INTRODUCTION

Osteoporosis is a progressive disease where the rate of bone loss is greater than that of new bone production.  Although bone density changes do occur with aging (Ding et al., 2002), osteoporotic bone typically demonstrates severely decreased mineral density, demonstrates degradation in biomechanical properties, and is more susceptible to fractures (Blain et al., 2008).  The primary location for bone loss and structural alteration is within the long bones of the body (e.g., tibia) where the rod and plate microarchitecture of the internal trabeculae is heavily implicated in the overall structural integrity of bone (Saha et al., 2000; Saha et al., 2010) .  Although there are several clinical tests that can determine the presence of bone loss (e.g., DXA, REMS), identifying early stages of the disease may be possible by examining the internal architecture of the trabecular bone (Chang et al., 2011).  Such internal structure information can be gathered from clinical CT (computer tomography) data and may provide early evidence of disease onset (Chen et al., 2018; Guha et al., 2022). To further expand the scope of using CT information to assess the trabecular structure and to compare its performance with other established modalities [e.g., MRI (Chang et al., 2008; Link et al., 1998; Majumdar et al., 1996; Wehrli et al., 2002) and HR-pQCT (Boutroy et al., 2005; Burrows et al., 2010)] we developed a set of anatomical specimens that might serve as models.  These models, with the internal trabeculae appropriately fixed, can be scanned at different intensities, within different scanners, and in different configurations to determine the validity and reliability of a scanned data set for assessing trabecular bone integrity.  Here we investigated appropriate substrates (i.e., paraffin, gelatin matrix, silicone polymer) that maintained the trabecular network and bone density in bone specimens used for comparative studies in osteoporosis. The process can then be replicated with bone specimens demonstrating various stages of osteoporosis.

MATERIALS AND METHODS

Overview
Fresh-frozen distal tibia specimens (n = 12) were separated from cadaveric human legs (5F/2M, range 68-95y) at mid-tibia for this study.  Each specimen was scanned using multi-row detector computer tomography (MDCT) at different times in the study to illustrate and/or quantify the preservation of microstructure and continuity of cortical and trabecular bone constituents.  Selected specimens were rescanned with micro computed tomography (micro-CT) to investigate the trabecular bone microstructure.  To determine an appropriate substance to effectively stabilize the trabecular network and approximate the marrow for scanning, specimens were impregnated with one of three stabilizing substrate conditions: paraffin wax (n=8), gelatin matrix (n=1, repeated), or silicone polymer (n=3). The basic design of the analysis was a repeated measures assessment within each condition, comparing specific trabecular bone measures to demonstrate correlations between the preprocessed and impregnated specimens.

MDCT Imaging
The MDCT scans were performed on a Siemens SOMATOM Force (Forchheim, Germany) scanner located at the University of Iowa Comprehensive Lung Imaging Center (ICLIC) research CT facility using the UHR CT mode.  For each scan, the tibial axis of each specimen was aligned with the scanner center using the reference of laser ray coordinates to achieve the highest spatial resolution.  An anterior-posterior projection CT scout scan was used to ensure the inclusion of the inferior tibial articular surface in the field-of-view.  The inferior tibial articular surface was used as a reference plane from which to define different sites for computation of bone microstructural measures. The following CT scan parameters were used ― single X-ray source spiral acquisition at 120 kV, 100 effective mAs, 1 second rotation speed, pitch factor: 1.0, number of detector rows: 64, scan time: 5.8 seconds, collimation: 64 × 0.6 mm.  Siemens z-UHR scan mode was applied enabling Siemens double-z sampling technology.  Images were reconstructed at in-plane resolution of 150 µm and 200 µm slice-spacing using Siemens’s special kernel Ur77u with Edge Technology to achieve high spatial resolution.  A Gammex RMI 467 Tissue Characterization Phantom (Gammex RMI, Middleton, WI) was separately scanned using the same protocol to calibrate CT intensity values to bone mineral density (BMD).

Micro-CT Imaging
Selected micro-CT (µCT) scans of cadaveric specimens were performed on an Xradia 520 Versa scanner (ZEISS, Oberkochen, Germany) at the Iowa Institute for Biomedical Imaging (IIBI) laboratory after removing soft tissue and disarticulating the tibia from the ankle joint. Specimens were thawed at room temperature before scanning to avoid subtle movement that can occur during thawing.  The following µCT parameters were used: 100 kV, 90 µA, 1601 projections over 360 degrees, exposure 4 sec per projection, scan time: 3 hours (approximately), scan-length: 36 mm (approximately), low-energy filter LE4 and standard beam hardening correction were applied, and images were reconstructed using a smooth 0.5 kernel with image array size 2048 x 2048 and 18.63 µm isotropic voxel size.

Initial Scans and Preparation
 As part of a larger project, the initial MDCT scan of each specimen was performed with intact surrounding soft tissue, prior to any processing.  After this initial MDCT scan, the distal tibia was isolated from the full lower-leg specimen with two nominally horizontal parallel cuts.  First, specimens were removed from the freezer (-20 °C), thawed for 24 hours, and allowed to reach room temperature.  All soft tissues (i.e., skin, muscle, tendon sheaths, etc.) were then dissected away from the tibia using a #10 scalpel.  The ankle joint capsule was opened, and the talus was disarticulated from the ankle mortise by transecting all capsule and ligaments of the joint.  The syndesmotic ligaments and the intraosseous membrane were then transected, and fibula removed.  To remove the entirety of the articular surface and the medial malleolus, the tibia was oriented with the incisura up and cut perpendicular to the tibial shaft in both the sagittal and coronal planes at a level 5 mm above (proximal to) the articular surface as visible in the incisura.  A second, parallel cut was then made 8 cm proximal to the first.  The distal tibia segment was then refrozen to -20 °C until the next scanning session.  The frozen distal tibia segments with intact marrow were brought to the µCT imaging facility, thawed for 24 hours, allowed to reach room temperature, and the first µCT scans were performed.

