Assessment of Upfront Selection Criteria to Prioritise Patients for Breath-hold Left-sided Breast Radiotherapy

Aims: Deep inspiratory breath-hold (DIBH) techniques for left breast and chest wall radiotherapy can reduce cardiac dose. We investigated the use of 'upfront selection' criteria for DIBH based on tumour bed position and whether cardiac shielding was used. Materials and methods: Four methods of selecting patients for DIBH were assessed retrospectively in a cohort of left breast and chest wall treatments. These were: (1) free breathing scan on all patients, selecting DIBH treatment for those with a predicted mean heart dose >= 3 Gy; (2) selective DIBH for those with maximum heart depth (MHD) on free breathing scan >= 1 cm; (3) use of an 'upfront selection process' using tumour bed position as initial selection and measurement of MHD on those not selected upfront; (4) DIBH on all. The methods were assessed on predicted mean heart dose, proportion needing two scans, sensitivity, specificity and the positive and negative predictive values. These were compared with method (1) as the gold standard. Results: In total 134 cases were analysed. The predicted mean heart dose in free breathing was >= 3 Gy in 28 (20.9%). Therefore, applying method (1), 28/134 (20.9%) would be selected for DIBH treatment. Applying method (2), 66/134 (49.2%) would be selected for DIBH treatment, all requiring two scans. Of these, 40/66 (60.6%) would receive <3 Gy in free breathing so are over-selected; 2/68 (2.9%) would have received >3 Gy in free breathing so failed to be selected. Selection using method (3) was similar to method (2), but only five patients required two planning scans; 61/134 (45.5%) cases would be selected for DIBH upfront and 5/134 (3.7%) after initial free breathing scan; 42/66 (63.6%) of those selected for DIBH treatment would receive <3 Gy in free breathing and 4/68 not selected (6%) would receive >3 Gy in free breathing. For methods (2) and (3) most patients not selected for DIBH would have had a mean heart dose of >= 3 Gy (64/68, 90%). Using method (3), 86% (95% confidence interval 67-96%) of patients with a mean heart dose >3 Gy would be selected for DIBH treatment. The estimated mean and standard error for the area under the receiver operator characteristic curve for MHD as a predictor for mean heart dose was 0.85 (0.03). Conclusion: This study supports the use of proposed an 'upfront selection process' as a means of selecting patients for treatment with DIBH and avoiding two radiotherapy planning scans. Calculation of MHD can be used as a surrogate for mean heart dose in the selection of cases for DIBH. (C) 2017 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.