Steroid metabolomics: machine learning and multidimensional diagnostics for adrenal cortical tumors, hyperplasias, and related disorders

Abstract

Abstract Steroid profiling applications have a long history primarily directed toward\xa0diagnosis of endocrine disorders of childhood. Technological advances in mass spectrometry enabling rapid, sensitive, and specific measurements of multiple steroids in biological fluids are now paving the way for numerous other applications, including diagnosis of adrenocortical carcinoma, primary aldosteronism, and different forms of hypercortisolism. Such analytical procedures that target combinations of steroids in a single biological sample have potential for efficient one-shot\xa0methods for diagnosis of multiple disorders of steroidogenesis. Moreover, within a specific disorder, such methods can facilitate subtyping for more rapid therapeutic stratification than allowed by current methods that rely on single measurements per sample at sequential time points in a diagnostic process. Combined with advances in computational mathematics, such as machine learning, it is now possible to move from traditional unidimensional approaches for interpreting diagnostic data to methods that can interpret patterns in data. Mass spectrometry–based steroidomics provides an ideal platform for advancing such multidimensional approaches for disease diagnosis and stratification. Bottlenecks that must be overcome to move forward include needs for laboratory harmonization and method certification combined with ingrained reliance on outmoded, but well-accepted, diagnostic methods and general inertia to take advantage of new technologies.