Bioinformatics analysis of the genes involved in the extension of prostate cancer to adjacent lymph nodes by supervised and unsupervised machine learning methods: The role of SPAG1 and PLEKHF2

Using machine learning to develop preoperative model for lymph node metastasis in patients with bladder urothelial carcinoma

BACKGROUND

Lymph node metastasis (LNM) is associated with worse prognosis in bladder urothelial carcinoma (BUC) patients. This study aimed to develop and validate machine learning (ML) models to preoperatively predict LNM in BUC patients treated with radical cystectomy (RC).

METHODS

We retrospectively collected demographic, pathological, imaging, and laboratory information of BUC patients who underwent RC and bilateral lymphadenectomy in our institution. Patients were randomly categorized into training set and testing set. Five ML algorithms were utilized to establish prediction models. The performance of each model was assessed by the area under the receiver operating characteristic curve (AUC) and accuracy. Finally, we calculated the corresponding variable coefficients based on the optimal model to reveal the contribution of each variable to LNM.

RESULTS

A total of 524 and 131 BUC patients were finally enrolled into training set and testing set, respectively. We identified that the support vector machine (SVM) model had the best prediction ability with an AUC of 0.934 (95% confidence interval [CI]: 0.903-0.964) and accuracy of 0.916 in the training set, and an AUC of 0.855 (95%CI: 0.777-0.933) and accuracy of 0.809 in the testing set. The SVM model contained 14 predictors, and positive lymph node in imaging contributed the most to the prediction of LNM in BUC patients.

CONCLUSIONS

We developed and validated the ML models to preoperatively predict LNM in BUC patients treated with RC, and identified that the SVM model with 14 variables had the best performance and high levels of clinical applicability.

Assessment of Prostate and Bladder Cancer Genomic Biomarkers Using Artificial Intelligence: a Systematic Review

PURPOSE OF REVIEW

The aim of the systematic review is to assess AI's capabilities in the genetics of prostate cancer (PCa) and bladder cancer (BCa) to evaluate target groups for such analysis as well as to assess its prospects in daily practice.

RECENT FINDINGS

In total, our analysis included 27 articles: 10 articles have reported on PCa and 17 on BCa, respectively. The AI algorithms added clinical value and demonstrated promising results in several fields, including cancer detection, assessment of cancer development risk, risk stratification in terms of survival and relapse, and prediction of response to a specific therapy. Besides clinical applications, genetic analysis aided by the AI shed light on the basic urologic cancer biology. We believe, our results of the AI application to the analysis of PCa, BCa data sets will help to identify new targets for urological cancer therapy. The integration of AI in genomic research for screening and clinical applications will evolve with time to help personalizing chemotherapy, prediction of survival and relapse, aid treatment strategies such as reducing frequency of diagnostic cystoscopies, and clinical decision support, e.g., by predicting immunotherapy response. These factors will ultimately lead to personalized and precision medicine thereby improving patient outcomes.

Pierre CL, Bimschleger H, Gao J, Cheng R, Mohammadi P, Chen H, Palmer AA, Polesskaya O, Jablonski MM. Genome-wide association study finds multiple loci associated with intraocular pressure in HS rats

Elevated intraocular pressure (IOP) is influenced by environmental and genetic factors. Increased IOP is a major risk factor for most types of glaucoma, including primary open angle glaucoma (POAG). Investigating the genetic basis of IOP may lead to a better understanding of the molecular mechanisms of POAG. The goal of this study was to identify genetic loci involved in regulating IOP using outbred heterogeneous stock (HS) rats. HS rats are a multigenerational outbred population derived from eight inbred strains that have been fully sequenced. This population is ideal for a genome-wide association study (GWAS) owing to the accumulated recombinations among well-defined haplotypes, the relatively high allele frequencies, the accessibility to a large collection of tissue samples, and the large allelic effect size compared to human studies. Both male and female HS rats (N = 1,812) were used in the study. Genotyping-by-sequencing was used to obtain ∼3.5 million single nucleotide polymorphisms (SNP) from each individual. SNP heritability for IOP in HS rats was 0.32, which agrees with other studies. We performed a GWAS for the IOP phenotype using a linear mixed model and used permutation to determine a genome-wide significance threshold. We identified three genome-wide significant loci for IOP on chromosomes 1, 5, and 16. Next, we sequenced the mRNA of 51 whole eye samples to find cis-eQTLs to aid in identification of candidate genes. We report 5 candidate genes within those loci: Tyr, Ctsc, Plekhf2, Ndufaf6 and Angpt2. Tyr, Ndufaf6 and Angpt2 genes have been previously implicated by human GWAS of IOP-related conditions. Ctsc and Plekhf2 genes represent novel findings that may provide new insight into the molecular basis of IOP. This study highlights the efficacy of HS rats for investigating the genetics of elevated IOP and identifying potential candidate genes for future functional testing.

