Autonomic neuropathy, a relatively common complication of several chemotherapy agents, can affect the vagus nerve and its pain inhibitory capacity, thus increasing sensitivity to pain. This study aimed to evaluate the relationships between autonomic parasympathetic function and the perception of (1) spontaneous pain; (2) experimental non-painful sensations; and (3) experimental painful sensations in chemotherapy-induced neuropathy patients.
Twenty-seven cancer patients with chemotherapy-induced polyneuropathy were enrolled (20 women, age 56.6 ± 7.9). Autonomic parameters of heart rate variability, deep-breathing and Valsalva ratios, experimental non-painful parameters of warm, cold and mechanical detection thresholds, and painful parameters of heat pain thresholds, pain rating of suprathreshold stimulus, mechanical temporal summation and conditioned pain modulation response were examined.
Autonomic parameters and spontaneous pain levels were not associated, yet autonomic parameters were positively correlated with non-painful sensations – milder autonomic neuropathy was accompanied by milder sensory neuropathy as indicated by several parameters, e.g., lower Valsalva ratio was correlated with higher warmth detection threshold (r = −0.465; p = 0.033). Autonomic parameters were, however, negatively correlated with painful sensations – lower parasympathetic-vagal activity was associated with higher pain sensitivity as indicated by several parameters, e.g., lower Valsalva ratio was correlated with higher pain rating of suprathreshold stimulus (r = −0.559; p = 0.008).
Diminished vagal function due to neuropathy is associated with, and may possibly underlie, pain disinhibition expressed as greater levels of experimental pain.
Cancer patients treated with chemotherapy agents may develop polyneuropathy with autonomic and/or sensory, potentially painful expressions.
Afferent vagal activity exerts inhibitory effect on the overall pain perception.
What does this study add?
There exists a correlation between severity of autonomic neuropathy and perceived levels of experimental pain, although not with levels of spontaneous pain.
Autonomic neuropathy has been widely described in diabetic patients (Young et al., 1986; Lanting et al., 1989; Veves et al., 1994; Lluch et al., 1998; Tentolouris et al., 2001; Vinik et al., 2003; Marthol et al., 2006; Gibbons and Freeman, 2010; Spallone et al., 2011; Kim et al., 2012) and, to a lesser extent, also in cancer patients following chemotherapy (Hirvonen et al., 1989; Ekholm et al., 1997, 2000). The vagus nerve, the longest autonomic nerve, mediates 75% of all parasympathetic activity and is one of the first to be affected.
Vagal nerve activation has been shown to inhibit pain perception. The mechanism probably involves activation of the nucleus of the solitary tract, where vagal afferents end, and its connections to various pain modulation areas, including the periaqueductal grey area and rostral ventromedial medulla (Benarroch, 1993, 2001). The pain-inhibiting role of the vagus was widely demonstrated in animals, where vagal stimulation or vagotomy caused an attenuation or augmentation of pain, respectively (Randich and Gebhart, 1992; Thurston and Randich, 1992; Ren et al., 1993; Khasar et al., 1998a, b, 2003; Weissman-Fogel et al., 2008). In humans, a reduction in experimental pain perception in response to stimulating vagal afferents was demonstrated among healthy subjects (Sedan et al., 2005; Busch et al., 2012) and epileptic patients undergoing vagal nerve stimulation (Kirchner et al., 2000, 2006; Ness et al., 2000). It was also demonstrated that lower parasympathetic activity, as indicated by the baroreflex sensitivity, was associated with higher pain sensitivity to the cold pressor test in fibromyalgia patients and healthy subjects (Duschek et al., 2007; Reyes del Paso et al., 2011). Thus, it can be assumed that autonomic neuropathy that includes the vagus nerve may lead to reduction of its pain inhibitory effect, which, in turn, can contribute to enhance pain sensation.
Still, only a few studies looked at the relationships between autonomic function and pain in neuropathy. Painful diabetic polyneuropathy (painful-PNP) patients show lower parasympathetic-vagal function compared with non-painful polyneuropathy (non–painful-PNP) patients (Gandhi et al., 2010). In addition, lower parasympathetic-vagal activation was associated with higher pain scoring in diabetic polyneuropathy patients (Gandhi et al., 2010; Spallone et al., 2011). Other studies, however, failed to find such a difference or an association between autonomic function and spontaneous pain levels in diabetic polyneuropathy patients (Veves et al., 1994; Krämer et al., 2004).
