Clinical Annotation Research Articles

The N6-methyladenosine Epitranscriptomic Landscape of Lung Adenocarcinoma.

Comprehensive m6A epitranscriptome profiling of primary tumors remains largely uncharted. Here, we profiled the m6A epitranscriptome of 10 non-neoplastic lung (NL) tissues and 51 lung adenocarcinoma (LUAD) tumors, integrating the corresponding transcriptome, proteome and extensive clinical annotations. We identified distinct clusters and genes that were exclusively linked to disease progression through m6A modifications. In comparison with NL tissues, we identified 430 transcripts to be hypo-methylated and 222 to be hyper-methylated in tumors. Among these genes, EML4 emerged as a novel metastatic driver, displaying significant hyper-methylation in tumors. m6A modification promoted the translation of EML4, leading to its widespread overexpression in primary tumors. Functionally, EML4 modulated cytoskeleton dynamics through interacting with ARPC1A, enhancing lamellipodia formation, cellular motility, local invasion, and metastasis. Clinically, high EML4 protein abundance correlated with features of metastasis. METTL3 small molecule inhibitor markedly diminished both EML4 m6A and protein abundance, and efficiently suppressed lung metastases in vivo.

9 Delving into molecular underpinnings of sarcomatoid/rhabdoid dedifferentioation in renal cancers using spatial profiling

Abstract Background Renal cell carcinomas (RCC) are characterized by their largely diverse clinical outcomes. Leveraging genomic diversity among individual tumors, our research during the past decade has concentrated on developing personalized medicine approaches for the most common and the most aggressive type of RCC, clear cell renal carcinoma (ccRCC). We generated the largest dataset of ccRCC including somatic genomic and clinical annotations for over 940 ccRCC patients, and developed a genomic classifier, based on mutational status of 12 RCC-relevant genes, which is able to stratify patients according to their risk of relapse after nephrectomy and/or death due to RCC. Furthermore, we noticed the presence of mesenchymal-like cellular phenotypes in tumors of high-risk patients, representsing a de-differentiation process, known as sarcomatoic or rhabdoid (S/R) de-differentation. Methods The current knowledge about molecular mechanisms that may drive S/R de-differentiation is primarily generated from molecular profiling of bulk tumors, collecting data from a mixture of cells that co-exist in the tumor milieu. Therefore, high-resolution studies that precisely define molecular characteristics of different cellular phenotypes associated with S/R features are missing. We have used spatial transcriptomic profiling to investigate molecular mechanisms that underline S/R de-differentationin RCC. While preserving tissue context, we applied spatial whole human transcriptome profiling to areas exhibiting S/R or clear cell phenotypes within the same tumor specimens to generate phenotype-specific transcriptome profiles. In addition, we applied whole-exome sequencing (WES) to independent areas that exhibit different differentiation phenotype withing same tumors. Results Pathway and network analysis of genesets with upregulation in each area have revealed meaningful differences in cellular pathways which are active in each phenotype. Specifically, we observed that dysregulation of extracellular matrix (ECM) is a hallmark of S/R areas within RCC tumors. Furthermore, WES of phenotypically-distinct areas within each tumor has shed new light son the genome evoltuion of S/R phenotypes. Conclusions S/R dedifferentiation in RCC tumors is characterized by specific genomic evolutionary patterns and substantial dysregulation of ECM components. DOD CDMRP Funding: yes

AURKB promotes immunogenicity and immune infiltration in clear cell renal cell carcinoma

BackgroundChromatin regulators (CRs) are capable of causing epigenetic alterations, which are significant features of cancer. However, the function of CRs in controlling Clear Cell Renal Cell Carcinoma (ccRCC) is not well understood. This research aims to discover a CRs prognostic signature in ccRCC and to elucidate the roles of CRs-related genes in tumor microenvironment (TME).MethodsExpression profiles and relevant clinical annotations were retrieved from the Cancer Genome Atlas (TCGA) and UCSC Xena platform for progression-free survival (PFS) data. The R package "limma" was used to identify differentially expressed CRs. A predictive model based on five CRs was developed using LASSO-Cox analysis. The model's predictive power and applicability were validated using K-M curves, ROC curves, nomograms, comparisons with other models, stratified survival analyses, and validation with the ICGC cohort. GO and GSEA analyses were performed to investigate mechanisms differentiating low and high riskScore groups. Immunogenicity was assessed using Tumor Mutational Burden (TMB), immune cell infiltrations were inferred, and immunotherapy was evaluated using immunophenogram analysis and the expression patterns of human leukocyte antigen (HLA) and checkpoint genes. Differentially expressed CRs (DECRs) between low and high riskScore groups were identified using log2|FC|> 1 and FDR < 0.05. AURKB, one of the high-risk DECRs and a component of our prognostic model, was selected for further analysis.ResultsWe constructed a 5 CRs signature, which demonstrated a strong capacity to predict survival and greater applicability in ccRCC. Elevated immunogenicity and immune infiltration in the high riskScore group were associated with poor prognosis. Immunotherapy was more effective in the high riskScore group, and certain chemotherapy medications, including cisplatin, docetaxel, bleomycin, and axitinib, had lower IC50 values. Our research shows that AURKB is critical for the immunogenicity and immune infiltration of the high riskScore group.ConclusionOur study produced a reliable prognostic prediction model using only 5 CRs. We found that AURKB promotes immunogenicity and immune infiltration. This research provides crucial support for the development of prognostic biomarkers and treatment strategies for ccRCC.

