An industry perspective on the role and utility of animal models of pain in drug discovery

G

N

M

Ap

GQ1

a

b

c

h

•••••

a

ARRA

KPTA

C

0h

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

ARTICLE IN PRESS Model

SL 29992 1–8

Neuroscience Letters xxx (2013) xxx– xxx

Contents lists available at ScienceDirect

Neuroscience Letters

jou rn al hom epage: www.elsev ier .com/ locate /neule t

ini-review

n industry perspective on the role and utility of animal models ofain in drug discovery

arth T. Whitesidea, James D. Pomonisb, Jeffrey D. Kennedyc,∗

Discovery Research, Purdue Pharma L.P., 6 Cedar Brook Drive, Cranbury, NJ 08512, United StatesAlgos Preclinical Services, 2848 Patton Road, Roseville, MN 55113, United StatesNeuroscience Discovery, Eli Lilly and Company, Lilly Corporate Center, Indianapolis, IN 46285, United States

i g h l i g h t s

Evidence that new pain drug discovery failures are due to insufficient efficacy is poorly documented.Discovery of new drugs requires data from many assays in addition to behavioral models.Pain models are used in drug discovery to rank order compounds and focus resources.Use of new pain models/endpoints to improve translational success first requires their validation.Pain model data analysis using effect size and NNT may create better alignment with clinical data.

r t i c l e i n f o

rticle history:eceived 1 July 2013eceived in revised form 14 August 2013ccepted 17 August 2013

eywords:redictive validityranslational modelsnalgesic drug development

a b s t r a c t

In recent years, animal behavioral models, particularly those used in pain research, have been increasinglyscrutinized and criticized for their role in the poor translation of novel pharmacotherapies for chronicpain. This article addresses the use of animal models of pain from the perspective of industrial drugdiscovery research. It highlights how, when, and why animal models of pain are used as one of the manyexperimental tools used to gain better understanding of target mechanisms and rank-order compoundsin the iterative process of establishing structure–activity relationships (SAR). Together, these models helpcreate an ‘analgesic signature’ for a compound and inform the indications most likely to yield success inclinical trials. In addition, the authors discuss some often under-appreciated aspects of currently used(traditional) animal models of pain, including how industry balances efficacy with side effect measures as
part of the overall conclusion of efficacy. This is provided to add perspective regarding current efforts todevelop new models and endpoints both in rodents and larger animal species as well as assess cognitiveand/or affective aspects of pain. Finally, the authors suggest ways in which efficacy evaluation in animalmodels of pain, whether traditional or new, might better align with clinical standards of analysis, citingexamples where applying effect size and NNT estimations to animal model data suggest that the efficacybar often may be set too low preclinically to allow successful translation to the clinical setting
© 2013 Published by Elsevier Ireland Ltd.

ontents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 002. Current status and new developments in animal models of pain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 003. The use of animal models of pain in an industry setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 004. What industry needs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 005. Recommendations/path forward for the future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 006. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00

Please cite this article in press as: G.T. Whiteside, et al., An industry perspecNeurosci. Lett. (2013), http://dx.doi.org/10.1016/j.neulet.2013.08.033

Conflict of interest statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

∗ Corresponding author. Tel.: +1 317 655 1731; fax: +1 317 276 7600.E-mail address: [email protected] (J.D. Kennedy).

304-3940/$ – see front matter © 2013 Published by Elsevier Ireland Ltd.ttp://dx.doi.org/10.1016/j.neulet.2013.08.033

tive on the role and utility of animal models of pain in drug discovery,

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00

dx.doi.org/10.1016/j.neulet.2013.08.033


http://www.sciencedirect.com/science/journal/03043940

http://www.elsevier.com/locate/neulet

mailto:[email protected]


ING Model

N

2 cience

1

acWcednfhaotbolpfia

dwoelbtnhtpsttnwcptnmdWlp

2o

tapeacstabbofs

38

39

40

41

42

43

44

45

46

47

48

49

50

51

52

53

54

55

56

57

58

59

60

61

62

63

64

65

66

67

68

69

70

71

72

73

74

75

76

77

78

79

80

81

82

83

84

85

86

87

88

89

90

91

92

93

94

95

96

97

98

99

100

101

102

103

104

105

106

107

108

109

110

111

112

113

114

115

116

117

118

119

120

121

122

123

124

125

126

127

128

129

130

131

132

133

134

135

136

137

138

139

140

141

142

143

144

145

146

147

148

149

150

151

152

153

154

155

156

157

158

159

160

161

162

ARTICLESL 29992 1–8

G.T. Whiteside et al. / Neuros

. Introduction

Discovering and developing novel drugs for use in humans isrduous. Obstacles are present at many levels, including biology,hemistry, intellectual property, and regulatory considerations.

hen pursuing unprecedented targets, these obstacles are asso-iated with even greater risk. As such, only a fraction of preclinicalffort will translate to successful clinical studies, a challenge forrug discovery in any therapeutic arena. Although pharmacoki-etic parameters were once a main reason for clinical development

ailures, this is no longer the case [19]. In recent years, criticismas been widely levied against the animal models used in researchnd development, and specifically regarding the predictive utilityf animal models of pain [4,23,27,39]. The widespread belief thathese models have limited or no translational value comes fromoth academia and industry, with many suggesting that the paucityf new analgesic drugs results from animal model data that are mis-eading in their conclusion of efficacy and/or poorly reflect clinicalain signs and symptoms [3,33,42]. With that said, it is the authors’rm belief that the current translational challenges should not inny way lessen the value of, or confidence in, animal models of pain.

Drug development efforts fail for numerous reasons: toxicology,ose-limiting side effects, failure to show improvement or other-ise differentiate versus SOC drugs, or poor selection of indication

r patient cohort, among others. While a few examples exist wherefficacy demonstrated in animal models of pain has failed to trans-ate to clinical efficacy [11,13,28], in general, such failures haveeen poorly documented with little published data. Thus, defini-ive conclusion that animal models yield ‘false positive’ data isot clearly supported. Likewise, it is all-but certain the converseas never been tested, namely identification of a ‘false negative’hrough clinical trial of a mechanism that failed to show efficacyreclinically. In this light, it’s worth noting that there have beenuccesses in translating preclinical efficacy to the clinic, includinghe approval of ziconitide, the still-evolving tanezumab story, andhe more recently published successful trial of a TRPV1 antago-ist compound, albeit in the context of third molar extraction [32];hether TRPV1 antagonist compounds prove efficacious in more

omplex, chronic pain conditions awaits data from further clinicalain studies. In this article, the authors put forward a current indus-ry perspective on existing, ‘traditional’ models of pain as well asew models being developed. We offer a rationale for how animalodels are used across the drug discovery process, one that may

iffer in some notable aspects from their use in academic research.hile recognizing the limitations of these models, we hope to high-

ight some of the current misconceptions around animal models ofain and suggest for consideration potential improvements.

