State of the Art ReviewsPain

# Lost but making progress—Where will new analgesic drugs come from?

See allHide authors and affiliations

Science Translational Medicine  13 Aug 2014:
Vol. 6, Issue 249, pp. 249sr3
DOI: 10.1126/scitranslmed.3008320

## Abstract

There is a critical need for effective new pharmacotherapies for pain. The paucity of new drugs successfully reaching the clinic calls for a reassessment of current analgesic drug discovery approaches. Many points early in the discovery process present significant hurdles, making it critical to exploit advances in pain neurobiology to increase the probability of success. In this review, we highlight approaches that are being pursued vigorously by the pain community for drug discovery, including innovative preclinical pain models, insights from genetics, mechanistic phenotyping of pain patients, development of biomarkers, and emerging insights into chronic pain as a disorder of both the periphery and the brain. Collaborative efforts between pharmaceutical, academic, and public entities to advance research in these areas promise to de-risk potential targets, stimulate investment, and speed evaluation and development of better pain therapies.

## INTRODUCTION

“Pain is a more terrible lord of mankind than even death itself” (Albert Schweitzer, 1931).

Pain is the most common reason people seek medical care (1). The Institutes of Medicine and the American Pain Society estimate that pain affects more than 100 million adults in the United States and costs about $635 billion each year in medical treatment and lost productivity (2, 3). These numbers will only increase as the world’s population ages. Current pharmacotherapy is dominated by well-established drug classes such as narcotic analgesics and nonsteroidal anti-inflammatory agents. These and other classes of drugs such as cyclooxygenase-2–selective inhibitors, anticonvulsants, and antidepressants are in widespread use for pain treatment. However, not all patients achieve meaningful pain relief, leaving a significant unmet medical need for new, safe, and effective treatments for both acute and chronic pain. The global market for current pain therapeutics is substantial, estimated at$50 billion in 2009, making pain an attractive therapeutic area to the pharmaceutical industry. This review considers issues that may limit, and approaches that may enhance, the probability of success in the discovery and development of new medicines for pain.

## PRECLINICAL CHALLENGES IN IDENTIFICATION OF PAIN TARGETS

### Target discovery for pain therapeutics

Pain is defined as an unpleasant sensory and emotional experience associated with actual or potential tissue damage, or described in terms of such damage. Tissue-damaging stimuli are detected by primary afferent neurons termed nociceptors. These cells innervate target tissues and convey excitatory signals to the spinal and medullary dorsal horns, and ultimately to many areas of the brain including the thalamus, somatosensory cortices (S1, S2), insula cortex, anterior cingulate cortex, periaqueductal gray, and other sites. Thus, tissue-damaging stimuli engage a distributed network of brain regions that transform nociceptive inputs into the human experience of pain (see below). The nociceptor has been a main target for new therapeutics because activation of these neurons reliably produces sensations of pain in humans (4, 5), and loss of function in these cells results in an inability to experience pain (6, 7).

As described in the accompanying Review (8), therapeutic targets for the development of new medications might be identified from studies that link gene mutations to altered pain sensation in specific patient populations (1, 9). Examples include calcium channels, where mutations have been identified in familial hemiplegic migraine (10) and migraine without aura (11), and sodium channels, where loss- and gain-of-function mutations in the SCN9A genes, encoding Nav1.7 voltage-gated sodium channels, can result in congenital insensitivity to pain or in abnormal pain such as erythromelalgia (1214). It is not yet clear, however, that targets identified through familial mutations will be useful targets for pain treatment in the general population; this possibility awaits testing in human trials [see accompanying Review (8)].

