The updated criteria for headaches, published in 2013 by the International Headache Society (IHS) (2013), consider primary headaches, secondary headaches, cranial neuralgias, central causes of facial pain, and all other headaches. Within this classification scheme, facial pain that is intermittent and shock-like, “recurring in paroxysmal attacks lasting from a fraction of a second to 2 min…without persistent background facial pain…without apparent cause other than neurovascular compression,” is considered classical trigeminal neuralgia, purely paroxysmal type. Facial pain meeting the criteria for classic trigeminal neuralgia, “with persistent facial pain of moderate intensity in the affected area,” is considered classic trigeminal neuralgia with concomitant persistent facial pain. Facial pain that occurs for more than 2 h each day for more than 3 months is considered persistent idiopathic facial pain. Thus, while this system enables some distinction between TN patients with and without vascular compression, further classification of idiopathic facial pain is needed.

Scrupulous examination of the results of MVD to treat TN reveals why classification of a patient’s clinical symptoms is germane to any discussion of TN’s etiopathogenesis. Because it relieves facial pain by eliminating vascular compression of the trigeminal nerve, MVD enables confirmation of the diagnosis of “vascular” TN. The results of several large case series by very experienced surgical teams yielding different success rates for the MVD operation, however, underscore the need for further refinement of the criteria used for classifying this disorder (Lee et al. 2014). Close examination of the inclusion and exclusion criteria reveal why this is so. In 2003 patients who underwent MVD for “typical TN” and 672 patients who underwent MVD for “atypical TN,” the proportions of patients who were free of facial pain at last long-term follow-up (i.e., more than one half of patients were followed for greater than 5 years) were 73.7 and 34.7 % in the typical and atypical groups, respectively. While the stark difference in outcomes between the two groups supports a causal link between typical TN and vascular compression and suggests other mechanisms must account for the majority of cases with constant facial pain, the association between clinical presentation and underlying mechanisms is far from perfect.

Another classification criteria of facial pain is the Burchiel classification system, wherein type 1 TN corresponds to pain that is constant less than 50 % of the time and type 2 TN corresponds to TN that is constant greater than 50 % of the time. Following publication of the Burchiel classification criteria (Burchiel 2003; Miller et al. 2009), however, Heros suggested in an editorial (Heros 2009) that type 2 TN ought to be subdivided into type 2a and type 2b. Type 2a TN begins intermittently and, over time, transitions toward developing a constant component. Type 2b TN, alternatively, begins insidiously as a constant or aching pain (Sekula et al. 2011). This implies that type 1 and 2a are pathologically related, while type 2b may arise via an independent mechanism. One speculative explanation for the rare responses to MVD in patients with type 2b TN includes iatrogenic injury to the trigeminal nerve during surgical manipulation resulting in a rhizotomy-like effect (Adams 1989).

The first evidence that there may be a link between vascular compression and TN was published by Dandy (1934), who observed an intracranial portion of the trigeminal nerve impinged by a blood vessel. The first evidence of a causal link between vascular compression and TN was provided by Gardner who performed, in 1959 (Gardner and Miklos 1959), the first vascular decompression resulting in symptomatic resolution of trigeminal neuralgia. This operation, however, was substantially developed and refined along with the use of the operating microscope by Jannetta. The results of MVD reported by Jannetta (and subsequently others) showed that treatment of the vascular compression resulted in resolution of a patient’s symptoms in some patients (Barker et al. 1996).

Further evidence of a causal link between vascular compression and TN comes from an analogous cranial nerve disorder, hemifacial spasm (HFS). The pathophysiology of hemifacial spasm, a syndrome of unilateral episodic twitching of the muscles around the eye and the lower face, is thought to be analogous to trigeminal neuralgia. Because the seventh cranial nerve is a motor nerve, however, intraoperative electromyography is routinely performed during MVD for HFS. This technique has been employed to show that immediate or delayed cessation of HFS’ signature “abnormal motor response” is often correlated with decompression of offending blood vessels providing indirect evidence that MVD is unlikely to work merely by causing trauma to the nerve.

While it is generally assumed that mechanical stimulation of the trigeminal nerve by the impinging blood vessel is ultimately responsible for the pain of TN, the close association between blood vessels and nerves throughout the body in addition to the relatively large mechanical forces regularly impinging on many peripheral nerves in the absence of pain suggests that mechanical stimulation alone cannot account for the pain of TN. Consequently, investigators have searched for additional underlying mechanisms. Most striking among these is the demyelination observed in the nerve at the site of compression.

Preclinical evidence in support of a link between demyelination and TN-like changes was obtained from Burchiel and colleagues who demonstrated that following iatrogenic demylination of cat and Macaca mulatta monkey trigeminal nerve roots, trigeminal nerve roots from both animals produced extra action potentials in response to stimulation (Burchiel 1980). Suture placement resulted in focal nerve injury and demyelination. Electrophysiological recordings were then performed at intervals of several weeks following suture implantation. The appearance of abnormal action potentials was correlated with the appearance of focal demyelination, which took at least 1 week to occur. Subsequent data from rodent models of peripheral nerve compression confirmed that any compression of a peripheral nerve of sufficient intensity to occlude the local blood supply to the nerve was capable for producing signs of neuropathic pain (Bennett and Xie 1988) and spontaneous activity in the compressed nerve (Kajander et al. 1992).

