Does Weight-bearing Versus Non-weight-bearing Pain Reflect Different Pain Mechanisms in Knee Osteoarthritis?: The Multicenter Osteoarthritis Study (MOST)

Abstract

Objective

Knee osteoarthritis (OA) is predominantly characterized by pain with weight-bearing activities. Pain at rest also occurs but the mechanisms for this are not clear. We evaluated the relations of nociceptive signal alterations to weight-bearing and non-weight-bearing pain in knee OA.

Design

We used data from a NIH–funded longitudinal cohort of older adults with or at risk of knee OA. We evaluated quantitative sensory testing (QST) measures (pressure pain threshold (PPT) at patellae and the wrist; mechanical temporal summation (TS); conditioned pain modulation (CPM)). Each WOMAC pain question was dichotomized as having at least moderate pain, and we further categorized them as weight-bearing pain and non-weight-bearing pain. We evaluated the relation of QST measures to each pain outcome using logistic regression, adjusting for potential confounders.

Results

2749 participants (5479 knees) were included (mean age 64±11, 57% female). Each SD unit decrease in patellar PPT was associated with greater odds of both weight-bearing pain (OR 1.51 (95% CI 1.27, 1.79)) and non-weight-bearing pain (OR 1.46 (1.20–1.77)), while wrist PPT was associated with greater odds of weight-bearing pain (OR 1.27 (1.15, 1.39)) but only with pain during sitting/lying (OR 1.20 (1.01, 1.43)). TS was significantly associated with greater odds of pain with walking and stairs (OR 1.11 (1.01, 1.23), 1.11 (1.03, 1.20), respectively). CPM was not associated with any pain outcomes.

Conclusions

Our findings challenge the hypothesis that non-weight-bearing pain may reflect greater pain sensitization and/or inefficient CPM than weight-bearing pain in knee OA, suggesting other mechanisms are likely responsible.

Introduction

Osteoarthritis (OA) affects about 500 million people worldwide and over 34 million people in the United States^1,2. The knee is a common site affected by OA, typically manifesting with pain during weight-bearing activities. This weight-bearing pain is thought to be reflective of nociceptive pain^3,4. Pain at rest, i.e., non-weight bearing pain, is also present in 23% to 81% of individuals with knee OA^5–8. Given that joint loading during sitting or lying, while not absent, is substantially less than during weight-bearing activities like walking or stair-climbing, pain at rest may not reflect mechanical nociception.

Understanding mechanisms that distinguish between weight-bearing and non-weight-bearing pain can help advance rational pain management strategies. However, few studies to date have investigated mechanisms underlying the differences in these types of pain. Diffuse pain around the knee, greater constant or intermittent pain, and psychological symptoms have been associated with pain at rest; however, weight-bearing pain was not evaluated⁹. In studies that evaluated both types of pain, neuropathic-like symptoms, pain catastrophizing, depression, bone marrow lesions and joint effusions were associated with both types of pain, but the associations were of larger magnitudes for weight-bearing pain (i.e., pain with walking, standing, stairs)^8,10.

An important and potentially targetable mechanism that may contribute to the experience of non-weight bearing pain is an alteration of nociceptive signaling. In animal models, persistent painful inputs can induce neuroplastic changes in the peripheral nervous system (i.e., peripheral sensitization), and similar changes can occur in the central nervous system (i.e., central sensitization)^11,12. Pain sensitization has been associated with pain presence and severity in knee OA^13,14, but its association with non-weight bearing pain has not been well-studied to date. Two prior small studies found that increased pain sensitivity measured by quantitative sensory testing (QST) at the knee or the hand was associated with non-weight-bearing pain but not with weight-bearing pain in knee OA¹⁵ or post-total knee replacement¹⁶. These studies suggest that pain sensitization (i.e., ascending facilitatory pathways) may contribute to non-weight-bearing pain to a greater degree than to weight-bearing pain. However, descending endogenous inhibitory signaling, also known as conditioned pain modulation (CPM), is another important nociceptive mechanism contributing to pain perception¹⁷, but has not yet been studied in this regard.

