The assessment of computed tomography (CT) through complex computer techniques can aid in both the objective quantification of lung fibrosis and outcome prediction in patients with idiopathic pulmonary fibrosis (IPF), according to a retrospective analysis recently published in Respirology.
Although traditional CT is an integral part of IPF assessment, it is subject to interobserver variation. Thus, there has been an interest in developing objective computer technology capable of detecting and quantifying lung fibrosis without interobserver bias.
Data-driven texture analysis (DTA) is one of these techniques. Evidence suggests that its measures of lung fibrosis directly correlate with lung function.
Read more about IPF prognosis
“We hypothesized that quantitative CT using a deep learning approach can stratify disease severity in patients with IPF,” the authors wrote.
The authors aimed to assess the prognostic value of DTA in IPF through a retrospective analysis that included 393 patients with IPF and used Cox analyses and linear mixed-effect modeling to correlate baseline DTA with outcomes.
Kaplan–Meier plots were constructed to evaluate transplant-free survival and progression-free survival in the DTA-stratified groups. Forced vital capacity (FVC), diffusing capacity of carbon monoxide (DLCO), and the composite physiologic index were compared with CT-derived metrics using Spearman’s rank correlation. The median follow-up was 2.7 years.
Over 30% of the patient population had an FVC% predicted of 80% or greater obtained within 90 days of baseline CT, with a median DTA score of approximately 30. The median age of the patients in the study was close to 70 years.
The results showed that, according to linear effect modeling, greater DTA scores were directly correlated with a greater decline in both FVC and DLCO. Patients with higher DTA scores also had an increased risk of mortality. As expected, Cox analyses revealed that a greater extent of fibrosis was linked with shorter transplant-free survival.
“In conclusion, these analyses in the [Australian Idiopathic Pulmonary Fibrosis Registry] show important associations between morphologic extent of pulmonary fibrosis, measured objectively on CT using a deep learning algorithm, and outcomes, independent of pulmonary function,” the authors wrote.
Humphries SM, Mackintosh JA, Jo HE, et al. Quantitative computed tomography predicts outcomes in idiopathic pulmonary fibrosis. Respirology. Published online July 25, 2022. doi:10.1111/resp.14333