Cancer gene expression profiles associated with clinical outcomes to chemotherapy treatments

Uniformly shaped harmonization combines human transcriptomic data from different platforms while retaining their biological properties and differential gene expression patterns

Introduction: Co-normalization of RNA profiles obtained using different experimental platforms and protocols opens avenue for comprehensive comparison of relevant features like differentially expressed genes associated with disease. Currently, most of bioinformatic tools enable normalization in a flexible format that depends on the individual datasets under analysis. Thus, the output data of such normalizations will be poorly compatible with each other. Recently we proposed a new approach to gene expression data normalization termed Shambhala which returns harmonized data in a uniform shape, where every expression profile is transformed into a pre-defined universal format. We previously showed that following shambhalization of human RNA profiles, overall tissue-specific clustering features are strongly retained while platform-specific clustering is dramatically reduced. Methods: Here, we tested Shambhala performance in retention of fold-change gene expression features and other functional characteristics of gene clusters such as pathway activation levels and predicted cancer drug activity scores. Results: Using 6,793 cancer and 11,135 normal tissue gene expression profiles from the literature and experimental datasets, we applied twelve performance criteria for different versions of Shambhala and other methods of transcriptomic harmonization with flexible output data format. Such criteria dealt with the biological type classifiers, hierarchical clustering, correlation/regression properties, stability of drug efficiency scores, and data quality for using machine learning classifiers. Discussion: Shambhala-2 harmonizer demonstrated the best results with the close to 1 correlation and linear regression coefficients for the comparison of training vs validation datasets and more than two times lesser instability for calculation of drug efficiency scores compared to other methods.

Reclassification of TCGA Diffuse Glioma Profiles Linked to Transcriptomic, Epigenetic, Genomic and Clinical Data, According to the 2021 WHO CNS Tumor Classification

In 2021, the fifth edition of the WHO classification of tumors of the central nervous system (WHO CNS5) was published. Molecular features of tumors were directly incorporated into the diagnostic decision tree, thus affecting both the typing and staging of the tumor. It has changed the traditional approach, based solely on histopathological classification. The Cancer Genome Atlas project (TCGA) is one of the main sources of molecular information about gliomas, including clinically annotated transcriptomic and genomic profiles. Although TCGA itself has played a pivotal role in developing the WHO CNS5 classification, its proprietary databases still retain outdated diagnoses which frequently appear incorrect and misleading according to the WHO CNS5 standards. We aimed to define the up-to-date annotations for gliomas from TCGA's database that other scientists can use in their research. Based on WHO CNS5 guidelines, we developed an algorithm for the reclassification of TCGA glioma samples by molecular features. We updated tumor type and diagnosis for 828 out of a total of 1122 TCGA glioma cases, after which available transcriptomic and methylation data showed clustering features more consistent with the updated grouping. We also observed better stratification by overall survival for the updated diagnoses, yet WHO grade 3 IDH-mutant oligodendrogliomas and astrocytomas are still indistinguishable. We also detected altered performance in the previous diagnostic transcriptomic molecular biomarkers (expression of SPRY1, CRNDE and FREM2 genes and FREM2 molecular pathway) and prognostic gene signature (FN1, ITGA5, OSMR, and NGFR) after reclassification. Thus, we conclude that further efforts are needed to reconsider glioma molecular biomarkers.

Transcriptomic Harmonization as the Way for Suppressing Cross-Platform Bias and Batch Effect

