To the dismay of millions of chronic pain sufferers, the search for novel analgesics over the past several decades hasn’t produced much relief. In fact, it’s been one big pain in the…well, you know.

Littering the field is a number of high profile failures that initially showed promise in research models, but flopped in clinical trials. Consider Pfizer’s FAAH1 inhibitor, designed to elevate a brain molecule that suppresses pain. It failed to relieve chronic knee pain in arthritic patients. AstraZeneca’s TRPV1 antagonist, which blocks a key protein enabling us to sense heat and feel pain, also showed little efficacy in relieving osteoarthritis or dental pain. Even more surprising, perhaps, were the results seen with several different NK1 receptor antagonists that performed poorly in clinical pain studies, despite proving effective for depression and emesis.

So what’s going on?

Researchers cite several reasons for the poor clinical outcomes including inadequate clinical trial design, high placebo responses, suboptimal compounds to test the target mechanism and poor animal model translatability. In all likelihood each of these reasons are partly to blame for the mediocre findings, but it’s the lack of translatability of the animal models that is occupying the spotlight these days. This is because translatability in the pain field faces significant challenges; the heterogeneous nature of the chronic pain population, the highly subjective experience of a chronic pain patient and differences in how we measure pain between animals and patients.

Older Models’ Limitations
Traditionally, preclinical animal pain models have largely focused on quantifying ‘reflexive’ withdrawal responses in rodents when thermal or mechanical stimuli are applied.  Such tests are quick and easily measured and thus have been widely adopted by academia and industry alike.

However, it is now widely accepted in the pain community that these types of animal models/measures have limitations. One of the major disparities is the way how pain is measured in the clinic versus how pain is measured in animals using these reflexive tests. In clinical studies patients most commonly log spontaneous/ongoing pain as their chief symptom whilst preclinical pain models have focused predominantly (because they are easy!) on thermal hyperalgesia –heightened sensitivity to heat or cold—and mechanical allodynia, such as heightened sensitivity to brush or prick. As a result, more attention is being given to finding ‘translatable’ pain models and ways of measuring pain.

While these changes could potentially lead us to better results in human trials, updating the models will require adjustments for researchers doing the preclinical testing.

Switching to an animal model that measures spontaneous pain, while perhaps a more ideal setup, won’t be easy, though. As animals can’t talk or fill in pain questionnaires, we need to find innovative ways to measure ongoing pain.

In Search of Translation
So, labs are looking at a number of innovative approaches to measure spontaneous pain, either directly or indirectly, in animals. Some of these measures include conditioned responses such as Conditioned Place Preference (memory of analgesic effects as manifested by preference for a chamber), measuring spontaneous pain behaviors such as flinching, gait/limb weight-bearing changes, burrowing and even pain-induced facial changes as described by Jeff Mogil’s lab at McGill University.

As chronic pain is broadly categorized as nociceptive— activation of pain-specific receptors usually due to tissue injury—and neuropathic (due to nerve injury), researchers have focused on establishing models in these areas. However, there are challenges in replicating all aspects of a painful clinical condition in an animal model. In contrast, certain acute experimental pain models—ones  using translatable stimuli such as capsaicin, UV burn or mustard oil—can be duplicated in Phase I study volunteers. As such, they can be invaluable in quickly and cost-efficiently understanding pharmacokinetics/pharmacodynamics prior to conducting more lengthy and costly Phase II trials.

Standard pain measures are also notoriously subjective in nature, which can create problems if you want to validate findings between researchers or across labs. Where possible, a greater use of non-subjective pain measures should help to reduce experimenter bias. The increased use of imaging endpoints, video capture and automated data analysis are all examples of technological developments moving the field in this direction. Together with blinded studies—a must for all pain studies—these improvements in study design should help reduce experimenter bias.

Of course, traditional models and ways of measuring pain still have their place in the lab. For instance, they can be used to screen and rank the in vivo potency of compounds, but they should also be complimented with newer-generation ‘more translatable’ models.  Recent advances suggest more clinically-relevant animal models and measures of pain can be developed, which might help overcome some of the negative findings from drug trials and lead to the development of more novel treatments.

And help legions of pain sufferers sleep a little easier.