Initial Studies with 11C-Vorozole PET Detect Overexpression of Intratumoral Aromatase in Breast Cancer.

Aromatase inhibitors are the mainstay of hormonal therapy in estrogen receptor-positive breast cancer, although the response rate is just over 50% and in vitro studies suggest that only two thirds of postmenopausal breast tumors overexpress aromatase. The goal of the present study was to validate and optimize PET with 11C-vorozole for measuring aromatase expression in postmenopausal breast cancer in vivo. Methods: Ten newly diagnosed postmenopausal women with biopsy-confirmed breast cancer were administered 11C-vorozole intravenously, and PET emission data were collected between 40 and 90 min after injection. Tracer injection and scanning were repeated 2 h after ingestion of 2.5 mg of letrozole. Mean and maximal SUVs and ratios to nontumor tissue in the contralateral breast were determined at baseline and after letrozole. Biopsy specimens from the same tumors were stained for aromatase using immunohistochemistry and evaluated for stain intensity and the percentage of immune-positive cells. Results: Seven of the 10 women (70%) demonstrated increased mean focal uptake of tracer (SUV ratio > 1.1) coinciding with the mammographic location of the lesion, whereas the other 3 women (30%) did not (SUV ratio ≤ 1.0). All patients with an SUV ratio above 1.1 had mean SUVs above 2.4, and there was no overlap (SUV ratio ≤ 1; SUVmean, 0.8-1.8). The SUV ratio relative to breast around tumor was indistinguishable from the ratio to contralateral breast. Pretreatment with letrozole reduced tracer uptake in most subjects, although the percentage of blocking varied across and within tumors. Tumors with a high SUV in vivo also showed a high immunohistochemical staining intensity. Conclusion: PET with 11C-vorozole is a useful technique for measuring aromatase expression in individual breast lesions, enabling noninvasive quantitative measurement of baseline and posttreatment aromatase availability in primary tumors and metastatic lesions.

Deep Neural Networks Improve Radiologists' Performance in Breast Cancer Screening.

We present a deep convolutional neural network for breast cancer screening exam classification, trained, and evaluated on over 200000 exams (over 1000000 images). Our network achieves an AUC of 0.895 in predicting the presence of cancer in the breast, when tested on the screening population. We attribute the high accuracy to a few technical advances. 1) Our network's novel two-stage architecture and training procedure, which allows us to use a high-capacity patch-level network to learn from pixel-level labels alongside a network learning from macroscopic breast-level labels. 2) A custom ResNet-based network used as a building block of our model, whose balance of depth and width is optimized for high-resolution medical images. 3) Pretraining the network on screening BI-RADS classification, a related task with more noisy labels. 4) Combining multiple input views in an optimal way among a number of possible choices. To validate our model, we conducted a reader study with 14 readers, each reading 720 screening mammogram exams, and show that our model is as accurate as experienced radiologists when presented with the same data. We also show that a hybrid model, averaging the probability of malignancy predicted by a radiologist with a prediction of our neural network, is more accurate than either of the two separately. To further understand our results, we conduct a thorough analysis of our network's performance on different subpopulations of the screening population, the model's design, training procedure, errors, and properties of its internal representations. Our best models are publicly available at https://github.com/nyukat/breast_cancer_classifier.