Genomic Landscape Alterations in Primary Tumor and Matched Lymph Node Metastasis in Hormone-Naïve Prostate Cancer Patients

Background: Prostate cancer (PCa) is a disease with a wide range of clinical manifestations. Up to the present date, the genetic understanding of patients with favorable or unfavorable prognosis is gaining interest for giving the appropriate tailored treatment. We aimed to investigate genetic changes associated with lymph node metastasis in a cohort of hormone-naïve Pca patients. Methods: We retrospectively analyzed data from 470 patients who underwent surgery for PCa between 2010 and 2020 at the Department of Urology, University of Catania. Inclusion criteria were patients with lymph node metastasis and patients with PCa with extra capsular extension (pT3) and negative lymph node metastasis. The final cohort consisted of 17 different patients (11 PCa with lymph node metastasis and 6 PCa without lymph node metastasis). Through the cBioPortal online tool, we analyzed gene alterations and their correlations with clinical factors. Results: A total of 688 intronic, synonym and nonsynonym mutations were sequenced. The gene with the most sequenced mutations was ERBB4 (83 mutations, 12% of 688 total), while the ones with the lower percentage of mutations were AKT1, FGFR2 and MLH1 (1 mutation alone, 0.14%). Conclusion: In the present study we found mostly concordance concerning the ERBB4 mutation between both primary PCa samples and matched lymph node metastasis, underlining that the identification of alterations in the primary tumor is extremely important for cancer prognosis prediction.

The PH Domain and C-Terminal polyD Motif of Phafin2 Exhibit a Unique Concurrence in Animals

Phafin2, a member of the Phafin family of proteins, contributes to a plethora of cellular activities including autophagy, endosomal cargo transportation, and macropinocytosis. The PH and FYVE domains of Phafin2 play key roles in membrane binding, whereas the C-terminal poly aspartic acid (polyD) motif specifically autoinhibits the PH domain binding to the membrane phosphatidylinositol 3-phosphate (PtdIns3P). Since the Phafin2 FYVE domain also binds PtdIns3P, the role of the polyD motif remains unclear. In this study, bioinformatics tools and resources were employed to determine the concurrence of the PH-FYVE module with the polyD motif among Phafin2 and PH-, FYVE-, or polyD-containing proteins from bacteria to humans. FYVE was found to be an ancient domain of Phafin2 and is related to proteins that are present in both prokaryotes and eukaryotes. Interestingly, the polyD motif only evolved in Phafin2 and PH- or both PH-FYVE-containing proteins in animals. PolyD motifs are absent in PH domain-free FYVE-containing proteins, which usually display cellular trafficking or autophagic functions. Moreover, the prediction of the Phafin2-interacting network indicates that Phafin2 primarily cross-talks with proteins involved in autophagy, protein trafficking, and neuronal function. Taken together, the concurrence of the polyD motif with the PH domain may be associated with complex cellular functions that evolved specifically in animals.

A transcriptomic signature for prostate cancer relapse prediction identified from the differentially expressed genes between TP53 mutant and wild-type tumors

For prostate cancer (PCa) patients, biochemical recurrence (BCR) is the first sign of disease relapse and the subsequent metastasis. TP53 mutations are relatively prevalent in advanced PCa forms. We aimed to utilize this knowledge to identify robust transcriptomic signatures for BCR prediction in patients with Gleason score ≥ 7 cancers, which cause most PCa deaths. Using the TCGA-PRAD dataset and the novel data-driven stochastic approach proposed in this study, we identified a 25-gene signature from the genes whose expression in tumors was associated with TP53 mutation statuses. The predictive strength of the signature was assessed by AUC and Fisher’s exact test p-value according to the output of support vector machine-based cross validation. For the TCGA-PRAD dataset, the AUC and p-value were 0.837 and 5 × 10^–13, respectively. For five external datasets, the AUCs and p-values ranged from 0.632 to 0.794 and 6 × 10^–2 to 5 × 10^–5, respectively. The signature also performed well in predicting relapse-free survival (RFS). The signature-based transcriptomic risk scores (TRS) explained 28.2% of variation in RFS on average. The combination of TRS and clinicopathologic prognostic factors explained 23–72% of variation in RFS, with a median of 54.5%. Our method and findings are useful for developing new prognostic tools in PCa and other cancers.

