Artificial intelligence defines protein-based classification of thyroid nodules.

Determination of malignancy in thyroid nodules remains a major diagnostic challenge. Here we report the feasibility and clinical utility of developing an AI-defined protein-based biomarker panel for diagnostic classification of thyroid nodules: based initially on formalin-fixed paraffin-embedded (FFPE), and further refined for fine-needle aspiration (FNA) tissue specimens of minute amounts which pose technical challenges for other methods. We first developed a neural network model of 19 protein biomarkers based on the proteomes of 1724 FFPE thyroid tissue samples from a retrospective cohort. This classifier achieved over 91% accuracy in the discovery set for classifying malignant thyroid nodules. The classifier was externally validated by blinded analyses in a retrospective cohort of 288 nodules (89% accuracy; FFPE) and a prospective cohort of 294 FNA biopsies (85% accuracy) from twelve independent clinical centers. This study shows that integrating high-throughput proteomics and AI technology in multi-center retrospective and prospective clinical cohorts facilitates precise disease diagnosis which is otherwise difficult to achieve by other methods.

An integrated automated multispectral imaging technique that simultaneously detects and quantitates viral RNA and immune cell protein markers in fixed sections from Epstein-Barr virus-related tumours.

BACKGROUND/AIM: Epstein-Barr virus (EBV) is an oncovirus that is commonly associated with the development of lymphomas and epithelial carcinomas. In the era of immunotherapy, histological evaluation of EBV-related cancers is currently a multi-sample, multi-technique process requiring separate time-consuming detection of EBV-encoded small RNAs by in situ hybridisation (ISH), and parallel labelling of sections for cancer-associated protein markers. METHODS: Using EBV-associated tumours as proof-of-concept for feasibility, here we developed an approach that allows simultaneous detection of EBV RNAs and multiple protein markers such as PD-L1, EBV-LMP, CD8, CD4, CD20, CD30 and CD15on a single tissue section based on our recently reported automated staining protocol. RESULTS: We successfully combined multiplex immunofluorescence (mIF) to detect 3 abovementioned protein markers involved in cancer, with ISH, and applied the protocol to f tissue samples from patients diagnosed with EBV-associated pulmonary lymphoepithelioma-like carcinoma (LELC), gastric carcinoma and Hodgkin's Lymphoma. Empowered by the Vectra 3 Automated Quantitative Pathology Imaging System, we demonstrate the utility and potential of this integrated approach to concurrently detect and quantitate viral RNA and protein biomarkers of immune and tumour cells. CONCLUSION: This study represents an important step forward in the research and diagnosis of EBV-associated cancers, and could be readily modified to include other proteins and RNA markers to apply to other malignancies. More importantly, the novel automated ISH-mIF protocol that we detailly described here could also be readily reproduced by most of the diagnostic and research lab to future projects that aim to look at both RNA and protein markers.

Genomic and Epigenomic Profiling of High-Risk Intestinal Metaplasia Reveals Molecular Determinants of Progression to Gastric Cancer.

Intestinal metaplasia (IM) is a pre-malignant condition of the gastric mucosa associated with increased gastric cancer (GC) risk. We performed (epi)genomic profiling of 138 IMs from 148 cancer-free patients, recruited through a 10-year prospective study. Compared with GCs, IMs exhibit low mutational burdens, recurrent mutations in certain tumor suppressors (FBXW7) but not others (TP53, ARID1A), chromosome 8q amplification, and shortened telomeres. Sequencing identified more IM patients with active Helicobacter pylori infection compared with histopathology (11%-27%). Several IMs exhibited hypermethylation at DNA methylation valleys; however, IMs generally lack intragenic hypomethylation signatures of advanced malignancy. IM patients with shortened telomeres and chromosomal alterations were associated with subsequent dysplasia or GC; conversely patients exhibiting normal-like epigenomic patterns were associated with regression.

Metabolomic prediction of treatment outcome in pancreatic ductal adenocarcinoma patients receiving gemcitabine.

PURPOSE: Resistance to gemcitabine remains a key challenge in the treatment of pancreatic ductal adenocarcinoma (PDAC), necessitating the constant search for effective strategies for a priori prediction of clinical outcome. While the existing studies focused on aberration of drug disposition genes and proteins as molecular predictors of gemcitabine treatment outcomes, the metabolic aberration associated with chemoresistance in clinical PDAC has been neglected. This exploratory study investigated the potential role of tissue metabolomics in characterizing the clinical treatment outcome of gemcitabine therapy. METHODS: Surgically resected tumors from PDAC patients who underwent gemcitabine-based adjuvant chemotherapy (n = 25) were subjected to metabotyping using gas chromatography/time-of-flight mass spectrometry (GC/TOFMS). RESULTS: A partial least-squares discriminant analysis (PLS-DA) model clearly distinguished patients who had favorable survival [overall survival (OS) > 24 months] from those who exhibited poorer survival (OS < 16 months) (Q 2 = 0.302). Receiver-operating characteristic analysis demonstrated the robustness of the PLS-DA model with an area under the curve of 1. PLS-DA revealed 19 marker metabolites (e.g., lactic acid, proline, and pyroglutamate) that shed insights into the chemoresistance of gemcitabine in PDAC. Particularly, tissue levels of lactic acid complemented transcript expression levels of human equilibrative nucleoside transporter 1 in distinguishing patients according to their overall survival. CONCLUSION: This work established proof-of-principle for GC/TOFMS-based global metabotyping of PDAC and laid the foundation for future discovery of metabolic biomarkers predictive of gemcitabine resistance in PDAC chemotherapy.

Genomic and proteomic characterization of ARID1A chromatin remodeller in ampullary tumors.

AT rich interactive domain 1A (ARID1A) is one of the most commonly mutated genes in a broad variety of tumors. The mechanisms that involve ARID1A in ampullary cancer progression remains elusive. Here, we evaluated the frequency of ARID1A and KRAS mutations in ampullary adenomas and adenocarcinomas and in duodenal adenocarcinomas from two cohorts of patients from Singapore and Romania, correlated with clinical and pathological tumor features, and assessed the functional role of ARID1A. In the ampullary adenocarcinomas, the frequency of KRAS and ARID1A mutations was 34.7% and 8.2% respectively, with a loss or reduction of ARID1A protein in 17.2% of the cases. ARID1A mutational status was significantly correlated with ARID1A protein expression level (P=0.023). There was a significant difference in frequency of ARID1A mutation between Romania and Singapore (2.7% versus 25%, P=0.04), suggestive of different etiologies. One somatic mutation was detected in the ampullary adenoma group. In vitro studies indicated the tumor suppressive role of ARID1A. Our results warrant further investigation of this chromatin remodeller as a potential early biomarker of the disease, as well as identification of therapeutic targets in ARID1A mutated ampullary cancers.