The aims of the current study were to evaluate the relationships between autonomic function and the perception of (1) spontaneous pain; (2) non-painful sensory thresholds; and (3) perception of painful sensory stimuli in chemotherapy-induced neuropathy patients. We hypothesized that patients with more severe autonomic neuropathy will demonstrate higher spontaneous pain ratings, more severe sensory neuropathy expressed by higher sensory thresholds and higher sensitivity to experimental painful stimuli. Chemotherapy-induced neuropathy patients have been less studied in relation to autonomic neuropathy.
The experiment was conducted by a single investigator (H.N-A) in the Clinical Neurophysiology Laboratory at the Technion in Haifa, Israel. The study was approved by the Rambam Health Care Ethics Committee, and all the participants signed the informed consent prior to the experiment. Patients were asked to refrain from pain relief medications for 6 h before the beginning of the experimental session. Each session lasted for approximately 1 h.
Twenty-seven cancer patients [20 women and 7 men, mean ± standard deviation (SD) age of 56.6 ± 7.9] were enrolled in the study. Table 1 depicts the type of cancer, chemotherapy agent that was used and analgesic used by the patient. The patients were recruited through the oncology departments of Rambam and Lin Medical Centers in Haifa, Israel. All patients met the following criteria: (1) treatment with a neurotoxic chemotherapy agent; (2) presence of neuropathy as clinically diagnosed by the treating oncologist or neurologist; (3) absence of other chronic pain or neurological disturbances; and (4) ability to communicate and understand the instructions of the study.
Patients scored the intensity of their spontaneous pain at the beginning of the experimental session and the maximum pain intensity they felt during the week prior to the experimental session using a numeric pain scale of 0 (no pain at all) to 10 (worst imaginable pain). We divided the patients into two groups according their maximum pain: Patients with spontaneous pain were assigned to the painful-PNP (> 0) group and patients with no spontaneous pain to the non–painful-PNP group.
2.2 Autonomic assessments
The role of the following parameters was to identify the autonomic neuropathy level and, more specifically, to focus on the parasympathetic-vagal function. There are various indirect ways to assess the parasympathetic-vagal function and we focus on heart rate variability (HRV) analysis. Three electrocardiogram (ECG) electrodes were mounted on each patient, for recording of heart beats, using the Suempathy100 device (Tehnik Seuss Medizin, Aue, Germany). Raw ECG data were visually inspected post-hoc for artefacts and abnormal beats, which were deleted.
2.2.1 Baseline rest
Heart rate was recorded for 5 min at rest. During the recordings, the patients were asked to sit quietly and breathe normally. HRV analysis, which is a non-invasive technique to assess the autonomic nervous system function, was performed using both time and frequency domain analyses. In the time domain analysis, the rMSSD (the root mean square of successive differences at rest), which reflects vagal-parasympathetic activity, was calculated. In addition, frequency domain analysis was performed to obtain high-frequency power (HF: 0.15–0.40 Hz, reflecting vagal-parasympathetic activity) and low-frequency power (LF: 0.04–0.15 Hz, reflecting sympathetic and parasympathetic activity) components. In this study, we calculated the ratio of LF/HF, which reflects the sympatho-vagal balance (von Borell et al., 2007).
2.2.2 Deep breathing test
Patients were asked to sit quietly and to breathe deeply at 6 breath cycles/min, repeatedly inhaling for 5 s and exhaling for 5 s, for 1 min. The ratio of the longest R-R interval and the shortest R-R interval was measured and the final deep breathing ratio was determined by averaging the results of three maximal ratios in order to examine the maximal vagal activation. Higher ratio indicates higher vagal-parasympathetic function.
2.2.3 Valsalva manoeuvre
In this test, patients were asked to blow into a mouthpiece attached to a manometer, holding it at a pressure of 40 mmHg for 15 s. The manoeuvre was repeated twice with approximately 1-min interval in-between. The Valsalva ratio was calculated as the longest R-R interval after the manoeuvre divided by the shortest R-R interval during the manoeuvre. The mean of the two Valsalva ratios was taken as the final value. Higher value indicates higher parasympathetic function.
2.3 Experimental non-painful sensations
The role of the following parameters was to identify the sensory neuropathy level.
