2.1 General observations

A total of 42 animals were included in the study. The average time to a locomotor score ≥4 was 12 days for SCC animals and 1 day for sham animals. Sporadic mild spontaneous spasms were initially observed in SCC animals, but the mobility, evaluated by the total distance travelled in the open field, was similar for both SCC and sham animals at all times tested (injury group: p = 0.564, time: p < 0.001; mixed model) (Table 1). There were no significant differences in general health (e.g., weight and fur coat) or behaviour between the injury groups, nor were any signs of distress observed (e.g., self-inflicted abrasions, spontaneous or handlingevoked vocalization).

2.2 Development of mechanical hypersensitivity

The animals were tested for mechanical sensitivity preinjury and 4, 6, and 12 weeks postinjury. Preinjury, no difference between the injury groups was observed (p = 0.131, Student's t-test) (Fig. 1a). Following injury, SCC animals (n = 20) had significantly lower at-level mechanical sensitivity thresholds than sham animals (n = 21) (p < 0.001, mixed model excluding preinjury). At-level mechanical sensitivity thresholds of SCC animals were, in contrast to the sham animals (p = 0.156, mixed model), significantly decreased postinjury compared with preinjury (p < 0.001, mixed model) remaining low throughout the test period. The thresholds were on average decreased by 71% (8.51-2.46 g) in the SCC animals as compared with preinjury, and at 6 and 12 weeks postinjury, SCC animals had a 65% lower threshold than sham animals. One SCC animal did not develop any mechanical hypersensitivity and was thus excluded from the drug study.

2.3 Development of anxiety

Anxiety levels were determined by the time spent in the open arms of the EPM. Preinjury, there was no difference between the injury groups (p = 0.884, Student's t-test) (Fig. 1b). An overall effect of injury group was observed (p = 0.036, mixed model excluding preinjury), and both SCC and sham animals experienced an initial decrease in time spent in open arms 2 weeks postinjury (p < 0.001, SCC and p = 0.003, sham; mixed model contrast), with a further decrease in SCC animals at 6weeks (p = 0.017, mixed model contrast) lasting throughout the test period. The decrease amounted on average to 85% at 6 and 12 weeks compared with preinjury. In sham animals, in contrast, there was no difference at 6weeks compared with preinjury (p = 0.101, mixed model contrast), nor were the levels at 6 and 12 weeks significantly different (p = 0.067, mixed model contrast). SCC animals spent less time (76% on average) in the open arms compared with sham animals at both 6 and 12 weeks postinjury (p < 0.001 and p = 0.003 respectively, Student's t-test). Repeating the mixed model analysis including the total distance travelled in the EPM as a covariate did not affect the above conclusions.

Fig. 1

Development of mechanical hypersensitivity (a) and anxiety (b) in animals with a 12.5mm SCC at bone segment T10. Data are presented as average at-level mechanical sensitivity thresholds (g)±S.E.M. ^*p < 0.05 compared with preinjury. ^#p < 0.05 compared with sham group. (A) SCC animals had a lower threshold 4, 6, and 12 weeks postinjury as compared with preinjury (p < 0.001, mixed model). There is no significant difference between 6 and 12 weeks postinjury (pre- and postpharmacological testing) (p = 0.191, mixed model contrast). (B) SCC animals spent less time in the open arms of the EPM 4, 6, and 12 weeks postinjury compared with preinjury (p = 0.036, mixed model excluding preinjury). There is no significant difference between 6 and 12 weeks postinjury (pre- and postpharmacological treatment). Data have been normalized to the mean value of the preinjury measurement.

2.4 Correlation between mechanical sensitivity and anxiety

Linear correlation between mechanical sensitivity and anxiety was evaluated based on the time spent in the open arms as a function of at-level mechanical sensitivity thresholds at 6 weeks postinjury. There was no apparent relationship (R² = 0.378, p = 0.091, Spearman ranked correlation) between the two parameters.

2.5 Effect of pregabalin, morphine, and midazolam on mechanical sensitivity

The effect on mechanical hypersensitivity of a single dose of an anxiolytic and two analgesics was examined in a subgroup of animals (n = 23). In SCC animals, a statistically significant effect of pregabalin (p < 0.001, mixed model contrast) and morphine treatment (p = 0.001, mixed model contrast) compared with saline treatment was observed, whereas treatment with midazolam resulted in no significant effect (p = 0.323, mixed model contrast) (Fig. 2a). Pregabalin and morphine both increased the at-level mechanical sensitivity thresholds to approximately the same level as seen in the saline-treated sham animals, but not to preinjury level. The at-level mechanical sensitivity thresholds for sham animals were not affected by either treatment (p = 0.178, mixed model).

