Article Text

Download PDFPDF

  1. Y. K. Tan1,
  2. C. Hong1,
  3. H. LI2,
  4. J. C. Allen Jr3,
  5. J. Thumboo1
  1. 1Singapore General Hospital, Department of Rheumatology and Immunology, Singapore, Singapore
  2. 2Singapore General Hospital, Health Services Research Unit, Singapore, Singapore
  3. 3Duke-NUS Medical School, Centre for Quantitative Medicine, Singapore, Singapore


Background: A novel combined thermal and ultrasound (CTUS) imaging approach in rheumatoid arthritis (RA) was recently shown to be superior to either imaging modality alone in terms of correlation with the 28-joint disease activity score (DAS28).

Objectives: To determine the performance of CTUS imaging in identifying RA patients with at least moderate disease activity (DAS28 > 3.2).

Methods: Bilateral hand (22 joints) thermal and ultrasound (US) imaging was performed. Thermal imaging provides the surface temperature readings at the joints with MAX, AVG and MIN derived per patient by summing the temperature differences with a control temperature, for the respective maximum (Tmax), average (Tavg) and minimum (Tmin) temperatures at each joint. US imaging assesses joint inflammation by summing up the power Doppler (PD) and grey-scale (GS) joint inflammation scores (graded 0-3 at each joint recess) at each joint to obtain the respective total PD and total GS scores per patient. CTUS imaging utilizes data from both thermal and US imaging to derive the MAX (PD), AVG (PD) and MIN (PD) by multiplying MAX, AVG and MIN by a factor of 2 when a patient’s Total PD > median score, which otherwise remained the same as the MAX, AVG and MIN. The results of the imaging parameters were compared between patients with DAS28 ≤ 3.2 and those with DAS28 > 3.2. Sensitivity (Sn), specificity (Sp) and receiver operating characteristic (ROC) curve analysis was performed to determine if the use of CTUS imaging can help identify patients with DAS28 > 3.2.

Results: In this cross-sectional study, 814 joints from 37 RA patients (75.7% female; 75.7% Chinese; baseline mean disease duration, 30.9 months; baseline mean DAS28, 4.43) were imaged. The mean (SD) values for the CTUS—but not single modality—imaging parameters (Table 1) were all significantly greater among patients with DAS28 > 3.2 versus those with DAS28 ≤ 3.2 (P-values were all <0.01). Based on cut-off levels of (a) MAX (PD) ≥ 94.5, (b) MIN (PD) ≥ 42.3 and (c) AVG (PD) ≥ 64.6 in identifying patients with DAS28 > 3.2, the respective area under the ROC curves (AUCs) (95%CIs) were (a) 0.731 (0.541, 0.921) with Sn = 58.1%; Sp = 100.0%; negative predictive value (NPV) = 31.6%; positive predictive value (PPV) = 100.0%; accuracy = 64.9%, (b) 0.758 (0.591, 0.925) with Sn = 61.3%; Sp = 100.0%; NPV = 33.3%; PPV = 100.0%; accuracy = 67.6% and (c) 0.763 (0.596, 0.931) with Sn = 61.3%; Sp = 100.0%; NPV = 33.3%; PPV = 100.0%; accuracy = 67.6%.

Conclusion: The severity of joint inflammation as detected by CTUS—but not single modality—imaging parameters were significantly greater among patients with DAS28 > 3.2 versus those with DAS28 ≤ 3.2. For the first time ever, by applying ROC analysis, this has helped to determine cut-off MAX (PD), MIN (PD) and AVG (PD) levels for identifying patients with DAS28 > 3.2; the usefulness of these cut-off levels will require further validation in independent RA cohorts.

Table 1.

Comparison of imaging parameters between patient groups.

Disclosure of Interests: None declared

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.