Psychophysical and functional imaging evidence supporting the presence of central sensitization in a cohort of osteoarthritis patients.код для вставкиСкачать
Arthritis & Rheumatism (Arthritis Care & Research) Vol. 61, No. 9, September 15, 2009, pp 1226 –1234 DOI 10.1002/art.24837 © 2009, American College of Rheumatology ORIGINAL ARTICLE Psychophysical and Functional Imaging Evidence Supporting the Presence of Central Sensitization in a Cohort of Osteoarthritis Patients STEPHEN E. GWILYM,1 JOHN R. KELTNER,1 CATHERINE E. WARNABY,1 ANDREW J. CARR,1 BORIS CHIZH,2 IAIN CHESSELL,2 AND IRENE TRACEY1 Objective. The groin pain experienced by patients with hip osteoarthritis (OA) is often accompanied by areas of referred pain and changes in skin sensitivity. We aimed to identify the supraspinal inﬂuences that underlie these clinical manifestations that we consider indicative of possible central sensitization. Methods. Twenty patients with hip OA awaiting joint replacement and displaying signs of referred pain were recruited into the study, together with age-matched controls. All subjects completed pain psychology questionnaires and underwent quantitative sensory testing (QST) in their area of referred pain. Twelve of 20 patients and their age- and sex-matched controls underwent functional magnetic resonance imaging (MRI) while the areas of referred pain were stimulated using cold stimuli (12°C) and punctate stimuli (256 mN). The remaining 8 of 20 patients underwent punctate stimulation only. Results. Patients were found to have signiﬁcantly lower threshold perception to punctate stimuli and were hyperalgesic to the noxious punctate stimulus in their areas of referred pain. Functional brain imaging illustrated signiﬁcantly greater activation in the brainstem of OA patients in response to punctate stimulation of their referred pain areas compared with healthy controls, and the magnitude of this activation positively correlated with the extent of neuropathic-like elements to the patient’s pain, as indicated by the PainDETECT score. Discussion. Using psychophysical (QST) and brain imaging methods (functional MRI), we have identiﬁed increased activity with the periaqueductal grey matter associated with stimulation of the skin in referred pain areas of patients with hip OA. This offers a central target for analgesia aimed at improving the treatment of this largely peripheral disease. INTRODUCTION Patients with osteoarthritis (OA) are historically considered to have a peripheral disease, with nociceptive damage at the joint level accounting solely for the pain they experience. Using this model, however, it has been appreciated for many years that anomalies exist in terms of the degree of pathology found at the joint and pain experienced by the patients (1). Two other clinical features seen in OA are also difﬁcult Supported by an unrestricted grant from GlaxoSmithKline. The Centre for Functional MRI of the Brain was supported by the MRC of the UK and Northern Ireland. Dr. Gwilym’s work was supported by the Arthritis Research Campaign and the Botnar Research Centre (NIHR Biomedical Research Unit). 1 Stephen E. Gwilym, MD, MRCS, John R. Keltner, MD, PhD, Catherine E. Warnaby, PhD, Andrew J. Carr, FRCS, Irene Tracey, PhD: University of Oxford, Oxford, UK; 2Boris Chizh, MD, PhD, Iain Chessell, PhD: GlaxoSmithKline, Cambridge, UK. Address correspondence to Stephen E. Gwilym, MD, MRCS, Centre for Functional MRI of the Brain (FMRIB), John Radcliffe Hospital, Headley Way, Oxford, OX3 9DU, UK. E-mail: firstname.lastname@example.org. Submitted for publication December 16, 2008; accepted in revised form June 17, 2009. 1226 to explain using an exclusively peripheral model of pain: the high incidence of referred pain (2– 4) and changes in skin sensitivity at sites away from the joint (5– 8). We use these latter observations as a mechanism to explore possible altered central processing of nociceptive stimuli in patients with OA of the hip akin to alterations found in patients with neuropathic pain. Classically, referred pain (somatic) is thought to occur because of segmental embryology, although since the term was initially used by Head (9), details as to the exact mechanism by which referred pain occurs have remained controversial. Several theories exist, but all include how higher centers misinterpret the peripheral source of nociception (10). The presence of referred pain in patients with painful OA is suggestive of central nervous system changes in nociceptive processing. Secondary hyperalgesia is a phenomenon of increased pain report and sensitivity to nociceptive stimuli at sites distant to the primary injury, and is thought to involve central sensitization (11). The literature relating to models of musculoskeletal pain suggest that OA is associated with enhanced nociceptive transmission at the dorsal horn (12,13), a hallmark of secondary hyperalgesia. This enhanced excitability of dorsal horn neurons to nociceptive inputs is termed central sensitization, and is manifested by Functional MRI Evidence of Central Sensitization in Hip OA 1) increased response to input from an injured or inﬂamed region, 2) increased response from regions adjacent to or remote from the injured/inﬂamed region, and 3) expansion of the receptive ﬁeld of the spinal cord neuron (14). As such, secondary hyperalgesia that is seen in patients with OA offers a peripheral manifestation of central sensitization