Citations

BackgroundPretreatment determination of renal cell carcinoma aggressiveness may help to guide clinical decision‐making.PurposeTo evaluate the efficacy of residual convolutional neural network using routine MRI in differentiating low‐grade (grade I–II) from high‐grade (grade III–IV) in stage I and II renal cell carcinoma.Study TypeRetrospective.PopulationIn all, 376 patients with 430 renal cell carcinoma lesions from 2008–2019 in a multicenter cohort were acquired. The 353 Fuhrman‐graded renal cell carcinomas were divided into a training, validation, and test set with a 7:2:1 split. The 77 WHO/ISUP graded renal cell carcinomas were used as a separate WHO/ISUP test set.Field Strength/Sequence1.5T and 3.0T/T2‐weighted and T1 contrast‐enhanced sequences.AssessmentThe accuracy, sensitivity, and specificity of the final model were assessed. The receiver operating characteristic (ROC) curve and precision‐recall curve were plotted to measure the performance of the binary classifier. A confusion matrix was drawn to show the true positive, true negative, false positive, and false negative of the model.Statistical TestsMann–Whitney U‐test for continuous data and the chi‐square test or Fisher's exact test for categorical data were used to compare the difference of clinicopathologic characteristics between the low‐ and high‐grade groups. The adjusted Wald method was used to calculate the 95% confidence interval (CI) of accuracy, sensitivity, and specificity.ResultsThe final deep‐learning model achieved a test accuracy of 0.88 (95% CI: 0.73–0.96), sensitivity of 0.89 (95% CI: 0.74–0.96), and specificity of 0.88 (95% CI: 0.73–0.96) in the Fuhrman test set and a test accuracy of 0.83 (95% CI: 0.73–0.90), sensitivity of 0.92 (95% CI: 0.84–0.97), and specificity of 0.78 (95% CI: 0.68–0.86) in the WHO/ISUP test set.Data ConclusionDeep learning can noninvasively predict the histological grade of stage I and II renal cell carcinoma using conventional MRI in a multiinstitutional dataset with high accuracy.Level of Evidence3Technical Efficacy Stage2

Abstract Purpose: With increasing incidence of renal mass, it is important to make a pretreatment differentiation between benign renal mass and malignant tumor. We aimed to develop a deep learning model that distinguishes benign renal tumors from renal cell carcinoma (RCC) by applying a residual convolutional neural network (ResNet) on routine MR imaging. Experimental Design: Preoperative MR images (T2-weighted and T1-postcontrast sequences) of 1,162 renal lesions definitely diagnosed on pathology or imaging in a multicenter cohort were divided into training, validation, and test sets (70:20:10 split). An ensemble model based on ResNet was built combining clinical variables and T1C and T2WI MR images using a bagging classifier to predict renal tumor pathology. Final model performance was compared with expert interpretation and the most optimized radiomics model. Results: Among the 1,162 renal lesions, 655 were malignant and 507 were benign. Compared with a baseline zero rule algorithm, the ensemble deep learning model had a statistically significant higher test accuracy (0.70 vs. 0.56, P = 0.004). Compared with all experts averaged, the ensemble deep learning model had higher test accuracy (0.70 vs. 0.60, P = 0.053), sensitivity (0.92 vs. 0.80, P = 0.017), and specificity (0.41 vs. 0.35, P = 0.450). Compared with the radiomics model, the ensemble deep learning model had higher test accuracy (0.70 vs. 0.62, P = 0.081), sensitivity (0.92 vs. 0.79, P = 0.012), and specificity (0.41 vs. 0.39, P = 0.770). Conclusions: Deep learning can noninvasively distinguish benign renal tumors from RCC using conventional MR imaging in a multi-institutional dataset with good accuracy, sensitivity, and specificity comparable with experts and radiomics.

Abstract The interpretation of accumulating genomic data with respect to tumor evolution and cancer progression requires integrated models. We developed a computational approach that enables the construction of disease progression models using static sample data. Application to breast cancer data revealed a linear, branching evolutionary model with two distinct trajectories for malignant progression. Here, we used the progression model as a foundation to investigate the relationships between matched primary and metastasis breast tumor samples. Mapping paired data onto the model confirmed that molecular breast cancer subtypes can shift during progression and supported directional tumor evolution through luminal subtypes to increasingly malignant states. Cancer progression modeling through the analysis of available static samples represents a promising breakthrough. Further refinement of a roadmap of breast cancer progression will facilitate the development of improved cancer diagnostics, prognostics, and targeted therapeutics. Significance: Analysis of matched primary and metastatic tumor samples supports a unidirectional, linear cancer evolution process and sheds light on longstanding issues regarding the origins of molecular subtypes and their progression relationships.

