Electrically Stimulated Axon Reflexes Are Diminished in Diabetic Small Fiber Neuropathies

Twelve patients (5 women and 7 men, mean age 46 ± 3.3 years) suffering from diabetic neuropathy were included in this study. The inclusion criterion was a clinically suspected small fiber impairment (SFN) because all patients suffered from neuropathic pain symptoms, such as burning sensation, lancinating, or spontaneous pain in the feet but not in the thighs. Among these patients were five patients with type 1 diabetes (two men and three women) and seven patients with type 2 diabetes (five men and two women). The type of neuropathic pain resulting from SFN was assessed by the medical history and neuropathy impairment score (18) and was recorded as burning, lancinating, or spontaneous pain.

Moreover, we recruited five diabetic subjects (four women and one man, mean age 43.6 ± 10.8 years; three patients with type 1 and two patients with type 2 diabetes) without clinical and electrophysiological evidence of neuropathy (NO-N). All patients underwent standard electrophysiological studies, such as the measurement of nerve conduction velocity of the peroneal and sural nerve (Table 1).

As a control group, 14 healthy subjects (8 women and 6 men, mean age 36.5 ± 2.3 years) without any neurological or dermatological disorders were included. In five subjects (two women and three men, mean age 41.4 ± 6 years) of this control group, reproducibility of the investigation was evaluated by retests performed with an interval of 12 months. All healthy volunteers were drawn from staff and relatives of the Department of Physiology, University of Erlangen-Nuremberg. All investigations were carried out in a temperature (23°C) and humidity (50% relative humidity) controlled laboratory. The time for acclimatization for all subjects was at least 1 h before starting the experiment.

Informed consent according to the Declaration of Helsinki was obtained from all participants and the study was approved by the local ethics committee.

Electrical stimulation was performed in the center of the right dorsum of the foot and in the center of the right ventral thighs 20 cm above knee level. Two microdialysis fibers (DermalDialysis, Erlangen, Germany) were inserted intradermally for a length of 1.5 cm by a 25-gauge cannula in every test area. This technique ensured that the fibers were located at a depth of ∼0.6 mm intradermally (16). The fibers were placed transversally to the axis of the limb for a distance of 3 mm. They were equipped with internal stainless steel wires (0.1 mm in diameter) for electrical stimulation. Each fiber was perfused with 0.9% NaCl solution by a microdialysis pump (Pump 22; Harvard Apparatus) at a constant flow rate of 4 μl/min.

After a baseline period of 60 min, when insertion-related vasodilatation had subsided (19), electrical stimulation was started. Electrical pulses (1Hz, 0.5 ms stimulus duration) were continuously delivered for 21 min via a constant current stimulator (DS7; Digitimer, Hertfordshire, U.K.). Electrical stimulation commenced with a current of 6 mA. Current was stepwise increased thereafter to 8, 10, 12.5, 15, 17.5, and 20 mA every 3 min.

In previous studies, this setting and the frequency of 1 Hz has proven to be most effective in provoking axon reflex vasodilation in human skin (17). This technique guarantees the induction of a flare being constant for even more than 1 h (20). With these parameters, electrical stimulation was well tolerated and no lasting side effects were observed.

Superficial blood flow was quantified using an LDI (Moor, London, U.K.). LDI scans (256 × 256 pixels, scan resolution 4 pixels/s, distance 50 cm to site of electrical stimulation) were recorded at baseline and at intervals every 3 min during the electrical stimulation (total of eight scans). The LDI scans were started 1 min after changing the current and took ∼115 s. The size of the scanned area was 144 cm², with the stimulation site in the center. Blood flux was calculated for each pixel by means of intensity of the Doppler shift of the backscattered laser light (arbitrary perfusion units [PUs]). Thereby, a two-dimensional map of the superficial blood flow is generated. Area and intensity of the neurogenic flare reaction were analyzed offline by dedicated software (MLDI 3.0; Moor, London, U.K.). Flare area was determined by total number of pixels (0.22 mm²/pixel) in which flux values exceeded the mean flux by two SDs from the baseline picture. Total flare areas >0.5 cm² on the feet and >1 cm² on the thighs were regarded as significant. The minimum current needed to induce a flare reaction according to this definition was noted as electrical threshold.

Patients and subjects were asked to quantify pain sensation during electrical stimulation by a numeric rating scale from 0 to 10, in which the value of 0 indicated “no pain” and 10 “maximum pain.” Pain ratings were obtained every minute during stimulation.

