Comparison between cone-beam breast-CT and full-field digital mammography for microcalcification detection depending on breast density.

The purpose of this study was to evaluate the impact of breast density on the diagnostic performance of cone-beam breast-CT (CBBCT) in comparison to full-field digital mammography (FFDM) for the detection of microcalcifications. This retrospective IRB-approved study was conducted between December 2015 and March 2017 and enrolled 171 women with Breast Imaging Reporting and Data System category 4 or 5 lesions on FFDM and additional CBBCT; 56 of which were ineligible. The inclusion was restricted to 83 women (90 breasts, 90 lesions) with microcalcifications. All lesions underwent histology or were monitored by FFDM and a clinical examination at least 2 years after enrollment. Two breast radiologists independently read each data set twice. Sensitivity, specificity and area under the curve were compared between the modalities. Thirty-two breasts (35.5%) were grouped as non-dense breasts (American College of Radiology types a/b) and 58 breasts (64.5%) as dense breasts (American College of Radiology types c/d). Histopathological assessment was performed in 61 of 90 breast lesions (32 malignant, 1 high-risk and 28 benign). Area under the curve was larger for FFDM than for CBBCT (P = .085). The sensitivity was significantly higher for FFDM compared to CBBCT (P = .009). The specificity showed no significant differences comparing FFDM (both readers: 0.62) versus CBBCT (reader 1: 0.76, reader 2: 0.60; P = .192). Inter-observer-reliability on BI-RADS readings was almost perfect for FFDM and moderate for CBBCT (κ = 0.84, κ = 0.54, respectively). Intra-observer agreement was substantial to almost perfect for both methods and readers. Compared with FFDM, CBBCT demonstrated non-comparable results for microcalcification detection in dense and non-dense breasts.

Novel Breast Imaging and Machine Learning: Predicting Breast Lesion Malignancy at Cone-Beam CT Using Machine Learning Techniques.

OBJECTIVE: The purpose of this study is to evaluate the diagnostic performance of machine learning techniques for malignancy prediction at breast cone-beam CT (CBCT) and to compare them to human readers. SUBJECTS AND METHODS: Five machine learning techniques, including random forests, back propagation neural networks (BPN), extreme learning machines, support vector machines, and K-nearest neighbors, were used to train diagnostic models on a clinical breast CBCT dataset with internal validation by repeated 10-fold cross-validation. Two independent blinded human readers with profound experience in breast imaging and breast CBCT analyzed the same CBCT dataset. Diagnostic performance was compared using AUC, sensitivity, and specificity. RESULTS: The clinical dataset comprised 35 patients (American College of Radiology density type C and D breasts) with 81 suspicious breast lesions examined with contrast-enhanced breast CBCT. Forty-five lesions were histopathologically proven to be malignant. Among the machine learning techniques, BPNs provided the best diagnostic performance, with AUC of 0.91, sensitivity of 0.85, and specificity of 0.82. The diagnostic performance of the human readers was AUC of 0.84, sensitivity of 0.89, and specificity of 0.72 for reader 1 and AUC of 0.72, sensitivity of 0.71, and specificity of 0.67 for reader 2. AUC was significantly higher for BPN when compared with both reader 1 (p = 0.01) and reader 2 (p < 0.001). CONCLUSION: Machine learning techniques provide a high and robust diagnostic performance in the prediction of malignancy in breast lesions identified at CBCT. BPNs showed the best diagnostic performance, surpassing human readers in terms of AUC and specificity.

Screening analysis of ubiquitin ligases reveals G2E3 as a potential target for chemosensitizing cancer cells.

Cisplatin is widely used against various tumors, but resistance is commonly encountered. By inducing DNA crosslinks, cisplatin triggers DNA damage response (DDR) and cell death. However, the molecular determinants of how cells respond to cisplatin are incompletely understood. Since ubiquitination plays a major role in DDR, we performed a high-content siRNA screen targeting 327 human ubiquitin ligases and 92 deubiquitinating enzymes in U2OS cells, interrogating the response to cisplatin. We quantified γH2AX by immunofluorescence and image analysis as a read-out for DNA damage. Among known mediators of DDR, the screen identified the ubiquitin ligase G2E3 as a new player in the response to cisplatin. G2E3 depletion led to decreased γH2AX levels and decreased phosphorylation of the checkpoint kinase 1 (Chk1) upon cisplatin. Moreover, loss of G2E3 triggered apoptosis and decreased proliferation of cancer cells. Treating cells with the nucleoside analogue gemcitabine led to increased accumulation of single-stranded DNA upon G2E3 depletion, pointing to a defect in replication. Furthermore, we show that endogenous G2E3 levels in cancer cells were down-regulated upon chemotherapeutic treatment. Taken together, our results suggest that G2E3 is a molecular determinant of the DDR and cell survival, and that its loss sensitizes tumor cells towards DNA-damaging treatment.

Damage-induced DNA replication stalling relies on MAPK-activated protein kinase 2 activity.

DNA damage can obstruct replication forks, resulting in replicative stress. By siRNA screening, we identified kinases involved in the accumulation of phosphohistone 2AX (γH2AX) upon UV irradiation-induced replication stress. Surprisingly, the strongest reduction of phosphohistone 2AX followed knockdown of the MAP kinase-activated protein kinase 2 (MK2), a kinase currently implicated in p38 stress signaling and G2 arrest. Depletion or inhibition of MK2 also protected cells from DNA damage-induced cell death, and mice deficient for MK2 displayed decreased apoptosis in the skin upon UV irradiation. Moreover, MK2 activity was required for damage response, accumulation of ssDNA, and decreased survival when cells were treated with the nucleoside analogue gemcitabine or when the checkpoint kinase Chk1 was antagonized. By using DNA fiber assays, we found that MK2 inhibition or knockdown rescued DNA replication impaired by gemcitabine or by Chk1 inhibition. This rescue strictly depended on translesion DNA polymerases. In conclusion, instead of being an unavoidable consequence of DNA damage, alterations of replication speed and origin firing depend on MK2-mediated signaling.