Introduction: Age related macular degeneration (AMD) causes legal blindness worldwide, with few therapeutic targets in early disease and no treatments for 80% of cases. Extracellular deposits, including drusen and subretinal drusenoid deposits (SDD; also called reticular pseudodrusen), disrupt cone and rod photoreceptor functions and strongly confer risk for advanced disease. Due to the differential cholesterol composition of drusen and SDD, lipid transfer and cycling between photoreceptors and support cells are candidate dysregulated pathways leading to deposit formation. The current study explores this hypothesis through a comprehensive lipid compositional analysis of SDD. Methods: Histology and transmission electron microscopy were used to characterize the morphology of SDD. Highly sensitive tools of imaging mass spectrometry (IMS) and nano liquid chromatography tandem mass spectrometry (nLC-MS/MS) in positive and negative ion modes were used to spatially map and identify SDD lipids, respectively. An interpretable supervised machine learning approach was utilized to compare the lipid composition of SDD to regions of uninvolved retina across 1873 IMS features and to automatically discern candidate markers for SDD. Immunohistochemistry (IHC) was used to localize secretory phospholipase A2 group 5 (PLA2G5). Results: Among the 1873 detected features in IMS data, three lipid classes, including lysophosphatidylcholine (LysoPC), lysophosphatidylethanolamine (LysoPE) and lysophosphatidic acid (LysoPA) were observed nearly exclusively in SDD while presumed precursors, including phosphatidylcholine (PC), phosphatidylethanolamine (PE) and phosphatidic acid (PA) lipids were detected in SDD and adjacent photoreceptor outer segments. Molecular signals specific to SDD were found in central retina and elsewhere. IHC results indicated abundant PLA2G5 in photoreceptors and retinal pigment epithelium (RPE). Discussion: The abundance of lysolipids in SDD implicates lipid remodeling or degradation in deposit formation, consistent with ultrastructural evidence of electron dense lipid-containing structures distinct from photoreceptor outer segment disks and immunolocalization of secretory PLA2G5 in photoreceptors and RPE. Further studies are required to understand the role of lipid signals observed in and around SDD.

The search for molecular species that are differentially expressed between biological states is an important step towards discovering promising biomarker candidates. In imaging mass spectrometry (IMS), performing this search manually is often impractical due to the large size and high-dimensionality of IMS datasets. Instead, we propose an interpretable machine learning workflow that automatically identifies biomarker candidates by their mass-to-charge ratios, and that quantitatively estimates their relevance to recognizing a given biological class using Shapley additive explanations (SHAP). The task of biomarker candidate discovery is translated into a feature ranking problem: given a classification model that assigns pixels to different biological classes on the basis of their mass spectra, the molecular species that the model uses as features are ranked in descending order of relative predictive importance such that the top-ranking features have a higher likelihood of being useful biomarkers. Besides providing the user with an experiment-wide measure of a molecular species' biomarker potential, our workflow delivers spatially localized explanations of the classification model's decision-making process in the form of a novel representation called SHAP maps. SHAP maps deliver insight into the spatial specificity of biomarker candidates by highlighting in which regions of the tissue sample each feature provides discriminative information and in which regions it does not. SHAP maps also enable one to determine whether the relationship between a biomarker candidate and a biological state of interest is correlative or anticorrelative. Our automated approach to estimating a molecular species' potential for characterizing a user-provided biological class, combined with the untargeted and multiplexed nature of IMS, allows for the rapid screening of thousands of molecular species and the obtention of a broader biomarker candidate shortlist than would be possible through targeted manual assessment. Our biomarker candidate discovery workflow is demonstrated on mouse-pup and rat kidney case studies.