Normalization of widespread hyperesthesia and facilitated spatial summation of deep-tissue pain in knee osteoarthritis patients after knee replacement




The modest association between radiographic joint damage and pain in osteoarthritis (OA) has led to the suggestion of facilitated central pain processing. This study evaluated the importance of ongoing tissue pathology in the maintenance of enhanced central pain processing.


Pain assessment was performed on 48 patients with symptomatic knee OA and 21 sex- and age-matched pain-free healthy control subjects. Twenty of the OA patients subsequently underwent total knee replacement surgery and were reassessed. Pressure–pain thresholds (PPTs) were recorded using a pressure algometer (both over and distant from the knee) and a double-chamber inflatable cuff mounted around the calf. Spatial summation was assessed by relating PPTs using the dual- and single-chamber cuff. Conditioned pain modulation (CPM) was assessed by recording the increase in PPT in response to experimental arm pain.


PPTs at the knee and at sites away from the knee were reduced in OA patients as compared with healthy pain-free control subjects (P < 0.0001). Cuff PPTs were decreased in OA patients as compared with the healthy controls (P < 0.05), who also exhibited a greater degree of spatial summation (P < 0.05). Whereas an elevation of PPTs was noted in the healthy controls in response to experimental arm pain (P < 0.0001), no such CPM was observed in the OA patients. Following joint replacement in the OA patients, there was a reduction in the widespread mechanical hyperesthesia, along with normalization of spatial summation ratios and restoration of CPM.


The widespread hyperesthesia and enhanced spatial summation observed in OA patients imply sensitized central pain mechanisms together with the loss of CPM. Normalization of the results following joint replacement implies that these central pain processes are maintained by peripheral input.

Symptoms in osteoarthritis (OA) vary considerably, although patterns of pain can be distinguished by different patients, varying from dull aches to sharp stabbing pains. Neuropathic pain descriptors reported by a subset of OA patients indicated the role of sensitization processes in one study (1). Use-related pain is common, although pain when at rest and pain during the night also occur (2). Imaging studies have shown only a modest correlation between radiologic findings and pain (3), although more recent studies using magnetic resonance imaging have shown associations with specific synovial and bone abnormalities (4, 5).

In painful musculoskeletal conditions, such as chronic low back pain or fibromyalgia, functional brain imaging studies have shown increased neural activity in brain structures involved in the processing of sensation, movement, cognition, and emotion (6). Similar findings have been reported in patients with symptomatic OA (7). Quantitative sensory testing complements functional imaging studies and permits both static and dynamic evaluation of pain mechanisms. Increased pain sensitivity in response to a variety of mechanical, thermal, and chemical stimuli has been reported in patients with OA at sites both over and away from symptomatic joints (8–11).

Continuous nociceptive input from an OA-damaged knee joint in animals drives central sensitization (12). Other features of central sensitization at the level of the dorsal horn include temporal and spatial summation, characterized by an increased pain response to repeated stimulation or to a larger area of stimulation, respectively (13). Facilitated temporal summation has been demonstrated in fibromyalgia pain patients (14–16) and more recently in patients with symptomatic knee OA (10). Spatial summation is an important neural mechanism for pain coding (17), but while there have been some studies on spatial summation in fibromyalgia (18, 19), no studies to date have investigated this phenomenon in patients with OA. Enhanced central summation is seen for other patient populations with chronic musculoskeletal pain as an indicator of central sensitization (20).

Neuronal activity within the spinal dorsal horn arising from peripheral nociceptive inputs can be modulated by powerful descending inhibitory and facilitatory mechanisms (21). An example is provided by the phenomenon of conditioned pain modulation (CPM) in humans, also known as diffuse noxious inhibitory control, which refers to a change in the response to a painful stimulus following the administration of a second conditioning pain stimulus (22). Musculoskeletal disorders in which this response is absent or blunted include OA of the hip (23) and knee (10), temporomandibular disorders (24), and fibromyalgia (25). Restoration of the CPM response has been reported in patients with hip OA following joint replacement surgery (23), suggesting that it is maintained by abnormal peripheral nociceptive inputs. Consistent with these reports, functional imaging studies of patients with hip OA show abnormal activity in the brainstem periaqueductal gray matter, an area known to be involved in the modulation of mechanical stimuli via descending modulatory pathways (7).

