Abstract
Offset analgesia reflects time-dependent, central nervous system (CNS) pain inhibition and refers to a dramatic drop in pain intensity after an offset of noxious stimulus intensity. Neuropathic and nociplastic pain conditions with strong CNS pathophysiologic mechanisms show deficits in offset analgesia. Whether offset analgesia is altered in more peripherally driven chronic nociceptive pain was unknown. Therefore, the primary goal of the current study was to determine whether chronic nociceptive pain is associated with changes in offset analgesia. We measured offset analgesia and sensory function using quantitative sensory tests (QST), patient-reported pain and function, and walking and stair climbing performance using standardized tasks in knee osteoarthritis patients with equivalent joint degeneration but Moderate-to-Severe (n=36) or Mild pain intensity (n=36) and Pain-free controls without knee osteoarthritis (n=30) matching for age, gender, and body mass index. Offset analgesia was significantly reduced in knee osteoarthritis groups compared with the Pain-free controls, with deficits occurring at both the nonpainful forearm and painful knee and in both genders. Greater deficits in offset analgesia were associated with more impairment in walking and stair climbing. Onset hyperalgesia, a novel measure of time-dependent pain facilitation, was reduced in women with Mild knee pain but not in men. These results suggest that deficits in temporal pain inhibition and gender-specific changes in temporal pain facilitation may contribute to pain and functional impairment in knee osteoarthritis, supporting further study of central pain modulation as a clinically relevant mechanism of chronic nociceptive pain.
Keywords: temporal contrast enhancement, psychophysics, central sensitization, movement-evoked pain, weight-bearing pain
Summary:
Offset analgesia is reduced in patients with knee osteoarthritis, implicating a deficit in central pain inhibition as a clinically relevant mechanism underlying chronic nociceptive pain.
Introduction
Knee osteoarthritis is common and debilitating [48]. There are many peripherally targeted, efficacious treatments [32; 48; 51]. In fact, the prevailing view is that this chronic nociceptive pain syndrome is strongly driven by peripheral pathophysiology [32; 48]. However, peripherally targeted treatments are not always successful, with up to 20% of knee replacements associated with continued pain after surgery [16; 28]. Central nervous system (CNS) mechanisms of sensitization have been identified in animal models of knee osteoarthritis, including amplification of ascending inputs, deficits in descending inhibition, and amplified descending facilitation [14; 32; 39; 55; 98]. However, the interplay among knee joint inflammation, joint degeneration, peripheral nervous system (PNS) sensitization, and CNS pain modulation in patients is unclear, in part due to limitations in measuring CNS processes noninvasively. This gap limits progress in the mechanistic understanding and clinical care of osteoarthritis.
Recent work has identified quantitative sensory tests (QST) which show promise in identifying CNS mechanisms of pain sensitization in patients [6; 35; 80; 91; 107]. Offset analgesia is a QST paradigm thought to reflect temporal processing of pain [18; 43; 64; 76; 112]. Coghill’s group first described accelerated reduction of pain intensity immediately following the offset of a noxious cutaneous heat stimulus [37]. Using a Peltier-based thermode, they developed a complex, suprathreshold stimulus during which a small reduction in noxious heat intensity dramatically reduced perceived pain intensity. This reduction was so profound it was similar to stopping the noxious heat completely, despite ongoing noxious range heat delivered to the skin. Since this discovery, offset analgesia has been shown to be mediated by the CNS [54; 56] and is influenced by expectations of pain [101]. Although offset analgesia is reduced in neuropathic pain and fibromyalgia [50; 69; 73] and deficits are observed with greater bodily pain in an aging population[66], the role of offset analgesia in chronic nociceptive pain remains unclear.
Given that peripheral processes poorly predict knee pain in osteoarthritis [6; 30], we reasoned that a deficit in offset analgesia reflecting pain disinhibition would be associated with greater pain intensity perceived with a given degree of joint degeneration. Prior studies have demonstrated that sensory profiles may contribute to pain syndromes [31; 58] consistent with this idea, although most work has identified CNS pain amplification processes rather than a deficit in CNS pain inhibition. We used an observational study design to examine the relationship between offset analgesia and knee osteoarthritis pain. We recruited patients with moderate-to-severe joint degeneration into one of two knee pain groups, with mild knee pain or moderate-to-severe knee pain, as well as a pain-free control group without knee osteoarthritis, balancing the groups during recruitment for age, body mass index (BMI), and gender. Patients also participated in other QST assessments and mobility tasks assessing function and pain with weight-bearing. With this approach, we aimed to test our primary hypothesis that knee osteoarthritis pain is associated with deficits in pain inhibition, measured with offset analgesia.
Materials and methods
Study design
Participants and Regulatory Information
Seventy-three participants with knee osteoarthritis pain and thirty pain-free volunteers without knee osteoarthritis provided signed informed consent to participate in this observational, 3-group, case-control study, which was approved by the University of Pittsburgh Institutional Review Board (Study #20060254). This study was preregistered on ClinicalTrials.gov (NCT05003323). This manuscript is the first publication of data from this clinical study.
Participant recruitment and enrollment
Participants were recruited from the University of Pittsburgh Clinical and Translational Science Institute (CTSI) Pitt+Me research registry, which includes more than 200,000 people from the Pittsburgh area reporting an interest in research study participation. After participants reported an interest in the study to the Pittsburgh CTSI Pitt+Me research registry, study staff contacted the participants in a screening phone call to determine study eligibility. Participants were also recruited from clinical sites, including an orthopedic office and an anesthesiology perioperative care clinic. If the participants were eligible and met enrollment goals (see “Study Group Assignment” below), they were scheduled for the single study visit, which took approximately 4 hours to complete. Participants were compensated financially for their travel and time.
Participants were required to be 45–80 years old to enroll in the study. Patients with knee osteoarthritis were included if they reported a history of knee osteoarthritis and had posterior-anterior weight-bearing knee X-rays showing moderate-to-severe knee arthrosis, determined by B.A. to be KL-grade 2–4 [49]. Participants from any group were excluded from the study if they endorsed any of the following: (1) current use of opioids or antidepressants (SSRIs were allowed, but other antidepressant types were not), (2) clinically unstable systemic illness or medical disorder judged to interfere with the study, (3) pregnancy, (4) inability to walk or climb stairs unassisted, (5) unable to complete electronic questionnaires in English, (6) tattoos at sensory testing sites, (7) history of brain surgeries, (8) history of knee replacement or open knee surgery on index knee (most painful knee), (9) inflammatory arthritis, (10) history of arthroscopic surgery on index knee within the last 3 months, (11) history of radiofrequency nerve ablation in knee, (12) history of intra-articular steroid injections in index knee within the last 30 days, (13) history of platelet enriched or hyaluronic acid injections in index knee within the last 30 days, (14) cognitive impairment affecting the ability to provide informed consent, understand directions, and participate in study procedures. Participants with knee osteoarthritis were excluded if they had a chronic pain condition that was more severe than their knee pain (e.g., fibromyalgia, CRPS); their knee pain must have been their primary pain diagnosis. If the participant has equal pain ratings in both knees, either knee was tested as long as all other inclusion criteria were met. Pain-free volunteers were excluded if they had a history of any chronic pain. Pain-free volunteers denied a history of osteoarthritis and denied knee pain at the time of screening and enrollment. Knee radiographs were not obtained for Pain-free volunteers. Current prescription opioid use was defined as either acute use of any prescription opioid in the last month or chronic opioid therapy defined as a 90-day supply or more of prescription opioids in the last year[25; 89]. This definition applied to weak (e.g. tramadol) and strong (e.g. hydrocodone) agonists. If participants had taken prescription opioids in the last year but did not meet criteria for current use, they remained eligible, and the specific opioid was recorded as a concurrent medication.
Study group assignment
Participants with knee osteoarthritis were assigned to either the Moderate-to-Severe knee pain (“Mod-Sev”) or Mild pain groups (“Mild”). During study recruitment and enrollment, groups were assigned during the screening phone call to allow the study team to match Mod-Sev, Mild, and Pain-free groups for gender, age decile, and body mass index. Patients were asked to rate their average knee pain on a verbal numeric rating scale of 0–10 (0 is no pain, 10 is worst pain imaginable). Participants who answered 1–3 were classified as Mild pain, while 4–10 ratings were classified as Mod-Sev pain. Final group assignment was based on our digital pain assessment questionnaires with the same prompt and NRS cutoffs. This final group assessment was then used to update our recruitment and enrollment goals for age, gender, and BMI.
Study visit timeline
Participants arrived at the lab, which was located within an outpatient clinical office. All participants were asked to stop NSAIDS and marijuana products (including CBD) for 7 days before their study visit and not to apply topical medications, such as diclofenac gel, on the knee for 24 hours before their study visit. During that time, they were allowed to take acetaminophen as needed to manage pain. Participants were also asked not to drink alcohol the night before or exercise the morning of the visit. They were instructed to take other medications as prescribed. The above protocol and inclusion/exclusion criteria meant that participants were not using opioids, NSAIDs, or marijuana products on the day of their study visit with a washout period depending on the drug class. These medications were recorded as concurrent medications if patients were taking them prior to participation in the study. These medications were specifically addressed since they are commonly used for osteoarthritis. Gabapentinoids and muscle relaxants were allowed, as these are typically used for other conditions and not osteoarthritis. Due to inclusion/exclusion criteria specifying knee osteoarthritis as the main pain problem, this was infrequently encountered.
After review and provision of informed consent, the participants answered a set of standardized questionnaires and completed a brief pain history and exam. The index knee was defined as the most painful knee for knee osteoarthritis patients or alternated between left and right knees for the Pain-free control group. Next, the research coordinator began the QST battery with the participant. First, the participant underwent heat pain threshold testing and then a heat pain calibration paradigm with a cutaneous heat thermode at the non-dominant forearm and index knee. A suprathreshold heat stimulus paradigm designed to measure offset analgesia and onset hyperalgesia was administered using the same setup and equipment as the heat pain threshold and calibration paradigm first on the forearm and then on the knee. Pinprick sensitivity was assessed between the forearm and knee offset analgesia / onset hyperalgesia paradigms. Other tests in the QST battery, including temporal summation, pressure pain threshold, and conditioned pain modulation (CPM), were then completed, each at both painful and nonpainful body sites. The QST battery was conducted in the following order: heat pain threshold at forearm, heat pain threshold at knee, heat stimulus calibration at forearm, heat stimulus calibration at knee, offset analgesia / onset hyperalgesia paradigm on forearm, pinprick test on elbow and knee, offset analgesia / onset hyperalgesia paradigm on knee, temporal summation on elbow and knee, pressure pain threshold on trapezius, thumb, and knee, CPM on trapezius and knee. We did not vary the order of body site testing or QST assessment, which may affect results due to order effects. However, similar to other QST batteries[46; 80; 84; 107], more intense tests, such as CPM, were done at the end to avoid test order effects. At the end of the sensory testing, participants completed a questionnaire, a 24-step stair climbing task, and a 6-minute walking task. After the mobility tasks, participants completed post-testing questionnaires. Participants wore a functional near-infrared spectroscopy cap during QST and mobility procedures, as part of a separate study. The study visit took approximately 4 hours to complete.
Questionnaires and pain history
Participants completed a survey including self-reported demographics, medical information, and standardized psychometric questionnaires, summarized and administered as previously done [2]. Briefly, the psychometric surveys included instruments measuring knee symptoms and function (KOOS, PROMIS-29), pain (PEG, PROMIS-29), social status (BSMSS), depression (BDI-II, PHQ-2, PROMIS-29), anxiety (STAI Y-1 and Y-2, GAD-2, PROMIS-29), sleep disturbance (PROMIS-29), fatigue (PROMIS-29), social participation (PROMIS-29), global mental and physical health (PROMIS Global Health 2a), positive outlook (HEAL Positive Outlook), impulsivity (BIS-11), pain catastrophizing (PCS), and fibromyalgia (Fibromyalgia Severity Scale). The KOOS generates five subscales with ordinal scores ranging from 0 (worst) to 100 (best) to assess knee pain, other symptoms, function in daily living, function in sport and recreation, and knee-related quality of life [81]. The PROMIS-29, PROMIS Global Health, and NIH HEAL Positive Outlook generate ordinal T-scores reflecting pain intensity and several mental and physical health domains [36; 40–42]. The PEG 3-item survey assesses average pain intensity (P), interference with enjoyment of life (E), and interference with general activity (G) [52]. The BSMSS generates an ordinal score reflecting the respondent’s education and occupation as well as the education and occupation of their parents and spouse [15].
The BDI-II generates a combined ordinal score reflecting depression with excellent psychometric properties across different populations.[106] The PHQ-2 measures depressed mood over the last 2 weeks.[34] The STAI measures both state (Y-1) and trait (Y-2) anxiety, producing an ordinal score reflecting apprehension, tension, nervousness, and arousal.[33; 94] The GAD-2 briefly identifies generalized anxiety disorder.[77] The PCS measures catastrophic thinking associated with pain incorporating magnification of pain-related symptoms, rumination about pain, feelings of helplessness, and pessimism about pain-related outcomes.[96] The BIS-11 measures attentional, motor, and non-planning impulsiveness,[74] which are associated with reward processing relevant to pain and addiction.[59; 60] The Fibromyalgia Severity Scale is widely used in fibromyalgia diagnosis.[109] A 74-region body map used previously by the research team was filled out by all patients to assess pain distribution [1; 85]. After sensory testing but before the mobility tasks, the situational pain catastrophizing scale (SPCS), measuring catastrophizing related to the QST pain experience,[17] was administered. After the mobility tasks were completed, the STAI Y-2 and SPCS were administered again, relative to the mobility tasks.
A semi-structured pain history and exam was performed after surveys were completed and before the quantitative sensory testing. Concurrent medications, past medical and past surgical history were initially extracted from the electronic medical record and then confirmed during this interview. During data analysis, ICD-10 diagnosis codes were assigned by reviewing past medical history text notes. Generalized or multi-joint arthritis was defined as osteoarthritis in three or more sites including the distal interphalangeal joints, the proximal interphalangeal joint, hip, knee, spine, ankle or foot [38; 67]. If a patient had bilateral knee osteoarthritis, this would count as 2 of these joints in this assessment. The pain assessment also confirmed that the patients did in fact meet American College of Rheumatology criteria for knee osteoarthritis [4].