2.3.1 Thermal detection thresholds
Cold and warm detection thresholds were examined on the right dorsal part of the foot in a random order. The thermal thresholds were detected using the thermal sensory analyzer (TSA2001, Medoc, Ramat Yishai, Israel) with a 30 × 30 mm Peltier surface thermode. The baseline temperature was set at 32.0 °C and was increased or decreased at a rate of 1 °C/s. The subjects were instructed to signal when the stimulus was first perceived as warm or cold. If a subject failed to signal before reaching the cut-off temperature of 50.0 or 0 °C, then this was recorded as the threshold temperature. The final threshold values were determined by averaging the results of three trials.
2.3.2 Mechanical detection thresholds
The mechanical detection thresholds were examined using von Frey monofilaments (Stoelting, Wood Dale, IL, USA), with each filament applied three times in an ascending sequence until the threshold was detected in at least two of the three trials. Then, the next lower von Frey filament was applied, and the lowest filament to be detected at least twice was determined as the mechanical detection threshold. This was examined on the right dorsal part of the foot.
2.4 Experimental painful sensations
The role of the following parameters was to identify the pain processing system, including pain perception and modulation.
2.4.1 Heat pain thresholds
Heat pain thresholds were detected on the right volar forearm using TSA in the method of limits. The baseline temperature was set at 32.0 °C and was increased or decreased at a rate of 1 °C/s. The patients were instructed to signal when the stimulus was first perceived as painful. If a subject failed to reach the pain threshold before the cut-off temperature of 50.0 °C, then this was recorded as the threshold temperature. The final threshold values were determined by averaging the results of three trials.
2.4.2 Mechanical temporal summation
In this test, the perceived intensity was reported using rating scale of 0 (no pain at all) to 100 (worst pain imaginable) for a single pinprick stimulus (using a von Frey filament of 6.45 Nm = 225.1 g) and then for the last stimulus in a series of 10 stimuli. The stimuli were given at a rate of 1 Hz, applied to the left volar forearm, within an area of 1 cm2. The temporal summation value was calculated as the difference between the scoring for the last stimulus in the series of 10 stimuli and the scorings for the single stimuli.
2.4.3 Suprathreshold pain – ‘test pain’ stimulus
The temperature that induced pain-60 was determined individually, as was previously described (see Granot et al., 2008). In general, subjects were exposed to a series of heat stimuli of 7-s plateau duration starting at the intensity of 1 °C above their individual heat pain threshold to their right volar forearm. The baseline temperature was set at 32.0 °C and was increased or decreased at a rate of 2 °C/s. After each stimulus, the subjects were asked to verbally report the level of pain. If this stimulus induced pain-60, which is the temperature that induces pain at the intensity of 60 (0–100 scale), then that temperature was chosen for the ‘test-pain’ stimulus; if not, then additional higher or lower temperatures were applied for the determination of the pain-60 temperature. Further, 10 min after finding the pain-60 temperature, the ‘test-pain’ stimulus was applied for 60 s to the right volar forearm; during this time, the subjects scored their pain levels every 10 s. In this test, the ‘test-pain’ stimulus was delivered alone and thus was termed ‘test stand-alone’.
2.4.4 Conditioned pain modulation (CPM) paradigm
The ‘test-pain’ stimulus was assessed before and during a painful conditioning (‘conditioned test-pain’); 15 min following the delivery of the ‘test stand-alone’ stimulus, the subjects were asked to place their left hand in a 46.5 °C water bath (Heto CBN 8-30 Lab equipment, Allerod, Denmark) for 60 s. During the first 30 s, the subjects scored their pain levels every 10 s. After 30 s, the ‘test pain’ stimulus was applied again to the right forearm. The subjects were asked to rate the intensity of the ‘test pain’ stimulus every 10 s after its onset while keeping their left hand in the water.
2.5 Study design
After meeting the inclusion criteria, all patients received an explanation about the study and they were asked to sign the inform consent. Then, the patients were interviewed about their disease characteristics and spontaneous pain. Then, the autonomic measurements were assessed in the same order: baseline, deep breathing test and the Valsalva manoeuvre. Following the autonomic measurements, the sensory stimuli were presented in the following order: starting from the least painful; thermal and mechanical detection thresholds; heat pain thresholds; mechanical temporal summation; application of the suprathreshold pain; and CPM.
