The utility of color normalization for AI-based diagnosis of hematoxylin and eosin-stained pathology images.

The color variation of hematoxylin and eosin (H&E)-stained tissues has presented a challenge for applications of artificial intelligence (AI) in digital pathology. Many color normalization algorithms have been developed in recent years in order to reduce the color variation between H&E images. However, previous efforts in benchmarking these algorithms have produced conflicting results and none have sufficiently assessed the efficacy of the various color normalization methods for improving diagnostic performance of AI systems. In this study, we systematically investigated eight color normalization algorithms for AI-based classification of H&E-stained histopathology slides, in the context of using images both from one center and from multiple centers. Our results show that color normalization does not consistently improve classification performance when both training and testing data are from a single center. However, using four multi-center datasets of two cancer types (ovarian and pleural) and objective functions, we show that color normalization can significantly improve the classification accuracy of images from external datasets (ovarian cancer: 0.25 AUC increase, p = 1.6 e-05; pleural cancer: 0.21 AUC increase, p = 1.4 e-10). Furthermore, we introduce a novel augmentation strategy by mixing color-normalized images using three easily accessible algorithms that consistently improves the diagnosis of test images from external centers, even when the individual normalization methods had varied results. We anticipate our study to be a starting point for reliable use of color normalization to improve AI-based, digital pathology-empowered diagnosis of cancers sourced from multiple centers. © 2021 The Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Synthesis of diagnostic quality cancer pathology images by generative adversarial networks.

Deep learning-based computer vision methods have recently made remarkable breakthroughs in the analysis and classification of cancer pathology images. However, there has been relatively little investigation of the utility of deep neural networks to synthesize medical images. In this study, we evaluated the efficacy of generative adversarial networks to synthesize high-resolution pathology images of 10 histological types of cancer, including five cancer types from The Cancer Genome Atlas and the five major histological subtypes of ovarian carcinoma. The quality of these images was assessed using a comprehensive survey of board-certified pathologists (n = 9) and pathology trainees (n = 6). Our results show that the real and synthetic images are classified by histotype with comparable accuracies and the synthetic images are visually indistinguishable from real images. Furthermore, we trained deep convolutional neural networks to diagnose the different cancer types and determined that the synthetic images perform as well as additional real images when used to supplement a small training set. These findings have important applications in proficiency testing of medical practitioners and quality assurance in clinical laboratories. Furthermore, training of computer-aided diagnostic systems can benefit from synthetic images where labeled datasets are limited (e.g. rare cancers). We have created a publicly available website where clinicians and researchers can attempt questions from the image survey (http://gan.aimlab.ca/). © 2020 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Integrative genomic analysis of matched primary and metastatic pediatric osteosarcoma.

Despite being the most common childhood bone tumor, the genomic characterization of osteosarcoma remains incomplete. In particular, very few osteosarcoma metastases have been sequenced to date, critical to better understand mechanisms of progression and evolution in this tumor. We performed an integrated whole genome and exome sequencing analysis of paired primary and metastatic pediatric osteosarcoma specimens to identify recurrent genomic alterations. Sequencing of 13 osteosarcoma patients including 13 primary, 10 metastatic, and 3 locally recurring tumors revealed a highly heterogeneous mutational landscape, including cases of hypermutation and microsatellite instability positivity, but with virtually no recurrent alterations except for mutations involving the tumor suppressor genes RB1 and TP53. At the germline level, we detected alterations in multiple cancer related genes in the majority of the cohort, including those potentially disrupting DNA damage response pathways. Metastases retained only a minimal number of short variants from their corresponding primary tumors, while copy number alterations showed higher conservation. One recurrently amplified gene, KDR, was highly expressed in advanced cases and associated with poor prognosis. © 2019 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Retrospective review using targeted deep sequencing reveals mutational differences between gastroesophageal junction and gastric carcinomas.

