• Open Access

Sensory Nerve Conduction and Somatosensory Evoked Potentials of the Trigeminal Nerve in Horses with Idiopathic Headshaking




Idiopathic headshaking (HSK) in horses is a distressing disorder in which the etiology and pathophysiology are unknown.


Differences in sensory function of the trigeminal nerve exist between healthy and affected horses.


Six healthy mature geldings and 6 mature geldings with idiopathic HSK.


Prospective study. Sensory nerve action and somatosensory evoked potentials studies were performed. The stimulus site comprised the gingival mucosa dorsal to the maxillary canine. A pair of recording electrodes was placed along the sensory pathway of the trigeminal complex at the infraorbital nerve (R1), maxillary nerve (R2), spinal tract of trigeminal (R3), and somatosensory cortex (R4). Sensory nerve action potential latency (ms), amplitude (μV), duration (ms), area under the curve (μVms), and conduction velocity (m/s) were calculated.


Threshold for activation of the infraorbital branch of the trigeminal nerve was significantly different between 5 affected (≤5 mA) and 6 control horses (≥10 mA). After initiation of an action potential, there were no differences in all parameters measured and no differences between left and right sides. A horse with seasonal HSK tested during a time of no clinical manifestations showed a threshold for activation similar to control horses.

Conclusions and Clinical Importance

This study confirms involvement of the trigeminal nerve hyperexcitability in the pathophysiology of disease. Further, results might support a functional rather than a structural alteration in the sensory pathway of the trigeminal complex that can be seasonal. The horse could serve as a natural animal model for humans with idiopathic trigeminal neuralgia.


area under the curve


conduction velocity




somatosensory evoked potential


sensory nerve action potential


sensory nerve conduction


sensory nerve CV

Idiopathic headshaking (HSK) is a spontaneously occurring disorder of mature horses affecting mainly geldings of Thoroughbred, Warmblood, Quarter horse and related breeds.[1-3] The disorder is characterized by shaking, tossing, or jerking movements of the head, muzzle rubbing, snorting, sneezing, striking at the face with the thoracic limbs, and having an anxious facial expression.[1, 2] Potential causes for the observed behavior such as oral/dental, temporomandibular joint, ocular, ear, or upper airway disease and temporohyoid osteoarthropathy must be ruled out.[2, 4] Absence of pathology and response to infraorbital nerve blocks have been advocated as diagnostic for the disorder.[5] These manifestations have been reported to be seasonal in 59% of cases with the majority occurring in the Spring and early Summer.[1] The disorder could impair performance and even daily life activities such as eating if severe. Treatment failure, self-inflicted trauma, compromised quality of life, and safety concerns for both the horse and those caring for the horse often lead to euthanasia of the animal.[6]

Clinical manifestations of affected horses have been compared to those of people with idiopathic trigeminal neuralgia who experience burning, itching, tingling, tickling, or electric-like pain sensation. This has led to the belief that the disorder in horses is also associated with trigeminal neuralgia.[7-10] Despite this belief, there have been limited studies to investigate the involvement of the trigeminal nerve in the etiopathogenesis of disease including infraorbital nerve blocks and compression.[5, 7, 10-13] Structural abnormalities within the nervous system have not been fully investigated or demonstrated in affected horses.[7] Environmental and hormonal factors have been considered in the development of the disorder.[1, 14] However, a specific cause has not been identified.[15]

Somatosensory evoked potentials (SEP) permit functional assessment of sensory conduction from the site of stimulus, along the sensory pathway (sensory nerve, dorsal root ganglion, and somatosensory spinal cord) to the site of cortical perception.[16] As the involvement of the trigeminal nerve in the pathogenesis of the disorder has not been fully investigated,[13] the purpose of this study was to describe a novel technique in the equine species for evaluation of the sensory function of the trigeminal complex (nerve and somatosensory) and to investigate whether differences exist in trigeminal nerve stimulation threshold and conduction velocity (CV) between healthy horses and horses with idiopathic HSK. We hypothesize that differences in sensory function of the trigeminal nerve exist between healthy and affected horses.

Materials and Methods


Control Group

Six mature healthy geldings were included in this group based on a normal physical and neurologic examination and with no history of HSK episodes in the previous years of the study. The criteria for inclusion for age were based on the reported mean age of 9 years (range from 1 to 30 years old) in affected horses.[1] Geldings were selected as the disorder is more common in geldings.[1, 5] The breeds were selected based on those reported to suffer from idiopathic HSK. These horses were available from this research herd. Control horses were used to validate the test and establish normal values for SEP resulting from stimulation of the maxillary branch of the trigeminal nerve for comparison with horses with idiopathic HSK.