. Current status and new developments in animal modelsf pain

Numerous animal models of pain have been designed as a meanso investigate mechanisms underlying nociceptive, inflammatory,nd nerve injury pain. It is beyond the scope and intent of thisaper to describe these models, their methodology, variations, andndpoints, but the reader is directed to reviews by Mogil [25]s well as Joshi and Honore [17] for more details. Most modelsouple a method for inducing a hypersensitive state (the model)uch as mechanical trauma or injection of an algogenic substanceo a behavioral assessment (the endpoint). Endpoints tradition-lly include either direct observation of non-evoked, spontaneousehaviors such as flinching, licking, biting or altered weight-


earing, or evoked responses such as paw withdrawal, vocalizationr struggle following application of a stimulus (heat, blunt pressure,ocused tactile probing, etc.). These behaviors are believed to repre-ent pain being experienced by the animal and have been referred

PRESS Letters xxx (2013) xxx– xxx

to as “pain stimulated” [38]. Advances are being made in estab-lishing disease models to represent osteoarthritis, fibromyalgia,post-operative, visceral, and thermal injury pain, among oth-ers. Likewise, the methodology for developing new endpoints isgrowing and now includes an array of measures such as elec-troencephalograms [10], alterations in sleep [22,37], movement(thigmotaxis) [12] or gait [30], changes in social or ‘well-being’behaviors such as burrowing [1], and choice (preference, aversion)paradigms [14,18,40], intracranial self-stimulation and other pain-depressed behaviors [26,29], even facial expressions [20]. It is safeto say that hundreds of model/endpoint combinations are now pos-sible and being described in the literature.

Interest has also grown in the development and implementationof large animal and/or naturally developing (‘naturalistic’) mod-els of pain based on the assumption these will show greater face,construct and, ultimately, predictive validity. There is support forthe view that new models utilizing novel endpoints are needed inorder to overcome the current translational impasse in the devel-opment of novel pain drugs [20,23]. A number of academic andindustrial groups are accelerating efforts around, for example, dogmodels of arthritis and nonhuman primate models of inflamma-tory or nerve injury pain. Although these models ultimately mayreveal insights into pain behaviors not readily apparent in rodentsand provide translational benefits from both PK/PD and toxicol-ogy perspectives, they also present a number of challenges, notthe least of which relates to greater heterogeneity in the manifes-tations of ‘spontaneous’ pain relative to that induced in rodentsby directed means. While rodent models utilizing homogeneousage, sex, weight, and strain are criticized as not being represen-tative of clinical pain, naturalistic models in larger animals makeestablishing robust endpoints and reproducibility between exper-iments much more difficult, even if they may more closely mimichuman-like disease heterogeneity. Power calculations for com-pound assessments in a typical behavioral study over a 3-point (½log) dose response typically suggest group sizes of 8–10 animals;including positive and negative control groups, this approaches atotal experiment size of 50 animals, a number not feasible whenusing larger animals. In addition, because injury and, hence, painseverity is not controlled as it is when applying a uniform insultto groups of rodents, intra- and inter-group variability is likelyto obscure conclusions of statistical efficacy. Assays using ther-mal or mechanical stimuli to evoke nocifensive responses in largeanimal models of pain are the same as those criticized for use inrodent models. Unfortunately, measurement of non-evoked painendpoints in larger species is still rudimentary and highly variable,although advances are being made [5,43]. The caution is to main-tain perspective regarding the fact that animal models, even thoseutilizing non-human primates, represent, at best, only an approx-imation of human biology and behavior. In an almost paradoxicalway, many, while acknowledging this gap, remain keen to anthro-pomorphize animal behavior.

Efforts to generate new behavioral data, particularly thoserevealing insights into cognitive and affective aspects of pain, arecertain to enrich our understanding of pain pathobiology and mayultimately increase translational success; however, in our view,there are other gaps in the translational chain more readily acces-sible and likely to yield positive results in the short term. Theseinclude better patient phenotyping and stratification, implemen-tation of trial designs that may help minimize placebo effects, and,in the preclinical realm, setting the bar higher with respect to whatis viewed as efficacious enough to merit advancement into devel-opment (discussed more below).


Gaining confidence in the predictive validity of any behavioralassessment is a lengthy process. Ideally, compounds proven clin-ically efficacious should have entered into development based, inpart, on robust efficacy in an animal model(s). Establishing such a

163

164

165

166


ING Model

N

cience

ctecydtbmltnvfnte[ntdtd[

3

mwtpcFcteeieammc

jt

FiiP

167

168

169

170

171

172

173

174

175

176

177

178

179

180

181

182

183

184

185

186

187

188

189

190

191

192

193

194

195

196

197

198

199

200

201

202

203

204

205

206

207

208

209

210

211

212

213

214

215

216

217

218

219

220

221

222

223

224

225

226

227

228

229

230

231

232

233

234

235

236

237

238

239

240

241

242

243

244

245

246

247

248

249

250

251

252

253

254



onnection for a new model could take many years. While back-ranslation is more rapid and provides some level of confidence instablished mechanisms, this is not possible if no clinically effica-ious treatments are available or for new mechanisms that haveet to be tested clinically. Only time and significant experience willetermine whether new models represent an improvement overhose currently in use. It is our opinion that any new model shoulde used in addition to, and not as a replacement for, establishedodels. All models are ‘predictive’ for some aspect of pain; the key

ies in determining which model/endpoint combination, whetherraditional or new, best reflects the pharmacology and clinical phe-otype of interest. Moreover, as more extensive pharmacologicalidation using new models accumulates, we would suggest that aailure to align efficacy across these and the traditional models doesot necessarily invalidate either but instead adds greater deptho our overall understanding of pain pathophysiology. Indeed, ifndpoints do align, as recently found for von Frey and burrowing21], the challenge for the field is to justify and demonstrate thatew models constitute an improvement as predictors of transla-ional success. The true test of this will come only with successfulevelopment of an unprecedented mechanism; we already knowhat clinically efficacious pain drugs back-translate into the tra-itional models both with respect to exposure and model type44].

