Laboratory tests of headache disorders – dawn of a new era?

Palpometer test

It is of great interest to investigate the degree of myofascial tenderness in primary headaches, particularly in TTH. In the new revision of the ICHD third edition beta (ICHD-3 beta) (1), frequent episodic (eTTH) and chronic TTH (cTTH) is subdivided into patients with or without pericranial tenderness, which is not based on patient reporting, but on manual palpation by the examiner. In 1994 Bendtsen et al. (7) presented the so-called palpometer, which consists of a force-sensing resistor (FSR) connected to a voltmeter. The FSR is a thin polymer film device that is attached to the fingertip by means of thin adhesive tape (7). The device allows manual palpation with a defined and constant pressure, so called pressure-controlled palpation (PCP). Unfortunately, only four studies have so far used PCP. Bendtsen et al. (8) showed that using conventional palpation even among experienced observers the interobserver variation was 25% and the intraobserver variation was 7%. However, using the PCP reduced the interobserver variation to 7% and the intraobserver variation to 3% (8). Thus, the large interobserver variation using conventional manual palpation is likely to result in over- or underdiagnosis of TTH with pericranial tenderness (1) depending on the investigator. PCP also demonstrated increased pericranial myofascial tenderness in cTTH (9), which was confirmed in a later study showing increased pain sensitivity in cTTH patients over the trapezius muscle, but not over the anterior tibial muscle (10). Using the PCP method, it was shown as expected that the stimulus-response function recorded from the trapezius muscle in healthy controls was best described as a power function (i.e. pain intensities increased in a positively accelerating fashion with increasing pressure intensities) but, unexpectedly, the relation between pressure and induced pain was linear in the highly tender muscles of cTTH patients (9). The same pattern was seen when comparing the most tender cTTH group (linear pattern) to the least tender cTTH group (power function pattern) (9). Later, Ashina et al. also showed increased pain sensitivity in the trapezius muscle in cTTH using PCP (10). PCP may thus be a useful approach to study the status and the effect of treatment in eTTH with cTTH patients. Whether increased pericranial myofascial tenderness exists in other primary headache conditions remains to be studied with PCP. The principle of using FSR for palpation was patented before the above-mentioned studies. The inventors never studied the equipment and never produced a palpometer for commercial use. This unfortunate situation has hampered widespread use of PCP in headache clinics. The method is, however, ready for implementation in headache centers that may have a copy of this relatively simple equipment made locally. In routine use it can help in subdividing TTH patients into those with and those without muscle tenderness, but it may also be useful in subtyping other primary headaches and cranial neuralgias. PCP has not been tested in migraine, but may be useful in assessing those with a possible muscle component as a trigger of the pain. Furthermore, PCP may be effective as a tool to monitor changes over time and a way to measure the effect of pharmacological and non-pharmacological treatment.

Neurophysiological studies

From a diagnostic perspective, it is of great interest to study differences between headache subtypes using neurophysiological tests. A vast amount of studies investigated functional changes between attacks and differences between migraineurs and healthy controls (11). Unfortunately, only a few studies have investigated differences between migraine and other headaches.

Visual-evoked potentials (VEPs)

Numerous studies have demonstrated that episodic migraine (EM) patients show a habituation deficit interictally, which reverses ictally (20,21). Visual-evoked responses to left hemifield checkerboard reversals have been recorded in EM patients and chronic migraine patients (CM) (interictally). Only EM had a habituation deficit interictally. Habituation was normal in CM and EM ictally (22). Interestingly, this habituation pattern changed in CM patients who remitted to EM after prophylactic topiramate treatment (23). To investigate possible central nervous system (CNS) generators of photo- and phonophobia, Sand and Vingen investigated if VEP or auditory evoked potential (AEP) amplitude correlated with light and sound discomfort thresholds in 21 migraine patients, but the study was negative (24). TTH showed no habituation deficit in VEPs, neither for eTTH (25,26) or cTTH (26). The same negative results were shown in cluster headache (CH) during and outside a cluster period (25). Blinded studies of the sensitivity and specificity in separating different types of headache are needed for these methods to be considered in the clinic.

Nociceptive-specific blink reflex (nBR)

All primary headaches are likely caused by activation and sensitization of the trigeminal nerve and/or the upper cervical nerves (27,28), which innervates the skull, eyes, nose and dura mater. It may therefore be diagnostically important to test the function of the trigeminal nerve. The trigeminal reflexes include the blink reflex, the masseter inhibitory reflex, the cornea reflex and the jaw reflex. The blink reflex is probably the most widely used in headache research. Its afferent leg is the trigeminal nerve and the efferent leg is the facial nerve with synapses in the brainstem. Following electrical stimulation of the supraorbital nerve, three components are detected: an ipsilateral early R1 component (onset latency 11 ms) and two polysynaptic bilateral medullary components R2 and R3 (onset latencies 33 and 84 ms) (29). A concentric surface electrode was developed (30) that produced a pin-prick-like pain. It preferentially stimulated nociceptive A-delta fibers without co-stimulation of non-nociceptive A-beta fibers and therefore mediated a nociceptive R2 response. The nociceptive R2 component was attenuated by local anesthetics (30) and central analgesics (31) and was therefore believed to be nociceptive specific. This nBR has been tested in migraine (32), TTH (33), CH (34) and TN (35).