that can be experimentally conﬁrmed. It is increasingly appreciated that the descending pain modulatory system with its inhibitory and facilitatory components (15) directly modulate nociceptive processing in the dorsal horn. Descending facilitation plays a pivotal role in the maintenance of central sensitization and accounts for the chronicity of pain states where central sensitization is a component (15–18). Animal studies have clearly proven that brainstem inﬂuences on dorsal horn excitability play a role in experimental pain models involving altered peripheral sensitivity (17,19). Furthermore, previous and current work from our laboratory has translated these ﬁndings from animal studies to humans using functional neuroimaging, identifying that brainstem areas within the descending modulatory network are involved in the ampliﬁcation of noxious mechanical stimuli in the area of secondary hyperalgesia (20,21). This work has also highlighted that brainstem regions are modulated by pharmacologic agents effective at treating neuropathic pain patients, providing support for their use as markers of central sensitization in humans. Therefore, based on our previous ﬁndings that use brainstem activity as a marker for central sensitization and other human literature supporting abnormalities in brainstem processing in patients with chronic pain (22–25), we hypothesize that in patients with OA, referred pain and altered skin sensitivity are likely to involve increased activity within the brainstem modulatory network in response to noxious mechanical provocation within their areas of referred pain. We focus this ﬁrst application to OA on the periaqueductal grey (PAG) matter because it has been most commonly reported as abnormal across all of these studies. We believe that by conﬁrming the presence of hyperalgesia and increased brainstem involvement in the pain manifestations experienced by patients with OA, this will have signiﬁcant implications in future rational analgesic therapy. Speciﬁcally, it will provide a rationale for exploring pharmacologic modulation of central sensitization, either directly (e.g., with agents reducing spinal hyperexcitability) or indirectly (e.g., by attenuating descending facilitation or facilitating descending inhibition) in OA patients with clinical and/or sensory signs of central sensitization. SUBJECTS AND METHODS Subjects. The Central Oxfordshire Research Ethics Committee provided ethical approval for this study. Twenty right-handed subjects were recruited from the orthopaedic outpatient clinic of the Nufﬁeld Orthopaedic Centre, Oxford. All of the subjects had right-sided hip pain secondary to primary OA of the hip and had been placed on the waiting list for hip arthroplasty. Subjects were 1227 excluded if they had previously undergone any form of orthopaedic surgery. During the screening interview, subjects were asked about the presence or absence of referred pain that was felt along the right thigh. All of the patients seen reported this phenomenon. Subject selection was conﬁrmed after screening for other chronic pain conditions, diabetes, and neurologic or psychiatric disorders. Controls. Twelve healthy right-handed control subjects were recruited through poster advertisements within our institution. Subjects were selected after screening for previous history of arthritis, chronic pain conditions, diabetes, and neurologic or psychiatric disorders. In both the patient and control groups, written informed consent was obtained. All of the subjects underwent comprehensive verbal screening to ensure that they did not meet any of the exclusion criteria for magnetic resonance experimentation. Methods. Patients and controls were invited to attend our unit, where a process of behavioral, psychophysical, and functional imaging experimentation took place, as outlined in Figure 1. Behavioral testing began with the completion of a number of self-administration questionnaires: the Beck Depression Inventory (BDI) (26), the Pain Catastrophizing Scale (PCS) (27), the Tampa Scale for Kinesiophobia (TSK) (28), the State Anxiety Index (STAI) and Trait Anxiety Index (TRAI) (29), and PainDETECT (30). PainDETECT has been developed and validated for the purpose of identifying neuropathic elements of a patient’s pain (30,31). The result is a composite score ranging from 0 to 39, where higher scores are more suggestive of neuropathic pain and lower scores are indicative of the pain being nociceptive. Following the completion of the questionnaires, all of the subjects and controls underwent a detailed examination prior to imaging in line with quantitative sensory testing (QST) protocols developed by the German Research Network on Neuropathic Pain (32). The elements of QST to be applied in the study were chosen after pilot work (data not shown) demonstrated the high yield of sensory abnormalities in patients with hip arthritis for the following domains: punctate stimulus detection threshold, punctate hyperalgesia, changes in thermal perception thresholds, and thermal pain threshold levels. For the chosen elements (punctate and thermal stimuli), a standardized protocol was developed that exactly replicated the methodologic techniques described by the German Research Network on Neuropathic Pain (32). Magnetic resonance imaging (MRI) data acquisition. Imaging was conducted using a 3T Varian-Siemens wholebody MR scanner (Varian, Palo Alto, CA). A