Assessment of intratumoral heterogeneity with mutations and gene expression profiles

Single-cell and bulk transcriptome analysis reveals tumor cell heterogeneity and underlying molecular program in uveal melanoma

BACKGROUND

Uveal melanoma (UM) is a rare and deadly eye cancer with high metastatic potential. Despite the predominance of malignant cells within the tumor microenvironment, the heterogeneity and underlying molecular features remain to be fully characterized.

METHODS

We analyzed single-cell transcriptomic profiling of 37,660 malignant cells from 17 UM tumors. A consensus non-negative factorization algorithm was used to decipher transcriptional programs underlying tumor cell-intrinsic heterogeneity. Tumor-infiltrated immune cells were computationally estimated from bulk transcriptomes and bioinformatics methods. A gene signature was derived for subtyping and prognostic stratification and validated in multicenter cohorts.

RESULTS

ScRNA-seq analysis revealed the existence of diverse subpopulations and transcriptional variability among malignant cells within and between tumors. Furthermore, we observed that the heterogeneity of malignant cell states and compositions correlated with genomic, immunological, and clinical characteristics. By identifying gene expression programs associated with malignant cell heterogeneity at the single cell level, UM samples were classified into two distinct intra-tumoral subtypes (ITMHlo and ITMHhi) with different prognoses and immune microenvironments. Finally, a machine learning-derived 9-gene signature was developed to translate single-cell information into bulk tissue transcriptomes for patient stratification and was validated in multicenter cohorts.

CONCLUSIONS

Our study provides insight into understanding the intra-tumoral heterogeneity of UM and its potential impact on patient stratification.

Precision Medicine in Veterinary Science

Precision medicine focuses on the clinical management of the individual patient, not on population-based findings. Successes from human precision medicine inform veterinary oncology. Early evidence of success for canines shows how precision medicine can be integrated into practice. Decreasing genomic profiling costs will allow increased utilization and subsequent improvement of knowledge base from which to make better informed decisions. Utility of precision medicine in canine oncology will only increase for improved cancer characterization, enhanced therapy selection, and overall more successful management of canine cancer. As such, practitioners are called to interpret and leverage precision medicine reports for their patients.

Multi-omics and artificial intelligence predict clinical outcomes of immunotherapy in non-small cell lung cancer patients

In recent years, various types of immunotherapy, particularly the use of immune checkpoint inhibitors targeting programmed cell death 1 or programmed death ligand 1 (PD-L1), have revolutionized the management and prognosis of non-small cell lung cancer. PD-L1 is frequently used as a biomarker for predicting the likely benefit of immunotherapy for patients. However, some patients receiving immunotherapy have high response rates despite having low levels of PD-L1. Therefore, the identification of this group of patients is extremely important to improve prognosis. The tumor microenvironment contains tumor, stromal, and infiltrating immune cells with its composition differing significantly within tumors, between tumors, and between individuals. The omics approach aims to provide a comprehensive assessment of each patient through high-throughput extracted features, promising a more comprehensive characterization of this complex ecosystem. However, features identified by high-throughput methods are complex and present analytical challenges to clinicians and data scientists. It is thus feasible that artificial intelligence could assist in the identification of features that are beyond human discernment as well as in the performance of repetitive tasks. In this paper, we review the prediction of immunotherapy efficacy by different biomarkers (genomic, transcriptomic, proteomic, microbiomic, and radiomic), together with the use of artificial intelligence and the challenges and future directions of these fields.

Globally invariant behavior of oncogenes and random genes at population but not at single cell level

Cancer is widely considered a genetic disease. Notably, recent works have highlighted that every human gene may possibly be associated with cancer. Thus, the distinction between genes that drive oncogenesis and those that are associated to the disease, but do not play a role, requires attention. Here we investigated single cells and bulk (cell-population) datasets of several cancer transcriptomes and proteomes in relation to their healthy counterparts. When analyzed by machine learning and statistical approaches in bulk datasets, both general and cancer-specific oncogenes, as defined by the Cancer Genes Census, show invariant behavior to randomly selected gene sets of the same size for all cancers. However, when protein-protein interaction analyses were performed, the oncogenes-derived networks show higher connectivity than those relative to random genes. Moreover, at single-cell scale, we observe variant behavior in a subset of oncogenes for each considered cancer type. Moving forward, we concur that the role of oncogenes needs to be further scrutinized by adopting protein causality and higher-resolution single-cell analyses.