Widespread mechanical hyperesthesia observed in symptomatic patients with OA may arise, at least in part, as a consequence of altered pain modulation at a spinal and/or supraspinal level that nevertheless remains contingent on abnormal peripheral neuronal inputs. In this study, we hypothesized that successful joint replacement surgery would lead to restoration of normal findings on sensory testing at sites away from damaged joints. The primary aim of this study was therefore to assess the reversibility of a range of psychophysical markers of pain processing in patients with symptomatic knee OA before and after knee replacement surgery. These included 1) pressure–pain thresholds, 2) spatial summation of pressure–pain, and 3) conditioning pain modulation.


Study subjects.

Forty-eight OA patients (average age 65 years [range 40–86 years]; 12 men) who had been diagnosed radiologically as having knee OA either unilaterally or bilaterally were recruited from rheumatology clinics. A subgroup of 20 of these OA patients (average age 68 years [range 48–86 years]; 6 men) were scheduled for knee replacement surgery, and these patients were further evaluated 5–28 weeks following surgery (60% reassessed after 9–18 weeks). Twenty-one sex- and age-matched healthy control subjects who had never had a painful joint disorder (average age 60 years [range 40–81 years]; 4 men) were recruited from Bart's and the London NHS Trust or were siblings or spouses of the patients.

Knee OA was diagnosed according to American College of Rheumatology classification criteria (26). Inclusion criteria were severe pain for >3 months and a pain score of ≥4 cm on a 0–10-cm visual analog scale (VAS). Clinical data, such as findings of the radiologic evaluation, current medications, and duration of pain were collected. Patients underwent bilateral weight-bearing, fixed-flexion posteroanterior and lateral radiography of the knees. Degenerative changes in the tibiofemoral joint were verified in each patient. Participants had no other pain problems or sensory dysfunctions (e.g., nerve damage) and were not mentally impaired. They were requested not to take analgesic medication 48 hours before the study assessment.

This study was approved by local ethics committee (approval 06/Q0603/138) and was conducted in accordance with the Declaration of Helsinki. All study subjects read and signed informed consent forms.

Study protocol.

Prior to assessment, all subjects were interviewed to obtain medical and medication history (including duration of pain symptoms) and a present pain intensity rating at rest, which was assessed using a 10-cm VAS anchored at no pain (0 cm) and maximal pain (10 cm). Examinations took place in a quiet, temperature-controlled room with the subject in a relaxed supine position. OA patients were asked to indicate the most affected painful knee joint. Assessment methods were carefully explained to all subjects before the assessment was begun so that they would be familiar with the procedure.

Sensory assessments consisted of the following 4 modalities. Pressure–pain thresholds (PPTs) were measured over several test sites (in the peripatellar region) and control sites (forearm and lower leg). PPTs were reassessed during tonic pain induced by ischemic exercise of the left arm (heterotopic stimulation to evoke CPM). Cuff PPTs were assessed at the lower leg. Spatial summation was assessed by 2 different volumes of cuff pressure stimulation. Bilateral assessments were performed, and the sequence (side and assessment procedures) was randomized. The same observer (TW) collected all of the data.

Pressure algometry.

A hand-held pressure algometer (Somedic Algometer Type II) was used for measuring PPTs. The probe (1 cm2) was placed perpendicular to the skin, and pressure was applied (30 kPa/second) until the subject perceived the pressure as pain, at which point he or she pressed a button, and the procedure was immediately terminated. The PPT was measured 2 or 3 times at each site, and the average was used for statistical analysis. The majority of PPTs were recorded 3 times but, occasionally, only twice because of experimental limitations. An ∼20–30-second interval between assessments was maintained.