head-only gradient coil was used with a birdcage radiofrequency coil for pulse transmission and signal reception. A whole-brain gradient echo-planar imaging sequence was used to acquire blood oxygenation level– dependent (BOLD) functional images with an echo time (TE) of 30 msec (43 contiguous 3-mm–thick slices, 64 ⫻ 64 matrix, ﬁeld of view 224 ⫻ 224, voxel size 3.5 ⫻ 3.5 ⫻ 3.5 mm3). The repetition 1228 Gwilym et al Figure 1. Illustration of the experimental paradigm performed by the subjects. All parts of the paradigm were performed on the same day. BDI ⫽ Beck Depression Inventory; PCS ⫽ Pain Catastrophizing Scale; TSK ⫽ Tampa Scale for Kinesiophobia; STAI ⫽ State Anxiety Index; TRAI ⫽ Trait Anxiety Index; vF ⫽ von Frey. time (TR) was 3 seconds, with 260 and 140 volumes collected for the cold and punctate stimuli experiments, respectively. The ﬁrst 5 volumes were discarded to permit equilibration of the BOLD signal. After acquisition of the functional scans, a T1 structural scan was acquired using a 3-dimensional fast low-angle shoot sequence with inversion recovery (TR 204 msec, TE 5.56 msec, inversion recovery time 1,100 msec, ﬂip angle 8°, isotropic volume acquisition, rectangular ﬁeld of view, voxel size 1 mm3) for coregistration purposes. Stimuli. Cold stimuli. Cold stimuli were generated using a 30 ⫻ 30 –mm ATS thermode (Medoc, Ramat Yishai, Israel) and applied to an area of skin 5 cm distal and 2 cm anterior to the greater trochanter of the right hip while functional MRI acquisitions were obtained. During piloting, and based on normal range data for thermal cutaneous pain sensitivity on the lower extremity, it was established that a cold thermal stimulus of 12°C applied for 10 seconds would be used to dichotomize subjects into those who were normoesthetic to cold pain and those who were hypoesthetic to cold pain. The thermode cycled through a cooling/warming paradigm 8 times, as shown in Figure 1. After the scanning session, subjects were asked to rate if the cold stimulus was painful (stinging, burning, or prickling) during the experiment or not. Twelve of the patient group and all 12 of the control group underwent cold stimuli. The remaining 8 patients in the patient group did not undergo cold experimentation due to equipment failure during the research period. Punctate stimuli. In the second part of the same functional imaging session, punctate stimuli were applied to an area 5 cm distal and 2 cm anterior to the greater trochanter using a 256-mN von Frey hair (MARSTOCKnervtest2, Fruhstorfer, Germany) while functional MRI scanning took place. A total of 15 stimuli were applied over a 7-minute experimental period, guided by a predetermined timing model incorporating a jitter for optimal event-related sampling of the BOLD response. All 20 patients and 12 controls underwent punctate stimulation during functional MRI scanning. Subjects were instructed to keep their eyes closed and to keep as still as possible during the experiments. Analysis of stimulus-evoked functional MRI signal changes. The process of functional MRI analysis was performed as detailed in our previous study (33). All of the analyses were performed using the Centre for Functional MRI of the Brain (FMRIB) Expert Analysis Tool, version 5.67, which is part of the FMRIB Centre Software Library (online at www.fmrib.ox.ac.uk/fsl). The following standard preprocessing was applied to each subject’s time series of functional MRI volumes: nonbrain removal using a brain extraction tool (34), motion correction (35), spatial smoothing using a Gaussian kernel of full width at half maximum of 5 mm, demeaning of each voxel time course, and nonlinear high-pass temporal ﬁltering (cutoff of 100 seconds). The functional MRI signal in response to thermal or punctate stimulation was modeled using a general linear model approach. The regressor of interest was constructed by convolving the stimulus input function with a gamma hemodynamic response function (mean ⫾ SD lag time 6 ⫾ 3 seconds). The estimated time courses of the head motion parameters (translation in the x, y, and z direction and rotation about the x, y, and z axis) were included as covariates of no interest to further control for the subject movement. Brain registration included coregistration of the functional scan onto the individual T1 high-resolution structural image, and then registration onto a standard brain (Montreal Neurological Institute 152 brain, Montreal, Quebec, Canada) using the FMRIB Centre Nonlinear Image Registration Tool. For individual subject analysis, a general linear model was applied to these data on a voxel-by-voxel basis using the FMRIB Centre improved linear model to produce parameter estimates of the BOLD response to thermal and mechanical stimulation. All of the group analyses were performed using the FMRIB Centre Local Analysis of Mixed Effects (36). Functional MRI Evidence of Central Sensitization in Hip OA 1229 Table 1. Results of psychophysical testing performed in line with the quantitative sensory testing methods described by the German Research Network on Neuropathic Pain* Punctate detection threshold, median mN Cool detection threshold, °C Warm detection threshold, °C Cold pain detection threshold, °C Heat pain detection threshold, °C Sharpness of 256-mN stimulus (log10)§ Patients, right side (n ⴝ 12) Controls, right side (n ⴝ 12) P 0.5 27.4 ⫾ 1.73 37.3 ⫾ 2.6 8.3 ⫾ 13.2 44.6 ⫾ 2.3 1.37 ⫾ 0.30 16 27.0 ⫾ 1.48 36.2 ⫾ 1.64 12.2 ⫾ 11.6 45.7 ⫾ 2.7 1.12 ⫾ 0.23 ⬍ 0.001† 0.556‡ 0.215‡ 0.862† 0.298† 0.028‡ * Values are the mean ⫾ SD unless otherwise indicated. Results show the actual and statistical differences between skin sensitivity in the region of referred pain in patients and an identical area of skin in control subjects. † By Mann-Whitney test between groups. ‡ By unpaired t-test between groups. § The results reported by subjects in response to stimulation with a 256-mN von Frey hair, measured using a rating scale of 0 –100, showed that patients found this punctate stimulation signiﬁcantly more sharp/painful (median sharpness rating of 12.9 and 29.4 for the control and patient groups, respectively). However, because the data are not normally distributed, they are reported as their logarithmic transform to generate normally distributed data for statistical analysis so they can be compared with those produced by the German Research Network on Neuropathic Pain (35). Several conditions were explored: 1) punctate stimuli from both patient and control groups (n ⫽ 12), 2) cold stimuli from both patient and control groups (n ⫽ 12), and 3) punctate data from the patient group with low PainDETECT scores and punctate data from the patient group with high PainDETECT scores, grouped using a median split of their scores. Unfortunately, cold data for the patients (n ⫽ 20) were incomplete due to equipment problems during the course of the study, and therefore comparisons of cold pain processing between high and low PainDETECT groups was not possible. For the purpose of group analysis, the signiﬁcance threshold was Z scores greater than 2.3, with a cluster threshold of P values less than 0.05 to correct for multiple comparisons. Finally, based on the a priori hypothesis of activation within the brainstem descending modulatory PAG region being present in patients with cutaneous areas of referred pain and secondary hyperalgesia, this anatomic region was masked by hand on each of the subjects’ high-resolution structural scans, and a region of interest analysis was performed as described by previous authors (37). The mean volume of the PAG mask was 15 voxels (643 mm3). RESULTS Behavioral and psychophysical results. Both the patient and control groups had a similar ratio of men to women (6:6 and 5:7, respectively), with a mean ⫾ SD age of 63 ⫾ 8 years and 64 ⫾ 9 years, respectively. The high PainDETECT and low PainDETECT groups also had similar demographics (6:4 and 5:5 for the sex ratio, respectively, and a mean ⫾ SD age of 67 ⫾ 8 years and 62 ⫾ 8 years, respectively). Analysis of the data obtained by the questionnaires showed no signiﬁcant difference between the patient and control groups for the following domains: BDI (mean ⫾ SD 3.92 ⫾ 4.62 and 6.67 ⫾ 4.23, respectively; P ⫽ 0.14 by Student’s t-test), STAI (mean ⫾ SD 46.92 ⫾ 5.18 and 43.75 ⫾ 7.10, respectively; P ⫽ 0.280 by Mann-Whitney test), TRAI (mean ⫾ SD 46.83 ⫾ 2.70 and 46.58 ⫾ 3.66, respectively; P ⫽ 0.954 by Mann-Whitney test), and PCS (mean ⫾ SD 11.17 ⫾ 0.64 and 13.17 ⫾ 10.39, respectively; P ⫽ 0.613 by Student’s t-test). Signiﬁcant differences were found between the controls and patients for the following domains: TSK (mean ⫾ SD 22.5 ⫾ 8.37 and 31.33 ⫾ 5.61, respectively; P ⫽ 0.006 by Student’s t-test), PainDETECT (mean ⫾ SD 0 ⫾ 0 and 12.33 ⫾ 5.55, respectively; P ⬍ 0.001 by Mann-Whitney test), and average pain score over the last 4 weeks (mean ⫾ SD 0 ⫾ 0 and 5.56 ⫾ 2.02, respectively; P ⬍ 0.001 by Student’s t-test). The QST data were assessed for normality of distribution and logarithmic transforms applied to non-normally distributed data, which is in line with previous descriptions (32). Parametric statistics were applied where normality was seen in the original or transformed data. The results of this analysis are shown in Table 1. Imaging results. During functional MRI data acquisition, patients were asked to identify discomfort (described as stinging, pricking, or pain) during the cold pain paradigm. Nine subjects in the control group felt discomfort, whereas only 4 of the patients recorded similar sensations (P ⬍ 0.05 by chi-square test). A comprehensive list of activated areas for all conditions for the controls and patients is shown in Table 2, with representative images of these same activations shown in Figure 2. Supporting the psychophysical data, where signiﬁcantly more controls than patients found the cold stimuli painful, the mixed-effects analysis with the thresholds of Z score ⬎2.3 and P ⬍ 0.05 showed signiﬁcant activation 1230 Gwilym et al Table 2. Areas of brain activation induced by different stimuli applied to the area of referred pain on the right thigh in patients and a matched area in controls* Punctate stimuli in Punctate stimuli in Cold stimuli in the patient group the control group the control group Right insula Left insula Right SII Left SII Right SI Left SI Right lateral occipital cortex Left premotor cortex Right premotor cortex Right middle temporal gyrus Left middle temporal gyrus Left supramarginal gyrus Right supramarginal gyrus Left anterior cingulate cortex Right precentral gyrus Left precentral gyrus Right middle frontal gyrus Left middle frontal gyrus Left frontal pole Right angular gyrus 3.93 4.13 4.31 4.67 4.27 4.33 3.70 3.06 4.26 4.22 3.67 3.62 4.32 - 3.82 4.38 4.63 5.18 4.76 4.97 4.28 4.71 - 3.66 3.19 3.75 3.9 3.44 3.98 3.54 3.46 3.51 2.94 3.75 * Values are the peak