Predictive accuracy of the ASIG and DETECT screening algorithms to identify pulmonary arterial hypertension in systemic sclerosis; a retrospective cohort study

Abstract Objective To evaluate the predictive accuracy of the Australian Scleroderma Interest Group (ASIG) algorithm and the DETECT algorithm to identify pulmonary arterial hypertension (PAH) in an Australian systemic sclerosis (SSc) cohort. Methods Algorithm performance evaluated using prospectively collected data from a cohort with annual serum NT-proBNP, TTE and RFT. Results From 2009 - 2022, 243 patients had sufficient data to apply both algorithms; 52 underwent right heart catheterisation (RHC), 33 were diagnosed with PAH (13.6%). The ASIG algorithm was positive in 105 patients (43.21%), 29 were diagnosed with PAH (27.62%). The DETECT algorithm was positive in 172 patients (70.78%), 30 were diagnosed with PAH (17.44%). The sensitivity of the ASIG algorithm for PAH was 87.88% (95% CI 71.80 – 96.60) and specificity was 31.58% (95% CI 12.58 – 56.55); the DETECT algorithm sensitivity was 90.91% (95% CI 75.67 – 98.08). Specificity and negative predictive value (NPV) of the DETECT algorithm were not able to be calculated as there were no true negative DETECT screen cases. The positive predictive value (PPV) of the ASIG algorithm was 69.05% (95% CI 52.91 – 82.38) and NPV was 60.00% (95% CI 26.24 – 87.84); DETECT algorithm PPV was 61.22% (95% CI 46.24 – 74.80). Three cases of PAH were not identified by either algorithm; all had co-existent interstitial lung disease (ILD). Conclusions ASIG and DETECT screening algorithms for PAH perform well in SSc. A small number of false negative screens in patients with ILD highlight the need to interpret PAH screens in SSc-ILD patients with caution.