Sense and sensibility – logical approaches to profiling in animal models

The recent topical review by Rice and colleagues on sensory profiling in animal models of neuropathic pain rightly makes a call for back-translation [24]. The authors state “… here, we focus on aligning sensory measurements made in animal models with current methods of clinical sensory assessment”. Sadly, calls may fall on deaf ears, but many basic scientists in pain do aim for translation. So, we are a little surprised that other than a series of question marks in the table, electrophysiology is not discussed as an endpoint. Perhaps we are nerds and have grown rather attached to our neurones, but neuronal activity is the currency of pain and can be measured in the periphery, spinal cord and brain. Action potentials build the pain experience. It is an old technique – in 1975, Besson and colleagues showed a suppressive effect of morphine on spinal neurones [17], a year later, Tony Yaksh showed the behavioural consequences in rats [29], and in 1978, spinal morphine was shown to be analgesic in patients [28]. An example of fast and accurate translation.

A few years later, diffuse noxious inhibitory controls (DNIC), acting on spinal neurones, whereby one pain inhibits another via descending inhibitions were first reported [16], and now form the basis for conditioned pain modulation (CPM), able to predict the response to duloxetine and tapentadol [20,30], in perfect alignment with the pharmacological basis for DNIC [2]. Logically it should follow that novel drugs or strategies that target descending modulation can be tested using a DNIC paradigm, and then subsequently examined in stratified patient groups exhibiting low CPM.

Temporal summation of pain is often measured as a proxy for central sensitisation; wind-up potentiation of spinal neurones represents a neural substrate of this phenomenon [15]. Both processes exhibit similar pharmacological dependencies e.g. sensitivity to ketamine [1,11] and these potentiated neuronal responses appear to be important in patients [4].

There are inherent difficulties in comparing perceptual outcomes determined with QST to endpoints in animals; by definition pain in animal models is inferred. However, the classical and elegant studies by Ron Dubner and colleagues highlighted the relationship between the fine-tuned intensity dependent coding of dorsal horn neurones and perceptual outcomes [12,19]. Importantly, human electrophysiological and micro-stimulation data from lateral thalamic pathways are available for comparison [6,7,18,22]. Here we show one example - a thalamic neurone where both stimulus-evoked and ongoing activity can be quantified in the same population of neurones (Figure 1A). These applied stimuli are very similar to those modalities used in clinical sensory assessment, and have been used in animals by many groups studying peripheral and central neuronal activity. Electrophysiological approaches are able to measure responses of sensory neurones to a wide range of stimuli, many of which are not used in behavioural studies. Using laser stimuli [26], and perturbing the system with UVB/heat [21], we have reported that spinal neurones in rats code various stimuli in a very similar manner to human psychophysics including QST.

Figure 1.

Example single unit histogram trace of a ventral posterolateral wide dynamic range neurone in a neuropathic rat. Various modalities of increasing intensity can be applied and the responses quantified (A). These responses can be pharmacologically modulated and pregabalin exerts modality selective inhibitory effects in SNL rats (B) (adapted from [23]).

Electrophysiology has also been successfully used to decipher the neuronal effects of inherited human channelopathies in vitro [10]. Pharmacology can be done on neuronal responses, looking at the effects of clinically used drugs, novel agents or if both are lacking, electrophysiology can be done in transgenic animals. The example we give is from our attempts to back-translate from the effects of pregabalin in patients, allied to the work by Ralf Baron and colleagues in particular, to use sensory phenotypes as surrogates of mechanisms and move to a personalised treatment of their pain [14]. As can be seen (Figure 1B), there are selective effects of the drug on mechanical and heat stimuli, but no effect on ongoing firing. Notably, the largest inhibitory effect is against noxious punctate mechanical stimulation, so let’s see if this can translate. In an overall negative clinical trial, there is evidence that HIV patients with pinprick hyperalgesia had relief from the drug [27]. One proposal in the review is that animal models, like the patients in this clinical trial, could be classified according to their sensory profile. We provide an example of how this can be applied to the spinal nerve ligation model based on neuronal recordings, and other appropriate measures could be added (Figure 2A).

Figure 2.

A standardised representation of sensory gain in the spinal nerve ligation model in rats (A). Historical neuronal data from the ventral posterolateral thalamus was compiled from sham/naïve (n=84) and SNL rats (n=111) to determine ‘normal’ levels of sensory coding and the degree of sensory gain in SNL rats. Windup was determined by collating historical data from dorsal horn recordings from sham/naïve (n=66) and SNL rats (n=61). (B) Post hoc analysis of the standardised inhibitory effects of pregabalin on thalamic neuronal responses (n=15) [23], and dorsal horn neurones (WU only, n=9) [5] in SNL rats. Measures represent: B – brush, LTM – low threshold punctate mechanical, HTM – high threshold punctate mechanical, IC – innocuous cooling, NC – noxious cooling, W- warm temperature, NH – noxious heat, WU – windup.

The authors go on to recommend that we should ‘avoid anthropomorphising human emotions to rodents’; surely the converse extrapolation should also be true? The review includes 16 references on thigmotaxis and burrowing, but only one on conditioned place preference. Without convincing evidence that these former assays will forward translate, we would caution against widespread adoption. The assays might be sensitive to analgesics, but is it clear which neurological process is being studied and what is the relatable endpoint in humans? Using the power of CPP and neuronal recordings to study evoked responses some rather interesting findings with gabapentin on evoked and ongoing pain have emerged [3].

The authors rightly note that reflexive endpoints do not always lead to clinical efficacy and their role requires re-evaluation. One final point that those carrying out behavioural approaches have to consider. Pain is most often measured in humans by an analogue scale (i.e. a rating scale of 0–10). Behavioural responses to painful stimuli should arise at the pain threshold – so around 1- 2 on the scale and this is what is measured by reflex responses in animals. Electrophysiology in anaesthetised animals can continue to study responses up to their maximum – 10 on the scale and so has the potential to study the ‘neuronal load’ of pain levels of 6–7 that patients in trials often have. Here, we have a preclinical investigation of drug efficacy where neurones have their responses only modulated by pregabalin in an intensity and modality dependent manner (see figure 1B and 2B), whereas the drug ‘normalises’ withdrawal thresholds in a behavioural study [13]. The behaviour, only gauging responses to low intensities, is not translating to the partial effectiveness of the drug in some neuropathic patients, but the neuronal characterisations may provide more insight.

Aligning sensory testing in animals and humans is a logical and progressive step towards a mechanism-based rationale for identifying suitable treatments. However, it is imperative to align the most appropriate endpoints, otherwise we risk further failed outcomes for patients and undermine translational research. We passionately feel that neurones provide a powerful insight into pain processing. We wished to be brief here and have commented upon several other related points already [25]. We are finishing a back-translational study based on a sensory profiling investigation of oxcarbazepine in neuropathy (for some bedtime reading see [8] and [9]). Thank you for holding, your call is very important to us and we’ll get back to you shortly…