Artificial Intelligence-Assisted Transcriptomic Analysis to Advance Cancer Immunotherapy

The emergence of immunotherapy has dramatically changed the cancer treatment paradigm and generated tremendous promise in precision medicine. However, cancer immunotherapy is greatly limited by its low response rates and immune-related adverse events. Transcriptomics technology is a promising tool for deciphering the molecular underpinnings of immunotherapy response and therapeutic toxicity. In particular, applying single-cell RNA-seq (scRNA-seq) has deepened our understanding of tumor heterogeneity and the microenvironment, providing powerful help for developing new immunotherapy strategies. Artificial intelligence (AI) technology in transcriptome analysis meets the need for efficient handling and robust results. Specifically, it further extends the application scope of transcriptomic technologies in cancer research. AI-assisted transcriptomic analysis has performed well in exploring the underlying mechanisms of drug resistance and immunotherapy toxicity and predicting therapeutic response, with profound significance in cancer treatment. In this review, we summarized emerging AI-assisted transcriptomic technologies. We then highlighted new insights into cancer immunotherapy based on AI-assisted transcriptomic analysis, focusing on tumor heterogeneity, the tumor microenvironment, immune-related adverse event pathogenesis, drug resistance, and new target discovery. This review summarizes solid evidence for immunotherapy research, which might help the cancer research community overcome the challenges faced by immunotherapy.

Expression heterogeneity of ABC-transporter family genes and chemosensitivity genes in gastric tumor, carcinomatosis and lymph node metastases

Introduction. Metastatic tumors (particularly gastric cancer) have been found to be characterized by heterogeneity between the primary tumor and metastases. This type of heterogeneity comes to the fore when treating primary-metastatic forms of tumor and is an important reason for the low effectiveness of their treatment. In this regard, comparative analysis of ABC-transporter gene expression and chemosensitivity genes will allow to characterize to a certain extent the resistance and sensitivity of primary tumor, carcinomatosis and metastases to therapy and provide the basis for personalized treatment approach.Aim. To evaluate expression heterogeneity of ABC-transporter genes and chemosensitivity genes in gastric tumor, carcinomatosis and lymph node metastases.Materials and methods. Overall 41 patients with disseminated gastric cancer stage IV with carcinomatosis of peritoneum were included in the investigation. All patients underwent surgery according to Roux palliative gastrectomy. After surgery patients underwent chemotherapy depending on indications. RNA was isolated using RNeasy Plus mini kit (Qiagen, Germany). The expression level of ABC transporter genes (ABCB1, ABCC1, ABCC2, ABCC5, ABCG1, ABCG2) and chemosensitivity genes (BRCA1, RRM1, ERCC1, TOP1, TOP2α, TUBβ3, TYMS, GSTP1) was assessed by reverse transcription polymerase chain reaction (RT-PCR) in primary tumor, carcinomatosis and lymph node metastases.Results. The expression levels of the genes under study were shown to vary widely. For ABC transporter genes, ABCG1 (3.1 ± 1.1; max 32.0), ABCG2 (7.9 ± 2.3; max 54.1), ABCG2 (9.6 ± 3.8; max 101.0) were the most expressed genes in gastric tumor tissue, carcinomatosis and lymph node metastasis, respectively. Hyperexpression among chemosensitivity genes at all three sites was characteristic only of TOP2α (17.2 ± 6.0; max. 161.9; 10.8 ± 4.1; max. 105.1; 35.3 ± 0.8; max. 439.6, respectively). We found that TOP2α and BRCA1 gene expression levels were higher in lymph node metastasis compared with gastric tumor tissue and carcinomatosis (at p = 0.005 and p = 0.001). Whereas ABCC1 gene expression was statistically significantly higher in carcinomatosis (p = 0.03).Conclusion. Thus, a high level of expression heterogeneity is observed in gastric cancer, which affects the expression patterns of various genes in different localizations. The expression profile can be used to determine the level of heterogeneity and approach to personalized therapy tactics.

