• Rheumatoid Arthritis
• Inflammation
• Ultrasonography
• Disease Activity

Optical spectral transmission imaging (OI) measures the blood-specific absorption of light transmitted through tissue.1–5 OI of joints may measure the level of inflammation in joint diseases such as rheumatoid arthritis (RA), where the inflammatory process is associated with angiogenesis and increased blood flow.

We compared OI by Hemics Full Hand Proto with colour Doppler (CD) ultrasonography as gold standard reference for inflammation. Sixty-two patients fulfilling the 2010 European League Against Rheumatism (EULAR)/American College of Rheumatology (ACR) criteria for RA, with no significant deformation of hands or wrists and no surgery for the last 3 months, had OI performed of the wrists and all metacarpophalangeal (MCP) and proximal interphalangeal (PIP) joints. Median age was 62 years, 76% were female, median DAS28 was 3.6 and 71% were anti-CCP-positive. Fifty-eight received disease-modifying antirheumatic drugs (synthetic, biological or in combination), five of these in combination with prednisolone. Four patients did not receive antirheumatic treatment. An OI score of each joint was calculated by an algorithm based on absorption of light at 660 and 810 nm. A predefined cut-off for inflammation based on results from another cohort using grey-scale (GS) and CD ultrasonography as gold standard (personal communication, Wouter Rensen, Hemics) was >0.521 (unitless). A GE Logiq E9 ultrasonography machine with a ML6–15 linear array transducer and CD settings optimised for slow flow was used for grading CD (0–3),6 following the EULAR recommendations regarding patient positioning and scanning planes,7 and OMERACT definitions were followed.8 ,9

An OI score >0.521 was found in 343 (25%) of 1364 joints, 91 of which had CD ultrasonography ≥1. An OI score ≤0.521 was found in 1021 (75%) of 1364 joints, 76 of which had CD ultrasonography ≥1. Hence, the kappa agreement with ultrasonography (US) was 0.23, sensitivity 54%, specificity 79%, positive predictive value 27% and negative predictive value 93%.

In comparison, swollen/not-swollen joints by clinical examination had a kappa agreement with US of 0.35, sensitivity 41% and specificity 93%; and tender/not-tender joints had a kappa agreement of 0.25, sensitivity 34% and specificity 91%. OI did not demonstrate an improved sensitivity compared with clinical examination of the joints when the cut-off of the OI score was set at specificity of 90% comparable with that of clinical examination (table 1, also other cut-offs were tested).

Area under the curve (AUC) of the receiver operating characteristic (ROC) curve of OI score versus the standard reference was 0.69, p<0.001. When the analysis was split on different joint regions, OI performed better for PIP joints (AUC=0.71, p=0.001) and MCP joints (AUC=0.60, p=0.003), than for wrist joints (AUC=0.58, not significant (NS), (figure 1). This may be due to a larger thickness, blood vessels near the wrist joints and sparse room for light passing through the bones constituting the wrist joint, leading to a lower signal-to-noise ratio. Similar results were obtained in sensitivity analyses using CD US ≥2 (AUC=0.71), CD US ≥2 OR US GS ≥2 (AUC=0.70) and CD US ≥1 OR US GS ≥1 (AUC=0.65) as alternative gold standards, and in a sensitivity analysis excluding 54/1364 joints with suboptimal OI images due to motion artefacts or significant under/overexposure (AUC=0.69).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Receiver operating characteristic (ROC) curves for OI versus colour Doppler ultrasound as standard reference at joint level. For comparison, the sensitivity and specificity of clinical examination (clinically swollen joint) is also marked. MCP, metacarpophalangeal; OI, optical spectral transmission imaging; PIP, proximal interphalangeal.

At the patient level, OI joint count correlated with patient global visual analogue scale (VAS) (r=0.34, p=0.006), C-reactive protein (CRP) (r=0.25, p=0.047), DAS28 (r=0.31, p=0.02) and health assessment questionnaire (HAQ) (r=0.35, p=0.005), using Spearman's rank correlations coefficient. This confirms that OI is capable of measuring disease activity in RA at the patient level, as demonstrated in a recent study by van Onna et al.10 It may, however, not be more accurate than clinical examination at the joint level, and only after further development, it may become a useful tool for fast, physician-independent, non-invasive assessment of joint inflammation in RA.