BACKGROUND: Adenocarcinomas of both the gastroesophageal junction and stomach are molecularly complex, but differ with respect to epidemiology, etiology and survival. There are few data directly comparing the frequencies of single nucleotide mutations in cancer-related genes between the two sites. Sequencing of targeted gene panels may be useful in uncovering multiple genomic aberrations using a single test. METHODS: DNA from 92 gastroesophageal junction and 75 gastric adenocarcinoma resection specimens was extracted from formalin-fixed paraffin-embedded tissue. Targeted deep sequencing of 46 cancer-related genes was performed through emulsion PCR followed by semiconductor-based sequencing. Gastroesophageal junction and gastric carcinomas were contrasted with respect to mutational profiles, immunohistochemistry and in situ hybridization, as well as corresponding clinicopathologic data. RESULTS: Gastroesophageal junction carcinomas were associated with younger age, more frequent intestinal-type histology, more frequent p53 overexpression, and worse disease-free survival on multivariable analysis. Among all cases, 145 mutations were detected in 31 genes. TP53 mutations were the most common abnormality detected, and were more common in gastroesophageal junction carcinomas (42% vs. 27%, p = 0.036). Mutations in the Wnt pathway components APC and CTNNB1 were more common among gastric carcinomas (16% vs. 3%, p = 0.006), and gastric carcinomas were more likely to have ≥3 driver mutations detected (11% vs. 2%, p = 0.044). Twenty percent of cases had potentially actionable mutations identified. R132H and R132C missense mutations in the IDH1 gene were observed, and are the first reported mutations of their kind in gastric carcinoma. CONCLUSIONS: Panel sequencing of routine pathology material can yield mutational information on several driver genes, including some for which targeted therapies are available. Differing rates of mutations and clinicopathologic differences support a distinction between adenocarcinomas that arise in the gastroesophageal junction and those that arise in the stomach proper.

Transcriptome resources and functional characterization of monoterpene synthases for two host species of the mountain pine beetle, lodgepole pine (Pinus contorta) and jack pine (Pinus banksiana).

BACKGROUND: The mountain pine beetle (MPB, Dendroctonus ponderosae) epidemic has affected lodgepole pine (Pinus contorta) across an area of more than 18 million hectares of pine forests in western Canada, and is a threat to the boreal jack pine (Pinus banksiana) forest. Defence of pines against MPB and associated fungal pathogens, as well as other pests, involves oleoresin monoterpenes, which are biosynthesized by families of terpene synthases (TPSs). Volatile monoterpenes also serve as host recognition cues for MPB and as precursors for MPB pheromones. The genes responsible for terpene biosynthesis in jack pine and lodgepole pine were previously unknown. RESULTS: We report the generation and quality assessment of assembled transcriptome resources for lodgepole pine and jack pine using Sanger, Roche 454, and Illumina sequencing technologies. Assemblies revealed transcripts for approximately 20,000 - 30,000 genes from each species and assembly analyses led to the identification of candidate full-length prenyl transferase, TPS, and P450 genes of oleoresin biosynthesis. We cloned and functionally characterized, via expression of recombinant proteins in E. coli, nine different jack pine and eight different lodgepole pine mono-TPSs. The newly identified lodgepole pine and jack pine mono-TPSs include (+)-α-pinene synthases, (-)-α-pinene synthases, (-)-ß-pinene synthases, (+)-3-carene synthases, and (-)-ß-phellandrene synthases from each of the two species. CONCLUSION: In the absence of genome sequences, transcriptome assemblies are important for defence gene discovery in lodgepole pine and jack pine, as demonstrated here for the terpenoid pathway genes. The product profiles of the functionally annotated mono-TPSs described here can account for the major monoterpene metabolites identified in lodgepole pine and jack pine.

Integrated genome and transcriptome sequencing identifies a novel form of hybrid and aggressive prostate cancer.

Next-generation sequencing is making sequence-based molecular pathology and personalized oncology viable. We selected an individual initially diagnosed with conventional but aggressive prostate adenocarcinoma and sequenced the genome and transcriptome from primary and metastatic tissues collected prior to hormone therapy. The histology-pathology and copy number profiles were remarkably homogeneous, yet it was possible to propose the quadrant of the prostate tumour that likely seeded the metastatic diaspora. Despite a homogeneous cell type, our transcriptome analysis revealed signatures of both luminal and neuroendocrine cell types. Remarkably, the repertoire of expressed but apparently private gene fusions, including C15orf21:MYC, recapitulated this biology. We hypothesize that the amplification and over-expression of the stem cell gene MSI2 may have contributed to the stable hybrid cellular identity. This hybrid luminal-neuroendocrine tumour appears to represent a novel and highly aggressive case of prostate cancer with unique biological features and, conceivably, a propensity for rapid progression to castrate-resistance. Overall, this work highlights the importance of integrated analyses of genome, exome and transcriptome sequences for basic tumour biology, sequence-based molecular pathology and personalized oncology.

