American
`Pai~C,
`Society
`
`RESEARCH
`
`EDUCATION
`
`TREATMENT
`
`ADVOCACY
`
`ELSEVIER
`
`The Journal of Pain, Vol 14, No 3 (March), 2013: pp 281-289
`Available online at www.jpain.org and www.sciencedirect.com
`
`The Osteoarthritis Knee Model: Psychophysical Characteristics and
`Putative Outcomes
`R. Norman Harden,*,y
`Gila Wallach,* Christine M. Gagnon,*,y
`Arzhang Zereshki,*
`Meryem Saracoglu,*,y
`Ai Mukai,*,y,x
`Maxine M. Kuroda,*,y
`Joseph R. Graciosa,*
`z
`and Stephen Bruehl
`*Center for Pain Studies, Rehabilitation Institute of Chicago, Chicago, Illinois.
`y
`Department of Physical Medicine and Rehabilitation, Northwestern University Feinberg School of Medicine, Chicago,
`Illinois.
`z
`Department of Anesthesiology, Vanderbilt University School of Medicine, Nashville, Tennessee.
`x
`Texas Orthopedics, Austin, Texas.
`
`Abstract: The knee osteoarthritis (KOA) model is a convenient and coherent archetype that is fre-
`quently used in pharmaceutical trials of drugs with analgesic and/or anti-inflammatory properties;
`yet, little is known about its specific pathophysiology. The presumed chronic inflammatory etiology
`of osteoarthritis suggests that nociceptive processes and neurogenic inflammation predominate in
`this condition. However, most chronic pain conditions are associated with changes in peripheral and
`central processing. Recent data corroborate this as an important mechanism in KOA. We compared psy-
`chophysical characteristics (including thermal Quantitative Sensory Testing); thermal, mechanical, and
`functional wind-up; thermal and mechanical aftersensations; and pressure algometry of 37 subjects
`with KOA with 35 age- and sex-matched controls. A third of the KOA subjects demonstrated hypoesthe-
`sia to vibration and the 4.56 von Frey fiber, yet few showed allodynia in their worse knee. The majority
`of subjects had abnormalities to pinprick (41% were hyperalgesic and 27% were hypoesthetic). Com-
`pared to controls, the more painful knee was hypoesthetic to cold detection and had greater thermal
`wind-up, lower pressure-pain thresholds, thermal and mechanical aftersensations, and twice the
`pain ratings of controls after stair climb. Substantial intraindividual differences were found in KOA sub-
`jects and controls for mechanical wind-up and algometric thresholds.
`Perspective: These results develop the KOA model and suggest mechanistic hypotheses. Certain of
`these tests may ultimately prove to be responsive, quasi-objective, and quantitative outcomes for re-
`search and lend empirical support to the notion of measurable sensitization in osteoarthritis.
`ª 2013 by the American Pain Society
`Key words: Knee osteoarthritis, central sensitization, peripheral sensitization, neuropathy, clinical out-
`comes.
`
`The knee osteoarthritis (KOA) model became popular
`
`during the development of various selective cycloox-
`ygenase-2 (COX-2) inhibitors. It had the presumed
`advantages of being a simple, heterogeneous model of
`chronic inflammation/nociception and served well in the
`traditional study designs and outcomes of the time.36,43
`
`Received May 31, 2012; Revised November 19, 2012; Accepted November
`25, 2012.
`Supported by the Nancy and Lawrence Glick Pain Research Fund and an
`unrestricted grant from GlaxoSmithKline.
`None of the authors have any relationships which may cause a conflict of
`interest with this work.
`Address reprint requests to R. Norman Harden, MD, Rehabilitation Insti-
`tute of Chicago, Center for Pain Studies, 446 E. Ontario, Suite 1011, Chi-
`cago, IL 60611. E-mail: nharden@ric.org
`1526-5900/$36.00
`ª 2013 by the American Pain Society
`http://dx.doi.org/10.1016/j.jpain.2012.11.009
`
`the mechanistic assumptions/
`some of
`However,
`conclusions about KOA may have been overly simplistic.
`There is accumulating evidence that peripheral
`and central neuropathic changes are at least as impor-
`as nociceptive/inflammatory mechanisms,24,55
`tant
`and that both peripheral
`sensitization (PS) and
`central sensitization (CS) likely play an important role
`in KOA.6,8-10,24,26,39,48,65
`Pain
`sensitization
`(both
`peripheral and central) is characterized by an increased
`response
`to
`noxious
`stimuli
`(hyperpathia/
`hyperalgesia,4,31,59 a painful
`response to normally
`innocuous stimuli [allodynia37,57], and wind-up40,69,70
`[all seen in our pilot work26,55,65]).