In migraine, the R2 amplitude is increased and latency decreased during migraine attacks, predominantly on the headache side (31,32), which likely reflects sensitization (Table 1(a)). Furthermore, a lack of habituation of the nBR response, which was inversely related to attack frequency, was found in the interictal phase (36–38). Interestingly, the habituation deficit was also found in asymptomatic individuals with a family history of migraine (37), suggesting that deficient nBR habituation is a marker of the predisposition to common migraine subtypes. In contrast, the rare familial hemiplegic migraine (FHM) subtypes 1 and 2 showed in one study increased habituation in comparison to healthy controls (39).

OPEN IN VIEWER

Table 1A.

Blink reflex migraine.

First author (year)	Participants	Mean age (range)	Timing of recordings	Methods	Main findings
Katsarava (2002) (32)	14 mixed migraine patients 14 CTRLs with sinusitis	36 (24–56)	2–8 h after migraine attack onset and interictally	nBR	Increase of R2 amplitude and decreased onset latency on the pain side during attacks only in migraine patients.
Kaube (2002) (31)	17 MO patients	40 (24–56)	≤6 h after migraine attack onset and interictally	nBR	Decreased R2 onset latencies during acute migraine attack compared with the headache-free interval, which was most pronounced on the headache.
Katsarava (2003) (36)	17 MO patients 15 CTRLs	40 (24–56) 35 (27–47)	≤6 h after migraine attack onset and interictally	nBR	Decreased habituation of the nBR responses outside migraine attacks, which normalized during attacks.
Di Clemente (2007) (37)	16 MO patients 14 CTRLs with first-degree relative with migraine 15 CTRLs	28 24 24	Interictally	nBR	Migraine patients interictally show decreased habituation, which is inversely related to attack frequency. A habituation decrease is also found in asymptomatic individuals with a family history of migraine.
Di Clemente (2005) (38)	15 MO 15 CTRLs	28 24	Interictally	nBR and pattern reversal-visual-evoked potentials	Positive correlation between habituation of pattern reversal-visual-evoked potentials and nBR habituation positively correlated with attack frequency.
Hansen (2011) (39)	9 FHM (5 FHM1, 4 FHM2) 7 CTRLs	38 (20–63) 29 (28–31)	Interictally	nBR	FHM patients are characterized by an increased habituation.
Coppola (2007) (40)	14 MO patients 15 CTRLS	31 30	Interictally	nBR	The nBR recovery curves were normal in MO patients compared to healthy CTRLs.
Avramidis (1998) (41)	19 MO patients 10 eTTH 30 CTRLs	38 36	Ictally	Conventional blink reflex	R2 amplitudes were low ictally in MO patients, compared with the healthy control group. The blink reflex was normal in all eTTH patients. Subcutaneous injection of sumatriptan increased R2 amplitude values to CTRL levels in 53% of MO patients.
de Tommaso (2002) (42)	50 MO patients 20 MA patients 35 CTRLs	34 37	Interictally	Conventional blink reflex	A reduced R3 threshold with a normal pain threshold in MO and MA patients; the R3 component was not significantly correlated with the pain thresholds in patients and controls. The R2 and R3 components were less influenced by the warning of the stimulus in MO and MA patients, in comparison with the control group.

MO: migraine without aura; CTRLs: healthy controls; FHM: familial hemiplegic migraine; eTTH: episodic tension-type headache; nBR: nociceptive-specific blink reflex; MA: migraine with aura.

CTTH, but not ETTH, displayed significantly lower R2 amplitude, negatively correlated with the frequency of headaches using conventional blink reflex recordings (43) (Table 1(b)). It was suggested that R2 suppression is a marker of decreased brainstem excitability, which may be the result of excessive descending inhibitory influences. In contrast, another study showed that the nBR parameters of cTTH patients were the same as in healthy controls (44). Furthermore, most studies using conventional BR have shown no difference between TTH and healthy individuals (45,46). No clear differences have been shown between migraine and TTH patients (45,47).

OPEN IN VIEWER

Table 1B.

Blink reflex tension-type headache.

First author (year)	Participants	Mean age (range)	Timing of recordings	Methods	Main findings
Aktekin (2001) (45)	20 MO patients 20 cTTH patients 32 eTTH patients 20 CTRLs	34 (20–54) 37 (27–50) 32 (17–45)	Interictally	Conventional blink reflex	Latency of the R1 and R2 and the amplitude of the R2 were not significantly different between the groups.
Lindelof (2009) (48)	20 cTTH patients 20 CTRLs	42 40	Ictally	Conventional blink reflex after hypertonic saline-induced neck pain	No clear difference in blink reflex integral after conditioning tonic neck pain between CTTH patients and CTRLs.
Sohn (2013) (43)	40 eTTH 32 cTTH 40 CTRLs	53 56 53	Ictally	nBR	CTTH patients displayed significantly lower amplitude and AUC values of the R2 component for the nBR compared with those displayed by ETTH patients and controls.
Peddireddy (2009) (44)	30 cTTH 30 CTRLs	47 48	Ictally	nBR	No different blink reflex response in CTTH patients.
Sand (2011) (46)	11 cTTH 11 migraine 9 CTRLs	39 33 37	NS	Conventional blink reflex	No R1 or R2 latency differences between the groups.
Filatova (2008) (47)	25 CM 25 cTTH 19 CTRLs	45 40	Ictally	Conventional blink reflex	Thresholds for R3 were lower in all patients. No differences between CM and CTTH were observed.