Statistics were calculated using an SPSS (SPSS 10.1 for Windows, Chicago, IL) software package. To identify significant differences between patient groups and control subjects, ANCOVA for repeated measures was performed. The ANCOVA incorporated seven consecutive measurements (during stimulation), one main factor (SFN subject or NO-N and control subjects), and age and sex as covariates. Since sphericity could not be assumed in all datasets, Greenhouse-Geisser correction was employed. Post hoc analysis was performed in order to allocate differences at the different stimulus intensities between SFNs or NO-Ns and control subjects. We used planned comparisons with t tests using Bonferroni’s correction.

Analyzing the test/retest reliability, we calculated the mean flare size and mean flux values from all stimulation intensities at thighs and feet and thereafter the Pearson correlation coefficient. Additionally, an ANOVA (seven consecutive measures and one main factor [test/retest]) was calculated. Covariate adjustment was not necessary in this analysis because both groups were identical.

Pain ratings in different subgroups were analyzed using Student’s t test for independent samples. To compare the current thresholds necessary to induce a flare, the Mann-Whitney U test was calculated. This nonparametric test was used because current intensities were ordinal scaled. Accordingly, the median current intensities and the ranges are given.

All further values are means ± SE. Statistical significance was considered at P < 0.05 or was corrected for multiple (i.e., seven) post hoc comparisons at P < 0.007.

As expected, the type 1 diabetic subjects included in our study were younger than type 2 diabetic subjects (mean age 35.2 ± 2.5 vs. 53.7 ± 2.7 years). The average duration of the disease was longer in type 1 diabetic subjects (15 ± 2.1 vs. 8 ± 1.4 years). Diabetes was well controlled in all patients at the time of the experiment. SFN patients showed neurophysiologic abnormalities proving large fiber neuropathy (LFN), but the extent of neuropathy showed significant variability. Patients with advanced LFN (absent plantar reflexes, highly pathological nerve conduction studies, and high neuropathy impairment score), patients with rather discrete LFN (normal plantar reflex, almost normal nerve conduction studies, and low neuropathy impairment score), and patients with intermediate symptoms were found in type 1 and 2 diabetic subjects. All patients showed neuropathic pain symptoms such as burning, lancinating, or spontaneous pain restricted in the feet, but not in the thighs, suggesting small fiber damage (Table 1).

Electrical stimulation provoked an intensity-dependent neurogenic flare. Thresholds for induction of flares were 8 mA (6–12.5 mA) in the thighs and 10 mA (6–20 mA) in the feet. Flare sizes were larger in women (feet: women 2.82 ± 0.76 cm² and men 1 ± 0.13 cm², P < 0.03; thighs: women 7.8 ± 1.25 cm² and men 4.2 ± 0.73 cm², P < 0.03) and in the thighs (6.5 ± 0.97 cm²) compared with the feet (1.9 ± 0.48 cm², P < 0.01) in control subjects. Flux values (flare intensity) were neither sex dependent nor was there a significant difference between thighs and feet (feet: women 291 ± 26 PU and men 215 ± 24 PU; thighs: women 266 ± 16 PU and men 239 ± 27 PU). Flare sizes at feet and thighs were age dependent (ANCOVA, P < 0.03). There was, however, no age dependence of flare intensity (flux).

Electrical stimulation elicited neurogenic flare in all patients (SFN subjects and NO-Ns). In SFN subjects, higher intensity of electrical stimulation was required to induce a flare reaction. Median threshold in the SFN group was 17.5 mA in both the thighs (8–20 mA) and feet (10–20 mA), and the differences compared with control subjects were significant (P < 0.001). In the NO-Ns, flare commenced at 10 mA (6–15 mA) in the feet and 8 mA (6–12.5 mA) in the thighs. This was not significantly different from the control group.

In the dorsal side of the feet, flare size was significantly reduced in SFN subjects (ANCOVA, P < 0.01). Post hoc analysis revealed significant differences with current intensities ≥15 mA (Fig. 1A). In the thighs, flare size was also significantly smaller in SFN patients (ANCOVA, P < 0.001). Post hoc analysis revealed significant differences at all stimulation intensities (Fig. 1B). For both stimulation sites, the differences between patients and control subjects became more obvious with increasing current. The flare intensity (flux) was not different between SFN subjects and control subjects in the feet or in the thighs.

Comparison between NO-Ns and control subjects revealed that flare size in the thighs was not different between patients and control subjects (ANCOVA) (Fig. 2B). However, flare size in the feet was even enlarged in NO-Ns (ANCOVA, P < 0.02). Post hoc analysis was not able to allocate significant differences (Fig. 2A). The flare intensity (flux) in the thighs (ANCOVA, P < 0.001) and in the feet (ANCOVA, P < 0.003) was increased in the NO-N group compared with the control group. Post hoc analysis was again not able to allocate significant differences with one exception, flux in the thighs at 6 mA (Fig. 1C and D).

