Effect of training-sample size and classification difficulty on the accuracy of genomic predictors

GSDME-mediated pyroptosis promotes anti-tumor immunity of neoadjuvant chemotherapy in breast cancer

Paclitaxel and anthracycline-based chemotherapy is one of the standard treatment options for breast cancer. However, only about 6-30% of breast cancer patients achieved a pathological complete response (pCR), and the mechanism responsible for the difference is still unclear. In this study, random forest algorithm was used to screen feature genes, and artificial neural network (ANN) algorithm was used to construct an ANN model for predicting the efficacy of neoadjuvant chemotherapy for breast cancer. Furthermore, digital pathology, cytology, and molecular biology experiments were used to verify the relationship between the efficacy of neoadjuvant chemotherapy and immune ecology. It was found that paclitaxel and doxorubicin, an anthracycline, could induce typical pyroptosis and bubbling in breast cancer cells, accompanied by gasdermin E (GSDME) cleavage. Paclitaxel with LDH release and Annexin V/PI doubule positive cell populations, and accompanied by the increased release of damage-associated molecular patterns, HMGB1 and ATP. Cell coculture experiments also demonstrated enhanced phagocytosis of macrophages and increased the levels of IFN-γ and IL-2 secretion after paclitaxel treatment. Mechanistically, GSDME may mediate paclitaxel and doxorubicin-induced pyroptosis in breast cancer cells through the caspase-9/caspase-3 pathway, activate anti-tumor immunity, and promote the efficacy of paclitaxel and anthracycline-based neoadjuvant chemotherapy. This study has practical guiding significance for the precision treatment of breast cancer, and can also provide ideas for understanding molecular mechanisms related to the chemotherapy sensitivity.

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.

Head and neck cancer of unknown primary: unveiling primary tumor sites through machine learning on DNA methylation profiles

BACKGROUND

The unknown tissue of origin in head and neck cancer of unknown primary (hnCUP) leads to invasive diagnostic procedures and unspecific and potentially inefficient treatment options for patients. The most common histologic subtype, squamous cell carcinoma, can stem from various tumor primary sites, including the oral cavity, oropharynx, larynx, head and neck skin, lungs, and esophagus. DNA methylation profiles are highly tissue-specific and have been successfully used to classify tissue origin. We therefore developed a support vector machine (SVM) classifier trained with publicly available DNA methylation profiles of commonly cervically metastasizing squamous cell carcinomas (n = 1103) in order to identify the primary tissue of origin of our own cohort of squamous cell hnCUP patient's samples (n = 28). Methylation analysis was performed with Infinium MethylationEPIC v1.0 BeadChip by Illumina.

RESULTS

The SVM algorithm achieved the highest overall accuracy of tested classifiers, with 87%. Squamous cell hnCUP samples on DNA methylation level resembled squamous cell carcinomas commonly metastasizing into cervical lymph nodes. The most frequently predicted cancer localization was the oral cavity in 11 cases (39%), followed by the oropharynx and larynx (both 7, 25%), skin (2, 7%), and esophagus (1, 4%). These frequencies concord with the expected distribution of lymph node metastases in epidemiological studies.

CONCLUSIONS

On DNA methylation level, hnCUP is comparable to primary tumor tissue cancer types that commonly metastasize to cervical lymph nodes. Our SVM-based classifier can accurately predict these cancers' tissues of origin and could significantly reduce the invasiveness of hnCUP diagnostics and enable a more precise therapy after clinical validation.

Correlation of an immune-related 8-gene panel with pathologic response to neoadjuvant chemotherapy in patients with primary breast cancers

Neoadjuvant chemotherapy (NACT)-induced pathologic complete response (pCR) is associated with a favorable prognosis for breast cancer. Prior research links tumor-infiltrating lymphocytes with breast cancer chemotherapy response, suggesting the tumor-immune microenvironment's role. The aim of this study was to evaluate the immune-related genes that exhibit associations with the response to NACT. In this study, we analyzed a total of 37 patients (aged 27-67) who received NACT as the first-line treatment for primary breast cancer, followed by surgery. This group consisted of nine patients (24.3 %) with estrogen receptor (ER)-positive/HER2-negative status, ten patients (27.0 %) with ER-positive/HER2-positive status, five patients (13.5 %) with ER-negative/HER2-positive status, and thirteen patients (35.1 %) with triple-negative breast cancer (TNBC). Among these patients, twelve (32.4 %) achieved a pCR, with eight (66.6 %) having HER2-positive tumors, and the remaining four having TNBC. To identify immune-related genes linked with pCR in subjects with breast cancer prior to NACT, we collected fresh tissues for next-generation sequencing. Patients with pCR had higher expressions of eight genes, KLRK1, IGJ, CD69, CD40LG, MS4A1, CD1C, KLRB1, and CA4, compared to non-pCR patients. The 8-gene signature was associated with good prognosis and linked to better relapse-free survival in patients receiving chemotherapy. The expression of these genes was involved in better drug response, displaying a positive correlation with the infiltration of immune cells. In conclusion, we have identified eight immune-related genes that are associated with a favorable prognosis and positive responses to drugs. This 8-gene signature could potentially provide prognostic insights for breast cancer patients undergoing NACT.

Notch-based gene signature for predicting the response to neoadjuvant chemotherapy in triple-negative breast cancer

BACKGROUND

While the efficacy of neoadjuvant chemotherapy (NACT) in treating triple-negative breast cancer (TNBC) is generally accepted, not all patients derive benefit from this preoperative treatment. Presently, there are no validated biomarkers to predict the NACT response, and previous attempts to develop predictive classifiers based on gene expression data have not demonstrated clinical utility. However, predictive models incorporating biological constraints have shown increased robustness and improved performance compared to agnostic classifiers.

METHODS

We used the preoperative transcriptomic profiles from 298 patients with TNBC to train and test a rank-based classifier, k-top scoring pairs, to predict whether the patient will have pathological complete response (pCR) or residual disease (RD) following NACT. To reduce overfitting and enhance the signature's interpretability, we constrained the training process to genes involved in the Notch signaling pathway. Subsequently, we evaluated the signature performance on two independent cohorts with 75 and 71 patients. Finally, we assessed the prognostic value of the signature by examining its association with relapse-free survival (RFS) using Kaplan‒Meier (KM) survival estimates and a multivariate Cox proportional hazards model.

RESULTS

The final signature consists of five gene pairs, whose relative ordering can be predictive of the NACT response. The signature has a robust performance at predicting pCR in TNBC patients with an area under the ROC curve (AUC) of 0.76 and 0.85 in the first and second testing cohorts, respectively, outperforming other gene signatures developed for the same purpose. Additionally, the signature was significantly associated with RFS in an independent TNBC patient cohort even after adjusting for T stage, patient age at the time of diagnosis, type of breast surgery, and menopausal status.

CONCLUSION

We introduce a robust gene signature to predict pathological complete response (pCR) in patients with TNBC. This signature applies easily interpretable, rank-based decision rules to genes regulated by the Notch signaling pathway, a known determinant in breast cancer chemoresistance. The robust predictive and prognostic performance of the signature make it a strong candidate for clinical implementation, aiding in the stratification of TNBC patients undergoing NACT.

PTCH1 and CTNNB1 emerge as pivotal predictors of resistance to neoadjuvant chemotherapy in ER+/HER2- breast cancer

Introduction

Endeavors in the molecular characterization of breast cancer opened the doors to endocrine therapies in ER+/HER2- breast cancer, increasing response rates substantially. Despite that, taxane-based neoadjuvant chemotherapy is still a cornerstone for achieving breast-conserving surgery and complete tumor resection in locally advanced cancers with high recurrence risk. Nonetheless, the rate of chemoresistance is high, and deselecting patients who will not benefit from chemotherapy is a significant task to prevent futile toxicities. Several multigene assays are being used to guide decisions on chemotherapy. However, their development as prognostic assays but not predictive assays limits predictive strength, leading to discordant results. Moreover, high costs impediment their use in developing countries. For global health equity, robust predictors that can be cost-effectively incorporated into routine clinical management are essential.

Methods

In this study, we comprehensively analyzed 5 GEO datasets, 2 validation sets, and The Cancer Genome Atlas breast cancer data to identify predictors of resistance to taxane-based neoadjuvant therapy in ER+/HER2- breast cancer using efficient bioinformatics algorithms.

