Spatial patterns of mineralization for human iliac crest cancellous bone were measured from images obtained by quantitative backscattered electron microscopy. Biopsies collected from vertebral fracture patients and healthy individuals with high or low bone formation rate (BFR_s) were examined (fracture/low BFRs:​ N = 12, fracture/high BFR_s:​ N = 10, normal/low BFR_s:​ N = 12, normal/high BFR_s:​ N = 15). 20 by 20 pixel square areas or smaller were sampled from superficial and deep remodeling packets. Mean (Z_mean) and standard deviation (SD) of mineralization were measured, and coefficients of variation (CV = SD / Z_mean) were calculated. Fast Fourier transform analysis was used to quantify the distribution of the mineral in the packets. “FFT_ratio” was defined as the ratio magnitude of the principal spatial frequency to the average atomic number density. A higher FFT_ratio occurred in specimens with more pronounced alternating layers of light and dark as visible in the backscattered electron image, which was defined as lamellar patterning. Two-way ANOVA revealed that the coefficients of variation of mineralization for both superficial and deep packets were significantly lower in fracture patients than in normal individuals. However, the interaction between turnover rate and group (fracture/non-fracture) indicated that the difference in packet CV occurred among the low turnover individuals and not among those with high turnover. Mean mineralization levels and CV between deep and superficial packets were highly correlated. Regressions of packet CV of mineralization and FFT_ratio were highly significant (p < 0.001) for all packets pooled and for packets divided by group (fracture/normal). However, analyses of packet CV and FFT_ratio by individual were variable (R² from 0.00338 to 0.700).

Packet-level mineralization variability may be associated with fracture toughness, and fracture patients had less variable packet-level mineralization. The result that the packet CV varied significantly between fracture and non-fracture individuals with low turnover suggests that for low turnover subjects without fracture, high variability in mineralization may have a protective effect. In high turnover patients, the accelerated turnover may prevent the lamellar variability from developing over time. Strong correlations between CV and Z_mean for both superficial and deep packets imply that newly formed bone is created similarly to older bone within an individual. Fourier transform results show that the mineralization variability found within packets is associated with lamellar patterning. Lamellar structure has been hypothesized to guide microcrack propagation in order to optimize bone strength and toughness. Osteoporotics with fracture had less pronounced lamellation than healthy normals and may be more prone to fracture.