Abstract Histiocytic neoplasms are clonal, hematopoietic disorders characterized by an accumulation of abnormal monocyte-derived dendritic cells or macrophages in Langerhans Cell (LCH) and non-Langerhans (non-LCH) histiocytoses, respectively. The discovery of the BRAF V600E mutation in ~50% of patients with LCH and the non-LCH Erdheim-Chester Disease (ECD) provided the first molecular target in these patients and novel insights into the pathogenesis of these disorders. However, recurrent mutations in the majority of the ~50% of BRAF V600E-wild type patients with non-LCH are unknown. Moreover, recurrent mutations outside of the MAP kinase pathway are undefined throughout histiocytic neoplasms. To address these issues, we performed whole exome sequencing (WES) of frozen biopsies from 24 patients with LCH (n=10) or ECD (n=14) paired with peripheral blood mononuclear cells. 13/24 patients also underwent RNA sequencing (RNA-seq). All mutations in activating kinases were validated by droplet-digital PCR, while targeted-capture next-generation sequencing validated all others. Both adult (n=18; n=2 with LCH) and pediatric cases (n=9; n=8 with LCH) were included. Using combined WES/RNA-seq, activating kinase alterations were identified in 100% of patients. In LCH, 60% and 40% had BRAF V600E and MAP2K1 mutations, respectively. In non-LCH 51%, 14%, 14%, and 7% were BRAFV600E, ARAF, MAP2K1, and NRAS mutant (Fig1A). Overall, a mean of 7 non-synonymous mutations per adult patient was identified (range 1-22) compared with 5 mutations per pediatric patient (range 4-9; p =ns). Mutations affecting diverse cellular processes were found to co-exist with kinase mutations including mutations in epigenetic modifiers and the p38/MAPK pathway. In addition to kinase point mutations, RNA-seq identified recurrent, in-frame kinase fusions-a first for these disorders. All identified fusions were validated using FISH and RT-PCR. This includes novel fusions in BRAF (RNF11-BRAF and CLIP2-BRAF), as well as therapeutically important fusions in ALK (2 separate KIF5B-ALK fusions) and NTRK1 (LMNA-NTRK1;Fig1B). Expression of each fusion in Ba/F3 cells conferred cytokine-independent growth. Importantly, the BRAF fusions were found to be sensitive to MEK inhibition but resistant to vemurafenib while the ALK fusions conferred sensitivity to the ALK inhibitors crizotinib or alectinib. We next interrogated a validation cohort of 37 BRAF V600E-wild type, non-LCH, formalin-fixed, paraffin-embedded tissue samples using targeted mutational profiling for MAP2K1, ARAF, NRAS, KRAS, and PIK3CA. This revealed activating mutations in MAP2K1 (32%; n=12), NRAS (16%; n=6), KRAS (11%; n=4), PIK3CA (8%; n=3), and ARAF (3%; n=1). Three of the investigated non-LCH patients with refractory disease and progressive organ dysfunction were treated with targeted therapies based on the discovery of novel kinase alterations described above. Treatment of 2 refractory MAP2K1- mutant, non-LCH patients with MEK inhibitors (trametinib or cobimetinib) resulted in dramatic clinical improvement (Fig1C). Both patients have been maintained on MEK inhibitor single-agent therapy with a sustained clinical response for >100 days. Further evidence of effective targeted inhibition was found in a refractory ECD patient carrying an ARAF S214A mutation. This patient failed to respond to 3 lines of prior therapies and suffered near blindness due to disease infiltration in the retina and optic nerves. Given a recent report of complete response to sorafenib in a lung cancer patient with an ARAF S214C mutation, we initiated sorafenib. Within 12 weeks, there was improvement in the patientÕs eyesight and decreased infiltrative disease, coinciding with >50% decrease in mutant ARAF DNA in plasma cell-free DNA. Whole exome and transcriptome sequencing identified activating kinase mutations or translocations in all patients with the common downstream effect of activating the MAPK pathway. The preliminary, dramatic, clinical efficacy observed with use of MEK and RAF inhibitors in MAP2K1 - and ARAF-mutated, non-LCH patients further supports the central role of targeting the MAPK pathway in these tumors. The discovery of the discussed mutations and fusions in diverse kinases provides critical new insights into the genetic events central to a spectrum of adult and pediatric histiocytic neoplasms. Figure 1. Figure 1. Disclosures Off Label Use: This abstract describes use of MEK inhibitors (both tremetinib and cobimetinib) as well as sorafenib for MEK1 and ARAF mutant histiocytosis. . Stephens:Foundation Medicine, Inc.: Employment, Equity Ownership. Miller:Foundation Medicine, Inc.: Employment, Equity Ownership. Ross:Foundation Medicine Inc.: Employment. Ali:Foundation Medicine Inc.: Employment. Hyman:Chugai Pharma: Consultancy; Biotherapeutics: Consultancy; Atara: Consultancy, Honoraria.