Heat stimulus pain threshold and calibration
Subjects were comfortably seated in an office chair in front of a small table for all sensory testing paradigms. A Q-sense 3 × 3 cm thermode (Medoc; Ramat Yishai, Israel) was applied to the non-dominant forearm and index knee and held with constant pressure by the research coordinator. For all heat testing, all forearm tests were completed before starting the knee tests. First, heat pain threshold was measured as previously done[2] in triplicate at three different locations on the forearm and lateral aspect of the index knee with participants verbally reporting the transition from heat to pain during a ramped heat stimulus (1.5 C°/sec). Specific instructions were for the participant to “say ‘now’ when the sensation changes from heat to pain.” The research coordinator hit a button when the participant indicated pain, so the temperature reached at the button press was recorded as the heat pain threshold. The heat stimulus maximum was set to 50°C with an interstimulus interval (ISI) of 30 seconds. Heat pain thresholds from the three skin locations were averaged, yielding mean thresholds for the forearm and index knee.
Next, based on the heat pain threshold, a previously described calibration protocol was used to individualize suprathreshold heat pain stimuli during the offset analgesia paradigm[2]. Briefly, participants were asked to rate pain intensity during an ascending series of 30-second noxious heat stimuli applied to the forearm and index knee with the Medoc thermode (Figure 1A). Participants rated their pain during the constant noxious heat stimuli in real time by operating a Computerized Visual Analogue Scale (COVAS, Medoc) with their dominant hand. The COVAS device features a horizontally positioned slider connected to a potentiometer inside the device and a 100-mm visual analogue scale (VAS) positioned externally along the slider. The anchors to the VAS were “no pain sensation” on the left (0/100 mm) and “most intense pain sensation imaginable” on the right (100/100 mm). Pain intensity was recorded using Medoc software. Real-time plots of the pain intensity curves were monitored by the experimenter during each 30-second stimulus, which involved a ramp to target temperature (1.5 C°/sec), a 30-second hold at target temperature, and then a ramp back to baseline temperature (32°C) at a rate of 6 C°/sec. Pain intensity and thermode temperature were not visible to the participant. The initial target temperature for the series of heat stimuli was the mean heat pain threshold rounded to the nearest whole number or, if the heat pain threshold was greater than 45°C, 45°C. Then, the target temperature was increased by 1 C° for each stimulus (ISI 80 seconds) until the participant reported a pain intensity of approximately 50/100 mm on the COVAS (within a window of 40–60/100 mm) at the end of the stimulus. This temperature was used as the individualized noxious heat temperature for that participant, termed “T1”. The maximum temperature cutoff for the target temperature was 47°C to prevent tissue damage. Each stimulus in the set of ascending heat stimuli was applied to one of three locations on the forearm or knee, rotating amongst them, so that the time between stimuli for a given skin site was ~4 minutes. A T1 for the forearm was determined before starting heat calibration on the knee.
Figure 1: Experimental setup.
A. The thermode was positioned on the lateral aspect of the index (most painful) knee or the volar surface of the non-dominant forearm. With the dominant hand, participants rated their pain intensity continuously using a slider on a computerized visual analogue scale (green box). B. A diagram reflecting the thermode temperature over time for a portion of the offset analgesia / onset hyperalgesia paradigm including a single trial of the offset, offset control, onset, and onset control stimuli. The dotted box represents the temperature plateau examined for offset analgesia and onset hyperalgesia. C. Example participant data from a single replicate of each stimulus during forearm testing. Calculating the pain change index point estimate for the offset and offset control (offcon) trial pair requires pain intensity values at points (a) and (b) which are then normalized to point (a) from the offset trial. Similarly, for the onset and onset control (oncon) trial pair, pain intensity values at (c) and (d) are used.
Offset analgesia and onset hyperalgesia paradigm
To measure offset analgesia, an “offset stimulus” as delivered to the skin using the Medoc thermode with participants reporting pain intensity in real time using the COVAS described above (Figure 1B&C). This technique was previously shown to elicit offset analgesia[2; 100]. The initial target temperature of the stimulus (“T1”) was tailored during the calibration phase outlined above to elicit 50/100 pain. As a control for adaptation, a constant noxious heat stimulus at T1 was also applied, termed “offset control”. To measure onset hyperalgesia, an “onset stimulus” and its constant control stimulus at the target temperature of T2 (“onset control”; T2 = 1 C° + T1) were applied, also with real time continuous pain intensity rating. Thus during the paradigm, a mixture of four suprathreshold heat stimuli were delivered through the Medoc thermode: offset, offset control, onset, and onset control (Figure 1B). In addition, a warm stimulus was delivered, termed “innocuous” (Inn, Inn = T1 – 5 C°). All five stimuli were repeated in triplicate. Stimulus order was pseudorandomized using three sequence patterns such that each stimulus type was completed by every 5th stimulus and then repeated by every 10th stimulus. Each stimulus type was applied to all testing sites at some point during the pattern. Stimulus patterns were counterbalanced in the knee pain groups, and different patterns were used on a given participant’s forearm and knee. Thus, fifteen heat stimuli were applied with an ISI of 80 seconds, rotating between three skin sites as done with the calibration procedure. Participants rated their pain during each stimulus with the COVAS as described above. The offset analgesia / onset hyperalgesia paradigm (15 stimuli) was first performed on the nondominant forearm and then on the index knee, with a break in between for pinprick testing. Detailed stimulus parameters are outlined as follows.
The offset stimulus started from baseline (32°C) with a rise rate 1.5 C°/sec to target temperature (T1) held for 5 seconds, then an increase in thermode temperature to T2, which was held for 5 seconds, then a decrease in temperature back to the T1 temperature which was held for 20 seconds, and finally a ramp (6 C°/sec) back to baseline temperature. Pain intensity during the time period around the decrease from T2 to T1 is examined for offset analgesia as previously done and outlined below.[2; 100]
The onset stimulus started from baseline (32°C) with a rise rate 1.5 C°/sec to target temperature (T2) held for 5 seconds, then a decrease in thermode temperature to T1 which was held for 5 seconds, then an increase in temperature back to T2 which was held for 20 seconds, and finally a ramp (6 C°/sec) back to baseline temperature. Pain intensity during the time period after the increase from T1 to T2 is examined for onset hyperalgesia as previously done and outlined below.[2]
The constant control stimuli started from baseline (32°C) with a rise rate 1.5 C°/sec to target temperatures (T1 or T2 for offset control and onset control, respectively), continued with a hold at T1 or T2 for 30 seconds, and then returned to baseline at a rate of 6 C°/sec.
The innocuous stimulus started from baseline (32°C) with a rise rate 1.5 C°/sec to target temperatures (T1–5°C), continued with a hold at that temperature for 30 seconds, and then returns to baseline at a rate of 6 C°/sec.
Pinprick and temporal summation
A Neuropen with a Neurotip was used for punctate stimulus and temporal summation testing [8] using protocols similar to prior studies [26; 61]. The Neuropen exerts 40 g of pressure when the tip is applied to the testing site until the reference marks are aligned on the pen [8]. To begin the pinprick test, a practice trial involving a single pinprick was applied to the non-dominant elbow and index knee. Participants answered “yes” or “no” to “Did that feel sharp?” and “Did that feel painful?”. Next, 2 trials of 3 consecutive pinpricks were applied to the same spot on the elbow and knee at 1 Hz, alternating between sites for each trial. Testing spots on a given site were approximately 1 inch apart for each trial. Participants were instructed to close their eyes when they were ready to begin each trial. At the end of each trial, participants were asked, “On a scale of 0 to 10, with 0 being no sharpness and 10 being the most intense sharpness imaginable, what was the maximum intensity of the sharpness you felt?” If participants answered “yes” to “Was the pin sensation painful?”, they were asked, “On a scale of 0 to 10, with 0 being no pain and 10 being the most intense pain imaginable, how painful was the pin sensation?”
For temporal summation, participants were instructed to give a numeric pain rating from 0–10 (0 being no pain, 10 being the most intense pain imaginable) as quickly as possible at the beginning, middle, and end of the trial when the coordinator said, “pain rating”. Two trials of 10 consecutive pinpricks were applied to the same spot on the elbow and knee at 1 Hz, alternating between sites for each trial.
Pressure pain threshold
An algometer (Wagner “Pain Test FPX50” device with 1 cm2 rubber tip; Greenwich, Connecticut) was used to measure pressure pain thresholds on bilateral trapezius, thumbs, and the index knee, using previously described protocols [26; 61]. The rate of pressure was manually increased at a rate of 1 pound of force per second, and efforts were made to ensure the pressure was a steadily increasing ramp. Participants were instructed to close their eyes when they were ready to begin and to say “now” when they first felt pain from the applied pressure. The trial order is as follows: trapezius, thumb contralateral to trapezius, index knee, other trapezius, other thumb. The starting trapezius was alternated for each participant. Two trials were completed on each site. After each trial, participants were asked, “On a scale of 0 to 10, with 0 being no pain and 10 being the most intense pain imaginable, what would you rate the pressure when it first felt painful?”
Conditioned pain modulation
To measure conditioned pain modulation (CPM), we used a concurrent application of pressure (“test” stimulus) and noxious cold (distracting stimulus) as previously described [26; 61]. First, pressure pain thresholds were determined on the index knee and contralateral trapezius. Then, participants immersed their hand ipsilateral to the index knee in a temperature-controlled water bath (Fisher Scientific; Hampton, New Hampshire) maintained at 10°C. Participants had their hand in the cold water for 5 seconds before pressure was applied to determine pressure pain threshold again. Participants were instructed to say “now” when they first felt pain from the pressure. After each trial, participants were asked, “On a scale of 0 to 10, with 0 being no pain and 10 being the most intense pain imaginable, what would you rate the pain from the cold water?” Two trials were completed on each site. Each test began with the knee followed by the trapezius. There was a 2-minute break between each trial.
Mobility tasks
To assess physical function and knee pain intensity, participants completed a 24-step stair evoked pain assessment [103] followed by a 6-minute walking task [108]. Before and after each task, participants rated their pain in both knees from 0–10 (0 is no pain, 10 is worst pain imaginable) while seated. After each task, participants also reported the maximum pain intensity felt in both knees while walking or stepping. For the stair evoked pain assessment, participants were asked to step fully up and down a single, aerobics step a total of 24 times, alternating the lead leg for each step. The total time spent stepping, including any pauses, was recorded with a stopwatch, and steps were tracked by the coordinator with a tally counter. For the walking task, participants were instructed to walk lengths between two cones approximately 83.5 feet apart along a straight hallway for 6 minutes. Participants were notified of the time and encouraged each minute. Time was monitored with a stopwatch. The number of completed lengths was measured with a tally counter, and distance from incomplete final lengths was calculated with a measuring wheel. Finally, the distance walked was calculated by multiplying lengths by 83.5 feet and adding the distance of the final, incomplete length.
Sample size justification and data analysis
The sample size for this study was planned based on prior studies of offset analgesia[2; 3; 100]. Using a two-sample t-test to achieve power of 0.9 with Bonferroni corrected α=0.05 / 3 = 0.0125 (for three comparisons across groups), 30 participants were required for each of three groups, using estimated effect sizes and standard deviations based on a meta-analysis analyzing offset analgesia in patients with chronic pain and Pain-free controls[100]. To provide additional power for multivariable regression across three groups, we recruited additional participants in each knee pain group. Matlab (R2023b, The MathWorks Inc, Natick, MA, USA), STATA (v17.0, StataCorp, College Station, TX, USA), and R (v4.1, Vienna, Austria, Europe) were used for the calculations and analysis described below.
Pain intensity timeseries curves during the offset, offset control, onset, onset control, and innocuous stimuli were analyzed in similar fashion to previously reported.[2; 100] In Matlab, raw timeseries data were aligned to time markers at the key temperature transition, around which offset analgesia and onset hyperalgesia are measured. For offset analgesia, this key transition is from T2 down to T1 temperature occurring at 10 seconds after reaching the target temperature plateau. For onset hyperalgesia, this key transition is from T1 up to T2 temperature occurring at 10 seconds after reaching temperature plateau. Timeseries data was then downsampled to 1 Hz. For each stimulus type, a mean pain intensity at each timepoint was calculated across the triplicates within each participant.
To test the primary hypothesis that offset analgesia differs in knee osteoarthritis pain, linear mixed effects modeling was performed in Stata. Prior to model construction, variables were graphically assessed for normality with histograms and QQ plots. For timeseries pain intensity data, we assessed for differences in offset analgesia, defined by the difference in pain intensity between offset and offset control stimuli, among the three knee pain groups: Pain-free controls, Mild pain, and Moderate-to-Severe pain. First, we constructed a nested linear mixed effects model with participant and body site set to random effects and fixed effects of a three-way interaction among stimulus (offset vs offset control), knee pain group (Pain-free, Mild pain, Moderate-to-Severe pain) and time. Then we calculated main effects of the interaction with prespecified post-hoc contrasts: [Mild – Pain-Free], [Mod-Sev – Pain-Free], and [Mod-Sev – Mild]. Model metrics are reported in the Results section. As part of model construction for timeseries data, residual covariance structure was determined to be 2nd or 3rd order autoregressive based on iterative fitting with different structures until we observed an improvement (reduction) in Akaike’s information criterion compared to simpler models. The same approach was used in secondary analysis to identify effects of knee pain group on onset hyperalgesia, except that data from the onset and onset control stimuli were used. To assess for group differences in pain intensity during the innocuous stimuli, there was no stimulus interaction term, since only the innocuous stimulus was examined. However, apart from this difference, the analysis approach was similar.