We therefore sought to determine the relation of pain sensitization and CPM to non-weight-bearing and weight-bearing pain in knee OA. We hypothesized that pain sensitization and inefficient CPM are associated with non-weight-bearing pain (i.e., pain at rest) more so than with weight-bearing pain.

Methods

Study Sample

The Multicenter Osteoarthritis Study (MOST) is a National Institutes of Health–funded longitudinal cohort of older adults with or at risk of knee OA. At baseline, 3,026 subjects, aged 50–79 years, were recruited from Birmingham, Alabama, and Iowa City, Iowa. Details of the cohort have been published elsewhere^18,19. The study protocol was approved by the institutional review boards at the University of Iowa, the University of Alabama at Birmingham, the University of California San Francisco, and Boston University Medical Center. In this cross-sectional study, we used data from the 144-month visit, which was the first visit at which conditioned pain modulation (CPM) was measured. Participants at the 144-month study visit comprised 2 cohorts: participants from the original cohort who were recruited between 2003 and 2005, had or were at risk of knee OA and were aged between 50–79 years at baseline (Original Cohort); and new participants recruited between 2016 and 2018, who were aged 45–69 years, with either no or only minimal knee pain and radiographic Kellgren-Lawrence grade ≤ 2 (New Cohort)^18,19. For these analyses, participants who had a history of knee replacement and peripheral neuropathy were excluded from this study.

Exposures

Quantitative Sensory Testing measures:

Quantitative sensory testing (QST) refers to a series of standardized psychophysical measures that are used to assess functioning of nociceptive signaling, enabling evaluation of pain sensitization and descending inhibitory pathways^20,21. We used the following measures:

Pressure pain threshold (PPT).

PPT is a reliable measure of sensitivity to nociception evoked by mechanical stimulation using a pressure algometer^22,23. We measured PPT at the right distal radioulnar joint (the wrist) and patellae. PPT at a local site (e.g., knee) is thought to reflect peripheral sensitization with or without central sensitization, while PPT at a distant non-diseased body site (e.g., wrist) is thought to reflect central sensitization^23,24. PPT was assessed using a handheld pressure algometer (1 cm² rubber tip; Wagner FDIX25) applied at a constant rate of 0.5 kg/second on the testing body site. PPT was defined as the point at which the participant verbally indicated that the pressure first changed to slight pain. The PPT at each anatomic site was calculated by averaging 3 trials. Lower PPT values indicate more pain sensitivity.

Mechanical Temporal Summation (TS).

TS, an amplified response to repetitive stimulation, is a sensitive measure of central augmented pain processing in chronic pain conditions including knee OA^20,23,24. We assessed TS using a standard set of weighted probes from 64–512 mN. Participants rated the pain experienced by each successive weighted probe being touched on the skin of the right distal radioulnar joint (the wrist) until a pain rating of ≥ 4/10 was achieved. If a pain rating of ≥ 4/10 was not achieved by the highest weighted probe (512 mN), then the examiner proceeded with that highest weighted probe²⁵. The selected probe was then applied at a frequency of 1 Hz for 10 seconds. Participants provided a pain rating at the completion of the train of 10 stimulations and 15 seconds post-stimulation. TS was defined as the difference between the highest post-stimulation pain rating and the initial pain rating. A post-stimulation pain rating greater than the initial pain rating was considered to be reflective of facilitated TS^23,24.

Conditioned Pain Modulation (CPM).

CPM evaluates the adequacy of the descending pain modulatory pathways^17,20. We assessed CPM using PPT as the test stimulus (PPT1) at the wrist as described above, with forearm ischemia using a blood pressure cuff as the conditioning stimulus²⁵. Briefly, the blood pressure cuff applied to the contralateral arm was inflated to 10mm Hg above systolic pressure. The participant was then instructed to perform hand grip squeezes until pain of at least 4/10 occurred in the forearm. PPT was then reassessed at the wrist 3 times and averaged (PPT2). CPM is considered to be inefficient when the post-conditioning stimulus PPT is the same or lower than the initial (pre-conditioning stimulus) PPT. Percent efficiency of CPM (%CPM) was computed as PPT2/PPT1, multiplied by 100; %CPM ≤ 100 indicates inefficient CPM^26,27.