(1) Background: Emergence of methods interrogating gene expression at high throughput gave birth to quantitative transcriptomics, but also posed a question of inter-comparison of expression profiles obtained using different equipment and protocols and/or in different series of experiments. Addressing this issue is challenging, because all of the above variables can dramatically influence gene expression signals and, therefore, cause a plethora of peculiar features in the transcriptomic profiles. Millions of transcriptomic profiles were obtained and deposited in public databases of which the usefulness is however strongly limited due to the inter-comparison issues; (2) Methods: Dozens of methods and software packages that can be generally classified as either flexible or predefined format harmonizers have been proposed, but none has become to the date the gold standard for unification of this type of Big Data; (3) Results: However, recent developments evidence that platform/protocol/batch bias can be efficiently reduced not only for the comparisons of limited transcriptomic datasets. Instead, instruments were proposed for transforming gene expression profiles into the universal, uniformly shaped format that can support multiple inter-comparisons for reasonable calculation costs. This forms a basement for universal indexing of all or most of all types of RNA sequencing and microarray hybridization profiles; (4) Conclusions: In this paper, we attempted to overview the landscape of modern approaches and methods in transcriptomic harmonization and focused on the practical aspects of their application.

Transcriptomic Portraits and Molecular Pathway Activation Features of Adult Spinal Intramedullary Astrocytomas

In this study, we report 31 spinal intramedullary astrocytoma (SIA) RNA sequencing (RNA-seq) profiles for 25 adult patients with documented clinical annotations. To our knowledge, this is the first clinically annotated RNA-seq dataset of spinal astrocytomas derived from the intradural intramedullary compartment. We compared these tumor profiles with the previous healthy central nervous system (CNS) RNA-seq data for spinal cord and brain and identified SIA-specific gene sets and molecular pathways. Our findings suggest a trend for SIA-upregulated pathways governing interactions with the immune cells and downregulated pathways for the neuronal functioning in the context of normal CNS activity. In two patient tumor biosamples, we identified diagnostic KIAA1549-BRAF fusion oncogenes, and we also found 16 new SIA-associated fusion transcripts. In addition, we bioinformatically simulated activities of targeted cancer drugs in SIA samples and predicted that several tyrosine kinase inhibitory drugs and thalidomide analogs could be potentially effective as second-line treatment agents to aid in the prevention of SIA recurrence and progression.

Gene Expression-Based Signature Can Predict Sorafenib Response in Kidney Cancer

Sorafenib is a tyrosine kinase inhibitory drug with multiple molecular specificities that is approved for clinical use in second-line treatments of metastatic and advanced renal cell carcinomas (RCCs). However, only 10–40% of RCC patients respond on sorafenib-containing therapies, and personalization of its prescription may help in finding an adequate balance of clinical efficiency, cost-effectiveness, and side effects. We investigated whether expression levels of known molecular targets of sorafenib in RCC can serve as prognostic biomarker of treatment response. We used Illumina microarrays to profile RNA expression in pre-treatment formalin-fixed paraffin-embedded (FFPE) samples of 22 metastatic or advanced RCC cases with known responses on next-line sorafenib monotherapy. Among them, nine patients showed partial response (PR), three patients—stable disease (SD), and 10 patients—progressive disease (PD) according to Response Evaluation Criteria In Solid Tumors (RECIST) criteria. We then classified PR + SD patients as “responders” and PD patients as “poor responders”. We found that gene signature including eight sorafenib target genes was congruent with the drug response characteristics and enabled high-quality separation of the responders and poor responders [area under a receiver operating characteristic curve (AUC) 0.89]. We validated these findings on another set of 13 experimental annotated FFPE RCC samples (for 2 PR, 1 SD, and 10 PD patients) that were profiled by RNA sequencing and observed AUC 0.97 for 8-gene signature as the response classifier. We further validated these results in a series of qRT-PCR experiments on the third experimental set of 12 annotated RCC biosamples (for 4 PR, 3 SD, and 5 PD patients), where 8-gene signature showed AUC 0.83.

Integration of DNA Microarray with Clinical and Genomic Data

DNA microarrays have been widely employed to understand cancer development. This technology is able to measure expression levels of a large numbers of genes or to genotype multiple regions of a genome in a massively parallel experiment. In addition, the detection of methylation patterns and gene copy number variations are also performed. Clinicians began to apply these findings in personalized medicine for the selection of cancer therapy according to the individual's cancer genomic profile. Because cancer is a complex disease it is of great value to integrate microarray data with genomic and clinical data. Here, we presented an overview of DNA microarray technology and discuss about benefits and challenging of microarray data integration.