To calculate point estimates of offset analgesia and onset hyperalgesia, extreme values of pain intensity curves during each trial were obtained in Matlab and normalized differences were calculated as previously done by our group and others (Figure 1)[2; 50; 65; 66; 69; 73; 100]. For offset analgesia, the maximum pain intensity in a window of 10 seconds centered around the T2-to-T1 step down was identified within a single offset stimulus trial (Figure 1C point “a”). Then, the minimum pain intensity following maximum was identified (Figure 1C point “b”). A normalized pain change index for offset analgesia was calculated: (offset min pain – offset max pain)/(offset max pain) * 100. The normalized pain change index has been used in multiple populations to examine offset analgesia, justifying its use in this study[2; 50; 65; 66; 69; 73; 100]. Next, the timepoints of offset minimum and maximum from that offset stimulus trial were applied to the offset control trial closest in stimulus order, and offset control pain intensity values were extracted. The normalized pain change index was then calculated normalized to the maximum pain intensity during the offset stimulus: (offset control pain at offset min timepoint – offset control pain at offset max timepoint) / (offset max pain) * 100. Calculation of the pain change index for the offset control stimulus allows for comparison with the offset stimulus index to disentangle pain change related to within-stimulus adaptation from offset analgesia. The mean of offset and offset control pain change index values was calculated across the 3 trials, pairing the closest neighboring offset and offset control stimulus as noted above.
For onset hyperalgesia, the pain change index was calculated in a similar manner but mirroring the normalized pain change index for offset analgesia. Within a given onset stimulus trial, the minimum pain intensity in a window of 10 seconds centered around the T1-to-T2 step up was identified (Figure 1C point “c”). Then, the maximum pain intensity following that minimum was determined (Figure 1C point “d”). These maximum and minimum values were used to calculate an onset pain change index: (onset max pain – onset min pain) / (100 – onset min pain) * 100. Motivated by the observation that onset hyperalgesia has an inverted behavioral response to offset analgesia,[2] the denominator used in this calculation is an inversion of the pain intensity rating scale. Finally, the onset control stimulus nearest in stimulus order to that onset stimulus was used to calculate an onset control pain change index: (onset control pain at onset max timepoint – onset control pain at onset min timepoint) / (100 – onset min pain) * 100. The mean onset and onset control pain change index values were calculated across the three trials, again pairing closest neighboring onset and onset control stimulus replicates.
Point estimates of offset stimulus, offset control, onset stimulus, and onset control pain change indices were then used in linear mixed effects models similar in approach as above and with model metrics reported in the Results section. These point estimates were chosen over timeseries data to maintain statistical power and allow for more complex models used in secondary analysis.
For descriptive analysis of continuous variables, the mean and standard deviation or median and interquartile range were reported. For categorical variables, counts and percentages were calculated. To assess across-group differences, 1-way ANOVA or Kruskal-Wallis tests were used for normally and non-normally distributed continuous variables respectively. A Chi2 test was used for categorical variables. Linear relationships between continuous variables were probed graphically and by calculating Pearson’s r. As part of a sensitivity analysis, knee osteoarthritis patients were grouped using their responses on the KOOS-pain subscale. The groups were divided by median split into low and high pain groups. Among knee osteoarthritis patients, median KOOS-pain was 58.3 (n=72, IQR 47.2 – 70.7).
Results
Participants with moderate-severe knee arthrosis (Kellgren-Lawrence grades 2–4) measured by standing knee radiographs were recruited into Mild and Moderate-to-Severe (Mod-Sev) knee pain intensity groups with final group assignment determined by knee pain intensity on the day of testing (Mild: pain intensity numeric rating scale, NRS <=3; Mod-Sev: pain NRS >=4; Supplemental Fig 1). These two groups and a Pain-free control group were recruited in parallel, matching for age decile, gender, and body mass index. Participant characteristics are shown in Table 1.
Table 1:
Sociodemographic characteristics of knee pain groups
| Pain-free | Mild Pain | Mod-Sev Pain | P-value | |
|---|---|---|---|---|
| n | 30 | 36 | 37 | |
| Age (years) | 66.2 (6.3) | 68.9 (5.9) | 66.7 (6.9) | 0.177 |
| Body mass index (kg/m2) | 29.9 (5.6) | 28.5 (5.7) | 31.4 (6.2) | 0.112 |
| Gender (endorsed “male”) | 12 (40) | 15 (42) | 14 (38) | 0.947 |
| Married | 22 (73.3) | 26 (72.2) | 16 (43.2) | 0.012 |
| Lives with others | 25 (83.3) | 26 (72.2) | 22 (59.5) | 0.099 |
| Race | 0.545 | |||
| White | 27 (90.0) | 32 (88.9) | 29 (78.4) | |
| Black | 3 (10.0) | 3 (8.3) | 6 (16.2) | |
| Other | 0 (0.0) | 1 (2.8) | 2 (5.4) | |
| Education level | 0.215 | |||
| Masters degree or higher | 13 (43.3) | 12 (33.3) | 9 (24.3) | |
| Associates or Bachelor degree | 12 (40.0) | 18 (50.0) | 15 (40.5) | |
| High school or some college | 5 (16.7) | 6 (16.7) | 13 (35.1) | |
| Work Status | 0.597 | |||
| Retired/Unemployed | 16 (53.3) | 24 (66.7) | 23 (62.2) | |
| Full Time | 8 (26.7) | 5 (13.9) | 5 (13.5) | |
| Part Time | 6 (20.0) | 7 (19.4) | 9 (24.3) | |
| Barratt Simplified Measure of Social Status | 45.3 (10.7) | 43.9 (8.8) | 34.7 (15.8) | 0.001 |
Data are shown as Mean (SD) or Count (%). P-values calculated with Chi-square tests for categorical variables and 1-way ANOVAs for continuous variables.
Offset analgesia is diminished in knee osteoarthritis
To test for alterations in offset analgesia, we applied offset and constant control stimuli to a nonpainful area, the volar forearm, and performed our primary analysis using the collected pain intensity data. One participant in the Moderate-to-Severe group partially completed quantitative sensory testing due to participant time constraints, and so was not included in the primary analysis. In a linear mixed-effects model of pain intensity during forearm stimulation with a three-way interaction between knee pain group, stimulus type (offset vs. control) and time with fixed effects for knee pain group and time, a random slope and intercept for stimulus type, and a random intercept for participant, we found a significant main effect of the three way interaction (n = 6324, n(stimulus) = 204, n(participant) = 102; log likelihood = −15885.64, Wald chi2(185) = 2667.42, p < 0.0001; post-estimation main effect of the 3-way interaction: chi2(60) = 89.13, p=0.0087). Two sets, one for the offset stimulus and one for constant offset control stimulus, of three contrasts at each timepoint were calculated to examine knee pain group differences. The contrasts compared: (1) the Moderate-to-Severe knee pain group with Pain-free control group, (2) the Mild knee pain group with the Pain-free control group, and (3) the Moderate-to-Severe with the Mild knee pain groups. The Moderate-to-Severe group had higher pain intensity than the Pain-free group following stimulus offset at timepoints consistent with offset analgesia. There was no statistically significant difference between the Mild group and the Pain-free group. Notably, the Moderate-to-Severe group had greater pain intensity compared with the Mild pain group at the peak of the pain intensity curve, prior to offset analgesia. There were no statistically significant differences across groups in pain intensity curves during the constant control stimulus. In the entire group, there was a main effect of stimulus (chi2(1) = 4.79, p = 0.029), and the offset stimulus produced lower pain intensity than the offset control stimulus at timepoints following temperature offset, consistent with offset analgesia (not shown). Within subgroups, contrasts at each timepoint did not achieve statistical significance, likely due to heterogeneity in timepoints of minimum values and subgroup sample size. Altogether, these data suggest that that greater knee osteoarthritis pain is associated with less offset analgesia at a remote, nonpainful site and that the observed results are not due to adaptation to heat pain measured with the constant control stimulus.
Given that the Moderate-to-Severe group had a greater peak pain intensity than the Mild pain group, an alternative explanation for observed group differences in Figure 2 is that heat hyperalgesia, rather a deficit in offset analgesia, accounts for elevated pain intensities after stimulus offset in the Moderate-to-Severe group. To address this possibility, we calculated a pain change index, normalizing pain intensity values to the maximum pain intensity within each participant and stimulus. Timeseries pain change index data are plotted in Figure 3. Using a similar analytic approach as with pain intensity, a linear mixed-effects model of pain change index was calculated (n = 6324, n(stimulus) = 204, n(participant) = 102; log likelihood = −23790.40, Wald chi2(185) = 1100.43, p < 0.0001). With the normalization, there was no statistically significant main effect of the three-way interaction (knee pain group X stimulus X time, chi2(60) = 67.08, p=0.25). However, there were significant main effects of two-way interactions (knee pain group X time, chi2(60) = 89.97, p=0.0074; stimulus X time, chi2(30) = 198.34, p<0.0001; knee pain group X stimulus, chi2(2) = 1.74, p=42). Although the knee pain group X stimulus main effect is not significant, examination of the plots and post-estimation contrasts (significant timepoints annotated in Figure 3) suggest this is because both knee osteoarthritis groups have very similar pain change indices during both offset and control stimuli. Importantly, during the offset stimulus, the Moderate-to-Severe group and the Mild group both have more positive pain change index values (reduced analgesia) compared to the Pain-free control group during offset analgesia. Given that the calculated index controls for potential heat hyperalgesia effects, these data support the conclusion that offset analgesia at the forearm is reduced in patients with knee osteoarthritis.
Figure 2: Pain intensity is greater after noxious stimulus offset in patients with knee osteoarthritis at a remote, nonpainful body site.
Cutaneous thermode temperature (top) and continuous pain intensity ratings (bottom, linear predictions from a mixed effects model) are plotted during the offset stimulus (left) and the constant control stimulus (right) applied to the forearm. Plotted lines are point estimates following linear mixed effects modeling with shaded areas reflecting 95% CI. Horizontal bars represent statistically significant differences of post-estimation contrasts across groups. Participants in the Moderate-to-Severe pain group (Mod-Sev, n=36) reported significantly greater pain after the stimulus offset (vertical grey line at 10-seconds) compared with Pain-free controls (n=30). Immediately following stimulus offset but while pain intensities were still high, the Mod-Sev group reported greater pain than the Mild group (n=36). There was no difference between Mild and Pain-free controls. There were no differences across groups during the constant offset control stimulus.
Figure 3: Centrally mediated offset analgesia is diminished in knee osteoarthritis.
During offset heat stimuli applied to the forearm, which is a nonpainful site reflecting centrally mediated pain processing, pain intensity change is more positive (less offset analgesia) in knee osteoarthritis patients with Mild (n=36) or Moderate-to-Severe (Mod-Sev, n=36) pain intensity compared with Pain-free controls (n=30). Timeseries data on the top rows reflecting thermode temperature (top) and pain change index (middle) are annotated with horizontal bars representing statistically significant differences of post-estimation contrasts following linear mixed effects modelling of pain intensity change across groups. The pain change index following stimulus offset, calculated as 100* (min pain – max pain)/(max pain), is shown on the bottom row in violin plots with individual participant data as grey dots and box and whisker plots reflecting median, IQR (box), and 1.5*IQR (whiskers). The left column shows data from the offset stimulus. The right column shows data from the constant offset control stimulus, which was also normalized to the maximum pain intensity during the offset stimulus. Line plots are estimates following linear mixed effects modeling with shaded areas reflecting 95% CI.
Offset analgesia was also measured at the most painful knee in the osteoarthritis groups and either right or left knees in the Pain-free control group. Using the same analytic approach as above, we found that pain intensity was greater in the Moderate-to-Severe knee pain group during the offset stimulus but not during the constant offset control stimulus (Figure 4; linear mixed effects model: n = 6324, n(stimulus) = 204, n(participant) = 102; log likelihood = −16048.2, Wald chi2(185) = 2025.47, p < 0.0001; post-estimation main effect of 3-way knee pain group X stimulus X time interaction: chi2(60) = 81.99, p=0.0312; post-estimation contrasts at each timepoint annotated in Figure 4). Calculation of pain change index to normalize pain intensities to the maximal pain reported during the heat stimuli yielded similar results to the forearm analysis (Figure 4; linear mixed effects model: n = 6324, n(stimulus) = 204, n(participant) = 102; log likelihood = −20383.737, Wald chi2(185) = 2709.64, p < 0.0001; post-estimation main effect of 3-way knee pain group X stimulus X time interaction: chi2(60) = 88.88, p=0.0091; post-estimation contrasts at each timepoint annotated in Figure 4).
Figure 4: Offset analgesia is reduced at the painful knee in knee osteoarthritis.
Pain intensity timeseries curves during the offset stimulus (top: thermode temperature, middle: pain intensity) show that pain is more positive (less analgesia) in knee osteoarthritis patients with Moderate-to-Severe (n=36) pain compared with Pain-free controls (n=30). No differences were observed with the constant control stimulus (right). Pain change index timeseries (middle row) and point estimates (bottom row) show that pain change index is more positive for knee osteoarthritis patients with Mild (n=36) or Moderate-to-Severe pain intensity during the offset stimulus compared with Pain-free controls. No difference in pain change index was observed during the constant offset control stimulus. Line plots are estimates following linear mixed effects modeling with shaded areas reflecting 95% CI. Violin plots include individual participant data as grey dots and box and whisker plots reflecting median, IQR (box), and 1.5*IQR (whiskers).
Deficits in offset analgesia are independent of testing site and gender.
To compare directly the remote, nonpainful site with the painful knee, pain timeseries curves were used to extract a single pain change index value for each participant (see violin plots in Figures 3 and 4) as previously done in patients and in studies comparing body sites[50; 65; 66; 69; 73; 100]. Solving a linear mixed effects model with body site nested within participant and a three-way interaction between knee pain group, stimulus (offset vs. constant control), and body site resulted in a nonsignificant 3-way interaction (linear mixed effects model: n = 408, n(body site) = 204, n(participant) = 102; log likelihood = −1847, Wald chi2(11) = 933.84, p < 0.0001; post-estimation main effect of 3-way knee pain group X stimulus X body site interaction: chi2(2) = 1.06, p=0.5894) and a nonsignificant main effect of body site (chi2(1) = 2.55, p=0.1105). However, there was a significant interaction between knee pain group and stimulus (chi2(2) = 7.69, p = 0.0213), similar to the above timeseries analyses. Post estimation testing revealed a significant difference between offset and control stimuli, confirming the presence of offset analgesia in each group ([offset – constant control] contrasts: Pain-free = −64.2 SE 3.4 p<0.001, Mild pain = −52.5 SE 3.1 p<0.001, Moderate-to-Severe pain = −53.2 SE 3.1, p<0.001) and significantly less (more positive) offset analgesia comparing knee osteoarthritis groups with the Pain-free control group ([Moderate-to-Severe – Pain-free]: offset stimulus 14.7 SE 5.2 p=0.005; [Mild – Pain-free]: offset stimulus 14.4 SE 5.2 p = 0.006). There were no statistically significant differences between Moderate-to-Severe and Mild pain groups during the offset stimulus. Importantly, there were no across group differences in the constant control stimuli. Contrasts of the contrasts were significantly different such that knee pain groups had less offset analgesia (less negative) than the Pain-free control group, without a significant difference between knee pain groups ([Mild(offset-control) – Pain-free(offset-control)] 11.7 SE 4.6 p=0.012; ([Mod-Sev(offset-control) – Pain-free(offset-control)] 10.9 SE 4.6 p=0.018; ([Mod-Sev(offset-control) – Mild(offset-control)] −0.7 SE 4.4 p=0.873). Finally, although a 2-way interaction between stimulus and body site was borderline significant (chi2(1) = 3.89, p = 0.0485), contrasts within body site (e.g. [Moderate-to-Severe – Pain-free] offset stimulus at the knee) showed similar results at both the forearm and knee with significant differences in offset analgesia and no differences in the constant control stimulus. Overall, these results indicate that body site does not impact pain change index differences observed across knee pain groups and stimulus types, supporting the notion that offset analgesia is reduced at both painful and nonpainful areas in knee osteoarthritis.