Fourteen-day test–retest reliability for PPT was 0.85–0.90 (intraclass coefficients) and for temporal summation was 0.61 (kappa statistic)^13,28. We did not assess test-retest reliability of CPM in this study.

Outcomes

The Western Ontario McMaster Universities Osteoarthritis Index (WOMAC) pain subscale (0–20 score) was used to assess knee-specific pain severity over the past 30 days²⁹. The pain subscale consists of 5 questions, 3 of which are related to weight-bearing pain (pain with walking, standing, and going up or down stairs), and 2 of which are related to non-weight-bearing pain (pain with sitting or lying, and with sleeping at night), each assessed on a 0–4 Likert scale (none (0), mild (1), moderate (2), severe (3), and extreme (4)). We dichotomized each of the individual WOMAC pain questions as having at least moderate pain (≥2)¹⁰. Additionally, we combined the 3 weight-bearing questions and the 2 non-weight-bearing questions as follows. We averaged the scores for the weight-bearing questions and for the non-weight-bearing questions (separately), and then dichotomized these average scores at a value of at least 2 as per our primary approach to denote having at least moderate pain overall for each pain category (weight-bearing, non-weight-bearing). In addition, we categorized people as having both types of pain if they met that threshold for both averages.

Statistical Analysis

We performed knee-specific analyses to separately evaluate the relation of each QST measure to presence of having at least moderate pain on each activity and combined activities for weight-bearing (i.e., walking, standing, stairs) and for non-weight-bearing (i.e., sitting/lying, sleeping at night) in separate models using logistic regression with generalized estimating equation (GEE) to account for correlations between two knees within an individual. We analyzed the QST exposures (all continuous) in the direction that would indicate more pain sensitivity, per standard deviation (SD) units to allow for comparisons across the exposures. As such, we analyzed PPT and %CPM per SD unit decrease while TS was analyzed per SD unit increase. Thus, for each of these analyses, a higher risk ratio represents a higher likelihood of the pain outcome.

As a sensitivity analysis, we evaluated the relation of each QST measure to a multi-level pain outcome defined as no pain, non-weight-bearing pain only, weight-bearing pain only, or both, using multinomial regression, with weight-bearing pain only as the referent group.

All analyses were adjusted for potential confounders including age, sex, BMI, race, depressive symptoms (defined as ≥16 on Center for Epidemiologic Studies Depression Scale³⁰), pain catastrophizing (one item from the Coping Strategies Questionnaire³¹), sleep quality (poor sleep was defined very bad or fairly bad sleep in the last 7 days from a Likert scale sleep question), and widespread pain using a body homunculus defined as pain being present above and below the waist, on the right and left sides of the body, and axial pain.³² Analyses were conducted using SAS version 9.4 (SAS Institute, Cary, North Carolina, USA).

Results

There were 2749 participants (5479 knees) eligible for this study (Supplementary Figure 1). The mean age was 64 years, and more than half of the participants were female (57%) and Caucasian (53%) (Table 1). Depressive symptoms, pain catastrophizing, poor sleep and widespread pain were present in 12%, 41%, 16%, and 34% of the participants, respectively. Overall mean values of PPT at the patellae and wrist, TS (i.e., different in pain rating pre- to post-train of stimuli), %CPM efficiency were 5.6kgf, 3.9kgf, 1.41 (0–10 pain scale), and 109%, respectively.