Backbone 1H, 15N, and 13C resonance assignments of the Phafin2 pleckstrin homology domain

Phafin2 is a peripheral protein that triggers cellular signaling from endosomal and lysosomal compartments. The specific subcellular localization of Phafin2 is mediated by the presence of a tandem of phosphatidylinositol 3-phosphate (PtdIns3P)-binding domains, the pleckstrin homology (PH) and the Fab-1, YOTB, Vac1, and EEA1 (FYVE) domains. The requirement for both domains for binding to PtdIns3P still remains unclear. To understand the molecular interactions of the Phafin2 PH domain in detail, we report its nearly complete 1H, 15N, and 13C backbone resonance assignments.

Comprehensive analysis of SPAG1 expression as a prognostic and predictive biomarker in acute myeloid leukemia by integrative bioinformatics and clinical validation

Background

Recently, an increasing number of studies have reported that sperm-associated antigen (SPAG) proteins play crucial roles in solid tumorigenesis, and may serve as potentially helpful biomarkers for cancer diagnosis and prognosis. However, very few studies systematically investigated the expression of SPAG family members and their clinical significance in acute myeloid leukemia (AML).

Methods

The expression of SPAGs and their prognostic significance in AML were determined by a systematic analysis on data gathered from public databases, and the results were validated in clinical samples.

Results

Using public data, we identified only increased SPAG1 expression negatively associated with survival in AML by Cox regression (P < 0.001) and Kaplan–Meier analysis (P < 0.001). The prognostic value of SPAG1 expression was further confirmed in other independent cohorts. Clinically, higher SPAG1 expression was significantly correlated with white blood cell counts (P = 0.014) and French–American–British (FAB) subtypes (P = 0.024). Moreover, higher SPAG1 expression was more common in + 8 patients (P = 0.034), rarely found with t(8;21) (P = 0.014), and correlated with FLT3 (P < 0.001) and DNMT3A mutations (P = 0.001). Despite these associations, multivariate analysis confirmed the independent prognostic value of SPAG1 expression in AML (P < 0.001). Notably, AML patients with higher SPAG1 expression may benefit from hematopoietic stem cell transplantation (HSCT), whereas patients with lower SPAG1 expression appeared less likely to benefit. Finally, we further validated that SPAG1 expression was significantly increased in newly diagnosed AML patients compared with normal controls (P < 0.001) and with AML patients who achieved complete remission (P < 0.001). Additionally, SPAG1 expression could act as a potentially helpful biomarker for the diagnosis and prognosis of AML (P < 0.001 and = 0.034, respectively).

Conclusions

Our findings demonstrated that SPAG1 overexpression may serve as an independent prognostic biomarker and may guide the choice between HSCT and chemotherapy in patients with AML.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12920-022-01193-0.

Construction of the circRNA-miRNA-mRNA Regulatory Network of an Abdominal Aortic Aneurysm to Explore Its Potential Pathogenesis

Background

Abdominal aortic aneurysm (AAA) is a progressive cardiovascular disease, which is a permanent and localized dilatation of the abdominal aorta with potentially fatal consequence of aortic rupture. Dysregulation of circRNAs is correlated with the development of various pathological events in cardiovascular diseases. However, the function of circRNAs in abdominal aortic aneurysm (AAA) is unknown and remains to be explored. This study is aimed at determining the regulatory mechanisms of circRNAs in AAAs. This study was aimed at exploring the underlying molecular mechanisms of abdominal aortic aneurysms based on the competing endogenous RNA (ceRNA) regulatory hypothesis of circRNA, miRNA, and mRNA.

Methods

The expression profiles of circRNAs (GSE144431), miRNAs (GSE62179), and mRNAs (GSE7084, GSE57691, and GSE47472) in human tissue sample from the aneurysm group and normal group were obtained from the Gene Expression Omnibus database, respectively. The circRNA-miRNA-mRNA network was constructed by using Cytoscape 3.7.2 software; then, the protein-protein interaction (PPI) network was constructed by using the STRING database, and the hub genes were identified by using the cytoHubba plug-in. The circRNA-miRNA-hub gene regulatory subnetwork was formed to understand the regulatory axis of hub genes in AAAs.