For each subject, 7 test sites in the peripatellar region, 1 control site on the tibialis anterior muscle (5 cm distal to the tibial tuberosity), and 1 control site on the extensor carpi radialis longus muscle (5 cm distal to the lateral epicondyle of the humerus) were located and marked on each side (Figure 1). The 7 peripatellar sites were located in relation to bone landmarks. Site 1 was located 2 cm proximal to the superior edge of the patella. Site 2 was located 2 cm proximal to the superior lateral edge of the patella. Site 3 was located 2 cm distal to the inferior lateral edge of the patella. Site 4 was located 2 cm distal to the inferior medial edge of the patella. Site 5 was located 3 cm medial to the midpoint on the medial edge of the patella. Site 6 was located 2 cm proximal to the superior medial edge of the patella. Site 7 was located at the center of the knee.

Figure 1.

Assessment sites for pressure algometry. A total of 7 test sites in the peripatellar region, 1 control site on the tibialis anterior (Tib Ant) muscle (5 cm distal to the tibial tuberosity), and 1 control site on the extensor carpi radialis longus muscle (5 cm distal to the lateral epicondyle of the humerus; site not shown) were evaluated.

Cuff pressure algometry.

The experimental setup consisted of a double-chamber, 13-cm–wide tourniquet cuff (a silicone high-pressure cuff, separated lengthwise into two chambers of equal size; VBM Medizintechnik), a computer-controlled air compressor, and an electronic 10-cm VAS (Aalborg University, Aalborg, Denmark). The cuff was connected to the compressor and wrapped around the midportion of the gastrocnemius–soleus muscles (∼20 cm distal to the knee). The pain intensity was recorded with the electronic VAS and sampled at 10 Hz. The 0-cm and 10-cm anchor scores on the electronic VAS were defined as “no pain” and “maximal pain,” respectively (27). The compression rate of the cuff was preset and controlled by the computer. The maximum pressure limit used was 100 kPa (760 mm Hg). A hand-held release button allowed the immediate termination of stimulation.

The cuff was automatically inflated (inflation rate 1 kPa/second). The subject was instructed to continuously rate on the VAS the pain intensity, beginning at the first sensation of pain, and to press the release button at “the pain intensity strong enough to make one feel like interrupting or stopping it.”

The pressure–pain threshold was defined as the moment of transition between strong pressure and painful pressure (i.e., the first time the VAS score exceeds 0). The pressure value at the termination of pressure inflation was defined as the pressure–pain tolerance, and the corresponding VAS score was defined as the pressure–pain limit. The mean of 3 recordings, with 5-minute intervals between, was used. Assessments were performed on both legs using the same protocol.

Spatial summation was evaluated after inflating one or both chambers of the tourniquet cuff randomly on both legs (equivalent to a stimulation area of 241 cm2 or 482 cm2, respectively). The subject was blinded as to whether a single chamber or both chambers were being inflated. The ratio of thresholds from the double-chamber cuff divided by the thresholds from the single-chamber cuff was used as an index for spatial summation.

Assessment of CPM.

Ischemic compression of the left arm was used as heterotopic noxious conditioning stimulation for evoking CPM. A cuff was wrapped around the middle of the arm (ipsilateral to the painful knee), with the lower rim of the tourniquet cuff (7.5-cm wide; VBM Medizintechnik) positioned 3 cm proximal to the cubital fossa. The pressure cuff was inflated above systolic pressure (200 mm Hg) for 10 minutes. After the target pressure was reached, the patient was asked to perform hand grips for 10 times or more until 4 cm on the VAS was reached (0 cm represented “no pain” and 10 cm represented “maximal pain”). When 4 cm on the VAS was reached, PPTs were assessed at point 3 and point 4 (Figure 1) (average value was used for CPM analysis), and cuff pressure–pain sensitivity was assessed once on the lower leg. The arm cuff was released once PPT and cuff pain assessments were finished (after a maximum of 10 minutes).