Given that offset analgesia may be influenced by gender[63; 65; 69], we examined gender effects in our study. Entering gender into a linear mixed effects model (body site nested within participant as random effects and fixed effects of a two-way interaction between knee pain group and stimulus (offset vs. constant control), body site, and gender) resulted in a significant main effect of both the knee pain group X stimulus interaction, as observed in the above models, and gender such that male participants had significantly less offset analgesia than female participants (Supplemental Fig 2; linear mixed effects model: n = 408, n(body site) = 204, n(participant) = 102; log likelihood = −1842.7, Wald chi2(7) = 926.95, p < 0.0001; main effect of gender chi2(1) 13.44, p<0.0002, main effect of knee pain group X stimulus chi2(2) = 7.56, p<0.023). In a separate nested linear mixed effects model, a three-way interaction between knee pain group, stimulus, and gender was not statistically significant. These results indicate that offset analgesia deficits in knee pain groups occur in both males and females, despite gender differences in offset analgesia.
In a sensitivity analysis, participants with knee osteoarthritis were grouped by their responses to the KOOS-pain subscale, which measures pain during activities over the preceding week. Using a KOOS-pain median split to divide participants into low and high pain groups and the same linear mixed effects modelling approach as above, offset analgesia was significantly attenuated in the low and high pain groups compared to Pain-free controls (linear mixed effects model: n = 408, n(bodysite) = 204, n(participant) = 102; log likelihood = −1848, Wald chi2(6) = 921.7, p < 0.0001; post-estimation main effect of 2-way knee pain group X stimulus while controlling for bodysite: chi2(2) = 9.83, p=0.0073; [high pain – pain free]: offset stimulus 14.5 SE 5.2 p=0.005; [low pain – Pain-free]: offset stimulus 14.5 SE 5.2 p = 0.005). There were no differences between high and low groups in offset analgesia, and there were no differences in the constant control stimulus. These results confirm the primary analysis, specifically finding that offset analgesia is attenuated in participants with knee osteoarthritis compared with Pain-free controls.
Knee osteoarthritis independently predicts offset analgesia controlling for pain quality, pain history, sensory function, and psychological characteristics.
Univariate differences across pain groups were examined to determine whether other relevant participant characteristics contributed to group differences in offset analgesia. There was no difference in age, gender, or BMI across the groups (Table 1), showing successful matching of the groups on these factors as part of our study design. Marital status and social status were significantly different. In Table 2, pain quality, pain history, pain impact, objective functional measures during the 6-minute walk task and stair climb task, pain during those tasks, “fibromyalgia-ness,” and pain distribution were all different across groups. Pain duration was no different between the Mild and Moderate-to-Severe pain groups. In Table 3, quantitative sensory testing results showed statistically significant across-group differences in pinprick hyperalgesia and temporal summation of pinprick pain at the painful knee. A statistically significant but small difference in T1 temperature suggests that participants in the Mild and Moderate-to-Severe pain groups may have widespread heat hyperalgesia, although heat pain thresholds were no different at the knee or forearm. Knee pressure pain threshold trended towards pressure hyperalgesia in the knee pain groups, although this was not statistically significant. Pressure pain thresholds at other sites were no different. CPM was no different between groups. Table 4 shows across-group differences in psychological characteristics with statistically significant differences in pain catastrophizing, depression, anxiety, impulsivity, and a global mental health self-assessment. Self-reported sleep and social engagement were also statistically significantly different. The frequency of NSAID, acetaminophen, and topical analgesic medication use was significantly different across groups (Table 5). There were no group differences in other medications or medical comorbidities (Supplemental Tables 1 and 2)
Table 2:
Pain and functional assessments
| Pain-free | Mild Pain | Mod-Sev Pain | p-value | |
|---|---|---|---|---|
| n | 30 | 36 | 37 | |
| Pain intensity now (NRS1) | 0.0 (0.0) | 1.6 (1.0) | 5.8 (1.5) | <0.001 |
| Worst pain intensity of last 7 days (NRS1) | 0.4 (0.9) | 5.0 (2.2) | 7.7 (1.6) | <0.001 |
| Average pain rating of last 7 days (NRS1) | 0.1 (0.2) | 3.0 (1.6) | 5.8 (1.6) | <0.001 |
| Pain duration (years) | - | 9.8 (8.2) | 9.5 (9.8) | 0.88 |
| KOOS Pain | 99.4 (1.1) | 66.9 (11.4) | 51.5 (12.8) | <0.001 |
| KOOS Symptoms | 74.3 (14.4) | 50.1 (11.7) | 46.0 (15.2) | <0.001 |
| KOOS Sport and Recreation Function | 97.3 (8.9) | 43.7 (21.7) | 24.8 (19.1) | <0.001 |
| KOOS Activities of Daily Living | 99.3 (2.0) | 75.5 (16.1) | 59.4 (15.6) | <0.001 |
| KOOS Quality of Life | 98.8 (4.8) | 47.1 (13.2) | 36.0 (16.7) | <0.001 |
| Neuropathic pain descriptors (yes)2 | 1 (3.3) | 11 (30.6) | 15 (40.5) | 0.002 |
| PROMIS Global Physical Health 2a (T-score) | 56.0 (5.5) | 50.6 (7.6) | 45.5 (6.7) | <0.001 |
| PROMIS Physical Function (T-score) | 57.2 (3.8) | 45.4 (7.1) | 41.3 (6.4) | <0.001 |
| PROMIS Pain Intensity T-score (T-score) | 42.0 (1.9) | 52.5 (7.3) | 59.2 (6.2) | <0.001 |
| Pain, Enjoyment of Life, and General Activity (PEG) score | 0.1 (0.3) | 2.7 (1.8) | 4.9 (2.0) | <0.001 |
| 6-minute walk task distance (feet) | 1483.8 (207.5) | 1396.6 (315.1) | 1166.7 (213.2) | <0.001 |
| Stair climbing task duration (seconds) | 80.2 (18.3) | 89.5 (21.0) | 105.7 (23.8) | <0.001 |
| Maximum pain intensity during walking task (NRS1) | 0.0 (0.0) | 1.8 (1.6) | 4.9 (3.0) | <0.001 |
| Maximum reported pain during stair climbing task (NRS1) | 0.0 (0.0) | 2.6 (2.1) | 5.9 (2.6) | <0.001 |
| Number of painful areas selected on body map | 0.3 (0.8) | 5.4 (4.6) | 7.0 (6.2) | <0.001 |
| Fibromyalgia symptom severity score | 0.9 (1.1) | 1.7 (1.4) | 3.1 (2.3) | <0.001 |
| Fibromyalgia widespread pain index | 0.7 (1.0) | 2.2 (1.7) | 2.2 (3.0) | 0.005 |
| Bilateral knee pain (yes) | 0 (0.0) | 17 (47.2) | 20 (54.1) | <0.001 |
| Generalized, multi-joint osteoarthritis (yes) | 0 (0.0) | 13 (36.1) | 17 (45.9) | <0.001 |
Data are shown as Mean (SD) or Count (%). P-values calculated with Chi-square tests for categorical variables and 1-way ANOVAs for continuous variables.
NRS = numeric rating scale.
positive for “neuropathic pain descriptors” if patients endorsed any of the following qualities: shooting, stabbing, sharp, hot, burning.
Table 3:
Quantitative sensory testing at the index knee and nonpainful body sites.
| Pain-free | Mild Pain | Mod-Sev Pain | p-value | |
|---|---|---|---|---|
| n | 30 | 36 | 37 | |
| Pressure pain threshold - knee (lbf, mean (SD)) | 13.6 (5.6) | 12.3 (4.9) | 10.9 (4.3) | 0.092 |
| Pressure pain threshold - trapezius (lbf, mean (SD)) | 12.7 (6.6) | 10.63 (4.3) | 10.7 (6.2) | 0.277 |
| Pressure pain threshold - thumb (lbf, mean (SD)) | 12.6 (5.6) | 10.8 (4.0) | 10.6 (5.0) | 0.217 |
| Cold water pain during CPM (NRS) (mean (SD)) | 5.2 (2.7) | 4.9 (2.3) | 6.1 (2.8) | 0.142 |
| CPM calculated index- trapezius (mean (SD)) | 202.2 (333.3) | 223.1 (269.0) | 134.0 (211.2) | 0.358 |
| CPM calculated index - knee (mean (SD)) | 146.1 (321.3) | 141.1 (330.3) | 112.2 (272.7) | 0.886 |
| Heat pain threshold - forearm (C°, mean (SD)) | 46.4 (2.0) | 46.5 (2.5) | 46.0 (2.7) | 0.724 |
| Heat pain threshold - knee (C°, mean (SD)) | 47.3 (1.4) | 47.8 (1.5) | 47.7 (2.0) | 0.528 |
| Pinprick painful - elbow (yes (%)) | 2 (6.7) | 5 (13.9) | 5 (13.9) | 0.587 |
| Pinprick painful - knee (yes (%)) | 2 (6.7) | 13 (36.1) | 13 (36.1) | 0.01 |
| Temporal summation - elbow (yes (%)) | 9 (30.0) | 13 (36.1) | 16 (44.4) | 0.474 |
| Temporal summation - knee (yes (%)) | 12 (40.0) | 18 (50.0) | 21 (58.3) | 0.333 |
| Temporal summation continuous - elbow (median [IQR])1 | 0.5 [0.0, 1.0] | 0.5 [0.0, 1.5] | 0.5 [0.00, 1.5] | 0.581 |
| Temporal summation continuous - knee (median [IQR])1 | 0.5 [0.0, 1.0] | 0.8 [0.0, 2.0] | 1.0 [0.5, 2.0] | 0.045 |
| Offset analgesia - forearm (pain change index, mean (SD)) | −83.5 (21.5) | −70.4 (23.8) | −66.4 (29.2) | 0.021 |
| Offset analgesia - knee (pain change index, mean (SD)) | −83.4 (21.0) | −67.8 (30.5) | −71.1 (32.3) | 0.077 |
| Onset hyperalgesia - forearm (pain change index, mean (SD)) | 46.0 (34.2) | 40.1 (27.3) | 45.2 (34.4) | 0.705 |
| Onset hyperalgesia - knee (pain change index, mean (SD)) | 43.2 (32.8) | 37.5 (24.2) | 44.2 (30.1) | 0.581 |
| Max pain during warm stimulus - forearm (median [IQR])2 | 2.8 [0.0, 11.6] | 8.5 [1.9, 15.3] | 6.5 [0.0, 19.6] | 0.326 |
| Max pain during warm stimulus – knee (median [IQR])2 | 0.2 [0.0, 5.4] | 1.5 [0.0, 8.4] | 1.3 [0.00, 5.5] | 0.652 |
| T1 Temperature used on forearm in C (median [IQR]) | 46 [45, 47] | 45 [45, 46] | 45 [45, 46] | 0.041 |
Data are shown as Mean (SD), Median [IQR], or count (%). Chi-square tests are used for categorical variables, 1-way ANOVAs are used for continuous variables, and Kruskal-Wallis tests are used for non-normal continuous variables denoted with the Median [IQR]. P-values are highlighted in bold if less than 0.05.
Temporal summation calculated as the difference in pain intensity (numeric rating scale) between the 10th and 1st stimulus. Positive values reflect increasing pain intensity.
Maximum of pain intensity continuously rated on the computerized visual analogue scale.
Table 4:
Assessments of mental health, fatigue, and sleep
| Pain-free | Mild Pain | Mod-Sev Pain | p-value | |
|---|---|---|---|---|
| n | 30 | 36 | 37 | |
| Pain Catastrophizing Scale | 1.1 (2.5) | 6.3 (7.1) | 11.0 (11.0) | <0.001 |
| Situational Pain Catastrophizing Scale after QST | 2.2 (2.2) | 2.4 (2.3) | 2.9 (4.3) | 0.678 |
| Situational Pain Catastrophizing Scale after Mobility Tasks | 0.0 (0.0) | 1.5 (3.9) | 2.5 (3.5) | 0.007 |
| Barratt Impulsiveness Scale 11 | 47.7 (8.6) | 52.1 (7.5) | 54.6 (9.7) | 0.006 |
| Patient Health Questionnaire 2 | 0.0 (0.2) | 0.4 (0.9) | 0.4 (0.9) | 0.092 |
| Beck Depression Inventory | 3.0 (3.3) | 5.1 (4.8) | 7.0 (6.8) | 0.013 |
| PROMIS 4-item Depression T-score | 41.7 (2.8) | 44.6 (7.3) | 44.8 (6.4) | 0.079 |
| Generalized Anxiety Disorder | 0.2 (0.5) | 0.4 (0.7) | 0.5 (1.0) | 0.13 |
| State Trait Anxiety Inventory 1 | 24.4 (5.0) | 27.4 (8.0) | 29.2 (9.6) | 0.048 |
| State Trait Anxiety Inventory 2 | 24.7 (4.9) | 28.7 (7.7) | 31.1 (8.2) | 0.002 |
| State Trait Anxiety Inventory 1 after testing | 21.4 (2.8) | 25.8 (6.7) | 25.5 (7.3) | 0.008 |
| PROMIS 4-item Anxiety T-score | 44.2 (5.8) | 45.8 (6.9) | 47.3 (7.8) | 0.204 |
| PROMIS 4-item Fatigue T-score | 36.8 (5.7) | 45.1 (7.8) | 47.5 (9.3) | <0.001 |
| PROMIS 6-item Sleep Disturbance T-score | 43.3 (5.8) | 45.0 (8.0) | 51.6 (8.5) | <0.001 |
| PROMIS Ability to Participate in Social Roles and Activities T-score | 63.1 (3.0) | 53.7 (7.2) | 50.7 (8.0) | <0.001 |
| PROMIS Global Mental Health 2a T-score | 58.3 (5.8) | 55.7 (7.2) | 52.5 (8.0) | 0.006 |
| HEAL Positive Outlook T-score | 43.4 (3.1) | 41.5 (5.2) | 41.3 (3.9) | 0.102 |
Data are shown as Mean (SD). 1-way ANOVAs are used for continuous variables.
