To investigate the relationship between the expression of serum long non-coding RNA growth arrest specific 5 (LncRNA-GAS5) and microRNA-221-3p (miR-221-3p) and the pulmonary hypertension (PH) in patients with idiopathic pulmonary fibrosis (IPF).

A total of 137 IPF patients (IPF group) and 70 healthy volunteers (control group) were enrolled at the Xi'an Hospital of Traditional Chinese Medicine from January 2021 to January 2023. According to the presence of PH, the IPF patients were divided into a PH group (n = 42) and a non-PH group (n = 95). Serum LncRNA-GAS5 and miR-221-3p expression levels were measured using quantitative real-time polymerase chain reaction (qPCR). Potential binding sites between LncRNA-GAS5 and miR-221-3p were predicted using the StarBase database, and Pearson correlation analysis assessed their association. Multivariate logistic regression was applied to identify independent factors influencing PH in IPF patients and to construct a predictive model. Receiver operating characteristic (ROC) curve analysis evaluated the predictive value of individual factors and the model.

Compared with the controls, IPF patients exhibited lower serum LncRNA-GAS5 and higher miR-221-3p expression (t = 6.489, 8.962; both P < 0.001). The incidence of PH in IPF patients was 30.66% (42/137). PH patients had significantly lower LncRNA-GAS5 and higher miR-221-3p levels compared with non-PH patients (t = 5.553, 5.694; both P < 0.001). LncRNA-GAS5 and miR-221-3p had binding sites and were negatively correlated in the serum of IPF patients (r = -0.788, P < 0.001). Multivariate logistic regression indicated that higher diffusion lung capacity of carbon monoxide as a percentage of predicted value (DLco%pred) [odds ratio (OR) = 0.864, 95% confidence interval (CI) (0.796, 0.938), P = 0.001] and LncRNA-GAS5 expression [OR = 0.463, 95%CI (0.282, 0.759), P = 0.002] were independent protective factors, while elevated B-type natriuretic peptide (BNP) [OR = 1.020, 95%CI (1.009, 1.032), P < 0.001] and miR-221-3p [OR = 2.269, 95%CI (1.423, 3.617), P = 0.001] were independent risk factors for PH in IPF patients. The predictive model equation was: Logit(P) = -1.518 - 0.042 × DLco%pred + 0.019 × BNP - 0.641 × LncRNA-GAS5 + 0.637 × miR-221-3p. The Hosmer-Lemeshow test indicated good model fit (χ² = 7.316, P = 0.412). ROC curve analysis showed that the model had an area under the curve (AUC) of 0.909, superior to DLco%pred (0.708), BNP (0.744), LncRNA-GAS5 (0.779), and miR-221-3p (0.773) alone (Z = 3.899, 3.633, 2.653, 2.760; all P < 0.001).

Low serum LncRNA-GAS5 and high miR-221-3p expression are closely associated with PH in IPF patients. The predictive model integrating LncRNA-GAS5, miR-221-3p, DLco%pred, and BNP demonstrates good predictive value for PH in this population.