Sensitive Multiplexed Spatial MicroRNA Profiling and Analysis Framework for Assessing Tumor Heterogeneity
Name
mohd-omarmohd-phd-eecs-2026-thesis.pdf
Size
15.59 MB
Format
Adobe PDF
Checksum (MD5)
04233e603c1dda23241628ff2675c749
Author(s)
Mohd, Omar Nazmi
Advisor(s)
Doyle, Patrick S.
Date Issued
February 2026
Publisher
Massachusetts Institute of Technology
Abstract
Many cancers exhibit intratumoral heterogeneity (ITH), a phenomenon that is characterized by variations in cellular composition and gene expression across tumor tissue. This spatial variation is often clinically consequential because it is associated with treatment resistance and disease recurrence. Characterizing such heterogeneity, however, is challenging because conventional assays often average signals across large tissue areas and obscure spatial structure, making spatial omics approaches that preserve tissue organization necessary. MicroRNAs (miRNAs) are a promising class of biomarkers and therapeutic targets in cancer, and their spatial distribution has been shown to reveal region-specific expression differences relevant to treatment outcome. However, there remains no widely adopted platform for spatial miRNA profiling that is both quantitative and multiplexed, particularly in formalin-fixed, paraffin-embedded (FFPE) tissue. As such, emerging spatial miRNA datasets are often targeted and quite small, which further increases the need for analysis methods that remain informative in data-scarce settings while still enabling cohort-level comparison and biologically grounded interpretation.
In this thesis, we developed a sensitive, quantitative, multiplexed, and spatially-resolved in situ miRNA assay for FFPE tissue using nanoliter well arrays with functionalized hydrogel posts, and we established an accompanying analysis framework for data-scarce spatial omics applications. We improved assay sensitivity and dynamic range through rolling circle amplification (RCA) and a novel data acquisition workflow that enabled detection down to 0.17 zeptomoles, supporting measurement of low-abundance clinically-relevant miRNAs in mouse breast tumor sections. To interpret these spatial miRNA outputs, we formulated a latent Dirichlet allocation (LDA)-based framework that learns interpretable miRNA topics for cohort-level tumor stratification and uses the structural similarity index measure (SSIM) to relate miRNA patterns to external spatial modalities. We applied the assay and analysis framework to a preclinical mouse BRCA1-mutated breast cancer case study of poly(ADP-ribose) polymerase (PARP) inhibitor-based therapies, demonstrating that early spatial miRNA patterns can stratify tumors by treatment response and reveal spatial associations with the underlying tissue architecture. We also present a preliminary application of both the assay and analysis framework to human BRCA1/2-mutated TNBC samples to assess cross-species generalizability of miRNA spatial patterns and translational potential.
MIT Department
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
Terms of Use
In Copyright - Educational Use Permitted
Copyright retained by author(s)
Persistent DSpace Link