Single-cell biology uncovers apoptotic cell death and its spatial organization as a potential modifier of tumor diversity in HCC

BACKGROUND AND AIMS

HCC is a highly aggressive and heterogeneous cancer type with limited treatment options. Identifying drivers of tumor heterogeneity may lead to better therapeutic options and favorable patient outcomes. We investigated whether apoptotic cell death and its spatial architecture is linked to tumor molecular heterogeneity using single-cell in situ hybridization analysis.

APPROACH AND RESULTS

We analyzed 254 tumor samples from two HCC cohorts using tissue microarrays. We developed a mathematical model to quantify cellular diversity among HCC samples using two tumor markers, cyclin-dependent kinase inhibitor 3 and protein regulator of cytokinesis 1 as surrogates for heterogeneity and caspase 3 (CASP3) as an apoptotic cell death marker. We further explored the impact of potential dying-cell hubs on tumor cell diversity and patient outcome by density contour mapping and spatial proximity analysis. We also developed a selectively controlled in vitro model of cell death using CRISPR/CRISPR-associated 9 to determine therapy response and growth under hypoxic conditions. We found that increasing levels of CASP3⁺ tumor cells are associated with higher tumor diversity. Interestingly, we discovered regions of densely populated CASP3⁺ , which we refer to as CASP3⁺ cell islands, in which the nearby cellular heterogeneity was found to be the greatest compared to cells farther away from these islands and that this phenomenon was associated with survival. Additionally, cell culture experiments revealed that higher levels of cell death, accompanied by increased CASP3 expression, led to greater therapy resistance and growth under hypoxia.

CONCLUSIONS

These results are consistent with the hypothesis that increased apoptotic cell death may lead to greater tumor heterogeneity and thus worse patient outcomes.

Machine Learning Predictor of Immune Checkpoint Blockade Response in Gastric Cancer

Predicting responses to immune checkpoint blockade (ICB) lacks official standards despite the discovery of several markers. Expensive drugs and different reactivities for each patient are the main disadvantages of immunotherapy. Gastric cancer is refractory and stem-like in nature and does not respond to immunotherapy. In this study, we aimed to identify a characteristic gene that predicts ICB response in gastric cancer and discover a drug target for non-responders. We built and evaluated a model using four machine learning algorithms for two cohorts of bulk and single-cell RNA seq to predict ICB response in gastric cancer patients. Through the LASSO feature selection, we discovered a marker gene signature that distinguishes responders from non-responders. VCAN, a candidate characteristic gene selected by all four machine learning algorithms, had a significantly high prevalence in non-responders (p = 0.0019) and showed a poor prognosis (p = 0.0014) at high expression values. This is the first study to discover a signature gene for predicting ICB response in gastric cancer by molecular subtype and provides broad insights into the treatment of stem-like immuno-oncology through precision medicine.

Potential advantages of FDG-PET radiomic feature map for target volume delineation in lung cancer radiotherapy

PURPOSE

To investigate the potential benefits of FDG PET radiomic feature maps (RFMs) for target delineation in non-small cell lung cancer (NSCLC) radiotherapy.

METHODS

Thirty-two NSCLC patients undergoing FDG PET/CT imaging were included. For each patient, nine grey-level co-occurrence matrix (GLCM) RFMs were generated. gross target volume (GTV) and clinical target volume (CTV) were contoured on CT (GTV_CT , CTV_CT ), PET (GTV_PET40 , CTV_PET40 ), and RFMs (GTV_RFM , CTV_RFM ,). Intratumoral heterogeneity areas were segmented as GTV_PET50-Boost and radiomic boost target volume (RTV_Boost ) on PET and RFMs, respectively. GTV_CT in homogenous tumors and GTV_PET40 in heterogeneous tumors were considered as GTV_{gold standard} (GTV_GS ). One-way analysis of variance was conducted to determine the threshold that finds the best conformity for GTV_RFM with GTV_GS . Dice similarity coefficient (DSC) and mean absolute percent error (MAPE) were calculated. Linear regression analysis was employed to report the correlations between the gold standard and RFM-derived target volumes.

