Cold allodynia is correlated to paroxysmal and evoked mechanical pain in complex regional pain syndrome (CRPS)

Patient material and diagnostic criteria

A total of 61 patients with CRPS 1 and 2 were selected from a large material of patients referred to the Section of Clinical Neurophysiology at Oslo University Hospital, Rikshospitalet. In order to diagnose CRPS, two different diagnostic criteria were used. If examined in the period 2004–2010 they were diagnosed according to previous International association for the study of pain (IASP) criteria [3]. As for patients investigated after 2010, we used the Budapest criteria [2]. All patients underwent clinical neurological examination, and quantitative sensory testing (QST). EMG/neurography was included in most patients (when possible due to pain). Patients were differentiated into three subgroups (A, B, C) based on results from thermal testing; (A) patients with temperature (cold – and/or heat) allodynia, (B) patients with QST-results compatible with small fiber degeneration (elevated warmth and/or cold detection thresholds), and (C) patients with both elevated thermal detection thresholds and temperature allodynia (features of both A and B).

The publication of results was considered by the Regional Ethical Committee (2015/688) as part of a quality improvement also for better treatment-options of a clinical material with no actual necessity of an ethical approval. We still obtained informed consent. All patients were investigated according to ethical guidelines and the Helsinki declaration.

This is a retrospective study, on a material of CRPS patients partly overlapping (but not entirely identical) with the material from two previous studies [20, 21]. In the present study, we have included more patients evaluated according to the Budapest criteria.

Anamnestic details

Based on previous medical records, specialist statements and results from examination at Oslo University Hospital, Rikshospitalet, we assessed the following: past medical history, heredity (family history), and details concerning any injury or surgery. We performed a detailed pain-assessment of (1) stimulus-dependent pain/evoked pain (allodynia, hyperalgesia) and (2) stimulus non-dependent pain (spontaneous ongoing and paroxysmal pain). Spontaneous ongoing pain is the stimulus-independent constantly ongoing pain of the patient, while paroxysmal pain is a stimulus-independent sudden appearance of pain, described as electric shocks, within the painful area, of seconds duration but with a large variety in frequency. Pain intensity of spontaneous ongoing pain was graded by the use of a 0–10 numeric rating scale (NRS), where 0 is no pain and 10 is worst imaginable pain. Paroxysmal pain was reported as present or not present. We also evaluated other symptoms of the patients, including symptoms of autonomic dysfunction and motor impairment (not further described in the present paper).

Clinical examination

Clinical investigation, which was performed in all patients at OUS-Rikshospitalet, was initiated by inspection. It was of particular interest to inspect both the injured and contralateral limb in order to compare and look for any asymmetry like oedema, abnormal sweating, discoloration of the skin, atrophy and trophic changes (thin and/or gleaming skin, as well as changes in hair growth). During examination we also assessed motor and sensory function. Evaluation of sensory aspects included screening for both hyperesthesia (increased sensation) and hypoesthesia (loss of/decreased sensation). Patients with hyperesthesia were further examined to consider the presence of mechanical allodynia (Pain due to a stimulus that does not normally provoke pain) [3] and/or hyperalgesia (Increased pain due to a stimulus that normally provoke pain) [3]. Allodynia to light touch was tested with a cotton swab and hyperalgesia to pin-prick/pressure.

Skin temperature was measured by a handheld measuring device (Somedics Tempett, Hőrby, Sweden) at the site of injury, adjacent areas and, at the contralateral limb. A temperature difference of >1 °C was considered significant [22].

EMG/neurography

In order to differentiate between CRPS 1 and 2, patients were examined with EMG/Neurography, with either a Dantec Counterpoint, Skovlunde, Denmark or a Dantec Keypoint-apparatus, Skovlunde, Denmark. Neurography was applied to investigate the relevant main nerves in the upper or lower extremities (median, ulnar, peroneal, tibial posterior, sural nerves). EMG of appropriate muscles was only performed, if necessary, in patients with CRPS for diagnosis of a nerve lesion. This examination was, however, not feasible in all patients due to pain, and an eventual peripheral nerve lesion was then diagnosed based on clinical findings (motor and sensory deficits within the innervation territory of a single nerve).