2.6 Statistical analysis
Autonomic parameters (rMSSD and LF/HF ratio during rest as well as deep breathing ratio and Valsalva ratio) of the painful-PNP and non–painful-PNP were not distributed normally and were compared using a non-parametric Wilcoxon rank-sum test.
Pain change value (30) was calculated as the difference between the pain ratings after 30 s minus pain ratings after 10 s of the ‘test stand-alone’ stimulus. The pain change value (60) after 60 s was calculated in a similar way. The CPM response was calculated by subtracting the average scores of the ‘test stand-alone’ stimulus from the scores of the ‘conditioned test-pain’ stimulus. For this, the average pain ratings during the first 30 s of the ‘test stand-alone’ stimulus were calculated. Negative values of CPM indicate a more efficient pain inhibition response.
The associations between the autonomic parameters and the spontaneous pain and sensory responses were assessed using Pearson's correlations. As these were originally exploratory, no correction is provided for multiple tests. The data are presented as means ± SDs. Statistical significance was defined as p < 0.05.
Statistical analyses were performed using JMP (SAS Institute, Cary, NC, USA).
The painful-PNP group consists of 13 patients (4 men, mean ± SD age of 55.7 ± 9.5) and the non–painful-PNP group consists of 14 patients (3 men, 57.5 ± 6.5). The two groups did not differ significantly in gender and age. In the painful-PNP, the level of spontaneous pain at the beginning of the experimental session was 3.7 ± 3.9. The maximum spontaneous pain level during the week prior to the experimental session was 7.1 ± 2.9.
3.1 Autonomic function and spontaneous pain
No difference was found between the two groups of patients in autonomic function. Time domain analysis of HRV revealed no difference in rMSSD [the median (range) was 17.9 (3.3–24.8) for the painful-PNP patients and 7.7 (4.5–26.6) for the non–painful-PNP patients; p = 0.237; see Table 2]. Similar findings were found in the LF/HF ratio [3.2 (0.9–9.1) and 4.4 (0.4–10.8) in the painful-PNP and non–painful-PNP patients, respectively, p = 0.878].
Table 2. The autonomic function in the painful and non-painful chemotherapy-induced neuropathy patients. Data are presented as median (range)
Painful neuropathy (n = 10; M-1)
Non-painful neuropathy (n = 13; M-4)
Difference between the two groups of patients
LF/HF, low frequency/high frequency; rMSSD, the root mean square of successive differences at rest.
p = 0.237
p = 0.878
Deep breathing ratio
p = 0.951
p = 0.972
The deep breathing ratio also did not differ between the two groups of patients [1.12 (1.02–1.41) and 1.14 (1.05–1.43) in the painful-PNP and non–painful-PNP patients, respectively, p = 0.951]. Similar results were found for the Valsalva ratio; there was no difference in the Valsalva ratio between the painful-PNP and non–painful-PNP patients [1.39 (1.04–1.97) and 1.42 (1.17–1.74), respectively, p = 0.972].
The correlations between the autonomic parameters and the spontaneous pain levels were tested only for the painful-PNP group. For this group, no correlations were found between any of the autonomic parameters and the two spontaneous pain parameters.
3.2 Correlations between autonomic and sensory functions
3.2.1 Experimental non-painful sensation
Our results suggested association between levels of autonomic and sensory neuropathy; in the non–painful-PNP patients, lower Valsalva ratio, which indicates lower parasympathetic-vagal function and more severe autonomic neuropathy, was correlated with higher cold detection thresholds in the foot (r = 0.615; p = 0.044) as well as with higher mechanical detection threshold in the foot (r = −0.597; p = 0.052). This, however, was not found for the painful-PNP patients or when combining the two groups together. In the same line, lower Valsalva ratio also exhibited a trend for association with higher warm detection threshold in the foot; r = −0.535, p = 0.090 and r = −0.615, p = 0.059 in the non–painful-PNP and painful-PNP patients, respectively. When combining these two groups, the correlation became significant (r = −0.465; p = 0.033; Fig. 1) probably due to the larger number in the combined group. This indicates that subjects with more severe autonomic neuropathy also exhibit more severe sensory neuropathy.
