Quantitative CT shows prospect in high risk COPD identification: Lancet

It is possible to identify chronic obstructive pulmonary disease (COPD) patients who are at a high probability of severe exacerbations using CT-based prediction models, says an article published in The Lancet Digital Health. 

Quantitative CT is being used to characterize lung illness; however, little is known about its further potential as a clinical tool for foretelling severe exacerbations. In order to build and evaluate quantitative CT-based models for foretelling severe chronic obstructive pulmonary disease exacerbations, Muhammad Chaudhary and colleagues carried out this work.

Researchers examined the SPIROMICS cohort, a multicenter study conducted at 12 clinical sites across the USA, which included people aged 40 to 80 years old from four strata: those who had never smoked, those who had smoked but had normal spirometry, those who had smoked and had mild to moderate COPD, and those who had smoked and had severe COPD. To create logistic regression classifiers for predicting severe exacerbations, 3-year follow-up data was employed. In addition to smoking status and CT-based assessments of density gradient texture and airway structure, predictors included sex, age, BMI, race, exacerbation history, pulmonary function, respiratory quality of life, and exacerbation history. 

Researchers validated their models in a subset of the COPD Genetic Epidemiology (COPDGene) cohort. The area under the receiver operating characteristic curve (AUC) was used to measure discriminative model performance, which was also compared with additional predictors such as exacerbation history and the BMI, dyspnoea, airflow obstruction, and exercise capacity (BODE) index. They used calibration plots and Brier ratings to assess model calibration.

The key findings of this study were:

SPIROMICS participants were enrolled between November 12, 2010 and July 31, 2015. 

COPDGene participants were registered between January 10, 2008, and April 15, 2011. 

The SPIROMICS cohort contained 1956 people with complete 3-year follow-up data: the group's mean age was 631 years (SD 92), and 1017 (52%) were males and 939 (48%) were women. 

434 (22%) of the 1956 individuals had a history of at least one major exacerbation. 

The AUC for at least one serious exacerbation within three years was 0854 for CT-based models, and 0931 for persistent exacerbations.

With low Brier values, the models were well calibrated. 

CT biomarkers had considerably higher AUCs than exacerbation history and the BODE index 0812 for predicting at least one severe episode during a 3-year follow-up. 

The external validation cohort includes 6965 persons with a mean age of 605 years (SD 89). The AUC for at least one severe exacerbation in this population was 0768.

Reference: 

Chaudhary, M. F. A., Hoffman, E. A., Guo, J., Comellas, A. P., Newell, J. D., Jr, Nagpal, P., Fortis, S., Christensen, G. E., Gerard, S. E., Pan, Y., Wang, D., Abtin, F., Barjaktarevic, I. Z., Barr, R. G., Bhatt, S. P., Bodduluri, S., Cooper, C. B., Gravens-Mueller, L., Han, M. K., … Reinhardt, J. M. (2023). Predicting severe chronic obstructive pulmonary disease exacerbations using quantitative CT: a retrospective model development and external validation study. In The Lancet Digital Health (Vol. 5, Issue 2, pp. e83–e92). Elsevier BV. https://doi.org/10.1016/s2589-7500(22)00232-1