Nociceptive two-point discrimination acuity and body representation failure in polyneuropathy

Subjects and study design

The study followed the Declaration of Helsinki and was approved by the Ethics Committee of the Rhineland Palatinate Medical Association Mainz, Germany (ID: 837.224.15(9962)). Informed written consent from patients and HC was obtained prior to the experiments. 20 patients (mean age 57.6 ± 13.9 years, 11 men), with painful and non-painful polyneuropathy, were enrolled after examination at the specialized neurological outpatient clinic of the Department of Neurology at the University Medical Center Mainz Germany. Inclusion criteria were age over 18 and fulfilling the diagnosis of polyneuropathy or sfPNP confirmed by experienced neurologists following the German guidelines for diagnosis of PNP (https://www.awmf.org/leitlinien/detail/ll/030-067.html) [11]. Exclusion criteria comprised tumor, infectious and autoimmune diseases. Diseases not related to PNP (e.g., mild hypertension, minor affective disorders, etc.) were recorded but not excluded. HC (n=20, nine men, mean age 50.6 ± 4.6 years) were enrolled if (i) age was between 45 and 65 years, (ii) no significant medical condition including pain was present and (iii) no intake of analgesics for at least one week.

Patients and HCs participated in two sessions separated by at least one week. Pain thresholds and two-point discrimination thresholds (2ptDT) for mechanical punctate stimuli (calibrated pinpricks) were assessed in the first session and 2ptDT for laser-evoked heat stimuli in the second session. Laser heat stimuli were used because standard thermal QST is not suited for the determination for 2ptDT. Prior to each test three training trials were performed. Measurements were processed bilaterally at four body areas: the dorsal side of hands, the volar forearms, at lateral calves and at the dorsal feet. The areas were chosen to spaciously map sensory deficits, since polyneuropathy is usually characterized by a proximal to distant gradient of neuropathic symptoms. The succession of the measurements at the different body areas and sides was randomized for each participant.

Pain thresholds

Two-point discrimination threshold for nociceptive pinprick (2ptDT_pinprick) and laser-evoked heat stimuli (2ptDT_laser)

Stimuli were delivered successively to determine the 2ptDT for both modalities. Assessment started from the maximum distance of 5 cm for hands and feet and 15 cm for arms and legs. With each following step, the distance between the two stimuli was decreased by 1 cm from proximally and distally (2 cm in total). For this purpose, a ruler with axis ticks every 1 cm was drawn on the respective body regions with an erasable pen. In the center, the axis ticks were fine graded (Figure 1) at each position, stimuli were first applied and then participants were asked to judge which stimulus was more proximal (first or second). The order of proximal or distal application of the first stimulus was randomized between subjects. Once the first incorrect judgement was given, a staircase design with a reduced step length of 0.2 cm distance was used to finally define the 2ptDT. A fixed stimulus intensity of 256 mN was used for pinprick, as this intensity was proposed to be clearly nociceptive [7]. The test was finished if the position of the second stimulus was correctly judged three consecutive times, or after a maximum of 22 trials. For analysis, the geometrical mean of all distances after the participant’s first incorrect response was calculated as the 2ptDT (cm). The 2ptDT for laser-evoked heat stimuli was assessed in the same manner, using the individual stimulus intensity determined during the pain threshold testing, i.e. the mean pulse intensity rated as pain of 2–4.

Figure 1:

Arm with marks drawn for the testing area of the pain threshold measurement (blue circle) and a scale for the two-point discrimination measurement (red line). Assessment started from the maximum distance of 5 cm for hands and feet and 15 cm for arms and legs. With each following step, the distance between the two stimuli was decreased by 1 cm from proximally and distally (2 cm in total). At each position, stimuli were first applied and then participants were asked to judge which stimulus was more proximal (first or second). Once the first incorrect judgement was given, a staircase design with a reduced step length of 0.2 cm distance was used to finally define the 2ptDT.

Assessment of body perception and limb representation disturbances

PNP patients often report swollen feet although objectively no changes are found. The Bath CRPS Body Perception Disturbance Scale (BBPDS) was adapted for our patients’ group [14, 15]. The scale was originally developed to capture pain-related changes of body perception of the affected limb of Complex Regional Pain Syndrome CRPS. Instead of asking about the one affected extremity, the questions referred to the feet. The questionnaire consists of seven items resulting in a maximum score of 57 points, with a higher score indicating a more pronounced body perception disturbance. Items 1–5 ask for patients’ ratings of body perception abnormalities on a 0 to 10 scale. Item six asks for a “desire to amputate”, and item seven includes patients’ descriptions of their whole body, that is simultaneously drawn by the investigator. The drawings finally had to be confirmed by the patient. Body representation disturbances are quantified by investigator’s rating on a 3-point scale (no distortion, distortion or severe distortion). For the present study, the BBPDS was translated into German following a forward-backward procedure. The patients were told to refer to both feet when answering the questions.

