No abstract

Purpose/Objectives(s): Women with biologically favorable early stage breast cancer are increasingly treated with accelerated partial breast techniques. However, many alternative techniques require costly specialized equipment not routinely available in most radiation oncology facilities. In addition, suboptimal cosmetic outcomes have been reported with the external beam technique, possibly related to large post-operative treatment volumes. To address these issues, we designed a phase I dose-escalation protocol to determine the maximally tolerated dose (MTD) of a single radiosurgery treatment delivered preoperatively to the intact tumor plus a small margin. Materials/Methods: Women aged 55 or older with clinically node negative, ER and/or PR+, HER2-, T1 invasive carcinomas were enrolled (n = 26). Patients with low/intermediate grade in situ disease <2cm were also included (n = 6). Breast MRI was required for target volume delineation. An intensity-modulated treatment plan was designed to deliver 15, 18, or 21Gy in a single fraction. An additional breast MRI, including T1-weighted, T2-weighted, diffusion-weighted and dynamic-contrast enhanced imaging, was obtained prior to lumpectomy which took place within 10 days of radiation treatment. Acute toxicity was assessed 3-4 weeks after radiation and any grade 3/4 toxicity possibly, probably, or definitely related to treatment was considered dose limiting. Tumor tissue was obtained from diagnostic and lumpectomy specimens. Immunohistochemistry (IHC) for Fas was performed on paraffin-embedded samples before and after radiation. A histoscore was created using the average membrane and cytoplasmic staining intensity multiplied by the percentage of positive cells. Results: Thirty-two women were treated, 8 each at the 15, 18, and 21Gy dose levels with an additional expansion cohort at the final 21Gy dose level. The maximally tolerated dose was not reached. Three patients required post-operative conventional radiation due to high-risk tumor features (ex. larger primary, nodal involvement). At a median follow-up of 6.8 months, primarily mild toxicities (grade 1-2 dermatitis, fibrosis, and pain) were noted. At 6 months (n = 20), all reported cosmetic outcomes are excellent or good. At 12 months (n = 10), 80% are excellent or good. Both patients with a fair/poor cosmetic outcome received radiosurgery plus post-operative conventional treatment; one experienced grade 3 breast atrophy. There have been no local or distant recurrences to date. Post-treatment MRIs were obtained in 20/32 patients, with early indicators of decreased cell density and increased vascular permeability. Sixteen patients had evaluable paired IHC and six demonstrated significant Fas up-regulation after radiation. The mean combined post-treatment histoscore was about twice as high as the mean pre-treatment score. Conclusion: Preoperative stereotactic radiotherapy to the intact breast tumor can be delivered with widely available clinical tools in a convenient single fraction, and provides a unique opportunity to study breast cancer radiation response. 21Gy did not yield dose-limiting toxicity and will be utilized for future studies. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P5-14-04.

Background: Deep learning, especially deep convolutional neural network (CNN), has emerged as a promising approach for many image recognition or classification tasks, demonstrating human or even superhuman performance. Used as feature extractor, some pre-trained CNN models can match or surpass the performance of domain-specific, “handcrafted” features. In this study, we aim to determine whether deep features extracted from digital mammograms using a pre-trained deep CNN are prognostic of occult invasive disease for patients with ductal carcinoma in situ (DCIS) on core needle biopsy. Materials and Methods: In this retrospective study, we collected digital mammography magnification views for 99 subjects with DCIS at biopsy, 25 of which were subsequently upstaged to invasive cancer. We utilized a deep CNN model that was pre-trained on non-medical images (e.g., animals, plants, instruments) as the feature extractor. Through a statistical pooling strategy, we extracted deep features at different levels of convolutional layers from the lesion areas, without sacrificing the original resolution or distorting the underlying topology. A multivariate classifier was then trained to predict which tumors contain occult invasive disease. This was compared to the performance of traditional “handcrafted” computer vision (CV) features previously developed specifically to assess mammographic calcifications. The generalization performance was assessed using Monte Carlo cross validation and receiver operating characteristic (ROC) curve analysis. Results: Deep features were able to distinguish DCIS with occult invasion from pure DCIS, with an area under the ROC curve (AUC-ROC) equal to 0.70 (95% CI: 0.68-0.73). This performance was comparable to the "handcrafted" CV features (AUC-ROC = 0.68, 95% CI: 0.66-0.71) that were designed with prior domain knowledge. Conclusion: In spite of being pre-trained on only non-medical images, the deep features extracted from digital mammograms demonstrated comparable performance to "handcrafted" CV features for the challenging task of predicting DCIS upstaging. Acknowledgments: This work was supported in part by NIH/NCI R01-CQA185138 and DOD Breast Cancer Research Program W81XWH-14-1-0473. Citation Format: Lo JY, Grimm LJ, Mazurowski MA, Baker JA, Marks JR, King LM, Maley CC, Hwang E-SS, Shi B. Prediction of occult invasive disease in ductal carcinoma in situ using deep learning features [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr GS5-04.