Convolutional Neural Network Using a Breast MRI Tumor Dataset Can Predict Oncotype Dx Recurrence Score.

BACKGROUND: Oncotype Dx is a validated genetic analysis that provides a recurrence score (RS) to quantitatively predict outcomes in patients who meet the criteria of estrogen receptor positive / human epidermal growth factor receptor-2 negative (ER+/HER2-)/node negative invasive breast carcinoma. Although effective, the test is invasive and expensive, which has motivated this investigation to determine the potential role of radiomics. HYPOTHESIS: We hypothesized that convolutional neural network (CNN) can be used to predict Oncotype Dx RS using an MRI dataset. STUDY TYPE: Institutional Review Board (IRB)-approved retrospective study from January 2010 to June 2016. POPULATION: In all, 134 patients with ER+/HER2- invasive ductal carcinoma who underwent both breast MRI and Oncotype Dx RS evaluation. Patients were classified into three groups: low risk (group 1, RS <18), intermediate risk (group 2, RS 18-30), and high risk (group 3, RS >30). FIELD STRENGTH/SEQUENCE: 1.5T and 3.0T. Breast MRI, T1 postcontrast. ASSESSMENT: Each breast tumor underwent 3D segmentation. In all, 1649 volumetric slices in 134 tumors (mean 12.3 slices/tumor) were evaluated. A CNN consisted of four convolutional layers and max-pooling layers. Dropout at 50% was applied to the second to last fully connected layer to prevent overfitting. Three-class prediction (group 1 vs. group 2 vs. group 3) and two-class prediction (group 1 vs. group 2/3) models were performed. STATISTICAL TESTS: A 5-fold crossvalidation test was performed using 80% training and 20% testing. Diagnostic accuracy, sensitivity, specificity, and receiver operating characteristic (ROC) area under the curve (AUC) were evaluated. RESULTS: The CNN achieved an overall accuracy of 81% (95% confidence interval [CI] ± 4%) in three-class prediction with specificity 90% (95% CI ± 5%), sensitivity 60% (95% CI ± 6%), and the area under the ROC curve was 0.92 (SD, 0.01). The CNN achieved an overall accuracy of 84% (95% CI ± 5%) in two-class prediction with specificity 81% (95% CI ± 4%), sensitivity 87% (95% CI ± 5%), and the area under the ROC curve was 0.92 (SD, 0.01). DATA CONCLUSION: It is feasible for current deep CNN architecture to be trained to predict Oncotype DX RS. LEVEL OF EVIDENCE: 4 Technical Efficacy: Stage 2 J. Magn. Reson. Imaging 2019;49:518-524.

Clinical Management of Uveal and Conjunctival Melanoma.

Ocular melanoma is a rare but potentially devastating malignancy arising from the melanocytes of the uveal tract, conjunctiva, or orbit; it represents less than 5% of all melanoma cases in the United States. The management of ocular melanoma varies depending on its anatomic origin, since uveal and conjunctival melanoma have distinct biologies and thus different treatment strategies. Uveal melanoma is the most common type of ocular melanoma and is characterized by activation of the mitogen-activated protein kinase (MAPK) pathway (among other signaling pathways) via mutations in GNAQ or GNA11. Despite primary radiation or surgical therapy, up to 50% of patients will eventually develop metastatic disease, for which there is no standard therapy and no treatment that has been shown to improve overall survival. The biology of conjunctival melanoma is less well characterized but has been associated with BRAF and NRAS mutations, and results in metastatic disease in 20% to 30% of cases. Clinical trials are currently ongoing to further evaluate and optimize the role of targeted therapies, as well as immunotherapies, as both adjuvant and metastatic treatment in uveal and conjunctival melanoma.

Transcriptional control of non-apoptotic developmental cell death in C. elegans.

Programmed cell death is an essential aspect of animal development. Mutations in vertebrate genes that mediate apoptosis only mildly perturb development, suggesting that other cell death modes likely have important roles. Linker cell-type death (LCD) is a morphologically conserved cell death form operating during the development of Caenorhabditis elegans and vertebrates. We recently described a molecular network governing LCD in C. elegans, delineating a key role for the transcription factor heat-shock factor 1 (HSF-1). Although HSF-1 functions to protect cells from stress in many settings by inducing expression of protein folding chaperones, it promotes LCD by inducing expression of the conserved E2 ubiquitin-conjugating enzyme LET-70/UBE2D2, which is not induced by stress. Following whole-genome RNA interference and candidate gene screens, we identified and characterized four conserved regulators required for LCD. Here we show that two of these, NOB-1/Hox and EOR-1/PLZF, act upstream of HSF-1, in the context of Wnt signaling. A third protein, NHR-67/TLX/NR2E1, also functions upstream of HSF-1, and has a separate activity that prevents precocious expression of HSF-1 transcriptional targets. We demonstrate that the SET-16/mixed lineage leukemia 3/4 (MLL3/4) chromatin regulation complex functions at the same step or downstream of HSF-1 to control LET-70/UBE2D2 expression. Our results identify conserved proteins governing LCD, and demonstrate that transcriptional regulators influence this process at multiple levels.