Headshaker Group

This group was comprised of 6 horses with a clinical diagnosis of idiopathic HSK based on the stereotypic behavior for which other disorders had been ruled out by the authors. Typical behavior consisted of sudden, violent flicks of the head, nose rubbing, excessive snorting, and apparent anxiety.[1] These horses were donated to our institution by owners because of intractable HSK, compromised quality of life, and for safety concerns. Owners responded to specific questions with regard to the type of HSK (vertical, horizontal, mild, violent movements, sudden) and associated signs (e.g, snorting, rubbing nose, trotting with head low in the ground, apparent anxiety), frequency, severity, seasonality, environmental factors, and physical activity. Written consent of donation was obtained from the owners, and the study was approved by the University of California, Davis Institutional Animal Care and Use Committee.

Anesthesia Protocol

Identical protocols were employed for both groups of horses. Following intravenous catheterization and premedication with 0.5 mg/kg xylazine,1 anesthesia was induced using 5 mg/kg of thiopental2 IV prepared as a 10% solution in a 90 mM aqueous sodium hydroxide solution. Anesthesia was maintained with 1.5% end-tidal isoflurane in oxygen and horses were mechanically ventilated. Lactated Ringer's solution was administered at a rate of 10 mL/kg/h IV throughout the study. Blood pressure was monitored using a dorsal metatarsal arterial catheter, and dobutamine was administered IV as needed to maintain a mean arterial pressure ≥70 mmHg. Heart rate and rectal temperature were monitored. External forced air warmers, blankets, and a heated water bed were used to avoid hypothermia. Electrophysiology recordings were started 1 hour after anesthetic induction to allow time for sedation and induction agent washout and establishment of steady-state conditions.


Sensory Nerve Conduction and SEP of the Maxillary Nerve

The nerve under study was selected based on signs of apparent facial discomfort, mainly of the muzzle area, displayed by affected horses. The maxillary nerve is sensory to the lower eyelid, maxillary teeth, upper lip, maxillary sinus, and nose.[17] The maxillary canine area is innervated by the infraorbital nerve (a branch of the maxillary nerve) and was selected as the stimulus site. An evoked potential system (Nicolet Viking IV3 ) was used for all sensory nerve conduction (SNC) and SEP studies. The stimulus rate was set at 3 Hz with a stimulus duration of 0.2 ms for all recordings. The sweep speed was set at 5 ms per division. Each recording was the average of 1,000 responses. Recordings were made at each of the following stimulus intensities (±1 mA): 2.5, 5, 10, 15, 20, and 25 mA. The acquisition sensitivity and band width varied for each recording area. Because of the expected short latency from the stimulus artifact to the onset of the peripheral response, a stimulus delay of 1 ms was used to ensure that the entire sensory nerve action potential (SNAP) was visible. Two sets of recordings were performed at each stimulus intensity to test the reproducibility of the data. As a negative control (to assess the level of background interference), additional recordings were performed on all horses with the stimulus intensity set at 0 mA.

The stimulus and recording sites were as follows (Fig 1). A surface stimulating unit (Nicolet S4033) with a pair of curved tips (cathode, anode) was placed on the dorsal gingival mucosa straddling the maxillary canine tooth. Three pairs of recording electrodes (Disposable EasyGrip monopolar electrode3) were placed along the sensory pathway of the trigeminal complex. These electrodes were teflon coated except for 5 mm at the distal tip. Recording site 1 (R1: infraorbital nerve) was located at the level of the infraorbital foramen with the exploring (active) electrode (27 G by 25 mm monopolar) placed SC above the nerve, and the reference electrode (27 G by 25 mm monopolar) placed 2.5 cm dorsally. Recording site 2 (R2: maxillary nerve) was located at the level of the maxillary foramen. The exploring electrode (25 G by 75 mm monopolar) was inserted so that the tip was located near the maxillary foramen and the reference electrode (27 G by 25 mm) was placed SC. Recording site 3 (R3: C1) was at the level of the 1st spinal cord segment on midline. The exploring electrode (26 G by 75 mm) was placed with the tip on the C1 vertebral lamina at midline, and the reference electrode (27 G by 25 mm) was placed SC near the contralateral (opposite to the side used for stimulation) wing of the atlas. A 4th recording site (R4: cortical) was located at the level of the frontoparietal cerebral cortex. The exploring electrode (subdermal needle electrode4 ) was placed SC at the intercanthus. The reference electrode (subdermal needle electrode4) was placed at the contralateral frontoparietal cortex (C3 or C4 by EEG designation).[18] This electrode placement was chosen for R4 to display positive events as upward deflections as is typical for cortical recordings (cortical SEP).[16, 19] All 4 tracings were recorded simultaneously, and displayed with negative up for R1, R2, and R3. An additional subdermal needle electrode4 was used as a ground and placed between the stimulating and 1st recording site. Surface temperature was maintained at a constant range between 35 and 37°C during recording.[20-22] A surface temperature probe was placed dorsal to the facial crest to monitor the regional temperature throughout the study period. SNC and SEP studies were performed bilaterally in 6 horses to determine whether differences between sides existed.