. The use of animal models of pain in an industry setting

Animal efficacy models constitute but one step in an array ofolecular, cell- and tissue-based biological assays that in concertith synthetic chemistry are used to screen many molecules and

hen to more thoroughly characterize smaller numbers of com-ounds with the goal of identifying one, or a few, candidates forlinical development. There are many steps in this process (seeig. 1), including creation and validation of cell lines used to con-lude appropriate activity at, and selectivity for, the moleculararget, pharmacokinetic assessments both in vitro and in animals,lectrophysiology assays, side-effect profiling, biomarker, targetngagement and other ‘translational’ models, as well as toxic-ty assessments. Together, these assays and models are used tostablish ‘converging lines of evidence’ in support of a target mech-nism’s role in pain and to build confidence that a candidateolecule, whether protein or small organic compound, by robustlyodifying the target’s activity, will demonstrate activity against a


linical pain condition(s).A lack of translatability of any of the aforementioned assays, not

ust in vivo efficacy assessments, may lead to a negative clinicalrial. A discussion of the ‘predictive validity’ of cell based assays

ig. 1. A simplified flow chart representing the use of biological assays, including in vivos indicated, although this will vary by company and project. Compounds move down “tnformation is fed back in order to improve the characteristics of newly synthesized compK, pharmacokinetics, MMs, millions.

PRESS Letters xxx (2013) xxx– xxx 3

would be counter-intuitive, as there is an appreciation that suchassays are highly contrived, artificial, and serve primarily as ameans to allow researchers to rank-order compounds accordingto a limited set of criteria. In contrast, such acknowledgment ofassay limitations is rarely, if ever, extended to the interpretationof data from animal models of pain, which are also contrived andartificial. It is sometimes underappreciated that in industry, thesemodels are used not only to determine efficacy interrogated inthe context of some insult, but also play critical roles in generat-ing credible dose–response curves that can be used to establishPK/PD relationships and help rank-order compounds as part of theSAR paradigm. These assays and models ultimately enable focus ofdevelopment resources onto those compounds with the greatestchance of achieving clinical success. As such, reproducibility andprecision are critical and mitigate the criticism of using homoge-neous (e.g., age, sex, and weight) experimental groups.

Because there are numerous combinations of model and end-point, it is not sensible to group all under a single heading of ‘animalmodels of pain’ and deem the lot neither valid nor predictive; amore specific analysis is warranted. For example, a simple nocicep-tive pain model such as tail flick is responsive to several classes ofcompounds given intrathecally in addition to systemically admin-istered opioids. If developing a new opioid drug, this assay maybe highly predictive of clinical efficacy. In addition, such a model,among the simplest in the spectrum of model/endpoint combi-nations, may even allow for accurate determinations of expectedefficacious exposure levels across a dose range, as supported bycomparative analysis of other molecules from the same class androute of administration for which both preclinical and clinical dataexist. Consider Fig. 2, which provides the authors’ opinion on theconfidence levels in pain models for predicting specific pharma-cology. Note that we have higher confidence, albeit for specifiedmechanisms, in the models with the longest history. This is in linewith our posit that while new models may constitute improve-ments, ultimately leading to greater predictive validity, this willtake years to establish.

The corollary is also true, in that currently used models mustbe used appropriately to have the best chance of providing a solidtranslational bridge to clinical pain. For example, the CFA modelmost closely recapitulates the etiology and symptomatology ofinflammatory arthritis conditions such as RA yet is often usedas a bridge to OA. It is unreasonable to expect efficacy achievedin the CFA model to accurately predict clinical efficacy in OA let


alone in animal models of OA. Furthermore, because clinically usedcompounds display efficacy against traditional models/endpoints,the development of new therapies will require they, too, betested in those assays; industry decision-makers and, ultimately,

assays, in industry. The range of compound numbers typically tested at each tierhrough the funnel” if pre-specified criteria in the biological assays are met, whileounds and increase the chances of a compound being advanced into development.

255

256

257

258


ARTICLE IN PRESSG Model

NSL 29992 1–8

4 G.T. Whiteside et al. / Neuroscience Letters xxx (2013) xxx– xxx

Neuropathic pain(trauma�c)

Acute pai n

Surgical pain

Osteoa rth ri� s

Inflammatory Pain

Chron ic Lo wBack pain

Fibromyalgia

Hot Plate / Tail flick(1952/1941)

Paw Incision(1996)

Carr / CFA(1964/1988)

SNL(1992)

MIA / MMT(1997/2002)

DRG compre ssion (1 998 )Lumbar canal stenosi s(1995)

Reserpine-inducedcatecholamine depl e�o n(2009 )

Opioids

Opioids / NSAIDS

Opioids /N SAIDS / C OX-2 s

Gabapen�noids, an�-convulsants, an�depressants

Lackin g

Lackin g

Gabape n�noids, an�dep ressant s

Indica�on/ Dis eas eStrength of Pharmacolo gical Vali da�o nAnimal Model

Fig. 2. Authors’ opinion on confidence levels regarding predictive validity across animal pain models in 7 distinct areas for the pharmacologies noted. The dates the assaysw Thosed , comM

pbppaiaTaPeabhipits

iteataatstblmdemcs

259

260

261

262

263

264

265

266

267

268

269

270

271

272

273

274

275

276

277

278

279

280

281

282

283

284

285

286

287

288

289

290

291

292

293

294

295

296

297

298

299

300

301

302

303

304

305

306

307

308

309

310

311

312

313

314

315

316

317

318

319

320

321

322

323

324

325

326

327

328

ere first described are in parenthesis. Confidence decreases from top to bottom.evelopment or continued validation of those models noted. Carr, carrageenan; CFAMT, medial meniscal tear; DRG, dorsal root ganglia.

ayors demand that any new drug differentiates from SOC drugsy demonstrating some aspect of superiority. And as additionalreclinical endpoints are developed, their sensitivities to differingharmacologies will have to be established. These are unlikely tolign perfectly [15], given that sensory, affective and cognitive man-festations of pain have different, if overlapping, neural substratesnd so would be expected to respond differently to treatments.hus, sorting out the relative importance of one endpoint versusnother will further complicate the ability to establish predictiveK/PD relationships. The trend toward increased use of evokedndpoints in the clinic, including QST, as a means to diagnosend stratify patient subpopulations [2] may also inform alignmentetween clinical and preclinical endpoints if these can be shown toave utility for assessing drug efficacy. Our prediction is that align-

ng preclinical model/endpoint with well-circumscribed diseasehenotypes and as well expanding the use of evoked endpoints,

n addition to currently used patient-reported outcome measureso assess clinical drug efficacy, will improve overall translationaluccess.