Clinical evidence of demyelination associated with TN was originally reported by Hilton and colleagues. Staining followed by electron microscopy revealed degraded central myelin at the site of arterial compression in a single patient (Hilton et al. 1994). This observation was subsequently replicated by Rappaport et al. (1997) who published an analysis of biopsy samples taken from the site of compression in 12 patients with trigeminal neuralgia undergoing microvascular decompression. Interestingly, in 11 of 12 patients, both demyelination and axonopathy were noted by biopsy even though only 7 of the 12 patients had intraoperatively confirmed arterial compression of the trigeminal nerve. Furthermore, the authors introduced the concept of the “ignition hypothesis” which holds that damage to the root entry zone of the trigeminal nerve induces parts of the trigeminal ganglion to develop autorhythmicity (Devor et al. 2002).

Jannetta hypothesized that compression-induced demyelination allowed the spread of ephaptic impulse transmission at these sites. These ephaptic impulses, in turn, are perceived as pain (Jannetta 1967). In addition to sites of dymelination at the site of vascular compression, demyelinating plaques of the trigeminal system have instead been implicated as the causative lesion in TN. In an MR study of six patients with symptomatic trigeminal neuralgia, Gass et al. (1997) demonstrated the presence of demyelinating plaques in the trigeminal fibers of all six patients, with the lesions ranging anywhere from pontine trigeminal nuclei to the junction of the central and peripheral myelin, midway through the cisternal segment of the trigeminal nerve. Furthermore, Cruccu et al. showed that patients with symptomatic TN were more likely to have a plaque within trigeminal afferents (i.e., mostly intrapontine) than patients with MS and trigeminal sensory dysfunction other than TN (Cruccu et al. 2009).

In addition to these imaging findings, there exists both intraoperative (Lazar and Kirkpatrick 1979; Love et al. 1998) and postmortem (Rushton and Olafson 1965) evidence for the presence of demyelinating lesions within the trigeminal afferent fibers. Lazar examined a biopsy specimen taken during a complete sensory rhizotomy performed on a patient with trigeminal neuralgia secondary to multiple sclerosis. Histological examination of the specimen demonstrated myelin degeneration consistent with an MS plaque occurring at the root entry zone of the trigeminal nerve. This finding was expanded upon by Love who examined six biopsy samples from patients with TN secondary to MS who underwent a partial sensory rhizotomy. Samples were taken from the most proximal part of the trigeminal nerve near the entry of afferent fibers into the pons. Histological examination showed numerous areas of directly abutting axons and a dearth of normal myelin. While all of these lesions were confined to the centrally myelinated portion of the trigeminal nerve, the area of demyelination extended proximally along the cisternal portion of the nerve to the transition between the distal and proximal portions of the trigeminal nerve (Love et al. 2001). Additionally, Olafson demonstrated, upon autopsy of deceased patient with right-sided symptomatic TN, a plaque “At the junction of the right fifth nerve and the pons” and again in the brainstem near the forth ventricle (Olafson et al. 1966).

One final line of evidence in support of the link between TN and demyelination comes from the observation that the incidence of TN is higher in patients with multiple sclerosis (MS) than in the general population (O’Connor et al. 2008). Additionally, when compared to TN patients without MS, TN patients with MS are more likely both to develop facial pain at a younger age and to develop bilateral facial pain. These facts raise the possibility that it is demyelination per se, if not ultimately plaque formation, rather than the combination of demyelination and vascular compression that is necessary for the manifestation of TN, because in contrast to patients with classical TN, most patients with MS and TN are much less likely to have a compressive vessel and, therefore, much less likely to respond to microvascular decompression (Resnick et al. 1996). Further evidence that the causal factor for the pain associated with TN is dymelination rather than demyelination with vascular compression comes from the observation that while MVD has a high cure rate with a low risk of sensory side effects in patients with type 1 or classical paroxysmal type TN or type 2a, the operation is largely ineffective in patients with other types of TN even when clear neurovascular compression is present.

While evidence in support of a link between vascular compression and demyelination/demyelinating plaques and TN is compelling, there are several lines of evidence suggesting that additional mechanisms are necessary for the manifestation of pain. Most prominent among these is the timing of the pain relief produced by MVD: the resolution of symptoms is almost instantaneous following MVD. In a landmark study, Bunge demonstrated that spontaneous central remyelination can occur in cats following spinal cord injury (Bunge et al. 1961). The authors, however, noted that the very first evidence of new myelin sheaths did not appear until 19 days following injury. If removal of the offending blood vessel does promote remyelination, it would not be expected to produce a clinical effect within seconds to hours. Additionally, as Adams suggests, demyelination is unlikely to account for the characteristic periods of remission and recurrence experienced by most patients with type 1 TN (Adams 1989). Myelin is simply too slow to regenerate in order to account for an episode of spontaneous remission. Additionally, it is unclear why myelin would spontaneously regenerate in the presence of unrelenting vascular compression.

Additional evidence in support of the suggestion that neither MVD nor demyelination alone is sufficient for the generation of pain comes from the relatively high incidence of asymptomatic vascular compression. Cadaveric observations have demonstrated vascular compression of the trigeminal nerve in 16–58 % of asymptomatic individuals, compared to 90–100 % of patients with TN (Haines et al. 1980; Hamlyn 1997a; b). These postmortem data must, of course, be viewed with caution, given the potential for structural changes that occur during the immediate postmortem period including loss of blood pressure and atrophy of the intracranial tissues that make identification of compression more difficult. Nevertheless, this relatively high incidence of neurovascular compression in asymptomatic individuals is consistent with findings in MRI studies. That is, recent technological and methodological advances have given rise to the ability to detect vascular compression of the cranial nerves radiographically, with a high degree of sensitivity and specificity (Sekula et al. 2014). During the dedicated examination of 200 trigeminal nerves in 100 unaffected individuals, vascular compression was observed in 87.5 % of the nerves studied. Moreover, 86 % of these compressive vessels were arteries and the compression occurred along the most proximal portion of the trigeminal nerve 58 % of the time (Peker et al. 2009).