MO: migraine without aura; cTTH: chronic tension-type headache; eTTH: episodic tension-type headache; nBR: nociceptive blink reflex; CTRLs: healthy controls; CM: chronic migraine; AUC: area under the curve; NS: not stated; h: hours.

In CH, the blink reflex was studied with conventional technique (Table 1(c)). The R2 amplitude on the pain side was significantly lower than on the non-pain side (p = 0.005) (34). The R2 area under the curve (AUC) was lower on both sides in episodic CH (ECH) compared to healthy controls interictally (49). Others have shown no significant change of R2 threshold, latency or AUC in CH patients compared to healthy controls (50). In contrast, by using the nociception-specific blink reflex, Holle et al. (51) demonstrated in ECH patients that the AUC ratio was increased only during the cluster periods. Furthermore, the nBR latency ratio (headache side/non-headache side) was decreased in all CH patients compared with healthy controls independent of CH subtype (51). Thus, conventional blink reflex recordings seem to show inhibition or no effect of the blink reflex in CH, whereas the nBR shows facilitation. Habituation of the blink reflex has also been studied in ECH and chronic CH (CCH) patients during a cluster period using nBR. There were no differences in habituation between CH and healthy controls or between headache side and non-headache side (51). In contrast, Perotta et al., using conventional blink reflex technique, found a reduced habituation in the R2 component on the pain side during the cluster period (52). Using nBR, in ECH habituation was significantly reduced on the pain side compared to the non-pain side and to healthy volunteers (49).

OPEN IN VIEWER

Table 1C.

Blink reflex and cluster headache.

First author	Participants	Mean age (range)	Timing of recordings	Methods	Main findings
Raudino (1990) (34)	12 CH	?	?	Conventional blink reflex	Amplitude of the contralateral R2 response on lower on the symptomatic side.
Holle (2012) (51)	8 ECH inside cluster 28 ECH outside 20 CCH 30 CTRLs	43 (19–60) 42 (19–67) 47 (19–60) 44 (20–68)	5 h > last attack	nBR	Habituation similar in CH subtypes and healthy controls. No side-to-side differences of habituation between headache sides.
Holle (2012) (51)	8 ECH inside cluster 28 ECH outside 20 CCH 30 CTRLs	43 (19–60) 42 (19–67) 47 (19–60) 44 (20–68)	5 h > last attack	nBR	nBR latency ratio (headache side/non-headache side) was decreased in all CH patients independent from CH subtype compared with healthy controls, indicating central facilitation at brainstem level. AUC ratio was increased in patients with ECH inside bout only.
Perrotta (2008) (52)	27 ECH patients 22 MO patients 20 CTRLs	37 (24–67) 36 (23–58) 32 (23–49)	Interictally	Conventional blink reflex	CH patients show decreased habituation of the R2 and R3 components on the symptomatic side. Deficits were even more pronounced than those observed in MO patients.
Coppola (2014) (49)	18 ECH 18 CTRLs	40 (27–52) 40 (26–60)	Interictally	nBR	Reflex area decreased on both sides in CH patients, whereas habituation was significantly decreased only on the headache side. Habituation slope positively correlated with the number of days since the onset of the bout and the daily attack frequency.
Lozza (1997) (50)	10 ECH during cluster 10 CTRLs	NS	Interictally	Conventional blink reflex after conditioning by supraorbital or index finger stimuli	No change of R2 threshold, latency or area in CH patients. After paired supraorbital stimuli, R2 recovered more rapidly in patients on the symptomatic side. After index stimulations, R2 recovery was more rapid both on symptomatic and non-symptomatic sides in patients compared to CTRLs.

CH: cluster headache; ECH: episodic cluster headache; CCH: chronic cluster headache; CTRLs: healthy controls; nBR: nociception-specific blink reflex; MO: migraine without aura; NS: not stated.

In TN, a joint report from the American Academy of Neurology (AAN) and European Federation of Neurological Societies (EFNS) (53) concluded that abnormal trigeminal reflexes are useful indicators of symptomatic TN with a sensitivity of 77%–93% and specificity of 91%–97%. This is based on evaluation of blink reflex testing of the R1 after supraorbital stimulation (for ophthalmic division), early masseter inhibitory reflex (SP1) after infraorbital stimulation (for maxillary division) and SP1 after mental stimulation or mandibular tendon reflex (for mandibular division). In classic (i.e. idiopathic) TN, Obermann et al. (54) showed prolonged nBR R2 latencies and reduced amplitudes comparing symptomatic and nonsymptomatic sides in 42 TN patients (Table 1(d)). Subdividing the TN patients into patient groups with and without chronic facial pain (18, respectively, and 24 patients) did not show differences in nBR results between groups. The authors concluded that this may be due to impairment of the trigeminal nociceptive system by demyelination and/or axonal dysfunction on the symptomatic side. In another study, the BR R2 component was delayed in nine of 28 TN patients, which was reduced to six of 28 after percutaneous glycerol rhizotomy (35). In contrast, Cruccu et al. demonstrated normal blink reflex R2 component latencies in nine TN patients (55).