Statistical analysis.

Results are presented as the mean ± SEM. Analysis of variance (ANOVA) was performed on PPTs using the following factors: study group (OA/control), side (affected/contralateral), and site (9 sites tested). A mixed-model ANOVA was used to analyze PPTs before and after surgery, with group factors side and site as well as repeated factor time (presurgery and postsurgery). A similar approach was used for cuff algometry parameters, where the ANOVA factors were subject (OA/control) and side (affected/contralateral). To analyze cuff algometry parameters before and after surgery, a mixed-model ANOVA was used, with group factor side and repeated factor time (presurgery/postsurgery). The CPM (i.e., how much the PPT increased with and without experimental arm pain) was analyzed by a mixed-model ANOVA, with group factor subject (OA/controls) and repeated factor modulation (with/without arm pain). The CPM data before and after surgery were examined with a repeated-measures ANOVA, with factors modulation (with/without arm pain) and time (presurgery/postsurgery). The Newman-Keuls (NK) test incorporating correction for multiple comparisons was used as a post hoc test in cases of significant ANOVA factors or interactions. Pearson's product-moment test was used for correlation analysis between VAS scores and PPT values (arm, tibialis anterior muscle, mean of peripatellar sites) in patients and controls; Bonferroni correction was applied to the P values to adjust for multiple correlations. P values less than or equal to 0.05 were considered significant.


Pain profiles in OA patients.

The knee OA patients in the present study had had their knee pain symptoms for a mean ± SEM of 80 ± 11 months. Their VAS scores for pain were significantly higher than those in the sex- and age-matched healthy control subjects (mean ± SEM 6.4 ± 0.3 cm versus 0.1 ± 0.1 cm; P < 0.0001). In 67% of the OA patients, the left side was designated as the most affected side. Among the OA patients who underwent total knee replacement surgery, the VAS scores were significantly reduced after surgery (2.6 ± 0.5 cm) as compared with before surgery (6.9 ± 0.4 cm; P < 0.0002).

Pressure algometry of the knee, lower leg, and arm.

Pressure–pain thresholds at the knee, the tibialis anterior muscle (spreading sensitization), and the extensor carpi radialis longus muscle (spreading sensitization) were significantly reduced in OA patients as compared with healthy controls. A two-way interaction between the most affected side and the study group (F[1,1] = 17.5, P < 0.0001 by ANOVA; P < 0.0001 by NK test) showed that PPTs in OA patients were significantly reduced as compared with those in healthy controls and that the most affected side in OA patients had significantly lower PPTs as compared with the contralateral side (Figure 2). Moreover, an interaction between sites and study group showed that all sites in OA patients, except sites 1 and 4, were significantly decreased as compared with healthy controls (F[1,1] = 2.06, P < 0.04 by ANOVA; P < 0.01 by NK test). Comparison of PPTs at all assessment sites showed that they were significantly different (F[8] = 151.2, P < 0.0001 by ANOVA; P < 0.002 by NK test). Site 4 presented the lowest PPTs as compared with all sites, and site 1 showed the highest PPTs as compared with all other sites except site 3.

Figure 2.

Pressure–pain thresholds (PPTs) in healthy control subjects (n = 21) and in patients with osteoarthritis (OA) of the knee (n = 46–48). PPTs were assessed at the knee (see Figure 1 for specific sites), the arm, and the tibialis anterior (Tib ant) muscle on the most affected side as well as on the contralateral side. All sites except 1 and 4 showed significantly reduced PPTs in the OA patients as compared with the healthy controls (∗ = P < 0.01 by Newman-Keuls test). The most affected side in the OA patients had significantly lower PPTs as compared with the contralateral side (# = P < 0.0001 by Newman-Keuls test). Values are the mean ± SEM.

PPTs were related to knee pain intensity (Figure 3). We found significant correlations between the VAS scores and the average PPTs on the knee (R = –0.37, P < 0.00003), the arm (R = –0.51, P < 0.00001), and the tibialis anterior muscle (R = –0.46, P < 0.00001).

