Construction and internal cohort verification of clinical-imaging-based nomogram for early diagnosis in Takayasu arteritis

TL;DR

This study created a predictive model combining clinical and imaging data to help diagnose Takayasu arteritis early.

Contribution

A new nomogram model integrating clinical, imaging, and radiomics features for early diagnosis of Takayasu arteritis was developed and validated.

Findings

The nomogram model achieved a C-index of 0.767 in the training set and 0.733 in the validation set.

Key predictors included vascular wall thickness, ESR, CRP, and radiomic features like uniformity and energy.

Abstract

Takayasu arteritis (TAK) is a chronic large-vessel vasculitis. This study aimed to develop and internally validate a nomogram model integrating clinical indicators, conventional imaging features, and radiomics features for the early diagnosis of TAK.

A total of 356 patients suspected of having TAK in our hospital were retrospectively included. They were randomly divided into a training set (n = 249) and a validation set (n = 107) in a ratio of 7:3. In the training set, Lasso regression was used to screen the influencing factors associated with TAK, and a Nomogram prediction model was constructed. The predictive efficacy and clinical application value of the model were evaluated by receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis (DCA).

In the training set, 75 cases (30.12%) were diagnosed with early-stage TAK, and in the validation set, 32 cases (29.91%) were diagnosed. There were no statistically significant differences in the incidence of TAK and clinical characteristics between the two groups (p > 0.05). In the training set, multivariate logistic regression identified the following independent predictors for early-stage TAK: intermittent claudication of the limbs, vascular murmur, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), the thickest part of the vascular wall, degree of vascular wall enhancement, and contrast were identified as risk factors (all OR > 1), while uniformity and energy was identified as a protective factor (OR < 1) (all p < 0.05). The C-index was 0.767 and 0.733, respectively. The mean absolute errors of the agreement between the predicted and actual values were 0.163 and 0.180, respectively. The results of the Hosmer-Lemeshow test were χ2 = 7.937, p = 0.440 and χ2 = 11.924, p = 0.155, respectively. The ROC curve showed that the areas under the curve (AUC) of the nomogram model for predicting whether patients were diagnosed with TAK in the early-stage diagnosis in the training set and validation set were 0.767 (95% CI: 0.684–0.850) and 0.733 (95% CI: 0.616–0.849) respectively, with sensitivities and specificities of 0.847, 0.660 and 0.720, 0.500, respectively.

This study successfully constructed and validated a comprehensive nomogram model, which can provide individualized and non-invasive risk assessment for the early diagnosis of TAK and contribute to clinical decision-making.

Figures