Epidemiological profiling of population-specific risk factors and validation of novel genetic variants and deep learning-driven mitosis detection in breast cancer patients
No Thumbnail Available
Date
2024-05
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Avinashilingam
Abstract
Breast cancer is a major global health issue and one of the most common
malignancy affecting women globally. Addressing the population-specific risk factors,
rare genetic alterations, and disease progression is challenging, and it is crucial for
developing targeted therapy and reducing the impact of breast cancer treatment on
affected individuals. The comprehensive four-phase study aimed to analyse breast cancer
patient's risk factors, genetic alterations, and disease progression. The investigation
commenced with an epidemiological survey to identify risk factors prevalent in breast
cancer patients residing in Tamil Nadu, followed by whole exome sequencing to explore
if there are any rare genetic variants in the patient's primary breast tumor, which may be
population-specific. Sanger sequencing was performed to validate the reliability of the
identified novel variants. With the advent of artificial intelligence (AI) in all fields,
especially the healthcare sector, we tried to use AI to detect breast cancer progression
accurately using histopathology images. The convolutional neural network (CNN) was
applied to histopathology images to assess the mitotic index for evaluating proliferation
in breast cancer patients. The results of an epidemiological study of 517 patients
highlighted the significance of demographic characteristics, lifestyle choices, and
treatment modalities for breast cancer. Our investigations ascertained that women aged
41-50 were the most affected, having a mean age of 47.40 (SD 11). The increased
incidence rate was observed in women with lower levels of education and who were
unemployed. Among reproductive factors, menopausal status was found to be a
significant determinant, with premenopausal women facing a higher risk (p<0.001).
Lifestyle factors showed that less physical exercise/activity (p<0.01), water intake
(p<0.05), and sleep duration (p<0.01) displayed significant associations with breast
cancer incidence. We also identified significant associations between comorbidities
(p<0.01), tumor location (p<0.001), treatment modalities (p<0.001), and breast cancer.
These findings add valuable insights into the complex web of factors influencing breast
cancer incidence and progression within the specific patient population. The second
phase involved whole exome sequencing (WES) of 6 primary breast cancer patient
samples (5-Female, 1-Male), which covers all the subtypes of breast cancer. We
identified 857 rare genetic variants and found that the common associated disease among
five patients is cardiomyopathy. The results indicate that complete cardiac evaluation is
necessary before breast cancer therapy. Functional enrichment analysis exhibited that the
rare variant candidate genes are mostly involved in biological processes that include
signal transduction, G-protein coupled receptor, gene expression regulation, synapse
assembly, and the cellular process of the genes occurred in the plasma membrane,
cytosol, and cytoplasm, and the molecular function of the genes mainly involved in
protein-protein binding. Dysregulation of these functional processes leads to oncogenic
properties of cancer cells. The third phase validated the novel genetic variants MYBPC3:
c.2816G>A and PTCH1: c.1889G>A for the first time through Sanger sequencing.
Interactome network mapping and KEGG pathway analysis showed that the interacted
genes of MYBPC3 and PTCH1 are mainly involved in the hedgehog signaling pathway,
p53 signaling pathway, circadian rhythm, and cardiomyopathy. Mutation in MYBPC3 and
PTCH1 leads to a mutation in the interacted genes, which alters the pathways and may
cause both breast cancer and cardiomyopathy. Epidemiological factors such as high blood
pressure, diabetes and disturbed sleep patterns of cancer patients may be one of the
driving forces causing these changes in health conditions in the specific population. In the
fourth phase, the convolutional neural network achieved high accuracy (0.966) in
detecting mitosis in histopathology images, offering a promising method for identifying
breast cancer progression at early stages which is usually missed out or neglected. We
are the first to identify mitosis in real-time patient histology images. Integrating
epidemiological, genetic, and image analysis approach opens the door for innovative
strategies for prevention, treatment, and comprehensive care in the battle against breast
cancer. The outcome of the study highlights personalized treatment options based on an
individuals' risk factors and genetic alterations thereby offering hope for improved patient
outcomes and a better quality of life.
Description
Keywords
Biotechnology