Elsevier

Academic Radiology

Volume 22, Issue 1, January 2015, Pages 70-80
Academic Radiology

Original Investigation
Comparison of the Quantitative CT Imaging Biomarkers of Idiopathic Pulmonary Fibrosis at Baseline and Early Change with an Interval of 7 Months

https://doi.org/10.1016/j.acra.2014.08.004Get rights and content

Rationale and Objectives

Median survival of patients with idiopathic pulmonary fibrosis (IPF) is 2–5 years. Sensitive imaging metrics can play a role in detecting early changes in therapeutic development. The aim of the present study was to compare known computed tomography (CT) histogram kurtosis and a classifier-based quantitative score to assess baseline severity and change over time in patients with IPF.

Materials and Methods

A total of 57 patients with at least baseline and paired follow-up scans were selected from an imaging database of standardized CT scans obtained from patients with IPF. CT histogram measurement of kurtosis and quantitative lung fibrosis (QLF) and quantitative interstitial lung disease (QILD) scores from a classification algorithm were calculated. Spearman rank correlations were used to assess associations between baseline severity and changes for all CT-derived measures compared to forced vital capacity (FVC) and carbon monoxide diffusion capacity (DLCO) (percent predicted).

Results

At baseline, mean (±SD) of kurtosis was 2.43 (±1.83). Mean (±SD) values of QLF and QILD scores were 20.7% (±13.4) and 43.3% (±20.0), respectively. All baseline histogram indices and QLF and QILD scores were correlated well with baseline FVC and DLCO. When assessing associations with changes in FVC and DLCO over time, only QLF score was statistically significant (ρ = −0.57; P < .0001 for FVC and ρ = −0.34; P = .025 for DLCO), whereas kurtosis was not.

Conclusions

Classifier-model-derived scores (QLF and QILD), based on a set of texture features, are associated with baseline disease extent and are also a sensitive measure of change over time. A QLF score can be used for measuring the extent of disease severity and longitudinal changes.

Section snippets

Patient Selection

A total of 57 subjects (45 males, 12 females, with mean FVC 65 ± 12%, DLCO 50 ± 14%) were selected from an imaging database of standardized CT scans with at least two scans obtained from patients with well-characterized IPF whose baseline scan dates ranged from January 2011 to September 2012. Diagnosis of IPF was based on the American Thoracic Society, European Respiratory Society, and Latin American Thoracic Association statement and was confirmed by HRCT and/or surgical lung biopsy (10). Of

Descriptive Summary Statistics

Mean duration between two CT scans was 7 months with standard deviation of 1.8 months. Table 1 lists mean, standard deviation, median, and interquartile range at baseline and follow-up of all three types of measures: PFT, CT histogram indices, and quantitative CT texture scores. It is recognized that most subjects demonstrate either stable or progressive disease over the period of study (mean of 7 months). Most of the measures showed statistically significant changes between baseline and

Discussion

For a quantitative imaging biomarker to be used as a predictive or surrogate outcome it must be shown to be robust enough to be implemented at multiple sites, reproducible, and related to clinically meaningful outcomes 36, 37. Quantitative CT has been used to measure the amount of lung involvement in patients with diffuse lung disease including nonspecific interstitial pneumonia and IPF. In patients with IPF the disease can vary from insidious and stable over time to being rapidly progressive

References (43)

  • T.E. King et al.

    Idiopathic pulmonary fibrosis: relationship between histopathologic features and mortality

    Am J Respir Crit Care Med

    (2001)
  • B. Ley et al.

    A multidimensional index and staging system for idiopathic pulmonary fibrosis

    Ann Intern Med

    (2012)
  • R.M. du Bois et al.

    Ascertainment of individual risk of mortality for patients with idiopathic pulmonary fibrosis

    Am J Respir Crit Care Med

    (2011)
  • A.U. Wells et al.

    Idiopathic pulmonary fibrosis: a composite physiologic index derived from disease extent observed by computed tomography

    Am J Respir Crit Care Med

    (2003)
  • P.I. Latsi et al.

    Fibrotic idiopathic interstitial pneumonia: the prognostic value of longitudinal functional trends

    Am J Respir Crit Care Med

    (2003)
  • G. Raghu et al.

    An official ATS/ERS/JRS/ALAT statement: idiopathic pulmonary fibrosis: evidence-based guidelines for diagnosis and management

    Am J Respir Crit Care Med

    (2011)
  • P.A. Hodnett et al.

    Fibrosing interstitial lung disease. A practical high-resolution computed tomography-based approach to diagnosis and management and a review of the literature

    Am J Respir Crit Care Med

    (2013)
  • G. Raghu et al.

    Idiopathic pulmonary fibrosis: clinically meaningful primary endpoints in phase 3 clinical trials

    Am J Respir Crit Care Med

    (2012)
  • V. Cottin

    Interstitial lung disease

    Eur Respir Rev

    (2013)
  • H.V. Woodcock et al.

    Reducing lung function decline in patients with idiopathic pulmonary fibrosis: potential of nintedanib

    Drug Des Devel Ther

    (2013)
  • H.I. Adamali et al.

    Current and novel drug therapies for idiopathic pulmonary fibrosis

    Drug Des Devel Ther

    (2012)
  • Cited by (95)

    • A Convolutional Neural Network Approach to Quantify Lung Disease Progression in Patients with Fibrotic Hypersensitivity Pneumonitis (HP)

      2022, Academic Radiology
      Citation Excerpt :

      Indexes derived from frequency histograms of lung CT scans are quantitative and reproducible features, which have shown sensitivity to disease severity and progression (12,13). Analysis of the lung parenchyma based on machine learning and deep learning techniques has rapidly advanced in recent years, providing characterization of lung texture and the relative contribution of different disease patterns (14–21). In particular, deep learning approaches for ILD pattern recognition and classification have achieved impressive results (22,23).

    • Explainability of radiomics through formal methods

      2022, Computer Methods and Programs in Biomedicine
    View all citing articles on Scopus

    Funding information: NIH NHLBI R01 HL 072424.

    View full text