Trigeminal neuralgia (TN) is a rare neuropathic pain disorder affecting the fifth cranial nerve and characterized by sharp paroxystic pain episodes lasting from a few seconds to two minutes.Reference Scrivani, Mathews and Maciewicz 1 , Reference Zakrzewska 2 Although TN was described several decades ago,Reference Rose 3 , Reference Cole, Liu and Apfelbaum 4 its physiopathology remains poorly understood.Reference Burchiel 5 - Reference Blumenfeld 7 According to DandyReference Dandy 8 and Janetta,Reference Jannetta 9 , Reference Jannetta 10 TN is caused by a microvascular-induced compression of the dorsal root of the trigeminal nerve, most commonly from the superior cerebellar artery. Despite its popularity, the microvascular compression hypothesis has been challenged by several studies showing that: (1) many patients suffering from TN do not show evidence of microvascular compressionReference Sindou, Howeidy and Acevedo 11 , Reference Ishikawa, Nishi and Aoki 12 and (2) microvascular compression is present in a large proportion of individuals who do not suffer from TN.Reference Hardy and Rhoton 13 - Reference Obermann and Katsarava 16
In the past two decades, several reports have suggested that TN could be linked to impaired autonomic nervous system (ANS) activity.Reference Noguchi, Hasegawa and Kobayashi 17 - Reference Kranzl and Kranzl 22 Until recently, however, evidence of ANS dysregulation in TN remained mostly speculative or anecdotal. For example, Noguchi et alReference Noguchi, Hasegawa and Kobayashi 17 reported the case of a 29-year-old woman suffering from symptomatic TN who was significantly relieved following a stellate ganglion block (a sympathetic ganglion). Similar results were reported by Manahan et alReference Manahan, Malesker and Malone 21 following a sphenopalatine ganglion block (a parasympathetic ganglion) on a 56-year-old woman.
More recently, Simms and colleaguesReference Simms and Honey 23 sent questionnaires and reviewed the medical records of 92 TN patients. They noted that a majority of these patients (i.e. 67% of the patients sampled) had at least one autonomic symptom associated with their pain. Although some patients reported regional sweating (a symptom that can be linked to sympathetic activation), the vast majority reported parasympathetic-like symptoms (e.g. excessive salivation, nasal congestion, excessive tearing). Taken together, these observations support the idea that TN could be linked to ANS dysfunctions, and prod researchers to pursue future work to better understand the role of the ANS in TN pain.
A next logical step to the study of Simms and colleaguesReference Simms and Honey 23 would be to evaluate the activity and reactivity of the ANS in a population of patients suffering from TN using objective sympathetic and parasympathetic measures. In this regard, heart rate variability (HRV) techniques, allowing the recording of both sympathetic and parasympathetic indices, 24 seem particularly appropriate. To our knowledge, no previous study used HRV techniques to evaluate the reactivity of cardiac ANS responses in patients with TN. Given this knowledge gap, the aim of this study was to assess and compare the reactivity of the ANS to experimental pain in a population of TN patients and healthy controls using HRV analyses. Based on the results of Simms et al, we postulated that TN patients would show increased sympathetic and/or parasympathetic reactivity when compared with healthy controls.
Methods
Participants
Twelve patients diagnosed with classical TN and 12 healthy controls participated in this study (mean age±standard deviation [SD]=62±11 years) (Table 1). TN patients were recruited from the list of patients scheduled for Gamma-knife surgery at the Gamma-knife surgery clinic of the Sherbrooke University Hospital. Diagnosis of classical TN was confirmed by a neurosurgeon using the International Classification of Headache Disorders criteria. 25 Patients with atypical or symptomatic TN (e.g. TN secondary to multiple sclerosis) or with symptoms suggesting postherpetic or deafferentation pain syndromes were excluded. None of our retained participants showed evidence of tactile, thermal, or pricking hypoesthesia. None showed signs of dysesthesia, hyperesthesia, or allodynia. In addition to undergoing a conventional and sensory-specific neurological examination (conducted by both the appointed neurosurgeon and GL), every participant underwent magnetic resonance imaging to rule out possible neuronal damage.