Thermal detection thresholds (quantitative sensory testing–QST)

Detection thresholds for warmth (WDT), cold (CDT), heat-pain (HPT) and cold-pain (CPT) were quantitatively determined by Thermotest (Somedic AB, Hörby, Sweden). A skin probe (thermode) of 25 mm × 50 mm was placed over the actual painful area, as well as over the contralateral area. Baseline was set at 32 °C with a 1 °C/sec rate of change. Patients were equipped with a registration device (handheld), allowing them to press the button when the perceived threshold/sensation of interest was reached. The upper and lower cut-off levels were 50 °C and 10 °C. The patients were asked to press the button upon the first sensation of cold or warmth. CDT was defined as the highest temperature perceived as cold, and WDT was the lowest temperature perceived as warmth. Similarly, the patients were asked to signal upon the first sensation of cold – or heat pain. CPT was defined as the highest temperature perceived as cold-pain, and HPT as the lowest temperature perceived as heat-pain. WDT and CDT were calculated from the average of five subsequent recordings. HPT and CPT were measured as the average of three successive recordings with an interstimulus interval of 10 s. Before QST skin temperature was measured by a handheld infrared thermometer (Somedics Tempett, Hőrby, Sweden) at the same site as the location of the thermode, as well as adjacent areas. Elevated thresholds to warmth and cold in the affected area are defined as elevated thresholds compared to thresholds in the normal contralateral area. As there is no definite temperature limit for thermal allodynia, we have defined heat allodynia as a lowered HPT (decrease in temperature) in the affected area compared to the contralateral side, and as the lowest temperature perceived as heat pain below 40 °C (below 2.5 percentile in an aged-matched normal material).

Cold allodynia is defined as a lowered CPT (increase in temperature) in the affected area compared to the contralateral side, with a minimum of 2 °C difference (higher temperature).

Statistical analysis

We investigated for significant correlations between the QST-defined subgroups (A, B, C). Only the statistical results of CRPS 1 and 2 merged are presented. We have also performed similar statistics of subgroups for CRPS 1 and 2 separately, which did not show any differing results, apart for some analyses that were inconclusive due to a limited number of patients with CRPS 2. A two-sided chi-square test (non-parametric test) was used to compare the distribution/prevalence of; (1) allodynia to light touch and (2) paroxysmal pain, between subgroups (A, B, C), as well as the distribution of autonomic dysfunction between patients with CRPS 1 and 2.

A Mann–Whitney U-test was used to compare subgroups for the following variables (1) Intensity of spontaneous pain (specified by numeric rating scale – NRS), and (2) QST values (thermal thresholds between painful and healthy locations). A p-value of less than 0.05 (5%) was defined as being statistically significant. There was no difference in clinical presentation or QST-results between CRPS 1 and 2, as also shown previously [6]. Linear correlations were calculated between heat- and cold pain thresholds separately for patients with and without paroxysmal pain. Statistical analyses were performed in Statistical Package for the Social Sciences (SPSS) (version 25).

Eliciting injuries and clinical presentation

As for the eliciting injuries, the mechanisms included squeeze injuries, fractures, stretch accidents, traffic-accidents and blunt injuries. In total six patients developed CRPS subsequently to surgery. The extremities were the most frequent areas of injury. No patients had pain in contralateral extremity. All 61 patients presented with spontaneous ongoing pain. As for features as evoked pain (both allodynia and hyperalgesia), paroxysmal pain, pain intensity and autonomic symptoms/findings, they are discussed in separate sections below. In total 23 (37.7%) patients were diagnosed with motor dysfunction in the form of muscle palsy.

