CT-Based Biomarkers for Predicting Progressive Phenotypes in Interstitial Lung Diseases: A Retrospective Cohort Study

Prediction of the progressive phenotype from baseline evaluation is challenging due to the variable clinical course of fibrosing interstitial lung disease (ILD). The aim of this study was to evaluate whether the Single Time Point Prediction (STP) score predicts disease progression better than other quantitative scores.

This is a retrospective two-centre study including patients with ILD other than idiopathic pulmonary fibrosis (IPF). Using an automated quantification system, quantitative lung fibrosis (QLF), quantitative ground-glass opacity (QGG), quantitative ILD (QILD; the sum of reticulation, honeycombing, and ground-glass opacity), quantitative normal lung (QNL), and STP scores were measured. Disease progression (DP) was defined as an absolute decline in forced vital capacity (FVC) ≥ 5% or DL_CO ≥ 10% predicted, death or lung transplantation.

Of 245 patients, 137 (55.9%) progressed during follow-up (median: 20.4 months). In the Kaplan-Meier analysis, patients with high QGG (≥ 10%), high QILD (≥ 30%), and high STP (≥ 30%) showed shorter DP free survival time than those without. After adjusting for age, sex, FVC, and types of non-IPF ILD, QGG scores ≥ 10%, QILD ≥ 30%, and STP ≥ 30% remained significant predictors of progression (hazard ratio [HR] = 1.71, p = 0.024; HR = 1.73, p = 0.006; and HR = 1.58, p = 0.020, respectively). Furthermore, higher QILD and QGG scores in STP-positive regions were independently associated with increased DP risk, whereas higher QNL in STP-negative regions was associated with lower DP risk.

The STP score is a useful imaging biomarker for predicting DP and, when combined with conventional quantitative CT scores, may improve risk stratification in patients with non-IPF ILD.