Additionally, the participants completed a computerized task, assessing body perception and representation distortion of lower extremities to evaluate in what way the perception of the feet differs between the groups. We focused on the feet because PNP commonly affects the lower limbs. Two sequences of four pictures were shown to the patients on a computer screen. The pictures showed feet distorted in length or width: −10, 0, +10, +20% distortion was applied and presented in a randomized order (supplementary Figure 1). Patients were asked which picture represented best the own perception of their feet.

Data analysis

Analyses were conducted using the SPSS software (version 23 for Windows, IBM SPSS Statistics, Chicago, USA). Statistical significance was considered at α-level of 0.05 (Bonferroni corrected α-level for multiple comparisons as indicated). Age-related differences between groups were tested by one-way analysis of variance. MPT, LHPT, and 2ptDT for pinprick and laser-evoked heat pain were log-transformed to approximate normal distribution. MPT thresholds were pre-processed according to the recommendations of the DFNS, i.e. log-transformation and calculation of z-values based on publicly available age-, sex- and body area-matched reference values [3, 16]. Body side (left vs. right) differences were assessed by analysis of variance for repeated measures (rm-ANOVA) for each parameter (MPT, LHPT and 2ptDT_pinprick and 2ptDT_laser laser-evoked heat pain) and body area, separately. Since there were no left-right sided differences, the values for the left and right body side were pooled for each of the four parameters.

Stimulus type–associated differences (punctate stimuli vs. laser-evoked heat) in 2ptDT acuity were tested using rm-ANOVA with “stimulus modality” as within-subject factor in each group. Body site effects were then explored by rm-ANOVA with “body area” (hand vs. arm vs. calf vs. foot), as within-subject factor and post-hoc Bonferroni corrected pairwise comparison. Differences in the 2ptDT acuity for nociceptive pinprick vs. laser-evoked heat stimuli were analyzed using one-way ANOVA for each group separately. Group comparison was subsequently performed for each parameter and body area by non-parametric statistics (Mann–Whitney U-test). This comparison was necessary due to the unequal group sizes that resulted from missing data, on the basis of dropouts, because of hyposensitivity to either the pinprick or laser heat stimuli which prevents the calculation of thresholds or 2-point discrimination.

Results were further explored by classifying the patients into predominantly painful and no-painful PNP groups. The classification was based on a cut-off value of pain rating of 3 (NRS 0–10) according to the findings of Ferrar et al., suggesting that relevant clinical changes on the numerical rating scale are indicated by differences of at least two scale units [17]. Comparisons were performed by pairwise analysis with Mann–Whitney U-tests. For association analysis among body perception, representation variables, thresholds and the 2ptDT thresholds, Spearman’s rho correlation coefficients were used.

Finally, for defining a classification (diagnostic) algorithm between groups, a stepwise discriminant analysis with cross-validation was calculated. We used the method of minimising Wilk’s lambda. The dependent variables were: MPT, LHPT, 2ptDT_pinprick and 2ptDT_laser from all body areas for classification of PNP vs. HC. Due to missing values, 32/40 subjects could be included. The missing values came from LHPT and 2ptDTlaser dropouts at the feet because heat pain was not felt at the upper safety limit of 600 mJ in these patients. For classification of painful vs. painless PNP MPT, LHPT, 2ptDT pinprick and 2ptDTlaser, bath body perception score and foot distortion were included. For this analysis, the pinprick and laser results at the feet were not included in order to get a reasonable number of patients with complete datasets for the analysis. 19/20 patients could be included in this way.

Study population and demographics

The mean age was different between groups; patients were 7 years older than controls (patients vs. controls: 57.6 vs. 50.6 years; F(1,0)=4.55, p=0.039). The sex was matched (n=11, n=9 men in the patient and control groups, respectively; χ²=0.40, p=0.53). Clinical characteristics of the patients are displayed in supplementary Figure 2. Patients were mildly affected and reported sensory symptoms, limb weakness was not reported. Sensory polyneuropathy was idiopathic in most cases (n=13 cases), seven patients had pure small fiber neuropathy. The mean current pain was 2.60 ± 2.44 (min/max: 0/8) on an 11-step numeric rating scale (NRS). For further investigations patients were divided into painful and no-painful PNP groups, based on a cut-off value of pain NRS rating of 3 (See section above). 10/20 patients could be classified as painful PNP.

Body area-related differences in HCs and PNP patients

Left-right differences did not surpass a Bonferroni adjusted threshold, which is shown in supplementary Figure 3. Left and right values were therefore averaged for each parameter and body area. The resulting mean values were used for further analysis. All four parameters (MPT, LHPT, 2ptDT_pinprick, 2ptDT_laser) differed significantly between body areas in both groups (see Table 1 for statistical details).