Autonomic abnormalities

Based on results from clinical examination at OUS-Rikshospitalet we evaluted the presence of clinical signs of different types of autonomic dysfunction. Altered skin temperature was the most common, found in 51 patients (83.6%) (Table 1). Discoloration of the skin is the second most common feature (75.4%), followed by oedema (57.4%), altered sweat pattern (44.3%) and trophic changes (29.5%). A similar pattern of distribution was found separate for CRPS 1 and 2 without significant differences. These findings are compatible with previous findings in parts of the same material [21].

Identification of CRPS-subgroups based on QST-results

Of a total of 61 patients diagnosed with CRPS 1 or 2, 14 (22.9%) patients were classified as Group A (thermal allodynia), 23 (37.7%) patients into group B (reduced thermal sensibility), and 24 (39.3%) patients into group C (both thermal allodynia and reduced thermal sensibility). All patients had some abnormality in QST results in the affected extremity compared to the contralateral control area.

In group A (n=14), a total of 10 patients had a combined heat and cold-allodynia, three patients a pure cold allodynia while one patient had heat-allodynia alone. Cold pain threshold was in the range between 23.7°C and 31.0 °C, which is clinically a severe cold allodynia. Heat pain threshold was in the range from 34.0 °C to 37.7 °C.

In group C (n=24), five patients had a combined cold and heat-allodynia, 17 patients a pure cold allodynia and two patients a pure heat-allodynia.

Cold pain threshold was in the range between 15 °C and 29.4 °C and heat pain thresholds between 33.8 °C and 39.0 °C. For group A and C combined (n=38), we found a total of 15 patients with combined cold and heat-allodynia, 20 patients with pure cold allodynia and three with only heat-allodynia. Cold allodynia was thus present in 35 patients and heat-allodynia in 18 patients.

Correlation between subgroups (A, B, C) and paroxysmal pain

Of a total material of 61 patients, 30 patients (49.1%) reported paroxysmal pain. We found a significant difference in the prevalence of paroxysmal pain between group A (10 out of 14 patients – 10/14–71.4%) and B (5/23–21.7%) (p-value=0.0028), as well as between group B (5/23–21.7%) and C (15/24–62.5%) (p-value=0.0047). Paroxysmal pain was more prevalent in patients with thermal allodynia (merging together group A + C, 25/38–65.8%) compared to patients without thermal allodynia (group B, 5/23–21.7%) (p-value=0.00085), and in particular cold allodynia. These figures are presented more detailed in Table 3.

Table 3:

Distribution of subgroups according to clinical variables.

Subgroups	Number of patients in different subgroups	Allodynia to light touch	Paroxysmal pain
A (thermal allodynia)	14/61 (22.9%)	12/ of total 43 (27.9%) 12/14 in group A (85.7%)	10 of total 30 (33.3%) 10/14 in group A (71.4%)
B (thermal hyposensibility)	23/61 (37.7%)	12/ of total 43 (27.9%) 12/23 in group B (52.2%)	5 of total 30 (16.7%) 5/23 in group B (21.7%)
C (thermal allodynia and hyposensibility)	24/61 (39.3%)	19 of total 43 (44.2%) 19/24 in group C (79.2%)	15 of total 30 (50.0%) 15/24 in group C 62.5%)
Total	61	43	30
A + C	38/61 (62.3%)	31 of total 43 (72.1%) 31/38 of group A + C (81.6%)	25 of total 30 (83.3%) 25/38 in group A + C (65.8%)

1. Table 3 shows the distribution of paroxysmal pain (stimulus non-dependent pain) and allodynia to light touch (stimulus-dependent pain/evoked pain) according to the different QST-defined subgroups (A, B, C) merged for CRPS type 1 and 2. The figures represent both the number of patients and the corresponding percentages.