Results

Gene expression and gene set enrichment analysis of 5 GEO datasets revealed the upregulation of 63 genes and the enrichment of CTNNB1-related oncogenic signatures in non-responsive patients. We validated the upregulation and predictive strength of 18 genes associated with resistance in the validation cohort, all exhibiting higher predictive powers for residual disease and higher specificities for ER+/HER2- breast cancers compared to one of the benchmark multi-gene assays. Cox Proportional Hazards Regression in three different treatment arms (neoadjuvant chemotherapy, endocrine therapy, and no systemic treatment) in a second comprehensive validation cohort strengthened the significance of PTCH1 and CTNNB1 as key predictors, with hazard ratios over 1.5, and 1.6 respectively in the univariate and multivariate models.

Discussion

Our results strongly suggest that PTCH1 and CTNNB1 can be used as robust and cost-effective predictors in developing countries to guide decisions on chemotherapy in ER +/HER2- breast cancer patients with a high risk of recurrence. The dual function of PTCH1 as a multidrug efflux pump and a hedgehog receptor, and the active involvement of CTNNB1 in breast cancer strongly indicate that PTCH1 and CTNNB1 can be potential drug targets to overcome chemoresistance in ER +/HER2- breast cancer patients.

Using biological constraints to improve prediction in precision oncology

Many gene signatures have been developed by applying machine learning (ML) on omics profiles, however, their clinical utility is often hindered by limited interpretability and unstable performance. Here, we show the importance of embedding prior biological knowledge in the decision rules yielded by ML approaches to build robust classifiers. We tested this by applying different ML algorithms on gene expression data to predict three difficult cancer phenotypes: bladder cancer progression to muscle-invasive disease, response to neoadjuvant chemotherapy in triple-negative breast cancer, and prostate cancer metastatic progression. We developed two sets of classifiers: mechanistic, by restricting the training to features capturing specific biological mechanisms; and agnostic, in which the training did not use any a priori biological information. Mechanistic models had a similar or better testing performance than their agnostic counterparts, with enhanced interpretability. Our findings support the use of biological constraints to develop robust gene signatures with high translational potential.

Schlafen 12 Slows TNBC Tumor Growth, Induces Luminal Markers, and Predicts Favorable Survival

The Schlafen 12 (SLFN12) protein regulates triple-negative breast cancer (TNBC) growth, differentiation, and proliferation. SLFN12 mRNA expression strongly correlates with TNBC patient survival. We sought to explore SLFN12 overexpression effects on in vivo human TNBC tumor xenograft growth and performed RNA-seq on xenografts to investigate related SLFN12 pathways. Stable SLFN12 overexpression reduced tumorigenesis, increased tumor latency, and reduced tumor volume. RNA-seq showed that SLFN12 overexpressing xenografts had higher luminal markers levels, suggesting that TNBC cells switched from an undifferentiated basal phenotype to a more differentiated, less aggressive luminal phenotype. SLFN12-overexpressing xenografts increased less aggressive BC markers, HER2 receptors ERBB2 and EGFR expression, which are not detectable by immunostaining in TNBC. Two cancer progression pathways, the NAD signaling pathway and the superpathway of cholesterol biosynthesis, were downregulated with SLFN12 overexpression. RNA-seq identified gene signatures associated with SLFN12 overexpression. Higher gene signature levels indicated good survival when tested on four independent BC datasets. These signatures behaved differently in African Americans than in Caucasian Americans, indicating a possible biological difference between these races that could contribute to the worse survival observed in African Americans with BC. These results suggest an increased SLFN12 expression modulates TNBC aggressiveness through a gene signature that could offer new treatment targets.

SCUBE2 as a Marker of Resistance to Taxane-based Neoadjuvant Chemotherapy and a Potential Therapeutic Target in Breast Cancer

Objective

Taxane-based neoadjuvant chemotherapy is the most common neoadjuvant approach in breast cancer, especially in human epidermal growth factor receptor 2 (HER2)-positive and triple-negative subtypes. However, chemoresistance is a problem in many patients, and success rates are low in estrogen receptor (ER)-positive breast cancer. The aim of this study was to identify predictive markers for resistance to taxane-based therapy, which may have a potential as therapeutic targets in breast cancer.

Materials and Methods

Three comprehensive breast cancer Gene Expression Omnibus datasets were analyzed to identify differentially expressed genes (DEGs) in breast cancer patients resistant to taxane-based neoadjuvant chemotherapy. Functional annotation clustering and enrichment analysis were performed on the DEGs list. A protein-protein interaction network was established with the upregulated genes. The predictive value and the differential expression of the central genes were validated in the extensive ROC Plotter database.

Results

Seventeen upregulated genes were found which were associated with resistance to taxane-based neoadjuvant therapy and high connectivity in the network analysis. ESR1, CCND1, and SCUBE2 emerged as the top three key genes associated with resistance. SCUBE2 displayed a high predictive power comparable to ESR1, and better than CCND1, the two commonly accepted markers. The predictive ability of SCUBE2 was higher in ER-positive and HER2-positive breast cancers.

Conclusion

These results suggest that SCUBE2 may be used as a predictive marker to guide decisions on neoadjuvant therapy. Emerging evidence about the role of SCUBE2 as a coreceptor involved in tumor progression and angiogenesis also suggests SCUBE2 as a potential therapeutic target. These points should be investigated in further studies.

CD4+ T cells drive an inflammatory, TNF-α/IFN-rich tumor microenvironment responsive to chemotherapy

While chemotherapy remains the first-line treatment for many cancers, it is still unclear what distinguishes responders from non-responders. Here, we characterize the chemotherapy-responsive tumor microenvironment in mice, using RNA sequencing on tumors before and after cyclophosphamide, and compare the gene expression profiles of responders with progressors. Responsive tumors have an inflammatory and highly immune infiltrated pre-treatment tumor microenvironment characterized by the enrichment of pathways associated with CD4⁺ T cells, interferons (IFNs), and tumor necrosis factor alpha (TNF-α). The same gene expression profile is associated with response to cyclophosphamide-based chemotherapy in patients with breast cancer. Finally, we demonstrate that tumors can be sensitized to cyclophosphamide and 5-FU chemotherapy by pre-treatment with recombinant TNF-α, IFNγ, and poly(I:C). Thus, a CD4⁺ T cell-inflamed pre-treatment tumor microenvironment is necessary for response to chemotherapy, and this state can be therapeutically attained by targeted immunotherapy.

High RAD51 gene expression is associated with aggressive biology and with poor survival in breast cancer

PURPOSE

Although the DNA repair mechanism is important in preventing carcinogenesis, its activation in established cancer cells may support their proliferation and aggravate cancer progression. RAD51 cooperates with BRCA2 and is essential in the homologous recombination of DNA repair. To this end, we hypothesized that RAD51 gene expression is associated with cancer cell proliferation and poor prognosis of breast cancer (BC) patients.

METHODS

A total of 8515 primary BC patients with transcriptome and clinical data from 17 independent cohorts were analyzed. The median value was used to divide each cohort into high and low RAD51 expression groups.

RESULTS

High RAD51 expression enriched the DNA repair gene set and was correlated with DNA repair-related genes. Nottingham histological grade, Ki67 expression and cell proliferation-related gene sets (E2F Targets, G2M Checkpoint and Myc Targets) were all significantly associated with the high RAD51 BC group. RAD51 expression was positively correlated with Homologous Recombination Deficiency, as well as both mutational burden and neoantigens that accompanied a higher infiltration of immune cells. Primary BC with lymph node metastases was associated with high expression of RAD51 in two cohorts. There was no strong correlation between RAD51 expression and drug sensitivity in cell lines, and RAD51 expression was lower after the neoadjuvant chemotherapy compared to before the treatment. High RAD51 BC was associated with poor prognosis consistently in three independent cohorts.

CONCLUSION

RAD51 gene expression is associated with aggressive cancer biology, cancer cell proliferation, and poor survival in breast cancer.

