Comprehensive Potential of Artificial Intelligence for Predicting PD-L1 Expression and EGFR Mutations in Lung Cancer: A Systematic Review and Meta-Analysis

OBJECTIVE

To evaluate the methodological quality and the predictive performance of artificial intelligence (AI) for predicting programmed death ligand 1 (PD-L1) expression and epidermal growth factor receptors (EGFR) mutations in lung cancer (LC) based on systematic review and meta-analysis.

METHODS

AI studies based on PET/CT, CT, PET, and immunohistochemistry (IHC)-whole-slide image (WSI) were included to predict PD-L1 expression or EGFR mutations in LC. The modified Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool was used to evaluate the methodological quality. A comprehensive meta-analysis was conducted to analyze the overall area under the curve (AUC). The Cochrane diagnostic test and I2 statistics were used to assess the heterogeneity of the meta-analysis.

RESULTS

A total of 45 AI studies were included, of which 10 were used to predict PD-L1 expression and 35 were used to predict EGFR mutations. Based on the analysis using the QUADAS-2 tool, 37 studies achieved a high-quality score of 7. In the meta-analysis of PD-L1 expression levels, the overall AUCs for PET/CT, CT, and IHC-WSI were 0.80 (95% confidence interval [CI], 0.77-0.84), 0.74 (95% CI, 0.69-0.77), and 0.95 (95% CI, 0.93-0.97), respectively. For EGFR mutation status, the overall AUCs for PET/CT, CT, and PET were 0.85 (95% CI, 0.81-0.88), 0.83 (95% CI, 0.80-0.86), and 0.75 (95% CI, 0.71-0.79), respectively. The Cochrane Diagnostic Test revealed an I2 value exceeding 50%, indicating substantial heterogeneity in the PD-L1 and EGFR meta-analyses. When AI was combined with clinicopathological features, the enhancement in predicting PD-L1 expression was not substantial, whereas the prediction of EGFR mutations showed improvement compared to the CT and PET models, albeit not significantly so compared to the PET/CT models.

CONCLUSIONS

The overall performance of AI in predicting PD-L1 expression and EGFR mutations in LC has promising clinical implications.

Protein structural context of cancer mutations reveals molecular mechanisms and candidate driver genes

Advances in protein structure determination and modeling allow us to study the structural context of human genetic variants on an unprecedented scale. Here, we analyze millions of cancer-associated missense mutations based on their structural locations and predicted perturbative effects. By considering the collective properties of mutations at the level of individual proteins, we identify distinct patterns associated with tumor suppressors and oncogenes. Tumor suppressors are enriched in structurally damaging mutations, consistent with loss-of-function mechanisms, while oncogene mutations tend to be structurally mild, reflecting selection for gain-of-function driver mutations and against loss-of-function mutations. Although oncogenes are difficult to distinguish from genes with no role in cancer using only structural damage, we find that the three-dimensional clustering of mutations is highly predictive. These observations allow us to identify candidate driver genes and speculate about their molecular roles, which we expect will have general utility in the analysis of cancer sequencing data.

Identification of oncogenes and tumor-suppressor genes with hepatocellular carcinoma: A comprehensive analysis based on TCGA and GEO datasets

Aim: Existing targeted therapies for hepatocellular carcinoma (HCC) are resistant and have limitations. It is crucial to find new HCC-related target genes. Methods: RNA-sequencing data of HCC were gathered from The Cancer Genome Atlas and Gene Expression Omnibus datasets. Initially, differentially expressed genes between normal and tumor tissues were identified from four Gene Expression Omnibus datasets, GSE36376, GSE102079, GSE54236, and GSE45267. GO terms and KEGG pathway enrichment analyses were performed to explore the potential biological functions of differentially expressed genes. A PPI network was constructed by using the STRING database, and up-regulated and down-regulated hub genes were defined through 12 topological approaches. Subsequently, the correlation bounded by up-regulated genes and down-regulated genes in the diagnosis, prognosis, and clinicopathological features of HCC was analyzed. Beyond a shadow of doubt, the key oncogene PBK and tumor suppressor gene F9 were screened out, and the specific mechanism was investigated through GSEA enrichment analysis and immune correlation analysis. The role of PBK in HCC was further verified by western blot, CCK8, transwell, and tube formation experiments. Results: CDCA5, CDC20, PBK, PRC1, TOP2A, and NCAPG are good indicators of HCC diagnosis and prognosis. The low expressions of F9, AFM, and C8B indicate malignant progression and poor prognosis of HCC. PBK was found to be closely related to VEGF, VEGFR, and PDGFR pathways. Experiments showed that PBK promotes HCC cell proliferation, migration, invasion, and tube formation in HUVEC cells. F9 was negatively correlated with the degree of immune infiltration, and low expression of F9 suggested a poor response to immunotherapy. Conclusion: The role of HCC-related oncogenes and tumor-suppressor genes in diagnosis and prognosis was identified. In addition, we have found that PBK may promote tumor proliferation through angiogenesis and F9 may be a predictor of tumor immunotherapy response.