Z score activity from the mixed-effects group analysis (corrected for multiple comparisons, Z score ⬎2.3, P ⬍ 0.05). Threshold activation was not achieved for any brain area on whole-brain analysis of the cold stimuli in the patient group at a Z score ⬎2.3. SII ⫽ secondary somatosensory cortex; SI ⫽ primary somatosensory cortex. in brain areas associated with pain perception when considering the contrast of more controls than patients (data not shown). There were no signiﬁcant activations (mixedeffects analysis, corrected for multiple comparisons, Z score ⬎2.3, P ⬍ 0.05) in the contrast of patients ⬎ controls in response to the cold stimulus. Considering punctate stimulation, the contrast analysis between the patient group and the control group, however, revealed signiﬁcantly greater activation in the patient group in the following regions: the anterior cingulate cortex, the right dorsolateral prefrontal cortex, the left middle frontal gyrus, and the left lateral occipital cortex (mixedeffects analysis, corrected for multiple comparisons, Z score ⬎2.3, P ⬍ 0.05). This patient group also reported this stimulus as signiﬁcantly more sharp compared with the control group (P ⫽ 0.028 by unpaired t-test). There were no activations that achieved threshold signiﬁcance (Z score ⬎2.3, P ⬍ 0.05) for the reverse contrast (controls ⬎ patients). Considering the whole-brain group means and contrasts between experimental groups (patient group mean, control group mean, patients ⬎ controls, and controls ⬎ patients), region of interest masks were applied to the PAG region based on our a priori hypothesis. The same group means and contrast analyses were performed, but limiting the results to this region of interest. The results of these analyses are shown in Figure 3A, and show an increased PAG activity in response to punctate stimulation of the referred pain area in patients (and homologous area in controls) compared with controls (patients ⬎ controls, mixed-effects analysis, corrected for multiple comparisons, Z score ⬎2.3, P ⬍ 0.05). Secondary analysis on our region of interest was done after performing a median split of the patients’ PainDETECT questionnaire scores into a low group and a high group. From this, low PainDETECT and high PainDETECT groups of patients were derived (mean ⫾ SD score 5.0 ⫾ 2.16 and 17.7 ⫾ 3.23, respectively; P ⬍ 0.0001 by Student’s t-test). These groups therefore represent those patients who display less (low PainDETECT score) and more (high PainDETECT score) neuropathic-like symptoms. From a psychophysical perspective, there was a signiﬁcant difference between the high and low PainDETECT groups in terms of how they rated the punctate stimulus, with the high PainDETECT group rating sharpness of the punctate stimulus signiﬁcantly higher than the low PainDETECT group (mean ⫾ SD sharpness rating 22.3 ⫾ 16.1 and 48.5 ⫾ 25.2, respectively; P ⬍ 0.05 by Student’s t-test). The imaging data derived from contrasting the activity in the PAG region of interest in response to punctate stimulation for the contrast of more high PainDETECT scores than low PainDETECT scores are shown in Figure 3B. This clearly shows a signiﬁcantly increased PAG activity in response to punctate stimulation of the referred pain area in the high PainDETECT group compared with the low PainDETECT group (mixed-effects analysis, corrected for multiple comparisons, Z score ⬎2.3, P ⬍ 0.05). The relationship between PAG activation and clinical manifestation of central sensitization in patients, as deﬁned by their PainDETECT score, is further shown by the signiﬁcant positive correlation of these 2 variables (Figure 4). Functional MRI Evidence of Central Sensitization in Hip OA Figure 2. Results of a mixed-effects analysis of the average group response for the 2 groups (controls and patients) in response to the 2 stimuli (cold and punctate). All of the results are corrected for multiple comparisons (Z score ⬎2.3, P ⬍ 0.05). The images shown are z ⫽ 8 and y ⫽ ⫺18, and are representative. All areas of activation above the threshold limits are listed in Table 2. Images are shown anatomically. SI ⫽ primary somatosensory cortex; SII ⫽ secondary somatosensory cortex; ACC ⫽ anterior cingulate cortex. Figure 3. A, Mixed-effects group analysis for periaqueductal grey (PAG) activation for the contrast of patients ⬎ controls in response to punctate stimuli (masked for PAG, thresholds of Z score ⬎2.3 and P ⬍ 0.05, corrected for multiple comparisons). B, Mixed-effects group analysis for PAG activation for the contrast of high PainDETECT ⬎ low PainDETECT (analysis masked for the PAG, thresholds of Z score ⬎2.3 and P ⬍ 0.05, corrected for multiple comparisons). 