HSF-1 activates the ubiquitin proteasome system to promote non-apoptotic developmental cell death in C. elegans.

Apoptosis is a prominent metazoan cell death form. Yet, mutations in apoptosis regulators cause only minor defects in vertebrate development, suggesting that another developmental cell death mechanism exists. While some non-apoptotic programs have been molecularly characterized, none appear to control developmental cell culling. Linker-cell-type death (LCD) is a morphologically conserved non-apoptotic cell death process operating in Caenorhabditis elegans and vertebrate development, and is therefore a compelling candidate process complementing apoptosis. However, the details of LCD execution are not known. Here we delineate a molecular-genetic pathway governing LCD in C. elegans. Redundant activities of antagonistic Wnt signals, a temporal control pathway, and mitogen-activated protein kinase kinase signaling control heat shock factor 1 (HSF-1), a conserved stress-activated transcription factor. Rather than protecting cells, HSF-1 promotes their demise by activating components of the ubiquitin proteasome system, including the E2 ligase LET-70/UBE2D2 functioning with E3 components CUL-3, RBX-1, BTBD-2, and SIAH-1. Our studies uncover design similarities between LCD and developmental apoptosis, and provide testable predictions for analyzing LCD in vertebrates.

PolyQ disease: misfiring of a developmental cell death program?

Polyglutamine (polyQ) repeat diseases are neurodegenerative ailments elicited by glutamine-encoding CAG nucleotide expansions within endogenous human genes. Despite efforts to understand the basis of these diseases, the precise mechanism of cell death remains stubbornly unclear. Much of the data seem to be consistent with a model in which toxicity is an inherent property of the polyQ repeat, whereas host protein sequences surrounding the polyQ expansion modulate severity, age of onset, and cell specificity. Recently, a gene, pqn-41, encoding a glutamine-rich protein, was found to promote normally occurring non-apoptotic cell death in Caenorhabditis elegans. Here we review evidence for toxic and modulatory roles for polyQ repeats and their host proteins, respectively, and suggest similarities with pqn-41 function. We explore the hypothesis that toxicity mediated by glutamine-rich motifs may be important not only in pathology, but also in normal development.

Control of nonapoptotic developmental cell death in Caenorhabditis elegans by a polyglutamine-repeat protein.

Death is a vital developmental cell fate. In Caenorhabditis elegans, programmed death of the linker cell, which leads gonadal elongation, proceeds independently of caspases and apoptotic effectors. To identify genes promoting linker-cell death, we performed a genome-wide RNA interference screen. We show that linker-cell death requires the gene pqn-41, encoding an endogenous polyglutamine-repeat protein. pqn-41 functions cell-autonomously and is expressed at the onset of linker-cell death. pqn-41 expression is controlled by the mitogen-activated protein kinase kinase SEK-1, which functions in parallel to the zinc-finger protein LIN-29 to promote cellular demise. Linker-cell death is morphologically similar to cell death associated with normal vertebrate development and polyglutamine-induced neurodegeneration. Our results may therefore provide molecular inroads to understanding nonapoptotic cell death in metazoan development and disease.

Conditional protein splicing: a new tool to control protein structure and function in vitro and in vivo.

Protein splicing is a naturally occurring process in which an intervening intein domain excises itself out of a precursor polypeptide in an autocatalytic fashion with concomitant linkage of the two flanking extein sequences by a native peptide bond. We have recently reported an engineered split VMA intein whose splicing activity in trans between two polypeptides can be triggered by the small molecule rapamycin. In this report, we show that this conditional protein splicing (CPS) system can be used in mammalian cells. Two model constructs harboring maltose-binding protein (MBP) and a His-tag as exteins were expressed from a constitutive promoter after transient transfection. The splicing product MBP-His was detected by Western blotting and immunoprecipitation in cells treated with rapamycin or a nontoxic analogue thereof. No background splicing in the absence of the small-molecule inducer was observed over a 24-h time course. Product formation could be detected within 10 min of addition of rapamycin, indicating the advantage of the posttranslational nature of CPS for quick responses. The level of protein splicing was dose dependent and could be competitively attenuated with the small molecule ascomycin. In related studies, the geometric flexibility of the CPS components was investigated with a series of purified proteins. The FKBP and FRB domains, which are dimerized by rapamycin and thereby induce the reconstitution of the split intein, were fused to the extein sequences of the split intein halves. CPS was still triggered by rapamycin when FKBP and FRB occupied one or both of the extein positions. This finding suggests yet further applications of CPS in the area of proteomics. In summary, CPS holds great promise to become a powerful new tool to control protein structure and function in vitro and in living cells.