`PS is conceptually related to inflammation/nocicep-
`tion2,8,9,48 and neurogenic inflammation,52 but this rela-
`tionship is poorly understood or characterized in human
`
`281
`
`

`

`282
`
`The Journal of Pain
`
`models. In a previous empirical study, a large cohort of
`799 knee pain subjects reported pain in 8 local areas
`(69% of subjects), in 4 regional areas (14% of subjects),
`or as diffuse (10% of subjects). In those with localized
`pain, the most commonly reported locations were in
`the medial (56%) and lateral (43%) joint lines. The pa-
`tella (44%) and medial region (38%) were the most com-
`monly reported by those with regional pain.63 It is likely
`that aspects of local and perhaps regional pain are due in
`part to PS, and that diffuse pain and secondary hyperal-
`gesia are primarily due to CS. However, neither of these
`hypotheses has been definitively studied, especially in os-
`teoarthritis (OA) pain.71
`Theoretically, CS is characterized by enhanced excit-
`ability (increased synaptic efficacy70,71) of dorsal horn
`neurons, indicating increased spontaneous activity of
`central pain generators (eg, substantia gelatinosa),
`enlarged receptive fields, increased responses on large
`and small caliber primary afferent fibers,17,40 and
`impaired diffuse noxious inhibitory controls.11 CS of dor-
`sal horn neurons occurs following tissue injury, inflam-
`mation, and pain (nociception22,37) and may occur
`concurrently with or after PS.
`It is likely that CS
`contributes to some components of hyperalgesia and
`allodynia experienced by patients with diseases such as
`OA.5,17,56,71 For
`instance,
`in the hip OA model,
`psychophysical
`testing
`reveals
`allodynic
`and
`hyperalgesic changes in the most painful area (PS and/
`or CS), bilaterally in some tests (CS), all of which
`resolved after successful surgery (total hip replacement
`or osteotomy), suggesting that the sensitization was
`driven or maintained by the OA pain generator.38 Noci-
`ception, inflammation, and PS may induce CS in the spi-
`nal cord via changes in the excitatory to inhibitory
`balance,14,57
`N-methyl-D-aspartate
`neuroamines
`second messengers,13,19,45,66,72 etc, which
`receptor,
`theoretically lead to psychophysical phenomena such as
`secondary hyperalgesia, allodynia, wind-up, and after-
`sensation.59,69,71 It is very likely that suprasegmental
`sensitization (ie, brainstem and higher) is also involved
`in OA pain.6 In arthritis patients, CS is thought to play
`an important role in the spectrum of pain report and be-
`haviors,5,56 and deep tissue pain and inflammation may
`be a more likely cause of CS than superficial cutaneous
`stimulation.5,50 Inflammation of large joints may be
`particularly effective at generating and maintaining CS
`in human patients.50,56 In neuropathic pain, it is known
`that ongoing nociceptive input dynamically maintains
`altered central processing, and that blocking the
`peripheral input causes central processing to revert to
`normal.22,37 Thus, the chronic and continuous nature of
`these spontaneous inflammatory/nociceptive processes
`in OA may be especially robust in generating and
`maintaining a CS state.5,29,56 Though the psychophysical
`outcomes for a variety of pain conditions have been
`reported,22,23,37,40,59,70 few formal assessments of the
`psychophysical correlates of CS in OA have been studied.4
`In this pilot study, the psychophysical correlates of neu-
`ropathy and PS and/or CS were compared between KOA
`subjects and age- and sex-matched controls using the
`KOA model in order to determine if KOA subjects show
`
`KOA Psychophysics and Outcomes
`
`the common correlates of sensitization seen in other
`chronic pain conditions. We will report elsewhere on
`the psychometric and demographic characteristics of
`KOA subjects.
`Aspects of this data were presented in abstract form at
`the annual meeting of the American Pain Society.26,55,65
`
`Methods
`Thirty-seven subjects with the diagnosis of KOA meet-
`ing inclusion criteria were recruited for a single ap-
`pointment at an urban academic pain research center
`by flyers, advertisements in local newspapers, at senior
`centers, and via websites. KOA subjects had experi-
`enced chronic pain of any reported severity $6 months
`in 1 or both knees. Physical examinations were conduct-
`ed by medical doctors on the team. Diagnosis was based
`on history and physical examination to determine if
`subjects fulfilled the criteria published by the Diagnos-
`tic and Therapeutic Criteria Committee of the American
`Rheumatism Association (ARA).1 Knee radiographs
`were not specifically obtained for diagnosis in this pilot;
`however, available radiographs of 8 subjects were eval-
`uated as supplements to the diagnosis. While the ‘‘clin-
`ical examination only’’ ARA criteria have a sensitivity
`comparable to the ARA criteria using ‘‘clinical examina-
`tion and radiography’’ and the ARA criteria using ‘‘clin-
`ical examination and laboratory results’’ (95, 91, and
`92%, respectively1), the specificity was lower in the cri-
`teria without radiographic or laboratory corroboration
`(69, 86, and 75%, respectively). Although greater spec-
`ificity is generally more desirable in research samples,27
`our decision to use the clinical ARA diagnostic scheme
`was based on practical considerations of funding; and
`in this pilot, we must accept the possibility of overdiag-
`nosis as compared to the other criteria. Subjects with se-
`vere or unmanaged medical or psychiatric disturbance
`or other chronic pain conditions or elbow pain were ex-
`cluded. All medication regimens were accepted. A con-
`venience sample of controls with no knee pain and no
`history of knee injury or damage was selected and
`matched by age and sex to the KOA subjects. All sub-
`jects were able to read and speak English and an in-
`formed consent was obtained from each subject. The
`project was approved by the Northwestern University
`Institutional Review Board and complied with the Hel-
`sinki accords.