1231 1232 Figure 4. Correlation between clinical manifestations of central sensitization (as shown by total score on the PainDETECT) and periaqueductal grey (PAG) activation in response to punctate stimulation in patients. Activation is deﬁned using the percentage change in the parameter estimate of the peak active voxel (vmax) within the masked region of interest of the PAG (r ⫽ 0.60, P ⫽ 0.006). DISCUSSION The discordance between the degree of articular pathology and pain experienced by patients with OA is a longstanding and validated observation. This anomaly may be partly explained by the variable modulation exerted on the peripheral nociceptive input to the spinal dorsal horn between individuals from the descending modulatory system. These descending systems affect dorsal horn excitability and as a result, cause the individual to alter the magnitude of nociceptive inputs relayed to supraspinal structures, where these signals are further processed to produce the experience of pain. In addition to the variability in pain perception with OA, some patients experience symptoms considered more typical of neuropathic pain, such as sudden electric shock sensations, allodynia, and referred pain. We ﬁrst used the patient history of referred pain, and then the presence of neuropathic-like elements of pain identiﬁed by the questionnaires plus alteration in the results of QST, to identify patients with speciﬁc neuropathic-like symptoms. We considered several of these symptoms to be manifestations of central sensitization, with the aim of investigating, by the use of functional neuroimaging, the involvement of supraspinal structures capable of modulating spinal nociceptive transmission. Central sensitization in musculoskeletal disease identiﬁed by lowered pressure pain thresholds (6,38), increased area of referred pain secondary to induced muscle pain (39), and hyperalgesia to mechanical stimuli (5,8) have been previously reported, and our results for punctate stimulation (Table 2) conﬁrm these ﬁndings in the areas of referred pain in our cohort of patients with hip OA. The work described here attempted to address the mechanisms behind these observations. The need for such insights in order to develop rational new therapies for OA pain has been recently highlighted (38). Functional MRI is Gwilym et al used because it offers a valid method of identifying how and where pain processing occurs in the human brain (40,41), and also important, how this process differs between groups. Activation within the brain areas associated with pain perception is shown in Figure 2. These results parallel the psychophysical response of the 2 groups to the stimulation modalities. Pain perception by the control group in response to cold stimulation, and by both groups in response to punctate stimulation, is associated with activity in areas commonly associated with the pain experience (primary and secondary somatosensory cortices, insular and prefrontal cortices). An absence of perceived pain in the patient group exposed to cold stimulation was associated with a lack of activation of similar pain-relevant brain regions. Signiﬁcant behavioral punctate hyperalgesia in the patient group compared with the normoalgesic control group was associated with additional activity in the patients’ anterior cingulate cortex and ipsilateral dorsolateral prefrontal cortex, together with the contralateral left middle frontal gyrus and left lateral occipital cortex activity. Other studies published in the literature investigating different clinical conditions with possible central sensitization present but not characterized as done in this study have shown additional active brain areas when compared with controls (42,43). However, those studies are methodologically quite different from ours in terms of stimulus modality and data analysis; additionally, those patients experienced higher pain compared with patients in our study and it is likely that this fact mostly accounts for any differences. In response to punctate stimulation in the patient group, the activation deﬁned using a region of interest PAG mask is signiﬁcantly greater compared with controls (Figure 3A). This is associated with clinical hyperalgesia to the stimulus. The a priori hypothesis that the brainstem would be preferentially activated in conditions where neuropathiclike symptoms are present was based on previous work investigating brainstem activity in experimentally induced secondary hyperalgesia and central sensitization (16, 20,21). We can therefore perhaps consider activity within the brainstem to be a biomarker of central sensitization in humans. Figure 3A shows a key ﬁnding from our study: that punctate hyperalgesia in patients is indeed associated with signiﬁcant PAG activity compared with controls, which is in line with our a priori hypothesis. The observation of increased brainstem activation during punctate hyperalgesia is further explored in Figure 3B by the division of the patients into 2 groups using the median split of their PainDETECT scores. The group with higher PainDETECT scores, who also had signiﬁcantly greater ratings of punctate hyperalgesia on sensory testing, showed a signiﬁcantly higher activation in the PAG region compared with patients with less neuropathic symptoms and signs. It therefore appears that PAG activity is related to the clinical manifestations of disease rather than the presence of disease alone, and this is further supported by the signiﬁcant positive correlation between PainDETECT scores and PAG activity (Figure 4). One potential confound in this study is the differences Functional MRI Evidence of Central Sensitization in Hip OA in medication use by the various groups of patients (controls versus patients, high PainDETECT versus low PainDETECT scores). Although both the patients and controls were excluded from the study if they were taking neuroleptic medications, there were predictable differences in the uptake of other categories of medications between patients and controls. Speciﬁcally, the patients were signiﬁcantly more likely to be taking opioids and nonsteroidal antiinﬂammatory drugs (NSAIDs; P ⬍ 0.001 by chi-square