In HCs, post-hoc pairwise comparisons of different modalities revealed that MPT was highest at the arms, and LHPT was highest at the feet. 2ptDT was highest at the calves for both modalities. In PNP-patients, post-hoc pairwise comparisons revealed a pattern that was different from HCs: MPT and LHPT were highest at the feet while 2ptDT (for both modalities) was highest at the calves. Details and statistical comparisons are shown in Figure 2.

Figure 2:

Pain thresholds and two-point discrimination thresholds at different body areas. For visualization of distributions the raw values are displayed; because of non-normality of the distributions, the log-transformed values were submitted for analysis. In HCs, post-hoc pairwise comparisons of different modalities revealed that MPT was highest at the arms and LHPT was highest at the feet. 2ptDT was worst at the calves for both modalities. In PNP-patients, post-hoc pairwise comparisons revealed a pattern that was different from HCs: MPT and LHPT were worst at the feet while 2ptDT (for both modalities) was worst at the calves. (Notes: Bonferroni corrected significant differences marked with *; MPT, mechanical pain threshold; LHPT, laser-evoked heat pain threshold; 2ptDT_pinprick, two-point discrimination threshold for nociceptive pinprick stimuli; 2ptDT_Laser, two-point discrimination threshold for nociceptive laser-evoked heat pain stimuli).

Acuity for 2ptDT_pinprick Vs. 2ptDT_laser

In HCs, comparisons revealed that 2ptDT_pinprick was smaller than 2ptDT_laser at all body sites (F(1,0)=[7.11–29.59], p<0.02). In the PNP group, 2ptDT_pinprick was only smaller at the arm (F(1,0)=9.19, p=0.008).

Differences in two-point discrimination between HCs and PNP patients

MPT, LHPT and 2ptDT_laser were not different between PNP patients and HCs, but PNP patients had greater 2ptDT_pinprick at hands (p=0.001), arms (p=0.009) and feet (p=0.001) than HCs. After Bonferroni correction only the 2ptDT_pinprick at hands met significance threshold (U=76.50, z=−3.34, p=0.001, corrected p<0.003 for 16 tests; mean 2ptDT_pinprick PNP patients 1.06 ± 1.05; HCs 0.48 ± 0.19). See Table 2 and Figure 3 for details.

Figure 3:

Comparison of the distribution of 2ptDT_pinprick stimuli between groups. For visualization of distributions, the raw values are displayed; because of non-normality of the distributions, the log-transformed values were submitted for analysis. MPT, LHPT and 2ptDT_laser were not different between PNP patients and HCs (data not shown), but PNP patients had greater 2ptDT_pinprick at hands (p=0.001), arms (p=0.009) and feet (p=0.001) than HCs. After Bonferroni correction only the 2ptDT_pinprick at hands met significance threshold (U=76.50, z=−3.34, p=0.001, corrected p<0.003 for 16 tests; mean 2ptDT_pinprick PNP patients 1.06 ± 1.05; HCs 0.48 ± 0.19). (Notes: *, significant at p<0.003 (Bonferroni corrected); MPT, mechanical pain threshold; LHPT, laser-evoked heat pain threshold; 2ptDT_pinprick, two-point discrimination threshold for nociceptive pinprick stimuli; 2ptDT_Laser, two-point discrimination threshold for nociceptive laser-evoked heat pain stimuli).

When dividing the PNP groups further in painful (n=10) and painless (n=10) PNP (cut-off value of pain rating of 3), Mann–Whitney U-tests revealed no differences between the painful and the non-painful PNP groups in any comparison (MPT, LHPT and 2ptDT of both modalities).

Body perception and representation

Bath CRPS body perception disturbance scale (BBPDS)

19 patients completed the BBPDS. Average sum score was 21.2 ± 6.9 (min/max: 12/32). The distribution of single items, the sum score, and the percentage of positively scored items, are displayed in supplementary Figure 4. Figure 4 shows examples of the individual drawings of mental body representation (BBPDS Item 7). Notes of the patients to the corresponding body party were included in german.

Figure 4:

Item seven of the BBPDS: Individual drawings (done by the investigator) of the mental body representation of patients number 8 and 17. Percentage values always indicate an enlargement of the body part as reported by the patient. (Notes: BBPDS, bath CRPS body perception disturbance Scale).

Discriminant analysis for prediction of group classification

In three steps (Wilk’s lambda=0.46), discriminant analysis (n=32) selected 2ptDT_pinprick at hands, 2ptDT_laser at feet and HPT at feet as the most predictive variables for correct classification of PNP vs. HC. Fisher’s linear discriminant function was (4.829 × 2ptDT_pinprick hands) + (5.804 × 2ptDT_laser feet) − (3,879 × HPT feet) − 11.653. All results above zero were predictive for the presence of PNP vs. HC. Employing this discrimination function, 85% of patients could be correctly classified (see supplementary Table 1).