Figure 1.

Needle electrode placement for sensory conduction study. (A) Skull depicting trigeminal complex and needle electrode placement: stimulating (S) and recording (R1, R2, R4) sites, R3 not shown. Diagram not shown at scale. (B) Stimulating tip placed dorsal to maxillary canine, (C) R1 at level of infraorbital foramen, (D) R2 at level of maxillary foramen, (E) R3 at level of C1 spinal cord segment, and (F) R4 at frontoparietal cerebral cortex. Stim, stimulus unit; G, ground; Exp, active or exploring electrode; Ref, reference electrode.

Acquisition sensitivities were set at 200, 100, 10, and 10 μV for recording sites 1, 2, 3, and 4, respectively. However, these sensitivities were variable depending on the amplitude of the signal recorded for each horse. Band widths for SNC and spinal SEP were 20 Hz to 3 KHz, and for cortical SEP were 20–250 Hz. The parameters analyzed included latency (ms), amplitude (μV), duration (ms), area under the curve (AUC, μVms), and (CV, m/s). By convention for SNC studies, the latency was measured at the onset of the negative (upward) deflection.[16, 19] The amplitude was measured from positive peak to negative peak for SNAP.[16, 19] Duration of SNAP was estimated from onset to return to baseline. The AUC was automatically calculated by computerized software (Nicolet Viking IV3). Using a fiberglass tape measure,5 the distance between the stimulus site and the R1 exploring electrode was measured from the cathode to the hub of R1, with the addition of 25 mm (length of the needle) for the distal CV calculation. Distances between R1 and R2 were determined by measuring from the hub of R1 to the insertion of R2 and subtracting 25 mm. Although not an accurate representation, surface distances between the insertions of R2 and R3 were used for the most proximal trigeminal CV.

Postmortem Evaluation

Horses were euthanized at the end of the study while under anesthesia with pentobarbital sodium 100 mg/kg IV. A full body routine necropsy was performed to confirm the absence of other possible causes of HSK. In addition, one of the authors (MA) removed the infraorbital nerve from the infraorbital canal after hand sawing the maxillary bone for histological evaluation (hematoxylin eosin and modified Gomori trichrome).

Statistical Analysis

Left- and right-sided recordings were compared by means of a paired t-test. Response versus no response at various stimuli (2.5, 5, 10, 15, 20, and 25 mA) was investigated to study threshold for response and compared between groups with a Fischer exact test. CV within groups was evaluated by repeated measurements ANOVA followed by the Bonferroni test post hoc. For this comparison within groups, nonresponders were assigned a value of “0.” For all parameters under study (latency, amplitude, duration, AUC, and CV), a t-test was used for comparison between groups at each recording site. For R4, the presence or absence of a signal was recorded. Latencies were measured for R4 on high stimulus intensities (20, 25 mA). Statistical significance was set at P value <.05 for all tests.



The study for both groups of horses was performed depending upon availability of animals.

Control Group

Six healthy adult geldings were included in this group. This group comprised horses of Thoroughbred (n = 3), Warmblood (n = 2), and Quarter horse (n = 1) breeds. The following information is reported as mean (median, range): age in years 10.6 (11.5, 5–14), body condition score (BCS) of 5–6 of 9 of published scoring system,[23] and body weight of 570.2 kg (520–621). The study was performed during the winter (n = 3 horses) and summer (n = 3) months.