3.2.2 Experimental painful sensation
In the non–painful-PNP group, lower rMSSD was correlated with lower heat pain thresholds (r = 0.602; p = 0.050) such that lower parasympathetic-vagal activity, which can indicate more severe autonomic neuropathy, was related to higher sensitivity to pain. This correlation was not significant in the painful-PNP group (r = 0.327; p = 0.357), but when combining the patients, the correlation became significant (r = 0.433; p = 0.050; Fig. 2).
Lower Valsalva ratio was correlated with higher average scoring of ‘test stand-alone’ stimulus such that lower parasympathetic-vagal activity was related to higher pain sensitivity: r = −0.586; p = 0.058 in the non–painful-PNP and r = −0.639; p = 0.046 in the painful-PNP. This was also highly correlated when the two groups were combined (r = −0.559; p = 0.008; Fig. 3).
In addition, lower Valsalva ratio was correlated with lower level of pain change value (30) (r = −0.638; p = 0.047 and r = −0.566; p = 0.069 for the painful-PNP and non–painful-PNP, respectively), such that lower parasympathetic-vagal activity was related to lower pain modulation. This was still significant when the two groups were combined (r = −0.495; p = 0.023; Fig. 4).
The CPM responses and the temporal summation values were not associated with any of the autonomic measurements. Sample size calculation revealed that even after theoretical tripling of our sample size, the correlations would not become significant.
Vagal nerve activation has been shown to inhibit experimental pain in animals (Randich and Gebhart, 1992; Thurston and Randich, 1992; Ren et al., 1993; Khasar et al., 1998a, b, 2003; Weissman-Fogel et al., 2008) and humans (Kirchner et al., 2000, 2006; Ness et al., 2000; Sedan et al., 2005; Busch et al., 2012; Napadow et al., 2012). Thus, vagal involvement in chemotherapy-induced neuropathy patients is expected to lead to reduction of the vagus nerve pain inhibitory effect and enhance pain. The current study found that (1) unexpectedly, there was no difference in the extent of autonomic neuropathy between painful and non–painful-PNP patients; (2) in line with our hypothesis, severity of autonomic neuropathy was associated with pain perception for experimental painful stimuli – milder autonomic neuropathy is accompanied by higher pain sensitivity; and (3) associations were found between parameters of autonomic and sensory neuropathies – milder autonomic neuropathy is accompanied by milder sensory neuropathy.
4.1 Autonomic function and spontaneous pain
The autonomic function in painful-PNP versus non–painful-PNP patients has been explored in a few studies, all in diabetic patients. To the best of our knowledge, only two of these studies reported a difference in the autonomic function between the two groups; however, this was found for only few of the several parameters that were tested (Gandhi et al., 2010; Spallone et al., 2011). The latter reported an association between lower parasympathetic-vagal activity and higher spontaneous pain scores, but this correlation was lost in their regression model, leading the authors to suggest that another factor, neuropathy severity, explains the finding. Our study failed to find a relationship between the level of parasympathetic-vagal activity and spontaneous pain level. In addition, we found no differences between painful-PNP and non–painful-PNP patients in the parasympathetic-vagal activity. This is in line with other studies that had similar results in diabetic patients (Veves et al., 1994; Tentolouris et al., 2001; Krämer et al., 2004). This may suggest that the level of spontaneous neuropathic pain is not related to the degree of parasympathetic-vagal activity. The pain inhibitory effect of the vagus, thus, does not seem to be a central mechanism in determining spontaneous pain levels in these patients.
4.2 Autonomic function and experimental non-painful sensations
Autonomic and sensory dysfunctions are common expressions of diabetic neuropathy, whose mutual association is debated; using tests of autonomic function, including changes in blood pressure when standing up and during the handgrip test for the sympathetic, and the Valsalva test and deep breathing test for the parasympathetic, Töyry et al. (1997) and Tentolouris et al. (2001) reported no somato-autonomic link, while Lluch et al. (1998) reported that autonomic neuropathy was more prevalent in diabetic patients with sensorimotor neuropathy compared to those without. Similarly, Kim et al. (2012) reported that 80% of diabetic patients with sensory neuropathy also have autonomic neuropathy. To the best of our knowledge, there are no data on such association for chemotherapy-induced neuropathies. In our study, we focused on assessing parasympathetic function by examining the deep breathing test, Valsalva test, and the time and frequency domains of the HRV. The correlations that were found between the level of autonomic and sensory expressions of neuropathy are in line with previous studies (Tentolouris et al., 2001; Gibbons et al., 2010).