Targeting the peripheral nervous system as a valid strategy for treatment of pain is evident by the clinical effectiveness of local anesthetics as well as from examples abundant in nature. Capsaicin, the piquant ingredient in hot chili peppers, activates the TRPV1 cation channel expressed on nociceptors, resulting in sensations of heat and pain; blockers of this channel are under investigation as potential new treatments for pain. The venom of bark scorpions produces pain by activation of the Nav1.7 sodium channel and recruitment of a second sodium channel, Nav1.8, that is required for generating action potentials (15). Some mice have mutations in Nav1.8 that prevent its recruitment after Nav1.7 activation, inhibiting the usual pain from the venom of these scorpions. These evolutionary strategies suggest that both of these sodium channel subtypes could be viable targets for pain therapies in humans (15). Peptides from fish and worm-hunting marine cone snails (for example, ω-conotoxin) selectively target N-type calcium channels that are critical for nociceptive transmission. Ziconotide represents successful translation of one of these peptides to the clinic (16). Other peptides from marine species are also being assessed for their potential as pain therapeutics (1719).

Molecular targets and mechanisms relevant to human pain states have been identified both from preclinical studies and from drug discovery programs on the basis of human pathophysiology and pharmacology. The NGF-TrkA signaling pathway is important for pain sensation, as shown by the fact that NGF administration produces pain and increased sensitivity to painful stimuli (2022). Additionally, mutations in NGFB and NTRK1 can result in pain insensitivity. Antibodies to NGF are clinically effective in reducing pain, albeit with potential side effect liabilities (23, 24), and are being evaluated in clinical trials (25, 26). Intriguing data from 2500 volunteers in the Twins UK cohort provide preliminary support for genetic links to experimental, thermal, mechanical, and chemical pain stimuli, suggesting possibilities for drug discovery through phenotyping (27). The angiotensin pathway has recently been implicated in animal models and in human pain conditions (2830). Therapeutic agents targeting the AT2 receptor have shown clinical efficacy (31).

The triptan drugs, serotonin 1B/1D agonists (32, 33), were the first class of drugs developed specifically for migraine. More recently, the cardinal role of calcitonin gene-related peptide (CGRP) in migraine has been recognized, and efforts are in process to target this mechanism. CGRP receptor antagonists and anti-CGRP antibodies (34) that modulate CGRP actions and trigeminovascular activation during migraine attacks are clinically effective and at various stages of development (3436). Drugs that affect neurotransmitters such as reuptake blockers (37, 38) and mixed function molecules [(39, 40) and see below] target central pain modulatory pathways and represent significant advances in therapy. For example, serotonergic-noradrenergic reuptake inhibitors can inhibit persistent pain but have little effect on acute pain (41). These compounds likely act in part via endogenous descending pain modulatory circuits (see below) (42).

Despite these successes, choosing targets in the preclinical setting remains difficult, and ultimately, validation only occurs in humans. The de-risking of novel potential targets for drugs is not straightforward and relies on convergent data from multiple systems, ranging from in vitro assays to in vivo function to confirmation of efficacy in pain models. The biology of pain is complex and, because it is a primary survival mechanism, has considerable redundancy and overlap with other sensory functions. Consequently, modulation of single proteins, the strategy most common in drug discovery, may not produce the desired effects in the general patient population. Many of the currently used drugs act on multiple pathways, receptors, and targets. For example, nonsteroidal anti-inflammatory drugs act on multiple inflammatory pathways (43), and even highly selective drugs such as the monoamine reuptake blockers (37, 38) elevate levels of serotonin and norepinephrine, which in turn affect more than 20 serotonergic and adrenergic receptor subtypes. Mixed μ-opioid agonist/monoamine reuptake inhibitor molecules modulate separate central nervous system (CNS) biological pathways, each of which has proven clinical efficacy [tapentadol or tramadol (39, 40)]. The effectiveness of these multitarget drugs that engage several CNS mechanisms suggests that poly-pharmacological strategies for the development of new therapies should continue to be explored (see below).