Figure 3.

Significant correlations between the knee pain intensity (as determined by scores on a 10-cm visual analog scale [VAS]) and the pressure–pain thresholds in the knee (mean of all peripatellar sites), arm (extensor carpi radialis longus muscle), and lower leg (tibialis anterior muscle) in the osteoarthritis patients and healthy controls. Each symbol represents an individual subject.

In the group of 20 patients who underwent total knee replacement of the most affected knee, the PPTs on both sides and at all sites were significantly increased after surgery as compared with the values before surgery (F[1,1] = 4.36, P < 0.04 by ANOVA; P < 0.0001 by NK test) (Figure 4). Presurgery and postsurgery PPTs at all sites on the most affected side were significantly reduced as compared with pre- and postsurgery PPTs on the contralateral side (P < 0.0001 by NK test).

Figure 4.

Pressure–pain thresholds (PPTs) in patients with osteoarthritis (OA) of the knee (n = 17–20) before versus after total knee replacement of the most affected knee. PPTs were assessed at the knee (see Figure 1 for specific sites), the arm, and the tibialis anterior (Tib ant) muscle on the most affected side as well as on the contralateral side. Compared with the values presurgery, the postsurgery PPTs were significantly increased at all sites (∗ = P < 0.0001 by Newman-Keuls test). PPTs at all sites on the most affected side were significantly reduced as compared with the contralateral side (# = P < 0.0001 by Newman-Keuls test). Values are the mean ± SEM.

Cuff algometry on the lower leg.

Cuff PPTs were significantly decreased in OA patients as compared with healthy controls for both sides as well as for single-chamber and double-chamber determinations at the lower leg (F[1] > 7.02, P < 0.009 by ANOVA; P < 0.008 by NK test) (Table 1). The spatial summation ratio was significantly reduced in OA patients as compared with healthy controls independently of assessment side (F[1] = 3.83, P ≤ 0.05 by ANOVA; P ≤ 0.05 by NK test) (Table 1).

Table 1. Cuff algometry findings in OA patients and healthy controls, as well as in a subgroup of OA patients before and after TKR*
 Single-cuff pain threshold, mean ± SEM kPaDouble-cuff pain threshold, mean ± SEM kPaSummation ratio, mean ± SEM
  • *

    Cuff pressure–pain thresholds and summation ratios were recorded on the lower leg in 46 osteoarthritis (OA) patients and 21 healthy control subjects, as well as before and after surgery in a subgroup of 20 OA patients who underwent total knee replacement (TKR) of the most affected knee.

  • P ≤ 0.05 versus controls, by Newman-Keuls test.

  • P < 0.006 versus presurgery, by Newman-Keuls test.

Most affected side   
 OA32.9 ± 2.025.2 ± 1.60.78 ± 0.03
 Controls40.4 ± 3.634.5 ± 2.70.87 ± 0.03
Contralateral side   
 OA34.0 ± 2.227.2 ± 1.80.83 ± 0.03
 Controls41.9 ± 4.335.3 ± 3.00.89 ± 0.05
Most affected side   
 OA before TKR31.2 ± 3.721.5 ± 2.60.70 ± 0.03
 OA after TKR34.7 ± 3.331.3 ± 3.40.91 ± 0.05
Contralateral side   
 OA before TKR30.3 ± 3.924.9 ± 3.50.85 ± 0.06
 OA after TKR33.3 ± 3.528.4 ± 3.10.87 ± 0.04

After total knee replacement, the cuff pain thresholds in response to double-chamber stimulation were significantly increased in both legs as compared with the presurgery values (F[1] = 8.59, P < 0.006 by ANOVA; P < 0.006 by NK test) (Table 1). An interaction analysis showed that the spatial summation ratio was increased after surgery versus before surgery, although only on the operated side (F[1,1] = 7.21, P < 0.01 by ANOVA; P < 0.001 by NK test).

