Mechanistic advances in the regulation of asthma by different exercise modalities: an integrative perspective on exercise load characteristics and asthma heterogeneity

Asthma is a heterogeneous chronic airway disease in which exercise may act both as a trigger of respiratory symptoms and as a beneficial non-pharmacological intervention. Training responses vary across exercise modalities and asthma phenotypes; inconsistent findings for lung function, airway inflammation, bronchial hyperresponsiveness (BHR), and exercise-induced bronchoconstriction (EIB) highlight the need for a mechanism-oriented framework. This review synthesizes current evidence on moderate-intensity continuous aerobic training, high-intensity interval training, resistance training, combined aerobic-resistance training, breathing exercises, mind-body exercise, and aquatic exercise from the perspective of their dominant ventilatory, metabolic, mechanical-muscular, respiratory-behavioral, and neuropsychological loads. These load characteristics interact with different asthma phenotypes and clinical-mechanistic traits, including type 2-high eosinophilic asthma, obesity-related or non-type 2 asthma, EIB-dominant asthma, severe asthma with physical deconditioning, and dysfunctional breathing or anxiety-related symptoms. Four key mechanistic axes are examined: the ventilatory load-airway epithelium-EIB axis, immune inflammation-BHR axis, skeletal muscle-metabolic-systemic inflammation axis, and respiratory behavior-autonomic regulation-symptom perception axis. Structured exercise training provides the strongest and most consistent benefits for exercise capacity, asthma control, symptoms, and health-related quality of life.,whereas effects on airway inflammation, BHR, lung function, and EIB remain heterogeneous and appear to depend more on baseline phenotype, disease control, environmental exposure, and training protocol. The proposed mechanism-informed framework is intended to support individualized exercise selection, risk stratification, and outcome monitoring rather than serve as a fixed phenotype-specific prescription algorithm.