A high-throughput pipeline for DNA/RNA/small RNA purification from tissue samples for sequencing.

High-throughput total nucleic acid (TNA) purification methods based on solid-phase reversible immobilization (SPRI) beads produce TNA suitable for both genomic and transcriptomic applications. Even so, small RNA species, including miRNA, bind weakly to SPRI beads under standard TNA purification conditions, necessitating a separate workflow using column-based methods that are difficult to automate. Here, an SPRI-based high-throughput TNA purification protocol that recovers DNA, RNA and small RNA, called GSC-modified RLT+ Aline bead-based protocol (GRAB-ALL), which incorporates modifications to enhance small RNA recovery is presented. GRAB-ALL was benchmarked against existing nucleic acid purification workflows and GRAB-ALL efficiently purifies TNA, including small RNA, for next-generation sequencing applications in a plate-based format suitable for automated high-throughput sample preparation.

Using next-generation sequencing for the diagnosis of rare disorders: a family with retinitis pigmentosa and skeletal abnormalities.

Linkage analysis with subsequent candidate gene sequencing is typically used to diagnose novel inherited syndromes. It is now possible to expedite diagnosis through the sequencing of all coding regions of the genome (the exome) or full genomes. We sequenced the exomes of four members of a family presenting with spondylo-epiphyseal dysplasia and retinitis pigmentosa and identified a six-base-pair (6-bp) deletion in GNPTG, the gene implicated in mucolipidosis type IIIγ. The diagnosis was confirmed by biochemical studies and both broadens the mucolipidosis type III phenotype and demonstrates the clinical utility of next-generation sequencing to diagnose rare genetic diseases.

The impact of whole genome and transcriptome analysis (WGTA) on predictive biomarker discovery and diagnostic accuracy of advanced malignancies.

In this study, we evaluate the impact of whole genome and transcriptome analysis (WGTA) on predictive molecular profiling and histologic diagnosis in a cohort of advanced malignancies. WGTA was used to generate reports including molecular alterations and site/tissue of origin prediction. Two reviewers analyzed genomic reports, clinical history, and tumor pathology. We used National Comprehensive Cancer Network (NCCN) consensus guidelines, Food and Drug Administration (FDA) approvals, and provincially reimbursed treatments to define genomic biomarkers associated with approved targeted therapeutic options (TTOs). Tumor tissue/site of origin was reassessed for most cases using genomic analysis, including a machine learning algorithm (Supervised Cancer Origin Prediction Using Expression [SCOPE]) trained on The Cancer Genome Atlas data. WGTA was performed on 652 cases, including a range of primary tumor types/tumor sites and 15 malignant tumors of uncertain histogenesis (MTUH). At the time WGTA was performed, alterations associated with an approved TTO were identified in 39 (6%) cases; 3 of these were not identified through routine pathology workup. In seven (1%) cases, the pathology workup either failed, was not performed, or gave a different result from the WGTA. Approved TTOs identified by WGTA increased to 103 (16%) when applying 2021 guidelines. The histopathologic diagnosis was reviewed in 389 cases and agreed with the diagnostic consensus after WGTA in 94% of non-MTUH cases (n = 374). The remainder included situations where the morphologic diagnosis was changed based on WGTA and clinical data (0.5%), or where the WGTA was non-contributory (5%). The 15 MTUH were all diagnosed as specific tumor types by WGTA. Tumor board reviews including WGTA agreed with almost all initial predictive molecular profile and histopathologic diagnoses. WGTA was a powerful tool to assign site/tissue of origin in MTUH. Current efforts focus on improving therapeutic predictive power and decreasing cost to enhance use of WGTA data as a routine clinical test.

A new frontier in personalized cancer therapy: mapping molecular changes.

