Abstract:

Objective

To investigate the risk factors of he progression of fibrotic connective tissue disease-associated interstitial lung disease （CTD-ILD） and constructing predictive models.

Methods

The baseline clinical data of patients with fibrosing CTD-ILD who were hospitalized in the Department of Pulmonary and Critical Care Medicine of the Fourth，Sixth and Eighth Medical Centers of the PLA General Hospital from January，2019 to December，2023 were collected. The patients were divided into a stable group and a progressive group （PPF group） based on whether progression occurred during follow-up，The variables were preliminarily screened by four different methods including LASSO regression，random forest，gradient boosting machine，and one-way analysis of variance. The screened covariates were incorporated into multivariate stepwise Logistic regression for three analysis to identify the risk factors for the progression of fibrotic. Based on multivariate stepwise regression，a predictive model for the progression of CTD-ILD was constructed and evaluated.

Results

The multivariate stepwise regression results demonstrated that a history of novel coronavirus infection（OR=8.735，95% CI：2.754 ～27.711，P＜0.001），DLCO%pred （OR=0.962，95% CI：0.930 ～0.995，P＜0.001），DLCO%pred ≥60%（OR=0.07，P =0.016），honeycomb shadow on HRCT（OR=7.685，95% CI：2.399～24.613，P＜0.001） ，combined pulmonary hypertension （OR=19.812，95% CI：4.152 ～94.534，P＜0.001）were independently related to the progression of fibrotic CTD-ILD. The area under the ROC curve （AUC） of the predictive model constructed based on stepwise regression was 0.92 in the training set，and the corresponding sensitivity，specificity，and 95%CI were 0.81，0.89，and 0.87～0.97，respectively； the AUC of the predictive model was 0.95 in the test set，and the corresponding sensitivity，specificity，and 95%CI were 0.82，0. 75，and 0. 88 ～1. 00，respectively. These results indicated that the model had an excellent discrimination ability. The calibration curves plotted based on the training set and the test set were basically consistent with the standard curves，which suggested that the model had a favorable calibration ability. The decision curve analysis （DCA） revealed that the model had a higher clinical net benefit within a large threshold in both the training set and the test set，which indicated that the model had favorable clinical utility.

Conclusion

A history of novel coronavirus infection，low level of DLCO%pred at baseline，combined pulmonary hypertension，and honeycomb shadow on HRCT are closely related to the progression of pulmonary fibrosis in patients with fibrotic CTD-ILD. The model constructed based on the above factors can accurately predict the progression of progressive fibrotic CTD-ILD and has certain clinical utility and generalization potency.

Key words: Connective tissue disease, Interstitial lung disease, Pulmonary fibrosis, Progressive pulmonary fibrosis, Risk factors