OPEN IN VIEWER

Table 1D.

Blink reflex trigeminal neuralgia.

First author	Participants	Mean age (range)	Timing of recordings	Methods	Main findings
Obermann (2007) (54)	24 classic TN 18 TN with concomitant chronic facial pain	65 (45–85) 62 (44–82)	Without pain medication for 24 h	nBR	Prolonged nBR latencies on symptomatic side. nBR results similar across groups.
Kumar (1995) (35)	28 TN before and after Gasserian ganglion glycerol rhizotomy 6 CTRLs	45 (28–70) NS	NS	Conventional blink reflex	Postoperatively R1 and ipsilateral R2 latencies from the side of the injection deteriorated and consensual R2 latency improved, signifying better function on the contralateral side following.
Cruccu (1990) (55)	30 TN 20 symptomatic trigeminal pain	NS (38–77) 50 (25–77)	NS	Conventional blink reflex	R1 and R2 delayed in 45% and 20% with symptomatic trigeminal pain. Only one (3%) with delayed R1 in TN patients.

TN: trigeminal neuralgia; nBR: nociception-specific blink reflex; CTRLs: healthy controls; NS: not stated; h: hours.

Overall, the nBR seems to give most valuable information only in migraine, CH and TN. If an increased R2 amplitude and decreased R2 latency is a sign of central sensitization, it is surprising that it has not been demonstrated clearly in TTH. The blink reflex may be useful in migraine and TN as a way of subclassifying patients, but whether this will have clinical and therapeutic consequences remain to be investigated.

Spontaneous EEG recordings

According to EFNS guidelines, EEG is not useful in the routine evaluation of headache patients (5) unless an underlying epilepsy disorder is suspected. However, EEG may be useful as a diagnostic tool to investigate neurophysiological phenotypes in primary headache. Three independent groups showed that MA patients interictally have interhemispheric asymmetry in frequency and power in the alpha range primarily in the occipital region in comparison to normal individuals (56–58), but MO patients did not (56). Repeated recordings in the same participants over the migraine cycle showed that the asymmetry increased in the preictal phase (<3 days before headache) (57). Other studies failed to reproduce these findings, but demonstrated instead interictal bilateral increase in theta power in the parieto-occipital regions (59) or globally (60) in mixed groups of MO and MA patients. The increase in theta power was positively correlated to headache intensity, but negatively to age and headache history (60). In the delta range power increased 36 hours (h) before the next attack compared with the interictal period (61).

At present, the overall understanding and clinical usefulness of EEG in headache disorders is non-existent. It remains to be shown whether high-density EEG with advanced computer analysis may improve the situation. So-called pharmaco-EEG has not yet been applied to headache studies.

QST

QST is a widely accepted tool to investigate patients in pain clinics. In neuropathic pain QST has been useful for 1) the identification of subgroups of patients with different underlying pain mechanisms; 2) prediction of therapeutic outcomes; and 3) quantification of therapeutic intervention results (62). In headache clinics, QST is not frequently used, but has been employed in scientific studies of migraine, CT, TTH, CH and TN (63–67) (Table 2(a) and (b)).

OPEN IN VIEWER

Table 2A.

Quantitative sensory testing migraine patients.

First author	Participants	Mean age (range)	Timing of recordings	Methods	Main findings
Burstein (2000) (63)	42 migraine patients	40 (25–67)	Ictally and interictally (>7 days)	Pain thresholds to cold, warm, and mechanical stimuli periorbital and forearm bilateral	During attacks 79% reported cutaneous allodynia, which occurred within the referred pain area on the ipsilateral head, or within and outside the ipsilateral head.
Zappaterra (2011) (65)	21 migraine patients 22 eTTH 11 cTTH 26 CTRLs 44 MOH	38 37 47 40 50	Interictally	Pain threshold to mechanical stimuli to temple, cheekbone and cervical areas bilateral	Pain thresholds are lower in all headache patient groups compared to CTRLs in temple and cheekbone areas. MOH patients also had lower pain thresholds in cervical areas and lower pain thresholds than any other group.
Burstein (2004) (68)	31 migraine patients	36 (17–55)	Interictally, within 1 h and 4 h of a migraine attack	Pain thresholds to cold, heat and mechanical stimuli at the periorbital site of the referred pain Pain thresholds to cold, heat and mechanical stimuli at the periorbital site of the referred pain	15% of allodynic vs. 93% of nonallodynic attacks were pain free within 2 h of triptan treatment.
Schoenen (2008) (69)	90 migraine patients	38	Ictally	Patients used a 2-cm brush to assess cutaneous allodynia in five zones on each side of the head	34% reported allodynia during attacks. Allodynia was not a significant explanatory variable for predicting treatment efficacy.
Weissman-Fogel (2003) (70)	34 migraine patients 28 CTRLs	26 25	Interictally	Mechanical, electrical and thermal pain thresholds and supra-threshold stimuli of the periorbital area and forearm skin. Temporal summation of pain induced by noxious tonic mechanical, electrical and heat pain	No differences between groups for pain thresholds to heat or electrical stimulation or pain scores for supra-threshold phasic stimuli. Migraine patients show a lower pain threshold for mechanical stimulation and higher pain scores on the temporal summation tests.
Schwedt (2011) (71)	20 EM 20 CM 20 CTRLs	32 (18–52) 31 (19–47) 36 (23–62)	Interictally >48 h	Heat, cold, and mechanical pain thresholds and heat and cold pain tolerance thresholds at forehead and forearm bilateral	EM more sensitive to heat and cold stimuli than CTRLs. CM more sensitive to heat pain, heat tolerance and cold tolerance (head region only) than CTRLs. No differences between CM and EM.
Kitaj (2005) (72)	40 EM 41 CM	42 (20–62) 41 (27–79)	Ictally and interictally	Heat, cold and mechanical pain threshold in four locations (bilateral ophthalmic, maxillary, C4/posterior neck, and forearm) bilateral	CM patients had lower pain thresholds than EM patients, both ictally and interictally.
Schwedt (2014) (73)	112 migraine patients 75 CTRLs	34 37	Interictally ≥48 h	Heat pain thresholds at forehead and forearm bilateral	Migraineurs had lower pain thresholds than CTRLs at all areas. There were positive correlations between time to next headache and pain thresholds.

