Influence of metastatic bone lesion type and tumor origin on human vertebral bone architecture, matrix quality, and mechanical properties

Abstract

Metastatic spine disease is incurable, causing increased vertebral fracture risk and severe patient morbidity. Here, we demonstrate that osteolytic, osteosclerotic, and mixed bone metastasis uniquely modify human vertebral bone architecture and quality, affecting vertebral strength and stiffness. Multivariable analysis showed bone metastasis type dominates vertebral strength and stiffness changes, with neither age nor gender having an independent effect. In osteolytic vertebrae, bone architecture rarefaction, lower tissue mineral content and connectivity, and accumulation of advanced glycation end-products (AGEs) affected low vertebral strength and stiffness. In osteosclerotic vertebrae, high trabecular number and thickness but low AGEs, suggesting a high degree of bone remodeling, yielded high vertebral strength. Our study found that bone metastasis from prostate and breast primary cancers differentially impacted the osteosclerotic bone microenvironment, yielding altered bone architecture and accumulation of AGEs. These findings indicate that therapeutic approaches should target the restoration of bone structural integrity.

Introduction

Vertebral bone metastases are found in up to 70% of patients with advanced cancers ⁽¹⁾, and for many patients, it is the first sign of malignancy ⁽²⁾. Pathologic vertebral fractures (PVF) occur in 15–30% of patients with spinal bone metastasis ⁽³⁾ when the affected vertebra cannot withstand daily loads. Initial fracture and any subsequent vertebral collapse can lead to severe impairment of quality of life ⁽⁴⁾ and higher healthcare costs ⁽⁵⁾. Up to 50% of the patients suffer from neurological deficits ⁽⁶⁾, leading to shortened patient survival ⁽⁷⁾ and a 3-year life expectancy ⁽⁸⁾. With recent improvements in the survival rates of patients with primary malignancies, PVF incidence is expected to rise ⁽⁹⁾. Identifying critical factors associated with PVF risk will improve the treatment and management of patients with metastatic spine disease.

Vertebral bone invasion with metastatic tumors is driven by cancer cell seeding, dormancy, and growth in the bone microenvironment ⁽¹⁰⁾, causing remarkable alterations in vertebral anatomy, vertebral bone architecture, and composition. Bone metastatic lesions are clinically classified based on radiological appearance ⁽¹¹⁾. Osteolytic bone (typical of breast, lung, urinary, and myeloma cancers) exhibits trabecular network’s rarefication with a low apparent bone density and includes lytic foci of varying size (Fig.1.A and includes lytic foci (Fig. 1A₁) of varying size (Fig. 1A₂). Typical of prostate and breast cancers, osteosclerotic bone (Fig. 1. B) exhibits higher apparent bone density, bone fraction (Fig. 1B₁), and dense trabecular network (Fig. 1B₂) on CT imaging ^(12,13). Mixed bone metastasis (typical of breast, lung, prostate cancers) ⁽¹⁴⁾ (Fig. 1C) exhibits regions containing both the destruction of the bone network (Fig. 1C₁) and sclerotic bone (Fig. 1C₂) within the vertebral anatomy. Although this subjective classification is widely used as part of the clinical management for patients at risk of vertebral fracture, it is prone to poor agreement between clinicians ⁽¹⁵⁾ and is imprecise for predicting PVF risk ⁽¹⁶⁾. Thus predicting PVF risk remains a significant clinical challenge ⁽¹⁷⁾.

Figure 1.

Sagittal and axial images of whole human vertebral specimens with osteolytic, osteosclerotic, and mixed bone metastasis with regions demonstrating typical bone features presented at higher magnification. In a human vertebra with osteolytic bone metastasis (A), rarefication of the bone network and at higher resolution (A₁ and A₂), destruction of bone tissue (A₂: red arrow) is observed. In vertebra with osteosclerotic bone metastasis (B), higher image resolution [B₁ and B₂] shows thickening of bone trabeculae, loss of bone marrow space, and extensive trabeculae remodeling. Both processes are observed in the human vertebra with mixed bone metastases (C) and when examining regions of the cancellous bone at higher image resolution [C₁ and C₂].

