Patient Care brings primary care clinicians a lot of medical news every day—it’s easy to miss an important study. The Daily Dose provides a concise summary of one of the website's leading stories you may not have seen.
On February 13, 2023, we reported on a study published in The Lancet Digital Health that aimed to develop and validate quantitative CT-based models for predicting severe chronic obstructive pulmonary disease (COPD) exacerbations.
To develop predictive models the research team drew on longitudinal data from 2 large well characterized multicenter study cohorts. From the ongoing Subpopulations and Intermediate Outcome Measures in COPD Study (SPIROMICS), the team analyzed individuals aged 40 to 80 years from 4 strata: those who never smoked, those who smoked but had normal spirometry, those who smoked and had mild-moderate COPD, and those who smoked and had severe COPD. They used 3-year follow-up data from SPIROMICS participants who had results from high resolution CT scans at total lung capacity to develop prediction models for severe exacerbations: age, sex, race, body mass index (BMI), forced expiratory volume in 1 second (FEV1), exacerbation history, smoking status, respiratory quality of life, and 2 CT-based measures of lung tissue: density gradient texture (CTDG) and airway structure (Pi10).
For external validation of the predictive models, Chaudhary et al use a subset of participants from the Genetic Epidemiology of COPD (COPDGene) cohort. Discriminative performance of each model was assessed using the area under the receiver operating characteristic curve (AUC). AUC was also compared with predictors including exacerbation history and the BODE (BMI, airflow obstruction, dyspnea, and exercise capacity) index. Calibration plots and Brier scores were used to evaluate model calibration.
For the CT biomarker-based models (ie, those including values for Pi10 and CTDG texture), the AUC was 0.854 (95% CI 0.852–0.855) for at least one severe exacerbation within 3 years and 0.931 (0.930–0.933) for consistent exacerbations (≥1 acute episode in each of the 3 years). Brier scores were low for all models. When the investigators compared the predictive quality of the CT biomarkers vs exacerbation history and BODE index for ≥1 severe event during the 3-year follow up, AUCs were significantly higher with CT biomarkers (0.854, 0.852–0.855) than with exacerbation history (3% difference; CT biomarkers 0.854 vs exacerbation history 0.823; P<.001) and BODE index (4% difference; CT biomarkers 0.854 vs BODE index 0.812; P<.001). Similar differences were observed in the external COPDGene validation cohort.
Note from authors
"Our study highlights the clinical utility and effectiveness of CT biomarkers for predicting severe and persistent exacerbations of COPD. Care providers can use the CT-based prediction models presented in this study to identify individuals at a higher risk of hospital admission or visits to the emergency department. Furthermore, these models can provide risk estimates of recurrent exacerbations that are another major source of burden associated with COPD. Unlike the highly variable previous exacerbation history, CT-based models can identify individuals who are at risk of a severe episode without necessarily having a history of exacerbations."