In addition to conducting experimental blinding and random-zation, establishing dose response curves to determine whetherhey inform/confirm both mechanism and PK/PD relationships, andnsuring motor deficits are ruled out as potential confounds againstccurate conclusions of efficacy, all of which we view as essen-ial [44], we further advocate the use of multiple models, species,nd endpoints (including non-evoked) combined with models offfect, cognition, and other behaviors that together are integratedo create an ‘analgesic signature’ for a given mechanism. Thiserves the purpose of not over-emphasizing any one model orechnique, given clear evidence from both clinical experience andack-translation studies that no drug/mechanism yields equiva-

ent efficacy in all models or on all endpoints; hence, no singleodel or technique should be used to drive a drug discovery and

evelopment effort. In addition, establishing efficacy across a vari-


ty of models reduces translational risk associated with the targetechanism, and observed variability in potency or maximal effects

an help inform both clinical trial design and patient populationelection.

areas highlighted in the red box are seen as areas of opportunity for new modelplete Freund’s adjuvant; SNL, spinal nerve ligation; MIA, monosodium iodoacetate;

4. What industry needs

The endgame for pharmaceutical companies is to bring newdrugs to market. The steps required to do so are numerous andspan, on average, 12 years or longer and involve a number of sci-ence and business functions. Drug discovery is often describedas beginning with target identification, proceeding to target val-idation, and culminating in identification and production of adevelopment candidate/potential new therapeutic agent. Each stepinvolves ‘go/no-go’ criteria that must be met in order to move tothe next step. As a consequence, the tools employed at each stepare likely to change as the process progresses, including the use ofanimal models.

In the target identification phase, animal models are very impor-tant in order to monitor gene expression or protein levels in specifictissues during disease progression. These studies utilize multiple,often complex, models to interrogate the role of a protein in theunderlying biology. In the target validation phase, genetically mod-ified animals may be used to ‘validate’ a target mechanism’s rolein pathobiology by determining the consequence of disrupting itsfunction; furthermore, at this stage, if any probe molecules areavailable to assess pharmacology, they typically exhibit poor drug-like properties. Upon reaching a decision to initiate a drug discoveryprogram aimed at a particular target, industry needs suitable ani-mal models of pain amenable to screening tens to hundreds ofcompounds in a manner that will enable description of efficacy andpotency in a dose-responsive fashion, pharmacokinetics, pharma-codynamics, and side effects, all within a time frame of months andwith a high degree of reproducibility and precision (see Fig. 1). Oncea suitable candidate is identified using screening and second tiermodels, industry frequently conducts additional compound charac-terization, including use of novel models/endpoints, that may morespecifically model a particular pain syndrome and may help informthe clinical conditions that should be targeted.


As such, the needs of the pharmaceutical industry vary over thecourse of the drug discovery process. At the target validation stage,the cost, throughput, and feasibility of pharmacological manipula-tion are of lesser importance than at later time points. Conversely,

329

330

331

332


ING Model

N

cience

oapiawipsd(tecsp

5

mtpcmt

rsdrppslmpi‘itiaifsat

mmvdcamrcs

tittpp

333

334

335

336

337

338

339

340

341

342

343

344

345

346

347

348

349

350

351

352

353

354

355

356

357

358

359

360

361

362

363

364

365

366

367

368

369

370

371

372

373

374

375

376

377

378

379

380

381

382

383

384

385

386

387

388

389

390

391

392

393

394

395

396

397

398

399

400

401

402

403

404

405

406

407

408

409

410

411

412

413

414

415

416

417

418

419

420

421

422

423

424

425

426

427

428

429

430

431

432

433

434

435

436

437

438

439

440

441

442

443

444

445

446

447

448

449

450

451

452

453

454

455

456

457

458



nce a target has been identified and preclinically validated, thenimal models used must be chosen based on relevance to theharmacology under investigation but also on throughput, sensitiv-

ty, ability to determine relative potency, reproducibility, and cost,mong others. The final stage of characterization is seen as adding tohat is known for a particular compound and enriching the preclin-

cal dataset. What might be considered an appropriate model in oneart of the process may not be suitable for other parts. Ultimately,uccessful drug discovery and development is not predicated oneveloping the ‘holy grail’ of animal models of neuropathic painor any other painful condition), but more-so on understandinghe benefits and limitations inherent to all the assays and mod-ls that together are used to establish confidence in a target andompounds so that data-driven decisions can be made when con-idering advancement of candidate compounds at each stage of therocess.

. Recommendations/path forward for the future

Our goal is to identify both strengths and weaknesses of ani-al models of pain in order to spur additional discussion, leading

o further innovation and progress. As part of the discussion, weropose recommendations for a path forward based on two mainomponents; the first pertains to the need for uniform reporting ofethodologies, the second involves data analysis and interpreta-

ion.We are not the first to call for uniform reporting standards and

efer the reader to published recommendations [33,34,41] thatpecify the types of information that should be included whenescribing in vivo experiments. Note that neither those nor theecommendations in this article call for uniform experimentalrotocols but for uniform reporting standards and enhanced trans-arency. Methodology variations across animal models of pain,uch as injecting different volumes and concentrations of forma-in, or ligation of the L5 rather than L5 and L6 spinal nerves,

ay ultimately represent a (small) step toward approximating thehenotypic heterogeneity of clinical trial subjects. Indeed, exper-

mental details should be considered important variables (neitherright’ nor ‘wrong’) but must be clearly described for anyone wish-ng to repeat a study. Not only is replication a key component ofhe scientific method, it is absolutely critical for making rational,nformed decisions in drug discovery. If data cannot be reproducedcross laboratories or across separate studies (to a level allow-ng consistent conclusions to be drawn), enthusiasm to proceedurther with development activities will be greatly diminished. Inhort, confirmation of experimental findings across laboratories isn often overlooked factor impacting predictive validity, yet onehat has been noted to be disturbingly poor [16,31].

Although we agree on the importance of continued refine-ent of currently used models as well as development of newodels and endpoints, we also assert that current approaches pro-

ide significantly more information than is often utilized. Efficacyeterminations driven solely by accepting statistically signifi-ant increases in response thresholds can lead to false positivesnd obscure understanding of relative efficacy. To this end, ani-al model data assessed through the lens of outcomes routinely

eported in pain clinical trials such as effect size and responder ratesan provide a broader and potentially more translatable under-tanding of efficacy.