Electrical stimulation was well tolerated and moderately painful. The pain rating raised with increasing current intensity. The pain ratings were slightly lower in the thighs (control subjects numeric rating scale 3.2 ± 0.5, SFN subjects 3.3 ± 0.5, and NO-Ns 2.4 ± 0.9) than in the feet (control subjects 3.5 ± 0.7, SFN subjects 3.9 ± 0.5, and NO-N 3.2 ± 0.8) but did not differ significantly between the different groups.

There was no difference between test and retest values regarding size and flux of the flare response (ANOVA, NS). Despite the limited number of subjects, we found a significant correlation between test and retest of the mean flare size (see research design and methods) (r = 0.83, P < 0.004). Separate calculation revealed a significant correlation at the feet (r = 0.91, P < 0.05). With greater flare sizes at the thighs, values were also positively correlated but without reaching significance (r = 0.74, P = 0.15) (Fig. 2). Flux values at both feet and thighs were not significantly correlated.

In the present study, we could demonstrate that the size but not the intensity of vasodilation of electrically induced axon reflex flare indicates C-fiber damage in patients with diabetic SFN. Reduction of flare size was not confined to the symptomatic feet, but was even more pronounced in the clinically unaffected thighs. Furthermore, our method not only detects C-fiber impairment but also detects increased flare size and intensity in NO-Ns, indicating increased release of neuropeptides, provided that peripheral C-fibers are intact.

Histamine flare, for example, is larger than an electrically-provoked flare (14), although histamine activates only a small proportion of afferent C-fibers. This is due to the huge receptive fields of histamine-sensitive C-fibers, which are a subgroup of mechanoinsensitive C-fibers. These fibers determine the axon reflex flare and have high electrical thresholds (21,22). If we applied stimulus intensities of ∼80 mA (20), most of these mechanoinsensitive fibers would be excited (23). In this case, the flare would become at least as large as after application of capsaicin (24). However, such high stimuli intensities cannot be used for routine clinical testing.

For clinical use, a standardized C-fiber stimulation protocol is essential. There are limitations of chemical stimulation (7). For instance, there is nonneurogenic vasodi-lation with acetylcholine or limited reproducibility with histamine due to seasonal variability (15). The electrical C-fiber stimulation has the advantage of intensity and frequency of current that can be well controlled. This standardized stimulation allows the detection of age dependence of axon reflexes, as it has been demonstrated for other neurophysiological parameters of peripheral nerve function.

The axon reflex flare further depends on the neurosecretory function of the terminals of activated primary afferent neurons (8). Our study confirms that flare sizes are larger in thighs compared with feet (14). As microneurographic data have shown, innervation territories of C-fiber neurons vary between body regions; mechanosensitive C-fibers show larger innervation territories in proximal body areas (25). The same may be the case for mechanoinsensitive C-fibers, although a previous study was unable to identify differences between lower legs and feet (21). On the other hand, the number of epidermal C-fiber terminals in skin punch biopsies increases from distal to proximal body regions (26,27). This first glance contradiction may be explained by a higher number of nerve terminals branching off from a limited number of C-fiber neurons at proximal body regions (21,25). Alternatively, function of nerve fibers in distal body regions also might decline earlier under physiological conditions, as suggested by the age-related and length-dependent decrease of nerve conduction in healthy subjects (28).

Pain ratings did not differ between control subjects and SFN patients, suggesting that there was no major impairment of the central connections of afferent fibers. Different fiber classes contribute to electrically-induced pain because pain was described as short and stinging (Aδ-fibers) with a delayed burning component (C-fibers) (29). In contrast to these proximally directed pain pathways, the distal efferent function of C-fibers was impaired not only in distal limbs but also in proximal skin areas. We plotted intensity-response relations and found that the thresholds for flare reaction were higher within the SFN group. A higher electrical threshold could be the result of different mechanisms; reduced epidermal nerve fiber density or increased electrical thresholds of single axons.

Nerve fiber reduction in diabetic skin leads to smaller axon reflex flares because fewer fibers are excited due to a statistically larger distance to the stimulation electrode and the reduced release of neuropeptides from the sparse axonal tree. If these neuropeptides (e.g., calcitonin gene- related peptide) remain below a critical level at the flare borders, the flare size shrinks (22). The overlap of innervation territories within the central flare makes flare intensity rather unchanged in SFN because there is a steep dose-response curve of calcitonin gene-related peptide-induced vasodilation.