Clinical signatures of genetic epilepsies precede diagnosis in electronic medical records of 32,000 individuals

PurposeAn early genetic diagnosis can guide the time-sensitive treatment of individuals with genetic epilepsies. However, most genetic diagnoses occur long after disease onset. We aimed to identify early clinical features suggestive of genetic diagnoses in individuals with epilepsy through large-scale analysis of full-text electronic medical records. MethodsWe extracted 89 million time-stamped standardized clinical annotations using Natural Language Processing from 4,572,783 clinical notes from 32,112 individuals with childhood epilepsy, including 1925 individuals with known or presumed genetic epilepsies. We applied these features to train random forest models to predict SCN1A-related disorders and any genetic diagnosis. ResultsWe identified 47,774 age-dependent associations of clinical features with genetic etiologies a median of 3.6 years before molecular diagnosis. Across all 710 genetic etiologies identified in our cohort, neurodevelopmental differences between 6 to 9 months increased the likelihood of a later molecular diagnosis 5-fold (P < .0001, 95% CI = 3.55-7.42). A later diagnosis of SCN1A-related disorders (area under the curve [AUC] = 0.91) or an overall positive genetic diagnosis (AUC = 0.82) could be reliably predicted using random forest models. ConclusionClinical features predictive of genetic epilepsies precede molecular diagnoses by up to several years in conditions with known precision treatments. An earlier diagnosis facilitated by automated electronic medical records analysis has the potential for earlier targeted therapeutic strategies in the genetic epilepsies.

Development of a prognostic risk model of uveal melanoma based on N7-methylguanosine-related regulators

BackgroundUveal melanoma (UVM) stands as the predominant type of primary intraocular malignancy among adults. The clinical significance of N7-methylguanosine (m7G), a prevalent RNA modifications, in UVM remains unclear.MethodsPrimary information from 80 UVM patients were analyzed as the training set, incorporating clinical information, mutation annotations and mRNA expression obtained from The Cancer Genome Atlas (TCGA) website. The validation set was carried out using Gene Expression Omnibus (GEO) database GSE22138 and GSE84976. Kaplan–Meier and Cox regression of univariate analyses were subjected to identify m7G-related regulators as prognostic genes.ResultA prognostic risk model comprising EIF4E2, NUDT16, SNUPN and WDR4 was established through Cox regression of LASSO. Evaluation of the model’s predictability for UVM patients’ prognosis by Receiver Operating Characteristic (ROC) curves in the training set, demonstrated excellent performance Area Under the Curve (AUC) > 0.75. The high-risk prognosis within the TCGA cohort exhibit a notable worse outcome. Additionally, an independent correlation between the risk score and overall survival (OS) among UVM patients were identified. External validation of this model was carried out using the validation sets (GSE22138 and GSE84976). Immune-related analysis revealed that patients with high score of m7G-related risk model exhibited elevated level of immune infiltration and immune checkpoint gene expression.ConclusionWe have developed a risk prediction model based on four m7G-related regulators, facilitating effective estimate UVM patients’ survival by clinicians. Our findings shed novel light on essential role of m7G-related regulators in UVM and suggest potential novel targets for the diagnosis, prognosis and therapy of UVM.

Immuno-oncologic profiling of pediatric brain tumors reveals major clinical significance of the tumor immune microenvironment

With the success of immunotherapy in cancer, understanding the tumor immune microenvironment (TIME) has become increasingly important; however in pediatric brain tumors this remains poorly characterized. Accordingly, we developed a clinical immune-oncology gene expression assay and used it to profile a diverse range of 1382 samples with detailed clinical and molecular annotation. In low-grade gliomas we identify distinct patterns of immune activation with prognostic significance in BRAF V600E-mutant tumors. In high-grade gliomas, we observe immune activation and T-cell infiltrates in tumors that have historically been considered immune cold, as well as genomic correlates of inflammation levels. In mismatch repair deficient high-grade gliomas, we find that high tumor inflammation signature is a significant predictor of response to immune checkpoint inhibition, and demonstrate the potential for multimodal biomarkers to improve treatment stratification. Importantly, while overall patterns of immune activation are observed for histologically and genetically defined tumor types, there is significant variability within each entity, indicating that the TIME must be evaluated as an independent feature from diagnosis. In sum, in addition to the histology and molecular profile, this work underscores the importance of reporting on the TIME as an essential axis of cancer diagnosis in the era of personalized medicine.