Fig. 2

Treatment effect on mechanical hypersensitivity (a) and anxiety (b) in animals with a 12.5mm SCC at T10. Data are presented as average at-level mechanical sensitivity thresholds (g)±S.E.M. ^*p < 0.05 compared with NaCl 0.9%. (A) SCC animals had an increase in at-level mechanical sensitivity threshold after treatment with 30 mg/kg pregabalin IP (p < 0.001, mixed model contrast) and 3 mg/kg morphine IP (p = 0.001, mixed model contrast) but no significant effect of midazolam 0.5 mg/kg IP (p = 0.323, mixed model contrast). Sham animals had no treatment effect of either drug (p = 0.178, mixed model). (B) SCC animals had a twofold increase in open arm activity after treatment with 30 mg/kg pregabalin IP (p < 0.001, mixed model contrast), whereas 3mg/kg morphine IP (p = 0.425, mixed model contrast) and 0.5 mg/kg midazolam IP (p = 0.675, mixed model contrast) had no significant effect. Sham animals had no treatment effect of either drug (p = 0.710, mixed model). Data have been normalized to the mean value of the saline measurement.

2.6 Effect of pregabalin, morphine, and midazolam on anxiety

The effect on anxiety of a single dose of one anxiolytic and two analgesic drugs was estimated in another subgroup of animals (n = 18). Pregabalin increased the open arm occupation of SCC animals by twofold (p < 0.001, mixed model contrast), though the absolute level was still substantially lower than the saline-treated sham animals (increase from 11% to 23%) and preinjury levels (Fig. 2b). The second analgesic drug morphine did not have any treatment effect (p = 0.425, mixed model contrast), and it appeared to decrease the open arm activity rather than increase it. The anxiolytic drug midazolam had no apparent effect compared with saline (p = 0.675, mixed model contrast). None of the pharmacological drugs were effective in sham animals (p = 0.710, mixed model). Repeating the mixed model analysis including the total distance travelled in the EPM as a covariate did not affect the above conclusions.

2.7 Effect of pregabalin, morphine, and midazolam on mobility

Side effects of the drugs, e.g., sedation and somnolence, were evaluated by the total distance travelled in the open field following administration (Table 1). An overall treatment effect was observed (p = 0.006, mixed model), but no group difference (p = 0.290, mixed model). Midazolam treatment of sham animals resulted in borderline significantly lower distances travelled than saline-treated sham animals (p = 0.054, mixed model contrast), but this effect was not observed for SCC animals. None of the other treatments had a significant effect on either group.

3.1 Pain and anxiety relationship

The relationship between pain and anxiety is still hypothetical. Several explanations have been suggested [5].

The mediational model, i.e., development of anxiety leads to increased pain perception or presence of anxiety may prevent the ability to experience pain relief. Only one factor changes, but affects the experience of the other.

Linear causality model, i.e., pain and anxiety are directly correlated; thus, either pain causes anxiety or anxiety causes pain. Changing one will equally change the other.

Coexistence model, i.e., pain and depression are related to SCI but remain independent. Changing one, does not necessarily affect the other.

In most of the previous studies it is hypothesized that anxiety is a consequence of persistent hypersensitivity (linear causality), supported by the fact that both morphine and gabapentin reversed pain and anxiety-like behaviour without having an effect on anxiety in the sham animals [35]. Furthermore, midazolam reduced anxiety in both groups without affecting mechanical hypersensitivity [32,36].

In this study, however, pregabalin and morphine reduced mechanical hypersensitivity, while pregabalin, but not morphine, reduced anxiety-like behaviour in SCC animals. Since anxiety could not be attenuated by pain relief (morphine), but only by the dual-acting pregabalin, this study supports the coexistence model: that following SCC, at least partly, anxiety-like behaviour develops independently of SCI pain. Since we did not observe the expected effect of midazolam on anxiety, the lack of effect of midazolam on hypersensitivity may be due to inadequate dosage.