A framework to predict the applicability of Oncotype DX, MammaPrint, and E2F4 gene signatures for improving breast cancer prognostic prediction

To improve cancer precision medicine, prognostic and predictive biomarkers are critically needed to aid physicians in deciding treatment strategies in a personalized fashion. Due to the heterogeneous nature of cancer, most biomarkers are expected to be valid only in a subset of patients. Furthermore, there is no current approach to determine the applicability of biomarkers. In this study, we propose a framework to improve the clinical application of biomarkers. As part of this framework, we develop a clinical outcome prediction model (CPM) and a predictability prediction model (PPM) for each biomarker and use these models to calculate a prognostic score (P-score) and a confidence score (C-score) for each patient. Each biomarker’s P-score indicates its association with patient clinical outcomes, while each C-score reflects the biomarker applicability of the biomarker’s CPM to a patient and therefore the confidence of the clinical prediction. We assessed the effectiveness of this framework by applying it to three biomarkers, Oncotype DX, MammaPrint, and an E2F4 signature, which have been used for predicting patient response, pathologic complete response versus residual disease to neoadjuvant chemotherapy (a classification problem), and recurrence-free survival (a Cox regression problem) in breast cancer, respectively. In both applications, our analyses indicated patients with higher C scores were more likely to be correctly predicted by the biomarkers, indicating the effectiveness of our framework. This framework provides a useful approach to develop and apply biomarkers in the context of cancer precision medicine.

An Immune-Associated Genomic Signature Effectively Predicts Pathologic Complete Response to Neoadjuvant Paclitaxel and Anthracycline-Based Chemotherapy in Breast Cancer

Breast cancer is now the leading cause of cancer morbidity and mortality among women worldwide. Paclitaxel and anthracycline-based neoadjuvant chemotherapy is widely used for the treatment of breast cancer, but its sensitivity remains difficult to predict for clinical use. In our study, a LASSO logistic regression method was applied to develop a genomic classifier for predicting pathologic complete response (pCR) to neoadjuvant chemotherapy in breast cancer. The predictive accuracy of the signature classifier was further evaluated using four other independent test sets. Also, functional enrichment analysis of genes in the signature was performed, and the correlations between the prediction score of the signature classifier and immune characteristics were explored. We found a 25-gene signature classifier through the modeling, which showed a strong ability to predict pCR to neoadjuvant chemotherapy in breast cancer. For T/FAC-based training and test sets, and a T/AC-based test set, the AUC of the signature classifier is 1.0, 0.9071, 0.9683, 0.9151, and 0.7350, respectively, indicating that it has good predictive ability for both T/FAC and T/AC schemes. The multivariate model showed that 25-gene signature was far superior to other clinical parameters as independent predictor. Functional enrichment analysis indicated that genes in the signature are mainly enriched in immune-related biological processes. The prediction score of the classifier was significantly positively correlated with the immune score. There were also significant differences in immune cell types between pCR and residual disease (RD) samples. Conclusively, we developed a 25-gene signature classifier that can effectively predict pCR to paclitaxel and anthracycline-based neoadjuvant chemotherapy in breast cancer. Our study also suggests that the immune ecosystem is actively involved in modulating clinical response to neoadjuvant chemotherapy and is beneficial to patient outcomes.

A systematic review of datasets that can help elucidate relationships among gene expression, race, and immunohistochemistry-defined subtypes in breast cancer

Scholarly requirements have led to a massive increase of transcriptomic data in the public domain, with millions of samples available for secondary research. We identified gene-expression datasets representing 10,214 breast-cancer patients in public databases. We focused on datasets that included patient metadata on race and/or immunohistochemistry (IHC) profiling of the ER, PR, and HER-2 proteins. This review provides a summary of these datasets and describes findings from 32 research articles associated with the datasets. These studies have helped to elucidate relationships between IHC, race, and/or treatment options, as well as relationships between IHC status and the breast-cancer intrinsic subtypes. We have also identified broad themes across the analysis methodologies used in these studies, including breast cancer subtyping, deriving predictive biomarkers, identifying differentially expressed genes, and optimizing data processing. Finally, we discuss limitations of prior work and recommend future directions for reusing these datasets in secondary analyses.

Expression of X-Linked Inhibitor of Apoptosis Protein (XIAP) in Breast Cancer Is Associated with Shorter Survival and Resistance to Chemotherapy

XIAP, the most potent inhibitor of cell death pathways, is linked to chemotherapy resistance and tumor aggressiveness. Currently, multiple XIAP-targeting agents are in clinical trials. However, the characterization of XIAP expression in relation to clinicopathological variables in large clinical series of breast cancer is lacking. We retrospectively analyzed non-metastatic, non-inflammatory, primary, invasive breast cancer samples for XIAP mRNA (n = 2341) and protein (n = 367) expression. XIAP expression was analyzed as a continuous value and correlated with clinicopathological variables. XIAP mRNA expression was heterogeneous across samples and significantly associated with younger patients' age (≤50 years), pathological ductal type, lower tumor grade, node-positive status, HR+/HER2- status, and PAM50 luminal B subtype. Higher XIAP expression was associated with shorter DFS in uni- and multivariate analyses in 909 informative patients. Very similar correlations were observed at the protein level. This prognostic impact was significant in the HR+/HER2- but not in the TN subtype. Finally, XIAP mRNA expression was associated with lower pCR rate to anthracycline-based neoadjuvant chemotherapy in both uni- and multivariate analyses in 1203 informative patients. Higher XIAP expression in invasive breast cancer is independently associated with poorer prognosis and resistance to chemotherapy, suggesting the potential therapeutic benefit of targeting XIAP.

Family Rank: A graphical domain knowledge informed feature ranking algorithm

MOTIVATION

When designing prediction models built with many features and relatively small sample sizes, feature selection methods often overfit training data, leading to selection of irrelevant features. One way to potentially mitigate overfitting is to incorporate domain knowledge during feature selection. Here, a feature ranking algorithm called 'Family Rank' is presented in which features are ranked based on a combination of graphical domain knowledge and feature scores computed from empirical data.

RESULTS

A simulated data set is used to demonstrate a scenario in which family rank outperforms other state-of-the-art graph based ranking algorithms, decreasing the sample size needed to detect true predictors by 2 to 3-fold. An example from oncology is then used to explore a real-world application of family rank.

AVAILABILITY

An implementation of Family Rank is freely available at https://cran.r-project.org/package=FamilyRank.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

High expression of CPNE5 and CPNE9 predicts positive prognosis in multiple myeloma

BACKGROUND

CPNEs are significant biomarkers which can affect the progression and prognosis of various tumor diseases. However, the prognosis role of CPNEs in multiple myeloma (MM) is still unclear.

OBJECTIVES

To investigate the prognosis role of CPNEs in MM.

METHODS

Seven hundred and thirty-five samples from two independent data sets were involved to analyze the clinical and molecular characteristics, and prognosis role of the expression of CPNE1-9 in MM.

RESULTS

MM patients with higher expressions of CPNE5 and CPNE9 had longer event-free survival (EFS) and overall survival (OS) compared with CPNE5low and CPNE9low expression groups (EFS: P= 0.0054, 0.0065; OS: P= 0.015, 0.016, respectively). Multivariate regression analysis showed that CPNE5 was an independent favorable predictor for EFS and OS (EFS: P= 0.005; OS: P= 0.006), and CPNE9 was an independent positive indicator for EFS (P= 0.002). Moreover, the survival probability and the cumulative event of EFS and OS in CPNE5highCPNE9high group were significantly longer than other groups.

CONCLUSIONS

High expressions of CPNE5 and CPNE9 might be used as positive indicators for MM, and their combination was a better predictor for the survival of MM patients.

High expression levels of centromere protein A plus upregulation of the phosphatidylinositol 3-kinase/Akt/mammalian target of rapamycin signaling pathway affect chemotherapy response and prognosis in patients with breast cancer

Centromere proteins (CENPs) are involved in mitosis, and CENP gene expression levels are associated with chemotherapy responses in patients with breast cancer. The present study aimed to examine the roles and underlying mechanisms of the effects of CENP genes on chemotherapy responses and breast cancer prognosis. Using data obtained from the Gene Expression Omnibus (GEO) database, correlation and Cox multivariate regression analyses were used to determine the CENP genes associated with chemotherapy responses and survival in patients with breast cancer. Weighted gene co-expression network and correlation analyses were used to determine the gene modules co-expressed with the identified genes and the differential expression of gene modules associated with the pathological complete response (PCR) and residual disease (RD) subgroups. CENPA, CENPE, CENPF, CENPI, CENPJ and CENPN were associated with a high nuclear grade and low estrogen and progesterone receptor expression levels. In addition, CENPA, CENPB, CENPC and CENPO were independent factors affecting the distant relapse-free survival (DRFS) rates in patients with breast cancer. Patients with high expression levels of CENPA or CENPO exhibited poor prognoses, whereas those with high expression levels of CENPB or CENPC presented with favorable prognoses. For validation between databases, the Cancer Genome Atlas (TCGA) database analysis also revealed that CENPA, CENPB and CENPO exerted similar effects on overall survival. However, according to the multivariate analyses, only CENPA was an independent risk factor associated with DRFS in GEO database. In addition, in the RD subgroup, patients with higher CENPA expression levels had a worse prognosis compared with those with lower CENPA expression levels. Among patients with high expression levels of CENPA, the PI3K/Akt/mTOR pathway was more likely to be activated in the RD compared with the PCR subgroup. The same trend was observed in TCGA data. These results suggested that high CENPA expression levels plus upregulation of the PI3K/Akt/mTOR signaling pathway may affect DRFS in patients with breast cancer.