Plasticity in circuits involved in emotional function has recently been suggested to be an important factor in promoting the transition from acute to chronic pain (44, 45), and it has been hypothesized that some individuals may be genetically predisposed to “chronification” (46). Chronic pain is often associated with comorbidities that are related to multiple CNS circuits. Despite the proven efficacy of analgesic drugs that affect multiple targets or mechanisms, few have achieved this by design. Mindfully addressing a network disease such as chronic pain will require new drug development paradigms and new drug screening approaches (4749). One strategy is to test mixtures that combine single-activity compounds for safety and efficacy. For instance, the combination of μ-opiate agonists with peripherally acting μ antagonists maintains effects in the CNS but avoids unwanted gastrointestinal adverse effects (50). This combination approach, especially for new chemical entities, may require rethinking of traditional drug registration processes. It may also complicate treatment of patients on multiple nonsynergistic or noncomplementary medications (51, 52).

An alternative method is to synthesize single multitarget drugs that can modulate multiple protein targets simultaneously (53, 54). This approach presents unique medicinal chemistry design challenges to optimize pharmacology at different targets within a single molecule. These combination agents would have an advantage over drug mixtures in getting regulatory approval because only one toxicological, metabolic, pharmacokinetic, and pharmacodynamic and safety profile would be required (55, 56). Whether developed by chance (39) or by design (57), these drugs have the potential to improve clinical outcomes (3740, 43, 5759). This strategy also may diminish side effect liability by modulating and stabilizing diverse neural circuits simultaneously. Combination therapies could also harness synergy or additivity: (i) subeffective or minimally effective doses of individual drugs could improve clinical efficacy for pain without enhancing side effects (51) or (ii) one component could inhibit the side effects of another analgesic component, allowing increased efficacy without penalty.

Nociception, pain, and translation

Most analgesics that are effective in humans also show some efficacy in preclinical models (60, 61). The reverse, however, is often not true. Preclinical investigations all too frequently identify pathways that modulate pain in animal assays, but molecules that target these pathways fail in clinical trials (6265). Examples include the substance P neurokinin-1 (NK-1) receptor antagonists (6668), the fatty acid amino hydrolase inhibitors that elevate endocannabinoid tone (69), and peripherally restricted cannabinoid CB1 receptor agonists (70). The reasons for such failures are not well understood but may include species differences in pain modulation pathways, that is, the molecular target identified in animals may not be a salient component of pain in patients. More recent strategies have emphasized starting with patients to identify approaches that can improve therapy and then “reverse-translating” to understand mechanism and refine molecules in the preclinical setting in a “bedside-to-bench-to-bedside” strategy.

Failures in clinical trials may also result from other, equally important factors. For example, there may be uncertainty in the underlying pathophysiological mechanisms promoting pain in specific patients, in the dose selected for the clinical trial, in the degree of target engagement required for efficacy, and in the complexity of the design of the clinical trial itself, including unexpected placebo response rates (see below). Strategies to test pain therapies in healthy volunteers as an early validation of molecule and mechanism may be flawed because normal healthy volunteers without disease may be poorly predictive. A straight-to-human strategy would, therefore, optimally test analgesic candidates in chronic pain patients in clinical trials. This approach is complicated by difficulties in selecting appropriate pain populations relevant to the mechanism being tested and would be associated with high early development costs. Thus, preclinical research will likely be needed to triage potential therapeutic targets for pain drug development. Animal- and cell-based assays can inform biochemical and pharmacological questions, provide pharmacodynamic endpoints, and estimate safety and mechanism-based and off-target side effect liability. As preclinical studies are likely to remain a requirement for drug development, approaches to improve their ultimate relevance to clinical outcomes are constantly being explored.