LRRC4 mediates the formation of circular RNA CD44 to inhibitGBM cell proliferation

Mounting evidence reveals that dysregulation of circular RNAs (circRNAs) is involved in the development of glioblastoma. Leucine-rich repeat-containing 4 (LRRC4) has been shown to suppress tumors in glioblastoma. However, whether LRRC4 can regulate the formation of circRNA is not yet understood. In this study, LRRC4 was found to interact with SAM68. LRRC4 promoted the generation of circCD44 by inhibiting the binding between SAM68 and CD44 pre-mRNA. Moreover, downregulated expression of circCD44 was found in glioblastoma multiforme (GBM) tissues and GBM primary cells. Re-expression of circCD44 significantly suppressed the proliferation, colony formation, and invasion of GBM cells and inhibited tumor growth in vivo. Mechanistically, circCD44 could regulate the expression of SMAD6 via sponging miR-326 and miR-330-5p involved in the progression of GBM. Thus, the LRRC4/SAM68/circCD44/miR-326/miR-330-5p/SMAD6 signaling axis could be a potential target for GBM treatment.

Experimental and Meta-Analytic Validation of RNA Sequencing Signatures for Predicting Status of Microsatellite Instability

Microsatellite instability (MSI) is an important diagnostic and prognostic cancer biomarker. In colorectal, cervical, ovarian, and gastric cancers, it can guide the prescription of chemotherapy and immunotherapy. In laboratory diagnostics of susceptible tumors, MSI is routinely detected by the size of marker polymerase chain reaction products encompassing frequent microsatellite expansion regions. Alternatively, MSI status is screened indirectly by immunohistochemical interrogation of microsatellite binding proteins. RNA sequencing (RNAseq) profiling is an emerging source of data for a wide spectrum of cancer biomarkers. Recently, three RNAseq-based gene signatures were deduced for establishing MSI status in tumor samples. They had 25, 15, and 14 gene products with only one common gene. However, they were developed and tested on the incomplete literature of The Cancer Genome Atlas (TCGA) sampling and never validated experimentally on independent RNAseq samples. In this study, we, for the first time, systematically validated these three RNAseq MSI signatures on the literature colorectal cancer (CRC) (n = 619), endometrial carcinoma (n = 533), gastric cancer (n = 380), uterine carcinosarcoma (n = 55), and esophageal cancer (n = 83) samples and on the set of experimental CRC RNAseq samples (n = 23) for tumors with known MSI status. We found that all three signatures performed well with area under the curve (AUC) ranges of 0.94-1 for the experimental CRCs and 0.94-1 for the TCGA CRC, esophageal cancer, and uterine carcinosarcoma samples. However, for the TCGA endometrial carcinoma and gastric cancer samples, only two signatures were effective with AUC 0.91-0.97, whereas the third signature showed a significantly lower AUC of 0.69-0.88. Software for calculating these MSI signatures using RNAseq data is included.

EGFR Pathway-Based Gene Signatures of Druggable Gene Mutations in Melanoma, Breast, Lung, and Thyroid Cancers

EGFR, BRAF, PIK3CA, and KRAS genes play major roles in EGFR pathway, and accommodate activating mutations that predict response to many targeted therapeutics. However, connections between these mutations and EGFR pathway expression patterns remain unexplored. Here, we investigated transcriptomic associations with these activating mutations in three ways. First, we compared expressions of these genes in the mutant and wild type tumors, respectively, using RNA sequencing profiles from The Cancer Genome Atlas project database (n = 3660). Second, mutations were associated with the activation level of EGFR pathway. Third, they were associated with the gene signatures of differentially expressed genes from these pathways between the mutant and wild type tumors. We found that the upregulated EGFR pathway was linked with mutations in the BRAF (thyroid cancer, melanoma) and PIK3CA (breast cancer) genes. Gene signatures were associated with BRAF (thyroid cancer, melanoma), EGFR (squamous cell lung cancer), KRAS (colorectal cancer), and PIK3CA (breast cancer) mutations. However, only for the BRAF gene signature in the thyroid cancer we observed strong biomarker diagnostic capacity with AUC > 0.7 (0.809). Next, we validated this signature on the independent literature-based dataset (n = 127, fresh-frozen tissue samples, AUC 0.912), and on the experimental dataset (n = 42, formalin fixed, paraffin embedded tissue samples, AUC 0.822). Our results suggest that the RNA sequencing profiles can be used for robust identification of the replacement of Valine at position 600 with Glutamic acid in the BRAF gene in the papillary subtype of thyroid cancer, and evidence that the specific gene expression levels could provide information about the driver carcinogenic mutations.