Deep generative neural network for accurate drug response imputation

Drug response differs substantially in cancer patients due to inter- and intra-tumor heterogeneity. Particularly, transcriptome context, especially tumor microenvironment, has been shown playing a significant role in shaping the actual treatment outcome. In this study, we develop a deep variational autoencoder (VAE) model to compress thousands of genes into latent vectors in a low-dimensional space. We then demonstrate that these encoded vectors could accurately impute drug response, outperform standard signature-gene based approaches, and appropriately control the overfitting problem. We apply rigorous quality assessment and validation, including assessing the impact of cell line lineage, cross-validation, cross-panel evaluation, and application in independent clinical data sets, to warrant the accuracy of the imputed drug response in both cell lines and cancer samples. Specifically, the expression-regulated component (EReX) of the observed drug response achieves high correlation across panels. Using the well-trained models, we impute drug response of The Cancer Genome Atlas data and investigate the features and signatures associated with the imputed drug response, including cell line origins, somatic mutations and tumor mutation burdens, tumor microenvironment, and confounding factors. In summary, our deep learning method and the results are useful for the study of signatures and markers of drug response.

High Expression of NRF2 Is Associated with Increased Tumor-Infiltrating Lymphocytes and Cancer Immunity in ER-Positive/HER2-Negative Breast Cancer

Nuclear factor erythroid 2-related factor 2 (NRF2) is a key modifier in breast cancer. It is unclear whether NRF2 suppresses or promotes breast cancer progression. We studied the clinical relevance of NRF2 expression by conducting in silico analyses in 5443 breast cancer patients from several large patient cohorts (METABRIC, GSE96058, GSE25066, GSE20194, and GSE75688). NRF2 expression was significantly associated with better survival, low Nottingham pathological grade, and ER-positive/HER2-negative and triple negative breast cancer (TNBC). High NRF2 ER-positive/HER2-negative breast cancer enriched inflammation- and immune-related gene sets by GSEA. NRF2 expression was elevated in immune, stromal, and cancer cells. High NRF2 tumors were associated with high infiltration of immune cells (CD8+, CD4+, and dendritic cells (DC)) and stromal cells (adipocyte, fibroblasts, and keratinocytes), and with low fraction of Th1 cells. NRF2 expression significantly correlated with area under the curve (AUC) of several drug response in multiple ER-positive breast cancer cell lines, however, there was no significant association between NRF2 and pathologic complete response (pCR) rate after neoadjuvant chemotherapy in human samples. Finally, high NRF2 breast cancer was associated with high expression of immune checkpoint molecules. In conclusion, NRF2 expression was associated with enhanced tumor-infiltrating lymphocytes in ER-positive/HER2-negative breast cancer.

Evaluation of Machine Learning-based Patient Outcome Prediction Using Patient-specific Difficulty and Discrimination Indices

Given the extensive use of machine learning in patient outcome prediction, and the understanding that the challenging nature of predictions in this field may considerably modify the performance of predictive models, research in this area requires some forms of context-sensitive performance metrics. The area under the receiver operating characteristic curve (AUC), precision, recall, specificity, and F1 are widely used measures of performance for patient outcome prediction. These metrics have several merits: they are easy to interpret and do not need any subjective input from the user. However, they weight all samples equally and do not adequately reflect the ability of predictive models in classifying difficult samples. In this paper, we propose the Difficulty Weight Adjustment (DWA) algorithm, a simple method that incorporates the difficulty level of samples when evaluating predictive models. Using a large dataset of 139,367 unique ICU admissions within the eICU Collaborative Research Database (eICU-CRD), we show that the classification difficulty and the discrimination ability of samples are critical aspects that need to be considered when comparing machine learning models that predict patient outcomes.

M1 Macrophage and M1/M2 ratio defined by transcriptomic signatures resemble only part of their conventional clinical characteristics in breast cancer

Tumor associated macrophages (TAMs) play a critical role in biology of various cancers, including breast cancer. In the current study, we defined "M1" macrophage and "M1"/"M2" ratio by transcriptomic signatures using xCell. We investigated the association between high level of "M1" macrophage or "M1"/"M2" ratio and the tumor immune microenvironment by analyzing the transcriptome of publicly available cohorts, TCGA and METABRIC. We found that "M1" high tumors were not associated with prolonged survival compared with "M1" low tumors, or with the response to neoadjuvant chemotherapy. "M1" high tumors were associated with clinically aggressive features and "M1" high tumors enriched the cell proliferation and cell cycle related gene sets in GSEA. At the same time, "M1" high tumors were associated with high immune activity and favorable tumor immune microenvironment, as well as high expression of immune check point molecules. Strikingly, all these results were mirrored in "M1"/"M2" ratio high tumors. In conclusion, transcriptomically defined "M1" or "M1"/"M2" high tumors were associated with aggressive cancer biology and favorable tumor immune microenvironment but not with survival benefit, which resembled only part of their conventional clinical characteristics.

Functional Transcription Factor Target Networks Illuminate Control of Epithelial Remodelling

Cell identity is governed by gene expression, regulated by transcription factor (TF) binding at cis-regulatory modules. Decoding the relationship between TF binding patterns and gene regulation is nontrivial, remaining a fundamental limitation in understanding cell decision-making. We developed the NetNC software to predict functionally active regulation of TF targets; demonstrated on nine datasets for the TFs Snail, Twist, and modENCODE Highly Occupied Target (HOT) regions. Snail and Twist are canonical drivers of epithelial to mesenchymal transition (EMT), a cell programme important in development, tumour progression and fibrosis. Predicted "neutral" (non-functional) TF binding always accounted for the majority (50% to 95%) of candidate target genes from statistically significant peaks and HOT regions had higher functional binding than most of the Snail and Twist datasets examined. Our results illuminated conserved gene networks that control epithelial plasticity in development and disease. We identified new gene functions and network modules including crosstalk with notch signalling and regulation of chromatin organisation, evidencing networks that reshape Waddington's epigenetic landscape during epithelial remodelling. Expression of orthologous functional TF targets discriminated breast cancer molecular subtypes and predicted novel tumour biology, with implications for precision medicine. Predicted invasion roles were validated using a tractable cell model, supporting our approach.

Cancer gene expression profiles associated with clinical outcomes to chemotherapy treatments

Background

Machine learning (ML) methods still have limited applicability in personalized oncology due to low numbers of available clinically annotated molecular profiles. This doesn’t allow sufficient training of ML classifiers that could be used for improving molecular diagnostics.

Methods

We reviewed published datasets of high throughput gene expression profiles corresponding to cancer patients with known responses on chemotherapy treatments. We browsed Gene Expression Omnibus (GEO), The Cancer Genome Atlas (TCGA) and Tumor Alterations Relevant for GEnomics-driven Therapy (TARGET) repositories.

Results

We identified data collections suitable to build ML models for predicting responses on certain chemotherapeutic schemes. We identified 26 datasets, ranging from 41 till 508 cases per dataset. All the datasets identified were checked for ML applicability and robustness with leave-one-out cross validation. Twenty-three datasets were found suitable for using ML that had balanced numbers of treatment responder and non-responder cases.

Conclusions

We collected a database of gene expression profiles associated with clinical responses on chemotherapy for 2786 individual cancer cases. Among them seven datasets included RNA sequencing data (for 645 cases) and the others – microarray expression profiles. The cases represented breast cancer, lung cancer, low-grade glioma, endothelial carcinoma, multiple myeloma, adult leukemia, pediatric leukemia and kidney tumors. Chemotherapeutics included taxanes, bortezomib, vincristine, trastuzumab, letrozole, tipifarnib, temozolomide, busulfan and cyclophosphamide.