Mutations in the genome of a normal cell can affect the function of its many genes and pathways. These alterations could eventually transform the cell from a normal to a malignant state by allowing an uncontrolled proliferation of the cell and formation of a cancer tumor. Each tumor in an individual patient can have hundreds of mutated genes and perturbed pathways. Cancers clinically presenting as the same type or subtype could potentially be very different at the molecular level and thus behave differently in response to therapy. The challenge is to distinguish the key mutations driving the cancer from the background of mutational noise and find ways to effectively target them. The promise is that such a molecular approach to classifying cancer will lead to better diagnostic, prognostic and personalized treatment strategies. This article provides an overview of advances in the molecular characterization of cancers and their applications in therapy.

Rare loss-of-function variants in type I IFN immunity genes are not associated with severe COVID-19.

A recent report found that rare predicted loss-of-function (pLOF) variants across 13 candidate genes in TLR3- and IRF7-dependent type I IFN pathways explain up to 3.5% of severe COVID-19 cases. We performed whole-exome or whole-genome sequencing of 1,864 COVID-19 cases (713 with severe and 1,151 with mild disease) and 15,033 ancestry-matched population controls across 4 independent COVID-19 biobanks. We tested whether rare pLOF variants in these 13 genes were associated with severe COVID-19. We identified only 1 rare pLOF mutation across these genes among 713 cases with severe COVID-19 and observed no enrichment of pLOFs in severe cases compared to population controls or mild COVID-19 cases. We found no evidence of association of rare LOF variants in the 13 candidate genes with severe COVID-19 outcomes.

Epigenomic programming in early fetal brain development.

Aim: To provide a comprehensive understanding of gene regulatory networks in the developing human brain and a foundation for interpreting pathogenic deregulation. Materials & methods: We generated reference epigenomes and transcriptomes of dissected brain regions and primary neural progenitor cells (NPCs) derived from cortical and ganglionic eminence tissues of four normal human fetuses. Results: Integration of these data across developmental stages revealed a directional increase in active regulatory states, transcription factor activities and gene transcription with developmental stage. Consistent with differences in their biology, NPCs derived from cortical and ganglionic eminence regions contained common, region specific, and gestational week specific regulatory states. Conclusion: We provide a high-resolution regulatory network for NPCs from different brain regions as a comprehensive reference for future studies.

A high-throughput protocol for isolating cell-free circulating tumor DNA from peripheral blood.

The analysis of cell-free circulating tumor DNA (ctDNA) is potentially a less invasive, more dynamic assessment of cancer progression and treatment response than characterizing solid tumor biopsies. Standard isolation methods require separation of plasma by centrifugation, a time-consuming step that complicates automation. To address these limitations, we present an automatable magnetic bead-based ctDNA isolation method that eliminates centrifugation to purify ctDNA directly from peripheral blood (PB). To develop and test our method, ctDNA from cancer patients was purified from PB and plasma. We found that allelic fractions of somatic single-nucleotide variants from target gene capture libraries were comparable, indicating that the PB ctDNA purification method may be a suitable replacement for the plasma-based protocols currently in use.

A pan-cancer analysis of alternative splicing events reveals novel tumor-associated splice variants of matriptase.

High-throughput transcriptome sequencing allows identification of cancer-related changes that occur at the stages of transcription, pre-messenger RNA (mRNA), and splicing. In the current study, we devised a pipeline to predict novel alternative splicing (AS) variants from high-throughput transcriptome sequencing data and applied it to large sets of tumor transcriptomes from The Cancer Genome Atlas (TCGA). We identified two novel tumor-associated splice variants of matriptase, a known cancer-associated gene, in the transcriptome data from epithelial-derived tumors but not normal tissue. Most notably, these variants were found in 69% of lung squamous cell carcinoma (LUSC) samples studied. We confirmed the expression of matriptase AS transcripts using quantitative reverse transcription PCR (qRT-PCR) in an orthogonal panel of tumor tissues and cell lines. Furthermore, flow cytometric analysis confirmed surface expression of matriptase splice variants in chinese hamster ovary (CHO) cells transiently transfected with cDNA encoding the novel transcripts. Our findings further implicate matriptase in contributing to oncogenic processes and suggest potential novel therapeutic uses for matriptase splice variants.