The relevance of effect size is exemplified in analyses of the rela-ionship between clinical trials and clinical practice as describedn the IMMPACT group consensus statement on understanding


he clinical importance of outcomes [7]. Of specific relevance ishe finding by Farrar et al. [8] that a 4 point change (on a 10oint NRS) or a 50% decrease in pain ratings for an individualatient equates to patient ratings of “very much improved”; a 2


point change (30% decrease) equates to “much improved”. Theseself-reported ratings illustrate a key component in determiningsuccessful interventions; a change in outcomes must not be onlystatistically significant but of a magnitude that will have a mean-ingful impact on patients’ perceptions of efficacy. These data alsosuggest that 30% is near the cut-off for a pain score reduction thatis perceived as clinically meaningful. An important question thenbecomes whether translation can be improved by making similardeterminations in preclinical studies.

One approach to answering this is to perform back-translationstudies in the animal models. For example, we can examine theefficacy of gabapentin for the treatment of pain associated withpost-herpetic neuralgia (PHN) and compare that to gabapentinefficacy in the spinal nerve ligation (SNL) model of neuropathicpain in rats. In patients that received gabapentin, average dailypain scores decreased approximately 33% from baseline, whereaspain scores in the placebo group decreased by 8% [35]. Further,approximately 43% of patients rated their pain as “moderately ormuch improved”, indicating treatment success in almost half of thepatients. These data can provide a framework for judging the effectsize of a meaningful drug effect in a successful clinical trial, butwhat does this effect look like in an animal model of neuropathicpain? We analyzed data from a study assessing gabapentin effectson SNL-induced mechanical allodynia (Algos Preclinical Services,internal data). In animals treated with 100 mg/kg gabapentin, themaximum possible effect (%MPE; analogous to % pain relief above)was calculated to be 65%. Further, 5 of 7 rats (71%) treated with thisdose showed a %MPE of 50% or more (Table 1). Although the dosingregimens are different in the two studies, the relative exposuresbetween rats and humans at the doses tested in these studies areremarkably similar (44).

This analysis also can be extended to other pain models anddrugs. In a clinical trial assessing duloxetine effects on pain asso-ciated with osteoarthritis of the knee [6], 24 h average pain scoresdecreased by approximately 41% in the duloxetine group and 29%in the placebo group. Further, approximately 28% of subjects in theduloxetine group reported pain relief of >50%, whereas 22% of sub-jects in the placebo group reported pain relief of at least 50%. In astudy assessing duloxetine efficacy on hind limb weight bearing inthe monosodium iodoacetate model of osteoarthritis pain in rats[24], duloxetine (30 mg/kg) produced a %MPE of 58%, while vehi-cle treated animals showed no effects. Responder rates were high,with 6/10 rats showing %MPE >50% in the 30 mg/kg group (Table 1).While exposure levels required for efficacy are much higher in ratsthan in humans (44), note that 30 mg/kg is typically consideredthe minimally effective dose in rats and, as such, is consistent withthe rodent greater effect size concept. And it is also worth notingthat for compounds where rodent and human efficacious exposuresvary, rodents consistently require higher exposures to see efficacy.As a result, preclinical models of pain are less, rather than more,likely to lead to advancement of false-positive drugs into clinicaltrials in contrast to what many commonly assume.

In both cases, the effect size in clinical trials is approximatelyhalf of that seen in animals. This does not mean that a metric of atwofold difference should be firmly applied to determine predictiv-ity. Instead, it simply demonstrates the importance of establishingbiologically meaningful effects; the examples cited provide anobjective, rational approach for helping determine what this mightlook like.

These two studies also demonstrate the significant confoundthat can arise due to placebo effects in clinical trials that aregenerally absent in preclinical studies. Though clearly an over-


simplification based on their complexity, if we subtract placeboeffects from those of gabapentin in the PHN study, the % pain relieffrom gabapentin is 25%. In the duloxetine study, the same sub-traction yields a % pain relief of 12%. In both cases, the pain relief

459

460

461

462


ARTICLE IN PRESSG Model

NSL 29992 1–8

6 G.T. Whiteside et al. / Neuroscience Letters xxx (2013) xxx– xxx

Table 1Comparison of the efficacy of gabapentin and duloxetine in clinical trials and preclinical studies.

Drug (dose) Indication or model tested Species % Reduction 50% Responder rate

Gabapentin (<3600 mg/day) PHN Human 33.3 NRa

Gabapentin (100 mg/kg) SNL Rat 65.4 71.4Duloxetine (60 mg/day) Knee OA Human 41.38 27.9Duloxetine (30 mg/kg) MIA Rat 58.1 60.0Duloxetine (60 mg/kg) MIA Rat 81.3 100.0

rted p

avslppss

ctvrbdf1it

esipstemfesobmssysped

t(c(aaitpteat

463

464

465

466

467

468

469

470

471

472

473

474

475

476

477

478

479

480

481

482

483

484

485

486

487

488

489

490

491

492

493

494

495

496

497

498

499

500

501

502

503

504

505

506

507

508

509

510

511

512

513

514

515

516

517

518

519

520

521

522

523

524

525

526

527

528

529

530

531

532

533

534

535

536

537

538

539

540

541

542

543

544

545

546

547

548

549

550

551

552

553

554

555

556

557

558

559

560

561

562

563

564

565

566

a NR – 50% responder rate not reported in paper; however, 43.2% of subjects repo

ttained was significant, and both products are widely used to treatarious painful conditions. One might consider that in preclinicaltudies, an effect size (as calculated by %MPE or % reversal) of ateast 60% at a non-impairing/confounding dose may be required toredict clinical efficacy; however, a much lower % reversal oftenroduces a statistically significant effect, emphasizing that effectize should be considered more important overall than statisticalignificance.

Another commonly used method for determining efficacy inlinical trials is number needed to treat (NNT). Briefly, NNT reflectshe number of patients that must be treated with a given inter-ention in order to see one successful outcome, typically 50% painelief. The NNT for gabapentin for treating PHN was calculated toe 3.2, indicating that just over 3 patients must be treated with therug in order to see >50% pain relief in 1 patient. In our calculationsrom a dataset of over 200 animals, the NNT for gabapentin (SNL;00 mg/kg) was calculated to be 2 [35], again showing that efficacy

n animal models is consistently greater at comparable exposureshan that observed in successful clinical trials.