OPEN IN VIEWER

Table 2B.

Quantitative sensory testing other primary headaches and trigeminal neuralgia.

First author	Participants	Mean age (range)	Timing of recordings	Methods	Main findings
Munksgaard (2013) (74)	21 MOH 40 CTRLs	40 40	Interictally ≥48 h	Cephalic and extracephalic pressure-pain thresholds and supra-threshold pressure-pain scores and extracephalic pain thresholds, supra-threshold pain scores and temporal summation for electrical stimulation	In MOH patients cephalic and extracephalic pressure-pain thresholds were lower than in controls. In supra-threshold pressure pain scores and extracephalic supra-threshold pain scores for electrical stimulation were lower in MOH. Cephalic supra-threshold pain scores decreased from baseline to 12-month follow-up.
Ellrich (2006) (66)	25 CCH and ECH 60 CTRLs	25 35	NS	Thermal and mechanical sensory functions in the periorbital region	Perception of warmth and cold and pressure pain were reduced on the cluster side as compared with the contralateral asymptomatic side. All patients showed reduced on the headache side in comparison to CTRLs.
Sinay (2003) (67)	9 TN 10 CTRLs	58 57	During painful periods outside attacks	Thermal sensory and pain thresholds	For all six trigeminal branches, warm sensation and heat pain thresholds were increased and cold sensation and cold pain thresholds were decreased in comparison to controls (except cold sensation on the contralateral side).
Maier (2010) (4)	94 TN 180 CTRLs	59 NS		Standardized QST	In 87% of TN the detection thresholds were abnormal in the pain area when compared to the reference range of healthy controls or the contralateral side. Patients with abnormal signs were distributed with decreased thresholds (30%), increased thresholds (19%) or a combination of these (38%).

MOH: medication-overuse headache; CTRLs: healthy controls; CCH: chronic cluster headache; ECH: episodic cluster headache; TN: trigeminal neuralgia; QST: quantitative sensory testing; NS: not stated; h: hours.

The pivotal study by Burstein et al. (63) employed repeated measurements of mechanical (via calibrated von Frey hairs) and thermal (via a thermode) pain thresholds of periorbital and forearm skin outside and during migraine attacks. In 79% of the patients, migraine attacks were associated with cutaneous allodynia occurring either within the referred pain area on the ipsilateral head, or within and outside the ipsilateral head (63). The authors suggested that non-allodynic patients have peripheral sensitization, whereas allodynic patients have central sensitization of second-order brainstem trigeminal neurons as well. Later, the Burstein group (68) examined the effect of sumatriptan (subcutaneous injection), rizatriptan and zolmitriptan in patients with or without allodynia, using the same methods. This was performed since it is well established in several clinical trials that taking triptan early, when pain is mild, is more effective than waiting until the pain is moderate or severe (64). The study showed that within 2 h of triptan treatment, patients were pain free in 15% of allodynic attacks vs 93% of nonallodynic attacks (68). Treating migraine attacks 1 h or 4 h after the onset of pain was equally ineffective in inducing a pain-free state in the presence of allodynia, and equally effective in the absence of allodynia. The finding was later challenged in a multicenter, double-blind pilot study (69) that evaluated the prevalence of self-reported brush allodynia and its relative influence on the efficacy of almotriptan 12.5 mg combined with either aceclofenac 100 mg or placebo in 112 migraine patients. The study showed that allodynia numerically reduced treatment success overall, but this effect was not significant for the primary outcome measure, and multivariate logistic regression analysis confirmed that headache intensity at treatment intake, rather than allodynia, significantly influenced most outcome measures. Furthermore, another study showed that migraine headache intensity can be reduced within 1 h by zolmitriptan nasal spray in spite of the presence of early cutaneous self-reported allodynia (70). And in another study (71) allodynia (based on a questionnaire) did not influence the efficacy of almotriptan.