Abstract Motivation Cancer subtype classification has the potential to significantly improve disease prognosis and develop individualized patient management. Existing methods are limited by their ability to handle extremely high-dimensional data and by the influence of misleading, irrelevant factors, resulting in ambiguous and overlapping subtypes. Results To address the above issues, we proposed a novel approach to disentangling and eliminating irrelevant factors by leveraging the power of deep learning. Specifically, we designed a deep-learning framework, referred to as DeepType, that performs joint supervised classification, unsupervised clustering and dimensionality reduction to learn cancer-relevant data representation with cluster structure. We applied DeepType to the METABRIC breast cancer dataset and compared its performance to state-of-the-art methods. DeepType significantly outperformed the existing methods, identifying more robust subtypes while using fewer genes. The new approach provides a framework for the derivation of more accurate and robust molecular cancer subtypes by using increasingly complex, multi-source data. Availability and implementation An open-source software package for the proposed method is freely available at http://www.acsu.buffalo.edu/~yijunsun/lab/DeepType.html. Supplementary information Supplementary data are available at Bioinformatics online.

Abstract Histiocytic neoplasms are clonal, hematopoietic disorders characterized by an accumulation of abnormal, monocyte-derived dendritic cells or macrophages in Langerhans cell histiocytosis (LCH) and non-Langerhans cell histiocytosis (non-LCH), respectively. The discovery of BRAFV600E mutations in approximately 50% of these patients provided the first molecular therapeutic target in histiocytosis. However, recurrent driving mutations in the majority of patients with BRAFV600E–wild-type non-LCH are unknown, and recurrent cooperating mutations in non-MAP kinase pathways are undefined for the histiocytic neoplasms. Through combined whole-exome and transcriptome sequencing, we identified recurrent kinase fusions involving BRAF, ALK, and NTRK1, as well as recurrent, activating MAP2K1 and ARAF mutations in patients with BRAFV600E–wild-type non-LCH. In addition to MAP kinase pathway lesions, recurrently altered genes involving diverse cellular pathways were identified. Treatment of patients with MAP2K1- and ARAF-mutated non-LCH using MEK and RAF inhibitors, respectively, resulted in clinical efficacy, demonstrating the importance of detecting and targeting diverse kinase alterations in these disorders. Significance: We provide the first description of kinase fusions in systemic histiocytic neoplasms and activating ARAF and MAP2K1 mutations in non-Langerhans histiocytic neoplasms. Refractory patients with MAP2K1- and ARAF-mutant histiocytoses had clinical responses to MEK inhibition and sorafenib, respectively, highlighting the importance of comprehensive genomic analysis of these disorders. Cancer Discov; 6(2); 154–65. ©2015 AACR. This article is highlighted in the In This Issue feature, p. 109

A Na+ current is present in human jejunal circular smooth muscle cells. The aim of the present study was to determine the role of the cytoskeleton in the regulation of the Na+ current. Whole cell currents were recorded by using standard patch-clamp techniques with Cs+ in the pipette to block K+currents. Cytochalasin D and gelsolin were used to disrupt the actin cytoskeleton and phalloidin to stabilize it. Colchicine was used to disassemble the microtubule cytoskeleton (and intermediate filaments) and paclitaxel to stabilize it. Acrylamide was used to disrupt the intermediate filament cytoskeleton. Perfusion of the recording chamber at 10 ml/min increased peak Na+ current recorded from jejunal smooth muscle cells by 27 ± 3%. Cytochalasin D and gelsolin abolished the perfusion-induced increase in Na+current, whereas incubation with phalloidin, colchicine, paclitaxel, or acrylamide had no effect. In conclusion, the Na+ current expressed in human jejunal circular smooth muscle cells appears to be regulated by the cytoskeleton. An intact actin cytoskeleton is required for perfusion-induced activation of the Na+ current.