Prediction of high risk of risk of cardiovascular complications in patients with ankylosing spondylitis

Rationale . In recent years, there has been increasing evidence of a high incidence of arterial hypertension (AH) among patients with ankylosing spondylitis (AS). Timely diagnosis and prevention require the development of methods for predicting the development of AH in patients with AS. Objective . To develop prognostic criteria for the development of AH in patients with AS using neural network technology. Material and methods . A total of 202 people were examined. Group 1 included 28 people with AH, Group 2 included 60 patients with AS without AH, Group 3 included 45 patients with AS and AH, and Group 4 included 69 people in the control group. The subjects underwent 24-hour blood pressure monitoring, complete blood count, determination of C-reactive protein and cholesterol, and a study of gene polymorphism: AGT (T704C), AGT (C521T); ACE (I/D); NOS3 (T789C), AGTR1 (A1166C). Results . As a result of the conducted logistic analysis, six parameters were identified that formed the basis of the model for predicting the high risk of hypertension in patients with AS. The structure of the multilayer perceptron included indicators of the presence of a burdened heredity for hypertension, carriage of the AGT and ACE gene polymorphisms, AS activity, the level of C-reactive protein and cholesterol. The accuracy of the model prediction was 95.2%, sensitivity — 88.9%, area under the curve (AUC) = 0.959 [95% CI 0.893–0.988], p < 0.001). To facilitate the use of the model, an online calculator was developed to identify the high risk of hypertension in patients with AS. Conclusion . The model for predicting the development of hypertension in patients with AS is highly informative and functional.