In group A + C, cold allodynia is more frequent than heat allodynia, being present in 92.1% (35/38) and 47.4% (18/38) of patients respectively, which constitute a total of 57.4% (35/61) patients with cold allodynia, and 29.5% (18/61) with heat allodynia. On the other hand, a total of 20 patients has pure cold allodynia (20/38–52.5%), vs. only three patients presenting with pure heat allodynia (3/28–7.9%). Of those with pure cold allodynia, 12 report paroxysmal pain (12/20–60.0%). Of three patients with pure heat allodynia, only one patient has paroxysmal pain. Sensitized heat pain thresholds (lower values) correlated to sensitized cold pain thresholds (higher values), both for patients with (linear correlation, r=−0.44; p<0.05) and without (r=−0.51; p<0.01) paroxysmal pain (Figure 1) indicating a certain link between sensitization to heat and cold. It is striking, that paroxysmal pain is particularly prevalent in those patients in whom heat- or cold pain can be induced by only minor changes of the normal resting skin temperature potentially indicating that such marginal temperature variations might facilitate paroxysmal pain.

Figure 1:

Scatterplot of cold- and heat pain threshold is shown for patients with (red circles) and without (filled squares) paroxysmal pain. For both patients’ groups higher cold pain threshold correlate to lower heat pain thresholds (linear correlation; paroxysmal pain: r=−0.44, p=0.023; no paroxysmal pain: r=−0.51, p=0.008). Note that many pain thresholds are within the range of normal skin temperatures for many patients.

Small fiber degeneration as a possible pathophysiological mechanism in CRPS

Patients with thermal hyposensitivity (elevated thermal detection thresholds, group B), were stratified assuming a degeneration of their small fibers [12, 23], which has been described in CRPS previously [12, 18, 23, 24].

Findings compatible with small fiber degeneration have been demonstrated in biopsies from patients with CRPS 1 of sural nerves and skin [12, 13]. Nevertheless, a high percentage (60%) of CRPS 1 patients had normal function of small fiber afferent pathways [6], as also supported by the present results. Furthermore, small fiber degeneration is present in many conditions, both painful and non-painful disorders, and is therefore not specific for CRPS [25, 26]. A non-structural dysfunction of small fibers does not need to be explained purely be peripheral mechanisms. Several reports have suggested involvement of central mechanisms in CRPS [1, 4, 27].

Classifying patients by sensory profiles has been suggested to further our understanding of underlying mechanisms, prognosis and validation of pharmacological efficiency for chronic pain conditions, including CRPS [10, 11]. However, the lack of different QST profiles between painful and painless neuropathy patients [28, 29] or in patients with peripheral nerve damage [30] has been disappointing and even in longitudinal studies improvements of sensory thresholds were not accompanied with pain reduction [31].

It is important to emphasize that small fiber degeneration is not directly linked to pain in patients with small fiber degeneration [10, 28, 32, 33]. Small fiber loss may explain loss of function in painful and painless disorders, but cannot explain the presence of pain [32], [33], [34], which instead may be linked to spontaneous activity of C-nociceptors (especially silent nociceptors), not possible to examine by routine QST [30, 32, 35]. Summarized, isolated small fiber degeneration is unlikely to be a major pathophysiological mechanism in CRPS.

Paroxysmal pain is linked to cold allodynia and touch-evoked allodynia

Our result that paroxysmal clinical pain was more prevalent in patients with evoked thermal allodynia may appear surprising. Paroxysmal pain is frequently reported in patients with neuropathic pain disorders [36, 37]. Several pathophysiological mechanisms have been discussed. Ochoa showed an association between paroxysmal pain and spontaneous activity in peripheral nociceptors [38]. Furthermore, studies have shown a link between spontaneous activity and hyperexcitability [30, 35].