test). The uptake of other categories of medications (antihypertensives, others) was not signiﬁcantly different between the patients and controls. Between the high PainDETECT score and low PainDETECT score groups, there were no signiﬁcant differences in the uptake of any category of medications (opioid analgesics, NSAIDs, antihypertensives, others; P ⬎ 0.05 using chi-square analysis for all comparisons). Current analgesic treatments address both peripheral and central processes in pain perception, although the bias within the arthritis ﬁeld remains toward the periphery (44). Animal studies using arthritis models investigating the analgesic efﬁcacy of centrally acting compounds known to be effective in neuropathic conditions (i.e., gabapentin) support the view that targeting both the peripheral origin and central promoters of pain may be a valuable approach in OA (45). Furthermore, there is some clinical evidence of efﬁcacy in OA of duloxetine, a monoamine reuptake inhibitor (46), suggesting that at least some patients with painful OA may beneﬁt from treatments targeting endogenous pain modulatory systems. Our results in humans support these new ﬁndings, and describe some neural markers that can be used to select patients whose pain is exacerbated and confounded by the likely presence of central sensitization. Pain in OA is often associated with hyperalgesia, referred pain, and spontaneous pain (pain at rest) (47,48). These features cannot be explained by peripheral changes and peripheral sensitization alone. We have shown in humans that the PAG matter is involved in the neuroplasticity associated with central sensitization in osteoarthritic pain. In the future, a mechanism-based classiﬁcation of the differing manifestations of osteoarthritic pain will promote development of mechanism-aligned analgesic strategies. To our knowledge, this work is the ﬁrst to explore the supraspinal mechanisms that underlie the clinical manifestations of referred pain and changes in skin sensitivity in OA patients. The need for such information has been recently highlighted (38). Understanding the implications of these bedside observations may allow patient care to be directed toward their individual pain phenotype, with patients displaying signs of central sensitization having both the peripheral and central components to their pain managed to improve patient outcomes. ACKNOWLEDGMENTS The authors would like to thank the surgeons at the Nufﬁeld Orthopaedic Centre, Oxford, for their assistance in identifying suitable patients for this study. 1233 AUTHOR CONTRIBUTIONS All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the ﬁnal version to be published. Dr. Gwilym had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design. Gwilym, Keltner, Carr, Chessell, Tracey. Acquisition of data. Gwilym, Keltner, Warnaby. Analysis and interpretation of data. Gwilym, Keltner, Chizh, Tracey. ROLE OF THE STUDY SPONSOR GlaxoSmithKline was involved in the hypothesis generation, study design, and partial funding of the project. The content of the manuscript was approved by our GlaxoSmithKline collaborators, but publication of the manuscript was not contingent on the approval of GlaxoSmithKline. REFERENCES 1. Kellgren JH, Lawrence JS. Rheumatism in miners. II. X-ray study. Br J Ind Med 1952;9:197–207. 2. Khan NQ, Woolson ST. Referral patterns of hip pain in patients undergoing total hip replacement. Orthopedics 1998; 21:123– 6. 3. Nakajima T, Ohtori S, Inoue G, Koshi T, Yamamoto S, Nakamura J, et al. The characteristics of dorsal-root ganglia and sensory innervation of the hip in rats. J Bone Joint Surg Br 2008;90:254 –7. 4. Street J, Lenehan B, Flavin R, Beale E, Murray P. Do pain referral patterns determine patient outcome after total hip arthroplasty? Acta Orthop Belg 2005;71:540 –7. 5. Farrell M, Gibson S, McMeeken J, Helme R. Pain and hyperalgesia in osteoarthritis of the hands. J Rheumatol 2000;27: 441–7. 6. Ordeberg G. Characterization of joint pain in human OA. Novartis Found Symp 2004;260:105–21. 7. Kosek E, Ordeberg G. Abnormalities of somatosensory perception in patients with painful osteoarthritis normalize following successful treatment. Eur J Pain 2000;4:229 –38. 8. Hendiani JA, Westlund KN, Lawand N, Goel N, Lisse J, McNearney T. Mechanical sensation and pain thresholds in patients with chronic arthropathies. J Pain 2003;4:203–11. 9. Head H. On disturbances of sensation with especial reference to the pain of visceral disease. Brain 1893;16:1–133. 10. Graven-Nielsen T. Fundamentals of muscle pain, referred pain, and deep tissue hyperalgesia. Scand J Rheumatol Suppl 2006;122:1– 43. 11. Treede RD, Meyer RA, Raja SN, Campbell JN. Peripheral and central mechanisms of cutaneous hyperalgesia. Prog Neurobiol 1992;38:397– 421. 12. Pinto M, Lima D, Tavares I. Neuronal activation at the spinal cord and medullary pain control centers after joint stimulation: a c-fos study in acute and chronic articular inﬂammation. Neuroscience 2007;147:1076 – 89. 13. Sharif Naeini R, Cahill CM, Ribeiro-da-Silva A, Menard HA, Henry JL. Remodelling of spinal nociceptive mechanisms in an animal model of monoarthritis. Eur J Neurosci 2005;22: 2005–15. 14. Schaible HG, Schmelz M, Tegeder I. Pathophysiology and treatment of pain in joint disease. Adv Drug Deliv Rev 2006; 58:323– 42. 15. Tracey I, Mantyh PW. The cerebral signature for pain perception and its modulation. Neuron 2007;55:377–91. 