Table 1:

Participant Characteristics

Descriptive Statistics	N = 2749 participants (5479 knees)
Age, mean ± SD (years)	63.9 ± 10.6
Women, n (%)	1563 (57)
Caucasian, n (%)	1456 (53)
BMI, mean ± SD (kg/m2)	29.5 ± 5.72
Depressive Symptom, n (%)	325 (12)
Pain Catastrophizing, n (%)	1134 (41)
Poor Sleep, n (%)	446 (16)
Widespread Pain, n (%)	936 (34)
PPT patellae (kgf) mean ± SD	5.60 ± 2.51
PPT wrist (kgf), mean ± SD	3.90 ± 2.0
TS* (0–10 pain scale), mean ± SD	1.41 ± 1.7
%CPM†, mean ± SD	109.1 ± 30.1

^*TS: temporal summation was calculated by post-stimulation pain rating (0–10) minus th initial pain rating

^†%CPM: % efficiency of conditioned pain modulation was computed as PPT2/PPT1, multiplied by 100

Abbreviations: BMI: body mass index; PPT: pressure pain threshold; TS: temporal summation; CPM: conditioned pain modulation

Of 5479 knees, 8.5%, 8.1% and 23.5% had at least moderate pain during walking, standing and stairs, respectively while 5.6% and 6.6% had at least moderate pain during sitting/lying and sleeping at night. Further, 6.3% and 4.3% had at least moderate pain for overall weight-bearing pain and non-weight-bearing pain, respectively.

Each SD unit decrease in patellar PPT (signifying more sensitization) was associated with 1.32 to 1.52 times significantly greater odds of having at least moderate pain on each of the individual WOMAC pain questions as well as for overall weight-bearing and overall non-weight-bearing pain (Figure 1A). The numerical effect estimates are presented in Supplementary Table 1.

Figure 1:

Relations of patellar PPT, wrist PPT, TS and CPM to individual and combined WOMAC pain

Outcomes: At least having moderate pain on individual WOMAC questions or at least moderate pain on the average of combined WB (or NWB) pain questions. Pressure pain threshold (PPT) and % conditioned pain modulation (CPM) are presented as standardized effect estimates per standard deviation unit decrease whereas temporal summation (TS) is presented as standardized effect estimates per standard deviation unit increase.

Similarly, each SD unit decrease in wrist PPT was associated with 1.27 to 1.36 times significantly greater odds of pain with weight-bearing activities (i.e., walking, standing, stairs, and overall weight-bearing activities).(Figure 1B) and 1.2 times significantly greater odds of having at least moderate pain during sitting/lying. However, wrist PPT was not significantly associated with pain during sleeping at night or overall non-weight bearing pain (aOR 1.13, 95%CI 0.97–1.31 and aOR 1.11, 95% CI 0.92–1.34, respectively).

Each SD unit increase in TS was associated with 1.11 times greater odds of having at least moderate pain with walking and stairs (aOR 1.11, 95% CI 1.01–1.23 and aOR 1.11, 95% CI: 1.03–1.20) (Figure 1C). TS was not significantly associated with pain during standing (OR 1.09, 95% CI 0.98,1.22), although this effect estimate was similar to that for pain with walking. Further, TS was not associated with pain during non-weight-bearing activities, with effect estimates being close to the null.

CPM was not associated with any of the five pain outcomes, with all aORs being close to the null (Figure 1D).

In sensitivity analyses, PPT at the patella and wrist had similar lower odds of having non-weight-bearing pain only compared with weight-bearing pain only (OR 0.77, 95% CI 0.58–1.02 and 0.61–0.98, respectively) (Supplementary Table 2). PPT at the patella was significantly associated with greater likelihood of having both types of pain (OR 1.20, 95% CI 1.02–1.42). There were no significant associations of TS or %CPM with non-weight-bearing pain compared with weight-bearing pain.

Discussion

We found that greater pain sensitization measured by patellar PPT was associated with both weight-bearing and non-weight-bearing pain. Wrist PPT and TS, both measures of central pain sensitization, were associated with weight-bearing pain; wrist PPT was also associated with non-weight-bearing pain. CPM, reflecting descending pain modulation, was not associated with any of the pain outcomes. These results were contrary to our hypothesis that pain sensitization and/or inefficient CPM contribute to non-weight-bearing pain. Our results therefore suggest that other mechanisms are likely important for understanding non-weight-bearing pain in knee OA, and that perhaps there is not a simple dichotomy between peripheral or central pain mechanisms to explain pain presence with different types of weight-bearing versus non-weight-bearing activities.