Differential NEUROD1, ASCL1, and POU2F3 Expression Defines Molecular Subsets of Bladder Small Cell/Neuroendocrine Carcinoma With Prognostic Implications

Small cell carcinomas (SMC) of the lung are now molecularly classified based on the expression of transcriptional regulators (NEUROD1, ASCL1, POU2F3, and YAP1) and DLL3, which has emerged as an investigational therapeutic target. PLCG2 has been shown to identify a distinct subpopulation of lung SMC with stem cell-like and prometastasis features and poor prognosis. We analyzed the expression of these novel neuroendocrine markers and their association with traditional neuroendocrine markers and patient outcomes in a cohort of bladder neuroendocrine carcinoma (NEC) consisting of 103 SMC and 19 large cell NEC (LCNEC) assembled in tissue microarrays. Coexpression patterns were assessed and integrated with detailed clinical annotation including overall (OS) and recurrence-free survival (RFS) and response to neoadjuvant/adjuvant chemotherapy. We identified 5 distinct molecular subtypes in bladder SMC based on the expression of ASCL1, NEUROD1, and POU2F3: ASCL1+/NEUROD1− (n = 33; 34%), ASCL1− /NEUROD1+ (n = 21; 21%), ASCL1+/NEUROD1+ (n = 17; 17%), POU2F3+ (n = 22, 22%), and ASCL1− /NEUROD1− /POU2F3− (n = 5, 5%). POU2F3+ tumors were mutually exclusive with those expressing ASCL1 and NEUROD1 and exhibited lower expression of traditional neuroendocrine markers. PLCG2 expression was noted in 33 tumors (32%) and was highly correlated with POU2F3 expression (P < .001). DLL3 expression was high in both SMC (n = 72, 82%) and LCNEC (n = 11, 85%). YAP1 expression was enriched in nonneuroendocrine components and negatively correlated with all neuroendocrine markers. In patients without metastatic disease who underwent radical cystectomy, PLCG2+ or POU2F3+ tumors had shorter RFS and OS (P < .05), but their expression was not associated with metastasis status or response to neoadjuvant/adjuvant chemotherapy. In conclusion, the NEC of the bladder can be divided into distinct molecular subtypes based on the expression of ASCL1, NEUROD1, and POU2F3. POU2F3-expressing tumors represent an ASCL1/NEUROD1-negative subset of bladder NEC characterized by lower expression of traditional neuroendocrine markers. Marker expression patterns were similar in SMC and LCNEC. Expression of PLCG2 and POU2F3 was associated with shorter RFS and OS. DLL3 was expressed at high levels in both SMC and LCNEC of the bladder, nominating it as a potential therapeutic target.

Role of TAP1 in the identification of immune-hot tumor microenvironment and its prognostic significance for immunotherapeutic efficacy in gastric carcinoma.

Gastric cancer (GC), a multifaceted gastrointestinal malignancy, is the fourth most prevalent contributor to cancer-related fatalities globally. As a member of the ATP-binding cassette (ABC) family, transporter associated with antigen processing 1 (TAP1) is crucial for conveying antigen peptides from the cytoplasm to the lumen of the endoplasmic reticulum and subsequently loading them onto the major histocompatibility complex (MHC) class I molecules. Recent studies have established the biological significance of TAP1 in upholding tumor survival and facilitating immune evasion by remodeling the tumor microenvironment (TME) and orchestrating immune infiltration. The study was conducted to elucidate the association of TAP1 expression with immunological characteristics, and sought to exploit the value of TAP1 as a biomarker reflecting the inflamed TME and immunotherapeutic response. RNA-sequencing profiles and clinical annotations were obtained from The Cancer Genome Atlas-stomach adenocarcinoma (TCGA-STAD) cohort and Gene Expression Omnibus (GEO) portal. Preprocessing was conducting using the limma package. Weighted gene co-expression network analysis (WGCNA) was used to identify gene modules and TAP1 co-expressed genes (CEGs) based on correlation patterns. Consensus clustering and silhouette analysis determined the optimal number of TAP1-related groups. Gene expression profiles were integrated and classified using the pamr package. The Estimation of STromal and Immune cells in MAlignant Tumors using Expression data (ESTIMATE) algorithm and single-sample gene set enrichment analysis (ssGSEA) were used to evaluate immunological characteristics. Differential expression analysis was conducted using the limma package. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed. Single-cell RNA sequencing (scRNA-seq) datasets were analyzed using the Seurat toolkit to characterize cell types. Within this investigation, no significant differences in TAP1 expression were observed among patients exhibiting various clinicopathological features, indicating that TAP1 expression was not specific to molecular subtypes. Subsequent analysis revealed a positive correlation between TAP1 and diverse immunological traits, encompassing immunomodulators, tumor-infiltrating immune cells, as well as immune checkpoints across multiple datasets. Besides, within a GC immunotherapy cohort, individuals displaying high TAP1 expression demonstrated an increased likelihood of achieving complete remission (CR) post-treatment, suggesting heightened sensitivity to immunotherapy. In the clinical cohort, TAP1 overexpression in GC patients was positively correlated with CD8. TAP1 appears linked to an inflamed TME and serves as a prospective biomarker for discerning immunological attributes and gauging immunotherapeutic responses in GC, particularly in identifying immune-reactive tumors.

