Participants

To evaluate our hypothesis, pain sensitivity to punctate stimuli was tested unilaterally on the most painful body site (paraspinal lumbar; location of measurement: vertebra Th12 to L5) and on two non-painful body sites (hand palmar, hand dorsum; unilaterally on the dominant right hand) in CLBP patients. Healthy control subjects were matched on age and gender to the patients (see Table 1). Testing in healthy control subjects was performed at the same regions on the lower back and the dominant hand.

Stimulus response function (SRF) to punctate mechanical stimuli

In accordance with our hypothesis, we used a test for mechanical pain thresholds from the Quantitative sensory testing (QST) according to the standardized protocol of the German Research Network on Neuropathic Pain (DFNS) [26, 27], i.e., the stimulus response function for pinprick stimuli that also allows mechanical pain sensitivity (MPS), another parameter of QST, to be determined. Testing was performed by a DFNS-trained investigator.

Stimulus response functions (SRF) to punctate mechanical stimuli were analysed using standard pinprick stimulators (cylindrical tip, 250 μm tip diameter) with fixed stimulus intensities that exerted forces of 8, 16, 32, 64, 128, 256, and 512 mN (MRC Systems GmbH, Heidelberg, Germany). All seven pinprick stimuli were applied in a balanced order, five times each at every test site (back paraspinal lumbar, hand palmar, hand dorsum). SRF was assessed as the relationship between the applied forces of the standard pinprick stimulators and the pain rating evoked by each of the pinprick stimulators [28, 29], i.e., participants were asked to rate the experienced pain intensity after each stimulation on a verbal rating scale (with 0 indicating “no pain”, and 100 indicating “maximal imaginable pain”). To avoid effects of sensitization or fatigue, successive stimuli were not applied at the same spot of skin, but some millimetres away from the preceding stimulation site.

Mechanical pain sensitivity (MPS) was also determined. MPS was assessed as the geometric mean of the pain ratings evoked by each of the seven pinprick stimuli (similar to the standard protocol of DFNS).

In order to exclude a potential influence of skin temperature on the results of testing, skin temperature was assessed for all body sites (back paraspinal lumbar, hand palmar, hand dorsum) before and after the testing. There were no significant skin temperature changes for any of the body sites before vs. after the testing or between patients and controls.

Questionnaires

Depression was assessed using a German version [30] of the Beck Depression inventory (BDI) [31]. None of the participants suffered from clinically manifest depression (cut-off BDI > 17; Table 1).

Disability was measured using the German version of the 24-item Roland Morris disability questionnaire (RDQ) [32–34]. RDQ scores range from 0 (no disability) to 24 (maximum disability). Five of the CLBP patients showed clinically-relevant disability (3 < RDQ score < 8); 1 patient reported a high level of disability (RDQ > 7); none of the control subjects reported clinically-relevant disability (Table 1).

Data evaluation

The experienced pain intensities of each of the seven pinpricks (SRF) as well as MPS were normally distributed in log space, and thus were log₁₀-transformed before statistical analysis in accordance with the recommendations of DFNS [27]. For SRF and MPS, a small constant (0.1) was added prior to log-transformation to avoid a loss of values due to zero rating [35]. Data transformed to secondary normal distribution were analysed with analysis of variance (ANOVA) for repeated measurements using the within-subject factors Pinprick (7 intensities from 8 to 512 mN) and Region (hand dorsum vs. hand palmar vs. lower back) as well as the between-subject factor Group (CLBP vs. HC). Post-hoc tests were performed using separate ANOVAs. Results were corrected for violations of sphericity using the Greenhouse-Geisser approach for epsilon correction of degrees of freedom (when appropriate). All statistical calculations were performed using SPSS 19 Software.

MPS for each CLBP patient was compared with the group means of healthy controls using z-scores. Z-scores above ‘0’ indicate a gain of function referring to the higher sensitivity of the CLBP patient to the tested stimuli compared to the healthy controls. Z-scores below ‘0’ indicate a loss of function when the CLBP patient is less sensitive to the tested stimuli compared with the healthy controls. Z-values below −1.96 or above +1.96 were considered as abnormal for diagnostic purposes (95% confidence interval [33, 34]). A t-test (two-sided for independent samples) was performed for MPS using the internet-based statistical freeware Simple Interactive Statistical Analysis (SISA; URL: http://www.quantitativeskills.com/sisa/) separately for healthy controls and CLBP patients to compare data with the DFNS reference data as proposed recently [33].