This challenge partly stems from the inability of clinical imaging to fully resolve the effect of bone metastases on bone microarchitecture ⁽¹⁸⁾ and quality and clinical assertions regarding the effect of bone metastases on the metastatic bone’s material properties. Derived from clinical observations of fracture patterns associated with the osteosclerotic and osteolytic bone metastasis ^(19–21), osteosclerotic bone is clinically assumed to indicate a materially stronger bone. In contrast, osteolytic bone is clinically assumed to indicate a materially weaker bone due to its radiographic appearance. At the bone tissue level, Stadelmann et al. ⁽²²⁾ found human vertebral osteosclerotic bone to possess lower elastic modulus and ductility, while osteolytic bone demonstrated little difference in either parameter than normal bone. These findings suggest that metastatic bone architecture and material properties are complex and unique to each bone metastasis phenotype.

Bone quality (i.e., bone microarchitecture, extracellular matrix composition, mineralization, markers of bone turnover, and microdamage) is increasingly gaining interest as a contributor of bone strength independently of its mineral density ⁽²³⁾. Beyond bone mass, fracture susceptibility is primarily associated with aging and disease-based alterations of bone proteins, predominantly type-I collagen ^(24–26). Nonenzymatic glycation (NEG), a common in vivo post-translational modification involving a reaction between glucose and lysine or hydroxylysine amino acid groups on type-I collagen, causes the formation of adducts known as advanced glycation end-products (AGEs). AGEs accumulation negatively affects bone fracture toughness (resistance to crack propagation) and mechanical properties ⁽²⁷⁾. A preclinical study by Bartlow et al. ⁽²⁸⁾ reported that radiation therapy increased AGEs formation in murine bone and was associated with bone embrittlement ⁽²⁹⁾. However, how bone metastasis phenotype affects AGEs accumulation in vertebral bone is unstudied and may aid in establishing new clinical modalities to reduce the risk of PVF.

This study investigated the effect of bone metastasis type on the architecture and organic matrix quality of human vertebral bone and evaluated the association between these effects and the vertebrae’s strength and stiffness. We hypothesized that 1) Metastatic lesion type differentially alters both microarchitecture and accumulation of AGEs; and 2) higher AGEs accumulation, indicative of suppressed bone remodeling ⁽³⁰⁾, is associated with lower vertebral strength and stiffness. For this purpose, multiple bone cores radiologically identified to contain osteolytic, osteosclerotic, and mixed bone metastasis were extracted from fourteen cadaveric vertebral bodies previously mechanically tested to measure their strength and stiffness ⁽²²⁾. Each bone core was µCT imaged at 10.5µm to evaluate bone microarchitecture and processed to measure the bone’s collagen and AGEs content.

Materials and Methods

Figure 2 provides a graphical summary of the study’s main experimental stages.

Figure 2.

Graphical summary of the study methodology. Fourteen metastatic human vertebral bodies (A) were mCT imaged at a resolution of 24.5mm (B) and mechanically tested (C) to measure strength and stiffness (D). Based on the mCT images, metastatic regions were identified, bone cores extracted (E), and each human bone core re-mCT was imaged at 10.5mm to evaluate the lesion’s effect on bone architecture (F). The imaged cores were theprocessed to evaluate the lesion’s effect on AGEs and collagen content (G)

2.1. Vertebral specimens.

a. Sample Preparation:

As part of a previous study ⁽²²⁾, we obtained nine cadaver spines (3 female, 6 male, age 49–71 years, mean age 54) from donors with solid cancer (3 breast, 3 lung, 2 prostate, and 1 kidney) through the Anatomy Gifts Registry (Hanover, MD). None of the donors had radiotherapy within 3 months before death. A standard spine imaging protocol was used to image the spines in a clinical CT scanner (Aquilion 64, Toshiba Medical, USA) ⁽²²⁾ with forty-five vertebral segments demonstrating osteolytic, osteosclerotic, or mixed bone metastasis extracted based on radiological review.

b. Mechanical Testing ⁽²²⁾:

In brief, each vertebral level was prepared to obtain plano-parallel vertebral body sections, 12mm to 25mm in height, and the vertebral segments imaged at 24.5 µm isotropic voxel size (µCT100, Scanco Medical, Bruettisellen) ⁽²²⁾. Based on our laboratory protocol ⁽³¹⁾, bone tissue was segmented, and the vertebral center of mass (COM) was computed ⁽³²⁾ to standardize the mechanical testing loading point.