Headshaker Group

Six adult geldings with idiopathic HSK of Quarter horse (n = 4), Thoroughbred (n = 1), and Sport horse cross (n = 1) breeds were included in this group. Mean age in years 11.8 (11.5, 10–15), BCS of 5–6 of 9, and body weight of 502.8 kg (362–610). Five of these horses displayed HSK at the time of the study. Four of these were studied during the summer, and 1 during the winter. The 6th horse, a seasonal HSK examined by one of the authors (JEM) the previous year while HSK, was not manifesting clinical signs at the time of the study (during winter).


Arterial blood pressures and carbon dioxide were all maintained within study parameters, and no horse exhibited hypoxemia or abnormal electrolyte values. Nerve stimulation caused a statistically significant 4 ± 3 (mean ± SD) beat/min heart rate reduction in headshaker horses; a smaller 1 ± 2 beat/min change occurred in normal control horses and was not statistically significant. Heart rate returned to baseline upon discontinuation of nerve stimulation.


From nonpublished cadaver studies by the authors and review of the literature,[17] the course of the infraorbital to maxillary nerve to the rest of the trigeminal sensory complex was determined. As the technique employed has not been described in horses, the stimulus intensity needed to elicit a response was unknown. Attempts to trigger SNAPs at low stimuli (2.5 mA) in 2 control horses failed. Sequential increases in stimulus intensity did not trigger SNAP until 20 mA were applied. SNAPs were observed, recorded, and values calculated for sites R1–R3. A regular muzzle twitch was observed upon stimulation at 20 and 25 mA. Gradual decreases to 15, 10, 5, and 2.5 mA were performed but neither SNAPs nor muzzle twitch were seen. An increase to 20 mA elicited SNAPs at all recording sites. The study was repeated at all stimulus intensities in both horses and left and right sides were evaluated.

The 1st horse with HSK became available and the study was performed starting at 2.5 mA to evaluate if a SNAP would be triggered. Action potentials were observed at all recording sites at 2.5 mA. Gradual increases of stimulus intensity resulted in SNAPs at all stimulus levels (2.5, 5, 10, 15, 20, and 25 mA). Regular muzzle twitch was observed at all stimulus intensities. Recordings were not obtained at stimulus intensities lower than 2.5 mA. The study was repeated and right and left sides evaluated. Subsequently, more horses were available and comparisons between left and right sides were investigated in control (n = 2/6) and HSK (n = 4/6) horses. There were no statistical differences in the parameters recorded between left and right sides.

The 6 remaining horses were initially stimulated at 25 mA and the stimulus intensity was gradually decreased down to 2.5 mA. The study was repeated and parameters recorded. There was statistical difference in response (threshold) versus no response at low stimuli (2.5 mA [P = .03], 5 mA [P = .008]) between groups of horses, but not at higher stimuli (10, 15, 20, 25 mA). Upon excluding the HSK horse who was not displaying signs at the time of the study from this comparison, the difference in threshold remained significant (2.5 mA [P = .015], 5 mA [P = .002]). All HSK horses displaying the behavior at the time of study (n = 5/6) had threshold onset of SNAPs at 2.5 (n = 4 horses) and 5 (n = 1) mA. The HSK horse not manifesting signs at the time of the study had recordable SNAPs at 10 mA, but not at lower intensities. Control horses had a stimulus threshold recordable at 10 (n = 3 horses), 15 (n = 1), and 20 (n = 2) mA. Recordings were obtained for all horses at stimulus intensity of 20 and 25 mA. All horses did not have visible SNAPs or muzzle twitch when the stimulus was turned off. See Figure 2 for representative SNAPs and SEPs at various stimulus intensities.

Figure 2.

Sensory action potentials at various stimulus intensities. Headshakers (A–E), control (F). (A) Stimulus intensity (SI) off, (B) SI 2.5 mA, (C) SI 5 mA, (D) SI 10 mA in headshaking, (E) SI 20 mA, (F) SI 25 mA. Divisions: 5 ms per horizontal division, variable amplitude (μV) per vertical division as indicated under each region, N, average of 1,000 responses.