In addition, diabetic neuropathy also directly increases electrical excitability by changes in ion channels distribution (calcium-activated potassium channels [30], and voltage-gated, Tetrodotoxin-resistant sodium channels [31]). Diabetic rat axons show a reduced inward rectification, hyperpolarization, and slower conduction velocity (32). In addition, there is evidence for smaller axonal diameters of unmyelinated fibers (33,34) in nerve biopsies from diabetic neuropathy patients. Smaller axonal diameters will further increase electrical thresholds. Interestingly, reduced C-fiber conduction velocities and increased electrical thresholds were recently demonstrated by microneurographic single fiber recordings in chronic pain states (35). Thus, our observation that electrical stimulation of higher intensity is required to provoke axon reflex flare in diabetic SFN patients might be a combined effect of axonal loss, changes of ion channel properties, and reduced axonal diameters. Irrespective of the relative contribution of single mechanisms, increased electrical thresholds of C-fibers will facilitate the differentiation between patients and control subjects.

Nonneuropathic impairment of skin perfusion, which can regularly be observed in diabetes, theoretically may also affect axon reflex vasodilation. However, reduction of skin perfusion will mainly reduce the intensity of the axon reflex flare, whereas flare size is based on the intact innervation causing the flare’s low vulnerability to vascular impairment (36,37).

Our results were very different in NO-Ns. Figure 1 overestimates the differences, since the majority of our NO-Ns were women and young. However, statistical analysis was adjusted for sex and age and the differences remained. Enhanced neuropeptide release from C-fibers in diabetes has been demonstrated in recent rat experiments (38). Our results may not be caused by increased excitability of diabetic C-fibers because the current thresholds for induction of the flare were unchanged and pain ratings were not different from control subjects. That is, the neuropeptide content of diabetic C-fibers must be increased. There is evidence that nerve growth factor (NGF) is upregulated in diabetes, possibly in response to minimal C-fiber damage (39). NGF, in turn, upregulates neuropetides in intact axons, whereas depletion of NGF in damaged axons leads to neuropeptide loss (40,41). We found no clinical evidence for C-fiber impairment in NO-Ns, and therefore, we assume that NGF-induced increase of neuropeptides in intact axons overrides a possible decrease of neuropetides in minimally lesioned axons. In SFN, however, axonal damage or depletion of neuropeptides predominates.

The noninvasive and painless quantitative sensory testing is a routine test for small fiber dysfunction. There is considerable variability of quantitative sensory testing results in neuropathy patients (42), and the data depend on active cooperation (43). Since the spatial extension of flare might be a surrogate of morphologic and functional changes of peripheral C-fibers and its independence of patients’ cooperation, it should be less variable than quantitative sensory testing.

An alternative tool to assess skin fibers is skin biopsies. They provide histological data and can be regarded as the gold standard. However, skin biopsies are more invasive than electrical stimulation, and quantitative analysis is time consuming (44). In clinically advanced diabetic neuropathy, skin biopsies show dermal fiber loss. This fiber loss was not restricted to symptomatic skin area (45). That is, skin biopsies corroborate our functional results that clinically nonaffected skin areas already show structural and functional impairment of thin nerve fibers (27). On the other hand, in patients with very early neuropathy and impaired glucose tolerance (45), skin biopsies demonstrated slightly reduced nerve fiber density but there was simultaneously an increased distal fiber branching, which counteracts fiber loss (4,46). The trigger for distal sprouting of primary afferent fibers may be NGF, which might also contribute to increased neuropeptide release and flare intensity in our NO-Ns, as discussed above. These striking analogies of histological data and the results presented herein should be further investigated in a combined study comparing epidermal fiber density and axon reflex flare in the same patients.

In conclusion, our results indicate that analysis of electrically-induced flare size might be a valuable tool for assessment of small fiber dysfunction in peripheral neuropathies. It may even be helpful in other pain disorders with C-fiber pathology, such as erythromelalgia or postherpetic neuralgia. Our method is sensitive enough to detect impairment of efferent neurosecretory C-fiber function earlier than impairment of afferent sensory function. It remains to be evaluated whether the analysis of electrical thresholds provides an additional measure for possible neuropathic changes.

This work was supported by the Deutsche Forschungsgemeinschaft, SFB 353, C3, and Bi 579-1 and the MAIFOR Program of the University of Mainz.

We thank B.B. Egbers and T. Hagerty for help with manuscript preparation.