Pragmatic approach to quantitative sensory testing in knee osteoarthritis: Measures of mechanical pain sensitivity predict concordant pain and structural status

Abstract

Purpose: While pain is the most important patient reported feature of OA, the pathophysiology of pain in OA is poorly understood with multiple mechanisms ranging from structural to neurogenic. Identifying pain mechanisms at a patient level will inform individualized treatment strategies and reducing the burden of OA. Quantitative Sensory Testing (QST) is a psychophysical technique being increasingly used to assess patients with pain from osteoarthritis (OA). However, studies vary in terms of the number of repeats conducted for each individual test as well as the modalities used. To inform future research, this study aimed to: 1. Investigate the trends seen across subsequent repeats for each individual test, and 2. Examine which modalities are best able to distinguish between congruent and discordant pain and structural status. Method(s): 1003 women aged 43-65 years from a single general practice in Chingford, UK were recruited in 1988-9 for a longitudinal population-based cohort primarily for studying OA and osteoporosis. The current study includes data from the 20-year follow-up visit. QST modalities included warm detect threshold, heat pain threshold, mechanical pain threshold and mechanical pain sensitivity at the sternum (distant site) and knee (local site). All thresholds were repeated 5 times apart from mechanical pain sensitivity, which was repeated 3 times. Anterio-posterior, fully-extended, weight bearing, bilateral knee radiographs were taken and knee WOMAC scores were also recorded. The trends of individual repeats for each modality were explored graphically for each test site. Paired student t-test was used to identify any significant differences between adjacent repeats. Subgroups were created using dichotomous splits of pain and radiographic osteoarthritis (ROA) status. Pain was defined as any pain captured using the Womac Pain subscore >0 and ROA was defined as Kellgren/Lawrence grade 2 or more creating 4 independent groups: ROA/pain, ROA/no pain, no ROA/no pain, no ROA/no pain. Wilcoxon- Mann-Whitney and multivariate regression modeling, adjusting for clustering of sites within a person, were used to compare the association between QST measures and knee group. We adjusted for the following potential confounding factors HADS, age and BMI. Result(s): In 462 women median age 71(67,76) years and mean BMI 27.7(5.0), 362 (39.6%) reported recent knee pain and ROAwas present in 714/958 (74.5%) of knees assessed. Figure 1 demonstrates the trend for individual repeats seen at the sternum for each modality tested. The same trends were seen at the knee (data not shown). When comparing responses by group, mechanical pain sensitivity showed significantly increased experimental sensitivity at both local (p = 0.0292) and distant sites (p = 0.0005) in the ROA/Pain group compared to the ROA/No Pain group. A non-significant increased sensitivity was also seen in the No ROA/Pain group compared to the No ROA/No Pain group. These findings remained when adjusting for confounders using the multivariate regression model, Table 1. Conclusion(s): These results suggest that three not five repeats are required for testing the QST modalities of heat pain threshold, mechanical pain threshold and mechanical pain sensitivity. For warm detect threshold the first measure should be omitted before calculating an average measure. Mechanical pain sensitivity was able to distinguish painful ROA positive knees from pain-free ROA positive knees. The fact that sensitivity at the sternum, as well as the knee, was able to predict concordant pain and structural status supports previous work showing that centrally mediated widespread pain sensitisation is present and highlights it potential use in future clinical research. (Figure Presented).