Underlying mechanisms of epithelial splicing regulatory proteins in cancer progression

Cancer is the second-leading disease-related cause of global mortality after cardiovascular disease. Despite significant advances in cancer therapeutic strategies, cancer remains one of the major obstacles to human life extension. Cancer pathogenesis is extremely complicated and not fully understood. Epithelial splicing regulatory proteins (ESRPs), including ESRP1 and ESRP2, belong to the heterogeneous nuclear ribonucleoprotein family of RNA-binding proteins and are crucial regulators of the alternative splicing of messenger RNAs (mRNAs). The expression and activity of ESRPs are modulated by various mechanisms, including post-translational modifications and non-coding RNAs. Although a growing body of evidence suggests that ESRP dysregulation is closely associated with cancer progression, the detailed mechanisms remain inconclusive. In this review, we summarize recent findings on the structures, functions, and regulatory mechanisms of ESRPs and focus on their underlying mechanisms in cancer progression. We also highlight the clinical implications of ESRPs as prognostic biomarkers and therapeutic targets in cancer treatment. The information reviewed herein could be extremely beneficial to the development of individualized therapeutic strategies for cancer patients.

Expression of neurofibromin 1 in colorectal cancer and cetuximab resistance

Neurofibromin 1 (NF1) is a tumor suppressor that has been previously reported to regulate RAS‑MAPK signaling. The present study investigated the possible relationship between NF1 expression and anti‑EGFR antibody (cetuximab) sensitivity in colorectal cancer cell lines. In addition, primary or metastatic colorectal cancer samples from patients treated with cetuximab were assessed for the association of cetuximab sensitivity. The quantities of the NF1 transcript, NF1‑related pathway enrichment and NF1 mutation profile were measured and investigated using RNA sequencing and targeted DNA sequencing. Based on growth inhibition and colony formation assay results, cell lines were designated to be cetuximab‑sensitive (NCI‑H508 and Caco2) or cetuximab‑resistant (KM12C and SM480). Western blotting revealed NF1 was highly expressed in cetuximab‑sensitive cell lines whilst there was little expression in their cetuximab‑resistant counterparts. Knocking down NF1 expression using small interfering RNA in the cetuximab‑sensitive cell lines enhanced the phosphorylation of MEK and ERK according to western blotting. NF1 knockdown also reduced apoptosis, as observed by the decreased number of apoptotic bodies by DAPI nuclear staining and reduced cleavage of caspase and poly‑(ADP ribose) polymerase. NF1 overexpression by transfection with GTPase‑activating protein‑related domain subunit rendered the cetuximab‑resistant cell lines, KM12C and SW480, more susceptible to cetuximab‑induced apoptosis. RNA sequencing of 111 RAS and BRAF^V600 wild‑type tumor samples collected from cetuximab‑treated patients with metastatic colorectal cancer revealed that the pre‑treatment NF1 expression levels were not associated with the cetuximab response. However, tumor samples obtained after cetuximab treatment displayed slightly lower NF1 transcript levels compared with those in the pre‑treatment samples, suggesting that exposure to the anti‑EGFR antibody may be associated with reduced NF1 expression levels. Next‑generation sequencing revealed that the frequency of inactivating mutations in NF1 were rare (1.8%) in patients with colorectal cancer and were not associated with the protein expression levels of NF1 except for in a small number of cases (0.5%), where the biallelic inactivation of NF1 was observed. To conclude, the present study showed that modification of NF1 expression can affect sensitivity to cetuximab in colorectal cancer cell lines, though a limitation exists in terms of its potential application as a biomarker for RAS and BRAF^V600 wild‑type tumors.

