Chadi Sargi1, Stephanie Ducharme-Benard2, Valerie Benard3, Rosalie-Selene Meunier4, Carolyn Ross5 and Jean-Paul Makhzoum6, 1University of Montreal, Laval, Canada, 2Hopital Sacre-Coeur, Montréal, QC, Canada, 3University of Montreal, Saint-Ambroise-de-Kildare, Canada, 4Hopital Sacre-Coeur, Universite de Montreal, Montréal, QC, Canada, 5University of Montreal, Montréal, QC, Canada, 6Hopital du Sacre-Coeur de Montreal, Montréal, QC, Canada
Background/Purpose: Giant cell arteritis (GCA) is the most common primary vasculitis but remains challenging to diagnose. In the past years, many probability tools have been developed to predict the presence of GCA. Clinical prediction tools include the GCA probability score (GCAPS), the GCA prediction model (Ing Score) and the Bhavsar-Khalidi (BK) score. Color doppler ultrasound (CDUS) prediction scores include the halo count and halo score. The objective of this study is to assess and compare the performance of these probability tools to predict a final diagnosis of GCA.
Methods: A monocentric, prospective, observational study, with systematic data collection was conducted from April to December 2021. Consecutive patients referred to our tertiary CDUS Fast-Track clinic with a suspected new-onset GCA were included. CDUS of temporal and axillary arteries was performed by experimented GCA specialists using a Canon XarioTM 200 Platinum series with an 18L7 probe. Final diagnosis of GCA was determined by the treating physician. A second confirmation of the final diagnosis was performed by a vasculitis specialist (blinded to CDUS results), 3 months after the initial Fast-Track clinic consultation. Clinical probability scores were calculated once final diagnosis was confirmed using data systematically collected at the initial visit. ROC curves were assessed to determine the best cut-off values for each prediction score.
Results: 200 patients with a suspected new-onset GCA were included: 58 with confirmed GCA and 142 without GCA.
Using a cut-off value of 9.5 points as positive, the GCAPS showed a sensitivity (Se) of 98.3% and a specificity (Sp) of 62.9%. Only 1 patient with confirmed GCA had a GCAPS below 9.5. The Ing score showed a Se of 82.7% and a Sp of 62.7% when using an intermediate probability level (level 3) or more as positive. The BK score showed a Se of 87.9% and a Sp of 71.1% when considering a moderate score (level 2) or more as positive. As for CDUS, a halo count of 1 or more was found to have a Se of 96.6% and a Sp of 97.9%, whereas a total halo score of 2 or more had a Se of 96.6% and a Sp of 84.5%.
The value of a concordant combination between clinical score and CDUS halo count (positive clinical-CDUS combination, or negative clinical-CDUS combination) was assessed to predict or exclude GCA. A GCAPS-CDUS concordant combination showed a Se of 100%, Sp of 97.7%, positive predictive value (PPV) of 96.5% and a negative predictive value (NPV) of 100%. A concordant Ing score-CDUS combination showed a Se of 97.9%, Sp of 97.8%, PPV of 95.9% and NPV of 98.8%. A concordant BK score-CDUS combination showed a Se of 98.0%, Sp of 98%, PPV of 96.2% and a NPV of 99.0%.
Conclusion: In our prospective cohort, we found that the GCAPS had the highest sensitivity, whereas the BK score had the highest specificity. Both scores are easy to calculate and use. The CDUS halo score and halo count both showed high sensitivity and specificity. Using a combination of a clinical score with CDUS halo count provided an accurate GCA prediction method which may be used in the setting of GCA Fast-Track clinics.
Disclosures: C. Sargi, None; S. Ducharme-Benard, None; V. Benard, None; R. Meunier, None; C. Ross, None; J. Makhzoum, None.