Examples such as these suggest that accepting a ‘low bar’ forfficacy in animal models of pain by focusing largely on achievingtatistically significant effects may translate to low or no efficacyn clinical trials, supporting our contention that criteria for com-ound advancement into clinical trials should be appropriatelytringent. The question this presents is how one can set an ‘efficacyhreshold’ preclinically that is analogous to a meaningful clinicalffect. While subject to a number of caveats, this might be approxi-ated by determining the efficacy (e.g., % reversal of hyperalgesia)

or a test compound in an animal model of pain compared tofficacy observed with a clinical SOC drug dosed to achieve an expo-ure equivalent to the human efficacious exposure. The efficacybserved with this dose/exposure now effectively sets the ‘efficacyar’ in models to a level known to be important to patients andoves away from a reliance solely on statistically significant rever-

als that can result in oversimplified conclusions that the treatmenttrategy ‘worked’. Study population homogeneity in animal studiesields decreased variability, increasing the likelihood of detectingtatistically significant effects for smaller effect sizes. In order torotect against Type 1 errors, preclinical studies should focus onffect size as well as statistical significance in order to make soundecisions on which compounds should be advanced.

It is not only efficacy that needs to be considered when assessinghe potential utility of novel compounds. A number of side effectse.g., sedation, cognitive impairment, GI irritation) can confoundonclusions of efficacy, and, as well, experimental parameterspotency, efficacy, exposure, etc.) can affect side effect assay resultsnd interpretations. The saying “the dose makes the poison” ispplicable here; no drug is without side effects or toxicity if givenn large enough quantities. True efficacy does not exist in isola-ion, but is balanced by tolerability. Research and developmentrograms typically determine a ratio of efficacy to side effects, or


herapeutic index (TI), calculated by dividing a pharmacologicallyffective dose/exposure by the minimal ataxic dose/exposure (ornother mechanistically relevant side effect assay). Most overtlyoxic effects can be identified well before development; however,

ain ratings to be “moderately or much improved”.

confounding or intolerable side effects remain a major challengefor drug discovery programs. Sometimes, these side effects, if well-understood, can be overcome. For example, the well-documentedhyperthermia seen after administration of TRPV1 antagonists wasconsidered problematic [9], but improved mechanistic understand-ing led to discovery of compounds with reduced hyperthermia thatretained efficacy [32]. Unfortunately, side effects for many com-pounds have no clear mechanistic underpinning; these ultimatelylimit doses and, hence, efficacy that can be achieved clinically. Aspart of the drug discovery process, industry routinely comparesunder identical experimental conditions the TIs of test compoundsto those of known compounds, whether SOC drugs or tool com-pounds targeting the same mechanism. Because differentiating anew pharmacotherapy from currently prescribed drugs requiresa balance between improved efficacy and reduced or less aver-sive side effects, animal models of pain must be paired withassays assessing potential side effects in order to both confidentlyconclude preclinical efficacy and more accurately predict clinicalsuccess.

6. Conclusions

It is our intent to stimulate discussion and debate and in doingso further progress the ultimate goal of all who work with ani-mal models of pain, namely to better understand human diseaseand develop more effective therapies. It is important to recognizethat animal model data are essential in the development of newanalgesic drugs, whether small or large molecule, and are usedto determine efficacious exposure levels and derivative preclinicalsafety margins. These data also inform dose ranges in early humanstudies and are required prior to initiating human trials. The ques-tion then becomes which animal pain model(s) should be utilizedto achieve these goals. We wholeheartedly support the concept ofadding depth to traditional models by considering supraspinallymediated contributions to pain in addition to spontaneous pain, butwe argue the pain research community would err if such modelsare developed as a means to discount or replace traditional mod-els. We propose assuming a more enlightened view of the value,appropriate uses and limitations of both traditional models as wellas novel models being developed. Furthermore, the drug discoveryprocess needs robust, reproducible and high throughput assays thatcan be used to screen and prioritize high numbers of compoundsand focus on those with the best chance of successful development.More complex and involved/lower throughput assays and models,including large animal models, are better applied later in the pro-cess on fewer compounds and when one may be able to better alignthe clinical methodology and patient population being targeted.We also propose that existing datasets contain richer informationthan is often utilized, and we have suggested options for alternativedata analyses that are more aligned to analysis of clinical datasets.It is our belief that these will improve conclusions of efficacy from


preclinical studies and should also help move us further from thetemptation to reduce complex datasets to simplistic assessmentsof ‘worked’ or ‘didn’t work’, inappropriate terminology for eitherpreclinical or clinical datasets. Finally, public–private-partnership

567

568

569

570


ING Model

N

cience

ctadrTpmbsetb

C

oC

UQ2

A

TAst

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

571

572

573

574

575

576

577

578

579

580

581

582

583

584

585

586

587

588

589

590

591

592

593

594

595

596

597

598

599

600

601

602

603

604

605

606

607

608

609

610

611

612

613

614

615

616

617

618

619

620

621

622

623

624

625

626

627

628

629

630

631

632

633

634

635

636

637

638

639

640

641

642

643

644

645

646

647

648

649

650

651

652

653

654

655

656

657

658

659

660

661

662

663

664

665

666

667

668

669

670

671

672

673

674

675

676

677

678

679

680

681

682

683

684

685

686

687

688

689

690

691

692

693

694

695

696

697

698

699

700

701

702

703

704

705

706

707

708

709

710

711

712

713

714

715

716

717



onsortia such as the European IMI (Innovative Medicines Initia-ive) and the US-sponsored ACTTION/PPRECISE groups, as wells others, in addition to their efforts in calling for uniform stan-ards for experimental reporting, might expand their leadershipoles to foster discussions on enhancing data analysis techniques.his could include helping to open doors into the analysis (andublication) of negative/failed clinical trial data, more extensiveeta-analyses of preclinical data across laboratories, promoting

etter alignment of animal model evoked endpoints with QST mea-ures, and the potential adoption of additional readouts such asffect size in preclinical models. Together, these efforts may serveo span what many perceive at present to be an unbridgeable gapetween preclinical research and clinical trial success.

onflict of interest statements

GTW is an employee of Purdue Pharma L.P. JDP is an employeef Algos Preclinical Services. JDK is an employee of Eli Lilly andompany.

ncited reference

[36].

cknowledgments

The authors thank Dr. Steve Negus, Dept. Pharmacology andoxicology, Virginia Commonwealth University and Dr. Steverneric, Neuroscience Discovery, Eli Lilly and Company for con-tructive discussions regarding this manuscript. The authors alsohank Ms. Eisa Sawyer for assistance with preparation of the figures.

eferences

[1] N. Andrews, E. Legg, D. Lisak, Y. Issop, D. Richardson, S. Harper, T. Pheby, W.Huang, G. Burgess, I. Machin, A.S.C. Rice, Spontaneous burrowing behaviour inthe rat is reduced by peripheral nerve injury or inflammation associated pain,Eur. J. Pain 16 (2012) 485–495.

