Explore the vast range of titles available.

BackgroundNeuroblastoma is the most common malignancy in infancy, accounting for 15% of childhood cancer deaths. Outcome for the high-risk disease remains poor. DNA-methylation patterns are significantly altered in all cancer types and can be utilised for disease stratification.MethodsGenome-wide DNA methylation (n = 223), gene expression (n = 130), genetic/clinical data (n = 213), whole-exome sequencing (n = 130) was derived from the TARGET study. Methylation data were derived from HumanMethylation450 BeadChip arrays. t-SNE was used for the segregation of molecular subgroups. A separate validation cohort of 105 cases was studied.ResultsFive distinct neuroblastoma molecular subgroups were identified, based on genome-wide DNA-methylation patterns, with unique features in each, including three subgroups associated with known prognostic features and two novel subgroups. As expected, Cluster-4 (infant diagnosis) had significantly better 5-year progression-free survival (PFS) than the four other clusters. However, in addition, the molecular subgrouping identified multiple patient subsets with highly increased risk, most notably infant patients that do not map to Cluster-4 (PFS 50% vs 80% for Cluster-4 infants, P = 0.005), and allowed identification of subgroup-specific methylation differences that may reflect important biological differences within neuroblastoma.ConclusionsMethylation-based clustering of neuroblastoma reveals novel molecular subgroups, with distinct molecular/clinical characteristics and identifies a subgroup of higher-risk infant patients.

The identification of cancer-specific vulnerability genes is one of the most promising approaches for developing more effective and less toxic cancer treatments. Cancer genomes exhibit thousands of changes in DNA methylation and gene expression, with the vast majority likely to be passenger changes. We hypothesised that, through integration of genome-wide DNA methylation/expression data, we could exploit this inherent variability to identify cancer subtype-specific vulnerability genes that would represent novel therapeutic targets that could allow cancer-specific cell killing. We developed a bioinformatics pipeline integrating genome-wide DNA methylation/gene expression data to identify candidate subtype-specific vulnerability partner genes for the genetic drivers of individual genetic/molecular subtypes. Using acute lymphoblastic leukaemia as an initial model, 21 candidate subtype-specific vulnerability genes were identified across the five common genetic subtypes, with at least one per subtype. To confirm the approach was applicable across cancer types, we also assessed medulloblastoma, identifying 15 candidate subtype-specific vulnerability genes across three of four established subtypes. Almost all identified genes had not previously been implicated in these diseases. Functional analysis of seven candidate subtype-specific vulnerability genes across the two tumour types confirmed that siRNA-mediated knockdown induced significant inhibition of proliferation/induction of apoptosis, which was specific to the cancer subtype in which the gene was predicted to be specifically lethal. Thus, we present a novel approach that integrates genome-wide DNA methylation/expression data to identify cancer subtype-specific vulnerability genes as novel therapeutic targets. We demonstrate this approach is applicable to multiple cancer types and identifies true functional subtype-specific vulnerability genes with high efficiency.

Precision in the diagnosis of diverse central nervous system (CNS) tumor types is crucial for optimal treatment. DNA methylation profiles, which capture the methylation status of thousands of individual CpG sites, are state-of-the-art data-driven means to enhance diagnostic accuracy but are also time consuming and not widely available. Here, to address these limitations, we developed Deep lEarning from histoPathoLOgy and methYlation (DEPLOY), a deep learning model that classifies CNS tumors to ten major categories from histopathology. DEPLOY integrates three distinct components: the first classifies CNS tumors directly from slide images (‘direct model’), the second initially generates predictions for DNA methylation beta values, which are subsequently used for tumor classification (‘indirect model’), and the third classifies tumor types directly from routinely available patient demographics. First, we find that DEPLOY accurately predicts beta values from histopathology images. Second, using a ten-class model trained on an internal dataset of 1,796 patients, we predict the tumor categories in three independent external test datasets including 2,156 patients, achieving an overall accuracy of 95% and balanced accuracy of 91% on samples that are predicted with high confidence. These results showcase the potential future use of DEPLOY to assist pathologists in diagnosing CNS tumors within a clinically relevant short time frame. A deep learning model is used to classify central nervous system tumors based on their DNA methylation profile directly from histopathology, and showed high accuracy in a large set of external validation cohorts, potentially informing downstream treatment.

Hepatocellular carcinoma (HCC), the most common type of liver cancer, has very poor outcomes. Current therapies often have low efficacy and significant toxicities. Thus, there is a critical need for the development of novel therapeutic approaches for HCC. We have developed a novel bioinformatics pipeline, which integrates genome-wide DNA methylation and gene expression data, to identify genes required for the survival of specific molecular cancer subgroups but not normal cells. Targeting these genes may induce cancer-specific “synthetic lethality”. Initially, five potential HCC molecular subgroups were identified based on global DNA methylation patterns. Subgroup-2 exhibited the most unique methylation profile and two candidate subtype-specific vulnerability or SL-like genes were identified for this subgroup, including TIAM1, a guanine nucleotide exchange factor encoding gene known to activate Rac1 signalling. siRNA targeting TIAM1 inhibited cell proliferation in TIAM1-positive (subgroup-2) HCC cell lines but had no effect on the normal hepatocyte HHL5 cell line. Furthermore, TIAM1-positive/subgroup-2 cell lines were significantly more sensitive to the TIAM1/RAC1 inhibitor NSC23766 compared with TIAM1-negative HCC lines or the normal HHL5 cell line. The results are consistent with a synthetic lethal role for TIAM1 in a methylation-defined HCC subgroup and suggest it may be a viable therapeutic target in this subset of HCC patients.