Group Characteristics
Healthy controls were recruited through local ads and were all community-dwelling individuals. They all had good general health and none suffered from a painful condition, with the exception of three participants who reported minor osteoarthritic pain. Every participant was asked to refrain from using short-term analgesics two hours before testing and from drinking coffee and smoking cigarettes six hours before testing. TN patients were also asked to stop all pain medications for a period of 24 hours before their appointment to minimize the effect that medication could have on pain perception and ANS measures. Group and patient characteristics are listed in Tables 1 and 2.
Patients characteristics
BID=twice daily, TID=three times daily, PRN, as needed, QD=each day
The experiment took place at the Centre de Recherche du Centre Hospitalier Universitaire de Sherbrooke (Sherbrooke, Quebec, Canada). The local institutional ethics committee approved the study's procedures and each participant provided informed written consent before participating.
Experimental Pain: Cold Pressor Test
Participants were asked to immerse their right arm for 5 minutes in a bath of circulating cold water maintained at 10°C. This procedure, commonly known as the cold pressor test (CPT), was used to trigger ANS responses to pain. Every 30 seconds, patients rated their immersion-induced pain intensity using a 0–100 numerical rating scale (0=no pain; 100=intolerable pain).
Cardiac Activity
Five-minute electrocardiogram recordings were obtained during baseline testing (rest) and during experimental pain testing. Heart rate (HR) was monitored with a three-lead (electrocardiogram) montage and sampled at 1000 Hz using a PowerLab recording system with Chart software (AD Instruments, Colorado). Instantaneous RR intervals were obtained from the electrocardiogram waveform with a peak detection algorithm to detect successive R waves. All data were manually checked to ensure that only normal to normal intervals were analyzed. We analyzed HRV in the frequency domain. Specifically, fast Fourier transforms were used to calculate the power spectral density of HR oscillations. Two components can be distinguished from short-term spectrum analysis of HRV: a low-frequency (LF) (0.04-0.15 Hz) and a high-frequency (HF) (0.15-0.4 Hz) component. These two components are used as indices of cardiac autonomic nervous system activity. 24 HR fluctuations in the LF range reflect baroreflex-mediated sympathetic activity associated with Mayer waves of blood pressure.Reference Bernardi, Leuzzi and Radaelli 26 , Reference Cevese, Gulli and Polati 27 On the other hand, HRV in the HF range is generated by respiratory sinus arrhythmia and constitutes a sensitive measure of cardiac vagal parasympathetic activity.Reference Bloomfield, Zweibel and Bigger 28 , Reference Akselrod, Gordon and Ubel 29 Normalized HF and LF values were computed and analysed. Sympathovagal LF/HF ratios were not analysed because the sympathovagal balance indexed by the LF/HF ratio is already accounted for when normalizing HF and LF values.Reference Burr 30
Data Analysis
Autonomic reactivity to pain was evaluated by calculating the delta score between baseline and CPT-induced responses [delta score=(CPT – baseline)]. HR and HRV delta scores were compared between the two groups using independent t-tests. To examine the relationships between ANS reactivity to experimental pain and the characteristics of TN patients, we performed Pearson correlational analyses between the change in LF and HF observed from baseline to CPT and a series of demographic and clinical variables obtained at the moment of the experiment. These demographic and clinical variables included the age and sex of the patient, disease duration, average intensity of pain paroxysms experienced in the preceding week (0-100 numerical rating scale), and the average number of daily pain paroxysms experienced in the preceding week. Statistical significance was set at p<0.05. All tests were performed using SPSS (version 18.0 for Windows, Chicago, IL).
Because of the relatively small number of subjects included in the study and in spite of histograms suggesting the presence of normally distributed data, nonparametric tests were additionally used to compare HR and HRV measures between the two groups (Mann-Whitney tests) and to evaluate the association between the HRV measures and the clinicodemographic variables (Spearman tests). Results confirmed no difference between parametric and nonparametric approaches; therefore, only parametric statistics are reported.
Results
Pain Perception
Every participant experienced the CPT as a painful procedure. The mean pain intensity score during the CPT was comparable between the two groups (mean±SD=48 ±27 for healthy controls and 63±26 for TN patients; t=−1.42; p=0.17). None of the TN participants experienced facial pain during the testing session.