Descending CPM.

In healthy control subjects, PPTs at the knee increased significantly from baseline values during ischemic exercise pain induced in the arm (F[1,1] = 72.7, P < 0.0001 by ANOVA; P < 0.0001 by NK test) (Figure 5A). In contrast, knee PPTs were further reduced from baseline by the ischemic exercise pain in the OA patients (P < 0.03 by NK test). After knee replacement, the CPM resulted in significantly increased PPTs (F[1,1] = 74.1, P < 0.0001 by ANOVA; P < 0.0002 by NK test).

Figure 5.

Pressure–pain thresholds (PPTs) (A) and cuff PPTs (B) in healthy control subjects (n = 20), patients with osteoarthritis (OA) of the knee (n = 44), and in a subgroup of OA patients who underwent total knee replacement of the most affected knee (n = 20). PPTs were assessed at the knee (mean of sites 3 and 4) before and during a conditioning stimulus (ischemic exercise pain in the arm). Cuff pain thresholds were assessed at the lower leg. Control subjects and OA patients postsurgery had significantly higher PPTs than did the entire group of OA patients and the OA patients presurgery, respectively (# = P < 0.0002; ## = P < 0.057 by Newman-Keuls test). ∗ = P < 0.03 versus baseline, by Newman-Keuls test. Values are the mean ± SEM.

In healthy control subjects, but not OA patients, cuff pain thresholds at the lower leg were significantly increased during heterotopic arm pain as compared with baseline values (F[1,1] = 3.97, P = 0.05 by ANOVA; P < 0.0007 by NK test) (Figure 5B). Postsurgery, the OA patients had a tendency toward increased cuff pain thresholds as compared with baseline levels (F[1,1] = 4.17, P = 0.055 by ANOVA; P < 0.006 by NK test).


Patients with symptomatic knee OA show widespread mechanical hyperesthesia of symptomatic joints (localized) as well as over noncontiguous sites in the leg and forearm (generalized). The observation of enhanced spatial summation in asymptomatic areas as well as the loss of CPM provides evidence of altered central pain processing. Normalization of spatial summation ratios and the restoration of CPM after successful joint replacement surgery imply that this altered processing arises in response to, and is maintained by, peripheral pathology.

Hyperesthesia and spontaneous pain in knee OA are most likely related to increased sensitivity of nociceptors located in the deep tissue (peripheral sensitization) and/or increased responses in the dorsal horn or supraspinal neurons (central sensitization) (8, 12, 20, 28–30). Mechanical hyperesthesia extending beyond the symptomatic joint has previously been reported in patients with OA affecting either the hip (23), knee (10), or carpometacarpal joints (31). Other studies have found that knee OA patients experience stronger pain and larger areas of referred pain to experimental muscle pain stimulation outside the affected joint (11). A similar example is patients with a long-lasting unilateral epicondylalgia, in whom PPTs assessed in the legs are significantly lower than those in healthy controls (32).

In the present study, PPTs over both the symptomatic knees and the asymptomatic knees as well as at 2 control sites over the shin and forearm were decreased in OA patients as compared with healthy controls. Spreading of the sensitization to the contralateral knee was observed and likely occurs because some knee OA patients have bilateral symptoms and because there may be contralateral subclinical changes in the patients presenting with only one symptomatic knee. It is also possible that in chronic diseases such as knee OA, central neuronal systems are sensitized bilaterally. The implication of such spreading sensitization is obvious. Together with spreading sensitization, the clinical pain manifestations start to spread; for example, a patient with a local muscle, tendon, or joint pain problem begins to experience pain in other regions. In some patients, a local pain problem develops into widespread pain, and comorbid pain conditions may develop.