Each vertebral segment was equilibrated in a 0.9% NaCl saline solution (one hour at room temperature) ⁽³²⁾, followed by a test to failure under monotonic (5 mm/min) axial compressive displacement ⁽²²⁾. With the test completed, the vertebral segment was wrapped in saline-soaked gauze and stored at − 20 °C in double plastic bags to await bone coring. Vertebral strength was defined as the maximum compressive force measured from the load-displacement curve. Vertebral stiffness was computed from a linear regression model’s coefficient fitted to the linear portion of the load-displacement curve (10 – 90% of the maximum compressive force). Segment’s demographics, bone mineral content values, strength, and structural stiffness values are summarized in Table 1 (Appendix). Based on the mechanical tests, fourteen vertebrae demonstrating either maximum or minimum strength values were selected to retrieve bone core samples.

2.2. Bone core samples.

a. Selection and Preparation:

For each selected vertebrae, the axial and sagittal vertebral μCT images ⁽²²⁾ were reviewed and, based on the bone radiological appearance, 4–6 locations in the body cross-section showing either osteolytic, osteosclerotic, mixed bone metastasis, or no radiological evidence of bone metastasis, were selected for bone core retrieval. Each core image-based X, Y coordinates were computed, and the core was extracted under constant water irrigation using a diamond-coated coring drill (5.08mm inner diameter, model 102055, Starlite industries, Rosemont, PA, USA) mounted on a CNC milling device. Each core was tagged, wrapped in saline-soaked gauze, and stored at − 20 °C in 1.5mL microtubes (MaxyClear, Fisher scientific, USA).

b. Bone core imaging:

The bone cores were immersed in saline using a custom imaging chamber and μCT imaged at 10.5μm voxel size (vivaCT 40; 70 kVp, 114mA, and 200ms integration time, Scanco Medical AG, Bassersdorf, Switzerland). Each core was re-tagged, wrapped in saline-soaked gauze, and stored in 1.5mL microtubes (Axygen™ MaxyClear, Fisher scientific, USA) at − 20 °C. Based on the radiological review of the µCT image data, each bone core was assigned a bone metastasis type (Appendix, Table A.1). A threshold value of 419 mgHA/cm³ was used for bone segmentation, and the µCT analysis software was used to compute the following bone architectural parameters: bone volume fraction, BV/TV (%), trabecular number, Tb.N (1/mm), trabecular thickness, Tb.N (mm), trabecular separation, Tb.Sp (mm), and connectivity density, Conn.D (1/mm3). Bone mineral content, BMC (mgHA), was computed from the complete bone core volume segmentation.

c. Measurement of the bone core total Fluorescent Advanced Glycation End-products:

Each imaged core was thawed, sectioned transversely to obtain two small cylindrical specimens (15–20 mm length x 5 mm diameter, ~40–50mg dry weight). The prepared bone specimens were defatted using isopropyl ether, lyophilized (freeze-dried) overnight, and hydrolyzed in 6N hydrochloric acid (HCL) based on dry weight for 16 hours. Fluorescence was measured for quinine sulfate standard (stock: 10μg/mL quinine per 0.1N sulfuric acid) and hydrolysate from the individual bone specimens at 360/460 excitation/emission ⁽³³⁾ using an Infinite 200 microplate reader (Tecan, Männedorf, Switzerland). The sample’s collagen content was calculated using a hydroxyproline standard of increasing concentration (stock: 2000 μg/mL L-hydroxyproline per 0.001 N HCL) ⁽³³⁾. The absorbance of the hydrolysates and hydroxyproline standard were subsequently measured at 570nm. Total fluorescent AGEs were computed as the amount of quinine per unit of collagen ⁽³³⁾ with fluorescence and absorbance measured using the microplate reader (Infinite 200, Tecan).

2.3. Statistical Analysis.

Statistical analyses were performed in JMP Pro (V14.3, SAS Institute, Inc., NC). Univariate analysis was applied to test for normality of distribution for bone architectural properties, BMC, AGEs, and collagen. Based on this analysis, we applied Kruskal–Wallis one-way ANOVA to test for the effect of bone metastasis phenotype (the independent variable) on; 1) vertebral strength, 2) vertebral stiffness, 3) imaged derived bone architectural parameters: (BMC, BV/TV, Tb.N, Tb.N, Tb.Sp, and Conn.D) and 4) measured AGEs and collagen content. Dunn With Control for Joint Ranks was used for post-hoc testing ⁽³⁴⁾.