Although recent failures of drugs with new mechanisms in clinical trials have prompted suggestions that preclinical research does not yield good mechanistic insights to human pain (63, 71, 72), this conclusion oversimplifies complex issues. Current understanding of human sensory neurobiology is largely based on information gathered in parallel from both animal and human studies, as well as from astute clinical observations during pharmacotherapy of diverse clinical conditions. For example, the excitability of neurons and the neurobiology of nociception, including the essential role of sodium channels in the action potential (73), are highly conserved across species. The relevance of sodium channels to pain is validated by the preclinical activity and clinical utility of local anesthetics and by evidence from human genetic studies noted above identifying mutations in the Nav1.7, Nav1.8, or Nav1.9 channels that modulate human pain (8). The mechanisms by which these mutations alter the biophysical properties of the channels and resulting neuronal excitability can explain the biology underlying the clinical syndrome (74). Other examples of highly conserved mechanisms also translate across species including the transducers of noxious stimuli (thermal or chemical stimuli) that can elicit sensations of pain [such as, for example, TRP channels (TRPV1 or TRPA1)] (75, 76). The general anatomy of pain systems is also similar in animals and humans, having been demonstrated in humans and characterized mechanistically in animals (7779): for example, pain and touch pathways are segregated (80, 81), and descending pathways play a critical role in modulating pain. Despite these areas of concurrence, there is still a shortfall in translating preclinical therapeutic findings to patients. Improving preclinical “pain” assessment including reverse translation from the human condition and devising approaches with higher predictive value remains a challenge for the pain community. Such improvements could include modeling clinical conditions in a more effective manner, including consideration of disease duration, effects of previous analgesic use, treatment response or resilience in subpopulations of animals, sex differences, and other considerations. However, the greatest translational barrier lies in the distinction between nociception and pain (82).

Although activation of nociceptors can elicit sensations of pain, the relationship between nociceptor activation and the human experience of pain is not linear. The human experience of pain is multidimensional and complex, and influenced by many factors that include modulation by brainstem and cortical circuits that change over time with the continuation of the pain state. Ultimately, the context (for example, threats to survival, athletic competition, availability of rewards, etc.) in which nociceptors are activated and the ultimate value of responding to pain, or suppressing pain, for the organism determine the human experience of pain (83). Such factors determine, and can override, the biological consequences of a specific pharmacological mechanism, as recently demonstrated in human imaging studies by the lack of analgesic action of hidden administration of potent opiate agonists in humans (82, 84). To date, insufficient emphasis has been placed preclinically on understanding the circuits and mechanisms mediating affective dimensions of pain that are the most troubling to patients (85, 86). Recent advances in understanding evaluation of pain preclinically have studied the brain circuits that are most likely to underlie the various components that together make up the human pain experience. It is, of course, not possible for animal models to capture the human experience that is pain (87, 88), and these models were not developed for this purpose. Nevertheless, deconstructing the pathways involved in pain responses in animals may inform and focus future clinical investigations.

Nociceptive mechanisms for pain resulting from noxious stimuli identified preclinically generally translate well to humans, but these acute conditions are usually not the primary medical need. After injury, patients often develop allodynia (increased sensitivity to normally innocuous stimuli). This symptom can be reliably reproduced in animal assays and human volunteers, making the development of therapies to prevent this escalation in patients seem achievable. Less understood, however, is why some patients show successful resolution of their pain, whereas others go on to develop chronic pain. Injuries to nerves (for example, after surgery) produce a variable incidence of chronic pain; estimates show that some 10 to 50% of patients undergoing surgery ultimately experience chronic pain (89). The reasons for this are unclear, but may include the nature of the injury, the age of the patient, psychological status, and genetic differences in pain susceptibility among patients (90, 91). Possible reasons for resistance to the development of chronic pain in genetically similar animals (9294) are currently being explored. Combined with strategies aimed at understanding endogenous factors (95, 96) associated with the natural resolution of pain, as observed in many chronic neuropathic pain patients over time (97), such studies could point to genetic and epigenetic clues to identify those patients who are most at risk to develop (98) or limit chronic pain, thereby linking the laboratory to the reality of the clinical world. Indeed, resolution from injury may be an active process that engages signaling molecules such as the resolvins and the protectins (95, 99).

Chronic “spontaneous” or ongoing pain, which is apparently independent of an external stimulus (100), presents a huge challenge for preclinical pain drug discovery research. Our relative inability to measure and study ongoing pain in preclinical models likely has impeded the discovery of new therapeutics (101103). A number of innovative preclinical approaches, however, are being developed that may be more relevant to ongoing pain (104108). These include operant protocols that assess complex animal behavior such as, for example, place conditioning and avoidance (105, 107, 109, 110), self-administration of analgesic drugs (111, 112), pain-induced facial expressions of rodents (facial scales) (113), wheel running (114, 115), burrowing (116, 117), and vocalization (118, 119). Determination of the potential translational value of these approaches must await the clinical assessment of drugs aimed at targets identified with these approaches.

