SUMMARY
There is a long-standing quest to better understand what causes our bones to break, especially as a devastating
and widespread consequence of osteoporosis. Beyond assessment of bone quantity (e.g., bone mineral density),
it is well-know that tissue-level quality plays a key role in determining bone strength and fragility. We and others
have shown that microscale tissue composition is intrinsically related to skeletal diseases; however, a limitation
of this approach is that it cannot capture features of the nanoscale building blocks of bone quality and integrity
(mineralized collagen fibrils and bundles on the order of 500 nm). Thus, analysis of bone composition at high
spatial resolution is needed to elucidate the nanoscale origins of impaired bone health. Additionally, there is a
paucity of research into the combined role of compositional properties in bone tissue quality, which is essential
to reveal the multifactorial nature underlying poor bone features in osteoporosis. To address these critical gaps
in knowledge, we aim to apply innovative approaches to enlighten the multifactorial relationship between
bone nanoscale composition and reduced bone tissue quality in osteoporosis. We hypothesize that
nanoscale compositional properties of bone are significant correlates to predict histological diagnosis and bone
morphologic features associated with osteoporosis. In Aim 1, we propose to determine and quantify nanoscale
compositional properties of healthy and osteoporotic bones. Readily available clinical bone biopsies will first be
evaluated for standard histopathological diagnosis, as well as by micro-computed tomography (microCT) of bone
morphometry. The samples will then be assessed by state-of-the-art optical photothermal infrared (O-PTIR)
spectroscopy and imaging, which allows breakthrough analysis of intact tissue composition at 500 nm spatial
resolution. Our supportive preliminary data show the acquisition and quantification of diverse bone nanoscale
compositional properties of mineral and collagen within individual osteon and trabeculae. Additionally, mineral
stoichiometry will be determined by scanning electron microscopy with energy dispersive X-ray spectroscopy
(SEM-EDX). With this rich dataset, we will perform a comprehensive analysis of comparisons and correlations
among bone features in healthy and osteoporotic bones to identify relevant nanoscale compositional properties
associated with typical osteoporotic bone quality. In Aim 2, we propose to apply machine learning methods to
elucidate the multifactorial relationship between bone nanoscale composition and osteoporosis. The goal will be
to predict histopathological diagnosis and morphometric features of normal and osteoporotic bones based on
input of combined nanoscale compositional properties. We will initially apply multivariable partial least square
(PLS) cross-validation, then focus on cutting-edge deep learning methods. This innovative approach will break
new ground towards elucidating which bone nanoscale compositional properties underlie poor bone quality in
osteoporosis and will inform future clinical studies into new therapeutic tissue targets to improve bone health.
