Models that include objective brain markers of breathlessness-expectation have the ability to predict, for the first time, which patients will experience clinically important improvements in chronic obstructive pulmonary disease (COPD)-linked breathlessness during pulmonary rehabilitation, according to clinical trial results published in Thorax.
Recognizing that baseline patient characteristics predictive of breathlessness remain unknown, the investigators sought to evaluate functional brain imaging markers of breathlessness-expectation as predictors of therapeutic response to pulmonary rehabilitation. Toward that end, the researchers assessed whether the brain-active agent D-cycloserine — which is known to influence expectation mechanisms — was able to modulate any predictive model.
The investigators conducted a randomized, controlled, double-blind, experimental medicine study (ClinicalTrials.gov identifier: NCT01985750) of D-cycloserine administered during pulmonary rehabilitation. The study evaluated 71 participants (18 female; average age, 71 years [range, 46 to 85 years]) with mild to moderate COPD who were recruited immediately prior to enrollment in a National Health Service-prescribed course of pulmonary rehabilitation. Baseline variables, including brain activity, clinical measures of pulmonary function, responses from self-report questionnaires, and drug allocation, were used to train machine-learning models to predict the outcome — which was a minimally clinically relevant change in the Dyspnea-12 (D-12) score.
Data for the current analysis were obtained at a baseline evaluation occurring at the beginning of a pulmonary rehabilitation course as well as at another evaluation performed upon completion of the pulmonary rehabilitation at 6 to 8 weeks. After the initial visit, the participants were randomly assigned to receive either oral D-cycloserine 250 mg or matched placebo. All of the participants received a single dose on 4 separate occasions 30 minutes prior to the onset of the first 4 pulmonary rehabilitation sessions. At baseline, the median Medical Research Council (MRC) DP-12 breathlessness score of the participants was 3, the median forced expiratory volume in 1 second/forced vital capacity (FEV1/FVC) was 0.55, and the median FEV1 percent predicted was 58.
Based on MRC DP-12 scores, participants were classified as “responders” (ie, those having a score of ≥3) or “nonresponders.” Overall, 41 of 71 participants in the primary dataset were classified as responders (24 in the D-cycloserine group and 17 in the placebo group), whereas 30 were nonresponders (13 in the D-cycloserine group and 17 in the placebo group). No statistically significant interaction between the responders and the nonresponders and the drug was identified with the use of χ2 analysis (P =.21).
We have shown that models including objective brain markers of breathlessness-expectation are able to predict, for the first time, which patients will have clinically important improvements in breathlessness over pulmonary rehabilitation.
Only models that included brain imaging markers of breathlessness-expectation had the ability to successfully predict improvements in D-12 score (sensitivity, 0.88; specificity, 0.77). The use of D-cycloserine was independently associated with improvement in breathlessness. Additionally, models that included questionnaires and clinical measures only did not predict outcomes (sensitivity, 0.68; specificity, 0.20).
A key limitation of the current study is the lack of validation of the model in an external dataset. Models with a large number of measures compared with events are associated with the risk for overfitting and demonstrate poor generalizability to novel datasets.
“We have shown that models including objective brain markers of breathlessness-expectation are able to predict, for the first time, which patients will have clinically important improvements in breathlessness over pulmonary rehabilitation,” the study authors concluded. “Such models could provide new insights into the mechanisms by which breathlessness may be targeted, paving the way for targeted behavioural and pharmacological interventions,” the researchers added.
Disclosure: Some of the study authors have declared affiliations with biotech, pharmaceutical, and/or device companies. Please see the original reference for a full list of authors’ disclosures.