Validation of a deep learning computer aided system for CT based lung nodule detection, classification, and growth rate estimation in a routine clinical population.

OBJECTIVE: In this study, we evaluated a commercially available computer assisted diagnosis system (CAD). The deep learning algorithm of the CAD was trained with a lung cancer screening cohort and developed for detection, classification, quantification, and growth of actionable pulmonary nodules on chest CT scans. Here, we evaluated the CAD in a retrospective cohort of a routine clinical population. MATERIALS AND METHODS: In total, a number of 337 scans of 314 different subjects with reported nodules of 3-30 mm in size were included into the evaluation. Two independent thoracic radiologists alternately reviewed scans with or without CAD assistance to detect, classify, segment, and register pulmonary nodules. A third, more experienced, radiologist served as an adjudicator. In addition, the cohort was analyzed by the CAD alone. The study cohort was divided into five different groups: 1) 178 CT studies without reported pulmonary nodules, 2) 95 studies with 1-10 pulmonary nodules, 23 studies from the same patients with 3) baseline and 4) follow-up studies, and 5) 18 CT studies with subsolid nodules. A reference standard for nodules was based on majority consensus with the third thoracic radiologist as required. Sensitivity, false positive (FP) rate and Dice inter-reader coefficient were calculated. RESULTS: After analysis of 470 pulmonary nodules, the sensitivity readings for radiologists without CAD and radiologist with CAD, were 71.9% (95% CI: 66.0%, 77.0%) and 80.3% (95% CI: 75.2%, 85.0%) (p < 0.01), with average FP rate of 0.11 and 0.16 per CT scan, respectively. Accuracy and kappa of CAD for classifying solid vs sub-solid nodules was 94.2% and 0.77, respectively. Average inter-reader Dice coefficient for nodule segmentation was 0.83 (95% CI: 0.39, 0.96) and 0.86 (95% CI: 0.51, 0.95) for CAD versus readers. Mean growth percentage discrepancy of readers and CAD alone was 1.30 (95% CI: 1.02, 2.21) and 1.35 (95% CI: 1.01, 4.99), respectively. CONCLUSION: The applied CAD significantly increased radiologist's detection of actionable nodules yet also minimally increasing the false positive rate. The CAD can automatically classify and quantify nodules and calculate nodule growth rate in a cohort of a routine clinical population. Results suggest this Deep Learning software has the potential to assist chest radiologists in the tasks of pulmonary nodule detection and management within their routine clinical practice.

Screening for modulators of cisplatin sensitivity: unbiased screens reveal common themes.

Cisplatin is a widely used chemotherapeutic agent to treat a variety of solid tumors. The cytotoxic mode of action of cisplatin is mediated by inducing conformational changes in DNA including intra- and inter-strand crosslink adducts. Recognition of these adducts results in the activation of the DNA damage response resulting in cell cycle arrest, repair, and potentially, apoptosis. Despite the clinical efficacy of cisplatin, many tumors are either intrinsically resistant or acquire resistance during treatment. The identification of cisplatin drug response modulators can help us understand these resistance mechanisms, provide biomarkers for treatment strategies, or provide drug targets for combination therapy. Here we discuss functional genetic screens, including one performed by us, set up to identify genes whose inhibition results in increased sensitivity to cisplatin. In summary, the validated genes identified in these screens mainly operate in DNA damage response including nucleotide excision repair, translesion synthesis, and homologous recombination.

The histone demethylase Jarid1b (Kdm5b) is a novel component of the Rb pathway and associates with E2f-target genes in MEFs during senescence.

Senescence is a robust cell cycle arrest controlled by the p53 and Rb pathways that acts as an important barrier to tumorigenesis. Senescence is associated with profound alterations in gene expression, including stable suppression of E2f-target genes by heterochromatin formation. Some of these changes in chromatin composition are orchestrated by Rb. In complex with E2f, Rb recruits chromatin modifying enzymes to E2f target genes, leading to their transcriptional repression. To identify novel chromatin remodeling enzymes that specifically function in the Rb pathway, we used a functional genetic screening model for bypass of senescence in murine cells. We identified the H3K4-demethylase Jarid1b as novel component of the Rb pathway in this screening model. We find that depletion of Jarid1b phenocopies knockdown of Rb1 and that Jarid1b associates with E2f-target genes during cellular senescence. These results suggest a role for Jarid1b in Rb-mediated repression of cell cycle genes during senescence.

Using large-scale RNAi screens to identify novel drug targets for cancer.

In recent years, the development and clinical implementation of targeted therapeutics have progressed significantly. The specific inhibition of components of signal transduction pathways controlling proliferation and survival has been a highly successful research strategy. However, cancer is a heterogeneous disease and, even within one type of cancer, different genetic alterations are associated with identical phenotypes. To advance the use of targeted therapeutics, it is not only essential to identify the crucial factors in the signal transduction networks that control cell proliferation and survival, but also to classify individual tumors according to genetic alterations that correlate with pathway activation. RNAi screening technologies have become established as an important strategy both to identify novel targets and to provide novel biomarkers that are crucial for the further development of personalized medicine. This feature review discusses different RNAi screening strategies and their contribution to the rapidly evolving field of targeted therapeutics.

Transforming growth factor-beta requires its target plasminogen activator inhibitor-1 for cytostatic activity.

The cytokine transforming growth factor beta (TGFbeta) has strong antiproliferative activity in most normal cells but contributes to tumor progression in the later stages of oncogenesis. It is not fully understood which TGFbeta target genes are causally involved in mediating its cytostatic activity. We report here that suppression of the TGFbeta target gene encoding plasminogen activator inhibitor-1 (PAI-1) by RNA interference leads to escape from the cytostatic activity of TGFbeta both in human keratinocytes (HaCaTs) and primary mouse embryo fibroblasts. Consistent with this, PAI-1 knock-out mouse embryo fibroblasts are also resistant to TGFbeta growth arrest. Conversely, we show that ectopic expression of PAI-1 in proliferating HaCaT cells induces a growth arrest. PAI-1 knockdown does not interfere with canonical TGFbeta signaling as judged by SMAD phosphorylation and induction of bona fide TGFbeta target genes. Instead, knockdown of PAI-1 results in sustained activation of protein kinase B. Significantly, we find that constitutive protein kinase B activity leads to evasion of the growth-inhibitory action of TGFbeta. Our data are consistent with a model in which induction of PAI-1 by TGFbeta is critical for the induction of proliferation arrest.