Cardiac HR Responses
No difference in HR (both at rest and during CPT) was observed between the two groups. Healthy controls and TN patients showed a similar increase in HR during the CPT (Table 3).
Heart rate (mean±SD)
BPM=beats per minute.
Autonomic Activity and Reactivity to Pain
There was no difference in resting sympathetic (LF) or parasympathetic activity (HF) between healthy controls and TN patients (Table 4). However, during the CPT, TN patients showed a greater increase in LF (t=−2.60, p<0.05) and a greater decrease in HF (t=2.23; p<0.05) when compared with healthy controls, indicating greater increment in sympathetic activity and greater decrement in parasympathetic activity during experimental pain, respectively (Figures 1 and 2).
Mean change in LF between baseline (resting condition) and CPT. Both healthy controls and TN patients had an increase in sympathetic activity (LF) during CPT. TN patients showed a greater increase in sympathetic activity when compared with healthy controls (*p<0.05).
Mean change in HF between baseline (resting condition) and CPT. Both healthy controls and TN patients showed a decrease in parasympathetic activity (HF) during CPT. TN patients showed a greater decrease in parasympathetic activity when compared with healthy controls (*p<0.05).
Baseline HRV values
To extend the results obtained with the LF and HF measures and better understand the role of the ANS in TN, we further explored HRV analysis in the time domain. More specifically, we calculated the root mean-square of successive differences between adjacent normal-to-normal intervals, the number of pairs of adjacent normal-to-normal intervals differing by more than 50 ms, and the SD of normal-to-normal intervalsReference Lotufo, Valiengo and Bensenor 31 , Reference Hilz and Dutsch 32 and compared results between the two groups of participants. Contrary to LF and HF analyses, no significant group differences were noted for root mean-square of successive differences, normal-to-normal intervals differing by more than 50 ms, and SD of normal-to-normal intervals (all p values >0.05).
Associations Between HRV Measures and Clinicodemographic Variables
The correlation coefficients obtained between the HRV measures and the demographic and clinical variables obtained at the moment of the experiment are presented in Table 5. As can be seen from the table, there was a negative association between the CPT-induced change in LF activity and the average number of daily pain paroxysms experienced by the patients in the preceding week. No other significant results were observed (all other p values >0.05).
Relationships between HRV measures and clinicodemographic variables
* Pearson correlation.
† Independent t-test.
Discussion
The purpose of this study was to test for differences in pain-evoked ANS activity between TN patients and healthy controls. Results indicate that TN patients and healthy controls have comparable autonomic cardiac responses at rest, but that in response to a tonic experimental pain challenge, TN patients show greater sympathetic arousal and parasympathetic withdrawal than healthy controls. The results also revealed the presence of an association between sympathetic reactivity to pain and the number of pain paroxysms experienced each day by TN patients in the preceding week. Surprisingly, the direction of the association suggests that the ANS of patients with few pain paroxysms is more reactive to experimental pain than the ANS of patients with frequent pain paroxysms. The exact reason for the negative association between sympathetic reactivity and the number of pain paroxysms experienced by TN patients remains unclear and surely merits future attention.
Cause or Consequence
At this time, it is difficult to determine if the altered autonomic reactivity to pain observed in TN patients is a cause or a consequence of TN pain. Some may for instance suggest that the increased sympathetic reactivity observed in TN patients represents a pathophysiological feature of TN, responsible (alone or combined with other factors) for the occurrence of pain episodes. This hypothesis can be supported by the observations of Noguchi et al,Reference Noguchi, Hasegawa and Kobayashi 17 who reported clinical improvement (i.e. decreased pain) in a TN patient following a sympathetic ganglion block.
Others, alternately, might argue that the differences noted in TN patients could be attributable to an adaptation of the nervous system in response to the frequent pain paroxysms experienced by these individuals. Somewhat supportive of this view are the results of Chalaye et al,Reference Chalaye, Lafrenaye and Goffaux 33 who noted the presence a positive relationship between the sympathetic response and the efficacy of pain inhibitory mechanisms (both triggered by the same CPT). In this regard, the increased sympathetic response noted in TN patients could be seen as an effective strategy to alleviate pain. Such an interpretation could help to explain why the number of pain paroxysms noted in this study in TN patients was lower in those with high sympathetic arousal. Future research is essential to better understand the link that exists between TN pain and ANS dysfunction.
