Relating clinical measures of pain with experimentally assessed pain mechanisms in patients with knee osteoarthritis

2.1 Material

Patients diagnosed with radiological and symptomatic KOA in 2007-2009 (n = 48) drawn from the Hospital patient files were contacted by telephone and asked if they were willing to undergo a pain assessment session. Seventeen still reporting pain (pain duration range 24–468 months) in their knee agreed to participate of whom eight had meanwhile undergone a TKA (24–48 months prior to the current study). Participants that had undergone TKA and participants that had not undergone TKA had similar clinical pain characteristics. The participants were not allowed to have pain or sensory dysfunctions in any other region than the knee and they should be able to collaborate. Furthermore they were asked to refrain from the use of any pain medication 24 h before the experiment. Demographics are shown in Table 1. This study was approved by the local ethics committee of the North Denmark Region (N-20100050) and conducted in accordance with the Helsinki Declaration. Before the study, oral and written information were provided to the participants, and written consent was obtained from all of the participants.

2.2 Clinical pain

Before the pain assessment session the participants were asked to rate their peak pain intensity in the previous 24 h using a 10 cm (0–10) VAS with the left end (0) marked as ‘no pain’ and the right end (10) by ‘worst pain possible’ (VAS max).

2.3 Protocol

The participants rested in a comfortable recumbent position during the pain assessment session. The pressure pain sensitivity in the peripatellar region and at TA of the affected leg was assessed by handheld pressure algometry and temporal summation of pressure pain was assessed using a computer-controlled pressure algometer. The pressure pain sensitivity of the lower leg was further evaluated by cuff-algometry including assessment of temporal and spatial summation. The conditioning pain modulation was evaluated by experimental tonic pain induced in the arm by cuff pressure stimulation (conditioning stimulation) and assessment of pressure pain thresholds (test stimulus) in the peripatellar region.

2.3.1 Assessment sites

Eight test sites in the peripatellar region and one site at TA (5 cm distal to the tibial tuberosity) were located in relation to bony landmarks and marked. Site 1) 2 cm distal to the inferior medial edge of patella; site 2) 2 cm distal to the inferior lateral edge of patella; site 3) 3 cm lateral to the midpoint on the lateral edge of patella; site 4) 2 cm proximal to the superior lateral edge of patella; site 5) 2 cm proximal to the superior edge of patella; site 6) 2 cm proximal to the superior medial edge of patella; site 7) 3 cm medial to the midpoint on the medial edge of patella; and site 8) at centre of patella (Fig. 1).

Fig. 1

PPT measurement sites.

2.4 Statistical analysis

Data are presented as mean values and standard deviations (SD). Data were normally distributed, confirmed by visual inspection of Q-Q plots. Pearsons Product Moment Correlations were assessed between the pain assessment parameters. To correct for multiple correlations P-values were Bonferroni corrected. The pain assessment parameters were set as independent variables in a multiple regression model where VAS max (Table 1) was set as the dependent variable. VAS max was chosen as the dependent variable since the intensity of the ongoing pain is related to the sensitisation in patients with KOA [14,16] and other chronic pain conditions [20]. Variables with a P-value of less than 0.20 in the univariate analysis were included in the multivariate regression. The construction of the multiple regression model followed the construction proposed by Hosmer et al. [21]. Variables in the final model with a P-value above 0.05 were only kept in the model if it caused more than 20% change in the estimate of the other variables. The assumptions of normality and constant variance of the model were verified using a normal probability plot of residuals and a plot of standardised residuals against standardised predicted value. Tolerance level was used to test for colinearity. Any variable that has a tolerance level of less than 0.01 were excluded from the model. Beta values were used as a measure of how much each independent variable affected the dependent variable. P-values less than 0.05 were considered to be significant. All analyses were done using IBM SPSS Statistics (Version 19).

4.1 Model predicting maximal clinical pain intensity in KOA

In the univariate analysis a significant negative association was found between the peak pain intensity in the previous 24 h (VAS max) and pressure pain thresholds assessed manually in the peripatellar region and at TA and computer-controlled pressure algometer in the peripatellar region. This is consistent with previous studies on osteoarthritis where reduced pressure pain thresholds were related to higher pain intensity [14,16,23]. Furthermore a significant positive relation was found between the peak pain intensity and the degree of temporal summation assessed by cuff algometry. This is also supported by previous studies on KOA and temporal summation [14]. Some care should be taken in the interpretation of the association between the peak pain intensity and the pain sensitivity parameters, since ten variables were applied in the univariate analysis, which could lead to a false positive error. However, as presented, the findings are in agreement with previous studies [14,16,23].