The sampling of multiple vertebrae per spine can introduce clustering (non-independence) of the data ⁽³⁵⁾. We fitted linear mixed-effects models (LMM) under different assumptions about the correlation structure among segments from the same spines to test this effect on the association of BMC, AGEs, and collagen with the bone architectural parameters. Based on the corrected Akaike information criterion (AICc), the independence structure best fits the data ⁽³⁵⁾ (i.e., a lack of correlation among segments from the same spine).

With multiple cores obtained per vertebra, we applied multivariable LMM regression to test the independent contribution of the bone core’s 1) age, 2) gender, 3) race, 4) BV/TV, 6) architectural parameters, and 7) AGEs and Collagen content to the vertebral bodies measured strength and structural stiffness. Given that the strength of three vertebrae was greater than the limits of the testing system (15kN), the true strength of these vertebrae was not ascertained (right-censored data). We repeated the multivariable model for the strength analysis using tobin regression (allowing for a censored observed range of the dependent variable, which, in this case, was vertebral strength>15kN). Based on AICc, we found the LMM model best fitted the association between the bone core data and the vertebral strength and stiffness. The LMM analyses were repeated for each bone metastasis type to test whether the model’s prediction for vertebral strength and stiffness was retained for individual bone metastasis types. Statistical significance was set at the 5% level.

Results

3.1. Metastatic bone type alters vertebral strength and stiffness.

ANOVA analysis found that bone metastasis phenotype significantly affects vertebral strength (p=0.0078) and stiffness (p=0.0208). Post-hoc analysis showed osteosclerotic vertebrae with higher strength (p=0.0049) and stiffness (p=0.0264) than osteolytic vertebrae, Fig.3.

Figure 3.

Human osteosclerotic vertebrae show higher strength and stiffness than human osteolytic vertebrae. The strength and stiffness of human vertebrae with mixed bone metastasis appear closer to that of osteosclerotic vertebrae, suggesting the ratio bone metastasis type affects mechanical behavior. In contrast to human vertebrae with a mix or osteolytic bone metastasis, osteosclerotic human vertebrae with high strength showed high variation in stiffness.

3.2. Bone metastases type modulates bone mineral content and architecture.

Univariate statistics revealed bone tissue mineral content and architectural indices to be non-normally distributed. Compared to mixed and osteolytic bone metastases, osteosclerotic bone showed higher BMC values (72%, p=0.0049, and 257%, p<0.0001, Fig.4A), Tb.N (34%, p=0.0023, and 69%, p<0.0001, Fig.4E) and Tb.Th (35%, p=0.0196 and 52.4%, p=0.0003, Fig.4C) but with lower Tb.Sp (60%, p=0.0003 and 68%, p<0.0001, Fig.4D). These differences manifested in higher Conn.D (63%, p=0.0023 and 93%, p<0.0001, Fig.4F) and BV/TV (57%, p=0.0010, and 76%, p<0.0001, Fig.4B). Bone with mixed bone metastasis had higher BMC (108%, p<0.0001, Fig.4A), BV/TV (82%, p<0.0001, Fig.4B), Tb.N (54%, p<0.0001, Fig.4E), Conn.D (83%, p=0.0005, Fig.4F) but lower Tb.Sp (−96%, p<0.0001, Fig.4D) than osteolytic bone (Fig 4). Our analysis further revel that osteosclerotic bone from breast cancer donors exhibited higher Conn.D, a median (Q1-Q3) of 133.6 (77.3–194.7), Fig.5A than osteosclerotic bone from prostate cancer donors,60.0 (44.0–70.2), the difference statistically significnat (p=0.0145), Fig.5A.

Figure 4.

Illustration of bone metastasis radiological type (osteosclerotic, n=14, Mixed, n=10 and osteolytic, n=31) effect on human cancellous bone tissue mineral content (BMC), bone volume fraction (BV/TV), and the following stereological indices [trabecular number (Tb.N) thickness (Tb.N), separation (Tb.Sp) and connectivity density (Conn.D)] analyzed for the bone vertebral bone cores.

Figure 5.

In vertebrae from doners with breast [B] and prostate [D], cancer primaries appear to affect distinct micro-architectural features in the affected cancellous bone (demonstrated by the difference in Conn.D), in osteosclerotic, but not osteolytic, bone metastasis [A]. The observed pattern of bone deposition may explain this difference. High-resolution imaging of bone core obtained from breast cancer vertebra [C] exhibits a high degree of osteoid-like material in the inter-trabecular space with relatively little change in trabecular thickness. By contrast, the bone core obtained from the prostate cancer vertebra [E] exhibits remarkable, uniform thickening of the trabeculae with little or no evidence of deposition of osteoid-like material.

