2.1 Spared nerve injury

The SNI model was developed by Decosterd and Woolf [16]. An incision is made at the lateral surface of the left thigh, and the proximal and distal parts of the biceps femoris muscle are separated to expose the sciatic nerve and its three terminal branches. The tibial and common peroneal nerves are tightly ligated and 2–3 mm of the nerves distal to the ligation removed. Any stretching or contact with the spared sural nerve should be avoided. The model affords high reproducibility. It is easy and fast to perform with a high operation success rate even for inexperienced operators. The SNIoperated animals have normal food intake, movements, grooming and growth [16].

Following SNI surgery, both rats and mice as early as two to three days after injury will develop long-term hypersensitivity to mechanical (Fig. 2), but not to thermal stimuli (Fig. 3) [17,18]. For rats, the hypersensitivity is of long duration (15 months or longer) [19]. For mice, the duration is at least 30 days [18]. An increased sensitivity to cold stimuli has also been observed in SNI-operated rats [16,19].

Fig. 3

Paw withdrawal latency in response to thermal stimulation before and after nerve injury, assessed in (A). SNI rats (n = 35), (B) SNI mice (n = 6), (C) SNT rats (n = 10) and (D) SNT mice (n = 9). The Hargreave’s plantar test (IITC Life Sciences, Woodland Hills, CA, USA) was conducted two or three times before SNI/SNT surgery and repeatedly up to 21 days after surgery. A radiant heat source, adjusted to give a pre-surgery withdrawal latency of 4–5 s and 8–9 s for mice and rats respectively, was directed to the lateral plantar surface of the hind paw. Measurements were done by manually starting and stopping the heat source, while the time was simultaneously recorded to the nearest 0.01 s by the apparatus. Measuring of thermal stimulation demand equal and stable skin temperature of area subjected for measurement [33]. This is ensured by performing the measurement at a set ambient temperature and where the paw has contact with the glass plate. The withdrawal latency was calculated as the mean of two or three measurements, which were separated by 5min. To prevent tissue damage, a cut-off of 20 s was used. Post-surgery latencies ipsilateral to nerve injury were significantly lower than pre-surgery latencies in the SNT model (^*p < 0.05, ^**p < 0.01, ^***p < 0.001; repeated measures one-way ANOVA with Bonferroni’s multiple comparison test). However, in the SNI model, post-surgical latencies were either not altered or slightly higher than pre-surgery latencies (^#p < 0.05, ^##p < 0.01; repeated measures one-way ANOVA with Bonferroni’s multiple comparison test) (in-house, unpublished data).

A disadvantage associated with the lesions in the peroneal and the tibial nerves leaving the sural nerve intact is that the procedure causes a degeneration of axons, which also limits the number of distal intact axons. Hence it will lead to difficulties in the performance of behavioral tests, as the ipsilateral paw includes uninjured areas close to the denervated areas [11,16]. An additional complication for the assessment of pain behavior in SNI animals is that the hypersensitive area in the limb is the lateral part of the paw that is technically difficult to test.

2.2 Spinal nerve ligation

The SNL model is associated with development of long-term hypersensitivity to mechanical and cold stimuli as well as spontaneous pain-like behavior. These behaviors develop quickly after ligation [20,21]. The development of heat hypersensitivity is more variable and not found in all experimental conditions.

The original version of the SNL model, developed by Kim and Chung, involved a tight ligation of both the L5 and L6 spinal nerves [20]. An incision is made along the spinal column and the left paraspinal muscles are separated from the spinous processes at the L4-S2 levels. Under a modular high-performance stereomicroscope, the L5 spinal nerve is isolated and 1–3 mm of the nerve is ligated distal to the dorsal root ganglia. Special care should be taken to avoid any damage to the L4 spinal nerve. Based on findings that only 0.4% of all sciatic afferents resides in the L6 dorsal root ganglions [22], the surgical procedure nowadays often involves an injury only to the L5 spinal nerve. Several studies have confirmed that this modification has a very limited effect on the outcome [20,23,24]. Since then the SNL model has also been used in mice with similar characteristics as seen in rats [25].

The behavioral signs resulting from SNL surgery, e.g. guarding, licking and lifting of the ipsilateral paw, may indicate clinical face validity. Hyperalgesia, allodynia-like behavior and spontaneous pain suggest high validity, as these symptoms may be present in patients with neuropathic pain. Compared to other models, the SNL model is not associated with autotomy, i.e. scratching and biting of the denervated hind paw. A likely explanation is that the hind paws still are innervated in the SNL model and that the sensation is not only spared but exaggerated. In contrast, in the CCI and SNI model, the hind limb or part of it is denervated and entirely insensate [12,26].

An interesting observation in the SNL model is that glutamate uptake was reduced by 72% in the ipsilateral dorsal horn, in comparison to sham operated rats six weeks after surgery. This was due to suppression of the excitatory amino acid transporters and/or functions and might be a common mechanism for sensitization regardless animal model, as SNL, SNT as well as CCI all lead to definitive decrease in expression and function of this transporter [27].

2.3 Spinal nerve transection

The SNT model was originally developed as a continuation and validation of the SNL model. Apparently, the SNT model produces similar results as did the SNL model (Figs. 2 and 3). Sheen and Chung [28] could subsequently conclude that the SNT model resembles the SNL model, and can be used as an animal model for neuropathic pain as well [28]. However, it has been suggested that this model lacks the local infiammatory component that is present in the CCI, partial sciatic nerve injury (PSNI) and SNL models [27,29].

As in the SNL model, an incision is made along the spinal column and the left paraspinal muscles are separated from the spinal processes at the L4-S2 levels. Under a modular high-performance stereomicroscope, the L5 spinal nerve is isolated and 1–3 mm of the nerve is excised distal to the dorsal root ganglia.

2.4 Chronic constriction injury

A chronic painful peripheral neuropathy develops after CCI surgery due to a peripheral infiammatory reaction in response to the ligatures. This is followed by a loss of most of the large myelinated Aβ-fibres, some myelinated Aδ-fibres and some small non-myelinated C-fibres [13]. Mixed signs of neuropathic pain and infiammatory components make the CCI model closest to mimicking neuropathic pain in humans [29]. The sciatic nerve receives input from L4, L5 and L6, which means that CCI affects a wider range of lumbar spinal cord levels, than does the more proximal SNT or SNL of L5 [27]. Thus, by affecting a wider range of lumbar spinal cord levels using CCI, it may result in a greater neurochemical and metabolic response than those in e.g. SNT [27]. CCI also leads to the development of allodynia-like behavior, hyperalgesia and spontaneous pain-like behaviors, which normally reach maximum 10–14 days after surgery [10,11,12,13]. CCI-operated animals develop allodynia-like behavior to cold and mechanical stimuli, and variable thermal hyperalgesia [30].

2.15 Partial sciatic nerve injury

In the PSNI model, allodynia-like behavior and hyperalgesia last for up to seven months [11,12,31]. In comparison to CCI, the PSNI model is associated with less infiammatory components [11]. Following injury, the animal develops a guarding behavior of the injured hind limb suggesting the possibility of spontaneous pain. Allodynia-like behavior to mechanical stimuli as well as thermal hyperalgesia and bilateral mechanical hyperalgesia are also developed [30].

A disadvantage concerning PSNI is the fact that an undefined number of neurons is ligated, making it complicated to relate the injury to a specific dorsal root ganglion [11]. However, this may be considered an advantage, as injury induced by PSNI might simulate neuropathic pain in humans where several dorsal root ganglions are included.