In the randomised TARGIT-A trial, risk-adapted targeted intraoperative radiotherapy (TARGIT-IORT) during lumpectomy was non-inferior to whole-breast external beam radiotherapy, for local recurrence. In the long-term, no difference was found in any breast cancer outcome, whereas there were fewer deaths from non-breast-cancer causes. TARGIT-IORT should be included in pre-operative consultations with eligible patients.
Abstract Background: Over the last years, the management of patients with node positive early breast cancer has gone through important innovations. On the medical side, new targeted therapies such as olaparib and abemaciclib have been introduced, with promising results on the invasive disease-free survival. Moreover, sparing axillary lymph node dissection has proven to be noninferior in terms of overall survival. However, no tools are currently available to predict lymph node involvement before definitive surgical evaluation. The aim of the study was to analyze clinical and pathological characteristics of patients with node positive early breast cancer to explore potential risk profiles associated with a ≥3 nodal involvement. Methods: The study retrospectively analyzed 335 node-positive breast cancer patients treated at the Breast Unit of the CRO Aviano National Cancer Institute, between 2017 and 2021. Data regarding primary tumor biological features, lymph node involvement and surgical approach were collected. Associations between clinico-pathological characteristics and ≥3 lymph node involvement were tested through stepwise logistic regression and the gradient boosting machine learning algorithm (GBM). Results: Among the 335 analyzed patients, 87.0% had a primary tumor < 5 cm, with a single positive lymph node in 73.3% of cases. Hormone receptors were mainly positive (respectively 93.5% and 83.4% for estrogen and progesterone receptors). Tumor grade was most frequently well differentiated (Grade 1 in 60.7%), with a Ki67 < 20% (59.5%). After multivariable logistic regression, a tumor size ≥ 3 cm (OR 3.24, CI95% 1.47-7.17, p = 0.004), the presence of massive lymphovascular stromal invasion (OR 2.50, CI95% 1.02-6.14, p = 0.045) and 2 or more positive sentinel lymph nodes at surgical evaluation (OR 6.08, CI95% 3.34-11.05, p < 0.001) were associated with a higher risk of identifying ≥ 3 positive lymph nodes after subsequent axillary dissection. Similar results were observed in the luminal-like cohort. A GBM machine learning model was then developed with a 0.77 Area Under the Curve. Features with the highest relative importance (RI) were single sentinel node involvement (RI 16.1873), followed by tumor size ≥ 3 cm (RI 10.2024), ≥2 positive sentinel lymph nodes (RI 8.5050) and lymphovascular stromal invasion (4.0217). Consistently, number of positive sentinel lymph nodes and tumor size were the predominant features in all top 20 GBM models. Conclusions: The present study explored the definition of risk profiles linked to 3 or more positive lymph nodes based on clinical and pathological features. It, moreover, tested the feasibility of developing machine learning classifiers to support future clinical decision-making. Due to the growing complexity of the adjuvant setting, finding a balance between minimally invasive surgical and staging approaches and risk definition for treatment personalization will become increasingly critical. Citation Format: Tania Pivetta, Brenno Pastò, Martina Urbani, Elisabetta Benozzi, Nicola De Pascalis, Tiziana Perin, Mario Mileto, Bruno Pasquotti, Erica Piccoli, Lorenzo Vinante, Chiara Bampo, Silvia Bolzonello, Mattia Garutti, Milena Nicoloso, Serena Corsetti, Simona Scalone, Lucia da Ros, Paola di Nardo, Camilla Lisanti, Simon Spazzapan, Barbara Belletti, Michele Bartoletti, Lorenzo Gerratana, Samuele Massarut, Fabio Puglisi. Clinical and biological predictors of lymph node involvement in patients with early breast cancer for adjuvant treatment personalization [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P4-02-06.