The coexistence model is further supported by the fact that we saw no significant correlation between postinjury time spent in open arms and at-level mechanical sensitivity thresholds, although this may be due to an insufficient number of animals. The development and level of pain have previously been reported to not be related to preinjury anxiety levels, but no reports on the relation to postinjury anxiety levels were included [47].

To the best of our knowledge all preclinical studies on pain and anxiety correlation use evoked sensitivity as outcome for pain-like behaviour. Evidence from clinical research with SCI patients suggests that anxiety may be correlated to spontaneous rather than evoked pain. Hence, patients with continuous pain had a higher score on the Hospital Anxiety and Depression Scale than those with intermittent or without pain, and thus 33% of SCI patients with continuous pain had probable clinically relevant anxiety compared with 21% of patients with intermittent and 6% without pain [48].

Finally, anxiety has also been related to a number of other consequences of SCI in humans [22], which highlights that anxiety may be generated and maintained by multiple factors and mechanisms.

Our results are in concordance with findings in clinical trials of SCI pain, showing effect of pregabalin on pain and anxiety [49,50] and of the opioid tramadol on pain [51].

Even though pregabalin reduced the anxiety behaviour in the SCC animals, it was not reversed to the level of the saline-treated sham animals (only to approx. 23%), whereas mechanical hypersensitivity was decreased to the level of the saline-treated sham animals. This suggests that pregabalin is more effective in reducing the sensory pain experience than the anxiety component, at least with the applied dosage. The observation is not likely caused by dosage issues, since the recommended dose of pregabalin for treatment of neuropathic pain and generalized anxiety is the same, although we cannot exclude that a longer treatment period may be required to modulate anxiety behaviour.

5.1 Animals

All experiments were performed according to the Ethical Guidelines of the International Association for the Study of Pain [52]. The study protocol and experimental design, animal housing, husbandry, and handing were in compliance with European and Danish legislation and approved by the Danish Animal Experiments Inspectorate (2006/561-1211).

In the experiment, 42 female Sprague-Dawley rats (200 g) (Taconic Farms Inc., Borup, Denmark) were used. All animals were housed in pairs in a temperature-controlled environment with a 12-h light/dark cycle. Food and water was available ad libitum. Special soft bedding was used after surgery to minimize contact pressure on the rat hind quarters. Animals were acclimatized to the housing facility for one week upon arrival from the vendor.

5.2 Model of central neuropathic pain: spinal cord contusion (SCC)

Animals were randomly assigned using random allocation software [53] and allocated to one of two groups: SCC (SCI by contusion, n = 21) and sham (laminectomy, but no contusion, n = 21).

The surgical procedure has been described in detail elsewhere [23]. In short, surgery was performed in an isolated facility intended for the purpose. Animals were anesthetized initially with 1ml/kg SC 1:1:2 Hypnorm^® (0.315 mg/ml fentanyl/10 mg/ml fluanisone)/Dormicum^® (5 mg/ml midazolam)/NaCl 0.9% (w/w). After laminectomy of bone segment T10, a 10 g rod was dropped 12.5mm onto spinal cord segments T12-13 using the MASCIS impactor (Rutgers, New Jersey)[39], generating a 1.0-2.0mmvertical compression of the cord (mild to moderate injury). The surgical site was closed in two layers with 4.0 re-absorbable sutures. Fluid replacement of 5ml 0.9% NaCl was administered IP, and the animals were allowed to recover under a heating lamp. Average time for the procedure was 1 h.

Postoperatively, animals required daily manual expression of the bladder until self-voiding (4 days on average). Responses to handling, fur condition, weight, porphyrin production (red tears), and signs of bladder infections (urine pH, colour, and clarity) were collected on daily inspections. A combination of sulfamethoxazole and trimethoprim (200 + 40mg/kg) (Tribissen^®, Boxmeer, the Netherlands) was administered prophylactically SC for 5 days postoperatively to prevent bladder infections.

Locomotor recovery and presence of spastic syndrome were evaluated regularly as described in Baastrup et al. [23]. A locomotor score of 4 (walks with deficit (balance and foot positioning)) out of 6 (normal walking) was required for performance of the behavioural testing.

Excessive grooming behaviour/autophagia was monitored [23] and any score >0 (normal) prompted topical treatment with Clindamycin gel 1% and grade >3 (penetration through subcutaneous layers) or refractory grade 2 (penetration through the dermis) served as human endpoint.