`Personal health history was reviewed following in-
`formed consent. Subjects had a physical examination
`with the study physician to confirm that inclusion/exclu-
`sion criteria were met and to ensure safety. The psycho-
`physical test battery was completed according to the
`protocols described below.
`
`Psychophysical Testing
`Subjects reported their current level of pain on the
`numeric rating scale (NRS), a well-validated measure of
`pain that ranges from 0 (‘‘no pain’’) to 10 (‘‘worst pain
`imaginable’’).16,34 Trained examiners (R.N.H., G.W., A.Z.,
`and A.M.) conducted all assessments. The sites tested
`were:
`
`

`

`Harden et al
` Medial knee; medial aspect of the more painful knee
`(‘‘worse’’) knee in the distribution of L3 and the sa-
`phenous nerve over the joint line.
` Lateral knee;
`lateral aspect of the more painful
`(‘‘worse’’) knee in the distribution of L5 and the com-
`mon fibular (peroneal) nerve over the joint line.
` Contralateral knee; medial and lateral aspects of the
`less painful knee.
` Contralateral elbow; elbow contralateral to the
`more painful knee in the distribution of C8 and the
`medial antebrachial cutaneous nerve over the me-
`dial joint line.
`For control subjects, the right knee was designated to
`correspond to the more painful (‘‘worse’’) knee of KOA
`subjects.
`
`Bedside Psychophysical Testing
`Bedside tests were conducted at the medial and lateral
`joint lines of the worse knee, contralateral knee, and
`contralateral elbow. Standard physical tests for hypoes-
`thesia and mechanical allodynia were performed using
`a 128-Hz tuning fork (vibration), a 1.5-inch boar bristle
`paint brush (dynamic mechanical stimuli), and a 4.56
`modified von Frey fiber (punctate mechanical stimuli,
`Touch-Test Sensory Evaluator; North Coast Medical, Gil-
`roy, CA). The 4.56 modified von Frey fiber is stiff but
`bends in testing and thus delivers a uniform stimulus of
`4 grams of force. Hypoesthesia and hyperalgesia to a nox-
`ious stimulus were evaluated by pinprick (static mechan-
`ical stimuli) using a 256-mN weighted pin (punctate
`mechanical stimulus).18,54
`
`Mechanical Pressure Stimulation
`
`Algometry
`The Fischer dolorimeter (Wagner Instruments, Green-
`
`wich, CT) with a 1-cm2 rubber disk was applied at a 90
`an-
`gle to the skin surface20 to measure pressure pain
`thresholds (PPTs). These algometric measures for obtain-
`ing PPT have demonstrated acceptable inter- and
`intrarater reliability.20,30,32,58,64 The PPTs were measured
`at the medial and lateral knee and at the contralateral
`elbow. Pressure was increased at a rate of approximately
`1 kg/second, and PPT (kg/cm2) was recorded when
`subjects verbally indicated that they first felt pain.67 PPTs
`were obtained from 2 trials with a recovery time (>5 min-
`utes) between trials.20,67 The values of both trials were
`averaged for each site, a method also previously shown
`to be reliable.46
`
`Thermal Quantitative Sensory Testing
`(tQST)
`Thermal detection and pain thresholds were assessed
`using an established protocol.15,25,73 The standard
`‘‘limits’’ program for
`the Medoc Thermal Sensory
`Analyzer
`(TSA-2001; Ramat Yishai,
`Israel) Peltier
`element-based stimulator25,44 was used. Thresholds
`were measured at the medial and lateral aspects of the
`worse knee and at the contralateral elbow. tQST has
`been shown to be useful
`in identifying small
`
`The Journal of Pain
`
`283
`
`unmyelinated (C) and small myelinated (A delta) fiber
`sensory
`abnormalities
`in subjects with diabetic
`neuropathy,
`small
`fiber
`neuropathies,
`uremic
`neuropathies, and demyelinating neuropathy.41,42,51,62
`
`Wind-Up (WU)
`
`Thermal Wind-Up (TWU)
`Thermal stimuli were delivered by the Medoc device
`following a fixed suprathreshold protocol that has been
`used in other studies.53 Three trains, each consisting of 5
`gradients of increasing heat, were delivered by a 30  30
`
`mm Peltier thermal probe. Each gradient began at 39
`C,
`
`
`rose to a peak temperature of 49
`C, and receded to 39
`C,
`
`with a rise and decline rate of 10
`C/second (2.4-second du-
`ration heat pulse). Participants were asked to rate their
`pain using the NRS at gradients 1, 3, and 5 in each of the
`3 trains. TWU was tested at the lateral and medial aspects
`of the worse knee and at the contralateral elbow.