Purpose: to predict the occurrence of late subcutaneous radiation induced fibrosis (RIF) after partial breast irradiation (PBI) for breast carcinoma by using machine learning (ML) models and radiomic features from 3D Biologically Effective Dose (3D-BED) and Relative Electron Density (3D-RED). Methods: 165 patients underwent external PBI following a hypo-fractionation protocol consisting of 40Gy/10 fractions, 35Gy/7 fractions and 28Gy/4 fractions, for 73, 60 and 32 patients, respectively. Physicians evaluated toxicity at regular intervals by the Common Terminology Adverse Events (CTAE) version 4.0. RIF was assessed every 3 months after the completion of radiation course and scored prospectively. RIF was experienced by 41 (24.8%) patients after average 5 years of follow up. The Hounsfield Units (HU) of the CT-images were converted into relative electron density (3D-RED) and Dose maps into Biologically Effective Dose (3D-BED), respectively. Shape, first-order and textural features of 3D-RED and 3D-BED were calculated in the PTV and breast. Clinical and demographic variables were also considered (954 features in total). Imbalance of the dataset was addressed by data augmentation using ADASYN technique. A subset of non-redundant features that best predict the data was identified by sequential feature selection. Support Vector Machines (SVM), ensemble machine learning (EML) using various aggregation algorithms and Naive Bayes (NB) classifiers were trained on patient dataset to predict RIF occurrence. Models were assessed using sensitivity and specificity of the ML classifiers and AUC of the score functions in repeated five-fold cross validation on the augmented dataset. Results: The SVM model with seven features was preferred for RIF prediction and scored sensitivity 0.83 (95% CI 0.80 - 0.86), specificity 0.75 (95% CI 0.71-0.77) and AUC of the score function 0.86 (0.85-0.88) on cross-validation. The selected features included cluster shade and Run Length Non-uniformity of breast 3D-BED, kurtosis and cluster shade from PTV 3D-RED, and 10th percentile of PTV 3D-BED. Conclusion: Textures extracted from 3D-BED and 3D-RED in the breast and PTV can predict late RIF and may help better select patient candidates to exclusive PBI.
The purpose of this study was to implement a machine learning model to predict skin dose from targeted intraoperative (TARGIT) treatment resulting in timely adoption of strategies to limit excessive skin dose.A total of 283 patients affected by invasive breast carcinoma underwent TARGIT with a prescribed dose of 6 Gy at 1 cm, after lumpectomy. Radiochromic films were used to measure the dose to the skin for each patient. Univariate statistical analysis was performed to identify correlation of physical and patient variables with measured dose. After feature selection of predictors of in vivo skin dose, machine learning models stepwise linear regression (SLR), support vector regression (SVR), ensemble with bagging or boosting, and feed forward neural networks were trained on results of in vivo dosimetry to derive models to predict skin dose. Models were evaluated by tenfold cross validation and ranked according to root mean square error (RMSE) and adjusted correlation coefficient of true vs predicted values (adj-R2 ).The predictors correlated with in vivo dosimetry were the distance of skin from source, depth-dose in water at depth of the applicator in the breast, use of a replacement source, and irradiation time. The best performing model was SVR, which scored RMSE and adj-R2 , equal to 0.746 [95% confidence intervals (CI), 95% CI 0.737,0.756] and 0.481 (95% CI 0.468,0.494), respectively, on the tenfold cross validation.The model trained on results of in vivo dosimetry can be used to predict skin dose during setup of patient for TARGIT and this allows for timely adoption of strategies to prevent of excessive skin dose.
