Computational prognostic indicators for breast cancer

Clinical decision support tool for breast cancer recurrence prediction using SHAP value in cooperative game theory

Background

Recurrence remains the primary cause of death in patients with breast cancer. Although machine learning can efficiently predict the prognosis of breast cancer patients, the black-box nature of the model may result in a lack of evidence for clinicians when making critical decisions.

Methods

In this study, our main objective was twofold: (1) to develop a clinical decision support tool for predicting the prognosis of breast cancer and (2) to identify and explore the key factors that influence breast cancer recurrence. To achieve this, we employed an explainable ensemble learning method called Shapley additive explanation (SHAP), which leverages cooperative game theory. Using real-world data from 1629 breast cancer patients, we analyzed and uncovered the key factors associated with breast cancer recurrence. Subsequently, we used these identified factors to create a recurrence prediction model and establish a decision mechanism for the tool. The proposed method not only provides accurate recurrence predictions but also offers transparent explanations for these predictions.

Results

By utilizing four key factors, namely, tumor size, clinical stage III, number of lymph node metastases, and age, our decision support tool for predicting breast cancer recurrence achieved significant improvements. The extra-tree model exhibited an increased area under the receiver operating characteristic curve (AUC) of 0.97, while the Random Forest model demonstrated an improved AUC of 0.96. We also offer a decision mechanism for a recurrence prediction model based on the identified key factors. This transparent and interpretable decision-making process facilitated by our explainable ensemble learning model enhances trust and promotes its applicability in clinical settings.

Conclusions

The proposed explainable ensemble learning method shows promising results in predicting breast cancer recurrence, outperforming existing methods with high accuracy and transparency. This advancement has the potential to significantly improve clinical decision-making and patient outcomes in breast cancer treatment.

Theranostic Interpolation of Genomic Instability in Breast Cancer

Breast cancer is a diverse disease caused by mutations in multiple genes accompanying epigenetic aberrations of hazardous genes and protein pathways, which distress tumor-suppressor genes and the expression of oncogenes. Alteration in any of the several physiological mechanisms such as cell cycle checkpoints, DNA repair machinery, mitotic checkpoints, and telomere maintenance results in genomic instability. Theranostic has the potential to foretell and estimate therapy response, contributing a valuable opportunity to modify the ongoing treatments and has developed new treatment strategies in a personalized manner. “Omics” technologies play a key role while studying genomic instability in breast cancer, and broadly include various aspects of proteomics, genomics, metabolomics, and tumor grading. Certain computational techniques have been designed to facilitate the early diagnosis of cancer and predict disease-specific therapies, which can produce many effective results. Several diverse tools are used to investigate genomic instability and underlying mechanisms. The current review aimed to explore the genomic landscape, tumor heterogeneity, and possible mechanisms of genomic instability involved in initiating breast cancer. We also discuss the implications of computational biology regarding mutational and pathway analyses, identification of prognostic markers, and the development of strategies for precision medicine. We also review different technologies required for the investigation of genomic instability in breast cancer cells, including recent therapeutic and preventive advances in breast cancer.

Deep Learning-Based Prediction Model for Breast Cancer Recurrence Using Adjuvant Breast Cancer Cohort in Tertiary Cancer Center Registry

Several prognosis prediction models have been developed for breast cancer (BC) patients with curative surgery, but there is still an unmet need to precisely determine BC prognosis for individual BC patients in real time. This is a retrospectively collected data analysis from adjuvant BC registry at Samsung Medical Center between January 2000 and December 2016. The initial data set contained 325 clinical data elements: baseline characteristics with demographics, clinical and pathologic information, and follow-up clinical information including laboratory and imaging data during surveillance. Weibull Time To Event Recurrent Neural Network (WTTE-RNN) by Martinsson was implemented for machine learning. We searched for the optimal window size as time-stamped inputs. To develop the prediction model, data from 13,117 patients were split into training (60%), validation (20%), and test (20%) sets. The median follow-up duration was 4.7 years and the median number of visits was 8.4. We identified 32 features related to BC recurrence and considered them in further analyses. Performance at a point of statistics was calculated using Harrell's C-index and area under the curve (AUC) at each 2-, 5-, and 7-year points. After 200 training epochs with a batch size of 100, the C-index reached 0.92 for the training data set and 0.89 for the validation and test data sets. The AUC values were 0.90 at 2-year point, 0.91 at 5-year point, and 0.91 at 7-year point. The deep learning-based final model outperformed three other machine learning-based models. In terms of pathologic characteristics, the median absolute error (MAE) and weighted mean absolute error (wMAE) showed great results of as little as 3.5%. This BC prognosis model to determine the probability of BC recurrence in real time was developed using information from the time of BC diagnosis and the follow-up period in RNN machine learning model.