Ras-dependent activation of BMAL2 regulates hypoxic metabolism in pancreatic cancer.

To identify drivers of malignancy in human pancreatic ductal adenocarcinoma (PDAC), we performed regulatory network analysis on a large collection of expression profiles from laser capture microdissected samples of PDAC and benign precursors. We discovered that BMAL2 plays a role in the initiation, progression, post resection survival, and KRAS activity in PDAC. Functional analysis of BMAL2 target genes led us to hypothesize that it plays a role in regulating the response to hypoxia, a critical but poorly understood feature of PDAC physiology. Knockout of BMAL2 in multiple human PDAC cell lines revealed effects on viability and invasion, particularly under hypoxic conditions. Loss of BMAL2 also affected glycolysis and other metabolic processes. We found that BMAL2 directly regulates hypoxia-responsive target genes. We also found that BMAL2 is necessary for the stabilization of HIF1A upon exposure to hypoxia, but destabilizes HIF2A under hypoxia. These data demonstrate that BMAL2 is a master transcriptional regulator of hypoxia responses in PDAC and may serve as a long-sought molecular switch that distinguishes HIF1A- and HIF2A-dependent modes of hypoxic metabolism.

LGG-12. INTEGRATED CLINICAL AND MOLECULAR CHARACTERIZATION OF 200 DISSEMINATED PEDIATRIC LOW-GRADE GLIOMAS

Abstract Pediatric low-grade gliomas (PLGG) have excellent outcomes overall but are a major clinical challenge when disseminated. Diffuse leptomeningeal glioneuronal tumor (DLGNT) is a recognised entity both clinically and pathologically, however many disseminated LGG (DLGG) fall outside this diagnosis. To better understand the clinical and molecular features of DLGG as well as risk factors for dissemination, we assembled an international consortium of 30 sites, contributing over 200 patients with clinical annotation, along with genomic and methylation profiling. DLGG have worse progression-free (PFS) and overall survival (OS) than PLGG overall (p&amp;lt;0.0001). Seventy (35%) presented with a localized mass and secondary dissemination, including some with initial gross-total resection. The most common growth pattern observed (n=77, 40%) is a dominant suprasellar/optic pathway tumor with leptomeningeal drop-metastases involving the brainstem and spinal cord. Only 27 (14%) patients had diffuse tumors (without an identifiable dominant mass), and these had significantly worse OS (p=0.03). The most common pathologic diagnosis was pilocytic/pilomyxoid astrocytoma (n=92; 53%). DLGNT comprised a minority of cases (n=23, 14%), with a trend (p=0.056) towards worse survival compared to other diagnoses. The most frequent molecular alteration was BRAF fusion (78/151 with molecular testing; 52%), followed by FGFR mutations or fusions (18/151; 12%). BRAF V600E was underrepresented (14/151; 9%). 1p deletion was rare (17 patients) but highly associated with DLGNT. Methylation classification (n=70) was highly concordant with histologic diagnoses rather than clinical behavior. Importantly, patients who received upfront targeted therapies (TT; BRAF and/or MEK inhibition, n=9) had better PFS compared to those receiving chemotherapy (n=146, p=0.05). Furthermore, patients who were treated sequentially with chemo then TT had longer PFS on TT (p=0.011). This study presents the largest cohort of DLGG to date, expanding our understanding of the clinical, pathologic, and molecular features of this disease and supports the use of upfront targeted therapy.

BIOM-03. IMMUNO-ONCOLOGIC PROFILING OF PEDIATRIC BRAIN TUMORS REVEALS MAJOR CLINICAL SIGNIFICANCE OF THE TUMOR IMMUNE MICROENVIRONMENT

Abstract With the success of immunotherapy in cancer, understanding the tumor immune microenvironment (TIME) has become increasingly important; however in pediatric brain tumors this remains poorly characterized. Accordingly, we developed a clinical immune-oncology gene expression assay, including the 18-gene tumor inflammation signature (TIS) as a marker of overall immune activation, and validated this with immunohistochemistry for key cell-type markers. We used this assay, in conjunction with public data from the Pediatric Brain Tumor Atlas (PBTA), to profile a diverse range of 1382 samples in total with detailed clinical and molecular annotation. Overall among the main types of pediatric brain tumors, low-grade gliomas (LGG) had the highest inflammation levels (regardless of genetic driver alteration) and medulloblastomas the lowest. In LGG we identified three distinct patterns of immune activation with prognostic significance in BRAF V600E-mutant tumors, specifically that higher inflammation predicted worse outcomes. In high-grade gliomas (HGG), we observed immune activation and T-cell infiltrates in tumors that have historically been considered exclusively immune cold. There were genomic correlates of inflammation levels, with BRAF-V600E-mutant HGG having the highest inflammation levels and suggesting these may be candidates for combination MEK inhibition and immunotherapy. In mismatch repair deficient HGG, we found that high TIS was a significant predictor of response to immune checkpoint inhibition, and demonstrated the potential for multimodal biomarkers (TIS plus tumor mutation burden) to improve treatment stratification. Importantly, while overall patterns of immune activation were observed for histologically and genetically defined tumor types, there was significant variability within each entity, indicating that the TIME must be evaluated as an independent feature from diagnosis. In sum, in addition to the histology and molecular profile, this work underscores the importance of reporting on the TIME as an essential axis of cancer diagnosis in the era of personalized medicine.