Preclinical studies of complex behaviors of animals with presumed chronic pain have pointed to brain circuits that may underlie the affective (aversive) components of human pain. Relief of pain, like relief of other aversive states, can be considered as a reward (120, 121). Clinical studies of experimental (122) and chronic pain (123, 124) have demonstrated that pain cessation has reward value. The mesolimbic dopaminergic reward pathway is activated after peripheral nerve block in injured animals (125), suggesting a neuroanatomical substrate for this effect. The responses of neurons within reward circuits to the onset or offset of nociceptive stimuli (126) are likely to serve as the neural basis of decisions of whether or not to respond to nociceptive stimuli based on context of safety and anticipated reward (83, 127). Such circuit-based analyses may be useful to detect and differentiate mechanisms specifically modulating pain from those that could induce overdrive in motivational processing in the absence of pain, such as the opioids (128).

## CLINICAL PAIN DRUG DISCOVERY CHALLENGES

Translational pharmacodynamic and pain biomarkers

Positron emission tomography (PET) is a nuclear imaging technique that uses radioactive tracers (i.e., ligands that bind to specific receptors), allowing measurement of target engagement of drug candidates. This technique can be used in humans and is an increasingly essential component of drug discovery programs (129). All too often, potential CNS drugs have apparently failed in the clinic, without an assessment of whether there was sufficient central target engagement, providing an inadequate test of the clinical hypothesis. PET imaging can demonstrate that clinical drug doses engage the desired target, allowing an adequate proof of concept within the safety and tolerability limits of the molecule. A trial that fails despite appropriate target engagement indicates that the targeted mechanism is not relevant to the tested pain condition. PET imaging is used in animal studies as well, revealing the exposure-occupancy relationships required to produce desired effects and facilitating molecule selection (that is, choosing the drug with the highest on-target occupancy combined with the lowest possible exposure to allow optimal safety margin).

Phase 1 clinical trials can also use pharmacodynamic biomarkers such as functional magnetic resonance imaging (fMRI) (measurement of blood flow as a surrogate for neural activity) (66) and pharmacological MRI (phMRI) imaging (fMRI to evaluate brain activity induced by a pharmacological agent) (130) to accelerate pain drug discovery programs (Fig. 1). Imaging of animals and humans reveals similarities in their responses to evoked pain (131, 132) and resting state activities (132), and phMRI shows correspondence of analgesic effects of drugs across species (133, 134). In addition, anatomical imaging shows parallel changes in morphometric measures of gray matter during neuropathic pain in rats (135) and humans (136, 137).

Establishment of disease biomarkers for pain or for analgesia requires validation for disease state and drug responses in well-characterized specific patient populations. In chronic pain conditions, the validation of biomarkers is complicated by the naturally variable and potentially reversible course of the disease. Pain is evaluated by asking patients or subjects to rate intensity on various scales (for example, numerical rating scale where 0 is for no pain and 10 for worst pain imaginable) (138, 139). These values are useful to evaluate the efficacy of therapeutics and can capture the intensity of pain, but they are not fully informative about mechanisms underlying pain. Platforms involving diagnostic assays and methodologies are being developed (140, 141) to discover biomarkers including cytokines and microRNA profiles (142, 143), brain metrics (44, 144, 145), cerebrospinal fluid and tissue changes (146), and skin biopsies (147, 148). These can potentially be used to monitor and track disease progression and therapeutic response in chronic pain syndromes. These markers could also be used in the future to follow disease-modifying effects of therapeutic interventions. For example, reversal of brain gray matter changes has been reported in patients with chronic pain after successful treatment or resolution of pain (149); predicting chronification of chronic back pain may be evaluated using functional imaging (150); genetic measures may predict pain sensitivity and persistence (91); and psychological states such as fear or catastrophizing may predict heightened pain after surgery (151), as well as persistent postsurgical pain (152).