Detailed analysis of SRF to pinprick stimuli

ANOVA on pain ratings to pinprick stimulation with factors Pinprick, Region, and Group showed significant main effects of factors Pinprick (F(6, 156) = 187.02; P < 0.001), Region (F(2, 52) = 11.57; P < 0.001), and Group (F(1, 26) = 7.912; P < 0.01). As expected, the main effect of factor Pinprick resulted from higher ratings to pinpricks with higher force (see Figure 1A-C). The main effect of factor Region resulted from overall higher pain ratings at the dorsum of the back compared with the hand (hand palmar vs. back: F(1, 26) = 16.62, P < 0.001; hand dorsum vs. back: F(1, 26) = 8.38, P < 0.01; hand palmar vs. hand dorsum: F(1, 26) = 5.99, P < 0.05). The main effect of Group is explained by overall higher pain ratings in the CLBP group as compared to HC (F(1, 26) = 7.91, P < 0.01, Cohen’s d = 1.44; see Figure 1A-C). Additionally, there were significant interactions between the factors Pinprick x Region (F(12, 312) = 16.53, ε = 1.63; P < 0.001) and Pinprick x Group (F(6, 156) = 9.07, ε = 3.28; P < 0.001). One-way ANOVAs for the interaction Pinprick x Group revealed significant differences in pain ratings between groups for the pinpricks 8 mN, 16 mN, 32 mN and 64 mN (all P < 0.05, corrected for multiple testing; all d > 1; range: 1.06 – 1.40). The perceived pain in CLBP patients was 3–5 times higher for low to moderate pinprick forces compared with healthy controls. The difference in pain rating between groups for the 128 mN pinprick was significant at an uncorrected level (P < 0.037, d = 0.83), whereas no difference in pain ratings between groups was observed for the 256 mN (P = 0.117, d = 0.61) and 512 mN pinpricks (P = 0.849; d = 0.08) (see Figure 1A-C). No significant interactions other than Pinprick x Group and Pinprick x Region were observed. In particular, there was no 3-way Pinprick x Region x Group interaction (F(12, 312) = 1.46, ε = 6.28; P = 0.19) indicating that the observed differences between groups and the interaction between Pinprick and Group were not different for hand and back.

Detailed analysis of MPS

ANOVA of MPS revealed main effects of factors Region (F(2, 52) = 11.63, P < 0.001) and Group (F(1, 26) = 7.90, P < 0.01), but no interaction between Region and Group (F(2, 52) = 1.93, P = 1.55. As Figure 2 shows, the main effect of factor Region resulted from significantly higher MPS for the back compared with the hand (back vs. hand palmar: F(1, 26) = 16.78, P < 0.001; back vs. hand dorsum: F(1, 26) = 8.41, P < 0.01; hand dorsum vs. hand palmar: F(1, 26) = 6.01, P < 0.05). The main effect of factor Group is explained by overall higher MPS in the CLBP group as compared to HC (F(1, 26) = 7.90, P < 0.01, d = 1.06; see Figure 2). MPS of the palmar site of the hand as well as the dorsum of the hand were highly correlated with the MPS at the back (z-score based correlations, hand palmar: R² = 0.78, P < 0.001; hand dorsum: R² = 0.83, P < 0.001). Furthermore, MPS of the palmar site and of the dorsum of the hand were also highly correlated (R² = 0.81, P < 0.001).

Incidence of reported pain to punctate stimuli

Figure 3 shows that the incidence of reported pain increased as a function of stimulus force on the back, hand palmar, and hand dorsum (Figure 3A-C, respectively). The population threshold, i.e., force at which 50% of pinprick stimuli were reported to be painful, was interpolated as 8 mN for the CLBP patients and 96 mN in healthy controls at the back. Additionally, there were significant differences between groups in the incidence of reported pain on the back for 32 mN (χ ² = 12.07; p < 0.05) and 64 mN (χ ² = 10.91; p < 0.05) (see Figure 3A). On the hand palmar, the population threshold was also interpolated as 8 mN for the CLBP patients and 96 mN in healthy controls. There were significant differences between groups in the incidence of reported pain at 32 mN (χ ² = 11.03; p < 0.05) and 64 mN (χ ² = 11.4; p < 0.05) on the hand palmar (see Figure 3B). For the dorsum of the hand the population threshold was interpolated as 8 mN for the CLBP patients and 48 mN in healthy controls. There were significant differences between groups in the incidence of reported pain at 8 mN (χ ² = 12.92; p < 0.05), 16 mN (χ ² = 12.48; p < 0.05) and 32 mN (χ ² = 10.60; p < 0.05) on the hand dorsum (see Figure 3C).

Comparison to reference data

We also compared our MPS data (detailed data on SRF are not publicly available in the DFNS reference data) with respect to the reference data from the German Research Network on Neuropathic Pain (DFNS). First, we compared the results of our HC group with the normative data. Results obtained at the hand of our HC group were in the range of the DFNS reference data as judged by the recommended quality self-control procedure (average z-score of 0.09 for HC subjects was < 95% confidence interval of DFNS reference data, recommended to be < 0.25; [36]) indicating a good agreement with the DFNS standard. The data of our HC group was almost perfectly matched to the reference data (t = 0.05, P = 0.62). Second, T-tests confirmed a significant group difference for MPS on the hand dorsum for the CLBP patients with respect to our HC group on the basis of z-transformed values (t = 2,96 P < 0.05) as well as with respect to the reference data (t = −2.65, P < 0.05). Furthermore, the MPS z-scores for the hand dorsum of CLBP patients normalized to our designated matched control group, and the z-scores normalized to the DFNS reference data [36], were significantly correlated (R² = 0.979, P < 0.001) underlining the robustness of results with respect to reference data.