Cutaneous pain thresholds (CPTs) in trigeminal innervated areas are lower in migraine, eTTH, cTTH and MOH in comparison to healthy controls (65,72). Migraine patients interictally had lower pain thresholds for mechanical stimulation, higher pain scores in a temporal summation test and increased summation of pain for repeated mechanical and electrical stimuli compared to controls (72). Schwedt et al. (73) tested thermal and mechanical pain thresholds in 20 EM, 20 CM and 20 healthy controls. Thermal pain thresholds and thermal pain tolerance were lower in migraine patients than in healthy controls, but there was no difference between EM and chronic migraine. This is in contrast to a study investigating EM and chronic migraine using QST in eight locations interictally (bilateral ophthalmic, maxillary, C4/posterior neck, and forearm) that revealed that CM patients had lower pain thresholds than EM patients on the neck, arm and maxillary region (74). Interestingly, it has been demonstrated that there is a positive correlation between the time to next headache and low heat pain threshold at the head and arm (75). Nociceptive threshold decreases in healthy individuals following infusion of nitroglycerine, a donor of the signaling molecule nitric oxide (NO) (76), which is known to induce migraine in migraine patients (77) and is likely involved in migraine pathophysiology. This may reflect central facilitation of nociception by NO in migraine, but warrants further studies.

A recent study showed that MOH patients had lower cephalic/extracephalic pressure-pain thresholds, higher cephalic supra-threshold pressure-pain scores than controls and higher extracephalic supra-threshold pain scores for electrical stimulation than controls (78). Twelve months after successful detoxification, cephalic supra-threshold pain scores decreased in MO patients. Furthermore, temporal summation was not found in MOH patients before withdrawal, in contrast to controls, but after detoxification temporal summation normalized (78). Interestingly, MOH patients had significantly lower CPTs in the temple and cheekbone areas in comparison to the other headache groups (65), which suggests pronounced central sensitization in MOH patients.

In a mixed group of 25 CCH and ECH patients during active cluster periods and during remission, thermal and mechanical sensory function in the periorbital region was compared to 60 healthy volunteers (66). Perception of heat and pressure pain were reduced on the cluster side compared to the contralateral asymptomatic side and compared to healthy controls (66).

In TN, nine TN patients without prior surgery were compared to healthy controls in all six trigeminal branches (67). For all six trigeminal branches, warm sensation and heat pain thresholds were increased and cold sensation and cold pain thresholds were decreased in comparison to controls (except cold sensation on the contralateral side) (67). As part of a huge study on different neuropathic pain conditions, Maier et al. (4) performed standardized QST on 94 TN patients. In 87% of TN patients, the detection thresholds were abnormal in the pain area, when compared to the reference range of healthy controls or the contralateral side. Patients with abnormal signs were distributed with decreased thresholds (30%), increased thresholds (19%) or a combination of these (38%).

Other approaches to sensory testing in headache patients are discomfort thresholds, which have been tested for visual (79,80) and auditory (80,81) stimuli. Thus, quantitative measurement of sound-induced discomfort showed that migraineurs were significantly more sensitive than controls both during and outside an attack, and there were no differences between MA and MO. Using quantitative thresholds for discomfort and pain with monocular and binocular light stimuli, migraine patients were more photophobic during attack than outside attack, and they were more sensitive to light than controls between attacks (79). Again, no differences in light sensitivity between MA and MO were shown. CH patients have also been shown to be more sensitive to light and sound-induced discomfort than controls, but only during cluster periods (80).

In summary, allodynia is a feature of the migraine attack, but it is also found in other primary and secondary headaches and there is no apparent difference between chronic and EH subtypes. The studies differ greatly in methodology, and even differ in the definition of allodynia, which makes it difficult to conclude. QST can detect subclinical sensory abnormalities in TN patients despite absence of clinically evident neurological deficits (1).

Headache provocation models

A powerful method to subclassify primary headaches is the use of pharmacological substances to induce headache (Table 3). In the most frequently used human headache model, patients are randomly allocated to receive intravenous (IV) infusion of the provoking substance or placebo (isotonic saline) in a double-blind, crossover design (82).

OPEN IN VIEWER

Table 3.

Provocation studies migraine.

Compound		Dose	Migraine-like attacks	Aura	References
Glyceryl trinitrate	MA	Intravenous 0.5 µg/kg/min	50%–67%	0–10%	(83,84)
		Sublingual 0.9 mg	41%	14%	(85)
	MO	Intravenous 0.5 µg/kg/min	80%–83%	0%	(83,86)
		Sublingual 0.9 mg	82%	0%	(85)
Hemiplegic migraine	Intravenous 0.5 µg/kg/min	13%–30%	0–18%	(86–88)
Sildenafil		100 mg per os	83%	0%	(89)
Histamine		Intravenous 0.5 µg/kg/min	70%	0%	(90)
CGRP	MA	Intravenous 1.5 µg/min	57%	28%	(91)
	MO	Intravenous 2 µg/min	50%	0%	(92)
Hemiplegic migraine	Intravenous 1.5 µg/min	9%–22%	0%	(93,94)
Dipyridamole		Intravenous 0.142 mg/kg/min	50%	0%	(95)
Vasoactive intestinal peptide		Intravenous 8 pmol/kg/min	0%	0%	(96)
PACAP38		Intravenous 10 pmol/kg/min	66%–73%	0%	(97,98)
Ciloztasol		200 mg	86%	0%	(99)
Prostaglandin I2		Intravenous 10 ng/kg/min	50%	0%	(100)