3.3. Bone metastases type differentially affect the association of bone volume fraction with bone architecture.

We found distinct associations between the phenotype of bone metastasis and alterations in the vertebral bone BV/TV, BMC, and architectural indices (Fig 6):

Figure 6.

Analysis of bone tissue obtained from the bone cores revealed that bone metastases type differentially affect the association between bone volume fraction and bone mineral content and architecture. In osteolytic bone (n=31), the association of BV/TV with lower BMC, Tb.N, and Conn.D but higher Tb.Sp and lack of significant association for Tb.N, suggests bone rarefaction via physical loss of bone tissue. In osteosclerotic bone (n=14), the association of higher BV/TV with higher BMC, Tb.N, Tb.N but lower Tb.Sp, and the surprising lack of association with Conn.D, suggests that the sclerotic tissue may occur via thicking of trabeculae and thdeposition of new osteoid-like material. Per its definition, bone with mixed bone metastasis (n=10) showed higher BV/TV associated with BMC and Tb.N, similar to osteosclerotic bone, with higher Conn.D similar to osteolytic bone,

• Osteolytic bone: BV/TV was associated with higher BMC (R²=0.86, p<0.0001), Tb.N (R²=0.74, p<0.0001) and Conn.D (R²=0.47, p<0.0001) and with lower Tb.Sp (R²=0.61, p<0.0001). BV/TV was not significantly associated with Tb.N (R²=0.01, p=0.6427).

• Osteosclerotic bone: BV/TV was associated with higher BMC (R²=0.42, p<0.0001), Tb.N (R²=0.69, p=0.0003), Tb.N (R²=0.31, p=0.0370) and with lower Tb.Sp (R²=0.51, p=0.0043). However, BV/TV and Conn.D was not significantly associated, (Fig 6).

• Mixed BM bone: Similar to the osteosclerotic bone, BV/TV was associated with higher BMC (R²=0.80, p=0.0004) and Tb.N (R²=0.84, p=0.0002). Similar to the osteolytic bone, BV/TV was associated with higher Conn.D (R²=0.64, p=0.0055). BV/TV was not significantly associated with Tb.Sp or Tb.N (Fig 6)

Multivariable analysis, used to investigate the independent contribution of BMC, architecture (Tb.N, Tb.Sp, Tb.N, Conn.D) and age, to BV/TV, found distinct differences between the BM phenotypes. In osteolytic bone, the model explained 88% of the variation in BV/TV (F:72.11, p<0.0001) with Tb.Sp (p<0.0001), Tb.N (p<0.0001) and Conn.D (p=0.0005) as significant independent correlates of BV/TV. In osteosclerotic bone, the model explained 94% of the variation in BV/TV (F:72.08, p<0.0001) with Tb.Sp (p<0.0001) and Tb.N (p<0.0001) significant independent correlates of BV/TV. In mixed bone metastasis, the model explained 98% of the variation in BV/TV (F: 73. 01, p=0.0024) with Tb.N (p=0.0001) and Tb.N (p=0.0405) significant independent correlates of BV/TV.

3.4. Accumulation of advanced glycation end-products (AGEs) is lesion type dependent.

We measured the amount of collagen and the modification of collagen by nonenzymatic crosslinking (by the accumulation of AGEs) to gain insight into the modification of bone matrix, particularly proteins, and identify matrix-based mechanisms contributing to the altered bone structural organization and mechanics. Bone metastasis phenotype significantly affected both AGEs accumulation (p=0.0004) and collagen content (p=0.0003) in the bone samples (Fig 7). Post-hoc comparisons showed osteosclerotic bone to exhibit significantly lower AGEs and higher collagen content compared to the osteolytic (a median difference of −159.4% (p=0.0012) and 60.8%, p<0.0001 respectively) and mixed bone metastases (−226.7%, p=0.0002 and 185.5%, p=0.0002) (Fig 7). There was no statistically significant difference between mixed bone metastasis and osteolytic bone for AGEs or collagen content. Independent of bone metastasis type, LMM analysis found AGEs correlated with higher Tb.Sp (F=8.7, p=0.0049) and lower BMC (f=8.44, p=0.0054), and Tb.N (F=11.0, p=0.0016). Of note, we observed no significant correlation for AGEs with Tb.N. No significant correlation was found between collagen content and the bone architecture indices.