Revisiting the Concept of Stress in the Prognosis of Solid Tumors: A Role for Stress Granules Proteins?

Simple Summary

Stress Granules (SGs) were discovered in 1999 and while the first decade of research has focused on some fundamental questions, the field is now investigating their role in human pathogenesis. Since then, evidences of a link between SGs and cancerology are accumulating in vitro and in vivo. In this work we summarized the role of SGs proteins in cancer development and their prognostic values. We find that level of expression of protein involved in SGs formation (and not mRNA level) could serve a prognostic marker in cancer. With this review we strongly suggest that SGs (proteins) could be targets of choice to block cancer development and counteract resistance to improve patients care.

Abstract

Cancer treatments are constantly evolving with new approaches to improve patient outcomes. Despite progresses, too many patients remain refractory to treatment due to either the development of resistance to therapeutic drugs and/or metastasis occurrence. Growing evidence suggests that these two barriers are due to transient survival mechanisms that are similar to those observed during stress response. We review the literature and current available open databases to study the potential role of stress response and, most particularly, the involvement of Stress Granules (proteins) in cancer. We propose that Stress Granule proteins may have prognostic value for patients.

Cyclin D1 targets hexokinase 2 to control aerobic glycolysis in myeloma cells

Cancer cells are characterized by the Warburg effect, a shift from mitochondrial respiration to oxidative glycolysis. We report here the crucial role of cyclin D1 in promoting this effect in a cyclin-dependent kinase (CDK)4/6-independent manner in multiple myeloma (MM) cells. We show that the cyclin D1 oncoprotein targets hexokinase 2 (HK2), a major glycolysis regulator, through two original molecular mechanisms in the cytoplasmic and nuclear compartments. In the cytoplasm, cyclin D1 binds HK2 at the outer mitochondrial membrane, and in the nucleus, it binds hypoxia-inducible factor-1α (HIF1α), which regulates HK2 gene transcription. We also show that high levels of HK2 expression are correlated with shorter event-free survival (EFS) and overall survival (OS) in MM patients. HK2 may therefore be considered as a possible target for antimyeloma therapy.

PDXGEM: patient-derived tumor xenograft-based gene expression model for predicting clinical response to anticancer therapy in cancer patients

Background

Cancer is a highly heterogeneous disease with varying responses to anti-cancer drugs. Although several attempts have been made to predict the anti-cancer therapeutic responses, there remains a great need to develop highly accurate prediction models of response to the anti-cancer drugs for clinical applications toward a personalized medicine. Patient derived xenografts (PDXs) are preclinical cancer models in which the tissue or cells from a patient’s tumor are implanted into an immunodeficient or humanized mouse. In the present study, we develop a bioinformatics analysis pipeline to build a predictive gene expression model (GEM) for cancer patients’ drug responses based on gene expression and drug activity data from PDX models.

Results

Drug sensitivity biomarkers were identified by performing an association analysis between gene expression levels and post-treatment tumor volume changes in PDX models. We built a drug response prediction model (called PDXGEM) in a random-forest algorithm by using a subset of the drug sensitvity biomarkers with concordant co-expression patterns between the PDXs and pretreatment cancer patient tumors. We applied the PDXGEM to several cytotoxic chemotherapies as well as targeted therapy agents that are used to treat breast cancer, pancreatic cancer, colorectal cancer, or non-small cell lung cancer. Significantly accurate predictions of PDXGEM for pathological response or survival outcomes were observed in extensive independent validations on multiple cancer patient datasets obtained from retrospective observational studies and prospective clinical trials.

Conclusion

Our results demonstrated the strong potential of using molecular profiles and drug activity data of PDX tumors in developing a clinically translatable predictive cancer biomarkers for cancer patients. The PDXGEM web application is publicly available at http://pdxgem.moffitt.org.

Testing Suicide Risk Prediction Algorithms Using Phone Measurements With Patients in Acute Mental Health Settings: Feasibility Study

BACKGROUND

Digital phenotyping and machine learning are currently being used to augment or even replace traditional analytic procedures in many domains, including health care. Given the heavy reliance on smartphones and mobile devices around the world, this readily available source of data is an important and highly underutilized source that has the potential to improve mental health risk prediction and prevention and advance mental health globally.

OBJECTIVE

This study aimed to apply machine learning in an acute mental health setting for suicide risk prediction. This study uses a nascent approach, adding to existing knowledge by using data collected through a smartphone in place of clinical data, which have typically been collected from health care records.

METHODS

We created a smartphone app called Strength Within Me, which was linked to Fitbit, Apple Health kit, and Facebook, to collect salient clinical information such as sleep behavior and mood, step frequency and count, and engagement patterns with the phone from a cohort of inpatients with acute mental health (n=66). In addition, clinical research interviews were used to assess mood, sleep, and suicide risk. Multiple machine learning algorithms were tested to determine the best fit.

RESULTS

K-nearest neighbors (KNN; k=2) with uniform weighting and the Euclidean distance metric emerged as the most promising algorithm, with 68% mean accuracy (averaged over 10,000 simulations of splitting the training and testing data via 10-fold cross-validation) and an average area under the curve of 0.65. We applied a combined 5×2 F test to test the model performance of KNN against the baseline classifier that guesses training majority, random forest, support vector machine and logistic regression, and achieved F statistics of 10.7 (P=.009) and 17.6 (P=.003) for training majority and random forest, respectively, rejecting the null of performance being the same. Therefore, we have taken the first steps in prototyping a system that could continuously and accurately assess the risk of suicide via mobile devices.

CONCLUSIONS

Predicting for suicidality is an underaddressed area of research to which this paper makes a useful contribution. This is part of the first generation of studies to suggest that it is feasible to utilize smartphone-generated user input and passive sensor data to generate a risk algorithm among inpatients at suicide risk. The model reveals fair concordance between phone-derived and research-generated clinical data, and with iterative development, it has the potential for accurate discriminant risk prediction. However, although full automation and independence of clinical judgment or input would be a worthy development for those individuals who are less likely to access specialist mental health services, and for providing a timely response in a crisis situation, the ethical and legal implications of such advances in the field of psychiatry need to be acknowledged.

Predicting treatment effects in unipolar depression: A meta-review

There is increasing interest in clinical prediction models in psychiatry, which focus on developing multivariate algorithms to guide personalized diagnostic or management decisions. The main target of these models is the prediction of treatment response to different antidepressant therapies. This is because the ability to predict response based on patients' personal data may allow clinicians to make improved treatment decisions, and to provide more efficacious or more tolerable medications to the right patient. We searched the literature for systematic reviews about treatment prediction in the context of existing treatment modalities for adult unipolar depression, until July 2019. Treatment effect is defined broadly to include efficacy, safety, tolerability and acceptability outcomes. We first focused on the identification of individual predictor variables that might predict treatment response, and second, we considered multivariate clinical prediction models. Our meta-review included a total of 10 systematic reviews; seven (from 2014 to 2018) focusing on individual predictor variables and three focusing on clinical prediction models. These identified a number of sociodemographic, phenomenological, clinical, neuroimaging, remote monitoring, genetic and serum marker variables as possible predictor variables for treatment response, alongside statistical and machine-learning approaches to clinical prediction model development. Effect sizes for individual predictor variables were generally small and clinical prediction models had generally not been validated in external populations. There is a need for rigorous model validation in large external data-sets to prove the clinical utility of models. We also discuss potential future avenues in the field of personalized psychiatry, particularly the combination of multiple sources of data and the emerging field of artificial intelligence and digital mental health to identify new individual predictor variables.