A novel recurrent mutation in ATP1A3 causes CAPOS syndrome.

BACKGROUND: We undertook genetic analysis of three affected families to identify the cause of dominantly-inherited CAPOS (cerebellar ataxia, areflexia, pes cavus, optic atrophy and sensorineural hearing loss) syndrome. METHODS: We used whole-exome sequencing to analyze two families affected with CAPOS syndrome, including the original family reported in 1996, and Sanger sequencing to assess familial segregation of rare variants identified in the probands and in a third, apparently unrelated family with CAPOS syndrome. RESULTS: We found an identical heterozygous missense mutation, c.2452G > A (p.(Glu818Lys)), in the Na⁺/K⁺ ATPase α3(ATP1A3) gene in the proband and his affected sister and mother, but not in either unaffected maternal grandparent, in the first family. The same mutation was also identified in the proband and three other affected members of the second family and in all three affected members of the third family. This mutation was not found in more than 3600 chromosomes from unaffected individuals. CONCLUSION: Other mutations in ATP1A3 have previously been demonstrated to cause rapid-onset dystonia-parkinsonism (also called dystonia-12) or alternating hemiplegia of childhood. This study shows that an allelic mutation in ATP1A3 produces CAPOS syndrome.

Distinct roles of KAP1, HP1 and G9a/GLP in silencing of the two-cell-specific retrotransposon MERVL in mouse ES cells.

BACKGROUND: In mouse embryonic stem cells (mESCs), transcriptional silencing of numerous class I and II endogenous retroviruses (ERVs), including IAP, ETn and MMERVK10C, is dependent upon the H3K9 methyltransferase (KMTase) SETDB1/ESET and its binding partner KAP1/TRIM28. In contrast, the H3K9 KMTases G9a and GLP and HP1 proteins are dispensable for this process. Intriguingly, MERVL retroelements are actively transcribed exclusively in the two-cell (2C) embryo, but the molecular basis of silencing of these class III ERVs at later developmental stages has not been systematically addressed. RESULTS: Here, we characterized the roles of these chromatin factors in MERVL silencing in mESCs. While MMERVK10C and IAP ERVs are bound by SETDB1 and KAP1 and are induced following their deletion, MERVL ERVs show relatively low levels of SETDB1 and KAP1 binding and are upregulated exclusively following KAP1 depletion, indicating that KAP1 influences MERVL expression independent of SETDB1. In contrast to class I and class II ERVs, MERVL and MERVL LTR-driven genic transcripts are also upregulated following depletion of G9a or GLP, and G9a binds directly to these ERVs. Consistent with a direct role for H3K9me2 in MERVL repression, these elements are highly enriched for G9a-dependent H3K9me2, and catalytically active G9a is required for silencing of MERVL LTR-driven transcripts. MERVL is also derepressed in HP1α and HP1ß KO ESCs. However, like KAP1, HP1α and HP1ß are only modestly enriched at MERVL relative to IAP LTRs. Intriguingly, as recently shown for KAP1, RYBP, LSD1 and G9a-deficient mESCs, many genes normally expressed in the 2C embryo are also induced in HP1 KO mESCs, revealing that aberrant expression of a subset of 2C-specific genes is a common feature in each of these KO lines. CONCLUSIONS: Our results indicate that G9a and GLP, which are not required for silencing of class I and II ERVs, are recruited to MERVL elements and play a direct role in silencing of these class III ERVs, dependent upon G9a catalytic activity. In contrast, induction of MERVL expression in KAP1, HP1α and HP1ß KO ESCs may occur predominantly as a consequence of indirect effects, in association with activation of a subset of 2C-specific genes.

Drug repositioning for personalized medicine.

Human diseases can be caused by complex mechanisms involving aberrations in numerous proteins and pathways. With recent advances in genomics, elucidating the molecular basis of disease on a personalized level has become an attainable goal. In many cases, relevant molecular targets will be identified for which approved drugs already exist, and the potential repositioning of these drugs to a new indication can be investigated. Repositioning is an accelerated route for drug discovery because existing drugs have established clinical and pharmacokinetic data. Personalized medicine and repositioning both aim to improve the productivity of current drug discovery pipelines, which expend enormous time and cost to develop new drugs, only to have them fail in clinical trials because of lack of efficacy or toxicity. Here, we discuss the current state of research in these two fields, focusing on recent large-scale efforts to systematically find repositioning candidates and elucidate individual disease mechanisms in cancer. We also discuss scenarios in which personalized drug repositioning could be particularly rewarding, such as for diseases that are rare or have specific mutations, as well as current challenges in this field. With an increasing number of drugs being approved for rare cancer subtypes, personalized medicine and repositioning approaches are poised to significantly alter the way we diagnose diseases, infer treatments and develop new drugs.

