Immune subtype identification and multi-layer perceptron classifier construction for breast cancer

Comparative Evaluation of Machine Learning Models for Subtyping Triple-Negative Breast Cancer: A Deep Learning-Based Multi-Omics Data Integration Approach

Objective: Triple-negative breast cancer (TNBC) poses significant diagnostic challenges due to its aggressive nature. This research develops an innovative deep learning (DL) model based on the latest multi-omics data to enhance the accuracy of TNBC subtype and prognosis prediction. The study focuses on addressing the constraints of prior studies by showcasing a model with substantial advancements in data integration, statistical performance, and algorithmic optimization. Methods: Breast cancer-related molecular characteristic data, including mRNA, miRNA, gene mutations, DNA methylation, and magnetic resonance imaging (MRI) images, were retrieved from the TCGA and TCIA databases. This study not only compared single-omics with multi-omics machine learning models but also applied Bayesian optimization to innovatively optimize the neural network structure of a DL model for multi-omics data. Results: The DL model for multi-omics data significantly outperformed single-omics models in subtype prediction, achieving a 98.0% accuracy in cross-validation, 97.0% in the validation set, and 91.0% in an external test set. Additionally, the MRI radiomics model showed promising performance, especially with the training set; however, a decrease in performance during transfer testing underscored the advantages of the DL model for multi-omics data in data consistency and digital processing. Conclusion: Our multi-omics DL model presents notable innovations in statistical performance and transfer learning capability, bearing significant clinical relevance for TNBC classification and prognosis prediction. While the MRI radiomics model proved effective, it requires further optimization for cross-dataset application to enhance accuracy and consistency. Our findings offer new insights into improving TNBC classification and prognosis through multi-omics data and DL algorithms.

Prediction of cell migration potential on human breast cancer cells treated with Albizia lebbeck ethanolic extract using extreme machine learning

Cancer is one of the major causes of death in the modern world, and the incidence varies considerably based on race, ethnicity, and region. Novel cancer treatments, such as surgery and immunotherapy, are ineffective and expensive. In this situation, ion channels responsible for cell migration have appeared to be the most promising targets for cancer treatment. This research presents findings on the organic compounds present in Albizia lebbeck ethanolic extracts (ALEE), as well as their impact on the anti-migratory, anti-proliferative and cytotoxic potentials on MDA-MB 231 and MCF-7 human breast cancer cell lines. In addition, artificial intelligence (AI) based models, multilayer perceptron (MLP), extreme gradient boosting (XGB), and extreme learning machine (ELM) were performed to predict in vitro cancer cell migration on both cell lines, based on our experimental data. The organic compounds composition of the ALEE was studied using gas chromatography-mass spectrometry (GC-MS) analysis. Cytotoxicity, anti-proliferations, and anti-migratory activity of the extract using Tryphan Blue, MTT, and Wound Heal assay, respectively. Among the various concentrations (2.5-200 μg/mL) of the ALEE that were used in our study, 2.5-10 μg/mL revealed anti-migratory potential with increased concentrations, and they did not show any effect on the proliferation of the cells (P < 0.05; n ≥ 3). Furthermore, the three data-driven models, Multi-layer perceptron (MLP), Extreme gradient boosting (XGB), and Extreme learning machine (ELM), predict the potential migration ability of the extract on the treated cells based on our experimental data. Overall, the concentrations of the plant extract that do not affect the proliferation of the type cells used demonstrated promising effects in reducing cell migration. XGB outperformed the MLP and ELM models and increased their performance efficiency by up to 3% and 1% for MCF and 1% and 2% for MDA-MB231, respectively, in the testing phase.

Characterization of immune checkpoints expression and lymphocyte densities of iranian breast cancer patients; the co-expression status and clinicopathological associates

BACKGROUND

Breast malignancies are now the most common and deadliest type of neoplasms among women worldwide. Novel therapeutic approaches are needed to combat advanced stages of breast cancer. In this study, we aimed to investigate the expression and co-expression status of three immune checkpoints (PD-1, PD-L1, and LAG-3), as well as tumor-infiltrating lymphocytes (TIL) scores, and to further establish their potential correlations with clinicopathologic features.

METHODS

We performed a retrospective study on 361 pathologic samples of breast cancer. Immunohistochemistry was performed to assess the status of the immune checkpoint markers, and H&E staining was used to score TILs. The correlations of the immune checkpoint markers of tumor cells and tumor-associated immune cells and TIL scores with clinicopathological characteristics were analyzed.

RESULTS

Out of 361 assessed samples, LAG-3 was positive in 51%, while IC PD-L1 and TC PD-L1 were detectable in 36% and 8.9%, respectively. Moreover, both IC PD-L1 and LAG-3 stained positively in 24.4% of samples. IC PD-L1 expression was significantly higher in tumors with higher nuclear, mitotic, and overall grades and tubule formation. In addition, TC PD-L1 and LAG-3 exhibited a similar trend for higher overall grading. Tumors with positive estrogen- and progesterone-receptor (ER and PR) expression had significantly lower IC PD-L1 and TC PD-L1 staining, while LAG-3 positivity was more prevalent in HER2 positive samples. Tumors that were positive for these biomarkers had significantly higher Ki-67 scores. LAG-3 expression showed significant correlations with PD-1 and IC PD-L1 expression. Besides, the co-expression of LAG-3 and IC PD-L1 was significantly more encountered in luminal B and triple-negative subtypes, compared to the luminal A subtype. Regarding TILs, their scoring was significantly higher in ER and PR negative and HER2 positive samples. Intriguingly, samples with positive staining for LAG-3, IC PD-L1, and TC PD-L1 had significantly higher TIL scorings.

CONCLUSIONS

Immune checkpoints show differentially different levels of expression in certain molecular subtypes of breast cancer. Moreover, they reveal a meaningful correlation with each other, proliferation indices, and histologic grades. Finally, a sizable proportion of breast cancers co-express PD-L1 and LAG-3, which will make them appropriate targets for future combined ICIs.