`
`Mechanical Wind-Up (MWU)
`Using a modified von Frey procedure, a 5.46 von Frey
`fiber was used to assess MWU.35,37 An initial NRS score
`was obtained after a single stimulus. Subsequently, 10
`stimulations were administered at a rate of 1/second
`within the same 1-cm2 area. Subjects were asked to re-
`port an NRS score immediately after each stimulus. A ter-
`minal NRS score was solicited, as per Rolke et al.54 MWU
`was tested at the same test sites used for TWU.
`
`Functional Wind-Up (FWU)
`A stair climb task was used as a measure of pain upon
`FWU. Subjects were asked to descend and then ascend
`a flight of 9 steps ‘‘as fast as you comfortably can.’’ Pain
`(NRS) was solicited immediately before and after the
`task.
`
`Aftersensation
`Aftersensation is described as ‘‘evoked pain outlasting
`the time of stimulation’’ in our study, pain that lingers
`after termination of pain-evoking external stimula-
`tion.21,71 For
`this
`study, data used to evaluate
`aftersensation were collected at the end of the TWU
`and the MWU protocols.
`Immediately following the
`final gradient of heat applied in the TWU protocol and
`after the last 5.46 von Frey fiber stimulation in the
`MWU protocol, subjects reported their NRS score every
`10 seconds until their ratings returned to NRS score of
`0 or until 3 minutes had elapsed, whichever came first.
`Aftersensation was recorded as present (NRS > 0) or as
`absent (NRS = 0).
`
`Data Analysis
`Descriptive summaries are presented as mean 6 stan-
`dard deviation for continuous variables; n (%) for ordinal
`and nominal (categorical) variables. Group differences in
`demographic characteristics were tested by independent
`samples t-tests or by chi-square (Fisher’s exact test
`depending on the distributional characteristics of the
`variables). A series of mixed-model analyses of variance
`
`

`

`284
`
`The Journal of Pain
`
`(ANOVAs) tested tQST, TWU, MWU, and PPT (algometric
`score) with Subject Type (KOA or control) as the
`between-subjects factor and Site (medial knee, lateral
`knee, contralateral knee, contralateral elbow) as the
`within-subjects factor. The interaction of Subject Type
`and Site was included in the mixed-model ANOVAs. In
`some instances, fewer sites were studied because of
`time considerations (testing ran over 3.5 hours in some
`cases). When Mauchly’s test indicated a violation of the
`sphericity assumption (equal variances in repeated mea-
`sures ANOVA), the Greenhouse-Geisser adjustment was
`used. Post hoc analyses (paired-sample t-tests) of identi-
`fied main effects were examined at a Bonferroni cor-
`rected a = .017. All data were analyzed using SPSS
`v.16.0 software (SPSS Inc, Chicago, IL).
`
`Results
`As expected, KOA subjects and controls were similar
`in age and gender. Though the psychometric features
`of the KOA subjects will be described in a separate pa-
`per, race/ethnicity is included here as further verifica-
`tion of the similarity of the KOA and control samples
`(Table 1). Sixty percent of the subjects had their more
`painful knee on their right side, and 40% on their
`left. A total of 30 (81%) subjects reported pain in
`both knees bilaterally. Controls were assessed for knee
`OA using the same criteria as the subjects and did not
`meet criteria.1
`
`Bedside Psychophysical Testing
`Only 4 (11%) KOA subjects demonstrated hypoesthe-
`sia to light touch (brush) in their worse knee; however,
`11 (31%) KOA subjects demonstrated hypoesthesia to vi-
`bration and to the 4.56 von Frey fiber in the worse knee.
`No more than 5 (14%) KOA subjects showed allodynia in
`the worse knee by light touch (brush), vibration, or the
`4.56 von Frey fiber (Table 2). All controls reported a nor-
`mal response to the brush test. On the vibration test, 1
`control was hypoesthetic in the left knee, and 2 were hy-
`poesthetic in the right knees. These findings are consis-
`tent with a large (A-beta) and medium (A-delta) fiber
`type neuropathy and/or specific receptor pathology.