Neuroblastoma pathogenesis: deregulation of embryonic neural crest development

Neuroblastoma (NB) is an aggressive pediatric cancer that originates from neural crest tissues of the sympathetic nervous system. NB is highly heterogeneous both from a clinical and a molecular perspective. Clinically, this cancer represents a wide range of phenotypes ranging from spontaneous regression of 4S disease to unremitting treatment-refractory progression and death of high-risk metastatic disease. At a cellular level, the heterogeneous behavior of NB likely arises from an arrest and deregulation of normal neural crest development. In the present review, we summarize our current knowledge of neural crest development as it relates to pathways promoting 'stemness' and how deregulation may contribute to the development of tumor-initiating CSCs. There is an emerging consensus that such tumor subpopulations contribute to the evolution of drug resistance, metastasis and relapse in other equally aggressive malignancies. As relapsed, refractory disease remains the primary cause of death for neuroblastoma, the identification and targeting of CSCs or other primary drivers of tumor progression remains a critical, clinically significant goal for neuroblastoma. We will critically review recent and past evidence in the literature supporting the concept of CSCs as drivers of neuroblastoma pathogenesis.

Incorporating genomic, transcriptomic and clinical data: a prognostic and stem cell-like MYC and PRC imbalance in high-risk neuroblastoma

BACKGROUND

Previous studies suggested that cancer cells possess traits reminiscent of the biological mechanisms ascribed to normal embryonic stem cells (ESCs) regulated by MYC and Polycomb repressive complex 2 (PRC2). Several poorly differentiated adult tumors showed preferentially high expression levels in targets of MYC, coincident with low expression levels in targets of PRC2. This paper will reveal this ESC-like cancer signature in high-risk neuroblastoma (HR-NB), the most common extracranial solid tumor in children.

METHODS

We systematically assembled genomic variants, gene expression changes, priori knowledge of gene functions, and clinical outcomes to identify prognostic multigene signatures. First, we assigned a new, individualized prognostic index using the relative expressions between the poor- and good-outcome signature genes. We then characterized HR-NB aggressiveness beyond these prognostic multigene signatures through the imbalanced effects of MYC and PRC2 signaling. We further analyzed Retinoic acid (RA)-induced HR-NB cells to model tumor cell differentiation. Finally, we performed in vitro validation on ZFHX3, a cell differentiation marker silenced by PRC2, and compared cell morphology changes before and after blocking PRC2 in HR-NB cells.

RESULTS

A significant concurrence existed between exons with verified variants and genes showing MYCN-dependent expression in HR-NB. From these biomarker candidates, we identified two novel prognostic gene-set pairs with multi-scale oncogenic defects. Intriguingly, MYC targets over-represented an unfavorable component of the identified prognostic signatures while PRC2 targets over-represented a favorable component. The cell cycle arrest and neuronal differentiation marker ZFHX3 was identified as one of PRC2-silenced tumor suppressor candidates. Blocking PRC2 reduced tumor cell growth and increased the mRNA expression levels of ZFHX3 in an early treatment stage. This hypothesis-driven systems bioinformatics work offered novel insights into the PRC2-mediated tumor cell growth and differentiation in neuroblastoma, which may exert oncogenic effects together with MYC regulation.

CONCLUSION

Our results propose a prognostic effect of imbalanced MYC and PRC2 moderations in pediatric HR-NB for the first time. This study demonstrates an incorporation of genomic landscapes and transcriptomic profiles into the hypothesis-driven precision prognosis and biomarker discovery. The application of this approach to neuroblastoma, as well as other cancer more broadly, could contribute to reduced relapse and mortality rates in the long term.

A c-Myc-regulated stem cell-like signature in high-risk neuroblastoma: A systematic discovery (Target neuroblastoma ESC-like signature)