Long Noncoding RNAs in Gastrointestinal Cancer: Tumor Suppression Versus Tumor Promotion

Approximately 80% of the human genome harbors biochemical marks of active transcription that its majority transcribes to noncoding RNAs, namely long noncoding RNAs (lncRNAs). LncRNAs are heterogeneous RNA transcripts that regulate critical biological processes such as cell survival and death. They involve in the progression of different cancers by affecting transcriptional and post-transcriptional modifications as well as epigenetic control of numerous tumor suppressors and oncogenes. Recent findings show that aberrant expression of lncRNAs is associated with tumor initiation, progression, invasion, and overall survival of patients with gastrointestinal (GI) cancers. Some lncRNAs play as tumor suppressors in all GI cancers, but others play as tumor promoters. However, some other lncRNAs might function as a tumor suppressor in one GI cancer, but as a tumor promoter in another GI cancer type. This fact highlights possible context dependency of the expression patterns and roles of at least some lncRNAs in GI cancer development and progression. Here, we review the functional relation of lncRNAs involved in the development and progression of GI cancer by focusing on their roles as tumor suppressor and tumor promoter genes.

NRF2 and the Ambiguous Consequences of Its Activation during Initiation and the Subsequent Stages of Tumourigenesis

NF-E2 p45-related factor 2 (NRF2, encoded in the human by NFE2L2) mediates short-term adaptation to thiol-reactive stressors. In normal cells, activation of NRF2 by a thiol-reactive stressor helps prevent, for a limited period of time, the initiation of cancer by chemical carcinogens through induction of genes encoding drug-metabolising enzymes. However, in many tumour types, NRF2 is permanently upregulated. In such cases, its overexpressed target genes support the promotion and progression of cancer by suppressing oxidative stress, because they constitutively increase the capacity to scavenge reactive oxygen species (ROS), and they support cell proliferation by increasing ribonucleotide synthesis, serine biosynthesis and autophagy. Herein, we describe cancer chemoprevention and the discovery of the essential role played by NRF2 in orchestrating protection against chemical carcinogenesis. We similarly describe the discoveries of somatic mutations in NFE2L2 and the gene encoding the principal NRF2 repressor, Kelch-like ECH-associated protein 1 (KEAP1) along with that encoding a component of the E3 ubiquitin-ligase complex Cullin 3 (CUL3), which result in permanent activation of NRF2, and the recognition that such mutations occur frequently in many types of cancer. Notably, mutations in NFE2L2, KEAP1 and CUL3 that cause persistent upregulation of NRF2 often co-exist with mutations that activate KRAS and the PI3K-PKB/Akt pathway, suggesting NRF2 supports growth of tumours in which KRAS or PKB/Akt are hyperactive. Besides somatic mutations, NRF2 activation in human tumours can occur by other means, such as alternative splicing that results in a NRF2 protein which lacks the KEAP1-binding domain or overexpression of other KEAP1-binding partners that compete with NRF2. Lastly, as NRF2 upregulation is associated with resistance to cancer chemotherapy and radiotherapy, we describe strategies that might be employed to suppress growth and overcome drug resistance in tumours with overactive NRF2.

Standardization of Somatic Variant Classifications in Solid and Haematological Tumours by a Two-Level Approach of Biological and Clinical Classes: An Initiative of the Belgian ComPerMed Expert Panel

In most diagnostic laboratories, targeted next-generation sequencing (NGS) is currently the default assay for the detection of somatic variants in solid as well as haematological tumours. Independent of the method, the final outcome is a list of variants that differ from the human genome reference sequence of which some may relate to the establishment of the tumour in the patient. A critical point towards a uniform patient management is the assignment of the biological contribution of each variant to the malignancy and its subsequent clinical impact in a specific malignancy. These so-called biological and clinical classifications of somatic variants are currently not standardized and are vastly dependent on the subjective analysis of each laboratory. This subjectivity can thus result in a different classification and subsequent clinical interpretation of the same variant. Therefore, the ComPerMed panel of Belgian experts in cancer diagnostics set up a working group with the goal to harmonize the biological classification and clinical interpretation of somatic variants detected by NGS. This effort resulted in the establishment of a uniform, two-level classification workflow system that should enable high consistency in diagnosis, prognosis, treatment and follow-up of cancer patients. Variants are first classified into a tumour-independent biological five class system and subsequently in a four tier ACMG clinical classification. Here, we describe the ComPerMed workflow in detail including examples for each step of the pipeline. Moreover, this workflow can be implemented in variant classification software tools enabling automatic reporting of NGS data, independent of panel, method or analysis software.

