Lasso-derived model for early prediction of systemic sclerosis based on vasculopathy assessment: a population-based cohort study.

Objective: Vasculopathy is a key feature of systemic sclerosis (SSc) and is critical for its diagnosis and prognosis. We aimed to comprehensively analyze vascular lesions in patients with SSc and establish an early diagnostic model based on these lesions.

Methods: We recruited 111 participants (45 healthy participants and 66 patients with SSc, mean age 49.75 ± 12.902 years). Age, sex, blood pressure, hand grip strength, skin thickness, proper palmar digital artery vascular index, skin blood flow index, and nailfold microcirculation were compared between the two groups. Applying Lasso regression for variable selection, we further developed a binary logistic regression model to analyze the diagnostic differences in disease occurrence based on the vascular injury status. We assessed the performance of the model using receiver operating characteristic (ROC) and calibration curve analyses to evaluate its diagnostic ability and determine the optimal cutoff value.

Results: Using Lasso regression analysis, we identified 10 key variables from 37 microcirculation parameters, including age, left hand grip strength, left peak systolic velocity (PSV), right PSV, right resistance index (RI), ischemic perfusion (IPU), ischemic reperfusion perfusion (IRPU), post-occlusive reactive hyperemia baseline (PORH BL), loop top length, and nailfold video-capillaroscopy (NVC) score. Among these, the NVC score (Cut-off point = 5.35, AUC = 0.845, SEN = 0.74, SPE = 0.87), PSV (Cut-off point = 11.38, AUC = 0.838, SEN = 0.82, SPE = 0.73), IRPU (Cut-off point = 111.3, AUC = 0.831, SEN = 0.61, SPE = 0.91), and Grip (Cut-off point = 22.8, AUC = 0.781, SEN = 0.79, SPE = 0.62) demonstrated high diagnostic value in predicting SSc. The binary logistic regression model based on these variables provides better interpretability for advanced diagnosis of microcirculation multidimensional integration. Compared to the scleroderma pattern model, this model exhibited superior performance, with an area under the curve of 0.929 (95% CI: 0.883-0.974).

Conclusion: Our results highlight the key role of the nailfold video-capillaroscopy score, grip strength, and peak systolic flow velocity of the proper palmar digital artery in predicting systemic sclerosis events.