Session Item

Incidental dose to the axilla in breast only radiotherapy (RT) can vary significantly with patient positioning, body habitus and field placement. Although extent of axillary surgery is the major risk factor for lymphoedema and shoulder morbidity, the dose to the axillary lateral thoracic junction (ALTJ), in the superior-lateral axilla, has previously been identified as a potential organ at risk for lymphoedema (1) (Fig. 1A). We report on incidental doses to shoulder and axillary structures in a prospective multi-centre cohort of patients treated with breast-conserving surgery. We also present a novel method to evaluate dose to regions of interest which could facilitate validation of new subregions of interest for RT toxicity.

Patients were selected from the REQUITE study (www.requite.eu) with the following criteria: full dose distribution available (primary + boost if delivered), supine position. Exclusion criteria were: prescribed axillary RT.  To assess doses in this cohort, all individual CTs and dose distributions were non-rigidly deformed to a single reference patient (NiftyReg), focusing on the breast region and 5cm superior and inferior to the breast. On this reference patient, lymph node regions (according to ESTRO guidelines), ALTJ (according to (1)) and humeral head were contoured and agreed by 2 clinical oncologists. The global accuracy of image-registration was assessed using normalised correlation (NCC). Dosimetric parameters for each structure were calculated for each including D2%, D50%, D95% and D98%.

171 patients were available for analysis, 114 treated to the whole breast (WB) and 57 treated to WB plus tumour bed simultaneous integrated boost (SIB) (Table 1). The global accuracy of image registration was excellent (NCC=0.9, ideal value 1). Doses to shoulder and axillary structures showed a large variation both within and between centres (Fig. 1B-D). In particular, for WB RT, D50% for level 1 was 46% (1.3-99.2%), WB plus SIB (46 (1.3-99.2) (Fig 1C). Similar variation was observed for internal mammary nodes and for other dose metrics (data not shown).

There is considerable variation between patients and between centres in the incidental dose to the axilla and the ALTJ region. This study highlights the necessity of investigating individual dose distributions when understanding the link between dose and toxicity in breast cancer patients. The proposed method demonstrates the feasibility of using deformable image registration techniques to facilitate efficient retrospective assessment and validation of newly identified regions of interest, without the need for resource-intensive contouring of hundreds of patients.  Further work is planned in the full REQUITE cohort to identify anatomical subregions particularly sensitive to dose, and link this with late arm/shoulder toxicity.

References
1.    Gross IJROBP 2019