[2] R. Baron, M. Förster, A. Binder, Subgrouping of patients with neuropathic painaccording to pain-related sensory abnormalities: a first step to a stratifiedtreatment approach, Lancet Neurol. 11 (2012) 999–1005.

[3] G. Blackburn-Munro, Pain-like behaviours in animals – how human are they?Trends Pharmacol. Sci. 25 (2004) 299–305.

[4] L.A. Blackshaw, Visceral pain readouts in experimental medicine, Neurogas-troenterol. Motil. 24 (2012) 891–894.

[5] D.C. Brown, R.C. Boston, J.C. Coyne, J.T. Farrar, Ability of the canine brief paininventory to detect response to treatment in dogs with osteoarthritis, J. Am.Vet. Med. Assoc. 233 (2008) 1278–1283.

[6] A.S. Chappell, D. Desaiah, H. Liu-Seifert, S. Zhang, V. Skljarevski, Y. Belenkov,J.P. Brown, Placebo-controlled study of the efficacy and safety of duloxetine forthe treatment of chronic pain due to osteoarthritis of the knee, Pain Pract. 11(2010) 33–41.

[7] R.H. Dworkin, D.C. Turk, K.W. Wyrwich, D. Beaton, C.S. Cleeland, J.T. Farrar, J.A.Haythornthwaite, M.P. Jensen, R.D. Kerns, D.N. Ader, N. Brandenburg, L.B. Burke,D. Cella, J. Chandler, P. Cowan, R. Dimitrova, R. Dionne, S. Hertz, A.R. Jadad, N.P.Katz, H. Kehlet, L.D. Kramer, D.C. Manning, C. McCormick, M.P. McDermott,H.J. McQuay, S. Patel, L. Porter, S. Quessy, B.A. Rappaport, C. Rauschkolb, D.A.Revicki, M. Rothman, K.E. Schmader, B.R. Stacey, J.W. Stauffer, T. von Stein, R.E.White, J. Witter, S. Zavisic, Interpreting the clinical importance of treatmentoutcomes in chronic pain clinical trials: IMMPACT recommendations, J. Pain 9(2008) 105–121.

[8] J.T. Farrar, J.P. Young, L. LaMoreaux, J.L. Werth, R.M. Poole, Clinical importanceof changes in chronic pain intensity measured on an 11-point numerical painrating scale, Pain 94 (2001) 149–158.

[9] N.R. Gavva, J.J. Treanor, A. Garami, L. Fang, S. Surapaneni, A. Akrami, F. Alvarez, A.Bak, M. Darling, A. Gore, G.R. Jang, J.P. Kessiak, L. Ni, M.H. Norman, G. Palluconi,M.J. Rose, M. Salfi, E. Tan, A.A. Romanovsky, C. Banfield, G. Davar, Pharmaco-logical blockade of the vanilloid receptor TRPV1 elicits marked hyperthermiain humans, Pain 136 (2008) 202–210.


10] M. Granmo, T. Jensen, J. Schouenborg, Nociceptive transmission to rat primarysomatosensory cortex – comparison of sedative and analgesic effects, PLoS ONE8 (2013) e53966.

11] R. Hill, NK1 (substance P) receptor antagonists – why are they not analgesic inhumans? Trends Pharmacol. Sci. 21 (2000) 244–246.

[

[


12] W. Huang, M. Calvo, K. Karu, H.R. Olausen, G. Bathgate, K. Okuse, D.L.H.Bennett, A.S.C. Rice, A clinically relevant rodent model of the HIV antiretro-viral drug stavudine induced painful peripheral neuropathy, Pain 154 (2013)560–575.

13] J.P. Huggins, T.S. Smart, S. Langman, L. Taylor, T. Young, An efficient ran-domised, placebo-controlled clinical trial with the irreversible fatty acid amidehydrolase-1 inhibitor PF-04457845, which modulates endocannabinoids butfails to induce effective analgesia in patients with pain due to osteoarthritis ofthe knee, Pain 153 (2012) 1837–1846.

14] M. Hummel, P. Lu, T.A. Cummons, G.T. Whiteside, The persistence of a long-term negative affective state following the induction of either acute or chronicpain, Pain 140 (2008) 436–445.

15] D.R. Huntjens, D.J. Spalding, M. Danhof, O.E. Della Pasqua, Differences in thesensitivity of behavioural measures of pain to the selectivity of cyclo-oxygenaseinhibitors, Eur. J. Pain 13 (2009) 448–457.

16] J.P.A. Ioannidis, Why most published research findings are false, PLoS Med. 2(2005) e124.

17] S.K. Joshi, P. Honore, Animal models of pain for drug discovery, Expert Opin.Drug Discov. 1 (2006) 341–352.

18] T. King, L. Vera-Portocarrero, T. Gutierrez, T.W. Vanderah, G. Dussor, J. Lai, H.L.Fields, F. Porreca, Unmasking the tonic-aversive state in neuropathic pain, Nat.Neurosci. 12 (2009) 1364–1366.

19] I. Kola, J. Landis, Can the pharmaceutical industry reduce attrition rates? Nat.Rev. Drug Discov. 3 (2004) 711–715.

20] D.J. Langford, A.L. Bailey, M.L. Chanda, S.E. Clarke, T.E. Drummond, S. Echols, S.Glick, J. Ingrao, T. Klassen-Ross, M.L. LaCroix-Fralish, L. Matsumiya, R.E. Sorge,S.G. Sotocinal, J.M. Tabaka, D. Wong, A.M.J.M. van den Maagdenberg, M.D. Fer-rari, K.D. Craig, J.S. Mogil, Coding of facial expressions of pain in the laboratorymouse, Nat. Methods 7 (2010) 447–449.

