Dose-volume effects of breast cancer radiation therapy on the risk of second oesophageal cancer

•252 oesophageal cancers (488 matched controls) after breast cancer were analysed.•The excess odds-ratio of oesophageal cancer linearly increased with median dose.•At a fixed median dose, larger V30 increased the risk of squamous cell carcinoma.•Median dose and V30 could be relevant metrics in treatment plan optimization.•The study highlights the role of highest doses in oesophageal cancer risk. To investigate the relationship between oesophagus dose-volume distribution and long-term risk of oesophageal cancer after radiation therapy for breast cancer. In a case-control study nested within a cohort of 289,748 ≥5-year survivors of female breast cancer treated in 1943–2003 in five countries, doses to the second primary cancer (DSPC) and individual dose-volume histograms (DVH) to the entire oesophagus were reconstructed for 252 oesophageal cancer cases and 488 matched controls (median follow-up time: 13, range: 5–37 years). Using conditional logistic regression, we estimated excess odds ratios (EOR) of oesophageal cancer associated with DVH metrics. We also investigated whether DVH metrics confounded or modified DSPC-related -risk estimates. Among the DVH metrics evaluated, median dose (Dmedian) to the entire oesophagus had the best statistical performance for estimating risk of all histological types combined (EOR/Gy = 0.071, 95% confidence interval [CI]: 0.018 to 0.206). For squamous cell carcinoma, the most common subtype, the EOR/Gy for Dmedian increased by 31% (95% CI: 3% to 205%) for each increment of 10% of V30 (p = 0.02). Adjusting for DVH metrics did not materially change the EOR/Gy for DSPC, but there was a borderline significant positive interaction between DSPC and V30 (p = 0.07). This first study investigating the relationship between oesophagus dose-volume distribution and oesophageal cancer risk showed an increased risk per Gy for Dmedian with larger volumes irradiated at high doses. While current techniques allows better oesophagus sparing, constraints applied to Dmedian and V30 could potentially further reduce the risk of oesophageal cancer.