In five steps (Wilk’s lambda=0.25), discriminant analysis (n=19) selected 2ptDT_pinprick at hands, MPT at calves and foot distortion as the most predictive variables for correct classification of PNP with vs. PNP without pain. Fisher’s linear discriminant function was (5.87 × 2ptDT_pinprick hands) − (4.562 × MPT calves) + (0.114 × foot length distortion) − 6.731. All results above zero were predictive for the presence of pain. Employing this discrimination function, 90% of patients could be correctly classified (see supplementary Table 2).

Why to investigate 2ptDT?

Evaluation of pinprick and laser heat pain thresholds revealed no differences between patients and HCs, neither at the upper nor the lower extremities. The reason is that the perception of painful skin stimuli in PNP depends on fiber loss but also on fiber sensitization, which might cancel each other out. This has already been studied several times for heat pain thresholds [2, 18], [19], [20], and our study shows that this statement is also valid for mechanical pain thresholds. This means testing pain thresholds does not make additional contributions to PNP diagnosis. However, we found that the pain 2ptD thresholds varied across body regions. We observed better 2ptD ability in the distal hairy skin of HCs, both on the arms and legs. The better resolution of pinprick stimuli might depend on the more focused stimulus, due to a smaller area of stimulation or on the type of activated afferent fibers (e.g., Aß/Aδ fibers vs. Aδ/C fibers) which convey the action potentials to finely graded somatotopic representations in the CNS [21, 22]. For heat pain, such a focused perception does not make sense in evolution. Thus, 2ptDT might report another quality of sensory perception, which not simply depends on innervation of the skin and could therefore be potentially useful for PNP diagnosis. Supporting this assumption, measuring 2ptD adds to the differentiation of PNP patients from controls. 2ptD_pinprick in PNP was significantly less accurate on the hands compared to controls. That was not the case on the lower limbs where PNP symptoms were present. We described a similar finding in previous studies. Neurogenic axon reflex flare testing showed that the proximal thigh is much more sensitive place than the feet to differentiate between PNP patients and HCs [23]. This means that despite neuropathic symptoms usually being length dependent and predominant at the feet, patients show generalized changes of dermal innervation [24]. Body areas that have a good sensory function in HCs, may be therefore better places to test for subtle changes than the symptomatic body area. Moreover, the results of the discriminant analyses indicate that testing unaffected and affected body regions with different modalities might be most conclusive for identification of mild neuropathy.

Body perception

Chronic pain and nerve damage can lead to distorted body perception, which are explained by changes in cortical representation of the affected body area [25], [26], [27], [28]. The present study shows a high prevalence of distorted body perception in PNP patients. Patients selected more distorted pictures, indicating that patients perceive and judge their feet larger than they are. Additionally, patients’ self-reported body perception disturbances in the BPPDS were highly prevalent in PNP. The BPPDS was initially developed for CRPS, a chronic pain disorder which is also characterized by highly distorted body image [29, 30], and impaired judgment about the length of the affected hand [31]. Remarkably, the BPPDS sum scores were even higher in the PNP patients than in a historic CRPS patient cohort [15]. We did not perform brain magnetic resonance imaging but it is likely that the above-mentioned sensory cortex reorganization, which should be the result of pain and nerve damage, underlies body perception disturbances in our patients as in other cohorts [32, 33]. We did not find a direct linear correlation between body distortion and pain thresholds or pain. There was only a subtle correlation of the perception of the foot length with 2ptD_pinprick at the arms, which seems peculiar at first glance, but as discussed above, the upper extremities may be a more sensitive place to test for pathology in PNP. These assumptions are supported by the results of the multivariable analysis. 2ptDT_pinprick at hands, MPT at calves and foot distortion predict neuropathic pain correctly in 90%. Although this has to be confirmed in a larger and independent cohort, and the following conclusion might be speculative, our result may confirm the complex relation between nerve fiber loss, cortical representation, pain and body perception of the affected feet in PNP. The complexity of this interaction explains why linear bivariate correlations are usually inconclusive.

Our study has limitations. First, the small sample size limits the generalizability of our findings. Second, we did not compare nociceptive with non-nociceptive 2ptD, which prevents judgement of the utility of these two modalities for diagnosis. Third, nor did we assess the limb function in detail. However, our patients were only mildly affected so that function scores usually are in the normal range. Forth, while all patients had a polyneuropathy based on routine clinical examinations including skin biopsy, the assessment was stopped if the diagnosis was confirmed. This means that we could not correlate e.g., warm or cold perception as surrogates for skin innervation with Aδ and C fibers with 2ptD. Finally, the patients were treated for their neuropathic pain. This might have affected pain, 2ptD and body perception.