Potential link between cold/heat allodynia and paroxysmal pain

Sensitized transduction mechanisms as sign of peripheral sensitization have been held responsible for heat and cold allodynia with TRPA1 and TRPM8 typically linked to cold allodynia and TRPV1 for heat allodynia. However, agonists for TRPA1 and TRPM8 did not provoke pain in patients with cold allodynia [39, 40] and thus, TRP-independent mechanisms are more probable. A simple physical effect of lower temperatures is slower regular sodium channel inactivation providing more time for competing conformation changes. These include an open channel block that can only resolve via another opening of the sodium channel inducing “resurgent currents”. Thereby repetitive discharge is induced that explains for example increased cold sensation in patients receiving oxaliplatin via NaV1.6 [41]. However, resurgent currents have also been reported for sodium channels of nociceptors such as NaV1.7 [42] and NaV1.8 [43], providing a mechanistic link for repetitive nociceptor firing in neuropathy [43]. Reduced activity of inhibitory potassium channels is another possible mechanism for cold-induced pain [44]. In addition to closure of cold-sensitive potassium channels [45] reduced expression of Kv 1.1/1.2 in neuropathic conditions has been shown to increase cold sensitivity of nociceptors [44]. Thus, repetitive opening of sodium channels and reduced inhibitory potassium currents facilitate nociceptor discharge and increase their discharge frequencies and might therefore underlie the higher prevalence of paroxysmal pain in the patients. Direct nerve recordings from normal menthol sensitive C-cold fibers in volunteers [46, 47] and spontaneously active C-nociceptors activated by cooling in patients with neuropathic pain [48] are in-line with this concept.

Correspondingly, higher temperatures can exert indirect effects on excitability via the physical effects rather than activation of TRPV1. Hyperpolarization of nociceptors by the simple physical warming effect on the sodium-potassium pump generally reduces their excitability by moving the membrane potential away from the activation threshold. However, if their resting membrane potential is substantially depolarized most of the NaV1.7 channels are kept in an inactivated state excluding them from action potential generation. In this situation, warming-induced hyperpolarization will “rescue” these inactivated sodium channels thereby paradoxically increasing their excitability as hypothesized in patients with hereditary erythromelalgia [49]. Obviously, the concomitant expression of heat and cold allodynia reflects the complexity of temperature-dependent excitability changes. Thus, assuming such excitability changes in patients suffering from heat and cold-allodynia it might have been expected that even minor incidental temperature changes can facilitate a paroxysmal pain pattern.

In addition to a correlation between cold allodynia and paroxysmal pain, we also found a link between thermal allodynia (primarily cold allodynia), and allodynia to light touch (dynamic mechanical allodynia) (gr. A vs. B p-value=0.038 and gr. A–C vs. B p-value=0.014). This might appear straight forward because underlying hyperexcitability is likely to explain both phenomena [5, 50], [51], [52], [53]. Touch-evoked allodynia is traditionally explained by spinal sensitization that switches the processing of low-threshold mechanosensory input (A-fibers) from touch to pain [54], [55], [56]. Ongoing peripheral C-nociceptor input is required to maintain spinal sensitization [57, 58]. It has been shown that in CRPS “pain intensity was greatest in patients with positive symptoms (particularly cold allodynia and thermal hyperalgesia), but was unrelated to negative symptoms (loss of heat pain or touch sensations)” [59]. Since spontaneous pain was present in all our patients, all with spontaneous continuous pain and no significant difference in pain intensity between groups, one might have expected allodynia to light touch to be independent of thermal allodynia. Neuropathy-related impairment of A-fiber function might contribute, but such an explanation would only include CRPS 2 but not CRPS 1. Alternatively, pain and hyperalgesia might differ in their optimum discharge frequencies [60] or more superficial location in the skin of the spontaneously active nociceptors might facilitate both, temperature hypersensitivity and the development of touch evoked allodynia.

Perspective

Our data suggest that the correlations between cold allodynia, touch evoked allodynia and paroxysmal clinical pain in patients with CRPS indicate sensitized peripheral nerve endings located superficially in symptomatic skin. Their superficial location makes them particularly sensitive to mild changes of the skin surface temperature. The combination of changed neuronal expression patterns and local inflammation in CRPS might facilitate the occurrence of cold allodynia. Irrespective of the exact mechanism it is important to note that the assumed involvement of these nociceptors not only in the evoked, but also in the paroxysmal clinical pain appears to be of therapeutical value. Based on their superficial location it appears promising to use topical analgesic strategies such as local anaesthetics for those patients with cold allodynia.