Assessment of mechanical pain sensitivity using hand-held pressure algometers is a well-established method and has good reliability, but concerns have arisen regarding interrater bias, difficulties maintaining a constant pressure compression rate, and the considerable physical effort needed for multiple assessments (33). Assessment of mechanical pain sensitivity with the use of an automated cuff pressure algometry technique excludes the manual involvement of a researcher and, hence, minimizes interexaminer variability and allows for control of both the compression rate and the configuration (27). Hyperesthesia to cuff stimulation on the lower leg has previously been demonstrated in fibromyalgia patients (34). The present study represents the first use of this technique in patients with arthritis and demonstrates the feasibility of using cuff algometry to measure hyperesthesia in OA patients. Our findings were similar to those observed using a conventional pressure algometer, with cuff pain thresholds being significantly lower in OA patients than in healthy controls.

Consistent with the results of a previous study assessing patients with knee OA (10), as well as a study of patients with hand OA (31), the present study found a modest association between PPTs away from the symptomatic joint and the reported joint pain, confirming a relationship between pain intensity and the degree of generalized mechanical hyperesthesia in OA patients. As the OA knee becomes sensitized and stimulates the central nervous system with excessive neuronal input, the adjacent neurons and additional spinal segments slowly become further sensitized (20). Peripheral sensitization of neuronal receptors is induced within the first hours after the onset of inflammation where the wide dynamic-range neurons and the nociceptive-specific neurons show increased responses to noxious and innocuous stimulation of the joint and lower the mechanical threshold to stimuli (35). It has been suggested that pain during knee OA may result from mild inflammation in such structures as the synovial layers (35). Assuming that the clinical nociceptive origin and pressure stimulation excite different populations of nociceptors, the hyperesthesia may also be due to a central summation or facilitatory mechanism. Spatial summation of neural activity and facilitation of previously ineffective dorsal horn synaptic connections might be of importance (36, 37).

Spatial summation refers to an increase in pain perception when larger areas of stimulation are used. To date, most research has focused on spatial summation of thermal stimuli (17, 38), whereas spatial summation of pressure–pain using cuff algometry has received less attention. Previous studies assessing spatial summation of pressure and heat pain in fibromyalgia patients did not find increased spatial summation (18, 19). In contrast, the present study demonstrated the presence of enhanced spatial summation in OA patients as compared with healthy controls, independently of assessment side, and the findings provide further evidence for altered central nociceptive processing in patients with demonstrable articular pathology.

Plastic changes or central sensitization within the neural organization of the spinal cord occur during development of joint inflammation (29, 35), which leads to facilitation of the spinal neuronal processing of the afferent inputs from nociceptive afferent fibers. This leads to increased sensitivity or reduced thresholds to non-noxious pressure on the inflamed joint and with some delay to the adjacent and noninflamed tissue, the latter indicating that spinal neurons expand their receptive fields (29, 35). In consequence, the cuff stimulation of a larger area will become more effective and will result in facilitated spatial summation. An alternative explanation relates to an interaction between spatial summation and descending inhibitory control. It has been shown that the descending inhibitory control is progressively recruited with larger stimulation areas (39), and with reduced descending inhibitory control, as found in the present study (see below), the outcome may be more effective spatial summation for the larger stimulation area. However, with the predominant facilitated spatial summation detected in the most affected leg and less in the contralateral leg, this mechanism is less plausible due to the generalized effects of CPM.

The perception of pain in response to a given painful stimulus can be reduced by application of another painful stimulus to an extrasegmental body location. This phenomenon, which is known as counterirritation analgesia, is the product of the endogenous pain modulation mechanism known as diffuse noxious inhibitory control or, as recently redefined, conditioned pain modulation (22). The loss of such modulation has been described in patients with different articular disorders, including OA (10, 23); however, the present study is the first to demonstrate the restoration of the phenomenon in patients with symptomatic knee OA following successful joint replacement surgery, which suggests that the chronic pain maintained the CPM dysfunction and that ongoing pain from one site may interact with CPM evoked by pain in another area. In patients with hip OA, restoration of the CPM response has been reported following joint replacement surgery (23). The restoration of CPM was significantly demonstrated by pressure algometry, and similar findings, although not statistically significant, were demonstrated by cuff algometry. The mechanism of descending control seems intact in rheumatoid arthritis patients with short-term or long-term disease as compared with controls (40).