Biophysical simulations and structure-based modeling of residue interaction networks in the tumor suppressor proteins reveal functional role of cancer mutation hotspots in molecular communication

In the current study, we have combined molecular simulations and energetic analysis with dynamics-based network modeling and perturbation response scanning to determine molecular signatures of mutational hotspot residues in the p53, PTEN, and SMAD4 tumor suppressor proteins. By examining structure, energetics and dynamics of these proteins, we have shown that inactivating mutations preferentially target a group of structurally stable residues that play a fundamental role in global propagation of dynamic fluctuations and mediating allosteric interaction networks. Through integration of long-range perturbation dynamics and network-based approaches, we have quantified allosteric potential of residues in the studied proteins. The results have revealed that mutational hotspot sites often correspond to high centrality mediating centers of the residue interaction networks that are responsible for coordination of global dynamic changes and allosteric signaling. Our findings have also suggested that structurally stable mutational hotpots can act as major effectors of allosteric interactions and mutations in these positions are typically associated with severe phenotype. Modeling of shortest inter-residue pathways has shown that mutational hotspot sites can also serve as key mediating bridges of allosteric communication in the p53 and PTEN protein structures. Multiple regression models have indicated that functional significance of mutational hotspots can be strongly associated with the network signatures serving as robust predictors of critical regulatory positions responsible for loss-of-function phenotype. The results of this computational investigation are compared with the experimental studies and reveal molecular signatures of mutational hotspots, providing a plausible rationale for explaining and localizing disease-causing mutations in tumor suppressor genes.

Identification and analysis of mutational hotspots in oncogenes and tumour suppressors

Background

The key to interpreting the contribution of a disease-associated mutation in the development and progression of cancer is an understanding of the consequences of that mutation both on the function of the affected protein and on the pathways in which that protein is involved. Protein domains encapsulate function and position-specific domain based analysis of mutations have been shown to help elucidate their phenotypes.

Results

In this paper we examine the domain biases in oncogenes and tumour suppressors, and find that their domain compositions substantially differ. Using data from over 30 different cancers from whole-exome sequencing cancer genomic projects we mapped over one million mutations to their respective Pfam domains to identify which domains are enriched in any of three different classes of mutation; missense, indels or truncations. Next, we identified the mutational hotspots within domain families by mapping small mutations to equivalent positions in multiple sequence alignments of protein domains We find that gain of function mutations from oncogenes and loss of function mutations from tumour suppressors are normally found in different domain families and when observed in the same domain families, hotspot mutations are located at different positions within the multiple sequence alignment of the domain.

Conclusions

By considering hotspots in tumour suppressors and oncogenes independently, we find that there are different specific positions within domain families that are particularly suited to accommodate either a loss or a gain of function mutation. The position is also dependent on the class of mutation. We find rare mutations co-located with well-known functional mutation hotspots, in members of homologous domain superfamilies, and we detect novel mutation hotspots in domain families previously unconnected with cancer. The results of this analysis can be accessed through the MOKCa database (http://strubiol.icr.ac.uk/extra/MOKCa).

Bioinformatics in translational drug discovery

Bioinformatics approaches are becoming ever more essential in translational drug discovery both in academia and within the pharmaceutical industry. Computational exploitation of the increasing volumes of data generated during all phases of drug discovery is enabling key challenges of the process to be addressed. Here, we highlight some of the areas in which bioinformatics resources and methods are being developed to support the drug discovery pipeline. These include the creation of large data warehouses, bioinformatics algorithms to analyse 'big data' that identify novel drug targets and/or biomarkers, programs to assess the tractability of targets, and prediction of repositioning opportunities that use licensed drugs to treat additional indications.

'Big data' approaches for novel anti-cancer drug discovery

INTRODUCTION

The development of improved cancer therapies is frequently cited as an urgent unmet medical need. Recent advances in platform technologies and the increasing availability of biological 'big data' are providing an unparalleled opportunity to systematically identify the key genes and pathways involved in tumorigenesis. The discoveries made using these new technologies may lead to novel therapeutic interventions. Areas covered: The authors discuss the current approaches that use 'big data' to identify cancer drivers. These approaches include the analysis of genomic sequencing data, pathway data, multi-platform data, identifying genetic interactions such as synthetic lethality and using cell line data. They review how big data is being used to identify novel drug targets. The authors then provide an overview of the available data repositories and tools being used at the forefront of cancer drug discovery. Expert opinion: Targeted therapies based on the genomic events driving the tumour will eventually inform treatment protocols. However, using a tailored approach to treat all tumour patients may require developing a large repertoire of targeted drugs.