Figure 7.

Analysis of human bone tissue obtained from the vertebral bone cores found bone with osteosclerotic bone metastasis contains lower AGEs and higher collagen content, indicative of higher bone remodeling, while, by comparison, the osteolytic bone higher AGEs and lower collagen content indicate lower remodeling rate and possible degradation of bone matrix.

3.5. BMC, AGEs, and bone architecture affect metastatic vertebral strength and stiffness.

LMM analysis showed that multiple sampling of bone cores per vertebrae had no significant effect on these associations (Wald p test >0.05). Univariate LMM analysis of the pooled data (independent of bone metastasis phenotype) revealed a higher BMC and lower AGEs associated with higher vertebral strength (F=37.38, p<0.0001 and F=12.72, p=0.0020) and stiffness (F=18.72, p<0.0001, and F=6.02, p=0.0206) (Fig 8). For bone architectural parameters, higher vertebral strength was associated with higher Tb.N (F=60.44, p<0.0001), Conn.D (F=54.75, p<0.0001), Tb.N (F=4.55, p=0.0380), and lower Tb.Sp (F=35.16, p<0.0001). Vertebral stiffness was associated with higher Tb.N (F=30.48, p<0.0001) and Conn.D (F=35.57, p<0.0001) but lower Tb.Sp (F=15.65, p=0.0001).

Figure 8.

tobbit regression showed higher BMC and lower AGEs values of the bone tissue associated with higher whole vertebral strength and stiffness. Note the higher variance in AGEs in the prostate bone and BMC in the breast cancer bone.

Multivariate LMM analysis, incorporating age, gender, BV/TV, BMC, architectural indices, and AGEs (modeled as fixed effects) with vertebral level set as a random variable, showed age (p=0.0012), Tb.Sp (p<0.0001) and Conn.D (p<0.0001) to be significant independent correlates of the measured vertebral strength. BV/TV was a significant independent correlate (p=0.0112) of vertebral stiffness with Tb.N at near significance as an independent correlate (p=0.0707).

Discussion

This study demonstrated that osteosclerotic, osteolytic, and mixed bone metastases uniquely alter BMC and the architectural features of cadaveric vertebral cancellous bone from donors with a range of solid sarcomas. Bone metastases types were associated with distinct differences in the accumulation of AGEs within the organic bone matrix, suggesting that lesions influence bone cellular activity/turnover and, consequently, bone matrix nanoindentation properties ⁽²²⁾. Gender also affected distinct differences in microarchitecture and AGEs accumulation in osteosclerotic but not osteolytic bone.

Preclinical murine models of osteolytic and mixed skeletal metastases ^(36–38) attributed the decrease in bone fraction to the significant but incomplete loss of physical trabeculae, i.e., a lower number and thinner bone trabeculae with reduced bone connectivity ^(38,39). Consistent with these reports, we found that a lower number of bone trabeculae with reduced connectivity, older age, and female gender were associated with greater BMC and BV/TV loss in human osteolytic vertebral bone. In contrast to preclinical models ^(38,39), we did not observe an association between changes in trabecular thickness and BV/TV in the human bone samples. However, our study’s strong association between lower bone connectivity and lower bone fraction suggests that osteolytic lesions affect bone loss in human bone by eliminating rather than thinning the bone trabeculae. Furthermore, we found that osteolytic bone tissue contained higher AGEs, suggesting that osteolytic bone exhibited limited new bone tissue formation and contained older and more fragile bone tissue ⁽²⁷⁾. The accumulation of AGEs is reported to be associated with reduced bone formation ⁽⁴⁰⁾, caused by reduced osteoblastic differentiation and activity ⁽⁴¹⁾, and reduced bone resorption, caused by reduced osteoclastic activity ⁽⁴²⁾. Our finding of more rods than plates in the osteolytic case is consistent with previous studies that showed AGEs are higher in trabecular rods than plates that are preferentially lost with age ⁽³³⁾.

