Chronic pain due to osteoarthritis (OA) may be aggravated by “central sensitisation”, whereby pain-processing pathways become sensitised by inflammatory and degenerative disease processes¹. We used 1H Magnetic Resonance Spectroscopy (MRS) to investigate whether there were biochemical changes in pain processing regions of the brain² that could be related to perceived pain in patients with hand OA.

MR acquisition: Data were acquired at 3T using PRESS single voxel localisation with TE 32ms TR 2000ms and 96 averages. Voxels were planned on 3D T1w images and placed in the anterior cingulate gyrus (voxel size 25 x 20 x 10 mm³) and insula cortex (voxel size 25 x 10 x 15 mm³). Metabolite levels were determined using LCModel™ as ratios to total creatine (tCr) to avoid the confound of CSF partial volumes. Data were only included for SNR > 8 and CRLB < 12% for the major metabolites of NAA, tCr, tCho, mI and Glx (Glu plus Gln).

Patients: Data were acquired on 32 hand OA patients (ages 49 to 76 yr) and 14 controls with no history of OA or chronic pain (ages 43 to 71 yr). Clinical scores were available for 22 patients as measured by visual analogue scale (VAS) for self-reported pain, painDETECT, the Australian and Canadian Hand Osteoarthritis Index (AUSCAN) and the Hospital Anxiety and Depression Scale (HADS).

Analysis: Statistical analyses were made with SPSS for linear correlations between MRS measures and clinical scores, Principal Component Analysis (PCA) with Varimax rotation, and a t-test for group comparisons.

An initial group comparison of was made with age-matched OA patients (63 ± 5 yr), for which there was a significant difference in VAS pain scores (n= 5 per group, VAS 5.2 ± 4 and VAS 8 ± 0.7 with p = 0.0079). There was no significant difference in NAA/tCr or tCho/tCr between these groups. However Glx/tCr was reduced, mI/tCr was increased and mI/Glx was significantly increased in patients (0.34 ± 0.03 v. 0.45 ± 0.08, p = 0.017) suggesting mI/Glx may be a useful biomarker relating to pain.

Over the full age range, there were no metabolite differences found between controls and OA participants in the anterior cingulate gyrus or insula cortex. Also, there were no age related changes of metabolite ratios in either region in the controls or in the anterior cingulate gyrus of OA patients. However, in the insula cortex of OA patients the mI/Glx ratio was positively correlated with age (R² = 0.29, p = 0.0018, see Figure 1) and with VAS pain score after co-varying for age (R² = 0.52, p = 0.018).

Since clinical scores of pain and related symptoms such as depression and anxiety are related, we performed Principal Component Analysis (PCA) across all clinical scores as a data reduction method to find which scores are related to MRS in the OA patient subgroup. Three PCs described 71% of the variance: PC1 was strongly weighted by AUSCAN pain and anxiety scores; PC2 by HADS anxiety and depression scores; PC3 by AUSCAN stiffness, painDETECT and VAS pain score. Using a ranked PC and mI/Glx correlation to minimise the effect of outliers, we found PC3 was correlated to mI/Glx (R² = 0.188, p = 0.049, see Figure 2), with a stronger correlation after co-varying for age (R² = 0.46, p = 0.041).

The correlation of insular cortex mI/Glx to PC3 of the clinical scores supports our initial hypothesis of this metabolite ratio being a biomarker relevant to perceived pain. However, how this metabolite ratio changes in relation to OA pain may not be straightforward, since in those age < 65 yr, mI/Glx is lower in OA patients than controls, and for patients aged > 65 yr there is elevated mI/Glx compared to controls (see Fig 1). Increased myo-Inositol has been observed in neuro-inflammation³ and an increase in glutamate has been associated with acute pain stimulation⁴. Hence there could be disease related changes in mI and Glx that are dependent on patient age and/or disease duration.