Glyceryl trinitrate (GTN)-induced migraine-like attacks in 80% of migraine sufferers but not in healthy participants (74). Migraineurs developed a delayed headache fulfilling criteria for induced migraine attacks peaking 5 h after end of the infusion compared to 10% after placebo (79). Other groups validated the model and found it to be reproducible and reliable (83,101). Using GTN provocation on two separate days, the maximal headache response differed in less than 50% and only one of 10 headache scores in 40% of healthy individuals (102). Furthermore, headache characteristics were also reproducible (102) and the ability of GTN to induce migraine did not vary according to the frequency of attacks in migraine patients (103). GTN was given sublingually (0.9 mg) to 168 MO patients, 22 MA patients and 57 healthy controls. The test sensitivity was 82% in MO, specificity 96% and accuracy 86% (82). In MA sensitivity was 14%, specificity 96% and accuracy 72% (101). However, with IV infusion 50%–67% of MA patients develop an MO attack (83,85). In FHM types 1 and 2, two very rare dominantly inherited subtypes of MA, GTN did not induce migraine attacks (86,87). Thus, GTN induced only delayed MO attacks in some migraine subtypes, which gives important information about migraine pathophysiology, but may also be useful as a diagnostic tool when investigating different migraine subtypes.

GTN has not been used to differentiate between TTH subtypes, but it caused both immediate and delayed headache (104). The delayed type resembled the TTH phenotype (104). Measurements of muscle hardness, myofascial tenderness and pain thresholds during the immediate type headache were not different after GTN compared to placebo (104), suggesting that there was no peripheral or central sensitization during the immediate headache. Unfortunately, no measurements were taken during delayed headache.

CH attacks can be triggered during a cluster period using GTN and do not differ from spontaneous attacks (101,105,106). GTN sublingually (0.9 mg) during the active phase had a sensitivity of 80.6%, a specificity of 100% and an accuracy of 93%. If the CH group also included patients in remission phase, the sensitivity dropped drastically to 61.9% (101), Thus, GTN induces attacks predominantly during an active cluster period.

Calcitonin gene-related peptide (CGRP) is a potent vasodilator (107) in cerebral and extracerebral arteries (107). It can be released upon stimulation of the trigeminal ganglion (108). Lassen et al. (92) infused CGRP or placebo for 20 min in 12 MO patients in a double-blind crossover study. All patients experienced headache and three patients fulfilled ICHD-2 criteria for MO. Using newer criteria for migraine-like headache in experimental models (94), 50% experienced migraine-like attacks after CGRP compared to only one after placebo. The effect of CGRP in MA was later explored by Hansen et al. (91). CGRP induced MO attacks in 57% of MA patients (91). Just like GTN, CGRP does not induce FHM attacks (93,94).

Pituitary adenylate cyclase activating polypeptide-38 (PACAP38) is a vasodilator distributed in human sensory (109) and parasympathetic nerves and ganglia (110). In MO patients PACAP38 infusion induced migraine attacks in 58%–73% (97,98). Other signaling molecules and pharmacological substances well known to induce migraine in migraine patients are prostaglandin I₂ (111), prostaglandin E₂ (110), histamine (90), sildenafil (89) and cilostazol (101).

In summary, various pharmacological substances are able to induce primary headaches. It would be very interesting to investigate if these models can predict therapeutic outcomes and if they can be used to measure therapeutic responses. At the Danish Headache Center, we have a long tradition of conducting these infusion tests, which are done relatively easy with trained and skilled personnel, but we acknowledge that other countries may have challenges in performing these tests in terms of internal review board acceptance, approval of drugs for infusion and patient compensation.

Intracranial pressure (ICP)

In relation to secondary headache associated with cerebrospinal fluid (CSF) alterations, there are magnetic resonance imaging (MRI) signs associated with either idiopathic intracranial hypertension (IIH) or headache attributed to spontaneous low CSF pressure. However, in some case series 10% of IIH patients may have a normal MRI (112). A case series of 10 patients with IIH without papilledema (IIHWOP) investigated continuous CSF pressure monitoring (Pcsf) (113). All patients had normal resting Pcsf defined as ICP < 15 mmHg, but during sleep, all patients had B-waves and 90% had plateau waves or near plateau waves. Increased Pcsf was seen mostly during sleep and was intermittent, suggesting that Pcsf elevation may be missed by a single spot-check low pressure measurement. The similarity between IIHWOP and chronic daily headache (CDH) suggests that continuous Pcsf monitoring in CDH patients may have an important diagnostic role that should be further investigated.