Brain imaging itself is a promising biomarker for chronic pain (144, 145) because it can track functional and morphological changes in multiple brain systems that can be related to the pain condition in animals and humans (135, 153155). Recent work has begun to define a neural signature of experimental pain in human subjects (156). In this approach, also used for depression, activated brain areas are recognized by a computer algorithm (machine learning) to define a state (such as pain) that can then be used to evaluate other individuals’ responses. This method has proved to be accurate for experimental pain, but its sensitivity in patients with chronic pain is not yet clear. Although imaging can provide insights into CNS responses to noxious stimuli, pain is a subjective phenomenon, and at present, the only reliable biomarker for pain is pain itself.

## CHRONIC PAIN AS A BRAIN DISORDER

Peripheral or CNS damage can result in maladaptations to initial nociceptive sensory processing that transform to a state of chronic pain. In the latter, there are alterations in multiple brain processes including motivational, cognitive, physiological (for example, sleep wake), and/or hedonic networks (Fig. 2). To date, therapies for chronic pain that target peripheral (non-CNS) mechanisms have generally not produced complete, sustained pain relief (157). Recent efforts have investigated the effects of chronic pain on the brain itself. Patients suffering from chronic pain show alterations in cognition (158) and decision making (134). There is significant comorbidity between pain and other negative emotional states such as depression and posttraumatic stress disorder (PTSD) (159, 160). Many chronically ill patients suffer from a lack of interest in life activities and diminished reward (anhedonia) (161, 162). Imaging analyses show decreases in cortical volume in patients with chronic pain and other conditions including migraine, irritable bowel syndrome, and back pain (163166), which appear in some to be reversible when chronic pain is resolved (149).

Preclinical studies have begun to corroborate these clinical observations by demonstrating the consequences of acute and chronic pain on reward (125), attention (167), and dynamic (reversible) changes in brain morphology (149, 168). These approaches can potentially address the disconnect between the complex features of chronic pain in humans and the short-term mechanisms that are commonly evaluated in preclinical studies on animals. Understanding the long-term consequences of pain on brain circuits is critical to improving assessment of potential treatments for chronic pain in humans.

Detecting and reporting clinical effects

Patients in clinical trials of pain drugs are heterogeneous with respect to genetics (169), past experience, past medication or drug history, and duration of disease. These factors likely account for the fact that some patients respond to pain treatment, whereas others do not (103). Clinical trials—unlike animal studies, which have homogeneous subject populations—often group patients with heterogeneous pain phenotypes, and probably different underlying pain mechanisms, into broad pain classifications and indications, and then evaluate responses subjectively. Together, these factors can produce large statistical variance, potentially masking clinical activity in subpopulations of patients and thus potential efficacy in these subgroups. Prescreening methods that better define pain genotype and phenotype and enriched trial designs (170) will likely improve clinical signal detection.

Publication of data on failed drug trials is important to prevent redundant and uninformative research, enabling our limited drug discovery and development resources to focus on new targets with improved probability of clinical success (171). The registration of drug trials at clincaltrials.gov provides an opportunity for greater transparency, although detailed descriptions of the clinical protocols may not be available unless results are published in peer-reviewed journals. Timely publication of trial protocols and results, especially if several groups are pursuing the same targets, also supports ethical drug development by preventing the exposure of new volunteers and patient cohorts to drug classes that have not shown efficacy.

## CHALLENGES TO THE PAIN RESEARCH COMMUNITY

The National Institutes of Health (NIH) and other federal agencies have supported academic pain research for decades, and this has furthered understanding of fundamental pain neurobiology preclinically and clinically, generating a significant literature base for future investigations. Recently, however, academic pain research has been weakened by reduced NIH funding (declining budget allocations, reduced number of grants being supported, diminished growth of NIH funding, and decreased relative dollar value) (172, 173). This has been compounded by the exodus of many large pharmaceutical companies from analgesic development, decreasing collaborative resources, and access to sophisticated tools and technologies. Reinvigorating industry-academia relations is critical for future pain drug discovery (174).

