Development and validation of a machine learning model of radiation-induced hypothyroidism with clinical and dose-volume features.

BACKGROUND AND PURPOSE: Radiation-induced hypothyroidism (RIHT) is a common but underestimated late effect in head and neck cancers. However, no consensus exists regarding risk prediction or dose constraints in RIHT. We aimed to develop a machine learning model for the accurate risk prediction of RIHT based on clinical and dose-volume features and to evaluate its performance internally and externally. MATERIALS AND METHODS: We retrospectively searched two institutions for patients aged >20 years treated with definitive radiotherapy for nasopharyngeal or oropharyngeal cancer, and extracted their clinical information and dose-volume features. One was designated the developmental cohort, the other as the external validation cohort. We compared the performances of machine learning models with those of published normal tissue complication probability (NTCP) models. RESULTS: The developmental and external validation cohorts consisted of 378 and 49 patients, respectively. The estimated cumulative incidence rates of grade ≥1 hypothyroidism were 53.5% and 61.3% in the developmental and external validation cohorts, respectively. Machine learning models outperformed traditional NTCP models by having lower Brier scores at every time point and a lower integrated Brier score, while demonstrating a comparable calibration index and mean area under the curve. Even simplified machine learning models using only thyroid features performed better than did traditional NTCP algorithms. The machine learning models showed consistent performance between folds. The performance in a previously unseen external validation cohort was comparable to that of the cross-validation. CONCLUSIONS: Our model outperformed traditional NTCP models, with additional capabilities of predicting the RIHT risk at individual time points. A simplified model using only thyroid dose-volume features still outperforms traditional NTCP models and can be incorporated into future treatment planning systems for biological optimization.

Free flap reconstruction in the head and neck region following radiotherapy: a cohort study identifying negative outcome predictors.

BACKGROUND: With the increased use of radiotherapy in the treatment of head and neck cancers, free tissue transfer in an irradiated field is now common. Reported outcomes with free tissue transfer reconstructions in irradiated areas are often unclear and contradictory. This is attributable in part to small sample size and heterogeneity in patients and procedures. The goals of this study were, first, to determine the outcome of head and neck reconstructions in an irradiated field using a large retrospective cohort and, second, to identify negative outcome predictors that could potentially be modified to reduce the morbidity in these procedures. METHODS: Patients undergoing free flap reconstruction in irradiated head and neck regions between July of 2005 and July of 2007 were identified. Charts, including operative and radiotherapy records, were reviewed. Multiple logistic regressions were performed to identify negative outcome predictors. RESULTS: During the study period at the authors' institution, 984 patients underwent head and neck free flap reconstruction. Of these, 137 had free tissue transferred to irradiated fields. Patients received an average radiotherapy dose of 64.5 Gy. The overall flap success rate was 96.4 percent. The postoperative complication rate was 47 percent. There was an overall 22 percent reoperation rate and a 28 percent infection rate. Segmental mandibulectomy, larger flap size, and infection were found to be significant negative outcome predictors. CONCLUSIONS: This study confirms that free flap transfer to previously irradiated head and neck areas has a success rate comparable to that of transfer to nonirradiated zones. However, previously irradiated patients have a high risk of complications. Infection, as a modifiable risk factor, should be treated aggressively.

Prediction of supraclavicular lymph node metastasis in breast carcinoma.

PURPOSE: Supraclavicular lymph node metastasis in breast cancer patients has a poor prognosis, and aggressive local treatment has usually resulted in severe morbidity. The purpose of this study was to select high-risk neck metastasis patients for prophylactic radiotherapy. METHODS: Between 1990 and 1998, 2658 consecutive invasive breast cancer patients underwent surgery and adjuvant therapy in the hospital. The median age was 47 years (range 22-92). The median follow-up period was 39 months. The following factors were analyzed: age, tumor size, tumor location, histologic type, histologic grade, estrogen and progesterone receptor status, DNA flow cytometry study results, number of positive axillary lymph nodes, use of chemotherapy, radiotherapy, and/or hormonal therapy, and level of involved axillary nodes. RESULTS: Of the 2658 patients, 113 (4.3%) developed supraclavicular lymph node metastasis during this period. Young age (< or =40 years), tumor size >3 cm, high histologic grade, angiolymphatic invasion, negative estrogen receptor status, synthetic phase fraction >4%, >4 positive nodes, and level II or III involved nodes were all significant for predicting neck metastasis in the univariate analysis. Three predictive factors were significant after multivariate analysis: high histologic grade, >4 positive nodes, and axillary level II or III involved nodes. In patients with axillary level I involved nodes and < or =4 positive nodes, the incidence was 4.4%. If axillary level III was involved, the rate of supraclavicular lymph node metastasis was 15.1%. CONCLUSION: The incidence of supraclavicular lymph node metastasis was higher in the groups with >4 positive nodes and in those with axillary level II or III involved nodes. Selective use of comprehensive radiotherapy for these high-risk patients will achieve good locoregional control.