Table 5:
Concurrent medications for pain management
| Pain-free | Mild Pain | Mod-Sev Pain | p-value | |
|---|---|---|---|---|
| n | 30 | 36 | 37 | |
| NSAID1 | 3 (10.0) | 21 (60.0) | 22 (59.5) | <0.001 |
| Acetaminophen | 1 (3.3) | 10 (28.6) | 12 (32.4) | 0.01 |
| Weak Opioid2 | 0 (0.0) | 2 (5.7) | 1 (2.7) | 0.395 |
| Strong Opioid2 | 0 (0.0) | 0 (0.0) | 3 (8.1) | 0.066 |
| Topical Pain1 | 0 (0.0) | 7 (20.0) | 10 (27.0) | 0.01 |
| Gabapentinoids | 0 (0.0) | 2 (5.7) | 3 (8.1) | 0.299 |
| Muscle Relaxants | 0 (0.0) | 2 (5.7) | 0 (0.0) | 0.142 |
| Medical Marijuana1 | 0 (0.0) | 1 (2.9) | 3 (8.1) | 0.218 |
Data are shown as Count (%). Chi-square tests are used for categorical variables.
Participants paused the use of these medications prior to the study visit.
Current use of opioids in the month preceding the study visit was exclusionary. If prescription opioids were used prior to one month and use did not meet chronic opioid therapy criteria, participants were allowed in the study and, in this case, opioids were recorded as a concurrently used medication.
Variables reflecting univariate across-group differences were then entered into a multivariable linear mixed effects model to determine whether pain group independently predicted offset analgesia. To avoid multicollinearity, variables reflecting similar characteristics were examined for pairwise correlations. If correlations were r > 0.75, then a single variable representing those variables was entered into the model. In the final multivariable linear mixed effects model with pain change index as the outcome and controlling for covariants, a group X stimulus interaction reflecting offset analgesia remained statistically significant (linear mixed effects model: n = 372, n(bodysite) = 186, n(participant) = 93; log likelihood = −1666, Wald chi2(30) = 913.65, p < 0.0001; post-estimation main effect of 2-way knee pain group X stimulus controlling for bodysite: chi2(2) = 6.11, p=0.0471). Group X stimulus independently predicted pain change index (LR test comparing models with and without group X stimulus: chi2(4) = 405.22, p<0.0001). The covariates included KOOS-pain, KOOS-symptoms, neuropathic pain characteristics (binary), stair climb time, 6MWT distance, stair climb max pain, 6MWT max pain, knee pinprick hyperalgesia (binary), knee temporal summation to pinprick (binary), forearm T1 temperature used, pain castrophizing scale, situational pain catastrophizing scale (relative to walking and stair climbing), impulsivity, global mental health, fatigue, sleep, social engagement, social status, marital status (binary), pain distribution, fibro WPI, fibro SSS, NSAID use (binary), acetaminophen (binary), topical analgesic (binary). A similar result was obtained using the KOOS-pain subscale median as a grouping variable.
In the multivariable models, across-group postestimation contrasts showed significant differences, albeit somewhat different than the less complex models described above. With the pain intensity group multivariable model, there was a significant difference between the Mild and Pain-free groups ([Mild – Pain-free]: offset stimulus 28.2 SE 9.1 p=0.002). However, the [Mod-Sev – Pain-free] contrast did not reach significance: offset stimulus 22.5 SE 12.2, p = 0.066. Other comparisons, including Moderate-to-Severe vs Mild offset stimulus and all across-group contrasts, were not statistically significant. Postestimation contrasts with the KOOS-pain group multivariable model were similar with both sets of results suggesting that offset analgesia deficits were significant in participants with Mild knee pain. With greater knee pain, these deficits were not as great, controlling for other factors known to be associated with greater knee pain, including QST evidence of sensitization, pain widespreadedness, and psychological factors included in the multivariable models above.
Since stimulus temperature was different across subgroups (Table 3, T1 temperature), it is possible that subgroup differences in offset analgesia arose from different stimulus intensity, such that knee osteoarthritis groups would have had similar offset analgesia if the T1 temperature were higher in those patients. Since the range of T1 temperatures in each subgroup was overlapping (Pain-free: 44°C-48°C; Mild: 43°C-47°C; Mod-Sev: 41°C-47°C) and nonoverlapping temperatures were used infrequently (percentage of tests at a given temperature out of all knee and forearm tests: overlapping temperatures 44°C-47°C 95.5%; nonoverlapping temperatures 41°C 1.5%, 42°C 1.5%, 43°C 1%, 48°C 0.5%;), we used regression modeling to determine whether offset analgesia was different across groups when controlling for T1 temperature used as a covariate. Solving a linear mixed effects model with the outcome variable set to pain change index, body site nested within participant, and a two-way interaction between knee pain group and stimulus (offset vs. constant control) with additional covariates of body site and T1 temperature (n = 408, n(body site) = 204, n(participant) = 102; log likelihood = −1847, Wald chi2(7) = 920.06, p < 0.0001), there was a significant main effect of the group X stimulus interaction (chi2(2) = 7.59, p=0.023), similar to the model without the T1 temperature variable. Controlling for stimulus temperature, the post-hoc contrast-of-contrasts comparing knee OA groups with the Pain-free controls remained significant ([Mild(offset-control) – Pain-free(offset-control)] 11.7 SE 4.7 p=0.013; ([Mod-Sev(offset-control) – Pain-free(offset-control)] 10.9 SE 4.7 p=0.019; ([Mod-Sev(offset-control) – Mild(offset-control)] −0.7 SE 4.5 p=0.873). These results suggest that differences in offset analgesia across subgroups are not due to differences in stimulus temperature.
Onset hyperalgesia reflecting pain amplification is reduced in women with mild but not moderate-to-severe knee pain compared to pain-free controls.
Since central pain modulatory pathways not only inhibit but also facilitate pain, we also examined whether onset hyperalgesia was different in knee pain groups by comparing an onset stimulus with a constant control control stimulus[2]. We did not find differences between groups in pain intensity during the onset stimulus, onset pain change index, or interactions between stimulus and knee pain group with either participant-level or timeseries mixed effects models (Supplemental Fig 3). In the whole group, the onset stimulus produced greater pain than the control stimulus at timepoints following the step up in temperature, consistent with onset hyperalgesia (not shown). Given gender differences observed in offset analgesia, we examined the effect of gender on onset hyperalgesia. Using nested linear mixed effects models of similar design to those for offset analgesia, we found a significant 3-way interaction between knee pain group, stimulus (onset vs control control), and gender such that there was less onset hyperalgesia, reflected by the [onset-control control] contrast, in the Mild knee pain group compared with the Pain-free control group without an effect of body site (Figure 5; nested linear mixed effects model of onset pain change index with group X stimulus X gender and body site fixed effects and random slopes of body site nested within participant: n = 408, n(bodysite) = 204, n(participant) = 102; log likelihood = −1910, Wald chi2(12) = 345.36, p < 0.0001; post-estimation main effect of 3-way group X stimulus X gender: chi2(2) = 6.14, p=0.0465; main effect of body site: chi2(1) = 1.35, p=0.2448; contrast of contrast [Mild(onset-control) – Pain-free(onset-control)] in females −18.22 SE 7.7 p=0.018). Other contrasts of contrasts were not statistically significant, reflected by overlap of 95% confidence intervals with 0 in Figure 5B. In women with Moderate-to-Severe knee pain, there was no difference in onset hyperalgesia with Pain-free control participants ([Mod-Sev(onset-control) – Pain-free(onset-control)] in females −8.62 SE 7.5 p=0.254). Similar results were obtained when controlling for stimulus temperature as a covariate in the above models. Taken together, this suggests that onset hyperalgesia is reduced in women with knee osteoarthritis in Mild knee pain and not in women with Moderate-to-Severe knee pain. Moreover, this is independent of adaptation, given the analysis approach included data from the constant control, and is likely a central mechanism, given a lack of body site effects. In a timeseries analysis incorporating gender, there was no statistically significant main effect of gender or a 4-way interaction (group X stimulus X time X gender), potentially due to sample size limitations.
Figure 5: Onset hyperalgesia is reduced in women with mild knee osteoarthritis pain.
Using linear mixed effects modeling, there was a significant 3-way interaction between stimulus (onset stimulus vs. onset control), group (Pain-free controls, Mild knee pain, Moderate-to-severe knee pain, “Mod-Sev”), and gender (male vs. female). In A., linear predictions of the pain change index are plotted for each stimulus, group, and gender with 95% CI. In B., contrast of contrasts results are plotted with 95% CI. The [Mild(onset-control) – Pain-free(onset-control)] contrast is statistically significantly negative, indicating that Mild pain patients had less onset hyperalgesia than Pain-free control participants.
Warm stimuli lasting 30-seconds were also applied to the forearm and knee as an assessment of heat allodynia. Consistent with greater pain intensity in the Moderate-to-Severe knee pain group during the offset stimulus (Figure 2), timeseries analysis showed the Moderate-to-Severe group reported statistically significantly greater pain during the warm stimulation compared with Pain-free controls at the nonpainful forearm and painful knee (Supplemental Fig 4). However, the maximum pain intensity during the warm stimulus was not statistically significant (Table 3). Notably, these differences in pain intensity are small in magnitude (3–6 mm / 100 mm on COVAS pain intensity).
We also examined whether offset analgesia and onset hyperalgesia might change over the course of the testing session. Applying linear mixed models, we found that the pain change index for the offset stimulus and onset stimulus did not change over three trials at the forearm or knee (Supplemental Fig 5). Although the control for offset analgesia did not change, we did observe an increase in pain change index during the constant control stimulus for onset hyperalgesia. This was associated with a decrease in the [onset-control] contrast with trial 3 compared to trial 1. However, there was no impact on differences across subgroups, with results of the model showing no across-group differences in onset hyperalgesia consistent with above models. For both offset analgesia and onset hyperalgesia, there was no effect of body site on trial effects. Taken together, we did not observe large group-level differences over repeated trials at the forearm or knee, suggesting that, at the group-level, attenuation or augmentation of offset analgesia and onset hyperalgesia does not occur.
Deficits in offset analgesia are associated with worse performance during walking and stair climbing.
Participants performed standardized mobility tasks known to provoke knee pain and used as objective measures of physical function. After QST measurements, participants walked for 6-minutes on a level surface and climbed a single step with 16 repetitions alternating the leading leg. Distance walked and time required for stair climbing were recorded. Statistically significant correlations were observed in all participants such that greater (more negative) offset analgesia at the forearm was associated with longer walking distance (Figure 6A; r = −0.30, n = 102, p = 0.002) and faster stair climbing (Figure 6B, r = 0.21, n = 102, p = 0.032). In a subgroup analysis, the Moderate-to-Severe knee pain group had a moderate negative correlation (Figure 6C, r = −0.41, n = 36, p = 0.0129). However, in the Mild pain group, this correlation did not meet statistical significance (r = −0.19, n = 36, p = 0.265), and there was no relationship in the Pain-free control group (r = −0.03, n = 30, p = 0.880). When subdivided by the maximum pain intensity reported immediately following walking, there was a moderate-strong negative correlation between offset analgesia and walking distance in patients with Moderate-to-Severe walking pain (Figure 6D, r = −0.58, n = 26, p = 0.0017). Similarly, correlations were not significant in the other subgroups (mild walking pain, r = −0.22, n = 34, p = 0.204; no walking pain, r = 0.05, n = 42, p = 0.743). Subgroup analysis of stair climbing showed similar trends, although no correlations reached statistical significance. Overall, this suggests that patients with less offset analgesia and greater knee arthritis pain, assessed either at the beginning of the study visit or after standardized pain-evoking tasks, have worse walking performance. On the other hand, patients with Moderate-to-Severe pain but robust offset analgesia have maintained walking ability.
Figure 6: Offset analgesia deficits are associated with worse objective performance during walking and stair climbing.
Scatter plots show the relationship between offset analgesia and walking distance during the 6-minute walk task (A., C., D.) or the duration required for climbing up and down an aerobic step 24 times (B.; y-axis = pain change index, more negative values reflect more analgesia). Statistically significant linear regression lines are plotted with a shaded region reflecting the 95% confidence interval. Lines without the shaded region were not statistically significant. In C., the participants are subdivided by knee pain group. In D. participants are subdivided by the maximum pain intensity (NRS 0–10) participants reported during the walk, assessed immediately afterwards.
Most other measures of central sensitization were not associated with pain or performance during walking or stair climbing tasks.
Most measures of pain sensitivity at a nonpainful body site, and therefore reflecting CNS sensitization processes, were not significantly correlated with walking or stair climbing performance. Temporal summation of pinprick pain, pressure pain thresholds, heat pain thresholds, heat hypersensitivity (T1 used and maximum pain during warm stimuli) at nonpainful sites were all not correlated with walking or stair climbing performance. Notably, CPM, reflecting centrally-mediated spatial pain inhibitory processes, was not correlated with mobility task outcomes either with the test stimulus of pressure pain applied to the trapezius (nonpainful) or knee (painful) body sites. Pinprick pain intensity at the nonpainful elbow did positively correlate with stair climb duration (r = 0.20, n = 100, p = 0.0413) and maximal pain during stair climbing (r = 0.34, n = 102, p = 0.005), without significant correlations for walking outcomes. Hand pain during cold water immersion positively correlated with maximum pain during walking (r = 0.23, n = 99, p = 0.02) and stair climbing (r = 0.26, n = 99, p = 0.0094) without significant correlation to functional measures. The calibrated temperature to elicit painful heat, T1, at the forearm negatively correlated with maximum pain during stair climbing (r = −0.25, n = 102, p = 0.013).