BACKGROUND: Thanks to their distinct physical properties proton therapy (PT) offers a superior dose distribution than photon radiation therapy. For brain and skull base (SB) tumors such dose sparing can be harnessed: a) in benign tumors, to achieve long-term better quality of life and reduced risk of neurocognitive deficits as well as radiation-related malignancies by delivering standard dose level; b) in malignant tumors or malignancies with local aggressiveness, to improve local control by safe delivery of tumor dose escalation. In the following we report an overview on our facility and the neuro-oncology program. METHODS/RESULTS: The Trento facility is built on over an area that will host the new Trento Hospital so that it will be a hospital-based PT center. As opposed to most centers using PT in the past, our facility is equipped with active beam delivery employing the spot scanning technique. A cyclotron provides a library of pencil beams of variable energy (70 - 226 MeV) and spot size (σ, 3 - 7 mm). Within the center, radiation therapy is offered in two treatment rooms. To allow for 360° rotation of the beamline these rooms are equipped with two gantries. In the treatment rooms, patients are placed on a robotic treatment table offering maximal variability and flexibility for patient positioning. The rooms are also equipped with devices enabling X-ray for treatment position verification. In a second step, the rooms will be equipped with cone-beam CT (room 1) and CT on-rails (room 2). Additionally there is one room equipped with horizontal beamline and equivalent beam scanning technology dedicated to quality assurance and preclinical research. For treatment planning imaging, the center is equipped with a dedicated CT and 1.5T MR. The staff is composed by professionals that experienced in leading particle centers in Villigen (CH), Heidelberg (GER), Boston (USA), Jacksonville (USA), Philadelphia (USA). Dedicated protocols for tumors of the brain and SB, for both adults and pediatric patients will be conducted based on the previous conducted trials on PT and potential expected benefit. As mentioned above, according to different clinical aims we are going to treat both benign and malignant tumors. The former group includes pituitary adenomas, schwannoma, benign meningioma, chondrosarcoma, craniopharyngioma, ependymoma, low-grade gliomas, medulloblastoma. In a second step, a PT radiosurgery program will be implemented for some of them. The latter group includes chordoma, high-grade gliomas, and atypical-malignant meningiomas. Regardless of the tumor type patients can be accepted also for re-irradiation. CONCLUSIONS: After a dedicated planning, construction and commissioning phase, patient treatment at Trento PT center is expected for summer 2014. Based on the already demonstrated as well as further potential clinical benefit neuro-oncology represents one of the most important field of interest.
Purpose: to predict eligibility for deep inspiration breath-hold (DIBH) radiotherapy (RT) treatment of patients with left breast cancer from analysis of respiratory signal, using Deep Bidirectional Long Short-Term Memory (BLSTM) recurrent neural networks. Methods: The respiratory traces from 36 patients who underwent DIBH RT were collected. The patients’ RT treatment plans were generated for both DIBH and free-breathing (FB) modalities. The patients were divided into two classes (patient eligible or not), based on the decrease of maximum dose to the left anterior descending (LAD) artery achieved with DIBH, compared to that achieved with FB and ΔDL. Patients with ΔDL > median value of ΔDL within the patient cohort were assumed to be those selected for DIBH. A BLSTM-RNN was trained for classification of patients eligible for DIBH by analysis of their respiratory signals, as acquired during acquisition of the pre-treatment computed tomography (CT), for selecting the window for DIBH. The dataset was split into training (60%) and test groups (40%), and the hyper-parameters, including the number of hidden layers, the optimizer, the learning rate, and the number of epochs, were selected for optimising model performance. The BLSTM included 2 layers of 100 neural units, each followed by a dropout layer with 20% dropout, and was trained in 35 epochs using the Adam optimizer, with an initial learning rate of 0.0003. Results: The system achieved accuracy, specificity, and sensitivity of, F1 score and area under the receiving operating characteristic curve (AUC) of 71.4%, 66.7%, 80.1%, 72.4%, and 69.4% in the test dataset, respectively. Conclusions: The proposed BLSTM-RNN classified patients in the test set eligible for DIBH with good accuracy. These results look promising for building an accurate and robust decision system to provide automated assistance to the radiotherapy team in assigning patients to DIBH.
Randomised evidence supports the use of partial breast irradiation (PBI) with targeted intraoperative radiotherapy (TARGIT-IORT) for early stage breast cancer, but prospective data from real-world adoption of this technique is also important. The aim of this study was to determine if the outcome reported in TARGIT-A trial could be replicated in large cohort of early stage breast cancer treated with TARGIT-IORT.
'%3e%3cpath fill='%23012169' d='M0 0v30h60V0z'/%3e%3cpath stroke='%23fff' stroke-width='6' d='m0 0 60 30m0-30L0 30'/%3e%3cpath stroke='%23C8102E' stroke-width='4' d='m0 0 60 30m0-30L0 30' clip-path='url(%23b)'/%3e%3cpath stroke='%23fff' stroke-width='10' d='M30 0v30M0 15h60'/%3e%3cpath stroke='%23C8102E' stroke-width='6' d='M30 0v30M0 15h60'/%3e%3c/g%3e%3c/svg%3e)