Once triggered, SNAP latency, amplitude, duration, AUC, and CV were not significantly different between groups at 10, 15, 20, and 25 mA stimulus intensities (see Table 1, data only shown for latency, amplitude, and CV; 25 mA data not shown) for each recording area. Latencies for R4 at 20 and 25 mA were not statistically different (controls 15.85 ± 0.44 [mean ± SE], HSK 15.84 ± 0.66). Local superficial recording temperature was maintained at 35°C for all horses. There was no statistical difference in nerve CV at the different stimulus intensities within the HSK group. There was a significant difference within the control group in CV between 2.5 and 5 mA (value of “0” for nonresponders) compared to 25 mA. At higher stimulus intensities, there were no differences in CV within the control group.

Table 1. Sensory nerve conduction of infraorbital/maxillary nerve. Means, SE, and range are provided for latency, amplitude, and conduction velocity.
Stim (mA)SiteControlsHSK
Mean (SE)Median (Range)Mean (SE)Median (Range)P-Value
  1. HSK, headshaking group; SNCV, sensory nerve conduction velocity; Stim, stimulus intensity in mA (left column); R1, recording site, same for R2 and R3; N, number of horses, SE, standard error; NA, not applicable; NS, not significant.

Latency (ms)
  (N = 0) (N = 4)  
2.5R1NANA2.48 (0.08)2.4 (2.3–2.7)NA
R2NANA4.03 (0.2)4.1 (3.4–4.6)NA
R3NANA5.42 (0.32)5.25 (4.9–6.5)NA
  (N = 0) (N = 5)  
5R1NANA2.34 (0.05)2.3 (2.2–2.6)NA
R2NANA4.13 (0.11)4.1 (3.9–4.6)NA
R3NANA5.24 (0.1)5.25 (4.9–5.65)NA
  (N = 3) (N = 6)  
10R12.34 (0.03)2.33 (2.3–2.4)2.13 (0.12)2.2 (1.6–2.4)NS
R24.19 (0.2)4.05 (3.95–4.7)3.83 (0.22)4.0 (2.9–4.5)NS
R35.45 (0.4)5.35 (4.8–6.3)5.14 (0.13)5.1 (4.7–5.55)NS
  (N = 4) (N = 6)  
15R12.43 (0.11)2.35 (2.3–2.8)2.01 (0.12)1.9 (1.45–2.45)NS
R24.23 (0.16)4.1 (3.95–4.7)3.66 (0.19)3.6 (2.7–4.5)NS
R35.4 (0.3)5.2 (4.8–6.3)4.83 (0.21)5.1 (4–5.45)NS
  (N = 6) (N = 6)  
20R11.99 (0.15)2.13 (1.7–2.9)2.15 (0.18)1.9 (1.45–2.45)NS
R23.66 (0.24)4.04 (3.3–4.5)3.96 (0.18)3.6 (2.7–4.5)NS
R34.84 (0.22)5.1 (4.4–5.85)5.09 (0.23)5.1 (4–5.6)NS
Amplitude (μV)
  (N = 0) (N = 4)  
2.5R1NANA89.52 (80.4)33 (0.6–375.5)NA
R2NANA50.78 (45.6)13.5 (0.9–213)NA
R3NANA0.41 (0.12)0.35 (0.2–0.7)NA
  (N = 0) (N = 5)  
5R1NANA38.6 (22.54)23.25 (0.3–148)NA
R2NANA63.575 (41.42)26.25 (3.9–4.6)NA
R3NANA0.65 (0.33)0.48 (0.2–2)NA
  (N = 3) (N = 6)  
10R12.88 (0.48)2.5 (2–4.5)62.07 (31.06)36.5 (0.6–225)NS
R22.78 (1.66)1.75 (0.6–7)99.97 (58.60)13.25 (0.6–420)NS
R30.15 (0.03)0.15 (0.1–0.2)0.68 (0.43)0.27 (0.1–2)NS
  (N = 4) (N = 6)  
15R17.1 (6.74)3 (2–24)146.22 (59.47)109.5 (5.5–478)NS
R22.34 (1.89)1.5 (0.6–7)78.06 (40.93)31 (2–242)NS
R30.14 (0.36)0.1 (0.1–0.2)0.89 (0.29)0.9 (0.05–2)NS
  (N = 6) (N = 6)  
20R149.45 (41.05)8.4 (5–295)142.49 (60.42)109.5 (2.45–478)NS
R216.64 (6.59)9.26 (5–52)78.44 (40.73)25.5 (3–242)NS
R30.57 (0.26)0.28 (0.15–2)1.43 (1.04)0.9 (0.1–5.15)NS
SNCV (m/s)
  (N = 0) (N = 4)  
2.5R1NANA72 (3.94)70 (65–85)NA
R2NANA118 (23.94)103 (76–200)NA
R3NANA257.3 (59.9)224 (113–433.5)NA
  (N = 0) (N = 5)  
5R1NANA72.75 (3.3)71.5 (64–85)NA
R2NANA94 (6.31)98.5 (74–109)NA
R3NANA282.6 (39.11)227 (211–408.5)NA
  (N = 3) (N = 6)  
10R180.38 (7.9)76.5 (68.5–100)81.25 (0.19)79.5 (68.5–103)NS
R2103.25 (10.56)105.5 (79–123)95.45 (5.13)100 (74–108)NS
R3262.38 (39.64)253.75 (188–354)259.8 (17.32)233 (130–436)NS
  (N = 4) (N = 6)  
15R177.9 (6.54)75 (68–100)82.43 (4.12)86 (63–97.5)NS
R2101.4 (8.18)103 (79–123)106.88 (11.12)95 (76–165.5)NS
R3278.7 (34.85)257.5 (188–354)259.75 (40.42)203 (174.5–540)NS
  (N = 6) (N = 6)  
20R186.15 (6.19)88 (66–106.5)84.67 (4.05)86 (66–97.5)NS
R2103.09 (4.23)100.5 (92.5–119)104.94 (10.67)100.5 (76–165.5)NS
R3302.77 (34.05)306.5 (173–457.14)286.5 (53.22)203 (166–540)NS

