁹, Met(O₂)¹¹]-SP dose dependently increased plasma extravasation (50), whereas CGRP reportedly increased vasodilatation but not plasma extravasation in the rat knee joint (51). Subsequent studies utilizing a more sensitive method of protein quantification in synovial fluid were able to document protein extravasation following CGRP infusion (52). Together, these studies suggest that neuropeptides, particularly SP and CGRP, released from capsaicin-sensitive sensory afferents induce neurogenic inflammation.

Prostaglandins are known to play a role in the development of inflammation and enhance sensitivity to various forms of stimulation. Many inflammatory maladies benefit from treatment with either nonsteroidal anti-inflammatory drugs or specific cyclo-oxygenase-2 (COX-2) inhibitors, which block the production of prostaglandins (53). Vasko et al. evaluated the effect of prostaglandins on the release of SP and CGRP from rat DRG cells in culture (54). They found that prostacyclin (PGI₂) and carba prostacyclin (CPGI₂) potentiated the capsaicin-induced release of immunoreactive SP and CGRP at concentrations of PGI₂ and CPGI₂ that had no effect on release by themselves. CPGI₂ also enhanced the release of SP and CGRP induced by bradykinin and potassium. In a related group of experiments, Vasko et al. showed that pretreatment of DRG cells with PGE₂ also enhanced the bradykinin-induced release of immunoreactive SP and CGRP (55). The bradykinin-induced release of peptides was attenuated by pretreatment with the COX inhibitor indomethacin, most likely because of inhibition of prostaglandin synthesis by the DRG neurons. Recent studies indicate that the PE3C and EP4 prostaglandin E₂ receptor subtypes may mediate the release of immunoreactive SP and CGRP (56). These studies provide further insights into the role of SP and CGRP in neurogenic inflammation, particularly with respect to the contribution of prostaglandins in the inflammatory processes.

In addition to skin (57) and conjunctiva, neurogenic inflammation has been documented in other tissues and organs such as the lower urinary tract (58), reproductive and genital organs (59), eye (60), stomach (61), intestine (62,63), respiratory tract (64,65), synovial joints (66), and the endocrine system (67,68) of laboratory animals or humans. Neurogenic inflammation within the dura mater has also been reported following electrical stimulation of the trigeminal ganglia or intravenous capsaicin administration (69), giving rise to the neurogenic inflammation theory of migraine.

The neurogenic inflammation theory of migraine pathogenesis was originally proposed by Moskowitz (69, 70, 71). This theory supposes that migraine pain is associated with inflammation and dilatation of blood vessels in the meninges, particularly the dura mater, following the
release of inflammatory neuropeptides from primary sensory afferents innervating the dural blood vessels (Fig. 33-1). Studies by Wolff, Chapman, and others set the foundation for this theory in the late 1950s and early 1960s. In their experiments, they collected subcutaneous perfusates from skin overlying painful migraine sites by infusing saline and subsequently withdrawing fluid. They found the specimens contained an active substance or substances they called “headache stuff” that was active in bioassays, but more importantly caused vasodilatation, capillary permeability, and increased sensitivity to pain when injected back into human skin at a different site (72, 73, 74). A bioassay measuring the relaxation of rat duodenum by the “headache stuff” was used for quantification of the active substance. The amount of active substance in the samples was related to the intensity of the headache pain. Together, these experiments suggest neurogenic inflammation may occur during a migraine attack.