5.3 Measurements of at-level mechanical sensitivity

Animals in groups of four were placed in individual 25cm × 10cm × 15cm Plexiglas boxes with black opaque walls (excluding the wall facing the investigator). A 30-min acclimatization period to the testing facility was allowed. Von Frey filaments (Semmes Weinstein nylon monofilaments, Stoelting, IL, USA) (0.07, 0.17, 0.69, 1.20, 2.04, 3.63, 5.50 g) were applied approx. 2 cm lateral to the incision site on the thorax until bending of the filament according to the up-down method [54]. Any brainstem reflex response (licking, guarding, struggling, vocalizing, jumping, and biting) [55] indicated a positive response, and results were converted to an at-level mechanical sensitivity threshold. All testing sessions were performed by the same trained investigator (CB). In sessions including pharmacological treatment, a pretreatment baseline threshold was measured.

5.4 Measurement of motility: open field

The effect of locomotor dysfunction on normal spontaneous exploratory behaviour, signs of spontaneous pain, and sedative side effects were evaluated in a black square 100cm × 100cm × 30cm open field. All animals were in turn placed in the same corner of the open field close to the wall and allowed to freely explore the environment for a test period of 5min. All trials were tracked and recorded using the Ethovision XT 4.0 software (Noldus Information Technology b.v., Wageningen, The Netherlands). The total distance travelled was calculated and used as outcome measure. The open field was thoroughly cleaned with 70% ethanol between the testing of each animal.

5.5 Measurement of anxiety: elevated plus maze (EPM)

The elevated plus maze (EPM) (Stoelting, IL, USA) consists of two opposing open and two closed arms (50cm × 10 cm) extending from a central platform (10cm × 0 cm), all made of odourless opaque gray plastic. The maze was elevated 40cm above floor level and equally illuminated. The animal was placed in the middle of the EPM facing an open arm and allowed to freely explore the environment for a test period of 10 min. All trials were tracked and recorded using the Ethovision XT 4.0 software (Noldus Information Technology b.v., Wageningen, The Netherlands). The time spent in the open arms was calculated by the software (outcome measure). After each test, the maze was carefully cleaned with 70% ethanol.

5.6 Experimental design

All 42 SCC and sham animals performed the anxiety, mechanical sensitivity, and motility tests prior to injury, 4 weeks postinjury, and 1 week after completion of the treatment study (12 weeks postinjury) to monitor temporal development. Mechanical sensitivity and anxiety tests were furthermore performed at 6weeks postinjury.

SCC and sham animals were randomly allocated to two separate test groups: one was tested for anxiety behaviour (n = 18), and one for mechanical sensitivity and motility testing (n = 24). All pharmacological treatments were performed 7-11 weeks postinjury. Each animal was treated with single doses of all four drugs in a balanced design, with each test session separated by 1 week to avoid carry-over effects and test habituation.

Testing was performed in an isolated testing facility to avoid inauspicious and unfavourable visual, olfactory, and auditory cues, minimize anticipatory hypervigilance and variations in stress levels induced by external factors [56,57]. Animals were tested in randomized order to avoid group bias. Drug solutions were prepared and drawn from identical vials, and the investigator was blinded both to injury groups and pharmacological treatments.

5.7 Drugs

Drug dosages were pregabalin 30 mg/kg, morphine 3mg/kg, midazolam 0.5 mg/kg, and 0.9% NaCl, based on previous studies with peripheral pain models. Pregabalin (pure substance) was supplied by Pfizer Inc., CT, USA; morphine (Morfin “SAD” 1mg/ml) was obtained from SAD, Copenhagen, Denmark; and midazolam (Midazolam “Hameln” 5mg/ml) was obtained from Hameln Pharmaceuticals, Hameln, Germany. All compounds were dissolved or diluted in saline (0.9% NaCl) and administered intraperitoneally 1ml/kg 45 min prior to testing.

5.8 Statistics

Data are presented as means ± SEM. Data for each test and each injury group were analyzed for temporal development and treatment effect with a linear mixed model including time/treatment and test sequence as a fixed effect and animals as a random effect, and for at-level mechanical sensitivity, the pretest threshold was included as a covariate. Appropriate post hoc contrast testing was predefined and performed. A p-value less than 0.05 indicates a statistical significance.

Results from the EPM test are graphically presented with data normalized to the mean value of the sham group (vehicle or preinjury).