Results

The present study identified 40 differentially expressed circRNAs (DECs) in the GSE144431, 90 differentially expressed miRNAs (DEmiRs) in the GSE62179, and 168 differentially expressed mRNAs (DEGs) with the same direction regulation (130 downregulated and 38 upregulated) in the GSE7084, GSE57691, and GSE47472 datasets identified regarding AAAs. The miRNA response elements (MREs) of three DECs were then predicted. Four overlapping miRNAs were obtained by intersecting the predicted miRNA and DEmiRs. Then, 17 overlapping mRNAs were obtained by intersecting the predicted target mRNAs of 4 miRNAs with 168 DEGs. Furthermore, the circRNA-miRNA-mRNA network was constructed through 3 circRNAs, 4 miRNAs, and 17 mRNAs, and three hub genes (SOD2, CCR7, and PGRMC1) were identified. Simultaneously, functional enrichment and pathway analysis were performed within genes in the circRNA-miRNA-mRNA network. Three of them (SOD2, CCR7, and PGRMC1) were suggested to be crucial based on functional enrichment, protein-protein interaction, and ceRNA network analysis. Furthermore, the expression of SOD2 and CCR7 may be regulated by hsa_circ_0011449/hsa_circ_0081968/hsa-let-7f-5p; the expression of PGRMC1 may be regulated by hsa_circ_0011449/hsa_circ_0081968-hsa-let-7f-5p/hsa-let-7e-5p.

Conclusion

In conclusion, the ceRNA interaction axis we identified may be an important target for the treatment of abdominal aortic aneurysms. This study provided further understanding of the potential pathogenesis from the perspective of the circRNA-related competitive endogenous RNA network in AAAs.

Applied machine learning in cancer research: A systematic review for patient diagnosis, classification and prognosis

Artificial Intelligence (AI) has recently altered the landscape of cancer research and medical oncology using traditional Machine Learning (ML) algorithms and cutting-edge Deep Learning (DL) architectures. In this review article we focus on the ML aspect of AI applications in cancer research and present the most indicative studies with respect to the ML algorithms and data used. The PubMed and dblp databases were considered to obtain the most relevant research works of the last five years. Based on a comparison of the proposed studies and their research clinical outcomes concerning the medical ML application in cancer research, three main clinical scenarios were identified. We give an overview of the well-known DL and Reinforcement Learning (RL) methodologies, as well as their application in clinical practice, and we briefly discuss Systems Biology in cancer research. We also provide a thorough examination of the clinical scenarios with respect to disease diagnosis, patient classification and cancer prognosis and survival. The most relevant studies identified in the preceding year are presented along with their primary findings. Furthermore, we examine the effective implementation and the main points that need to be addressed in the direction of robustness, explainability and transparency of predictive models. Finally, we summarize the most recent advances in the field of AI/ML applications in cancer research and medical oncology, as well as some of the challenges and open issues that need to be addressed before data-driven models can be implemented in healthcare systems to assist physicians in their daily practice.

Site-Specific and Common Prostate Cancer Metastasis Genes as Suggested by Meta-Analysis of Gene Expression Data

Anticancer therapies mainly target primary tumor growth and little attention is given to the events driving metastasis formation. Metastatic prostate cancer, in comparison to localized disease, has a much worse prognosis. In the work presented here, groups of genes that are common to prostate cancer metastatic cells from bones, lymph nodes, and liver and those that are site-specific were delineated. The purpose of the study was to dissect potential markers and targets of anticancer therapies considering the common characteristics and differences in transcriptional programs of metastatic cells from different secondary sites. To that end, a meta-analysis of gene expression data of prostate cancer datasets from the GEO database was conducted. Genes with differential expression in all metastatic sites analyzed belong to the class of filaments, focal adhesion, and androgen receptor signaling. Bone metastases undergo the largest transcriptional changes that are highly enriched for the term of the chemokine signaling pathway, while lymph node metastasis show perturbation in signaling cascades. Liver metastases change the expression of genes in a way that is reminiscent of processes that take place in the target organ. Survival analysis for the common hub genes revealed involvements in prostate cancer prognosis and suggested potential biomarkers.

Robust proportional overlapping analysis for feature selection in binary classification within functional genomic experiments

In this paper, a novel feature selection method called Robust Proportional Overlapping Score (RPOS), for microarray gene expression datasets has been proposed, by utilizing the robust measure of dispersion, i.e., Median Absolute Deviation (MAD). This method robustly identifies the most discriminative genes by considering the overlapping scores of the gene expression values for binary class problems. Genes with a high degree of overlap between classes are discarded and the ones that discriminate between the classes are selected. The results of the proposed method are compared with five state-of-the-art gene selection methods based on classification error, Brier score, and sensitivity, by considering eleven gene expression datasets. Classification of observations for different sets of selected genes by the proposed method is carried out by three different classifiers, i.e., random forest, k-nearest neighbors (k-NN), and support vector machine (SVM). Box-plots and stability scores of the results are also shown in this paper. The results reveal that in most of the cases the proposed method outperforms the other methods.