Postmortem Evaluation

Postmortem evaluation confirmed lack of other causes for the displayed HSK behavior (oral/dental, ocular, ear, airway [nasal, laryneal, guttural pouch, sinus] disease). Routine evaluation of the brain did not reveal abnormalities. There was no evidence of axonopathy or demyelination of the infraorbital nerve in affected horses (Fig 3).

Figure 3.

Histological evaluation of the infraorbital nerve. Cross-section representing part of the infraorbital nerve at a distance half way between the infraorbital and maxillary foramina in a horse with headshaking. Note multiple nerve fascicles surrounded by perineurium (P) containing axons surrounded by myelin and endoneurium (E). Blood vessels (BV) can be observed between nerve fascicles. Stained with modified Gomori Trichrome, bar 200 μm.


This study confirms involvement of the trigeminal nerve in the pathophysiology of idiopathic HSK in horses. Further, the results from the present study might suggest a functional rather than a structural alteration in the sensory pathway of the trigeminal complex as part of the pathophysiology of disease. This study validates sensory conduction evaluation of the trigeminal nerve (maxillary branch) in healthy horses, and provides reference data for its comparison with horses with suspected sensory trigeminal neuropathies.

An important finding was a significant difference in stimulus threshold values for sensory potentials, with headshakers having a lower threshold for activation than healthy horses. This means that affected horses are more sensitive or hyper-responsive to minimal stimuli. This finding might explain why some environmental factors such as sunlight, wind, noise, or exercise (innocuous to most individuals) can trigger episodes of uncontrollable HSK in affected horses.[1, 2, 8] The effect of season in nerve conduction was not investigated here. Although the number of control horses was small with half of the horses evaluated during the winter and half in the summer months, differences in nerve conduction were not observed among these horses. Evaluating healthy horses during the same season as that of horses with seasonal HSK would have avoided a possible confounding factor if any. However, as 41% of horses with HSK are nonseasonal,[2] evaluation of healthy horses at different seasons of the year for its comparison with affected horses is valid. All HSK horses displaying the characteristic behavior at the time of the study (n = 5/6) had a threshold for activation of 5 mA or less. The 1 horse with HSK not displaying the behavior at the time of the study (winter) was a seasonal HSK (previous 2 years). The threshold for activation in this horse was 10 mA, similar to that of 3 control horses. The 3 remaining control horses had an activation threshold of 15 mA or more. This study suggests that threshold for the activation of the trigeminal nerve is lower in horses with HSK during the time of clinical manifestations, and that the threshold may be higher when affected horses are not displaying the behavior. However, more horses are needed to confirm or refute this assertion. It would have been beneficial to repeat the study in this horse during the time of clinical manifestations, and repeat during remission. These findings appear to support a functional rather than a structural alteration in the sensory pathway of the trigeminal nerve as part of the pathophysiology of the disease. Additionally, there was no evidence of axonopathy or demyelination in the nerves examined. However, a full stereological, morphometric, histochemical, and ultrastructural evaluation of the trigeminal complex was not performed in this study and needs investigation (currently under study by the authors).