A Qualitative Transcriptional Signature for Predicting Extreme Resistance of ER-Negative Breast Cancer to Paclitaxel, Doxorubicin, and Cyclophosphamide Neoadjuvant Chemotherapy

For estrogen receptor (ER)-negative breast cancer patients, paclitaxel (P), doxorubicin (A) and cyclophosphamide (C) neoadjuvant chemotherapy (NAC) is the standard therapeutic regimen. Pathologic complete response (pCR) and residual disease (RD) are common surrogate measures of chemosensitivity. After NAC, most patients still have RD; of these, some partially respond to NAC, whereas others show extreme resistance and cannot benefit from NAC but only suffer complications resulting from drug toxicity. Here we developed a qualitative transcriptional signature, based on the within-sample relative expression ordering (REO) of gene pairs, to identify extremely resistant samples to PAC NAC. Using gene expression data for ER-negative breast cancer patients including 113 pCR samples and 137 RD samples from four datasets, we selected 61 gene pairs with reversal REO patterns between the two groups as the resistance signature, denoted as NR61. Samples with more than 37 signature gene pairs that had the same REO patterns within the extremely resistant group were defined as having extreme resistance; otherwise, they were considered responders. In the GSE25055 and GSE25065 dataset, the NR61 signature could correctly identify 44 (97.8%) of the 45 pCR samples and 22 (95.7%) of the 23 pCR samples as responder samples, respectively; it also identified 13 (16.9%) of 77 RD samples and 8 (21.1%) of 38 RD samples as extremely resistant samples, respectively. Survival analysis showed that the distant relapse-free survival (DRFS) time of the 14 extremely resistant cases was significantly shorter than that of the 108 responders (P < 0.01; HR = 3.84; 95% CI = 1.91–7.70) in GSE25055. Similar results were obtained in GSE25065. Moreover, in the integrated data of the two datasets with 94 responders and 21 extremely resistant samples identified from RD patients, the former had significantly longer DRFS than the latter (P < 0.01; HR = 2.22; 95% CI = 1.26–3.90). In summary, our signature could effectively identify patients who completely respond to PAC NAC, as well as cases of extreme resistance, which can assist decision-making on the clinical therapy for these patients.

The therapeutic response of ER+/HER2- breast cancers differs according to the molecular Basal or Luminal subtype

The genomics-based molecular classifications aim at identifying more homogeneous classes than immunohistochemistry, associated with a more uniform clinical outcome. We conducted an in silico analysis on a meta-dataset including gene expression data from 5342 clinically defined ER+/HER2- breast cancers (BC) and DNA copy number/mutational and proteomic data. We show that the Basal (16%) versus Luminal (74%) subtypes as defined using the 80-gene signature differ in terms of response/vulnerability to systemic therapies of BC. The Basal subtype is associated with better chemosensitivity, lesser benefit from adjuvant hormone therapy, and likely better sensitivity to PARP inhibitors, platinum salts and immune therapy, and other targeted therapies under development such as FGFR inhibitors. The Luminal subtype displays potential better sensitivity to CDK4/6 inhibitors and vulnerability to targeted therapies such as PIK3CA, AR and Bcl-2 inhibitors. Expression profiles are very different, showing an intermediate position of the ER+/HER2- Basal subtype between the ER+/HER2- Luminal and ER- Basal subtypes, and let suggest a different cell-of-origin. Our data suggest that the ER+/HER2- Basal and Luminal subtypes should not be assimilated and treated as a homogeneous group.

Multiple Myeloma DREAM Challenge reveals epigenetic regulator PHF19 as marker of aggressive disease

While the past decade has seen meaningful improvements in clinical outcomes for multiple myeloma patients, a subset of patients does not benefit from current therapeutics for unclear reasons. Many gene expression-based models of risk have been developed, but each model uses a different combination of genes and often involves assaying many genes making them difficult to implement. We organized the Multiple Myeloma DREAM Challenge, a crowdsourced effort to develop models of rapid progression in newly diagnosed myeloma patients and to benchmark these against previously published models. This effort lead to more robust predictors and found that incorporating specific demographic and clinical features improved gene expression-based models of high risk. Furthermore, post-challenge analysis identified a novel expression-based risk marker, PHF19, which has recently been found to have an important biological role in multiple myeloma. Lastly, we show that a simple four feature predictor composed of age, ISS, and expression of PHF19 and MMSET performs similarly to more complex models with many more gene expression features included.

Flexible Data Trimming Improves Performance of Global Machine Learning Methods in Omics-Based Personalized Oncology

(1) Background: Machine learning (ML) methods are rarely used for an omics-based prescription of cancer drugs, due to shortage of case histories with clinical outcome supplemented by high-throughput molecular data. This causes overtraining and high vulnerability of most ML methods. Recently, we proposed a hybrid global-local approach to ML termed floating window projective separator (FloWPS) that avoids extrapolation in the feature space. Its core property is data trimming, i.e., sample-specific removal of irrelevant features. (2) Methods: Here, we applied FloWPS to seven popular ML methods, including linear SVM, k nearest neighbors (kNN), random forest (RF), Tikhonov (ridge) regression (RR), binomial naïve Bayes (BNB), adaptive boosting (ADA) and multi-layer perceptron (MLP). (3) Results: We performed computational experiments for 21 high throughput gene expression datasets (41–235 samples per dataset) totally representing 1778 cancer patients with known responses on chemotherapy treatments. FloWPS essentially improved the classifier quality for all global ML methods (SVM, RF, BNB, ADA, MLP), where the area under the receiver-operator curve (ROC AUC) for the treatment response classifiers increased from 0.61–0.88 range to 0.70–0.94. We tested FloWPS-empowered methods for overtraining by interrogating the importance of different features for different ML methods in the same model datasets. (4) Conclusions: We showed that FloWPS increases the correlation of feature importance between the different ML methods, which indicates its robustness to overtraining. For all the datasets tested, the best performance of FloWPS data trimming was observed for the BNB method, which can be valuable for further building of ML classifiers in personalized oncology.

Prognostic role of minichromosome maintenance family in multiple myeloma

Multiple myeloma (MM) is a plasma cell malignancy. The minichromosome maintenance (MCM) family involve in DNA replication and is vital in limiting replication in cell cycle. The prognostic role of MCMs in MM is still unclear. We took four independent GEO datasets to analyze the relationship between the expression of MCMs and myeloma progression and survival. The expression of MCMs showed an upward trend with myeloma progression in 205 patients. High MCM2/3/4/6/8 expression was associated with both poor EFS and OS (all p < 0.050). Multivariate analysis demonstrated that high MCM2 expression, B2M, and LDH were independent risk factors. Moreover, the combination of MCM2/B2M and MCM2/LDH was a better tool in prognostication. In conclusion, high MCM2 expression is an independent adverse prognostic factor and could be used as a prognostic biomarker in MM.

High expression of chaperonin-containing TCP1 subunit 3 may induce dismal prognosis in multiple myeloma

The prognosis role of CCT3 in MM and the possible pathways it involved were studied in our research. By analyzing ten independent datasets (including 48 healthy donors, 2220 MM, 73 MGUS, and 6 PCL), CCT3 was found to express higher in MM than healthy donors, and the expression level was gradually increased from MGUS, SMM, MM to PCL (all P < 0.01). By analyzing three independent datasets (GSE24080, GSE2658, and GSE4204), we found that CCT3 was a significant indicator of poor prognosis (all P < 0.01). KEGG and GSEA analysis showed that CCT3 expression was associated with JAK-STAT3 pathway, Hippo signaling pathway, and WNT signaling pathway. In addition, different expressed genes analysis revealed MYC, which was one of the downstream genes regulated by JAK-STAT3 pathway, was upregulated in MM. This confirms that JAK-STAT3 signaling pathway may promote the progress of disease which was regulated by CCT3 expression. Our study revealed that CCT3 may play a supporting role at the diagnosis of myeloid, and high expression of CCT3 suggested poor prognosis in MM. CCT3 expression may promote the progression of MM mainly by regulating MYC through JAK-STAT3 signaling pathway.

ThalPred: a web-based prediction tool for discriminating thalassemia trait and iron deficiency anemia

Background

The hypochromic microcytic anemia (HMA) commonly found in Thailand are iron deficiency anemia (IDA) and thalassemia trait (TT). Accurate discrimination between IDA and TT is an important issue and better methods are urgently needed. Although considerable RBC formulas and indices with various optimal cut-off values have been developed, distinguishing between IDA and TT is still a challenging problem due to the diversity of various anemic populations. To address this problem, it is desirable to develop an improved and automated prediction model for discriminating IDA from TT.

Methods

We retrospectively collected laboratory data of HMA found in Thai adults. Five machine learnings, including k-nearest neighbor (k-NN), decision tree, random forest (RF), artificial neural network (ANN) and support vector machine (SVM), were applied to construct a discriminant model. Performance was assessed and compared with thirteen existing discriminant formulas and indices.

Results

The data of 186 patients (146 patients with TT and 40 with IDA) were enrolled. The interpretable rules derived from the RF model were proposed to demonstrate the combination of RBC indices for discriminating IDA from TT. A web-based tool ‘ThalPred’ was implemented using an SVM model based on seven RBC parameters. ThalPred achieved prediction results with an external accuracy, MCC and AUC of 95.59, 0.87 and 0.98, respectively.

