With the aid of a computational tool, researchers say they have identified key phenotypes among patients with severe asthma that can help predict who is likely to benefit and not benefit from treatment with systemic corticosteroids (CS).
Using a machine learning algorithm, the researchers examined around 100 variables among adult patients enrolled in the National Heart, Lung, and Blood Institute-funded Severe Asthma Research Program (SARP).
Wei Wu, PhD, a computational biologist at Carnegie Mellon University, who co-developed the computer algorithm, told MedPage Today that approximately a dozen variables were associated with systemic CS response.
Age at asthma onset, patient weight, race, and scores on a quality-of-life questionnaire were among the variables used to predict individual patient response in the computational tool.
For the study, Wu and colleagues validated the approach in 346 adult asthma patients participating in SARP. The study was published online in the American Journal of Respiratory and Critical Care Medicine.
In a press statement, co-author Sally Wenzel, MD, of the University of Pittsburgh Medical Center Asthma Institute, said that predicting how patients with severe asthma will respond to systemic CS therapies is an important, unmet medical challenge.
“I see so many patients in my clinic who have been ravaged by the side effects of corticosteroids. Weight gain, extreme emotions, inability to sleep, glaucoma, and thinning of the skin are among the possible side effects of corticosteroid pills and injections, so physicians would like to prescribe them only to patients they know will benefit from them,” she said.
The researchers set out to come up with an integrated, cluster approach to identify differential responses to systemic CS approaches using an unsupervised multi-view learning approach.
Wu explained that multiple kernel K-means clustering was applied to 100 clinical physiologic, inflammatory, and demographic characteristics from the 346 study participants included in the analysis. All had paired pre and 2- to 3-week post-triamcinolone sputum data. Machine learning techniques selected the main baseline variables that predicted cluster assignment.
Multiple kernel clustering revealed the following four asthma clusters with different CS responses:
Clusters 1 and 2 were younger patients who were modestly CS-responsive allergic asthmatics with relatively normal lung function, separated by contrasting sputum neutrophil and macrophage percentages post steroids
Cluster 3 identified late-onset asthmatics with low lung function, high baseline eosinophilia, and the most CS responsiveness
Cluster 4 was primarily comprised of young obese females with severe airflow limitation and little eosinophilic inflammation; these patients were the least responsive to systemic CSs
“Of the four clusters identified, only one cluster, Cluster 3 (28% of the overall population), would be widely recognized as highly CS responsive,” the researchers wrote. “Not surprisingly, patients in this cluster had the highest baseline eosinophilia, markedly obstructed lung function, and the most nasal polyps. They also had the highest value of [T2 gene mean] by post hoc analysis.”
Despite their older age and later age at asthma onset, these patients had the greatest improvement in obstruction and inflammation, the team reported.
Clusters 1 and 2 were only modestly responsive with no improvement in Asthma Control Questionnaire 6 scores — most likely because the subjects in these clusters had near normal baselines.
Wu, Wenzel, and colleagues noted that it would not be possible to predict how responsive these patients might be during an acute episode of worsening, when their lung function would likely be lower.
“On the other hand, this analysis identified Cluster 4 as the least CS responsive, despite a rather severe baseline airflow limitation,” the team wrote, further noting that after treatment, patients in Cluster 4 were still worse than other clusters, “with almost no change in lung function, and even a small decrease in maximal (post albuterol) lung function, particularly in the change in [forced vital capacity].”
The researchers noted that while the reason for this was not clear, both the discovery and test set were consistent. “Thus, CS responses in Cluster 4 are likely complex. In these patients, CSs could detrimentally ‘stiffen the lungs/airways’ to decrease bronchodilator responses,” they suggested. “Although not typically thought of in relation to asthma, CSs can increase stiffness of extracellular matrix, particularly in the eye in relation to CS-induced glaucoma. A similar effect could occur in airway matrix, which decreases bronchodilator responses in susceptible patients. While further confirmation is needed, the implications could be important.”
SARP is funded by the National Heart, Lung, and Blood Institute.
The researchers declared no relevant relationships with industry related to this study.