21] W. Lau, C. Dykstra, S. Thevarkunnel, L.B. Silenieks, I.A.M. de Lannoy, D.K.H.Lee, G.A. Higgins, A back translation of pregabalin and carbamazepine againstevoked and non-evoked endpoints in the rat spared nerve injury model of neu-ropathic pain, Neuropharmacology June (2013), pii: S0028-3908(13)00237-2.

22] L.J. Leys, K.L. Chu, J. Xu, M. Pai, H.S. Yang, H.M. Robb, M.F. Jarvis, R.J. Radek, S.McGaraughty, Disturbances in slow-wave sleep are induced by models of bilat-eral inflammation, neuropathic, and postoperative pain, but not osteoarthriticpain in rats, Pain 154 (2013) 1092–1102.

23] J. Mao, Current challenges in translational pain research, Trends Pharmacol. Sci.33 (2012) 568–573.

24] C.L. Marker, J.D. Pomonis, S.L. Gottshall, Abstract, measuring use-related pain ina rat model of osteoarthritis, in: 13th IASP World Congress on Pain, Montreal,Canada, 2010.

25] J.S. Mogil, Animal models of pain: progress and challenges, Nat. Rev. Neurosci.10 (2009) 283–294.

26] S.S. Negus, T.W. Vanderah, M.R. Brandt, E.J. Bilsky, L. Becerra, D. Borsook, Pre-clinical assessment of candidate analgesic drugs: recent advances and futurechallenges, J. Pharmacol. Exp. Ther. 319 (2006) 507–514.

27] B.G. Oertel, J. Lötsch, Clinical pharmacology of analgesics assessed with humanexperimental pain models: bridging basic and clinical research, Br. J. Pharmacol.168 (2013) 534–553.

28] A.C. Pande, R.E. Pyke, M. Greiner, S.A. Cooper, R. Benjamin, M.W. Pierce, Anal-gesic efficacy of the kappa-receptor agonist, enadoline, in dental surgery pain,Clin. Neuropharmacol. 19 (1996) 92–97.

29] G. Pereira Do Carmo, G.W. Stevenson, W.A. Carlezon, S.S. Negus, Effects of pain-and analgesia-related manipulations on intracranial self-stimulation in rats:further studies on pain-depressed behavior, Pain 144 (2009) 170–177.

30] M.J. Piesla, L. Leventhal, B.W. Strassle, J.E. Harrison, T.A. Cummons, P. Lu,G.T. Whiteside, Abnormal gait, due to inflammation but not nerve injury,reflects enhanced nociception in preclinical pain models, Brain Res. 1295 (2009)89–98.

31] F. Prinz, T. Schlange, K. Asadullah, Believe it or not: how much can we relyon published data on potential drug targets? Nat. Rev. Drug Discov. 10 (2011)328–329.

32] H. Quiding, B. Jonzon, O. Svensson, L. Webster, A. Reimfelt, A. Karin, R.Karlsten, M. Segerdahl, TRPV1 antagonistic analgesic effect: a randomizedstudy of AZD1386 in pain after third molar extraction, Pain 154 (2013)808–812.

33] A.S.C. Rice, D. Cimino-Brown, J.C. Eisenach, V.K. Kontinen, M.L. Lacroix-Fralish,Ian Machin (on behalf of the Preclinical Pain Consortium), J.S. Mogil, T. Stohr,Animal models and the prediction of efficacy in clinical trials of analgesic drugs:a critical appraisal and call for uniform reporting standards, Pain 139 (2008)243–247.

34] A.S.C. Rice, R. Morland, W. Huang, G.L. Currie, E.S. Sena, M.R. Macleod, Trans-parency in the reporting of in vivo pre-clinical pain research: the relevance andimplications of the ARRIVE (Animal Research: Reporting In Vivo Experiments)guidelines, Scand. J. Pain 4 (2013) 58–62.

35] M. Rowbotham, N. Harden, B. Stacey, P. Bernstein, L. Magnus-Miller, Gabapentinfor the treatment of postherpetic neuralgia, J. Am. Med. Assoc. 280 (1998)1837–1842.

36] E.S. Sena, H.B. van der Worp, P.M.W. Bath, D.W. Howells, M.R. Macleod, Publi-cation bias in reports of animal stroke studies leads to major overstatement of


efficacy, PLoS Biol. 8 (2010) e1000344.37] A. Silva, M.L. Andersen, S. Tufik, Sleep pattern in an experimental model of

osteoarthritis, Pain 140 (2008) 446–455.38] G.W. Stevenson, J. Cormier, H. Mercer, C. Adams, C. Dunbar, S.S. Negus, E.J.

Bilsky, Targeting pain-depressed behaviors in preclinical assays of pain and

718

719

720

721

722


ING Model

N

8 cience

[

[

[

[

[

8 (2013) e58125.

723

724

725

726

727

728

729

730

731

732

733

734

735

736

737

738



analgesia: drug effects on acetic acid-depressed locomotor activity in ICR mice,Life Sci. 85 (2009) 309–315.

39] A. Taneja, V.L. Di lorio, M. Danhof, O. Della Pasqua, Translation of drug effectsfrom experimental models of neuropathic pain and analgesia to humans, DrugDiscov. Today 17 (2012) 837–849.

40] E.L. van der Kam, J.D. Vry, K. Schiene, T.M. Tzschentke, Differential effects of


morphine on the affective and the sensory component of carrageenan-inducednociception in the rat, Pain 136 (2008) 373–379.

41] H.B. van der Worp, D.W. Howells, E.S. Sena, M.J. Porritt, S. Rewell, V. O’Collins,M.R. Macleod, Can animal models of disease reliably inform human studies?PLoS Med. 7 (2010) e1000245.

[

PRESS Letters xxx (2013) xxx– xxx

42] C.J. Vierck, P.T. Hansson, R.P. Yezierski, Clinical and pre-clinical pain assess-ment: are we measuring the same thing? Pain 135 (2008) 7–10.

43] M.B. Walton, E. Cowderoy, D. Lascelles, J.F. Innes, Evaluation of constructand criterion validity for the ‘Liverpool Osteoarthritis in Dogs’ (LOAD) clini-cal metrology instrument and comparison to two other instruments, PLoS ONE


44] G.T. Whiteside, A. Adedoyin, L. Leventhal, Predictive validity of animal painmodels? A comparison of the pharmacokinetic–pharmacodynamic relation-ship for pain drugs in rats and humans, Neuropharmacology 54 (2008)767–775.

739

740

741

742


Documents

An industry perspective on the role and utility of animal models of pain in drug discovery