Conclusion

ThalPred and an interpretable rule were provided for distinguishing IDA from TT. For the convenience of health care team experimental scientists, a web-based tool has been established at http://codes.bio/thalpred/ by which users can easily get their desired screening test result without the need to go through the underlying mathematical and computational details.

A Predictor of Pathological Complete Response to Neoadjuvant Chemotherapy Stratifies Triple Negative Breast Cancer Patients with High Risk of Recurrence

We developed a test to predict which patients will achieve pathological complete response (pCR) to neoadjuvant chemotherapy (NAC) and which will have residual disease (RD). Gene expression data from pretreatment biopsies of patients with all breast cancer subtypes were combined into a 519-patient cohort containing 177 TNBC patients. Two RNA classifiers of 16 genes each were sequentially applied to the total cohort, classifying patients into 3 distinct classes. The test performance was further validated in an independent 304-patient cohort. The test accurately identified 70.5% (79/112) of pCR and 83.5% (340/407) of RD patients in the total population, and 75.0% (45/60) of pCR and 75.2% (88/117) of RD patients in the TNBC subset. For the independent cohort, the test identified 91.5% RD patients in the total population and 86.2% RD patients in the TNBC subset. However, the identification of pCR in both total and TNBC population are as low as 21.1% and 30%, respectively. The TNBC RD patients were subdivided by our classifiers, with one class showing significantly higher levels of Ki67 expression and having significantly poorer survival rates than the other classes. This stratification of patients may allow predicted residual disease classes to be assigned an alternative therapy.

Targetable mechanisms driving immunoevasion of persistent senescent cells link chemotherapy-resistant cancer to aging

Cellular senescence is a tumor suppressive mechanism that can paradoxically contribute to aging pathologies. Despite evidence of immune clearance in mouse models, it is not known how senescent cells (SnCs) persist and accumulate with age or in tumors in individuals. Here, we identify cooperative mechanisms that orchestrate the immunoevasion and persistence of normal and cancer human SnCs through extracellular targeting of natural killer receptor signaling. Damaged SnCs avoid immune recognition through MMPs-dependent shedding of NKG2D-ligands reinforced via paracrine suppression of NKG2D receptor-mediated immunosurveillance. These coordinated immunoediting processes are evident in residual, drug-resistant tumors from cohorts of >700 prostate and breast cancer patients treated with senescence-inducing genotoxic chemotherapies. Unlike in mice, these reversible senescence-subversion mechanisms are independent of p53/p16 and exacerbated in oncogenic RAS-induced senescence. Critically, the p16INK4A tumor suppressor can disengage the senescence growth arrest from the damage-associated immune senescence program, which is manifest in benign nevi lesions where indolent SnCs accumulate over time and preserve a non-pro-inflammatory tissue microenvironment maintaining NKG2D-mediated immunosurveillance. Our study shows how subpopulations of SnCs elude immunosurveillance, and reveals secretome-targeted therapeutic strategies to selectively eliminate -and restore the clearance of- the detrimental SnCs that actively persist after chemotherapy and accumulate at sites of aging pathologies.

The effect of uncertainty in patient classification on diagnostic performance estimations

Background

The performance of a new diagnostic test is typically evaluated against a comparator which is assumed to correspond closely to some true state of interest. Judgments about the new test’s performance are based on the differences between the outputs of the test and comparator. It is commonly assumed that a small amount of uncertainty in the comparator’s classifications will negligibly affect the measured performance of a diagnostic test.

Methods

Simulated datasets were generated to represent typical diagnostic scenarios. Comparator noise was introduced in the form of random misclassifications, and the effect on the apparent performance of the diagnostic test was determined. An actual dataset from a clinical trial on a new diagnostic test for sepsis was also analyzed.

Results

We demonstrate that as little as 5% misclassification of patients by the comparator can be enough to statistically invalidate performance estimates such as sensitivity, specificity and area under the receiver operating characteristic curve, if this uncertainty is not measured and taken into account. This distortion effect is found to increase non-linearly with comparator uncertainty, under some common diagnostic scenarios. For clinical populations exhibiting high degrees of classification uncertainty, failure to measure and account for this effect will introduce significant risks of drawing false conclusions. The effect of classification uncertainty is magnified further for high performing tests that would otherwise reach near-perfection in diagnostic evaluation trials. A requirement of very high diagnostic performance for clinical adoption, such as a 99% sensitivity, can be rendered nearly unachievable even for a perfect test, if the comparator diagnosis contains even small amounts of uncertainty. This paper and an accompanying online simulation tool demonstrate the effect of classification uncertainty on the apparent performance of tests across a range of typical diagnostic scenarios. Both simulated and real datasets are used to show the degradation of apparent test performance as comparator uncertainty increases.

Conclusions

Overall, a 5% or greater misclassification rate by the comparator can lead to significant underestimation of true test performance. An online simulation tool allows researchers to explore this effect using their own trial parameters (https://imperfect-gold-standard.shinyapps.io/classification-noise/) and the source code is freely available (https://github.com/ksny/Imperfect-Gold-Standard).

Parsimonious Gene Correlation Network Analysis (PGCNA): a tool to define modular gene co-expression for refined molecular stratification in cancer

Cancers converge onto shared patterns that arise from constraints placed by the biology of the originating cell lineage and microenvironment on programs driven by oncogenic events. Here we define consistent expression modules reflecting this structure in colon and breast cancer by exploiting expression data resources and a new computationally efficient approach that we validate against other comparable methods. This approach, Parsimonious Gene Correlation Network Analysis (PGCNA), allows comparison of network structures between these cancer types identifying shared modules of gene co-expression reflecting: cancer hallmarks, functional and structural gene batteries, copy number variation and biology of originating lineage. These networks along with the mapping of outcome data at gene and module level provide an interactive resource that generates context for relationships between genes within and between such modules. Assigning module expression values (MEVs) provides a tool to summarize network level gene expression in individual cases illustrating potential utility in classification and allowing analysis of linkage between module expression and mutational state. Exploiting TCGA data thus defines both recurrent patterns of association between module expression and mutation at data-set level, and exemplifies the polarization of mutation patterns with the leading edge of module expression at individual case level. We illustrate the scalable nature of the approach within immune response related modules, which in the context of breast cancer demonstrates the selective association of immune subsets, in particular mast cells, with the underlying mutational pattern. Together our analyses provide evidence for a generalizable framework to enhance molecular stratification in cancer.

Gene Expression-Based Predictive Markers for Paclitaxel Treatment in ER+ and ER- Breast Cancer

One of the objectives of precision oncology is to identify patient’s responsiveness to a given treatment and prevent potential overtreatments through molecular profiling. Predictive gene expression biomarkers are a promising and practical means to this purpose. The overall response rate of paclitaxel drugs in breast cancer has been reported to be in the range of 20–60% and is in the even lower range for ER-positive patients. Predicting responsiveness of breast cancer patients, either ER-positive or ER-negative, to paclitaxel treatment could prevent individuals with poor response to the therapy from undergoing excess exposure to the agent. In this study, we identified six sets of gene signatures whose gene expression profiles could robustly predict nonresponding patients with precisions more than 94% and recalls more than 93% on various discovery datasets (n = 469 for the largest set) and independent validation datasets (n = 278), using the previously developed Multiple Survival Screening algorithm, a random-sampling-based methodology. The gene signatures reported were stable regardless of half of the discovery datasets being swapped, demonstrating their robustness. We also reported a set of optimizations that enabled the algorithm to train on small-scale computational resources. The gene signatures and optimized methodology described in this study could be used for identifying unresponsiveness in patients of ER-positive or ER-negative breast cancers.