Increased sensitivity of the trigeminal nerve in horses with HSK was also supported by the induced bradycardia noted upon stimulation of the infraorbital nerve. The horses' heart rates returned to baseline rate upon discontinuation of the stimulus. This observation supports an induced trigeminovagal reflex.[24] Trigeminovagal (trigeminocardiac) reflexes have been documented in people upon stimulation (mainly surgical manipulation) of branches of the trigeminal nerve.[24] The oculocardiac reflex is the most well-recognized induced reflex via stimulation of the opthalmic branch of the trigeminal nerve.[24] Hypotension associated with trigeminovagal reflex was not observed in affected horses because dobutamine was used to maintain mean arterial blood pressure ≥70 mmHg in all horses studied. This reflex was not induced in control horses.

The latency at onset was not statistically different between groups of horses. Latencies of SEP have been reported to be delayed in people with idiopathic trigeminal neuralgia,[25] which was not observed in this study. Despite similar clinical manifestations in both species, the pathophysiology or duration of disease might be different. The duration of the wave that provides an estimate of temporal dispersion[20] was not different between groups of horses. The waveforms varied at different sites of recording (R1, R2, R3, R4). However, the waveform for each recording region did not vary between groups of horses (Fig 2). Waveforms for R1 and R2 sites consisted of a single negative peak (SNAP; Fig 2). A polyphasic potential was recorded at R3 which represents SNAP propagation into spinal tract of trigeminal nerve and depolarization of trigeminal nucleus. R4 represents the primary somatosensory cortical evoked potential. Depth of anesthesia can alter the appearance of the cortical somatosensory potential. Anesthesia in these horses was maintained at the same constant level to minimize variations in the recordings.

Although a difference in amplitude was noted for all stimulus intensities between groups of horses, this was not statistically different. The amplitude varied substantially among horses within both groups. Marked variations in amplitude of sensory action potentials are common in humans.[20] Factors that may influence amplitude include density of sensory innervation (sensory axon density), body mass (depth of nerve from skin surface), type of electrodes (needle versus surface), position of electrodes, and pre-existing neuropathy.[20] In the present study, a possible explanation of this variability in amplitude includes volume conduction differences as the result of body mass and recording electrode placement. Anatomical landmarks were used consistently for placement of recording electrodes. Once a SNAP was obtained, fine movements of the recording electrodes were made to obtain the best visible SNAP. Higher or lower amplitudes will be seen if the recording electrodes are close or far from the nerve, respectively.[20] Axonopathies result in decreased amplitude which was not the case of horses with HSK. Sample size in the present study was small. A larger number of horses might be needed to determine if significant differences in amplitude occur. SNAPs are more readily recognized than SEP because of their large amplitude as seen in Figure 2.[26]

Conduction velocities at the recording regions were not significantly different between groups. Variability in CV between sites is the result of difficulties in the accurate measurement of distances as the nerve is superficial distally (and easier to measure). The stimulus to R1 values are likely representative of the true infraorbital nerve CV. In contrast, the infraorbital to maxillary (R1–R2) and especially the maxillary to trigeminal nucleus (R2–R3) values are exaggerated by the use of distances that are considerably longer than the actual segments of nerve examined. However, the measurements were consistent for all horses, therefore comparisons are valid. Measurement of the nerve at necropsy would have provided more accurate distance information for each horse. High stimulus intensities might have also impacted distal latencies and CV in that the point of nerve excitation might have moved away from the cathode and closer to R1. Substantial slowing in CV implies demyelination of the sensory fibers[20] which was not the case of horses with HSK. There was no slowing in CV at any area of recording in HSK which indicates that conduction at the pre- and postganglionic level is comparable to that of the control group. It was essential to maintain the local temperature at a narrow range (35°C) in these horses, as CV can be altered substantially by changes in temperature.[20-22, 27] Lower temperatures slow down impulse propagation and increase amplitude of SNAP.[20] In contrast, impulse propagation is faster at higher temperatures.[20, 21] Correction factors to adjust for CV for temperature variation were not determined in this study. Published correction factors for sensory CV in limbs of horses and ponies were not used here as variations in temperature (head versus limbs) and correction factors differences might exist and therefore, might not be applicable to the head.[27] Further, it has been suggested the use of correction formulas to adjust CV for temperature in circumstances on which temperature cannot be normalized.[22] Differences in CV with sex and age (infants versus adults) have been described in people.[20] Sensory conduction velocities of the palmar and plantar digital nerves were reported to be significantly different between horses and ponies.[28] Differences in CV based on sex, age, breed, and height were not investigated here.

