Autoantibodies to 14-3-3η Improves Discriminative Performance of CRP and HLA-B27 to Differentiate People with Radiographic axSpA from Those with Mechanical Back Pain

Objectives Reducing diagnostic delay for people presenting with back pain but who have axial spondyloarthritis (axSpA) has become a clinical imperative. The absence of a simple blood test that is accessible by referring physicians contributes to this delay. This study evaluates the performance of auto-antibodies to 14-3-3η in combination with existing clinical parameters in an established Edmonton radiographic (r-axSpA) patient cohort. Methods Serum samples from 159 patients with a rheumatologist confirmed diagnosis and classified as r-axSpA (AS-modified New York criteria) were tested on the anti-14-3-3η multiplex assay. The serum differential expression of autoantibodies to 14-3-3η, CRP and HLA-B27 between patients with r-axSpA and MBP was evaluated by ROC AUC. Regression models combining the auto 14-3-3η AAb Score derived from the Bath study were tested in 2 predictive models: 1 with CRP and HLA-B27 markers only and the other adding Age and Sex variables. Corresponding specificity (SP), sensitivity (SN), predictive values (PPV, NPV), likelihood (LR) and diagnostic odds ratios (DxOR) are provided. The Akaike Information Criterion (AIC) statistical measure was reported to compare the relative quality of predictive models. Results Mean age of patients with r-axSpA was 42 (69% male) and for MBP, 30 years (59% male). Differential expressions of each of the biomarkers delivered significant ROC AUCs; 0.70 for autoantibodies to 14-3-3η, 0.81 for CRP and 0.84 for HLA-B27. The latter 2 delivered respectively higher DxOR (6.2, 11.1, 24.1) as expected, considering that they were used in patient diagnosis. The model that included the auto 14-3-3η AAb score with CRP and HLA-B27 added had better discrimination of patients with axSpA vs MBP with an AUC of 0.93 and DxOR of 39.6 that further improved to an AUC of 0.94 and DxOR of 95.3 when age and sex variables were added. As shown below (Table 1), the AIC of the predictive models supported the full model as the best fit. Table 1. Marker and model performance Conclusion This study demonstrates that auto 14-3-3η AAb score when added to currently used CRP and HLA-B27 markers, improves the differentiation of people with MBP from established r-axSpA. While the pretest probability of CRP and HLA-B27 performance is higher in this cross-sectional study than would be expected in an early referral cohort, including 14-3-3η AAb score may reduce diagnostic delay and potentially enable more confident and prompt initiation of therapy to improve outcomes.