Hepatocellular carcinoma (HCC), the most common type of liver cancer, has very poor outcomes. Current therapies often have low efficacy and significant toxicities. Thus, there is a critical need for the development of novel therapeutic approaches for HCC. We have developed a novel bioinformatics pipeline, which integrates genome-wide DNA methylation and gene expression data, to identify genes required for the survival of specific molecular cancer subgroups but not normal cells. Targeting these genes may induce cancer-specific \"synthetic lethality\". Initially, five potential HCC molecular subgroups were identified based on global DNA methylation patterns. Subgroup-2 exhibited the most unique methylation profile and two candidate subtype-specific vulnerability or SL-like genes were identified for this subgroup, including TIAM1, a guanine nucleotide exchange factor encoding gene known to activate Rac1 signalling. siRNA targeting TIAM1 inhibited cell proliferation in TIAM1-positive (subgroup-2) HCC cell lines but had no effect on the normal hepatocyte HHL5 cell line. Furthermore, TIAM1-positive/subgroup-2 cell lines were significantly more sensitive to the TIAM1/RAC1 inhibitor NSC23766 compared with TIAM1-negative HCC lines or the normal HHL5 cell line. The results are consistent with a synthetic lethal role for TIAM1 in a methylation-defined HCC subgroup and suggest it may be a viable therapeutic target in this subset of HCC patients.

Precision oncology is becoming increasingly integral to clinical practice, demonstrating notable improvements in treatment outcomes. While molecular data provide comprehensive insights, obtaining such data remains costly and time-consuming. To address this challenge, we developed Path2Omics, a deep learning model that predicts gene expression and methylation from histopathology for 23 cancer types. Path2Omics was trained on 20,497 slides (9,456 formalin-fixed and paraffin-embedded (FFPE) and 11,041 fresh frozen (FF)) from 8,007 patients across 23 The Cancer Genome Atlas cohorts. When tested on FFPE slides, the most readily available format in clinical pathology practice, the integrated model outperformed its individual FF and FFPE components, robustly predicting nearly 5,000 genes on average, approximately five times more than our recently published DeepPT model. Externally evaluated on seven independent cohorts, Path2Omics robustly predicted the expression of approximately 4,400 genes, yielding a 30% increase over the FFPE model alone. Finally, we demonstrate that the inferred gene expression is nearly as effective as the actual values in predicting patient survival and treatment response. These results lay the basis for using Path2Omics to advance precision oncology from histopathology slides in a speedy and cost-effective manner.

BackgroundPrecision oncology is increasingly vital in clinical settings, improving treatment outcomes through molecular profiling. However, obtaining transcriptomic and methylation data remains expensive and time-consuming. Here, we present Path2Omics, a deep learning framework capable both gene expression and DNA methylation directly from histopathology slides across 30 cancer types.MethodsUnlike existing approaches that rely solely on FFPE slides for training, Path2Omics leverages both FFPE and FF slides by constructing two separate models: one based on the FFPE slides, called ‘FFPE model’, and another based on FF slides, called ‘FF model’. Our final ‘integrated model’ combines both predictions from the FFPE model and FF model.ResultsIn five-fold cross-validation on the TCGA cohort, FFPE models achieved an average of 3,057 well-predicted genes. The FF models, however, achieved double the predictive coverage, with an average of 6,311 well-predicted genes. To assess model generalizability, we applied the pre-trained models to 7 external datasets comprising of 1,323 slides (1,163 FFPE and 160 FF). Surprisingly, despite 6 of 7 datasets consisting solely of FFPE slides, the FF models still outperformed the FFPE model, achieving an average of 3,691 well-predicted genes compared to 3,404. The integrated model further improved performance, achieving an average of 4,391 well-predicted genes. Importantly, the well-predicted genes are strongly enriched in immune-related pathways across most cancer types, highlighting the potential to develop a predictor of patient response to immunotherapy directly from pathology slides.To demonstrate the potential translational value of the inferred gene expression, we developed models to predict patient overall survival and treatment response to cancer therapies using the inferred gene expression profiles. The results were comparable to those achieved with measured gene expression and outperformed the ‘direct’ modes, which relied solely on pathology slides without inferred gene expression as an intermediate.ConclusionsOur results demonstrate that integrating two different types of slide preparations while training gene expression and methylation predictors significantly enhance their prediction accuracy, even when only one slide type is available during inference. Our findings also highlight the model’s utility for downstream clinical tasks such as survival analysis and treatment response prediction, underscoring its potential as a scalable tool for precision oncology.

All cancers are associated with massive reorganisation of cellular epigenetic patterns, including extensive changes in the genomic patterns of DNA methylation. However, the huge scale of these changes has made it very challenging to identify key DNA methylation changes responsible for driving cancer development. Here, we present a novel approach to address this problem called methylation mapping. Through comparison of multiple types of B-lymphocyte derived malignancies and normal cell populations, this approach can define the origins of methylation changes as proliferation-driven, differentiation-driven and disease-driven (including both cancer-specific changes and cancer absent changes). Each of these categories of methylation change were found to occur at genomic regions that vary in sequence context, chromatin structure and associated transcription factors, implying underlying mechanistic differences behind the acquisition of methylation at each category. This analysis determined that only a very small fraction (about 3%) of DNA methylation changes in B-cell cancers are disease related, with the overwhelming majority (97%) being driven by normal biological processes, predominantly cell proliferation. Furthermore, the low level of true disease-specific changes can potentially simplify identification of functionally relevant DNA methylation changes, allowing identification of previously unappreciated candidate drivers of cancer development, as illustrated here by the identification and functional confirmation of SLC22A15 as a novel tumour suppressor candidate in acute lymphoblastic leukaemia. Overall, this approach should lead to a clearer understanding of the role of altered DNA methylation in cancer development, facilitate the identification of DNA methylation targeted genes with genuine functional roles in cancer development and thus identify novel therapeutic targets.