Quantitative sensory testing over the painful knee, reflecting some combination of peripheral and central mechanisms, did reveal some correlations with mobility outcomes. Pressure pain threshold at the lateral knee positively correlated with walking distance (r = 0.35, n = 103, p = 0.0002) such that higher thresholds (lower sensitivity) were associated with longer walking distances. Pinprick sensitivity at the painful knee was associated with walking and stair climbing function and pain (walking distance: r = −0.28, n = 102, p = 0.0038; stair climbing duration: r = 0.22, n = 100, p = 0.0272; walking max pain: r = 0.35, n = 102, p = 0.0003; stair climbing max pain: r = 0.43, n = 102, p < 0.0001).
In knee osteoarthritis patients in which knee radiographs were available (n=73; pain-free controls did not have knee radiographs), we observed different frequencies of KL grade in Mild versus Mod-Sev pain groups with more KL grade 2 and 4 scores in the Mod-Sev group and more KL grade 3 scores in the Mild group (Chi2(2) = 9.18, p=0.01). However, there were no statistically significant effects of knee joint degeneration, defined by KL grade, on quantitative sensory tests, including offset analgesia. KL grade 4 patients had higher pain during stair climbing (pain NRS; grade 2: 3.8 (SD 3.1), grade 3: 3.8 (SD 2.7), grade 4: 5.8 (SD 2.4), 1-way ANOVA p = 0.049). There was a trend towards greater stair climbing time with greater KL grade (grade 2: 90.6 s (SD 26.1 s), grade 3: 96.0 s (SD 21.2 s), grade 4: 109.1 s (SD 23.3 s), 1-way ANOVA p = 0.06). Walking distance was greatest in KL grade 3 patients (grade 2: 1263 ft (SD 287 ft), grade 3: 1362 ft (SD 276 ft), grade 4: 1116 ft (SD 270 ft), 1-way ANOVA p = 0.015). Significant univariable Pearson’s correlations with walk distance included BMI, pin and pressure sensitivity at the index knee, situational catastrophizing related to walking, and PROMIS measures of fatigue, sleep, social engagement, physical function, and pain interference. A 1-way ANOVA showed a main effect of race on walk distance. There was no association between walk distance and the following variables: age, gender, temporal summation, heat pain threshold, CPM, anxiety, depression, marital status, or living situation. Using a likelihood ratio test comparing multivariable linear regression models, a model including offset analgesia at the forearm fit better than a model without offset analgesia controlling for the above significant univariable predictors of walking distance as well as age and gender (offset model: F(14, 54) = 11.45, p<0.0001, R2 = 0.748; smaller model: F(13, 55) = 10.16, p<0.0001, R2 = 0.706; likelihood ratio test Chi2(1) = 10.6, p=0.001). Overall, this demonstrates that, in knee osteoarthritis, offset analgesia independently predicts performance on the 6-minute walk test.
Discussion
Currently, peripheral pathophysiologic mechanisms are thought to predominate in chronic nociceptive pain [32; 48]. In this study, we find evidence that deficits in offset analgesia, reflecting reduced CNS-mediated pain inhibition, are associated with worse pain and functional performance in knee osteoarthritis, a common chronic nociceptive pain syndrome. Using suprathreshold heat stimuli at the nonpainful forearm, we demonstrate that participants with knee osteoarthritis have reduced offset analgesia compared with Pain-free control participants controlling for adaptation. Offset analgesia deficits were independent of body site tested, suggesting a CNS mechanism. Our study suggests that CNS temporal pain modulation is a clinically relevant mechanism in knee osteoarthritis.
The temporal aspects of pain have been studied in multiple settings, each relating time-dependent nociceptive processing to pain behavior. In one framework, time-dependent pain prediction underlies behavioral adaptations to avoid pain and seek pain relief [5; 13; 18; 29; 62; 76; 86–88]. In another, learning and expectations, which are also time-dependent, underly placebo analgesia and nocebo hyperalgesia [11; 12; 20; 21; 29; 44]. In both settings, there is a time-ordered processing of (1) incoming nociceptive information, (2) prediction generation in the context of the current state of the organism, and (3) action selection, manifesting as behavioral responses [29]. This fundamental brain mechanism can be measured in patients with offset analgesia, which involves the time-ordered processing of noxious input, a prediction of decreased pain at the offset of the strong nociceptive input, and the behavioral manifestation of dramatically reduced pain intensity.
We and others suggest that time-ordered pain processing contributes to chronic pain and, in light of our current findings, we propose a model in which osteoarthritis pain shapes time-ordered nociceptive processing through changes in pain prediction. In knee osteoarthritis, nociceptive input during the gait cycle includes pain offsets immediately followed by increasing nociceptive input or “onsets” when forces return to the joint. After nociceptive offsets with rest, pain consistently returns with use of the joint. Both scenarios lead to a learned deemphasis of nociceptive offsets in prediction processing since these nociceptive offsets less reliably predict pain relief. Ultimately, this leads to behavioral manifestations of increased pain intensity and decreased functional performance on painful tasks, such as walking and stair climbing. This model is consistent with our observation that pain intensity is higher after stimulus offset in the Moderate-to-Severe knee pain group but not the Mild group (Figure 2) and that offset analgesia deficits are correlated with worse pain and performance on the 6-minute walk and stair climbing tasks (Figure 6A&B). In patients with worse knee pain and disability, nociceptive offsets have been deemphasized, and this is associated with reduced offset analgesia.
Somewhat counter to this explanatory model are our observations that (1) there is no difference between Mild and Moderate-to-Severe knee pain groups in offset analgesia after normalizing to maximum pain intensity (Figure 3) and that (2) there is a large range in offset analgesia at a given knee pain intensity. First, it may be that normalized offset analgesia is less important in prediction weighting than “raw” pain intensity. Second, we anticipate that there is population-level variation in pain prediction -- patients may be better or worse at updating pain prediction weights. Cognitive and affective-motivational factors influence other pain modulatory processes, including placebo analgesia and CPM [7; 10; 70; 104; 105], making this possibility likely. Alternatively, nociceptive onsets could develop greater emphasis in pain prediction, rather than deemphasizing offsets. This is less likely since knee osteoarthritis patients did not have elevated onset hyperalgesia[2]. Differences in past pain experiences may also contribute to variation in pain prediction, such that offset deemphasis makes patients susceptible to worse knee osteoarthritis pain and impairment. Although our model of deemphasized offset weighting is compelling, future prospective studies are required to elucidate the causal relationships this model suggests.
It is likely that offset analgesia is centrally mediated. Offset analgesia can occur even when the offset stimulus is decomposed and applied to different, remote body areas [54]. Multiple neuroimaging modalities show activations consistent with descending inhibitory pathways [3; 24; 53; 64; 92; 111], which in animal models are known to decrease nociceptive input at the spinal cord [23; 72; 110]. Brain regions implicated in offset analgesia overlap with placebo analgesia [3; 64; 105; 111; 114]. Brain stimulation and lesion studies of offset analgesia implicate CNS structures [22; 82; 93]. Although there is some evidence for peripheral mechanisms of offset analgesia [9; 57; 65; 79], there was no effect of ischemic conduction block of peripheral A-delta fibers on offset analgesia [56], arguing against an exclusively peripheral mechanism. Interestingly, the one brain lesion study of offset analgesia implicates the cerebellum [82], which has both well-known motor functions and recently-appreciated roles in pain and placebo analgesia [19; 95]. Our observation of a correlation between offset analgesia and performance on painful motor tasks provides additional evidence for a role for the cerebellum in knee osteoarthritis pain processing.
Gender differences in temporal pain modulation have been reported previously. In studies of pain-free healthy participants with reported gender comparisons, offset analgesia in males is greater than females, although the difference was not large [45; 69]. Other studies have reported no gender differences [65; 66; 68; 102]. In patients with clinical pain, gender differences are unclear due to different factors: a population restricted to only one gender, significantly skewed gender distributions, no formal comparisons reported, or epidemiologic sampling including Pain-free individuals [50; 66; 73; 97; 99]. In our study, we observe offset analgesia is less in males than females in a large sample of patients with chronic nociceptive pain. We also find that onset hyperalgesia is reduced only in women in the Mild pain group, suggesting a gender-specific protective mechanism such that less central pain facilitation leads to less knee pain. Future work in patients with chronic pain with sample sizes large enough to allow gender-based analysis will be required to clarify potential gender differences.
Several studies have examined offset analgesia in patients with clinical pain. Offset analgesia is reduced compared with Pain-free controls in mixed neuropathic pain cohorts and fibromyalgia [43; 50; 53; 69; 73; 113]. At the population level, offset analgesia is associated with pain assessed by questionnaire [66], although pain diagnosis or classification was unknown in this study. In most of these studies, stimuli are applied to nonpainful regions, most commonly the forearm, finding deficits in offset analgesia when compared with Pain-free control participants [39; 50; 73; 90; 99]. In migraine, a deficit in offset analgesia was only observed when stimulating the trigeminal region and not the forearm [99]. The current study is the first to our knowledge to find that offset analgesia is reduced in chronic nociceptive pain, specifically knee osteoarthritis. Regarding body site, we found no difference between nonpainful forearm and painful knee stimulation. Taken together, our results suggest that, in a subset of patients with knee osteoarthritis, the somatic sensory system shows deficits in descending temporal inhibition, manifest by body-wide reductions in offset analgesia. This reduction is clinically relevant since it is associated with pain and functional impairment during clinically important weight-bearing tasks, such as walking.
The potential relationship between offset analgesia deficits and nociplastic pain is intriguing. Nociplastic pain is defined as pain arising from altered nociception without identified tissue damage, epitomized by fibromyalgia[47]. Offset analgesia is reduced in fibromyalgia[73]. In the current study, we identified deficits in offset analgesia affecting the whole body, which could be considered a nociplastic mechanism. However, the patients did not have widespread pain since fibromyalgia was exclusionary and the widespread pain index was low. Similarly, while CPM is often deficient or paradoxically pain amplifying in fibromyalgia[71; 78], we did not observe changes in CPM in knee osteoarthritis, although others have[6; 27]. Given that offset analgesia and CPM reflect distinct mechanisms underlying endogenous pain inhibition (temporal and spatial, respectively) [64], it is possible that we have identified a nociplastic mechanism in knee osteoarthritis patients which is temporally defined. This is distinct from heightened temporal summation or heat hyperalgesia, observed in nociplastic pain[71], since we found that offset analgesia deficits in knee osteoarthritis patients remained when controlling for temporal summation or T1 temperature used in multivariable regression. Overall, our results and others[75; 83] imply inter-individual variability in multiple nociplastic-like mechanisms in knee osteoarthritis. Prospective studies with comprehensive measures, including offset analgesia, are needed to understand the clinical impact of this heterogeneity.
Because we were not able to selectively manipulate offset analgesia, our understanding is correlational rather than proving causality. However, with a large sample size, we do find significant deficits in offset analgesia in patients with knee osteoarthritis, suggesting that impaired predictive pain inhibition contributes to osteoarthritis pain. Importantly, reduced predictive pain inhibition is associated with greater intermittent movement-evoked pain and objective functional performance. We anticipate this CNS mechanism has clinical implications for osteoarthritis pain management, e.g. movement-evoked pain during physical therapy, weight-bearing pain during recovery from joint replacement, etc. Overall, this study adds to our understanding of CNS-mediated sensitization processes in knee osteoarthritis and identifies a promising mechanistic target to reduce joint pain and prevent disability.
Supplementary Material
Acknowledgements
This study was supported by the University of Pittsburgh and the National Institutes of Health (NINDS K23NS123429). Recruitment via Pitt+Me was supported by the National Institutes of Health through Grant Number UL1 TR001857, KL2 TR001856, and/or TL1 TR001858. The authors report no conflicts of interest.
Data availability statement:
Data presented in this paper are available upon reasonable request.