A data-driven interactome of synergistic genes improves network-based cancer outcome prediction

Robustly predicting outcome for cancer patients from gene expression is an important challenge on the road to better personalized treatment. Network-based outcome predictors (NOPs), which considers the cellular wiring diagram in the classification, hold much promise to improve performance, stability and interpretability of identified marker genes. Problematically, reports on the efficacy of NOPs are conflicting and for instance suggest that utilizing random networks performs on par to networks that describe biologically relevant interactions. In this paper we turn the prediction problem around: instead of using a given biological network in the NOP, we aim to identify the network of genes that truly improves outcome prediction. To this end, we propose SyNet, a gene network constructed ab initio from synergistic gene pairs derived from survival-labelled gene expression data. To obtain SyNet, we evaluate synergy for all 69 million pairwise combinations of genes resulting in a network that is specific to the dataset and phenotype under study and can be used to in a NOP model. We evaluated SyNet and 11 other networks on a compendium dataset of >4000 survival-labelled breast cancer samples. For this purpose, we used cross-study validation which more closely emulates real world application of these outcome predictors. We find that SyNet is the only network that truly improves performance, stability and interpretability in several existing NOPs. We show that SyNet overlaps significantly with existing gene networks, and can be confidently predicted (~85% AUC) from graph-topological descriptions of these networks, in particular the breast tissue-specific network. Due to its data-driven nature, SyNet is not biased to well-studied genes and thus facilitates post-hoc interpretation. We find that SyNet is highly enriched for known breast cancer genes and genes related to e.g. histological grade and tamoxifen resistance, suggestive of a role in determining breast cancer outcome.

EarlyR signature predicts response to neoadjuvant chemotherapy in breast cancer

BACKGROUND

EarlyR gene signature uses ESPL1, SPAG5, MKI67, PLK1 and PGR to classify ER+ breast cancer (ER+ BC) into EarlyR-Low, EarlyR-Int, and EarlyR-High risk strata and is prognostic in patients treated with adjuvant chemotherapy. The ability of EarlyR to predict pathological complete response (pCR) and long-term survival following neoadjuvant chemotherapy (NACT) is evaluated herein.

MATERIALS

The ability of EarlyR gene signature to predict pCR was assessed in publicly available Affymetrix microarray datasets (Cohort A; n = 659; 74 pCR events) derived from NACT-treated ER+ BC patients. Distant relapse-free survival (DRFS) results were analyzed in patients treated with NACT and adjuvant hormone therapy (AHT) (n = 281) and compared with patients treated with AHT alone (n = 455) (Cohort B; n = 736; 142 events).

RESULTS

In cohort A, EarlyR was a significant predictor of pCR (p = 5.8 × 10-11) (EarlyR-Low, n = 400, pCR = 40, 5%; EarlyR-Int, n = 69, pCR = 7, 15% and EarlyR-High, n = 190, pCR = 47, 24%). In EarlyR-Low of Cohort B, the 5-year DRFS was not significantly (p = 0.55) different between NACT + AHT [0.81 (95%CI 0.73-0.90)] and AHT-only [0.85 (95%CI 0.81-0.90)]. In contrast, in EarlyR-High, the 5-year DRFS was higher (p = 0.019) in NACT + AHT [0.81 (95%CI 0.70-0.93)] as compared to AHT-only [0.60 (95%CI 0.51-0.71)].

CONCLUSIONS

High EarlyR is strongly associated with pCR in patients treated with neoadjuvant chemotherapy. EarlyR also predicts poor DRFS outcomes for patients in EarlyR-High not receiving NACT, and improved survival in NACT-treated EarlyR-High patients. EarlyR is not only a prognostic assay but also a predictive assay that identifies patients, who are also likely to respond to chemotherapy.

pAKT pathway activation is associated with PIK3CA mutations and good prognosis in luminal breast cancer in contrast to p-mTOR pathway activation

Numerous studies have focused on the PI3K/AKT/mTOR pathway in estrogen receptor positive (ER) breast cancer (BC), as a linear signal transduction pathway and reported its association with worse clinical outcomes. We developed gene signatures that reflect the level of expression of phosphorylated-Serine473-AKT (pAKT) and phosphorylated-Serine2448-mTOR (p-mTOR) separately, capturing their corresponding level of pathway activation. Our analysis revealed that the pAKT pathway activation was associated with luminal A BC while the p-mTOR pathway activation was more associated with luminal B BC (Kruskal-Wallis test p < 10^-10). pAKT pathway activation was significantly associated with better outcomes (multivariable HR, 0.79; 95%CI, 0.74-0.85; p = 2.5 × 10^-10) and PIK3CA mutations (p = 0.0001) whereas p-mTOR pathway activation showed worse outcomes (multivariable HR,1.1; 95%CI, 1.1-1.2; p = 9.9 × 10^-4) and associated with p53 mutations (p = 0.04). in conclusion, our data show that pAKT and p-mTOR pathway activation have differing impact on prognosis and suggest that they are not linearly connected in luminal breast cancers.

Genomic Landscape and Endocrine-Resistant Subgroup in Estrogen Receptor-Positive, Progesterone Receptor-Negative, and HER2-Negative Breast Cancer

Estrogen receptor-positive, progesterone receptor-negative, and human epidermal growth factor receptor 2 (HER2)-negative (ER+PR-HER2-) breast cancer comprise a special type of breast cancer that constitutes ~10% of all breast cancer patients. ER+PR-HER2- tumor benefits less from endocrine therapy, while its genomic features remain elusive. In this study, we systematically assessed the multiomic landscape and endocrine responsiveness of ER+PR-HER2- breast cancer.

Methods: This study incorporated five cohorts. The first and second cohorts were from the Surveillance, Epidemiology, and End Results database (n=130,856) and Molecular Taxonomy of Breast Cancer International Consortium (n=1,055) for analyzing survival outcomes and endocrine responsiveness. The third cohort was from The Cancer Genome Atlas (n=630) for multiomic analysis and endocrine-resistant subgroup exploration. The fourth cohort, from the MD Anderson database (n=92), was employed to assist gene selection. The fifth cohort was a prospective observational cohort from Fudan University Shanghai Cancer Center (n=245) that was utilized to validate the gene-defined subgroup by immunohistochemistry (IHC).

Results: Clinically, ER+PR-HER2- tumors showed lower endocrine responsiveness than did ER+PR+HER2- tumors. Genomically, copy number loss or promoter methylation of PR genes occurred in 75% of ER+PR-HER2- tumors, collectively explaining PR loss. ER+PR-HER2- tumors had higher TP53 (30.3% vs. 17.0%) and lower PIK3CA mutation rates (25.8% vs. 42.7%) and exhibited more ZNF703 (21.5% vs. 13.6%) and RPS6KB1 (18.5% vs. 7.8%) amplification events than ER+PR+HER2- tumors. Among ER+PR-HER2- tumors, nearly 20% were of the PAM50-defined non-luminal-like subgroup and manifested lower endocrine sensitivity scores and enriched biosynthesis, metabolism and DNA replication pathways. We further identified the non-luminal-like subgroup using three IHC markers, GATA3, CK5, and EGFR. These IHC-defined non-luminal-like (GATA3-negative, CK5-positive and/or EGFR-positive) tumors received limited benefit from adjuvant endocrine therapy.

Conclusion: ER+PR-HER2- breast cancer consists of clinically and genomically distinct groups that may require different treatment strategies. The non-luminal-like subgroup was associated with reduced benefit from endocrine therapy.

The Expression of Actin-Related Protein 2/3 Complex Subunit 5 (ARPC5) Expression in Multiple Myeloma and its Prognostic Significance

Background

The aim of this study was to analyze the prognostic value of ARPC5 in patients with multiple myeloma (MM).

Material/Methods

MM gene expression studies GSE6477, GSE31162, GSE24080, and GSE19784 were obtained and analyzed. The expression of ARPC5 was assessed in normal plasma cells, baseline MM cells, and relapsed MM cells. Univariate and multivariable analyses were used to determine the relationship between ARPC5 expression and clinical characteristics and survivals of MM patients. Quantitative PCR was used to detect the expression ARPC5 in bone marrow mononuclear cells of MM patients and normal controls. GSEA was conducted to identify associated mechanisms.

Results

ARPC5 expression was significantly increased in baseline MM cells compared to normal plasma cells (P=0.0414). Meanwhile, ARPC5 was significantly increased in relapsed MM cells compared to baseline MM cells (P<0.0001). ARPC5 expression was significantly associated with β2-microglobin (P=0.047), serum lactate dehydrogenase (P=0.007), and rates of aspirate plasma cells (P=0.007). Meanwhile, patients in the ARPC5 high expression group were associated with poor overall survival (P=0.0027) and event-free survival (P=0.0102) compared to those in the ARPC5 low expression group. Multivariable analysis indicated that ARPC5 was an independent prognostic factor for MM patients. Quantitative PCR demonstrated that ARPC5 was significantly increased in MM patients. GSEA results indicated that ARPC5 might affect cellular growth of myeloma cells through mammalian target of rapamycin (mTOR)C1 signaling pathway.

Conclusions

ARPC5 could be treated as an independent biomarker for patients with MM.