KEY POINTS
- This retrospective Memorial Sloan Kettering study included 82 patients with abdominal cancers treated using voluntary deep-inspiration breath hold between January and July 2022. Importantly, all patients had successfully tolerated DIBH at simulation within the planned 3-mm gating window.
- Before CT simulation, patients performed three 20-second practice breath holds. Abdominal-wall displacement was monitored with infrared markers, and the resulting breath-hold level was used to define a 3-mm gating window that was carried forward to treatment.
- Despite successful simulation, only 56 of 82 patients (68.3%) completed treatment without modification. 22 patients (26.8%) required a modified DIBH gating window and another 4 (4.9%) had to be re-simulated and treated with free breathing—an overall intolerance rate of 31.7%.
- Breath-hold level at simulation was the only predictor remaining statistically significant after correction for multiple testing (false-discovery rate 4%). Each 1-mm increase in abdominal-wall breath-hold displacement was associated with approximately a 7% increase in intolerance risk, and modeled risk increased sharply beyond 25 mm.
- Patient age, performance status, respiratory or musculoskeletal comorbidity, anxiety and other tested clinical variables were weaker predictors than characteristics of the actual respiratory trace. Non-English preference showed a notable signal toward poorer tolerance but did not reach statistical significance because of the small subgroup.
- A random-forest model combining respiratory-trace and clinical variables achieved only median AUC 0.63 with repeated 10-fold cross-validation. This is insufficient for autonomous patient selection and suggests that the main actionable information may be the breath-hold trace itself rather than the multivariable algorithm.
- In the 15-patient subset with intrafraction fiducial monitoring, residual internal motion was significantly greater among DIBH-intolerant patients (p=0.02). Predicted intolerance also showed a moderate correlation with residual motion (r=0.46), although this did not reach statistical significance (p=0.08).
CLINICAL TAKEAWAY
Passing DIBH at simulation does not guarantee that a patient will tolerate the same gate throughout abdominal radiotherapy: almost one in three needed a change during treatment. Avoiding unnecessarily deep breath holds—particularly abdominal-wall displacement above about 25 mm—may improve tolerability, while the prediction model itself is not accurate enough for clinical decision-making.