In contrast to osteolytic, osteosclerotic bone regions were radiologically characterized by high BMC and BV/TV ⁽¹¹⁾. In human bone from prostate cancer patients, osteosclerotic bone exhibits an increased number of bone trabeculae and bone surface area and a lower degree of structural anisotropy than healthy bone ^(12,13). Consistent with these reports, our stereological analysis showed trabecular thickening and decreased intertrabecular distance with higher BMC in the osteosclerotic bone (up to 30% of samples had BV/TV values > 50%, suggesting a near-solid bone). However, in contrast to studies in human ^(12,13) and mouse ⁽³⁹⁾ osteosclerotic vertebral bone, neither new bone trabeculae nor higher bone connectivity was associated with higher BMC. Indeed, similar to previously reported histological observation of unorganized woven structure in newly remodeled osteosclerotic bone ⁽⁴³⁾, and in concert with trabecular thickening and higher BV/TV, we found the osteosclerotic bone to contain more collagen and exhibit lower AGEs values indicative of increased matrix production in the newly deposited osteoid-like mineralized bone tissue. These findings suggest that increased BMC occurs via extensive bone deposition and remodeling of the existing bone trabeculae network.

This study found differences in the bone architecture of osteosclerotic bone obtained from breast and prostate cancer donors. In osteosclerotic bone obtained from breast cancer donors, we observed an abundant deposition of mineralized bone-like tissue in the marrow spaces, increased number of bone trabeculae, and lower intertrabecular spacing with little evidence of trabecular thickening. In contrast, osteosclerotic bone obtained from prostate cancer showed opposite micro-architectural features. Endocrine ⁽⁴⁴⁾ and paracrine ⁽⁴⁵⁾ factors modulate the interaction of breast and prostate cancer-derived metastatic tumor cells, including the differentiation, proliferation, and maturation of osteoblasts and the inhibition of osteoclasts ⁽⁴⁶⁾, resulting in uncontrolled bone formation ⁽⁴⁷⁾. Although the mechanism underlying the observed differences in osteosclerotic bone architecture is unknown, this novel finding suggests that primary cancer (breast vs. prostate) may result in distinct differences in micro-architectural features of the osteosclerotic bone tissue. Whether these differences affect, the osteosclerotic vertebral bone failure process is under active investigation in our laboratory.

Preclinical studies reported bone with mixed bone metastasis to exhibit increased osteolysis, indicated by increased osteoclast number and decreased bone volume with regions of enhanced yet immature bone growth ⁽³⁸⁾. Furthermore, in canine models of prostate cancer, lesions yielding mixed bone metastasis exhibited lower trabecular number and spacing (generalized loss of trabecular bone rather than trabecular thickness) ⁽⁴⁸⁾. In agreement with these reports, this study’s stereological analysis found that human bone with mixed bone metastasis exhibited features typical of osteosclerotic (higher bone trabecular thickness associated with higher BMC) and osteolytic (moderate association of BMC with Conn.D) bone. However, in these cases, BMC was weakly associated with either trabecular number or separation compared to osteolytic or osteosclerotic bone tissues. Although research on the effect of mixed bone metastasis on human bone microarchitecture is rare, our findings in humans and the previous finding of murine models ⁽³⁸⁾ suggest that incomplete osteosclerotic activity rather than excessive resorption dominate the structural changes in bone with mixed/span>bone metastasis.

We found that AGEs normalized to bone collagen were significantly higher in bone with osteolytic and mixed bone metastasis than with osteosclerotic bone metastasis. Newly formed and younger bone matrix contains lower AGEs ⁽⁴⁹⁾. The presence of lower AGEs in osteosclerotic bone metastasis is likely to result from increased collagen production (See Fig 5), present in newly formed osteoid-like mineralized bone. In contrast, osteolytic, mixed bone metastasis, and older trabeculae are higher in AGEs ⁽³³⁾. Our results agree with previous studies demonstrating an increase in AGEs, namely pentoside (PEN) and carboxymethylysine (CML), with osteolytic involvement ^(50,51). Here, we measured total fluorescent AGEs in bulk, including PEN, to highlight the role of sugar-induced AGEs in osteolytic and mixed bone metastasis for the first time. Our finding of high AGEs in osteolytic bone tissue compared to the mixed and osteosclerotic bone tissue suggests reduced bone formation leaving behind older bone tissue ⁽³³⁾. Thus, the accumulation of AGEs may also modify cell-matrix interactions altering subsequent osteoblastic and osteoclastic differentiation, influencing bone architecture and mechanical properties, and contributing to the differences observed between osteolytic and osteosclerotic bone.