c-Myc dysregulation is hypothesized to account for the ‘stemness’ – self-renewal and pluripotency – shared between embryonic stem cells (ESCs) and adult aggressive tumours. High-risk neuroblastoma (HR-NB) is the most frequent, aggressive, extracranial solid tumour in childhood. Using HR-NB as a platform, we performed a network analysis of transcriptome data and presented a c-Myc subnetwork enriched for genes previously reported as ESC-like cancer signatures. A subsequent drug-gene interaction analysis identified a pharmacogenomic agent that preferentially interacted with this HR-NB-specific, ESC-like signature. This agent, Roniciclib (BAY 1000394), inhibited neuroblastoma cell growth and induced apoptosis in vitro. It also repressed the expression of the oncogene c-Myc and the neural ESC marker CDK2 in vitro, which was accompanied by altered expression of the c-Myc-targeted cell cycle regulators CCND1, CDKN1A and CDKN2D in a time-dependent manner. Further investigation into this HR-NB-specific ESC-like signature in 295 and 243 independent patients revealed and validated the general prognostic index of CDK2 and CDKN3 compared with CDKN2D and CDKN1B. These findings highlight the very potent therapeutic benefits of Roniciclib in HR-NB through the targeting of c-Myc-regulated, ESC-like tumorigenesis. This work provides a hypothesis-driven systems computational model that facilitates the translation of genomic and transcriptomic signatures to molecular mechanisms underlying high-risk tumours.

Systematic computation with functional gene-sets among leukemic and hematopoietic stem cells reveals a favorable prognostic signature for acute myeloid leukemia

Background

Genes that regulate stem cell function are suspected to exert adverse effects on prognosis in malignancy. However, diverse cancer stem cell signatures are difficult for physicians to interpret and apply clinically. To connect the transcriptome and stem cell biology, with potential clinical applications, we propose a novel computational “gene-to-function, snapshot-to-dynamics, and biology-to-clinic” framework to uncover core functional gene-sets signatures. This framework incorporates three function-centric gene-set analysis strategies: a meta-analysis of both microarray and RNA-seq data, novel dynamic network mechanism (DNM) identification, and a personalized prognostic indicator analysis. This work uses complex disease acute myeloid leukemia (AML) as a research platform.

Results

We introduced an adjustable “soft threshold” to a functional gene-set algorithm and found that two different analysis methods identified distinct gene-set signatures from the same samples. We identified a 30-gene cluster that characterizes leukemic stem cell (LSC)-depleted cells and a 25-gene cluster that characterizes LSC-enriched cells in parallel; both mark favorable-prognosis in AML. Genes within each signature significantly share common biological processes and/or molecular functions (empirical p = 6e-5 and 0.03 respectively). The 25-gene signature reflects the abnormal development of stem cells in AML, such as AURKA over-expression. We subsequently determined that the clinical relevance of both signatures is independent of known clinical risk classifications in 214 patients with cytogenetically normal AML. We successfully validated the prognosis of both signatures in two independent cohorts of 91 and 242 patients respectively (log-rank p < 0.0015 and 0.05; empirical p < 0.015 and 0.08).

Conclusion

The proposed algorithms and computational framework will harness systems biology research because they efficiently translate gene-sets (rather than single genes) into biological discoveries about AML and other complex diseases.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-015-0510-7) contains supplementary material, which is available to authorized users.

Can multiple SNP testing in BRCA2 and BRCA1 female carriers be used to improve risk prediction models in conjunction with clinical assessment?

BACKGROUND

Several single nucleotide polymorphisms (SNPs) at different loci have been associated with breast cancer susceptibility, accounting for around 10% of the familial component. Recent studies have found direct associations between specific SNPs and breast cancer in BRCA1/2 mutation carriers. Our aim was to determine whether validated susceptibility SNP scores improve the predictive ability of risk models in comparison/conjunction to other clinical/demographic information.

METHODS

Female BRCA1/2 carriers were identified from the Manchester genetic database, and included in the study regardless of breast cancer status or age. DNA was extracted from blood samples provided by these women and used for gene and SNP profiling. Estimates of survival were examined with Kaplan-Meier curves. Multivariable Cox proportional hazards models were fit in the separate BRCA datasets and in menopausal stages screening different combinations of clinical/demographic/genetic variables. Nonlinear random survival forests were also fit to identify relevant interactions. Models were compared using Harrell's concordance index (1 - c-index).

RESULTS

548 female BRCA1 mutation carriers and 523 BRCA2 carriers were identified from the database. Median Kaplan-Meier estimate of survival was 46.0 years (44.9-48.1) for BRCA1 carriers and 48.9 (47.3-50.4) for BRCA2. By fitting Cox models and random survival forests, including both a genetic SNP score and clinical/demographic variables, average 1 - c-index values were 0.221 (st.dev. 0.019) for BRCA1 carriers and 0.215 (st.dev. 0.018) for BRCA2 carriers.

CONCLUSIONS

Random survival forests did not yield higher performance compared to Cox proportional hazards. We found improvement in prediction performance when coupling the genetic SNP score with clinical/demographic markers, which warrants further investigation.