`The von Frey test elicited hypoesthetic response in 2 sub-
`jects. Only 1 control was hyperalgesic to pinprick in the
`contralateral (left) elbow. Further statistical testing of
`
`Table 1. Demographic Characteristics by Subject
`Type*
`
`VARIABLE
`
`KOA (N = 37)
`
`CONTROL (N = 35)
`
`Age
`Sex (% female)
`Race/ethnicity
`White non-Hispanic
`African American
`Hispanic
`Other
`
`64.8 (11.6)
`28 (76)
`
`24 (65)
`11 (30)
`1 (3)
`1 (3)
`
`*Data are n (%) except for age (mean 6 SD).
`
`62.1 (9.5)
`22 (63)
`
`29 (83)
`4 (11)
`1 (3)
`1 (3)
`
`KOA Psychophysics and Outcomes
`Table 2. Bedside Psychophysical Testing
`Responses to Sensory Stimuli (Light Touch,
`Vibration, Mechanical) and Pain (Pinprick) in
`KOA Subjects (n [%]) by Site*
`
`SENSORY DETECTION
`
`N
`
`NORMAL HYPOESTHETIC
`
`ALLODYNIC
`
`4 (11)
`2 (6)
`1 (3)
`
`3 (8)
`2 (6)
`0
`
`3 (8)
`2 (6)
`0
`
`Dynamic mechanical stimuli (brush)
`Worse knee
`36
`29 (81)
`Contralateral knee
`35
`31 (87)
`Contralateral elbow 34
`33 (97)
`Vibration (128-Hz tuning fork)
`11 (31)
`22 (61)
`Worse knee
`36
`7 (20)
`26 (74)
`Contralateral knee
`35
`1 (3)
`33 (97)
`Contralateral elbow 34
`Static mechanical stimuli (4.56 von Frey fiber)
`5 (14)
`Worse knee
`36
`20 (56)
`11 (31)
`4 (11)
`Contralateral knee
`35
`27 (77)
`4 (11)
`2 (6)
`Contralateral elbow 34
`29 (85)
`3 (9)
`Pain
`n Normal Hypoesthetic Hyperalgesic
`Static mechanical stimuli (pinprick)
`Worse knee
`37
`12 (33)
`Contralateral knee
`35
`26 (74)
`Contralateral elbow 34
`31 (91)
`
`10 (27)
`5 (14)
`2 (6)
`
`15 (41)
`4 (11)
`1 (3)
`
`*Medial aspect of worse knee, medial aspect of contralateral knee, and medial
`aspect of contralateral elbow.
`
`the differences between KOA subjects and controls was
`not feasible due to the small number of responders to
`bedside psychophysical testing.
`
`Algometry
`Compared to controls, KOA subjects had lower PPTs
`at all 3 sites, suggesting a diffuse process (CS). PPTs
`between KOA subjects and controls differed most nota-
`bly for the medial and lateral knee than for the elbow
`(Fig 1).
`
`tQST
`KOA subjects detected cold at a lower tempera-
`ture (ie, higher threshold; hypoesthesia) than controls
`(F[1, 70] = 7.395, P = .008, partial h2 = .10) in the medial
`and lateral knee. However, KOA subjects and controls
`did not differ on any of the other tQST measures
`
`•..
`
`..............
`
`···· ...
`
`7
`
`6
`
`5
`
`N'
`~ 4
`i
`f3
`"-
`
`2
`
`1
`
`0
`
`• • ,. •• Control
`
`-KOA
`
`M edial knee
`
`Lateral knee
`
`Contra lateral
`elbow
`
`Figure 1. Algometry in worse knee and contralateral elbow vs
`control. Mean PPTs differed by location (F[2, 138] = 39.2, P = .000,
`partial h2 = .36). KOA subjects had lower thresholds to pain de-
`tection at all 3 sites compared to control subjects (F[1, 69] = 27.6,
`P = .000, partial h2 = .29).
`
`

`

`Harden et al
`Table 3. tQST by Subject Type and Site
`
`The Journal of Pain
`
`285
`
`MEDIAL WORSE KNEE
`
`LATERAL WORSE KNEE
`
`CONTRALATERAL ELBOW
`
`KOA
`
`CONTROL
`
`KOA
`
`CONTROL
`
`KOA
`
`CONTROL
`
`Detection
`Cold
`Warm
`Pain
`Cold
`Hot
`
`27.4 (2.8)*
`37.2 (3.0)
`
`10.5 (10.3)
`45.5 (3.5)
`
`28.9 (1.8)
`36.4 (2.7)
`
`9.2 (9.0)
`45.8 (2.9)
`
`24.1 (5.8)*
`41.8 (4.6)
`
`4.2 (7.8)
`47.9 (2.2)
`
`26.5 (2.9)
`40.0 (3.9)
`
`4.8 (8.3)
`47.9 (1.9)
`
`28.4 (2.1)
`35.0 (1.5)
`
`13.1 (9.5)
`44.3 (3.9)
`
`29.1 (2.3)
`34.8 (1.6)
`
`9.2 (9.6)
`45.2 (3.4)
`
`NOTE. Data are mean
`*P < .01.