Trigeminal Neuralgia and Trigeminal Autonomic Cephalalgias
In 1997, Goadsby and Lipton suggested that many headache disorders could be grouped together based on the presence of prominent autonomic features.Reference Goadsby and Lipton 34 Goadsby and Lipton used the appellation “trigeminal autonomic cephalalgia” to describe this group of headaches, characterized by short-lasting, unilateral, and extremely severe pain associated with blatant cranial autonomic symptoms (e.g. tearing, ptosis, rhinorrhea). Today, the appellation trigeminal autonomic cephalalgia is a widely accepted throughout the medical community, and includes disorders such as cluster headache (CH), chronic paroxysmal hemicrania, and short-lasting unilateral neuralgiform headache attacks with conjunctival injection and tearing (SUNCT) syndrome.Reference Goadsby and Lipton 34 Despite the similarities between these disorders and TN (i.e., short-lasting and severe pain usually affecting one side of the face), TN was not classified under the trigeminal autonomic cephalalgia family, largely because of the absence of salient autonomic features.Reference Goadsby and Lipton 34 , Reference May 35 The present results, as well as the observations of Simms et al.,Reference Simms and Honey 23 challenge this view by showing that TN may also be linked with autonomic features.
Chronic Pain Syndromes and ANS Dysfunctions
Autonomic dysfunctions have been documented in numerous other painful conditions, including irritable bowel syndrome (IBS),Reference Tousignant-Laflamme, Goffaux and Bourgault 36 fibromyalgia,Reference Cohen, Neumann and Shore 37 complex regional pain syndrome,Reference Vogel, Gradl and Ockert 38 temporomandibular disorder,Reference Eze-Nliam, Quartana and Quain 39 and cluster headaches.Reference Tassorelli, Micieli and Osipova 40 For instance, studies looking at ANS arousal among fibromyalgia patients found that these patients have increased sympathetic and decreased parasympathetic activity at rest.Reference Cohen, Neumann and Shore 37 , Reference Furlan, Colombo and Perego 41 - Reference Stein, Domitrovich and Ambrose 44 Fibromyalgia patients also exhibit greater sympathetic and weaker parasympathetic cardiac reactivity during painful stimuli,Reference Chalaye, Goffaux and Bourgault 45 a pattern of changes associated to increased HR response to pain (see del Paso et alReference Reyes del Paso, Garrido and Pulgar 46 and Chalaye et alReference Chalaye, Goffaux and Bourgault 45 ). On their hand, Tousignant-Laflamme et alReference Tousignant-Laflamme, Goffaux and Bourgault 36 noted that IBS patients and healthy controls had an opposite autonomic response during the CPT—that is, decreased parasympathetic/increased sympathetic reactivity for healthy controls and increased parasympathetic/decreased sympathetic reactivity for IBS patients. Compared with healthy controls, IBS patients also showed a lower HR response during the CPT.Reference Tousignant-Laflamme, Goffaux and Bourgault 36 Finally, in an elegant study published by Tassorelli et al,Reference Tassorelli, Micieli and Osipova 40 the authors recorded the pupillary and cardiovascular (blood pressure) responses to the CPT in a population of individuals with cluster headaches and healthy controls. Compared with the healthy controls, patients with cluster headaches had an abnormal pupillary and cardiovascular response, pointing towards a general sympathetic hyperactivation and concomitant pupillary sympathetic hypofunction.
In the present study, there was no difference between TN patients and healthy controls for the ANS measures recorded at rest, but we observed a similar pattern to that of del Paso et al,Reference Reyes del Paso, Garrido and Pulgar 46 Chalaye et al,Reference Chalaye, Goffaux and Bourgault 45 and Tassorelli et alReference Tassorelli, Micieli and Osipova 40 during exposure to a painful stimulus (i.e. decreased parasympathetic and/or increased sympathetic reactivity to pain in pain patients compared with healthy controls). However, contrary to Chalaye et al,Reference Chalaye, Goffaux and Bourgault 45 Reyes del Paso et al,Reference Reyes del Paso, Garrido and Pulgar 46 and Tousignant-Laflamme et al,Reference Tousignant-Laflamme, Goffaux and Bourgault 36 we observed no difference in HR between pain patients and healthy controls. The similarities and differences observed between the current results (obtained in a TN population) and the results of Chalaye et alReference Chalaye, Goffaux and Bourgault 45 and Reyes del Paso et alReference Reyes del Paso, Garrido and Pulgar 46 (fibromyalgia population), Tousignant-LaflammeReference Tousignant-Laflamme, Goffaux and Bourgault 36 (IBS population) and Tassorelli et alReference Tassorelli, Micieli and Osipova 40 (CH population) suggest that autonomic abnormalities are probably a common feature of many painful conditions, but that each condition has a distinct ANS dysfunction profile.