RNA Sequencing Data for FFPE Tumor Blocks Can Be Used for Robust Estimation of Tumor Mutation Burden in Individual Biosamples

Tumor mutation burden (TMB) is a well-known efficacy predictor for checkpoint inhibitor immunotherapies. Currently, TMB assessment relies on DNA sequencing data. Gene expression profiling by RNA sequencing (RNAseq) is another type of analysis that can inform clinical decision-making and including TMB estimation may strongly benefit this approach, especially for the formalin-fixed, paraffin-embedded (FFPE) tissue samples. Here, we for the first time compared TMB levels deduced from whole exome sequencing (WES) and RNAseq profiles of the same FFPE biosamples in single-sample mode. We took TCGA project data with mean sequencing depth 23 million gene-mapped reads (MGMRs) and found 0.46 (Pearson)–0.59 (Spearman) correlation with standard mutation calling pipelines. This was converted into low (<10) and high (>10) TMB per megabase classifier with area under the curve (AUC) 0.757, and application of machine learning increased AUC till 0.854. We then compared 73 experimental pairs of WES and RNAseq profiles with lower (mean 11 MGMRs) and higher (mean 68 MGMRs) RNA sequencing depths. For higher depth, we observed ~1 AUC for the high/low TMB classifier and 0.85 (Pearson)–0.95 (Spearman) correlation with standard mutation calling pipelines. For the lower depth, the AUC was below the high-quality threshold of 0.7. Thus, we conclude that using RNA sequencing of tumor materials from FFPE blocks with enough coverage can afford for high-quality discrimination of tumors with high and low TMB levels in a single-sample mode.

CTR-DB, an omnibus for patient-derived gene expression signatures correlated with cancer drug response

To date, only some cancer patients can benefit from chemotherapy and targeted therapy. Drug resistance continues to be a major and challenging problem facing current cancer research. Rapidly accumulated patient-derived clinical transcriptomic data with cancer drug response bring opportunities for exploring molecular determinants of drug response, but meanwhile pose challenges for data management, integration, and reuse. Here we present the Cancer Treatment Response gene signature DataBase (CTR-DB, http://ctrdb.ncpsb.org.cn/), a unique database for basic and clinical researchers to access, integrate, and reuse clinical transcriptomes with cancer drug response. CTR-DB has collected and uniformly reprocessed 83 patient-derived pre-treatment transcriptomic source datasets with manually curated cancer drug response information, involving 28 histological cancer types, 123 drugs, and 5139 patient samples. These data are browsable, searchable, and downloadable. Moreover, CTR-DB supports single-dataset exploration (including differential gene expression, receiver operating characteristic curve, functional enrichment, sensitizing drug search, and tumor microenvironment analyses), and multiple-dataset combination and comparison, as well as biomarker validation function, which provide insights into the drug resistance mechanism, predictive biomarker discovery and validation, drug combination, and resistance mechanism heterogeneity.