A diagnostic challenge of KIT p.V559D and BRAF p.G469A mutations in a paragastric mass.

A patient with gastrointestinal stroma tumor (GIST) and KIT p.V559D and BRAF p.G469A alterations was referred to our institutional molecular tumor board (MTB) to discuss therapeutic implications. The patient had been diagnosed with B-cell chronic lymphocytic leukemia (CLL) years prior to the MTB presentation. GIST had been diagnosed 1 month earlier. After structured clinical annotation of the molecular alterations and interdisciplinary discussion, we considered BRAF/KIT co-mutation unlikely in a treatment-naïve GIST. Discordant variant allele frequencies furthermore suggested a second malignancy. NGS of a CLL sample revealed the identical class 2 BRAF alteration, thus supporting admixture of CLL cells in the paragastric mass, leading to the detection of 2 alterations. Following the MTB recommendation, the patient received imatinib and had a radiographic response. Structured annotation and interdisciplinary discussion in specialized tumor boards facilitate the clinical management of complex molecular findings. Coexisting malignancies and clonal hematopoiesis warrant consideration in case of complex and uncommon molecular findings.

Deep Learning–Based Clustering of OCT Images for Biomarker Discovery in Age-Related Macular Degeneration (PINNACLE Study Report 4)

PurposeWe introduce a deep-learning-based biomarker proposal system for the purpose of accelerating biomarker discovery in age-related macular degeneration (AMD). DesignRetrospective analysis of a large dataset of retinal optical coherence tomography (OCT) images. ParticipantsA total of 3,456 adults aged between 51 and 102 years of whom OCT images were collected under the PINNACLE project. MethodsOur system proposes candidates for novel AMD imaging biomarkers in OCT. It works by first training a neural network using self-supervised contrastive learning to discover, without any clinical annotations, features relating to both known and unknown AMD biomarkers present in 46,496 retinal OCT images. To interpret the learned biomarkers, we partition the images into 30 subsets, termed clusters, that contain similar features. We conduct two parallel 1.5-hour semi-structured interviews with two independent teams of retinal specialists to assign descriptions in clinical language to each cluster. Descriptions of clusters achieving consensus can potentially inform new biomarker candidates. Main Outcome MeasuresWe checked if each cluster showed clear features comprehensible to retinal specialists, if they related to AMD and how many described established biomarkers used in grading systems as opposed to recently proposed or potentially new biomarkers. We also compared their prognostic value for late stage wet and dry AMD against an established clinical grading system and a demographic baseline model. ResultsOverall, both teams independently identified clearly distinct characteristics in 27 of 30 clusters, of which 23 were related to AMD. Seven were recognised as known biomarkers used in established grading systems and 16 depicted biomarker combinations or subtypes that are either not yet used in grading systems, were only recently proposed, or were unknown. Clusters separated incomplete from complete retinal atrophy, intraretinal from subretinal fluid and thick from thin choroids, and in simulation outperformed clinically-used grading systems in prognostic value. ConclusionsUsing self-supervised deep learning we were able to automatically propose AMD biomarkers going beyond the set used in clinically established grading systems. Without any clinical annotations, contrastive learning discovered subtle differences between fine-grained biomarkers. Ultimately, we envision that equipping clinicians with discovery-oriented deep-learning tools can accelerate discovery of novel prognostic biomarkers.

Time-of-Flight MRA of Intracranial Aneurysms with Interval Surveillance, Clinical Segmentation and Annotations

Intracranial aneurysms (IAs) are present in 2–6% of the global population and can be catastrophic upon rupture with a mortality rate of 30–50%. IAs are commonly detected through time-of-flight magnetic resonance angiography (TOF-MRA), however, this data is rarely available for research and training purposes. The provision of imaging resources such as TOF-MRA images is imperative to develop new strategies for IA detection, rupture prediction, and surgical training. To support efforts in addressing data availability bottlenecks, we provide an open-access TOF-MRA dataset comprising 63 patients, of which 24 underwent interval surveillance imaging by TOF-MRA. Patient scans were evaluated by a neuroradiologist, providing aneurysm and vessel segmentations, clinical annotations, 3D models, in addition to 3D Slicer software environments containing all this data for each patient. This dataset is the first to provide interval surveillance imaging for supporting the understanding of IA growth and stability. This dataset will support computational and experimental research into IA dynamics and assist surgical and radiology training in IA treatment.

