Applying Full Spectrum Analysis to a Raman Spectroscopic Assessment of Fracture Toughness of Human Cortical Bone. Issue 10 (October 2017)

• Record Type: Journal Article
• Authors: Makowski, Alexander J.; Granke, Mathilde; Ayala, Oscar D.; Uppuganti, Sasidhar; Mahadevan-Jansen, Anita; Nyman, Jeffry S.
• Is Part Of: Applied spectroscopy. Volume 71:Issue 10(2017)
• Page Start: 2385
• Page End: 2394
• Publication Date: 2017-10
• Subjects: Bone quality -- Raman spectroscopy -- toughness -- multivariate -- bone mineral density -- fracture risk; Spectrum analysis -- Periodicals; 543.505
• DOI: 10.1177/0003702817718149
• Languages: English
• ISSNs: 0003-7028

Abstract

A decline in the inherent quality of bone tissue is a † Equal contributors contributor to the age-related increase in fracture risk. Although this is well-known, the important biochemical factors of bone quality have yet to be identified using Raman spectroscopy (RS), a nondestructive, inelastic light-scattering technique. To identify potential RS predictors of fracture risk, we applied principal component analysis (PCA) to 558 Raman spectra (370–1720 cm –1 ) of human cortical bone acquired from 62 female and male donors (nine spectra each) spanning adulthood (age range = 21–101 years). Spectra were analyzed prior to R-curve, nonlinear fracture mechanics that delineate crack initiation (Kinit ) from crack growth toughness (Kgrow ). The traditional ν1 phosphate peak per amide I peak (mineral-to-matrix ratio) weakly correlated with Kinit (r = 0.341, p = 0.0067) and overall crack growth toughness (J-int: r = 0.331, p = 0.0086). Sub-peak ratios of the amide I band that are related to the secondary structure of type 1 collagen did not correlate with the fracture toughness properties. In the full spectrum analysis, one principal component (PC5) correlated with all of the mechanical properties (Kinit : r = − 0.467, Kgrow : r = − 0.375, and J-int: r = − 0.428; p < 0.0067). More importantly, when known predictors of fracture toughness, namely age and/or volumetric bone mineral density (vBMD), were included in general linear models as covariates, several PCs helped explain 45.0% (PC5) to 48.5% (PC7), 31.4% (PC6), and 25.8% (PC7) of the variance in Kinit, Kgrow, and J-int, respectively. Deriving spectral features from full spectrum analysis may improve the ability of RS, a clinically viable technology, to assess fracture risk.