RESULTS

Entropy, contrast, and Haralick correlation (H-correlation) were selected for tumor segmentation. The threshold values of 80%, 50%, and 10% have the best conformity of GTV_RFM-entropy , GTV_RFM-contrast , and GTV_{RFM-H-correlation} with GTV_GS , respectively. The linear regression results showed a positive correlation between GTV_GS and GTV_RFM-entropy (r = 0.98, p < 0.001), between GTV_GS and GTV_RFM-contrast (r = 0.93, p < 0.001), and between GTV_GS and GTV_{RFM-H-correlation} (r = 0.91, p < 0.001). The average threshold values of 45% and 15% were resulted in the best segmentation matching between CTV_RFM-entropy and CTV_RFM-contrast with CTV_GS , respectively. Moreover, we used RFM to determine RTV_Boost in the heterogeneous tumors. Comparison of RTV_Boost with GTV_PET50-Boost MAPE showed the volume error differences of 31.7%, 36%, and 34.7% in RTV_{Boost-entropy} , RTV_{Boost-contrast} , and RTV_{Boost-H-correlation} , respectively.

CONCLUSIONS

FDG PET-based radiomics features in NSCLC demonstrated a promising potential for decision support in radiotherapy, helping radiation oncologists delineate tumors and generate accurate segmentation for heterogeneous region of tumors.

Copy Number Alterations as Novel Biomarkers and Therapeutic Targets in Colorectal Cancer

In colorectal cancer, somatic mutations have played an important role as prognostic and predictive biomarkers, with some also functioning as therapeutic targets. Another genetic aberration that has shown significance in colorectal cancer is copy number alterations (CNAs). CNAs occur when a change to the DNA structure propagates gain/amplification or loss/deletion in sections of DNA, which can often lead to changes in protein expression. Multiple techniques have been developed to detect CNAs, including comparative genomic hybridization with microarray, low pass whole genome sequencing, and digital droplet PCR. In this review, we summarize key findings in the literature regarding the role of CNAs in the pathogenesis of colorectal cancer, from adenoma to carcinoma to distant metastasis, and discuss the roles of CNAs as prognostic and predictive biomarkers in colorectal cancer.

A primer on machine learning techniques for genomic applications

High throughput sequencing technologies have enabled the study of complex biological aspects at single nucleotide resolution, opening the big data era. The analysis of large volumes of heterogeneous "omic" data, however, requires novel and efficient computational algorithms based on the paradigm of Artificial Intelligence. In the present review, we introduce and describe the most common machine learning methodologies, and lately deep learning, applied to a variety of genomics tasks, trying to emphasize capabilities, strengths and limitations through a simple and intuitive language. We highlight the power of the machine learning approach in handling big data by means of a real life example, and underline how described methods could be relevant in all cases in which large amounts of multimodal genomic data are available.

Establishment of a Gene Signature to Predict Prognosis for Patients with Lung Adenocarcinoma

Accumulating evidence indicates that the reliable gene signature may serve as an independent prognosis factor for lung adenocarcinoma (LUAD) diagnosis. Here, we sought to identify a risk score signature for survival prediction of LUAD patients. In the Gene Expression Omnibus (GEO) database, GSE18842, GSE75037, GSE101929, and GSE19188 mRNA expression profiles were downloaded to screen differentially expressed genes (DEGs), which were used to establish a protein-protein interaction network and perform clustering module analysis. Univariate and multivariate proportional hazards regression analyses were applied to develop and validate the gene signature based on the TCGA dataset. The signature genes were then verified on GEPIA, Oncomine, and HPA platforms. Expression levels of corresponding genes were also measured by qRT-PCR and Western blotting in HBE, A549, and PC-9 cell lines. The prognostic signature based on eight genes (TTK, HMMR, ASPM, CDCA8, KIF2C, CCNA2, CCNB2, and MKI67) was established, which was independent of other clinical factors. The risk model offered better discrimination between risk groups, and patients with high-risk scores tended to have poor survival rate at 1-, 3- and 5-year follow-up. The model also presented better survival prediction in cancer-specific cohorts of age, gender, clinical stage III/IV, primary tumor 1/2, and lymph node metastasis 1/2. The signature genes, moreover, were highly expressed in A549 and PC-9 cells. In conclusion, the risk score signature could be used for prognostic estimation and as an independent risk factor for survival prediction in patients with LUAD.