Abstract

Background: GCA is a critically ischemic disease with protean manifestations that require urgent diagnosis and treatment. It is vital to make a secure diagnosis urgently, not only to confirm GCA but also to exclude GCA mimics. The Southend GCA Probability Score (SGCAPS) stratifies patients into low, intermediate, and high-risk categories based on demographics, symptoms, physical signs, and C-reactive protein levels in the blood[1,2]. EULAR recommendations advocate ultrasonography as the first investigation for suspected GCA[3]. Objectives: We developed a GCA prediction tool that sequentially combines clinical assessment, as determined by the SGCAPS, with the results of quantitative ultrasonography. Methods: HAS GCA (IRAS#264294) is a prospective, European multi-centre, inception cohort study. The final clinical diagnosis was established at six months. SGCAPS and quantitative ultrasonography of temporal and axillary arteries, i.e. Halo count[4], Halo Score[4], OMERACT GCA Score (OGUS)[5], were performed at diagnosis. Prediction models for GCA diagnosis were developed by multivariable logistic regression analysis with SGCAPS and each of the three ultrasonographic scores as predicting variables. Intraclass correlation coefficients for inter- and intra-rater reliability were obtained in a separate patient-based reliability exercise with 5 patients and 5 observers. Results: 229 unselected consecutive patients with suspected new-onset GCA were referred to fast-track clinics: 84 GCA versus 145 GCA mimics (controls) were recruited. SGCAPS and all three ultrasonographic scores discriminated well between GCA and non-GCA patients. SGCAPS was higher in GCA vs non-GCA, confirmed by the ROC analysis with an AUC of 0.918. the number of patients with confirmed GCA at 6 months was 0 (0%) in low-risk (LRC), 17 (24%) in intermediate (IRC) and 67 (74%) in high-risk (HRC) patients (Figure 1). A reliability exercise showed that the inter-rater and intra-rater reliability was high for all three ultrasonographic scores. The prediction model combining SGCAPS with the Halo count, termed HAS-GCA Score, was the most accurate model, with a C statistic of 0.969 (95%CI 0.952 to 0.990). The HAS-GCA Score could classify 169/229 (74%) patients into either the low or high-probability groups, with misclassification observed in 2/105 (2%) and 2/64 (3%) of these patients, respectively (Figure 2). A nomogram was created to easily apply the score in daily practice. Conclusion: A prediction tool for GCA (HAS-GCA score), combining SGCAPS and the Halo count, reliably confirms/excludes GCA from GCA mimics in fast-track clinics. These findings require confirmation in an independent, multi-centre study. REFERENCES: [1] Laskou et al. Clin Exp Rheumatol 2019. [2] Sebastian et al. RMD Open 2020. [3] Dejaco et al. ARD 2023. [4] Van Der Geest et al. ARD 2020. [5] Dejaco et al. ARD 2022. Acknowledgements: We thank Dr. Suzanne Arends for her statistical advice on the logistic regression bootstrapping and reliability analyses. We thank Fiona Coath for her initial involvement in the study design and participation in the reliability exercise. We thank Abdul Kayani and Mohammad Tariq for their clinical support. We thank Sue Inness and Jo Jackson for their advice on the study design. Disclosure of Interests: Alwin Sebastian: None declared, Kornelis van der Geest Roche, AbbVie, Alessandro Tomelleri: None declared, Pierluigi Macchioni: None declared, Giulia Klinowski: None declared, Carlo Salvarani: None declared, Diana Prieto-Peña Lilly, Novartis, AbbVie, Amgen, Merck Sharp & Dohme, Edoardo Conticini GSK and Chiesi., Muhammad Khurshid: None declared, Lorenzo Dagna: None declared, Elisabeth Brouwer Mark, Bhaskar Dasgupta Novartis, Abbvie, Sanofi.