Trigeminal neuralgia in people is characterized by a recurrent sudden, severe sharp electric-shock-like pain.[29] People experience pain in one side of the jaw or cheek and it affects one or more divisions of the trigeminal nerve, most commonly the maxillary nerve.[29] Lateralization of signs has not been noted in horses with HSK.[2] The current study further supports no lateralization of altered SNC as thresholds were low bilaterally in horses with HSK. However, only the maxillary nerve was evaluated in these horses. Studies in humans have revealed spontaneous action potentials that evoke sustained discharges.[29] Through the stimulation of the maxillary nerve in these horses, a trigeminal-facial reflex was induced as observed by lip twitching. This twitching did not interfere with recording and did not generate movement artifacts. Movement can be a major limiting artifact when performing electrophysiology studies[30] but was not the case in this study. The American Association of Electrodiagnostic Medicine recommends optimal recording of SEP by the observation of muscle twitch or muscle twitch plus sensory threshold at each stimulus intensity.[31] This serves to standardize SEP testing by ensuring that the number of depolarized nerve fibers contributing to SEP is similar.[26] Visible muscle twitch threshold is sufficient for evoking SEP in people.[26] Muscle twitching in this study was associated with each stimulus which consistently evoked SEP.

Results from this study also provide baseline data for potential therapeutic management of suspected sensory trigeminal neuropathies including idiopathic HSK. For example, whether the administration of therapeutic agents might alter the threshold of activation in horses with idiopathic HSK, a topic currently under study by the authors. Whereas the sensory function of the trigeminal nerve is involved in the pathophysiology of HSK, findings from this study did not identify a specific cause. The authors previously investigated whether a possible association between latent EHV-1 infection and idiopathic HSK exists but found no apparent causative association.[15] Further investigation is needed to determine why horses with idiopathic HSK have lower stimulus thresholds for activation, what the exact mechanism is for nerve activation, and why some horses are seasonally afflicted and others are not. It is known that environmental factors have been associated with the development of HSK manifestations in affected horses.[1, 2]

In conclusion, this study provides objective evidence in support of sensory trigeminal nerve involvement in the pathophysiology of idiopathic HSK in horses. Affected horses have low stimulus thresholds for nerve activation, which is evident by triggered clinical manifestations upon exposure to apparently innocuous stimuli. Once a SNAP was triggered, there were no differences in latency, amplitude, and CV between both groups of horses, with the only difference being the activation threshold which might suggest a functional rather than a structural alteration in affected horses. This might be further supported by the findings in one horse with HSK with no clinical manifestations at the time of the study, who had thresholds comparable to control horses. Lateralization has not been reported in the clinical setting and was not seen in this study. Two reflexes were induced by stimulation of the infraorbital nerve; the trigeminal-facial in both groups of horses and trigeminovagal in horses with HSK. Lastly, as clinical manifestations in both humans and horses with trigeminal neuralgia are similar, and in both the sensory component of the trigeminal nerve is affected; horses could be a potential natural animal model for human disease. However, demyelination as reported in people with idiopathic trigeminal neuralgia was not evident in this study based on comparable SNAP latencies and conduction velocities between affected and control horses, and lack of histological alterations in horses with HSK. Additionally, other electrophysiological features of demyelination such as temporal dispersion, polyphasia, and conduction block were not seen in affected horses.[32]


The authors thank Ms Cindy McClin for organizing the data, and Mr John Doval for technical support. The study was carried out at UCD and supported by gifts from private donors to the Equine and Comparative Neurology Research Group at UCD.

Dedication: This study is dedicated in loving memory of Dr Terrell A. Holliday; founding father of veterinary neurology, neurosurgery, and electrophysiology.

Conflict of Interest Declaration: Authors disclose no conflict of interest.


  1. 1

    Anased; Akorn Inc, Decatur, IL

  2. 2

    CGeneTech Inc, Indianapolis, IN

  3. 3

    Nicolet Biomedical Inc, Madison, WI

  4. 4

    Subdermal E-2 Grass electrode; Astro-Med Inc, West Warwick, RI

  5. 5

    Soft retractable tape measure; Prym-Dritz, Spartenburg, SC