Conditioned pain modulation results from the activation of brainstem inhibitory projections that, in turn, act to postsynaptically inhibit spinal and trigeminal wide dynamic-range neurons (41). It is believed to reflect the overall effect of complex facilitatory and inhibitory mechanisms of pain processing. This balance has also been suggested to be important for the spreading of pain and hyperesthesia. Less-efficient CPM mechanisms have been reported in patients with musculoskeletal pain conditions, such as myofascial temporomandibular disorders (42), chronic low back pain (43), fibromyalgia (44), and chronic tension-type headaches (45). Such a reduced potency of the descending control makes the entire neuraxis more vulnerable to pain (46).

Of the 20 OA patients who underwent knee replacement, 18 reported reduced VAS scores for pain within 6 months of their operation, with all patients eventually reporting a reduction in the pain score as compared with their preoperative score. Postoperatively, patients had reduced hyperesthesia, normalized spatial summation of pain, and an active descending pain modulation, indicating that the peripheral nociceptive drive is highly important for maintenance of facilitated central pain mechanisms in this musculoskeletal pain condition. Interestingly, facilitated spatial summation is expressed more in the most affected side and reverted more efficiently in the affected leg after knee replacement as compared with the contralateral leg. Depending on the mechanisms of central sensitization, it has been hypothesized to strongly affect the neuroanatomic structures related to nociceptive locus (i.e., the knee) and expand to other structures (e.g., contralaterally) over time (20). Such an expansion theory may explain the current data with predominant effects in the most affected leg.

Although the postsurgical assessment time was relatively different between patients, no significant correlations between the postsurgical time and reductions in pain intensity or sensitivity were found. The majority of patients were assessed after 9–18 weeks postsurgery, although in a larger study cohort with followup assessments at more time points, such correlations may exist. Importantly, the hypersensitivity was significantly reduced postsurgically with the current difference between time points for followup assessments.

The experimental design of the present study did not include reassessment of the healthy controls or the OA patients who did not undergo surgery. Test–retest quantitative sensory testing in patients with knee OA has been reported, with adequate-to-excellent intraclass correlation coefficients when assessed within less than a 10-day interval (47–49), with pressure algometry being the most reliable (49). Studies with longer intervals between assessments are lacking, but it is likely that disease progression results in more pronounced hypersensitivity, in contrast to the current findings of reduced hypersensitivity postsurgery. Importantly, several factors other than central sensitization, such as medications, functional changes related to improvements in levels of physical activity, anxiety, and mood may play a role in the reduced hypersensitivity.

The current data clearly substantiate the need for adequate pain treatment in patients with localized pain conditions, avoiding spreading pain and hyperesthesia, which may become a significant disabling problem in OA patients.

The differences in pain mechanisms between the OA and asymptomatic sex- and age-matched healthy subjects observed in this study provide evidence for facilitated nociceptive processing in this disorder. The widespread reduction in mechanical pain thresholds, the presence of enhanced spatial summation, and the loss of pain modulation by a conditioning stimulus suggest sensitized central mechanisms. Normalization of the responses, including a reduction in spatial summation and restoration of the response to a conditioning stimulus, implies that these processes are maintained by peripheral pathology.


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. Graven-Nielsen 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. Graven-Nielsen, Wodehouse, Langford, Arendt-Nielsen, Kidd.

Acquisition of data. Graven-Nielsen, Wodehouse, Langford, Kidd.

Analysis and interpretation of data. Graven-Nielsen, Wodehouse, Langford, Arendt-Nielsen, Kidd.


AstraZeneca had no role in the study design or in the collection, analysis, or interpretation of the data, the writing of the manuscript, or the decision to submit the manuscript for publication. Publication of this article was not contingent upon approval by AstraZeneca.