Faculty Bibliography

PURPOSE/OBJECTIVE:Isocitrate dehydrogenase (IDH) mutant gliomas are a distinct glioma molecular subtype for which no effective molecularly-directed therapy exists. Low-grade gliomas, which are 80-90% IDH mutant, have high RNA levels of the cell surface Notch ligand DLL3. We sought to determine DLL3 expression by immunohistochemistry in glioma molecular subtypes and the potential efficacy of an anti-DLL3 antibody drug conjugate (ADC), rovalpituzumab tesirine (Rova-T), in IDH mutant glioma. EXPERIMENTAL DESIGN/METHODS:We evaluated DLL3 expression by RNA using TCGA data and by immunohistochemistry in a discovery set of 63 gliomas and 20 non-tumor brain tissues and a validation set of 62 known IDH wildtype and mutant gliomas using a monoclonal anti-DLL3 antibody. Genotype was determined using a DNA methylation array classifier or by sequencing. The effect of Rova-T on patient-derived endogenous IDH mutant glioma tumorspheres was determined by cell viability assay. RESULTS:Compared to IDH wildtype glioblastoma, IDH mutant gliomas have significantly higher DLL3 RNA (P<1x10-15) and protein by immunohistochemistry (P=0.0014 and P<4.3x10-6 in the discovery and validation set, respectively). DLL3 immunostaining was intense and homogeneous in IDH mutant gliomas, retained in all recurrent tumors, and detected in only 1 of 20 non-tumor brains. Patient-derived IDH mutant glioma tumorspheres overexpressed DLL3 and were potently sensitive to Rova-T in an antigen-dependent manner. CONCLUSIONS:DLL3 is selectively and homogeneously expressed in IDH mutant gliomas and can be targeted with Rova-T in patient-derived IDH mutant glioma tumorspheres. Our findings are potentially immediately translatable and have implications for therapeutic strategies that exploit cell surface tumor-associated antigens.

Visual inspection of histopathology slides is one of the main methods used by pathologists to assess the stage, type and subtype of lung tumors. Adenocarcinoma (LUAD) and squamous cell carcinoma (LUSC) are the most prevalent subtypes of lung cancer, and their distinction requires visual inspection by an experienced pathologist. In this study, we trained a deep convolutional neural network (inception v3) on whole-slide images obtained from The Cancer Genome Atlas to accurately and automatically classify them into LUAD, LUSC or normal lung tissue. The performance of our method is comparable to that of pathologists, with an average area under the curve (AUC) of 0.97. Our model was validated on independent datasets of frozen tissues, formalin-fixed paraffin-embedded tissues and biopsies. Furthermore, we trained the network to predict the ten most commonly mutated genes in LUAD. We found that six of them-STK11, EGFR, FAT1, SETBP1, KRAS and TP53-can be predicted from pathology images, with AUCs from 0.733 to 0.856 as measured on a held-out population. These findings suggest that deep-learning models can assist pathologists in the detection of cancer subtype or gene mutations. Our approach can be applied to any cancer type, and the code is available at https://github.com/ncoudray/DeepPATH .