References
- [1].Alter BJ, Anderson NP, Gillman AG, Yin Q, Jeong JH, Wasan AD. Hierarchical clustering by patient-reported pain distribution alone identifies distinct chronic pain subgroups differing by pain intensity, quality, and clinical outcomes. PLoS One 2021;16(8):e0254862. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Alter BJ, Aung MS, Strigo IA, Fields HL. Onset hyperalgesia and offset analgesia: Transient increases or decreases of noxious thermal stimulus intensity robustly modulate subsequent perceived pain intensity. PLoS One 2020;15(12):e0231124. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Alter BJ, Santosa H, Nguyen QH, Huppert TJ, Wasan AD. Offset analgesia is associated with opposing modulation of medial versus dorsolateral prefrontal cortex activations: A functional near-infrared spectroscopy study. Mol Pain 2022;18:17448069221074991. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, Christy W, Cooke TD, Greenwald R, Hochberg M, et al. Development of criteria for the classification and reporting of osteoarthritis. Classification of osteoarthritis of the knee. Diagnostic and Therapeutic Criteria Committee of the American Rheumatism Association. Arthritis Rheum 1986;29(8):1039–1049. [DOI] [PubMed] [Google Scholar]
- [5].Apkarian AV, Baliki MN, Geha PY. Towards a theory of chronic pain. Prog Neurobiol 2009;87(2):81–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Arant KR, Katz JN, Neogi T. Quantitative sensory testing: identifying pain characteristics in patients with osteoarthritis. Osteoarthritis Cartilage 2022;30(1):17–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Ashar YK, Chang LJ, Wager TD. Brain Mechanisms of the Placebo Effect: An Affective Appraisal Account. Annu Rev Clin Psychol 2017;13:73–98. [DOI] [PubMed] [Google Scholar]
- [8].Aspinall SL, Leboeuf-Yde C, Etherington SJ, Walker BF. Feasibility of using the Neuropen for temporal summation testing. Pain Manag 2019;9(4):361–368. [DOI] [PubMed] [Google Scholar]
- [9].Asplund CL, Kannangath A, Long VJE, Derbyshire SWG. Offset analgesia is reduced on the palm and increases with stimulus duration. Eur J Pain 2021;25(4):790–800. [DOI] [PubMed] [Google Scholar]
- [10].Atlas LY. A social affective neuroscience lens on placebo analgesia. Trends Cogn Sci 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].Atlas LY, Wager TD. How expectations shape pain. Neurosci Lett 2012;520(2):140–148. [DOI] [PubMed] [Google Scholar]
- [12].Babel P, Bajcar EA, Adamczyk W, Kicman P, Lisinska N, Swider K, Colloca L. Classical conditioning without verbal suggestions elicits placebo analgesia and nocebo hyperalgesia. PLoS One 2017;12(7):e0181856. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Baliki MN, Apkarian AV. Nociception, Pain, Negative Moods, and Behavior Selection. Neuron 2015;87(3):474–491. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Bannister K, Dickenson AH. The plasticity of descending controls in pain: translational probing. J Physiol 2017;595(13):4159–4166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Barratt W The Barratt Simplified Measure of Social Status (BSMSS) measuring SES., Vol. 2017, 2006. [Google Scholar]
- [16].Beswick AD, Wylde V, Gooberman-Hill R, Blom A, Dieppe P. What proportion of patients report long-term pain after total hip or knee replacement for osteoarthritis? A systematic review of prospective studies in unselected patients. BMJ Open 2012;2(1):e000435. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Campbell CM, Kronfli T, Buenaver LF, Smith MT, Berna C, Haythornthwaite JA, Edwards RR. Situational Versus Dispositional Measurement of Catastrophizing: Associations With Pain Responses in Multiple Samples. Journal of Pain 2010;11:443–453.e442. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Cecchi GA, Huang L, Hashmi JA, Baliki M, Centeno MV, Rish I, Apkarian AV. Predictive dynamics of human pain perception. PLoS Comput Biol 2012;8(10):e1002719. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Chen C, Niehaus JK, Dinc F, Huang KL, Barnette AL, Tassou A, Shuster SA, Wang L, Lemire A, Menon V, Ritola K, Hantman A, Zeng H, Schnitzer MJ, Scherrer G. Neural circuit basis of placebo pain relief. Nature 2024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Colagiuri B, Quinn VF, Colloca L. Nocebo Hyperalgesia, Partial Reinforcement, and Extinction. J Pain 2015;16(10):995–1004. [DOI] [PubMed] [Google Scholar]
- [21].Colloca L, Barsky AJ. Placebo and Nocebo Effects. N Engl J Med 2020;382(6):554–561. [DOI] [PubMed] [Google Scholar]
- [22].Cosentino G, Antoniazzi E, Cavigioli C, Tang V, Tammam G, Zaffina C, Tassorelli C, Todisco M. Repetitive Transcranial Magnetic Stimulation of the Human Motor Cortex Modulates Processing of Heat Pain Sensation as Assessed by the Offset Analgesia Paradigm. J Clin Med 2023;12(22). [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23].De Preter CC, Heinricher MM. The ‘in’s and out’s’ of descending pain modulation from the rostral ventromedial medulla. Trends Neurosci 2024;47(6):447–460. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Derbyshire SW, Osborn J. Offset analgesia is mediated by activation in the region of the periaqueductal grey and rostral ventromedial medulla. Neuroimage 2009;47(3):1002–1006. [DOI] [PubMed] [Google Scholar]
- [25].Dunn KM, Saunders KW, Rutter CM, Banta-Green CJ, Merrill JO, Sullivan MD, Weisner CM, Silverberg MJ, Campbell CI, Psaty BM, Von Korff M. Opioid prescriptions for chronic pain and overdose: a cohort study. Ann Intern Med 2010;152(2):85–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Edwards RR, Dolman AJ, Martel MO, Finan PH, Lazaridou A, Cornelius M, Wasan AD. Variability in conditioned pain modulation predicts response to NSAID treatment in patients with knee osteoarthritis. BMC Musculoskelet Disord 2016;17:284. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Fernandes C, Pidal-Miranda M, Samartin-Veiga N, Carrillo-de-la-Pena MT. Conditioned pain modulation as a biomarker of chronic pain: a systematic review of its concurrent validity. Pain 2019;160(12):2679–2690. [DOI] [PubMed] [Google Scholar]
- [28].Fernández-de-Las-Peñas C, Florencio LL, de-la-Llave-Rincón AI, Ortega-Santiago R, Cigarán-Méndez M, Fuensalida-Novo S, Plaza-Manzano G, Arendt-Nielsen L, Valera-Calero JA, Navarro-Santana MJ. Prognostic Factors for Postoperative Chronic Pain after Knee or Hip Replacement in Patients with Knee or Hip Osteoarthritis: An Umbrella Review. J Clin Med 2023;12(20). [DOI] [PMC free article] [PubMed] [Google Scholar]
- [29].Fields HL. How expectations influence pain. Pain 2018;159 Suppl 1:S3–S10. [DOI] [PubMed] [Google Scholar]
- [30].Finan PH, Buenaver LF, Bounds SC, Hussain S, Park RJ, Haque UJ, Campbell CM, Haythornthwaite JA, Edwards RR, Smith MT. Discordance between pain and radiographic severity in knee osteoarthritis: findings from quantitative sensory testing of central sensitization. Arthritis Rheum 2013;65(2):363–372. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Frey-Law LA, Bohr NL, Sluka KA, Herr K, Clark CR, Noiseux NO, Callaghan JJ, Zimmerman MB, Rakel BA. Pain sensitivity profiles in patients with advanced knee osteoarthritis. Pain 2016;157(9):1988–1999. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32].Fu K, Robbins SR, McDougall JJ. Osteoarthritis: the genesis of pain. Rheumatology (Oxford) 2018;57(suppl_4):iv43–iv50. [DOI] [PubMed] [Google Scholar]
- [33].Gaudry E, Vagg P, Spielberger CD. Validation of the State-Trait Distinction in Anxiety Research. Multivariate behavioral research 1975;10(3):331–341. [DOI] [PubMed] [Google Scholar]
- [34].Gilbody S, Richards D, Brealey S, Hewitt C. Screening for depression in medical settings with the Patient Health Questionnaire (PHQ): a diagnostic meta-analysis. J Gen Intern Med 2007;22(11):1596–1602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [35].Greco CM, Wasan AD, Schneider MJ, Mehling W, Williams DA, Darwin J, Harte SE. Biobehavioral Assessments in BACPAC: Recommendations, Rationale, and Methods. Pain Med 2023;24(Suppl 1):S61–s70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [36].Greco CM, Yu L, Johnston KL, Dodds NE, Morone NE, Glick RM, Schneider MJ, Klem ML, McFarland CE, Lawrence S, Colditz J, Maihoefer CC, Jonas WB, Ryan ND, Pilkonis PA. Measuring nonspecific factors in treatment: item banks that assess the healthcare experience and attitudes from the patient’s perspective. Qual Life Res 2016;25(7):1625–1634. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [37].Grill JD, Coghill RC. Transient analgesia evoked by noxious stimulus offset. J Neurophysiol 2002;87(4):2205–2208. [DOI] [PubMed] [Google Scholar]
- [38].Gullo TR, Golightly YM, Cleveland RJ, Renner JB, Callahan LF, Jordan JM, Kraus VB, Nelson AE. Defining multiple joint osteoarthritis, its frequency and impact in a community-based cohort. Seminars in arthritis and rheumatism 2019;48(6):950–957. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [39].Havelin J, Imbert I, Cormier J, Allen J, Porreca F, King T. Central Sensitization and Neuropathic Features of Ongoing Pain in a Rat Model of Advanced Osteoarthritis. J Pain 2016;17(3):374–382. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [40].Hays RD, Bjorner JB, Revicki DA, Spritzer KL, Cella D. Development of physical and mental health summary scores from the patient-reported outcomes measurement information system (PROMIS) global items. Qual Life Res 2009;18(7):873–880. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [41].Hays RD, Schalet BD, Spritzer KL, Cella D. Two-item PROMIS® global physical and mental health scales. J Patient Rep Outcomes 2017;1(1):2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [42].Hays RD, Spritzer KL, Schalet BD, Cella D. PROMIS(®)-29 v2.0 profile physical and mental health summary scores. Qual Life Res 2018;27(7):1885–1891. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [43].Hermans L, Calders P, Van Oosterwijck J, Verschelde E, Bertel E, Meeus M. An Overview of Offset Analgesia and the Comparison with Conditioned Pain Modulation: A Systematic Literature Review. Pain Physician 2016;19(6):307–326. [PubMed] [Google Scholar]
- [44].Hoffman GA, Harrington A, Fields HL. Pain and the placebo: what we have learned. Perspect Biol Med 2005;48:248–265. [DOI] [PubMed] [Google Scholar]
- [45].Honigman L, Yarnitsky D, Sprecher E, Weissman-Fogel I. Psychophysical testing of spatial and temporal dimensions of endogenous analgesia: conditioned pain modulation and offset analgesia. Experimental brain research 2013;228:493–501. [DOI] [PubMed] [Google Scholar]
- [46].Izumi M, Hayashi Y, Saito R, Oda S, Petersen KK, Arendt-Nielsen L, Ikeuchi M. Detection of altered pain facilitatory and inhibitory mechanisms in patients with knee osteoarthritis by using a simple bedside tool kit (QuantiPain). Pain reports 2022;7(3):e998. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [47].Kaplan CM, Kelleher E, Irani A, Schrepf A, Clauw DJ, Harte SE. Deciphering nociplastic pain: clinical features, risk factors and potential mechanisms. Nature reviews Neurology 2024;20(6):347–363. [DOI] [PubMed] [Google Scholar]
- [48].Katz JN, Arant KR, Loeser RF. Diagnosis and Treatment of Hip and Knee Osteoarthritis: A Review. JAMA 2021;325(6):568–578. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [49].Kellgren JH, Lawrence JS. Radiological assessment of osteo-arthrosis. Ann Rheum Dis 1957;16(4):494–502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [50].Kobinata H, Ikeda E, Zhang S, Li T, Makita K, Kurata J. Disrupted offset analgesia distinguishes patients with chronic pain from healthy controls. Pain 2017;158(10):1951–1959. [DOI] [PubMed] [Google Scholar]
- [51].Kolasinski SL, Neogi T, Hochberg MC, Oatis C, Guyatt G, Block J, Callahan L, Copenhaver C, Dodge C, Felson D, Gellar K, Harvey WF, Hawker G, Herzig E, Kwoh CK, Nelson AE, Samuels J, Scanzello C, White D, Wise B, Altman RD, DiRenzo D, Fontanarosa J, Giradi G, Ishimori M, Misra D, Shah AA, Shmagel AK, Thoma LM, Turgunbaev M, Turner AS, Reston J. 2019 American College of Rheumatology/Arthritis Foundation Guideline for the Management of Osteoarthritis of the Hand, Hip, and Knee. Arthritis care & research 2020;72(2):149–162. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [52].Krebs EE, Lorenz KA, Bair MJ, Damush TM, Wu J, Sutherland JM, Asch SM, Kroenke K. Development and initial validation of the PEG, a three-item scale assessing pain intensity and interference. J Gen Intern Med 2009;24(6):733–738. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [53].Li T, Zhang S, Ikeda E, Kobinata H. Functional connectivity modulations during offset analgesia in chronic pain patients: an fMRI study. Brain imaging and behavior 2022. [DOI] [PubMed] [Google Scholar]
- [54].Ligato D, Petersen KK, Morch CD, Arendt-Nielsen L. Offset analgesia: The role of peripheral and central mechanisms. Eur J Pain 2018;22(1):142–149. [DOI] [PubMed] [Google Scholar]
- [55].Lockwood SM, Bannister K, Dickenson AH. An investigation into the noradrenergic and serotonergic contributions of diffuse noxious inhibitory controls in a monoiodoacetate model of osteoarthritis. J Neurophysiol 2019;121(1):96–104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [56].Luebke L, Lopes CG, Myka Y, Lumma A, Adamczyk WM, Carvalho GF, Scholten-Peeters GGM, Luedtke K, Szikszay TM. Assessing the Influence of Nonischemic A-Fiber Conduction Blockade on Offset Analgesia: An Experimental Study. J Pain 2024;25(10):104611. [DOI] [PubMed] [Google Scholar]
- [57].Luebke L, von Selle J, Adamczyk WM, Knorr MJ, Carvalho GF, Gouverneur P, Luedtke K, Szikszay TM. Differential Effects of Thermal Stimuli in Eliciting Temporal Contrast Enhancement: A Psychophysical Study. J Pain 2024;25(1):228–237. [DOI] [PubMed] [Google Scholar]