In our study, pain sensitization, as assessed by patellar PPT, was associated with both weight-bearing and non-weight-bearing pain whereas a recent small study found that patellar PPT was associated with pain at rest but not with weight-bearing¹⁵, though their sample predominantly comprised those with moderate pain during walking so there may not have been a sufficient range in pain severity to evaluate this. Our results suggest factors at the joint level may contribute to pain under both weight-bearing and non-weight-bearing conditions. One such joint-level factor that could contribute to both pain and pain sensitization is inflammation. Joint effusion and synovitis have been associated with both weight-bearing and non-weight bearing pain in some^10,33, but not all^15,34, studies. Nonetheless, it is recognized that inflammation can sensitize peripheral nociceptors leading to peripheral sensitization³⁵. We have previously reported that synovitis and effusion were associated with patellar PPT²⁸. There are also data to suggest that different inflammatory cytokines may have differential associations with pain at rest versus with weight-bearing; synovial fluid IL-6, IL-8, and TNFα have been associated with pain during gait or standing, while only TNFα was associated with night pain and pain with sitting or lying³³.

Central sensitization may also be associated with inflammation. Peripheral inflammatory mediators can lead to neuroplastic changes in the central nociceptive system, resulting in widespread hyperalgesia and ascending pain facilitation^12,35. We found that both wrist PPT and TS were significantly associated with weight-bearing pain. Wrist PPT was also associated with pain during sitting/lying; however, the effect estimate was slightly smaller than that for weight-bearing outcomes. Central sensitization is an amplified responsiveness to stimuli in the central nervous system³⁶ and therefore greater stimuli (e.g., greater mechanical loading to the knee) could feasibly have an impact on nociceptive signaling in the central nervous system. Thus, weight-bearing activities may be expected to have greater effects on ascending nociceptive facilitated signaling than non-weight-bearing activities where there may be less nociceptive input.

CPM, which aims to evaluate the efficiency of descending pain modulatory pathways^17,20, was not associated with any of the pain outcomes in our study. With point estimates close to the null for each of the activities, our findings suggest that descending pain modulation is unlikely to play an important role in differentiating pain with versus without weight-bearing activities.

A strength of our study is its large size, and evaluation of both ascending and descending nociceptive pathways using QST. Our sample consisted of a community-based cohort that reflects a broad range of age and disease duration, contributing to generalizability. Limitations include our cross-sectional study design, which precludes conclusions regarding the direction of the associations. The study sample had a low prevalence of individuals with at least moderate pain with sitting or lying, although a similar low prevalence was noted for having at least moderate pain with walking and standing; thus, the lack of associations with non-weight-bearing pain is unlikely related to lack of sufficient precision. We cannot rule out potential for measurement error for our CPM assessment. The effect estimates were relatively small across many of our results; however, given that pain is multifactorial in nature, these results can provide insights regarding the potential contribution of pain sensitization to weight-bearing and/or non-weight-bearing pain. Although we controlled for several relevant potential confounders, residual confounding may remain. When we additionally adjusted for sleep quality and radiographic knee OA severity, the results did not materially change (Supplementary Table 3), which is consistent with prior findings²⁸. We were unable to assess whether duration of pain may have impacted these associations, though we previously did not note an association between duration of knee OA and pain sensitization¹³.

In conclusion, our findings challenge the hypothesis that non-weight-bearing pain may reflect greater pain sensitization and inefficient CPM than weight-bearing pain. Other mechanisms, such as inflammation, may be responsible for non-weight-bearing pain in knee OA. Alternatively, the same mechanisms may underlie both weight-bearing and non-weight-bearing pain without distinction between the two.

Supplementary Material

1

Supplemental Figure 1: CONSORT Flow Diagram