MosaicMPI: a framework for modular data integration across cohorts and -omics modalities.

Advances in molecular profiling have facilitated generation of large multi-modal datasets that can potentially reveal critical axes of biological variation underlying complex diseases. Distilling biological meaning, however, requires computational strategies that can perform mosaic integration across diverse cohorts and datatypes. Here, we present mosaicMPI, a framework for discovery of low to high-resolution molecular programs representing both cell types and states, and integration within and across datasets into a network representing biological themes. Using existing datasets in glioblastoma, we demonstrate that this approach robustly integrates single cell and bulk programs across multiple platforms. Clinical and molecular annotations from cohorts are statistically propagated onto this network of programs, yielding a richly characterized landscape of biological themes. This enables deep understanding of individual tumor samples, systematic exploration of relationships between modalities, and generation of a reference map onto which new datasets can rapidly be mapped. mosaicMPI is available at https://github.com/MorrissyLab/mosaicMPI.

The N6-methyladenosine methylation landscape stratifies breast cancer into two subtypes with distinct immunological characteristics.

N6-methyladenosine (m6A) methylation modification affects the tumorigenesis and metastasis of breast cancer (BC). This study investigated the association between m6A regulator-mediated methylation modification patterns and characterization of the tumour microenvironment in BC, as well as their prognostic importance. Public gene expression data and clinical annotations were collected from The Cancer Genome Atlas (TCGA) database, the Gene Expression Omnibus website and the METABRIC program. We analysed the genetic expression, gene-gene interactions, gene mutations and copy number variations using R software. The data were screened for risk genes using the Cox risk regression model, and we developed an algorithm for risk score and its predictive value. Compared to adjacent normal tissue, we identified 16 differentially expressed m6A regulators in BC, including six writers and 10 readers. Under unsupervised clustering, two distinguished modification patterns were identified, cluster C1 and C2. Compared to m6A cluster C2, cluster C1 was found to be more involved in immune-related pathways, with a relatively higher immune score and stromal score (P < 0.05). Patients were divided into two groups based on their risk scores for survival analysis. The patients in the high-risk score group had significantly worse overall survival than patients in the low-risk score group, (P < 0.0001). The TCGA database validation revealed the same prognostic tendency. In summary, our study showed distinct m6A regulator modification patterns contribute to the immunological heterogeneity and diversity of BC. The development of m6A gene signatures and the m6A score aid in the prognostic prediction of patients with BC.

Frequency of pharmacogenomic variants affecting efficacy and safety of anti-cancer drugs in a south Asian population from Sri Lanka

BackgroundTherapy with anti-cancer drugs remain the cornerstone of treating cancer. The effectiveness and safety of anti-cancer drugs vary significantly among individuals due to genetic factors influencing the drug response and metabolism. Data on the pharmacogenomic variations in Sri Lankans related to anti-cancer therapy is sparse. As current treatment guidelines in Sri Lanka often do not consider local pharmacogenomic variants, this study aimed to explore the diversity of pharmacogenomic variants in the Sri Lankan population to pave the way for personalized treatment approaches and improve patient outcomes.MethodsPharmacogenomic data regarding variant-drug pairs of genes CYP2D6, DPYD, NUDT15, EPAS1, and XRCC1 with clinical annotations labelled as evidence levels 1A-2B were obtained from the Pharmacogenomics Knowledgebase database. Their frequencies in Sri Lankans were obtained from an anonymized database that was derived from 541 Sri Lankans who underwent exome sequencing at the Human Genetics Unit, Faculty of Medicine, University of Colombo. Variations in DPYD, NUDT15, and EPAS1 genes are related to increased toxicity to fluoropyrimidines, mercaptopurines, and sorafenib respectively. Variations in CYP2D6 and XRCC1 genes are related to changes in efficacy of tamoxifen and platinum compounds, respectively. Minor allele frequencies of these variants were calculated and compared with other populations.ResultsMAFs of rs1065852 c.100 C > T (CYP2D6), rs3918290 c.1905 + 1G > A (DPYD), rs56038477 c.1236G > A (DPYD), rs7557402 c.1035–7 C > G (EPAS1), rs116855232 c.415 C > T (NUDT15*3), and rs25487 c.1196 A > G (XRCC1) were: 12.9% [95%CI:10.9–14.9], 1.5% [95%CI:0.8–2.2], 1.2% [95%CI:0.5–1.8], 37.7% [95%CI:34.8–40.6], 8.3% [95%CI:6.7–10.0], and 64.0% [95%CI:61.1–66.8], respectively. Frequencies of rs1065852 c.100 C > T (CYP2D6), rs7557402 c.1035–7 C > G (EPAS1), and rs25487 (XRCC1) were significantly lower in Sri Lankans, while frequencies of rs116855232 c.415 C > T (NUDT15*3) and rs56038477 c.1236G > A (DPYD) were significantly higher in Sri Lankans when compared to some Western and Asian populations.ConclusionSri Lankans are likely to show lower toxicity risk with sorafenib (rs7557402 c.84,131 C > G) and, higher toxicity risk with fluoropyrimidines (rs56038477 c.1236G > A) and mercaptopurine (rs116855232 c.415 C > T), and reduced effectiveness with tamoxifen (rs1065852 c.100 C > T) and platinum compounds (rs25487). These findings highlight the potential contribution of these genetic variations to the individual variability in anti-cancer dosage requirements among Sri Lankans.