The current productivity crisis in the pharmaceutical industry has spawned the necessity to develop new business models and approaches for open innovation (175) that link external opportunities to internal research and development. New risk- and cost-sharing consortia models that require a “cultural transformation” around intellectual property within both industry and academia (176) are now being explored (Fig. 3). These consortia promote precompetitive sharing of information, early-space interactions with academia, support for investigator-initiated study protocols, imaging consortia, and early assessment of drug candidates through biotechnology companies or newly emerging programs at large universities, private-public partnerships, as well as other national initiatives (Table 1).

Table 1 Initiatives for cooperative science and analgesic drug development.
View this table:

## SUMMARY AND NEXT STEPS

In a recent review on the global burden of disease, chronic pain (including tension headache, migraine, low back pain, musculoskeletal pain, osteoarthritis, and diabetic neuropathy) ranked as one of the most significant health issues—above common psychiatric disorders such as major depressive disorder and anxiety disorder—underscoring the compelling need for more research and investment into treatments for chronic pain (177) (Fig. 4).

Pain is one of the most important global health issues, contributing substantially to humankind’s burden of disease. As a complex brain disease with high unmet medical need, pain is still inadequately treated, despite significant investment in pain research and advances in understanding of its neurobiology. There are many contributors to this situation: (i) difficulties in preclinical target assessment, (ii) poor chemical druggability of targets validated in human studies, (iii) a paucity of clinically qualified biomarkers, (iv) a lack of translational assays (robust, reproducible measures that can predict efficacy across species or from healthy to chronic pain conditions) to guide proof-of-concept testing, (v) poor target engagement of molecules, and (vi) small effect sizes in generalized pain populations that increase the difficulties of informative clinical trials. Drug development is challenging for many CNS disorders (178182). The problems of analgesic drug development are set in this bleak landscape and the sad context of a retreat from CNS drug discovery by large pharmaceutical companies, despite regulatory encouragement (183). Analgesics often require costly proof-of-concept clinical trials and have a low probability of success (184) compared to other disease areas. Although pain may be considered to be no different than other complex neurological or psychiatric disorders in its translational difficulties, our failure to produce new pain therapeutics is particularly disappointing given our substantial understanding of pain pathways and pain neurobiology.

The de-risking of drug development in academia (185) and industry is essential to encourage further investment in new pain therapies, a need that is fueling the generation of standardized guidelines for analgesic drug development in industry (186, 187). Public-private consortia are forming to enhance clinical pain research, particularly with scientifically validated and clinically qualified human experimental pain models, including patients (188). Such consortia can define new clinical pain paradigms as well as common standards for data capture, processing, and evaluation, so that methodologies and analytical techniques can be easily shared between academia and industry (189, 190). These collaborative efforts (191) will require us to overcome diverse societal, financial, and personal interests to enable rapid progress. Patients are waiting for more effective treatments for pain.

There are, nonetheless, many reasons for optimism. The human genetics of pain are helping to inform decisions on drug discovery targets. Knowledge of the fundamental adaptive changes in the peripheral and CNS that underlie the progression, and possibly the resolution, of pain is growing. New translational efficacy models that bridge the gap between the laboratory and the clinic by the use of common endpoints are being developed and validated through known pharmacological mechanisms. Improvements in imaging of brain pain circuits are progressing rapidly. Biomarkers are being identified that can help to prioritize early decision making and ensure that mechanisms and clinical hypotheses are truly tested. Finally, our understanding of preclinical and clinical chronic pain biology continues to increase, and new approaches promise success in disease modification.

## REFERENCES AND NOTES

1. Funding: Supported in part by grants from the NIH (F.P. and D.B.). Competing interests: The authors declare that they have no competing interests.
View Abstract