16. Urban MO, Gebhart GF. Supraspinal contributions to hyperalgesia. Proc Natl Acad Sci U S A 1999;96:7687–92. 17. Suzuki R, Dickenson A. Spinal and supraspinal contributions to central sensitization in peripheral neuropathy. Neurosignals 2005;14:175– 81. 1234 18. Gebhart GF. Descending modulation of pain. Neurosci Biobehav Rev 2004;27:729 –37. 19. Coderre TJ, Katz J, Vaccarino AL, Melzack R. Contribution of central neuroplasticity to pathological pain: review of clinical and experimental evidence. Pain 1993;52:259 – 85. 20. Zambreanu L, Wise RG, Brooks JC, Iannetti GD, Tracey I. A role for the brainstem in central sensitisation in humans: evidence from functional magnetic resonance imaging. Pain 2005;114:397– 407. 21. Lee MC, Zambreanu L, Menon DK, Tracey I. Identifying brain activity speciﬁcally related to the maintenance and perceptual consequence of central sensitization in humans. J Neurosci 2008;28:11642–9. 22. Becerra L, Morris S, Bazes S, Gostic R, Sherman S, Gostic J, et al. Trigeminal neuropathic pain alters responses in CNS circuits to mechanical (brush) and thermal (cold and heat) stimuli. J Neurosci 2006;26:10646 –57. 23. Sandrini G, Rossi P, Milanov I, Serrao M, Cecchini AP, Nappi G. Abnormal modulatory inﬂuence of diffuse noxious inhibitory controls in migraine and chronic tension-type headache patients. Cephalalgia 2006;26:782–9. 24. Mayer EA, Berman S, Suyenobu B, Labus J, Mandelkern MA, Naliboff BD, et al. Differences in brain responses to visceral pain between patients with irritable bowel syndrome and ulcerative colitis. Pain 2005;115:398 – 409. 25. Wilder-Smith CH, Schindler D, Lovblad K, Redmond SM, Nirkko A. Brain functional magnetic resonance imaging of rectal pain and activation of endogenous inhibitory mechanisms in irritable bowel syndrome patient subgroups and healthy controls. Gut 2004;53:1595– 601. 26. Beck AT, Ward CH, Mendelson M, Mock J, Erbaugh J. An inventory for measuring depression. Arch Gen Psychiatry 1961;4:561–71. 27. Sullivan MJ, Stanish W, Waite H, Sullivan M, Tripp DA. Catastrophizing, pain, and disability in patients with softtissue injuries. Pain 1998;77:253– 60. 28. Vlaeyen JW, Kole-Snijders AM, Boeren RG, van Eek H. Fear of movement/(re)injury in chronic low back pain and its relation to behavioral performance. Pain 1995;62:363–72. 29. Smith RC, Lay CD. State and trait anxiety: an annotated bibliography. Psychol Rep 1974;34:519 –94. 30. Freynhagen R, Baron R, Gockel U, Tolle TR. PainDETECT: a new screening questionnaire to identify neuropathic components in patients with back pain. Curr Med Res Opin 2006; 22:1911–20. 31. Bennett MI, Attal N, Backonja MM, Baron R, Bouhassira D, Freynhagen R, et al. Using screening tools to identify neuropathic pain. Pain 2007;127:199 –203. 32. Rolke R, Baron R, Maier C, Tolle TR, Treede RD, Beyer A, et al. Quantitative sensory testing in the German Research Gwilym et al 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. Network on Neuropathic Pain (DFNS): standardized protocol and reference values. Pain 2006;123:231– 43. Smith SM. Overview of fMRI analysis. Br J Radiol 2004;77: S167–75. Smith SM. Fast robust automated brain extraction. Hum Brain Mapp 2002;17:143–55. Jenkinson M, Smith S. A global optimisation method for robust afﬁne registration of brain images. Med Image Anal 2001;5:143–56. Woolrich MW, Behrens TE, Beckmann CF, Jenkinson M, Smith SM. Multilevel linear modelling for FMRI group analysis using Bayesian inference. Neuroimage 2004;21:1732– 47. Poldrack RA. Region of interest analysis for fMRI. Soc Cogn Affect Neurosci 2007;2:67–70. Imamura M, Imamura ST, Kaziyama HH, Targino RA, Hsing WT, de Souza LP, et al. Impact of nervous system hyperalgesia on pain, disability, and quality of life in patients with knee osteoarthritis: a controlled analysis. Arthritis Rheum 2008;59:1424 –31. Bajaj P, Bajaj P, Graven-Nielsen T, Arendt-Nielsen L. Osteoarthritis and its association with muscle hyperalgesia: an experimental controlled study. Pain 2001;93:107–14. Schweinhardt P, Lee M, Tracey I. Imaging pain in patients: is it meaningful? Curr Opin Neurol 2006;19:392– 400. Borsook D, Becerra L. Functional imaging of pain and analgesia: a valid diagnostic tool? Pain 2005;117:247–50. Gracely RH, Petzke F, Wolf JM, Clauw DJ. Functional magnetic resonance imaging evidence of augmented pain processing in ﬁbromyalgia. Arthritis Rheum 2002;46:1333– 43. Giesecke T, Gracely RH, Grant MA, Nachemson A, Petzke F, Williams DA, et al. Evidence of augmented central pain processing in idiopathic chronic low back pain. Arthritis Rheum 2004;50:613–23. Gwilym SE, Pollard TC, Carr AJ. Understanding pain in osteoarthritis. J Bone Joint Surg Br 2008;90:280 –7. Bove SE, Laemont KD, Brooker RM, Osborn MN, Sanchez BM, Guzman RE, et al. Surgically induced osteoarthritis in the rat results in the development of both osteoarthritis-like joint pain and secondary hyperalgesia. Osteoarthritis Cartilage 2006;14:1041– 8. Sullivan MD, Bentley S, Fan MY, Gardner G. A single-blind, placebo run-in study of duloxetine for activity-limiting osteoarthritis pain. J Pain 2009;10:208 –13. Schaible HG, Ebersberger A, von Banchet GS. Mechanisms of pain in arthritis. Ann N Y Acad Sci 2002;966:343–54. Konttinen YT, Kemppinen P, Segerberg M, Hukkanen M, Rees R, Santavirta S, et al. Peripheral and spinal neural mechanisms in arthritis, with particular reference to treatment of inﬂammation and pain. Arthritis Rheum 1994;37:965– 82.