Upload
jeffrey-d
View
215
Download
2
Embed Size (px)
Citation preview
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
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-
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
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).
tive on the role and utility of animal models of pain in drug discovery,
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
ARTICLESL 29992 1–8
G.T. Whiteside et al. / Neuros
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
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
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
tive on the role and utility of animal models of pain in drug discovery,
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-
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
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.
tive on the role and utility of animal models of pain in drug discovery,
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
ARTICLESL 29992 1–8
G.T. Whiteside et al. / Neuros
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
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
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
PRESS Letters xxx (2013) xxx– xxx 5
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-
tive on the role and utility of animal models of pain in drug discovery,
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
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
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
tive on the role and utility of animal models of pain in drug discovery,
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
ARTICLESL 29992 1–8
G.T. Whiteside et al. / Neuros
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.
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
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.
[
[
PRESS Letters xxx (2013) xxx– xxx 7
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
tive on the role and utility of animal models of pain in drug discovery,
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
ARTICLESL 29992 1–8
G.T. Whiteside et al. / Neuros
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
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
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
tive on the role and utility of animal models of pain in drug discovery,
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