To explore the clinical biomarkers that predict therapeutic response to bronchial thermoplasty (BT) in severe asthma. We prospectively recruited patients with severe asthma who completed three sessions of bronchial thermoplasty. Baseline demographics, Asthma Control Questionnaire-5 (ACQ5) scores, eosinophil counts, fractional exhaled nitric oxide (FeNO), spirometry, impulse oscillometry (IOS), endobronchial optical coherence tomography (EB-OCT), and BT activation counts were recorded. All subjects were followed for two years and classified as responders or non-responders according to ACQ5 improvement ≥0.5 points. Thirty patients were included (22 responders, 8 non-responders). Compared with non-responders, responders had lower body weight, BMI, and triglycerides, along with more negative X5 values, higher RV/TLC ratios, greater Collagen Type III (COL3) expression, and larger airway luminal areas with thinner airway walls on EB-OCT. Receiver operating characteristic (ROC) analysis demonstrated that body weight (AUC = 0.759), BMI (AUC = 0.733), triglycerides (AUC = 0.694), X5 (AUC = 0.938), RV/TLC (AUC = 0.756), COL3 (AUC = 0.846), and EB-OCT indices including airway luminal area from the 3rd to 6th generation (Ai3-6), 7th to 9th generation (Ai7-9), and airway wall area percentage from the 3rd to 6th generation (Aw%3-6) showed moderate-to-good discriminatory power (AUC range: 0.761-0.830). Multivariable logistic model integrating BMI, X5, and Ai3-6 achieved better discrimination (AUC = 0.988) in predicting response to BT. More negative baseline X5, lower triglyceride levels, EB-OCT-derived thinner airway walls and larger luminal areas, and higher COL3 expression, but not BT activation number, may help identify asthma patients most likely to benefit from BT and serve as potential predictors of its long-term efficacy.