Stimulus response function to punctate pinprick stimuli at the back

Our data revealed significant changes on the back for punctuate mechanical pinprick stimuli, especially pronounced at low to moderate stimulus intensities. SRFs (Figure 1A) showed manifold increases of pain ratings for low to moderate intensities compared to healthy controls. SRFs converged for higher intensities. Compared to healthy controls pain ratings to pinpricks were increased by a factor of 2 to 6. Beyond the frequent finding of muscle pain thresholds, this is the first report demonstrating hyperalgesia to pinprick stimuli, specifically the pronounced allodynia to low-intensity pinprick stimuli in CLPB patients. These changes are a hallmark of centrally mediated hyperalgesia, and are found in experimental models of central sensitization (neurogenic hyperalgesia [29, 36–40] reviewed in [41, 42]) as well as in neuropathic pain patients [43–46]. Specificity for facilitation of this stimulus modality has also been shown in animal models of central sensitization at the level of the spinal dorsal horn, thalamus, and amygdala [47–50].

Stimulus response function to punctate pinprick stimuli on the hand

Interestingly, we also found significant changes to punctate pinprick stimuli on the hand (MPS, stimulus response function), especially pronounced at low to moderate stimulus intensities (see SRF and pain incidence in Figures 1B-C, 3B-C) in CLBP patients. This finding may be of principal importance because it demonstrates changes in pain sensitivity extending far beyond the painful lumbar back. More specifically, there was no significant effect of factor Region suggesting that the increase in sensitivity in CLBP patients to mechanical punctate stimuli is similar for the back and the hand. The mechanisms for this widespread pain sensitivity are largely unknown. Potential explanations include e.g. altered attentional processes, loss of inhibitory pain control, or higher vulnerability in CLBP patients.

A further potential explanation may be related to plasticity at supraspinal levels, for which the thalamus is a prime candidate, since it forms the next relay of the ascending pathway. Receptive field sizes in the thalamus are much larger than in the spinal cord, both in the specific pathways of lateral thalamus projecting to the somatosensory cortices, and in the non-specific pathways of medial thalamus projecting to the anterior cingulate cortex and amygdalae, where receptive field sizes can encompass whole quadrants, or body sides [51]. Several findings in animals support the involvement of the thalamus. In one such study, facial mechanical allodynia spreading from the affected to the contralateral side was accompanied by enhanced thalamic transmission [52]. Contralateral spread of mechanical hyperalgesia and allodynia also appears in more severe cases of migraine in human patients [47, 52]. Similar mechanisms have been found in nociceptive plasticity of the spinal or medullary dorsal horn and thalamic plasticity [53, 54]. The modality specificity of experimentally-induced mechanical hyperalgesia was also reported as preserved in the amygdala receiving ascending projections from the medial thalamus, as well as descending input from the anterior cingulate cortex and insula [48]. Data on extrasegmental spread however are not yet published. Nevertheless, reverse modulation has been demonstrated in patients with fibromyalgia after two terms of successful local treatment of tender points with widespread relief of signs of hyperalgesia [55].

Alternatively, a relative insufficiency of descending control may passively, or descending facilitation may actively, produce an expansion of receptive field sizes both segmentally and extrasegmentally. Human and animal data have demonstrated that experimentally-induced secondary hyperalgesia involves descending facilitation and spreads rapidly into neighbouring segments [49, 56]. However, there is no animal data so far demonstrating that sustained spinal nociceptive input may spread sensitization for more than a few segments, e.g. from lumbar to cervical level. Another alternative explanation for the hypersensitivity might be an a priori increased sensitivity to mechanical stimuli in CLBP patients compared to HC. Longitudinal approaches might test this assumption.

Limitations and further directions

Our sample size of 14 subjects in each group is relatively small, so the study should be extended to larger sample sizes and different centres. Furthermore, we only tested female subjects, whereas previous studies have demonstrated gender differences in pain thresholds (Magerl et al. 2010). Our study should therefore be extended to the investigation of SR functions in male subjects. Additionally, we did not follow the menstrual cycle of our participants although there are known variations in pain perception during the cycle. Nevertheless, in our small sample, the sensory changes we report are robust and the effect sizes large (CLBP vs. HC, MPS back: d = 1.08; MPS hand dorsum: d = 0.90; MPS hand palmar: d = 1.07).

The data do not allow us to distinguish between the two major hypotheses for this higher-order effect, i.e., sensitization after an aversive event vs. susceptibility. Longitudinal studies might help to solve this question. Alternatively, identifying subjects with a specific pinprick hyperalgesia without CLBP (or other chronic pain symptoms) might be informative about the independence of the observed phenomena.