It can be argued that the length differences between the relatively short vagus nerve and the longer somatosensory limb nerves should contradict an association between the extents of autonomic and sensory neuropathies, and only when a severe peripheral neuropathy is observed the autonomic neuropathy will emerge. However, results from previous studies and the present study do not support this claim since autonomic impairment can develop even when peripheral somatosensory neuropathy is in an early stage.
4.3 Autonomic function and experimental painful sensations
In the current study, lower parasympathetic-vagal activity was related to higher experimental pain sensitivity as indicated by lower heat pain thresholds, higher pain ratings of suprathreshold stimuli and lesser pain adaptation. A previous study conducted in fibromyalgia patients, who do not sustain autonomic neuropathy, found, in accord with our results, that lower parasympathetic-vagal activity, as indicated by the low baroreflex sensitivity during rest and during a painful stimulus, was related to higher pain sensitivity to cold pressor test (Duschek et al., 2007; Reyes del Paso et al., 2011). In healthy subjects, Appelhans and Luecken (2008), however, found that lower resting LF power, which represents both sympathetic and parasympathetic activity, was associated with higher pain sensitivity, as reflected by lower pain thresholds and lower stimulation temperature required to induce pain at the intensity of 50 (pain-50). The HF parameter was surprisingly, unrelated to any of the pain measures (Appelhans and Luecken, 2008). This might stem from lack of normalization in the HRV analysis, and the fact that the data were gathered from healthy subjects. It might be that the range of variation in vagal activity in patients with neuropathy is greater than in healthy controls, allowing for significant correlations. Another study, in which unprocessed heart rate values were used as the autonomic parameter, showed that in men, higher increase in heart rate (increased sympathetic and decrease parasympathetic response) during painful stimulation of hot water immersion was associated with higher pain intensity (Tousignant-Laflamme et al., 2005). These studies highlight the relationships between autonomic activity and experimental pain, suggesting that for both healthy subjects and patients, the vagus nerve expresses an inhibitory effect on pain, such that loss of vagal function can lead to higher pain perception.
Notably, no correlations were found between the autonomic parameters and the pain modulatory mechanisms, as expressed by CPM response and temporal summation value. Temporal summation is the clinical expression of wind-up in which there is an increase in pain perception in response to repetitive peripheral noxious stimulation, which is probably expressing increased excitability and firing of dorsal horn neurons and possibly neurons at higher levels (Curatolo et al., 1997; Baranauskas and Nistri, 1998). The CPM paradigm examines the ‘diffuse noxious inhibitory controls’ phenomenon (Yarnitsky et al., 2010), which is a spino-bulbo-spinal loop (Le Bars et al., 1979). Imaging studies have demonstrated that many brain regions affect CPM, including the thalamus, primary and secondary somatosensory cortex, anterior and middle cingulate cortex, supplementary motor area, insula, prefrontal cortex and amygdala (Song et al., 2006; Piché et al., 2009; Sprenger et al., 2011). Thus, it might be that vagal influence is diluted by many other influences and was not detectable in our sample.
A limitation of this cross-sectional study is the aetiological heterogeneity of our patients. However, the physiological neural interactions in the neuropathy situation should not be affected by the type of chemotherapy that induced the neuropathy and should not interfere with drawing conclusions from the results. In the current study, there were more female patients in our sample. Although there are controversial results regarding the role of gender in pain perception (Racine et al., 2012a, b), we cannot rule out such a difference that can affect our results.
In conclusion, our findings highlight the influences of autonomic function on pain sensitivity. This was demonstrated in our clinical sample for experimental pain, showing that diminished vagal function in patients is associated with increased experimental pain perception. As one among many pathophysiological factors, this effect was not found for spontaneous pain levels, suggesting that the vagal effect on spontaneous pain is less dominant.
H.N-A., Y.G., M.S, T.T-S, D.P and D.Y. conceived and designed the experiments. H.N-A. performed the experiments. H.N-A and E.S. analysed the data. H.N-A and D.Y. wrote the manuscript. Y.G., D.P., E.S., M.S and T.T-S. are responsible for critical revision of manuscript. All authors discussed the results and commented on the manuscript.