Algorithmically Deduced FREM2 Molecular Pathway Is a Potent Grade and Survival Biomarker of Human Gliomas

Gliomas are the most common malignant brain tumors with high mortality rates. Recently we showed that the FREM2 gene has a role in glioblastoma progression. Here we reconstructed the FREM2 molecular pathway using the human interactome model. We assessed the biomarker capacity of FREM2 expression and its pathway as the overall survival (OS) and progression-free survival (PFS) biomarkers. To this end, we used three literature and one experimental RNA sequencing datasets collectively covering 566 glioblastomas (GBM) and 1097 low-grade gliomas (LGG). The activation level of deduced FREM2 pathway showed strong biomarker characteristics and significantly outperformed the FREM2 expression level itself. For all relevant datasets, it could robustly discriminate GBM and LGG (p < 1.63 × 10-13, AUC > 0.74). High FREM2 pathway activation level was associated with poor OS in LGG (p < 0.001), and low PFS in LGG (p < 0.001) and GBM (p < 0.05). FREM2 pathway activation level was poor prognosis biomarker for OS (p < 0.05) and PFS (p < 0.05) in LGG with IDH mutation, for PFS in LGG with wild type IDH (p < 0.001) and mutant IDH with 1p/19q codeletion(p < 0.05), in GBM with unmethylated MGMT (p < 0.05), and in GBM with wild type IDH (p < 0.05). Thus, we conclude that the activation level of the FREM2 pathway is a potent new-generation diagnostic and prognostic biomarker for multiple molecular subtypes of GBM and LGG.

Machine Learning Applicability for Classification of PAD/VCD Chemotherapy Response Using 53 Multiple Myeloma RNA Sequencing Profiles

Multiple myeloma (MM) affects ~500,000 people and results in ~100,000 deaths annually, being currently considered treatable but incurable. There are several MM chemotherapy treatment regimens, among which eleven include bortezomib, a proteasome-targeted drug. MM patients respond differently to bortezomib, and new prognostic biomarkers are needed to personalize treatments. However, there is a shortage of clinically annotated MM molecular data that could be used to establish novel molecular diagnostics. We report new RNA sequencing profiles for 53 MM patients annotated with responses on two similar chemotherapy regimens: bortezomib, doxorubicin, dexamethasone (PAD), and bortezomib, cyclophosphamide, dexamethasone (VCD), or with responses to their combinations. Fourteen patients received both PAD and VCD; six received only PAD, and 33 received only VCD. We compared profiles for the good and poor responders and found five genes commonly regulated here and in the previous datasets for other bortezomib regimens (all upregulated in the good responders): FGFR3, MAF, IGHA2, IGHV1-69, and GRB14. Four of these genes are linked with known immunoglobulin locus rearrangements. We then used five machine learning (ML) methods to build a classifier distinguishing good and poor responders for two cohorts: PAD + VCD (53 patients), and separately VCD (47 patients). We showed that the application of FloWPS dynamic data trimming was beneficial for all ML methods tested in both cohorts, and also in the previous MM bortezomib datasets. However, the ML models build for the different datasets did not allow cross-transferring, which can be due to different treatment regimens, experimental profiling methods, and MM heterogeneity.

Identification and Validation of an Immune-Associated RNA-Binding Proteins Signature to Predict Clinical Outcomes and Therapeutic Responses in Glioma Patients

The prognosis of patients with glioma is largely related to both the tumor-infiltrating immune cells and the expression of RNA-binding proteins (RBPs) that are able to regulate various pro-inflammatory and oncogenic mediators. However, immune-associated RBPs in glioma remain unexplored. In this study, we captured patient data from The Cancer Genome Atlas (TCGA) and divided them into two immune subtype groups according to the difference in infiltration of immune cells. After differential expression and co-expression analysis, we identified 216 RBPs defined as immune-associated RBPs. After narrowing down processes, eight RBPs were selected out to construct a risk signature that proven to be a novel and independent prognostic factor. The patients were divided into high- and low-risk groups on the basis of risk score. Higher risk scores meant worse overall survival and higher expression of human leukocyte antigen and immune checkpoints such as PD1 and CTLA4. In addition, analyses of pathway enrichment, somatic mutation, copy number variations and immuno-/chemotherapeutic response prediction were performed in high- and low-risk groups and compared with each other. For the first time, we demonstrated a novel signature composed of eight immune-associated RBPs that was valuable in predicting the survival of glioma patients and directing immunotherapy and chemotherapy.