NTCP Modeling of Hypothyroidism after Supraclavicular-Directed Radiation Therapy for Breast Cancer

John Roberson,Huakang Huang,Wei-Huan Hou,Samuel Ryu,A. Stessin
DOI: https://doi.org/10.1016/j.ijrobp.2020.07.1033
2020-01-01
Abstract:Adjuvant radiation therapy (RT) is frequently employed in the treatment of early stage breast cancer. In select cases, patients receive radiation treatment to the supraclavicular (SCV) fossa to cover regional lymphatics. Due to the proximity of the thyroid to this fossa, hypothyroidism (HT) is a potential complication of SCV-directed RT. We performed a retrospective analysis to evaluate potential predictors of hypothyroidism in breast cancer patients who receive SCV-directed RT and to develop the first normal tissue complication probability (NTCP) model to predict HT specific to breast cancer patients. 192 breast cancer patients without pre-existing thyroid disorder who received SCV-directed RT were analyzed. Patients were treated with conventionally fractionated external beam RT to 44-50.4 Gy in 1.8-2 Gy per fraction with either intensity-modulated or three-dimensional conformal RT, often followed by an electron or photon boost to the tumor bed or scar. Patients' records were reviewed for clinical risk factors as well as the development of clinical or subclinical HT. Thyroid glands were retrospectively contoured, and individual dosimetric data was correlated with the development of hypothyroidism using univariable and multivariable logistic regression. Pearson correlation matrices were used to exclude factors demonstrating significant multicollinearity. An NTCP model was subsequently created based on this logistic regression, and model performance was evaluated using area under the receiver operating characteristic curve (AUC). Hypothyroidism developed in 37 patients (19.3%) subsequent to RT with 21 developing clinical and 16 developing subclinical HT. Median time to onset was 25 months (range: 2-83 months). The estimated incidence of HT was 9.4% at 2 years, 18.6% at 3 years, and 25.0% at 5 years. Only volume receiving less than 20 Gy (VS20) was a significant prognostic factor for HT on multivariable analysis with median VS20 of 5.58 cc (range: 0-14.63 cc) for patients who developed HT and 7.49 cc (range: 0.65-72.48 cc) for those who did not (p = 0.001). An NTCP model using VS20 was created. VS20 ≥ 10 cc was associated with a risk of HT less than 10%. The AUC value for this NTCP was 0.703 (95% CI 0.608-0.798). The odds ratio for developing HT with a VS20 less than 10 cc was 3.81 (95% CI: 1.11-13.10). For patients with smaller thyroids unable to meet this criterion, a second NTCP model demonstrated mean dose (Dmean) ≤ 11.8 Gy was associated with a risk of HT less than 10%. Our results demonstrate that VS20 is a significant predictor for the development of post-treatment hypothyroidism. NTCP modeling within our patient cohort suggests that VS20 ≥ 10 cc or Dmean ≤ 11.8 Gy may be recommended dosimetric guidelines to minimize the risk of HT in breast cancer patients receiving SCV-directed RT.
What problem does this paper attempt to address?