Background: There are no predictive biomarkers of ipilimumab (IPI) toxicity. Of metastatic melanoma (MM) patients (pts) receiving IPI (3 mg/kg), 35% require systemic therapies to treat immune-related adverse events (irAEs) and 20% must terminate treatment [1]. Here we tested the hypothesis that a pre-existing autoantibody (autoAb) profile is predictive of IPI irAEs.
Method(s): We measured autoAb levels in pre- and post-treatment sera from MM pts who received IPI (3 mg/kg) monotherapy on a proteome microarray containing ~ 20,000 unique full-length human proteins (HuProt array, CDI Laboratories). Clinical data were prospectively collected with protocol-driven follow-up. IrAEs were categorized by CTCAE guidelines as none (grade 0), mild (grade 12), or severe (grade 34). AutoAb levels were standardized using median quantile normalization and considered positive hits if > 2-SD above the peak array signal and differed by >= 2-fold with p < 0.05 between toxicity groups (Non-parametric Analysis/Wilcox test).
Result(s): Seventy-eight sera from 37 MM pts were analyzed. Antibodies against CTLA-4 were significantly elevated post IPI treatment (p < 0.0001), validating the assay. The pre-treatment levels of 190 IgG autoAbs were significantly different in pts who experienced irAEs (n = 28) compared to those with no irAEs (n = 9). Comparison of severe irAE (n = 9) and no irAE (n = 9) groups revealed 129 IgG auto- Abs that significantly differed in pre-treatment sera. Localization and pathway analysis (UniProt, KEGG, Reactome) showed 81/190 (43%) of the autoAbs targeted nuclear and mitochondrial antigens and were enriched in metabolic pathways (p = 0.015). AutoAbs associated with irAEs did not correlate with treatment response.
Conclusion(s): AutoAbs to antigens enriched in metabolic pathways prior to treatment may predict IPI-induced toxicity in MM. The subcellular localization of targeted antigens could explain the autoimmune toxicities associated with IPI. Studies in larger cohorts and in pts receiving other checkpoint inhibitors and/or combination therapies are essential to determine the validity of the data. If validated, our results would support the discovery of the first toxicity predictor in cancer immunotherapy

Venous thromboembolism (VTE), comprised of deep vein thrombosis and pulmonary embolism, is a common health problem both in the United States and worldwide, with significant associated morbidity and mortality. Despite multiple known genetic and situational risk factors, an estimated 30% of all events remain classified as idiopathic, demonstrating a significant knowledge gap in the pathophysiology VTE. While platelets are well established as an essential contributor to thrombus formation and there has been recent interest in the role of neutrophil extracellular traps, specific cell types and pathways involved in the pathogenesis of VTE remain uncertain. In this study, our primary aims were to define a unique transcriptional signature for VTE and to identify the types of cells and specific pathways involved in development of VTE. Whole blood was collected in PAX gene tubes and RNA sequencing for coding mRNA was performed in an unbiased manner in 201 patients with prevalent VTE as well as 43 healthy controls. We used a bioinformatics approach to develop a unique signature for VTE by identifying differentially expressed genes, developing cell-type modules, and ascertaining pathways driving differentially expressed transcripts. We performed additional analyses on subgroups of patients with idiopathic VTE, patients with incident VTE, and VTE patients matched to healthy controls by age and sex. We went on to use machine learning methods to learn models that best differentiate VTE patients from healthy controls and validated it on a left out test set within our VTE population. Genes specific to neutrophils, erythrocytes, and platelets, in that order, were most significantly upregulated in patients with VTE compared to healthy controls. Genes related to T-cells were downregulated. Pathway analysis revealed upregulated neutrophil activation and degranulation, erythrocyte differentiation and homeostasis, and platelet degranulation. A gene signature of 217 transcripts was outstanding at differentiating patients with VTE versus healthy controls (AUC 0.94). Following adjustment for age, sex, and race/ethnicity our genetic signature remained significantly robust at differentiating patients with VTE versus controls (AUC 0.83). Our expression signature remained stable across patients with idiopathic VTE (AUC 0.93), and in patients who went on to develop future VTE events (AUC 0.95). In summary, we have demonstrated a whole blood transcriptional signature for prevalent and incident VTE. Genes related to neutrophils, erythrocytes, and platelets are upregulated in patients with VTE and genes related to T-cells were downregulated. These findings suggest an active role of cell types once thought to be passively entrapped within thrombus and provide new areas of study to establish the pathophysiology of VTE

There is a considerable need to identify those individuals with prostate cancer who have indolent disease. We propose that genes that control adult stem cell homeostasis in organs with slow turnover, such as the prostate, control cancer fate. One such gene, KLF4, overexpressed in murine prostate stem cells, regulates their homeostasis, blocks malignant transformation, and controls the self-renewal of tumor-initiating cells. KLF4 loss induces the molecular features of aggressive cancer and converts PIN lesions to invasive sarcomatoid carcinomas; its re-expression in vivo reverses this process. Bioinformatic analysis links these changes to human cancer. KLF4 and its downstream targets make up a gene signature that identifies indolent tumors and predicts recurrence-free survival. This approach may improve prognosis and identify therapeutic targets for advanced cancer.