Microdamage Generated by Fatigue Loading in Cancellous Bone Does Not Occur
Preferentially Near Resorption Cavities
Matthew G. Goff, M.S.1, Floor M. Lambers, PhD2, Thu M. Nguyen2, Jed Sung2, Clare M. Rimnac3, Christopher J. Hernandez2.
1
Cornell, ithaca, NY, USA, 2Cornell University, Ithaca, NY, USA, 3Case Western Reserve University, Cleveland, OH, USA.
Disclosures:
M.G. Goff: None. F.M. Lambers: None. T.M. Nguyen: None. J. Sung: None. C.M. Rimnac: None. C.J. Hernandez: None.
Introduction: Accumulation of microscopic cracks and other tissue damage in bone is a key contributor to the development of
fatigue and insufficiency fractures [4]. In engineering components, microscopic cracks tend to initiate and accumulate during
cyclic loading at the location of stress risers such as notches or holes. In bone, resorption cavities are believed to act as stress
risers [2]. It is not known if resorption cavities promote the generation of microdamage in cancellous bone because threedimensional imaging is required to confirm the association between resorption cavities and microdamage (to avoid missing
correlations out of plane) and few three-dimensional imaging techniques have the resolution to reliably detect resorption
cavities in cancellous bone (1.5 µm or smaller voxel size) [6]. Here we test the hypothesis that microdamage formed during
fatigue loading occurs preferentially near resorption cavities. Specifically we determine the spatial association between
microdamage and resorption cavities in human vertebral cancellous bone under conditions of fatigue loading.
Methods: Cylindrical cores of cancellous bone were collected from the third lumbar vertebral body of 9 donors (7 male, 2
female, age range of 62-92 yrs.). The cancellous bone cores were cyclically loaded between 0N and a normalized stress σ/E0 of
0.0035 mm/mm (where E0 is the initial Young’s modulus) until the modulus was reduced by 32 - 82%. The specimens were then
submitted to lead uranyl acetate staining to identify microdamage [5].
Micro-computed tomography images of each specimen were collected using a machine that allows for scanning sub regions of a
large specimen at a high resolution (VersaXRm-520, Xradia, Pleasanton, CA). A region of interest (3 mm diameter, 3 mm length)
from the center of each vertebral core (8 mm diameter) was scanned with a 1.5 µm voxel size. Bone and microdamage were
automatically segmented using the two material Otsu method [3]. Resorption cavities were identified by bone surface texture
(the “scalloped surface”) and traced in three-dimensions by trained observers [1]. A total of 11 - 40 resorption cavities were
identified within each specimen.
We tested the proximity of microdamage and resorption cavities in three ways: 1) point proximity; 2) object volume proximity;
and 3) object number proximity. Point proximity of cavities near microdamage was determined as the proportion of randomly
selected points within a region of microdamage near a resorption cavity (pdamage) relative to the proportion of randomly selected
points from regions without microdamage (pnodamage). In a similar manner, the proportion of randomly selected points on
resorption cavities that were near microdamage (pcavity) was compared to the proportion of randomly selected points on
quiescent bone surface near microdamage (pnocavity). Object volume proximity was determined by comparing the percent of
resorption cavity surface (ES) and total bone surface (BS) that was near microdamage. Object number proximity was determined
by expressing the percent of all resorption cavities near microdamage as well as the percent of all microdamage objects that
were near resorption cavities. Each proximity measure was evaluated at four distances: 133 µm (average trabecular thickness),
67 µm, 34 µm and 17 µm (~2 osteoblast diameters). Comparisons between groups were made using ANOVA.
Results: The eroded surface (ES/BS) was 2.78 ± 0.92% (mean ± SD), number of cavities per bone surface was 0.75 ± 0.26 (1/mm 2)
and the damage volume fraction (DV/BV) was 3.83 ± 1.95%. Microdamage was broadly distributed throughout the cancellous
bone (Fig. 1). Point proximity assays found that the proportion of points of microdamage near resorption cavities (p damage = 0.21
± 0.09) was similar to the proportion of randomly selected points in bone (pnodamage = 0.20 ± 0.12). The proportion of points on a
resorption cavity that were near microdamage (pcavity = 0.84 ± 0.14) was similar to the proportion of randomly selected points on
the bone surface (pnocavity = 0.87 ± 0.13). The percent of eroded surface near microdamage was similar to the percentage of total
bone surface near microdamage and ES/BS near microdamage was similar to the whole specimen ES/BS (Fig. 2). Although most
of the resorption cavities were near microdamage events, only a small portion of microdamage events were near resorption
cavities (Fig. 3). Altering the distance used to define proximity did not have an effect on the relative likelihood that microdamage
and resorption cavities were near one another. No correlations between microdamage event size and probability of a spatial
correlation were observed, nor did the size of individual resorption cavities appear to be related to the spatial correlation assays.
Fig 1. An image of microscopic damage (red) and resorption cavities (blue) in the trabecular bone (tan) is shown.
Fig 2. The percent of eroded surface near microdamage was not different from the percent of total surface near microdamage.
Fig 3. While the majority of resorption cavities neighbored microdamage events (square dots), few microdamage events were
near resorption cavities (round dots) suggesting that microdamage is distributed in a manner independent of resorption cavities
Discussion: The current study provides an exhaustive test of spatial correlations between microdamage and resorption cavities
and does not support the idea that microdamage generated by fatigue loading occurs preferentially near resorption cavities. The
point proximity assays found pdamage to be so similar to pnodamage and pcavity to be so similar to pnocavity that over 1600 specimens
would be necessary to detect the observed difference with a power of 0.80 (α = 0.05). The object volume proximity and object
number proximity assays are also consistent with the idea that microdamage is not preferentially located near resorption
cavities. It is possible that microdamage caused by fatigue loading is more closely related to other characteristics of bone (tissue
material quality including heterogeneity) or that resorption cavities occur preferentially in regions of the trabecular structure
experiencing lower habitual stress/strain (as suggested by bone adaptation theory).
Significance: The accumulation of microscopic tissue damage during multiple loading events is a contributor to the development
of fatigue fractures and insufficiency fractures. The current work does not support the idea that resorption cavities influence the
location of microdamage caused by fatigue loading in otherwise normal human cancellous bone suggesting that other aspects of
bone tissue quality are more influential in determining the formation of microdamage.
Acknowledgments: NIAMS/NIH AR057362
References: 1. Goff, MG, et al. (2012). Bone 51(1): 28-37. 2. Hernandez, CJ (2008). Bone 42(6): 1014-20. 3. Otsu, N (1979). IEEE
Transactions on Systems Man and Cybernetics 9(1): 62-66. 4. Pentecost, RL, et al. (1964). JAMA 187: 1001-4. 5. Tang, SY, et al.
(2007). Bone 40(5): 1259-64. 6. Tkachenko, EV, et al. (2009). Bone 45(3): 487-92.
ORS 2014 Annual Meeting
Poster No: 1493