Vertebral cancellous bone, representing a foam-based structure designed to achieve high strength and stiffness with relatively small mass amounts, forms the predominant load-carrying structure in the vertebral body ⁽⁵²⁾. Studies in mouse ⁽³⁹⁾ and human ⁽⁵³⁾ bone samples highlighted BMC, BV/TV, and specific architectural features (higher trabecular thickness and lower inter-trabecular spacing) as determinants of metastatic bone strength. Although we tested a limited number of human vertebral specimens demonstrating a range of bone metastasis types, we found that metastatic lesion’s effect on bone mineral content and architectural features was strongly associated with vertebral strength and stiffness. Previous findings demonstrated that metastatically involved bone experienced higher strains localized to areas of degraded osteolytic bone ⁽⁵⁴⁾. In our study, severely disrupted structural synergism between bone stereology (trabecular separation, number, and connectivity) and the tissue mineral content suggests that this altered structural loading may affect early failure initiation and loss of vertebral strength.

The contribution of bone “quality” to pathologic bone strength remains unclear. The accumulation of AGEs in aged bone tissue is linked with altered architecture (higher in rods than plates ⁽³³⁾ and reduced toughness ^(55,56), whichin turn reduces bone resistance to fracture. Consistent with these studies, we found higher bone AGEs accumulation associated with lower vertebral strength. However, the osteosclerotic bone showed high BMC, more collagen, and low AGEs accumulation. Such differences indicate that osteosclerotic bone tissue is different from osteolytic bone tissue and may consist of newly deposited and less mineralized woven bone ⁽⁵⁷⁾ that exhibits slightly lower mean indentation modulus ^(22,50) mean hardness ⁽⁵⁰⁾, and ductility ⁽²²⁾. Indeed, our stereology analysis suggests the osteosclerotic vertebrae’s high strength and stiffness are achieved by the abundant deposition of new bone material rather than improved bone structural network and bone material properties. Consistent with Burke et al. ⁽³⁹⁾, our observations indicate that metastatic lesions change bone mineral content, structure and composition to affect vertebral mechanical behavior.

Our study has several limitations. Cancer patients are exposed to extensive oncological treatments, including chemotherapy, immunotherapy, and radiotherapy, affecting bone architecture and density and, consequently, vertebral fracture risk ^(58,59). Although we had the donor’s medical history, primary cancer diagnosis, chemotherapy, pharmacological, and surgical treatments, we had little information regarding radiotherapy and, if so, which levels were treated. Hence, we cannot discount that these treatments could have affected the measured vertebral bodies’ strength and stiffness and collagen and AGEs values. The fourteen vertebral bodies employed in this study were from nine donors with a range of primary cancers and were exposed to mechanical testing as part of a previous study. Our bone cores were selected from these mechanically tested vertebral bodies. Although the micro-CT images did not reveal the bone architecture to be damaged, we cannot assume that the previous testing did not impose mechanical damage at the bone tissue level.

Given that in cortical bone, both magnitude and the type of damage affected indentation derived material parameters and other mechanical properties of ⁽⁶⁰⁾, we did not measure the bone cores’ strength and stiffness. Our previous work demonstrated whole vertebral BMC and the computation of vertebral strength based on homogenised finite element analysis to provide moderate [BMC: R²=0.66, p<0.001] and strong [hFE: R²=0.78, p<0.001] prediction of pathologic vertebral strength ⁽²²⁾, suggesting that the mechanical properties of the pathologic vertebral body are driven predominantly by spatial distribution and densitometric properties of the lesioned bone. This study finding of the association between bone metastasis’s type and change in bone architectural indices and AGEs suggests that additional work is needed to understand better whether the effect of metastasis on bone quality significantly contributes to the strength of whole human pathologic vertebrae.

In conclusion, metastatic tumors affect human vertebral bodies by altering bone mineral content, microarchitecture, and extracellular matrix composition. Primary cancers vary significantly within lesion types, particularly for osteolytic bone metastasis. Each lesion type also confers a distinct alteration of bone structure and, to a lesser extent, material properties independently of the primary tumor origin. Therapeutic approaches should, therefore, target the restoration of bone structural integrity. The relationships highlighted in this study may aid in forming the foundation for technologies to assess the risk of pathologic fracture clinically.

Data Availability Statement:

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.