- [58].Maier C, Baron R, Tolle TR, Binder A, Birbaumer N, Birklein F, Gierthmuhlen J, Flor H, Geber C, Huge V, Krumova EK, Landwehrmeyer GB, Magerl W, Maihofner C, Richter H, Rolke R, Scherens A, Schwarz A, Sommer C, Tronnier V, Uceyler N, Valet M, Wasner G, Treede RD. Quantitative sensory testing in the German Research Network on Neuropathic Pain (DFNS): somatosensory abnormalities in 1236 patients with different neuropathic pain syndromes. Pain 2010;150(3):439–450. [DOI] [PubMed] [Google Scholar]
- [59].Margari F, Lorusso M, Matera E, Pastore A, Zagaria G, Bruno F, Puntillo F, Margari L. Aggression, impulsivity, and suicide risk in benign chronic pain patients - a cross-sectional study. Neuropsychiatric disease and treatment 2014;10:1613–1620. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [60].Marino EN, Rosen KD, Gutierrez A, Eckmann M, Ramamurthy S, Potter JS. Impulsivity but not sensation seeking is associated with opioid analgesic misuse risk in patients with chronic pain. Addict Behav 2013;38(5):2154–2157. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [61].Martel MO, Petersen K, Cornelius M, Arendt-Nielsen L, Edwards R. Endogenous Pain Modulation Profiles Among Individuals With Chronic Pain: Relation to Opioid Use. J Pain 2019;20(4):462–471. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [62].Mulders D, Seymour B, Mouraux A, Mancini F. Confidence of probabilistic predictions modulates the cortical response to pain. Proc Natl Acad Sci U S A 2023;120(4):e2212252120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [63].Nahman-Averbuch H, Dayan L, Sprecher E, Hochberg U, Brill S, Yarnitsky D, Jacob G. Sex differences in the relationships between parasympathetic activity and pain modulation. Physiol Behav 2016;154:40–48. [DOI] [PubMed] [Google Scholar]
- [64].Nahman-Averbuch H, Martucci KT, Granovsky Y, Weissman-Fogel I, Yarnitsky D, Coghill RC. Distinct brain mechanisms support spatial vs temporal filtering of nociceptive information. Pain 2014;155(12):2491–2501. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [65].Naugle KM, Cruz-Almeida Y, Fillingim RB, Riley JL. Offset analgesia is reduced in older adults. Pain 2013;154:2381–2387. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [66].Naugle KM, Cruz-Almeida Y, Fillingim RB, Riley JL 3rd. Loss of Temporal Inhibition of Nociceptive Information Is Associated With Aging and Bodily Pain. J Pain 2017;18(12):1496–1504. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [67].Nelson AE. Multiple joint osteoarthritis (MJOA): What’s in a name? Osteoarthritis Cartilage 2023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [68].Niesters M, Dahan A, Swartjes M, Noppers I, Fillingim RB, Aarts L, Sarton EY. Effect of ketamine on endogenous pain modulation in healthy volunteers. Pain 2011;152:656–663. [DOI] [PubMed] [Google Scholar]
- [69].Niesters M, Hoitsma E, Sarton E, Aarts L, Dahan A. Offset analgesia in neuropathic pain patients and effect of treatment with morphine and ketamine. Anesthesiology 2011;115:1063–1071. [DOI] [PubMed] [Google Scholar]
- [70].Nir RR, Yarnitsky D, Honigman L, Granot M. Cognitive manipulation targeted at decreasing the conditioning pain perception reduces the efficacy of conditioned pain modulation. Pain 2012;153(1):170–176. [DOI] [PubMed] [Google Scholar]
- [71].O’Brien AT, Deitos A, Triñanes Pego Y, Fregni F, Carrillo-de-la-Peña MT. Defective Endogenous Pain Modulation in Fibromyalgia: A Meta-Analysis of Temporal Summation and Conditioned Pain Modulation Paradigms. J Pain 2018;19(8):819–836. [DOI] [PubMed] [Google Scholar]
- [72].Ossipov MH, Morimura K, Porreca F. Descending pain modulation and chronification of pain. Curr Opin Support Palliat Care 2014;8(2):143–151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [73].Oudejans LCJ, Smit JM, van Velzen M, Dahan A, Niesters M. The influence of offset analgesia on the onset and offset of pain in patients with fibromyalgia. Pain 2015;156(12):2521–2527. [DOI] [PubMed] [Google Scholar]
- [74].Patton JH, Stanford MS, Barratt ES. Factor structure of the Barratt impulsiveness scale. Journal of clinical psychology 1995;51(6):768–774. [DOI] [PubMed] [Google Scholar]
- [75].Petersen KK, O’Neill S, Blichfeldt-Eckhardt MR, Nim C, Arendt-Nielsen L, Vaegter HB. Pain profiles and variability in temporal summation of pain and conditioned pain modulation in pain-free individuals and patients with low back pain, osteoarthritis, and fibromyalgia. Eur J Pain 2024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [76].Petre B, Tetreault P, Mathur VA, Schurgin MW, Chiao JY, Huang L, Apkarian AV. A central mechanism enhances pain perception of noxious thermal stimulus changes. Sci Rep 2017;7(1):3894. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [77].Plummer F, Manea L, Trepel D, McMillan D. Screening for anxiety disorders with the GAD-7 and GAD-2: a systematic review and diagnostic metaanalysis. Gen Hosp Psychiatry 2016;39:24–31. [DOI] [PubMed] [Google Scholar]
- [78].Potvin S, Marchand S. Pain facilitation and pain inhibition during conditioned pain modulation in fibromyalgia and in healthy controls. Pain 2016;157(8):1704–1710. [DOI] [PubMed] [Google Scholar]
- [79].Robinson CJ, Erik Torebjork H, LaMotte RH. Psychophysical detection and pain ratings of incremental thermal stimuli: A comparison with nociceptor responses in humans. Brain Research 1983;274:87–106. [DOI] [PubMed] [Google Scholar]
- [80].Rolke R, Magerl W, Campbell KA, Schalber C, Caspari S, Birklein F, Treede RD. Quantitative sensory testing: a comprehensive protocol for clinical trials. Eur J Pain 2006;10(1):77–88. [DOI] [PubMed] [Google Scholar]
- [81].Roos EM, Roos HP, Lohmander LS, Ekdahl C, Beynnon BD. Knee Injury and Osteoarthritis Outcome Score (KOOS)--development of a self-administered outcome measure. J Orthop Sports Phys Ther 1998;28(2):88–96. [DOI] [PubMed] [Google Scholar]
- [82].Ruscheweyh R, Kühnel M, Filippopulos F, Blum B, Eggert T, Straube A. Altered experimental pain perception after cerebellar infarction. Pain 2014;155:1303–1312. [DOI] [PubMed] [Google Scholar]
- [83].Rutter-Locher Z, Norton S, Denk F, McMahon S, Taams LS, Kirkham BW, Bannister K. A randomised controlled trial of the effect of intra-articular lidocaine on pain scores in inflammatory arthritis. Pain 2024;165(11):2578–2585. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [84].Sachau J, Otto JC, Kirchhofer V, Larsen JB, Kennes LN, Hüllemann P, Arendt-Nielsen L, Baron R. Development of a bedside tool kit for assessing sensitization in patients with chronic osteoarthritis knee pain or chronic knee pain after total knee replacement. Pain 2021. [DOI] [PubMed] [Google Scholar]
- [85].Scherrer KH, Ziadni MS, Kong JT, Sturgeon JA, Salmasi V, Hong J, Cramer E, Chen AL, Pacht T, Olson G, Darnall BD, Kao MC, Mackey S. Development and validation of the Collaborative Health Outcomes Information Registry body map. Pain reports 2021;6(1):e880. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [86].Seymour B Pain: A Precision Signal for Reinforcement Learning and Control. Neuron 2019;101(6):1029–1041. [DOI] [PubMed] [Google Scholar]
- [87].Seymour B, Crook RJ, Chen ZS. Post-injury pain and behaviour: a control theory perspective. Nat Rev Neurosci 2023;24(6):378–392. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [88].Seymour B, O’Doherty JP, Koltzenburg M, Wiech K, Frackowiak R, Friston K, Dolan R. Opponent appetitive-aversive neural processes underlie predictive learning of pain relief. Nat Neurosci 2005;8(9):1234–1240. [DOI] [PubMed] [Google Scholar]
- [89].Shen Y, Bhagwandass H, Branchcomb T, Galvez SA, Grande I, Lessing J, Mollanazar M, Ourhaan N, Oueini R, Sasser M, Valdes IL, Jadubans A, Hollmann J, Maguire M, Usmani S, Vouri SM, Hincapie-Castillo JM, Adkins LE, Goodin AJ. Chronic Opioid Therapy: A Scoping Literature Review on Evolving Clinical and Scientific Definitions. J Pain 2021;22(3):246–262. [DOI] [PubMed] [Google Scholar]
- [90].Shulman J, Zurakowski D, Keysor J, Jervis K, Sethna NF. Offset analgesia identifies impaired endogenous pain modulation in pediatric chronic pain disorders. Pain 2020. [DOI] [PubMed] [Google Scholar]
- [91].Smith SM, Dworkin RH, Turk DC, Baron R, Polydefkis M, Tracey I, Borsook D, Edwards RR, Harris RE, Wager TD, Arendt-Nielsen L, Burke LB, Carr DB, Chappell A, Farrar JT, Freeman R, Gilron I, Goli V, Haeussler J, Jensen T, Katz NP, Kent J, Kopecky EA, Lee DA, Maixner W, Markman JD, McArthur JC, McDermott MP, Parvathenani L, Raja SN, Rappaport BA, Rice ASC, Rowbotham MC, Tobias JK, Wasan AD, Witter J. The Potential Role of Sensory Testing, Skin Biopsy, and Functional Brain Imaging as Biomarkers in Chronic Pain Clinical Trials: IMMPACT Considerations. J Pain 2017;18(7):757–777. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [92].Sprenger C, Stenmans P, Tinnermann A, Buchel C. Evidence for a spinal involvement in temporal pain contrast enhancement. Neuroimage 2018;183:788–799. [DOI] [PubMed] [Google Scholar]
- [93].Stacheneder R, Alt L, Straube A, Ruscheweyh R. Effects of Transcranial Direct Current Stimulation (t-DCS) of the Cerebellum on Pain Perception and Endogenous Pain Modulation: a Randomized, Monocentric, Double-Blind, Sham-Controlled Crossover Study. Cerebellum 2023;22(6):1234–1242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [94].Starr CJ, Houle TT, Coghill RC. Psychological and sensory predictors of experimental thermal pain: a multifactorial model. J Pain 2010;11(12):1394–1402. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [95].Strick PL, Dum RP, Fiez JA. Cerebellum and nonmotor function. Annu Rev Neurosci 2009;32:413–434. [DOI] [PubMed] [Google Scholar]
- [96].Sullivan MJL, Bishop SR, Pivik J. The Pain Catastrophizing Scale: Development and validation. Psychological Assessment 1995;7:524–532. [Google Scholar]
- [97].Suzan E, Treister R, Pud D, Haddad M, Eisenberg E. The effect of hydromorphone therapy on psychophysical measurements of the descending inhibitory pain systems in patients with chronic radicular pain. Pain Med 2015;16(1):168–175. [DOI] [PubMed] [Google Scholar]
- [98].Syx D, Tran PB, Miller RE, Malfait AM. Peripheral Mechanisms Contributing to Osteoarthritis Pain. Curr Rheumatol Rep 2018;20(2):9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [99].Szikszay TM, Adamczyk WM, Carvalho GF, May A, Luedtke K. Offset analgesia: somatotopic endogenous pain modulation in migraine. Pain 2020;161(3):557–564. [DOI] [PubMed] [Google Scholar]
- [100].Szikszay TM, Adamczyk WM, Luedtke K. The Magnitude of Offset Analgesia as a Measure of Endogenous Pain Modulation in Healthy Participants and Patients With Chronic Pain: A Systematic Review and Meta-Analysis. Clin J Pain 2019;35(2):189–204. [DOI] [PubMed] [Google Scholar]
- [101].Szikszay TM, Adamczyk WM, Panskus J, Heimes L, David C, Gouverneur P, Luedtke K. Psychological mechanisms of offset analgesia: The effect of expectancy manipulation. PLoS One 2023;18(1):e0280579. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [102].Szikszay TM, Lévénez JLM, von Selle J, Adamczyk WM, Luedtke K. Investigation of correlations between pain modulation paradigms. Pain Med 2021. [DOI] [PubMed] [Google Scholar]
- [103].Treister R, Suzan E, Lawal OD, Katz NP. Staircase-evoked Pain May be More Sensitive Than Traditional Pain Assessments in Discriminating Analgesic Effects: A Randomized, Placebo-controlled Trial of Naproxen in Patients With Osteoarthritis of the Knee. Clin J Pain 2019;35(1):50–55. [DOI] [PubMed] [Google Scholar]
- [104].Valentini E, Martini M, Lee M, Aglioti SM, Iannetti GD. Seeing facial expressions enhances placebo analgesia. Pain 2014;155(4):666–673. [DOI] [PubMed] [Google Scholar]
- [105].Wager TD, Atlas LY. The neuroscience of placebo effects: connecting context, learning and health. Nature Reviews Neuroscience 2015;16:403–418. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [106].Wang YP, Gorenstein C. Psychometric properties of the Beck Depression Inventory-II: a comprehensive review. Revista brasileira de psiquiatria (Sao Paulo, Brazil : 1999) 2013;35(4):416–431. [DOI] [PubMed] [Google Scholar]
- [107].Wasan AD, Alter BJ, Edwards RR, Argoff CE, Sehgal N, Walk D, Moeller-Bertram T, Wallace MS, Backonja M. Test-Retest and Inter-Examiner Reliability of a Novel Bedside Quantitative Sensory Testing Battery in Postherpetic Neuralgia Patients. J Pain 2020;21(7–8):858–868. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [108].Wideman TH, Finan PH, Edwards RR, Quartana PJ, Buenaver LF, Haythornthwaite JA, Smith MT. Increased sensitivity to physical activity among individuals with knee osteoarthritis: relation to pain outcomes, psychological factors, and responses to quantitative sensory testing. Pain 2014;155(4):703–711. [DOI] [PubMed] [Google Scholar]
- [109].Wolfe F, Clauw DJ, Fitzcharles MA, Goldenberg DL, Hauser W, Katz RL, Mease PJ, Russell AS, Russell IJ, Walitt B. 2016 Revisions to the 2010/2011 fibromyalgia diagnostic criteria. Seminars in arthritis and rheumatism 2016;46(3):319–329. [DOI] [PubMed] [Google Scholar]
- [110].Yao D, Chen Y, Chen G. The role of pain modulation pathway and related brain regions in pain. Rev Neurosci 2023;34(8):899–914. [DOI] [PubMed] [Google Scholar]
- [111].Yelle MD, Oshiro Y, Kraft RA, Coghill RC. Temporal filtering of nociceptive information by dynamic activation of endogenous pain modulatory systems. J Neurosci 2009;29(33):10264–10271. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [112].Yelle MD, Rogers JM, Coghill RC. Offset analgesia: A temporal contrast mechanism for nociceptive information. Pain 2008;134:174–186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [113].Zhang S, Li T, Kobinata H, Ikeda E, Ota T, Kurata J. Attenuation of offset analgesia is associated with suppression of descending pain modulatory and reward systems in patients with chronic pain. Mol Pain 2018;14:1744806918767512. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [114].Zunhammer M, Spisák T, Wager TD, Bingel U. Meta-analysis of neural systems underlying placebo analgesia from individual participant fMRI data. Nat Commun 2021;12(1):1391. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data presented in this paper are available upon reasonable request.