`
`
`
`C (6 SD) from 37 KOA subjects and 35 controls.
`
`(warm detection, cold pain, hot pain) (Table 3). Sensitiv-
`ity differed by site for both KOA subjects and controls.
`The lateral knee showed the least sensitivity (highest
`detection and pain thresholds) of all 3 sites, followed
`by the medial knee, with the contralateral elbow exhib-
`iting the most sensitivity. No significant differences by
`subject type or interaction of subject type  site were
`observed.
`
`WU
`
`TWU
`Mean NRS score was calculated for gradients 1 and 5
`over the 3 trains, and difference scores obtained. KOA
`subjects had higher NRS difference scores than controls,
`indicating greater TWU regardless of site, consistent
`with a diffuse process (CS). No significant differences by
`site or interaction of subject type  site were observed
`(Fig 2).
`
`MWU
`Mean NRS score was calculated for taps 1, 5, and 10,
`and difference scores obtained between taps 10 and 1.
`The medial aspect of the knee demonstrated higher
`MWU than the lateral aspect of the knee regardless of
`subject type, consistent with a local process (PS). No sig-
`nificant differences by subject type or interaction of sub-
`ject type  site were observed (Fig 3).
`
`10
`
`TWU
`
`9
`.;;- 8
`a: 7
`z
`'F 6
`8
`"' 5
`C
`'ii
`C .. 3
`4
`0..
`
`QI
`::i:
`
`2
`
`1
`0
`
`-
`
`•·······························
`.. .... -!-::+..-.-. .. .... :.::::::i
`
`-MedialKOA
`
`··~·· MedialC
`
`-
`
`LateralKOA
`
`...... LateralC
`
`-ElbowKOA
`
`•• ,. •• Elbowc
`
`FWU
`Compared to controls, KOA subjects reported in-
`creased pain following stair climb (P < .001). Moreover,
`mean NRS score reported by KOA subjects after stair
`climb (NRS 3.8 6 2.9) was twice that reported before stair
`climb (NRS 1.8 6 2.4). This may be most consistent with
`a local process (PS).
`
`Aftersensation
`Interestingly, thermal aftersensation was only seen in
`the lateral knee at 30 seconds (P = .008), and mechani-
`cal aftersensation was only found in the medial knee at
`30 seconds (P = .034), 60 seconds (P = .029), and 180 sec-
`onds (P = .029). Other trends can be observed in the
`data (Table 4), but no aftersensation was seen in the
`contralateral elbow, suggesting a local process (PS).
`
`Discussion
`This pilot study sought data that would be useful in de-
`veloping a better understanding of features and charac-
`teristics of the KOA model. This model received a great
`deal of attention during the development of various se-
`lective COX-2 inhibitors. It had the presumed advantages
`of being a straightforward, homogeneous model of
`chronic inflammation at a simple hinge joint and the sur-
`rounding structures, and indeed it performed well with
`traditional study designs and outcomes.36,43 However,
`
`MWU
`
`.;;-
`a: z
`'F 0
`u
`"' C
`'ii
`.. QI
`0..
`C
`
`::i:
`
`10
`
`9
`
`8
`
`7
`
`6
`
`5
`
`4
`
`3
`
`2
`
`1
`
`0
`
`-MedialKOA
`
`··~·· MedialC
`
`-
`
`LateralKOA
`
`···•·· LateralC
`
`-ElbowKOA
`
`•• ,. •• ElbowC
`
`Gradient 1
`
`Gradient 3
`
`Gradient 5
`
`Tap 1
`
`Tap2
`
`Tap3
`
`Figure 2. TWU in worse knee and contralateral elbow versus
`control. KOA subjects demonstrated greater TWU than controls
`(F[1, 26] = 11.9, P = .002, partial h2 = .32) regardless of site. No
`main effect for location and interaction of subject type  loca-
`tion.
`
`Figure 3. MWU in worse knee and contralateral elbow versus
`control. MWU differed by location (F[2, 87] = 6.3, P = .005, partial
`h2 = .11); the medial aspect of the knee demonstrated higher
`MWU than the lateral portion of the knee (P < .002). No differ-
`ences by subject type or interaction of subject type  site.
