Abstract A009: Precision pre-cancer detection via functional metabolic heterogeneity metrics from label-free, high resolution tissue fluorescence and transcriptomic data

Abstract While hallmarks of cancers have been established and updated for nearly fifteen years, pre-cancerous lesions remain less well-defined. Label-free, non-invasive, three-dimensional imaging approaches with micron-scale resolution have the potential to provide important functional and morphological insights regarding the progression of pre-cancers, and thus to enable the development of precision diagnostic hallmarks. Towards this goal, we have performed an iimaging study with freshly excised human cervical epithelial biopsies from 62 patients, including 33 benign, 14 low-grade and 15 high-grade squamous intra-epithelial lesions (LSILs and HSILs, respectively). We rely on endogenous fluorescence contrast from reduced nicotinamide adenine dinucleotide (NADH) and oxidized flavin adenine dinucleotide (FAD) and associated flavoproteins to acquire two-photon excited fluorescence (TPEF) images every four microns across the epithelial depth. We validate our findings with gene set enrichment analysis of three different publicly available datasets including cervical benign, pre-cancerous, and cancerous tissues. We analyze the images to extract quantitative metrics of the redox state and mitochondrial organization of the cells along with their depth-dependent variations. In addition, we quantify epithelial thickness and the cell nuclear to cytoplasmic ratio values across the epithelium. We observed decreases in all metrics of metabolic function heterogeneity as a function of depth in HSILs compared to non-HSILs. Specifically, changes in the overall redox state and mitochondrial organization of HSILs is consistent with enhanced glycolysis, fatty acid oxidation and synthesis. However, a more nuanced assessment of the overall TPEF NADH intensity changes, also indicate enhancements in oxidative phosphorylation. These metabolic HSIL changes are consistent with analysis of the transcriptomic data, confirming a distinct metabolic profile for HSILS vs cancers. Finally, predictive quadratic discriminant analysis algorithms leveraging a combination of the morphological and metabolic functional metrics extracted from the TPEF images yield a high sensitivity and specificity (>90% and 70%, respectively) for the classification of HSILs from non-HSILs. Importantly, this high diagnostic accuracy is maintained when metrics from only two depths, along the superficial and basal cell layers, are utilized. Together, these results highlight the potential of label-free, TPEF to yield high precision metrics of tissue morphology and function non-invasively. Transcriptomic data validate sensitivity to changes in several key metabolic pathways leading to a distinct profile for high grade cervical pre-cancers compared to benign and cancer tissues. These findings motivate TPEF imaging studies in vivo, with the potential of dynamic monitoring of pre-cancerous lesion evolution to yield enhanced understanding of factors that drive the progression of pre-cancers to cancers. Ultimately, such knowledge will be critical for developing precision diagnostics and therapies. Citation Format: Irene Georgakoudi, Christopher Polleys, Pramesh Singh, Hong-Thao Thieu, Elizabeth Genega, Narges Jahanseir, Andrea Zuckerman, Francisca Rius Diaz, Abani Patra, Afshin Beheshti. Precision pre-cancer detection via functional metabolic heterogeneity metrics from label-free, high resolution tissue fluorescence and transcriptomic data [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Functional and Genomic Precision Medicine in Cancer: Different Perspectives, Common Goals; 2025 Mar 11-13; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2025;85(5 Suppl):Abstract nr A009.