According to the clonal model of tumor evolution, trunk alterations arise at early stages and are ubiquitous. Through the characterization of early stages of hepatocarcinogenesis, we aimed to identify trunk alterations in hepatocellular carcinoma (HCC) and study their intra- and inter-tumor distribution in advanced lesions. A total of 151 samples representing the multistep process of hepatocarcinogenesis were analyzed by targeted-sequencing and a single nucleotide polymorphism array. Genes altered in early lesions (31 dysplastic nodules [DNs] and 38 small HCCs [sHCC]) were defined as trunk. Their distribution was explored in: a) different regions of large tumors (43 regions, 21 tumors), and b) different nodules of the same patient (39 tumors, 17 patients). Multinodular lesions were classified as intrahepatic metastases (IMs) or synchronous tumors based on chromosomal aberrations. TERT promoter mutations (10.5%) and broad copy-number aberrations in chromosomes 1 and 8 (3-7%) were identified as trunk gatekeepers in DNs and were maintained in sHCCs. Trunk drivers identified in sHCCs included TP53 (23%) and CTNNB1 (11%) mutations, and focal amplifications or deletions in known drivers (6%). Overall, TERT, TP53 and CTNNB1 mutations were the most frequent trunk events and at least one was present in 51% of sHCCs. Around 90% of mutations in these genes were ubiquitous among different regions of large tumors. In multinodular HCCs, 35% of patients harbored IMs; 85% of mutations in TERT, TP53 and/or CTNNB1 were retained in primary and metastatic tumors. Trunk events in early stages (TERT, TP53, CTNNB1 mutations) were ubiquitous across different regions of the same tumor and between primary and metastatic nodules in >85% of cases. This concept supports the knowledge that single biopsies would suffice to capture trunk mutations in HCC. Trunk alterations arise at early stages of cancer and are shared among all malignant cells of the tumor. In order to identify trunk alterations in HCC, we characterized early stages of hepatocarcinogenesis represented by dysplastic nodules and small lesions. Mutations in TERT, TP53 and CTNNB1 genes were the most frequent. Analyses in more advanced lesions showed that mutations in these same genes were shared between different regions of the same tumor and between primary and metastatic tumors, suggesting their trunk role in this disease.

Colorectal cancer is one of the most common cancers worldwide, accounting for an annual estimated 1.8 million incident cases. With the increasing number of colonoscopies being performed, colorectal biopsies make up a large proportion of any histopathology laboratory workload. We trained and validated a unique artificial intelligence (AI) deep learning model as an assistive tool to screen for colonic malignancies in colorectal specimens, in order to improve cancer detection and classification; enabling busy pathologists to focus on higher order decision-making tasks. The study cohort consists of Whole Slide Images (WSI) obtained from 294 colorectal specimens. Qritive’s unique composite algorithm comprises both a deep learning model based on a Faster Region Based Convolutional Neural Network (Faster-RCNN) architecture for instance segmentation with a ResNet-101 feature extraction backbone that provides glandular segmentation, and a classical machine learning classifier. The initial training used pathologists’ annotations on a cohort of 66,191 image tiles extracted from 39 WSIs. A subsequent application of a classical machine learning-based slide classifier sorted the WSIs into ‘low risk’ (benign, inflammation) and ‘high risk’ (dysplasia, malignancy) categories. We further trained the composite AI-model’s performance on a larger cohort of 105 resections WSIs and then validated our findings on a cohort of 150 biopsies WSIs against the classifications of two independently blinded pathologists. We evaluated the area under the receiver-operator characteristic curve (AUC) and other performance metrics. The AI model achieved an AUC of 0.917 in the validation cohort, with excellent sensitivity (97.4%) in detection of high risk features of dysplasia and malignancy. We demonstrate an unique composite AI-model incorporating both a glandular segmentation deep learning model and a classical machine learning classifier, with excellent sensitivity in picking up high risk colorectal features. As such, AI plays a role as a potential screening tool in assisting busy pathologists by outlining the dysplastic and malignant glands.

Circulating tumour DNA (ctDNA) has the potential to be a specific biomarker for the monitoring of tumours in patients with colorectal cancer (CRC). Here, our aim was to develop a personalised surveillance strategy to monitor the clinical course of CRC after surgery. We developed patient-specific ctDNA assays based on multiplexed detection of somatic mutations identified from patient primary tumours, and applied them to detect ctDNA in 44 CRC patients, analysing a total of 260 plasma samples. We found that ctDNA detection correlated with clinical events – it is detectable in pre-operative but not post-operative plasma, and also in patients with recurrent CRC. We also detected ctDNA in 11 out of 15 cases at or before clinical or radiological recurrence of CRC, indicating the potential of our assay for early detection of metastasis. We further present data from a patient with multiple primary cancers to demonstrate the specificity of our assays to distinguish between CRC recurrence and a second primary cancer. Our approach can complement current methods for surveillance of CRC by adding an individualised biological component, allowing us not only to point to the presence of residual or recurrent disease, but also attribute it to the original cancer.

Assessment of severity of liver fibrosis is essential in the management of non-alcoholic fatty liver disease (NAFLD). Second Harmonic Generation (SHG) microscopy is a novel optical tissue imaging system that provides automated quantification of fibrosis based on unique architectural features of collagen. This study aims to develop and validate a SHG-based index for automated staging of liver fibrosis in patients with NAFLD. SHG microscopy was performed on archived liver biopsy specimens from 83 patients with NAFLD. A unique algorithm was developed to identify specific SHG parameters that correlated with fibrosis stage. The accuracy of the algorithm was compared against clinical assessment by experienced liver histopathologists using the Brunt fibrosis staging and further validated using the leave-one-out cross-validation method. Mean age of the study cohort was 51.8 ± 11.7 years, with 41% males. A fibrosis index (SHG B-index) was developed comprising 14 unique SHG-based collagen parameters that correlated with severity of NAFLD fibrosis in a continuous fashion. The SHG B-index had excellent correlation with Brunt fibrosis stage (Spearman’s correlation 0.820, p1.76 had an overall diagnostic accuracy of 98.5% for prediction of presence of bridging fibrosis (Brunt stage ≥3) with sensitivity of 87.5%, specificity 98.0%, positive predictive value 96.6% and negative predictive value 92.6%. The SHG B-index is a unique SHG-based index that provides accurate automated assessment of fibrosis stage in NAFLD patients.