`
`

`

`286
`KOA Psychophysics and Outcomes
`The Journal of Pain
`Table 4. Aftersensation by Site, Subject Type, Wind-Up Parameter, and Time
`
`10 SECONDS
`
`30 SECONDS
`
`60 SECONDS
`
`180 SECONDS
`
`SITE
`
`Medial knee
`
`Lateral knee
`
`Contralateral elbow
`
`SUBJECT
`
`KOA
`Control
`P value*
`KOA
`Control
`P value*
`KOA
`Control
`P value*
`
`THERMAL
`AFTER-
`SENSATION
`
`MECHANICAL
`AFTER-
`SENSATION
`
`THERMAL
`AFTER-
`SENSATION
`
`MECHANICAL
`AFTER-
`SENSATION
`
`THERMAL
`AFTER-
`SENSATION
`
`MECHANICAL
`AFTER-
`SENSATION
`
`THERMAL
`AFTER-
`SENSATION
`
`MECHANICAL
`AFTER-
`SENSATION
`
`10 (27)
`4 (11)
`.167
`12 (32)
`4 (11)
`.060
`10 (27)
`5 (14)
`.298
`
`6 (16)
`1 (3)
`.124
`7 (19)
`1 (3)
`.066
`5 (14)
`2 (6)
`.476
`
`10 (27)
`3 (8)
`.080
`12 (32)
`2 (6)
`.008
`7 (19)
`3 (8)
`.354
`
`8 (22)
`1 (3)
`.034
`4 (11)
`1 (3)
`.392
`4 (11)
`0
`.128
`
`8 (22)
`2 (6)
`.103
`6 (16)
`2 (6)
`.297
`5 (14)
`1 (3)
`.225
`
`6 (16)
`0
`.029
`4 (11)
`0
`.128
`1 (3)
`0
`1.00
`
`8 (22)
`2 (6)
`.103
`6 (16)
`1 (3)
`.124
`3 (8)
`1 (3)
`.657
`
`6 (16)
`0
`.029
`3 (8)
`0
`.260
`1 (3)
`0
`1.00
`
`NOTE. Data are n (%) of 37 KOA subjects and 35 controls who reported NRS > 0 at 10, 30, 60, and 180 seconds following termination of the TWU and MWU protocols.
`*Two-sided Fisher’s exact test P values.
`
`it has become apparent that the simplistic view of this
`model does not adequately encompass the mechanistic
`complexity of KOA; in particular, there is accumulating
`evidence that peripheral and central neuropathic
`changes are at least as important as its nociceptive/
`inflammatory mechanisms.6,24,26,55,65,71 Our
`data
`corroborate a pathophysiology involving considerably
`more mechanistic
`heterogeneity
`in KOA than
`previously thought.
`
`Neuropathy
`Compared to controls, KOA subjects showed hypoes-
`thesia to cold detection in the medial and lateral knee
`with the lateral knee showing a more pronounced hypo-
`esthesia than the medial (most likely a regional dysfunc-
`tion). This may indicate a transducer dysfunction and/or
`fiber type specific neuropathy, as warm detection and
`heat and cold pain thresholds did not differ between
`our KOA subjects and controls. Arendt-Nielsen et al
`also found medial versus lateral differences in psycho-
`physical responses in their KOA subjects.4,68 Moreover,
`our KOA subjects showed hypoesthesia to punctate
`mechanical stimuli and vibration in the more painful
`knee, all consistent with peripheral neuropathy as
`a mechanism. Another possible (and not mutually
`exclusive) mechanism may be cytokine/inflammatory
`damage at transducer or nerve.
`All indicators of pain sensitization, including hyperalge-
`sia,59 allodynia,37,57 wind-up,40,69,70 and aftersensation,71
`were observed in our pilot KOA work26,55,65 and in this
`project. Both PS and CS have been implicated as
`mechanisms underlying pain in OA.3-5,31 However,
`technology for distinguishing PS from CS has not been
`validated.
`PS is likely related to local inflammation2,8-10,39,48 and
`neurogenic inflammation,52 but these hypothetical/con-
`ceptual correlations are very poorly characterized in hu-
`man models. Certain aspects of our data support PS and/
`or neurogenic inflammation in KOA. Very few of the
`
`KOA subjects showed allodynia in the worse knee
`(only) either by light touch, vibration, or the 4.56 von
`Frey fiber (Table 2). In our MWU protocol, more KOA sub-
`jects than controls reported NRS > 0 after 10 seconds spe-
`cifically in the medial knee.