In the present study, the model consisting of pressure pain thresholds assessed manually at TA and temporal summation assessed using a cuff-algometer, accounted for 55% of the variance in peak pain in previous 24 h (VAS max). The beta values in the final model indicate that the impact of the two independent variables (and thereby hyperalgesia and temporal summation) on peak pain in previous 24 h are almost equivalent. A change of one SD in the independent variable would result in a change of –0.56 (more pain lower PPT) and 0.50 (more pain more summation) SD in pain, respectively. In the final model pressure pain thresholds assessed manually at TA and temporal summation assessed using a cuff-algometer were significant predictors of pain. A previous study developed a similar model combining different PPT-measurements to predict pain intensity in KOA and found a R² of 0.61 [16], but the present study is the first to combine different quantitative sensory tests (QSTs) reflecting different peripheral and central pain mechanisms in the prediction of pain in KOA. In a study comparing patients with chronic whiplash-associated disorder to healthy controls the partial least squares regression showed a R² of 0.49 when trying to identify which QSTs most powerful differed between the two groups. Furthermore it was found that PPT was one of the most important variables when trying to differentiate the two groups and inter-correlations between pain intensity variables and PPT in the chronic whiplash-associated disorder group in a multivariate context [24]. Hence the model in the present study is fully equivalent to previous studies trying to account for clinical pain measures using experimentally assessed pain mechanism in chronic pain states.

4.2 Association between pain modalities

There were positive correlations between handheld PPT knee and handheld PPT TA, between computer PPT knee and handheld PPT knee, between computer PPT knee and handheld PPT TA and between VAS sum knee and VAS sum TA. An association between these parameters was expected as they all reflects different aspects of sensitisation [12,25]. In addition, previous studies found high correlations within the same pain assessment modality [26,27], but not between modalities [27,28]. We did not have the possibility to perform a full personality or a selected gene phenotyping of the participants. Hence we cannot, as in the presented studies rule out a trait-related general high pain sensitivity. However, assessing the PPT from TA is a way to investigate spreading sensitisation [12], and hence an indicator for central sensitisation as previously shown in KOA [14], unilateral epicondylitis [29], chronic low back pain [30] and whiplash [31]. In the present study we confirmed the previous finding [14] of a relationship between both peripatellar and TA PPTs and the clinical pain intensity. This support the proposition that central sensitisation is an important factor in chronic musculoskeletal pain [32].

The lack of other correlations between the other QST methods highlight the importance of applying different tests and different pain modalities when assessing the reorganised pain system as different methods add complementary information. A similar lack of correlation between different types of QST stimulation have previously been described [26,27,33,34]. One modality is truly insufficient when assessing experimentally the status and complexity of pain and a multimodal and optimally a multi-tissue assessment regime is needed [27,35].

The present study is limited by the sample size, the fact that it does not take differences in pain duration and other characteristics (gender, age, etc.) of the participants or TKA versus no TKA into account. For instance, both pain intensity [20] and duration of the pain [36] affect the degree of widespread muscle hyperalgesia and the area where the patient experiences pain. The inclusion of these aspects in this study would however have made an evaluation of the influence of the different characteristics impossible and at the same time increased the complexity of the model making the interpretation harder. Future studies should evaluate these aspects in relation to the prediction of pain in the future leading to a combined model to predict pain. Furthermore a limitation is the use of only pressure pain stimulation. Cold pain thresholds have for example been found predictive for patients developing chronic pain after a whiplash injury [37]. Different pain modalities represent different specific pathways of the pain and should therefore ideally be combined when assessing pain in an experimental setting [27,33].

However this is the first study to combine different mechanistic experimental pain measures in prediction pain in KOA and may form the basis for selecting a set of pain measures to mechanistically phenotype OA patients. This may be developed to predict high risk sensitised patients prior to surgery or selecting patients for the most optimal treatment.