Gene expression profiling of oxidative stress response of C. elegans aging defective AMPK mutants using massively parallel transcriptome sequencing.

BACKGROUND: A strong association between stress resistance and longevity in multicellular organisms has been established as many mutations that extend lifespan also show increased resistance to stress. AAK-2, the C. elegans homolog of an alpha subunit of AMP-activated protein kinase (AMPK) is an intracellular fuel sensor that regulates cellular energy homeostasis and functions in stress resistance and lifespan extension. FINDINGS: Here, we investigated global transcriptional responses of aak-2 mutants to oxidative stress and in turn identified potential downstream targets of AAK-2 involved in stress resistance in C. elegans. We employed massively parallel Illumina sequencing technology and performed comprehensive comparative transcriptome analysis. Specifically, we compared the transcriptomes of aak-2 and wild type animals under normal conditions and conditions of induced oxidative stress. This research has presented a snapshot of genome-wide transcriptional activities that take place in C. elegans in response to oxidative stress both in the presence and absence of AAK-2. CONCLUSIONS: The analysis presented in this study has enabled us to identify potential genes involved in stress resistance that may be either directly or indirectly under the control of AAK-2. Furthermore, we have extended our current knowledge of general defense responses of C. elegans against oxidative stress supporting the function for AAK-2 in inhibition of biosynthetic processes, especially lipid synthesis, under oxidative stress and transcriptional regulation of genes involved in reproductive processes.

Evolution of an adenocarcinoma in response to selection by targeted kinase inhibitors.

BACKGROUND: Adenocarcinomas of the tongue are rare and represent the minority (20 to 25%) of salivary gland tumors affecting the tongue. We investigated the utility of massively parallel sequencing to characterize an adenocarcinoma of the tongue, before and after treatment. RESULTS: In the pre-treatment tumor we identified 7,629 genes within regions of copy number gain. There were 1,078 genes that exhibited increased expression relative to the blood and unrelated tumors and four genes contained somatic protein-coding mutations. Our analysis suggested the tumor cells were driven by the RET oncogene. Genes whose protein products are targeted by the RET inhibitors sunitinib and sorafenib correlated with being amplified and or highly expressed. Consistent with our observations, administration of sunitinib was associated with stable disease lasting 4 months, after which the lung lesions began to grow. Administration of sorafenib and sulindac provided disease stabilization for an additional 3 months after which the cancer progressed and new lesions appeared. A recurring metastasis possessed 7,288 genes within copy number amplicons, 385 genes exhibiting increased expression relative to other tumors and 9 new somatic protein coding mutations. The observed mutations and amplifications were consistent with therapeutic resistance arising through activation of the MAPK and AKT pathways. CONCLUSIONS: We conclude that complete genomic characterization of a rare tumor has the potential to aid in clinical decision making and identifying therapeutic approaches where no established treatment protocols exist. These results also provide direct in vivo genomic evidence for mutational evolution within a tumor under drug selection and potential mechanisms of drug resistance accrual.

De novo genome sequence assembly of a filamentous fungus using Sanger, 454 and Illumina sequence data.

Sequencing-by-synthesis technologies can reduce the cost of generating de novo genome assemblies. We report a method for assembling draft genome sequences of eukaryotic organisms that integrates sequence information from different sources, and demonstrate its effectiveness by assembling an approximately 32.5 Mb draft genome sequence for the forest pathogen Grosmannia clavigera, an ascomycete fungus. We also developed a method for assessing draft assemblies using Illumina paired end read data and demonstrate how we are using it to guide future sequence finishing. Our results demonstrate that eukaryotic genome sequences can be accurately assembled by combining Illumina, 454 and Sanger sequence data.