Automated Classification of Physiologic, Glaucomatous, and Glaucoma-Suspected Optic Discs Using Machine Learning.

In order to generate a machine learning algorithm (MLA) that can support ophthalmologists with the diagnosis of glaucoma, a carefully selected dataset that is based on clinically confirmed glaucoma patients as well as borderline cases (e.g., patients with suspected glaucoma) is required. The clinical annotation of datasets is usually performed at the expense of the data volume, which results in poorer algorithm performance. This study aimed to evaluate the application of an MLA for the automated classification of physiological optic discs (PODs), glaucomatous optic discs (GODs), and glaucoma-suspected optic discs (GSODs). Annotation of the data to the three groups was based on the diagnosis made in clinical practice by a glaucoma specialist. Color fundus photographs and 14 types of metadata (including visual field testing, retinal nerve fiber layer thickness, and cup-disc ratio) of 1168 eyes from 584 patients (POD = 321, GOD = 336, GSOD = 310) were used for the study. Machine learning (ML) was performed in the first step with the color fundus photographs only and in the second step with the images and metadata. Sensitivity, specificity, and accuracy of the classification of GSOD vs. GOD and POD vs. GOD were evaluated. Classification of GOD vs. GSOD and GOD vs. POD performed in the first step had AUCs of 0.84 and 0.88, respectively. By combining the images and metadata, the AUCs increased to 0.92 and 0.99, respectively. By combining images and metadata, excellent performance of the MLA can be achieved despite having only a small amount of data, thus supporting ophthalmologists with glaucoma diagnosis.

Abstract PO4-03-01: Enhancing Breast Cancer Subtyping: APIS Breast Cancer Subtyping Kit Proliferative Signature's Potential for Improved Proliferation Assessment and Recurrence Prediction

Abstract Background: Ki67, a widely recognized biomarker for measuring tumor cell proliferation in breast cancer (BC), plays a crucial role in classifying BC into luminal A and luminal B subtypes, subsequently aiding treatment decisions. However, the inconsistency in Ki67 assessment and the lack of standardization in methods and interpretation of immunohistochemistry (IHC) staining has hindered its recommendation for routine diagnostic use. The standardization of measurement techniques, especially for Ki67, holds immense potential for improving patient treatment outcomes. The APIS Breast Cancer Subtyping Kit (APIS BCS Kit) addresses these challenges by utilizing RT-qPCR to accurately determine the mRNA expression of BC markers, including Ki67, along with a novel four-gene proliferative signature, comprising four markers (MKI67, CCNA2, KIF23, and PCNA) associated with proliferation and expressed throughout all cell cycle stages, thereby providing a more accurate measure of tumor proliferation. In this study, we present the preliminary findings of our ongoing evaluation of the APIS BCS Kit proliferative signature's ability to distinguish between Luminal A and Luminal B subtypes and to determine the risk of recurrence in hormone receptor-positive (HR+) patients. Method: A total of N=153 retrospectively collected HR+/HER2 negative samples were provided by University of Basel. For each sample the following clinical annotations were included: relapse free survival, treatment regimen (Chemotherapy (ChT) or Hormone Therapy (HT)) as well as OncotypeDx (ODx) risk score (RS). N=40 samples underwent additional testing using PAM50 (Prosigna®) test. Only samples collected since 2018 were included, resulting in 10 observed relapse events (N=7 in HT cohort and N=3 in ChT cohort). RNA isolation from tissue sections and gene expression analysis were performed as per APIS BCS Kit instructions for use. The agreement of subtype calls between IHC, APIS BCS Kit, and PAM50 was assessed. The correlation between RS obtained from ODx, PAM50 and the APIS BCS Kit proliferation signature was also explored. Results: The agreement between IHC and the APIS BCS Kit subtype call, utilizing Ki67 alone as a measure of proliferation, was 67.3% (103/153). Replacing Ki67 with the APIS BCS Kit proliferation signature increased the concordance to 69.9% (107/153). The concordance between IHC and PAM50 subtype calls was comparable at 62.5% (25/40). The agreement between the APIS BCS Kit and PAM50 subtype call was notably higher (80%; 32/40). Significant association was observed between the APIS BCS Kit proliferation signature and ODx RS (P=.0001), and PAM50 RS (P=.002). In the HT cohort, both ODx and the APIS BCS Kit proliferation signature accurately predicted recurrence (by assigning high RS) in n=2 cases. In some cases both the APIS BCS Kit and ODx erroneously assigned low risk and subsequently missed n=4 recurrence events. Notably, APIS BCS Kit proliferation signature correctly identified one recurrence event (high RS), which was missed by ODx (low RS). Conclusion: The APIS Breast Cancer Subtyping Kit exhibits improved agreement with the PAM50 molecular subtype in contrast to the IHC-derived subtype, indicating that molecular testing may be a more suitable method for determining proliferation and distinguishing between luminal A and B subtypes. Furthermore, there is a significant correlation between the RS generated by APIS Breast Cancer Subtyping Kit proliferation signature and both PAM50 and Oncotype RS. These initial findings indicate that the APIS Breast Cancer Subtyping Kit proliferation signature has the potential to accurately identify patients at risk of recurrence, although additional studies including a larger proportion of events are needed to validate these results. Citation Format: Marcus Vetter, Anna Gasior, Elena-Diana Chiru, Joanna Gorniak, Leanne Gough, Mathew Harrison, Sara Rollinson, Simone Muenst, Christian Kurzeder. Enhancing Breast Cancer Subtyping: APIS Breast Cancer Subtyping Kit Proliferative Signature's Potential for Improved Proliferation Assessment and Recurrence Prediction [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO4-03-01.