Limitations
An important limitation of the present study concerns the relatively small number of participants tested. Small sample size impedes the statistical power of the analyses, thus increasing the probability of committing a type II error. Perhaps a larger sample size would have allowed us to detect other ANS abnormalities (e.g. between-group differences in sympathetic and parasympathetic activity at rest). Future studies need to be carried out before any final conclusion can be made regarding the exact pattern of ANS dysregulation in TN.
The results of the present study are derived from a single autonomic test and a limited number of ANS measures. Of importance, the differences observed between the healthy controls and TN individuals were noted only for the LF and HF measures; no significant group differences being present for the other time domain variables (root mean-square of successive differences, normal-to-normal intervals differing by more than 50 ms, SD of normal-to-normal intervals). Several other tests (e.g. cold face test, slow deep breathing) and measures (e.g. skin blood flow, pupillometry) can been used to evaluate the activity and reactivity of the ANS (see for instance Hilz and DutschReference Hilz and Dutsch 32 ). The use of the cold face test could provide more information on the role of the parasympathetic nervous system in the pathophysiology of TN.Reference Khurana, Watabiki and Hebel 47 , Reference Heath and Downey 48 However, the possibility of triggering intense pain episodes in TN individuals with a cold face test is quite high, hence raising important ethical and scientific concerns about the use of cold face tests in this population. Other tests evaluating parasympathetic reactivity (e.g. slow deep breathing, Valsalva maneuver) should perhaps be preferred.
Moreover, the continued use of medication in the group of TN patients raises the possibility that the pattern of ANS results observed is due to medication use rather than to the presence or absence of TN. In a recent review, Lotufo et alReference Lotufo, Valiengo and Bensenor 31 observed a trend for higher baseline LF values in patients with epilepsy receiving anticonvulsant drugs. In the present study, we observed no difference in resting autonomic indices between healthy controls and TN patients. We also asked TN patients to stop all pain medications (including antiepileptic drugs) for 24 hours before their appointment. Nevertheless, given the long half-life of antiepileptic drugs,Reference Rang, Dale, Ritter, Flower and Henderson 49 we cannot exclude the possibility that antiepileptic drugs are contributing to the observed effects. Future studies should be wary of this potential confound.
Conclusions
In the present study, we sought to compare the autonomic reactivity to pain between TN patients and healthy controls. Our results showed that, for a comparable degree of experimental pain, TN patients showed greater increase in cardiac sympathetic activity and a greater decrease in cardiac parasympathetic activity during CPT compared with healthy controls. Although it is unclear if this altered autonomic response to experimental pain is also present during TN pain paroxysms, the present pattern of results opens interesting research avenues for the understanding of TN physiopathology and for the development of new treatment approaches (e.g. sympathetic blockers) to ease TN symptoms.
Acknowledgements and Funding
The authors would like to thank Brigitte Cabana for her assistance and the Centre de Recherche du Centre Hospitalier Universitaire de Sherbrooke for their support. Serge Marchand was supported by the Canadian Institutes of Health Research. Guillaume Léonard and Philippe Goffaux are supported by Natural Sciences and Engineering Research Council of Canada and Fonds de Research Santé (Quebec).
Disclosures
SM has the following disclosures: Lilly Canada, consultant, honoraria, speaker, grant recipient, grants; AstraZeneca, consultant, honoraria, speaker, grant recipient, grants; Jensen, consultant, honoraria, speaker; Pfizer, consultant, honoraria, speaker, grant recipient, grants.
GL, PC, PG, DM, and IG do not have anything to disclose.