In a case series of 40 patients with orthostatic headache and diffuse gadolinium pachymeningeal enhancement, 18% had opening pressures that were within normal limits (7–14 cm H₂0) (114). It has been suggested that a head-down tilt could be an accurate screening test for low CSF pressure in patients with daily headache (115). This would be an easy, inexpensive, noninvasive method that could be used before lumbar puncture with pressure measurement, radioisotope cisternography, CT or MR myelography or dynamic CT. A decrease in headache intensity during a 20-degree head-down tilt was seen in patients who had low CSF pressure (115).

Hannerz et al. investigated patients initially diagnosed with cTTH in three separate studies and found that 41%–68% had opening lumbar pressure of 20–25 cm H₂O (116–118), which could point to headache attributes to IIH in non-obese patients according to the ICHD-2 criteria, but not the latest ICHD-3 criteria. CTTH patients in general had an increase in headache intensity following a 15-degree head-down tilt (116–118). In a recent study a surprisingly large number (43%) of patients diagnosed with medically refractory chronic migraine with normal MRI and no papilledema showed an opening pressure above 25 cm H₂O (119). In 85 CM patients, 12% had increased ICP without papilledema, transient visual obscurations or visual fields (120) and in another study 61 CM patients, 5% had increased ICP without papilledema (121). Thus, some patients with chronic migraine may actually have unrecognized headache attributed to low or high CSF pressure. Furthermore, some patients with chronic headache may have normal fundoscopy but nevertheless an increased ICP (119).

In conclusion, measurement of ICP is very important in patients suspected of having IIH or low-pressure headache. A reasonable body of evidence suggests that it may also be worthwhile in patients with chronic headache, as some have increased pressure without papilledema. Monitoring pressure continuously improves such measurements, but this needs further validation.

Autonomic tests in primary headaches

In a German population-based sample an epidemiological study using questionnaires showed that 27% of migraine patients have accompanying unilateral autonomic symptoms (130). Several studies have shown dysfunction (both hypo- and hyperfunction) of the autonomic system in migraine patients, but no clear autonomic deficits have emerged from these studies (131). It would be of interest to compare migraine patients with or without autonomic symptoms using autonomic tests to elucidate whether they have different phenotypes, but, to our knowledge, this has never been performed.

The pupil dilates by sympathetic activity and contracts by parasympathetic activity. Miosis and ptosis during an attack indicate dysfunction of the sympathic innvervation, Horner’s syndrome. Seven out of 11 CH patients showed ocular signs of a postganglionic sympathetic lesion following an irritation test to the eye (120). Following corneal stimulation, ECH patients show reduced mydriasis bilaterally, but most pronounced on the pain side during but not outside the cluster period (132). Even with pain stimulation of extracephalic structures such as the sural nerve, pupil dilation was attenuated on the pain side during cluster periods but not during remission or in CCH (133). The cold pressor test (CPT) during and outside a bout showed absent miosis (134). Naloxone, an opioid-receptor antagonist, blocked miosis response in healthy controls but induced miosis in CH patients (134). This points to an oculocephalic sympathetic hypofunction in CH patients as well as an altered opioid neuromodulation (134). Thirty-eight ECH patients had significantly reduced average pupillary constriction velocity on the CH pain side compared to healthy controls. CH patients displayed bilaterally reduced pupillary average velocities, which indicates a bilaterally reduced cranial parasympathetic tone in CH patients in remission phase, with significant lateralization to the CH pain side (135).

Sweating can be measured by evaporimetry, by which water loss from the body surface can be measured rapidly and accurately (136). Thermoregulatory flushing and sweating are impaired on the pain side in CH patients (137), despite the fact that ipsilateral forehead sweating occurs during CH attacks (138). CH patients are also hypersensitive to acetylcholine analog pilocarpin on the ipsilateral side (137). Whether these findings may be useful in subcategorizing CH patients is not known.

Changes in heat loss can be measured by thermovision cameras as a marker of either sympathetic (cutaneous vasoconstriction, heat gain) or parasympathetic (cutaneous vasodilation, heat loss) changes. In 1984 thermography was recorded during 209 headache attacks both in migraine and TTH patients (135). During unilateral headache attacks, heat loss was found in the frontotemporal region on the pain side in patients with pain relief following superficial temporal artery (STA) compression. This thermographic asymmetry was abolished after pressure was applied over the STA. Throbbing headache was also associated with unilateral heat loss. Interestingly, 74% of patients with unilateral heat loss had headache response to ergotamine compared to 44% of patients without heat loss (139). No differences were found between migraine and TTH patients in terms of heat loss. Using thermography, it has been proposed that migraine patients have an area in the face with less heat loss on the pain side, which is also found interictally (140). This cold patch was later suggested to be a prognostic index marker in MA and CH patients as the cold patch was shown to be attenuated following successful treatment (141). However, a following validation study failed to show a similar pattern (142). In CH an ipsilateral heat loss in the orbital region in 63% of CH patients during attacks (143) was shown. Asymmetry in thermography seems to be a relatively sensitive but not a very headache phenotype-specific finding. Thus, in 1000 thermograms of headache patients, 85% of MO patients, 89% of MA patients, 86% of CH patients and 85% of post-traumatic headache patients had abnormal thermograms (defined as an arbitrary thermal difference of 0.5℃ between homologous regions of the face) compared to 13% of healthy controls (144). Based on the above, thermography does not seem to be of benefit in headache diagnostics, as the results so far have been unspecific.