Abstract PO5-24-11: An integrated approach for comprehensive molecular and tumor microenvironment characterization of invasive lobular carcinoma

Abstract Introduction: Although recent studies have shown that invasive lobular carcinoma (ILC) is associated with worse long-term outcomes compared to invasive breast cancer of no special type (NST), they are treated in a similar manner. Therefore, classification approaches to help understand the molecular characteristics associated with the different BC subtypes are needed to personalize therapy. Here, we describe an integrated analytical approach to comprehensively characterize ILC samples. Methods: We collected genomic and transcriptomic data for 16,087 ILC and NST samples, with clinical and pathological annotation, from 59 datasets and used an internally developed algorithm based on clustering performed after gene expression analysis to classify them into 5 intrinsic subtypes: Basal, Luminal A (LumA), Luminal B (LumB), HER2 high, and HER2 low. We performed gene set enrichment analyses and utilized the DESeq2 algorithm to perform differential expression analysis of RNA-seq data. We performed molecular grade subtyping using methods described by Antysheva et al. (Cancer Res., 2022). Samples that were listed as ILC or had an inactivated CDH1 gene by mutation, deletion or low expression (&amp;lt; −2.5 MAD after median scaling) were defined as having an ILC histomolecular subtype. Using methods described by Bagaev et. al. (Cancer Cell, 2021), 29 functional gene expression signatures were selected, and unsupervised dense Louvain clustering was performed to identify TME subtypes. The Mann-Whitney U test was used for statistical analysis. The tertiary lymphoid structures (TLS) gene signature was evaluated using methods described by Sautès-Fridman et al. (Nat. Rev. Cancer, 2019). Results: In the ILC samples (n=1,887), we identified LumA (46%), LumB (40%), Basal (8%,), HER2 high (3%), and HER2 low (3%) subtypes. In contrast, the NST samples were classified as LumB (41%), LumA (25%), Basal (19%), HER2 high (9%), and HER2 low (5%). Molecular grade subtyping revealed that 67% of ILC samples were low grade and 33% were high grade, compared to NST where 55% were high grade and 45% were low grade. Notably, among high-grade ILC, LumB (64%) and Basal (23%) subtypes were predominant. Key differentially expressed genes included downregulated CDH1 and upregulated MMP23B, IGF1, and TFAP2B across LumA and LumB ILC samples. Further analysis revealed downregulation of TACSTD2 and CDH3 in the Basal subtype for ILC samples compared to NST samples. Transcriptomic-based TME subtyping revealed distinct TME distributions across the BC subtypes and molecular grades for ILC samples (Table 1). In high-grade ILC LumB samples, 38% had a Desert (D; predictive of poor immunotherapy response) and 39% had an Immune-Enriched (IE; predictive of favorable immunotherapy response) TME. In low-grade ILC LumB samples we found that while 51% had a D TME, only 14% had an IE TME. We found an overrepresentation of the TLS signature [Solid TLS] in ILC compared to NST samples, and higher matrix remodeling and CAF signatures in LumA compared to the LumB subtype among low-grade ILC samples. Samples with high CD274 expression levels had higher percentages of T cells and M1 macrophages, and LumA samples had significantly higher (p &amp;lt; 0.001) CD274 expression compared to the other BC subtypes. Conclusion: These results provide in-depth molecular and tumor microenvironment characterization of ILC. Further optimization of this analytical approach could lead to the development of more effective therapeutic strategies. Table 1. ILC subtypes classification Citation Format: Jason Mouabbi, Konstantin Chernyshov, Oleg Baranov, Vladimir Kushnarev, Polina Turova, Anna Butusova, Sofia Menshikova, Jessica Brown, Nikita Kotlov, Patrick Clayton, Krystle Nomie, Nathan Fowler, Debu Tripathy. An integrated approach for comprehensive molecular and tumor microenvironment characterization of invasive lobular carcinoma [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-24-11.