`In contrast,
`in our
`aftersensation protocol, more KOA subjects than con-
`trols appeared to have early aftersensations in the lateral
`knee that ceased by 60 seconds (ie, reported NRS > 0) lo-
`cally. These findings were not seen in the contralateral el-
`bow (thus not diffuse or conceptually consistent with
`widespread CS [Table 4]). These findings and the FWU
`pain following stair climb are perhaps most suggestive
`of PS. Traditionally, the findings mentioned above are
`also said to be characteristic of CS, but if they only occur
`locally they may be more indicative of a peripheral pro-
`cess, which may be cytokine/inflammatory mediated,
`neurogenic inflammation, and/or peripheral neuropa-
`thy.9
`CS is said to be characterized by hyperalgesia,59 allody-
`nia,37,57 wind-up,40,69,70 and diffuse psychophysical
`changes occurring remote from the affected joint in
`the KOA model.3-5,71 Regional
`sensitization is
`less
`clearly PS or CS (as best illustrated by such extensively
`studied
`syndromes
`as
`complex
`regional
`pain
`syndrome28). Sensitization of dorsal horn and supraseg-
`mental neurons following joint lesions and inflamma-
`nociception22,37)
`tion
`(also
`known
`as
`perhaps
`contribute to the regional hyperalgesia experienced by
`such as OA,5,6,17,22,56
`patients with diseases
`thus
`possibly playing an important role in the spectrum of
`pain report and behaviors seen in KOA.5,56 Our findings
`of lower threshold to algometric testing and higher
`TWU in KOA subjects rather than controls, regardless of
`site (including contralateral elbow), suggests a central,
`and probably cerebral, process.
`In future designs,
`diffuse signs of general (central) sensitization—such as
`assessing algometric threshold at fibromyalgia tender
`points67 or tQST at nonjoint sites such as medial fore-
`arm4—should be methodically studied, as these hypo-
`thetically distinguish CS. It may prove useful to develop
`
`

`

`Harden et al
`
`technologies to distinguish peripheral from regional
`from central processes.
`
`Psychophysical Testing as Outcomes
`Psychometric testing in OA disorders (particularly
`pain rating scales and the Western Ontario and McMas-
`ter Universities Arthritis Index (WOMAC7) are currently
`the most common (and responsive) outcomes in OA re-
`search paradigms.12,60,61 Subjective, yet quantified, pain
`ratings are commonly used to follow a patient’s
`response to analgesic treatment33,34; although having
`good clinical value for within-subject comparison,
`between-subject contrasts are less reliable.49 Unfortu-
`nately, the OA pain literature is based entirely on self-
`report type scales, and the fully subjective nature of
`these measures makes them less satisfactory for re-
`search or statistical analysis.27,49 Thus, there is a critical
`need to shift the paradigm away from subjective
`measures
`(where possible)
`toward more objective
`methodologies,
`such as mechanical and thermal
`quantitative sensory testing for detecting possible
`neuropathic pain components. The KOA model lends
`itself to that. Specifically, cold detection at the joint
`(perhaps PS or neuropathy),
`joint
`line algometry
`(perhaps most characteristic of CS), MWU (perhaps PS,
`
`at the medial knee) and the fixed 49
`TWU paradigm
`(perhaps PS, regional), early thermal aftersensation in
`the lateral knee (perhaps PS), and late mechanical
`aftersensation in the medial knee (perhaps PS) may
`prove to be responsive quasi-objective outcomes for re-
`search in this model.4,26,55,65 Reported pain difference
`
`References
`
`1. Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K,
`Christy W, Cooke TD, Greenwald R, Hochberg M, Howell D,
`Kaplan D, Koopman W, Longley S, Mankin H, McShand DJ,
`Medsger T, Meenan R, Mikkelsen W, Moskowitz R,
`Murphy W, Rothschild B, Segal M, Sokoloff L, Wolfe F: Devel-
`opment of criteria for the classification and reporting of
`osteoarthritis. Classification of osteoarthritis of the knee.
`Diagnostic and Therapeutic Criteria Committee of the Amer-
`ican Rheumatism Association. Arthritis Rheum 29:
`1039-1049, 1986
`
`2. Andrew D, Greenspan JD: Mechanical and heat sensitiza-
`tion of cutaneous nociceptors after peripheral inflamma-
`tion in the rat. J Neurophysiol 82:2649-2656, 1999
`
`3. Arendt-Nielsen L, Graven-Nielsen T: Central sensitization
`in fibromyalgia and other musculoskeletal disorders. Curr
`Pain Headache Rep 7:355-361, 2003
`
`4. Arendt-Nielsen L, Nie H, Laursen MB, Laursen BS,
`Madeleine P, Simonsen OH, Graven-Nielsen T: Sensitization
`in patients with painful knee osteoarthritis. Pain 149:
`573-581, 2010
`
`5. Bajaj P, Graven-Nielsen T, Arendt-Nielsen L: Osteoarthritis
`and its association with muscle hyperalgesia: An experimen-
`tal controlled study.