Removing Marrow
After the first µCT scans were completed, specimens were placed in a series of acetone baths to remove water and marrow.  Specimens were placed in a volume of acetone that was equivalent to 10X the volume of the specimen and allowed to bathe, at room temperature, for 3-5 days.  The specimens were then drained and placed in a similar volume of fresh acetone for another 3-5 days.  This series was continued until the specific gravity of the remaining acetone was < 0.798, indicating that the specimens were effectively dehydrated.  Once dehydrated, specimens were moved to a (24h, 4%) formalin bath to properly fix any remaining tissue.  Specimens used for the paraffin wax and gelatin matrix substrates were drained, returned to a fresh acetone bath for rinsing, and then air dried just prior to impregnation to minimize flammability risks associated with the heated substrates.

Intermediate Scans
Both MDCT and µCT scans were performed on a subset of specimens (n=4) after the marrow removal and prior to impregnation with paraffin to determine if any microstructural damage occurred during the de-marrowing process.

Impregnating Specimens
Specimens were impregnated with either histology-grade paraffin, gelatin matrix, or silicone polymer to preserve the trabecular arrangement for prolonged/repeated scanning and cross modality comparisons.

For the paraffin and gelatin matrix procedure, tibial specimens were drained of acetone and dried in a fume hood (1h) prior to being placed in a molded container within a vacuum chamber. This precaution was set given the flammability of acetone.  Specimens were supported within the mold by the addition of rubber stoppers that served to occupy volume and provide easy release of the specimens.  Liquid paraffin (65°C) paraffin, or gelatin mixture (70 °C) was added to the mold to fully submerge the specimen with an additional volume such that there was a minimum 2 cm above the height of the specimen (Fig. 1B). The lid of the vacuum chamber was secured as the vacuum gradually increased (57.6 kPa) and then that vacuum was sustained for 15 minutes.  The vacuum rate was controlled by manually adjusting intake valves that were attached in-line with the manometer.  This gradual pressure reduction, coupled with the sustained vacuum, permitted air/acetone within the bone specimens to escape without aerosolizing the liquid paraffin/gelatin (Fig. 1C).  After the specimen was filled with liquid paraffin/gelatin, the mold containing the specimen was removed from the vacuum chamber and allowed to harden for 24 hours (Fig. 1D).  The paraffin block was removed from the mold and divided along break lines to free the tibial specimen.

The procedures for impregnating specimens (n=3) with silicone polymer (NC-S10) followed standard protocol for the plastination process (Henry et al., 2019).  Caution was taken to minimize leakage by curing the specimens with the inferior aspect of the tibia submerged in polymer. Any hardened polymer was removed from the external aspect of the specimen for CT analysis. Specimens were impregnated and fully cured/hardened (Fig. 1E).

 Follow-up Scans
MDCT and µCT scans were performed on a subset of specimens (n=4) after impregnating with paraffin.  Only MDCT scans were completed on the remaining specimens.

Image Registration and Processing
MDCT intensity values in Hounsfield unit (HU) were converted to BMD (mg/cm³) using matching scans of a Gammex RMI 467 Tissue Characterization Phantom (Gammex RMI, Middleton, WI, USA) and an automated custom algorithm (Chen et al., 2018).  µCT images were converted to BMD using a vendorsupplied calibration phantom with four rod inserts at different material densities. For quantitative microstructural analysis, MDCT scans were interpolated to obtain 150 µm isotropic voxels, while µ-CT scans were down-sampled to 50 µm voxels. Tibia bone was first segmented using a previously validated algorithm based on thresholding and multi-scale connectivity and morphological analysis (Li et al., 2015).  Then a quasi-cylindrical axial region of interest (ROI) at 4-8% of tibia length was selected on the first MDCT scan of each specimen using 30% peel from the outer cortical surface (Li et al., 2015). Spherical sample regions, each of diameter of 7.05 mm, were randomly selected over the quasi-cylindrical ROI to compute and compare target measures over matching sample regions after different experimental steps.  To evaluate the preservation of trabecular bone micro-structure after de-marrowing, 25 sample regions were selected from each of the four specimens scanned after de-marrowing but prior to paraffin impregnation (n = 4x25). Fifty sample regions were selected from each of five specimens used to compare between microstructural metrics before de-marrowing and after paraffin impregnation at MDCT and µCT imaging (n = 5x50).  All other MDCT scans and µCT scans of a given specimen were registered to the first MDCT scans to determine the matching ROIs.  Within each ROI, the following trabecular bone microstructure measures (Table 1.) were computed.  Pearson correlation coefficient was computed to show the relationship for microstructural metrics over sample regions before de-marrowing and after de-marrowing but prior to paraffin impregnation. It was also evaluated to show the association of microstructural metrics before de-marrowing and after paraffin impregnation at MDCT and µCT imaging.

RESULTS

Determining which impregnated specimens would best represent models for further study required a demonstration of sufficient marrow clearing.  This clearing was necessary to ensure that the impregnated specimens demonstrated appropriate filling and adequately surrounded the trabecular network.  Perhaps more importantly, the de-marrowed and impregnated specimens must also demonstrate a stable and undisturbed trabecular network.

De-marrowed Specimens
Qualitative inspection of registered axial µCT scans from a specimen in the preprocessed and de-marrowed state illustrates that the marrow (preprocessed) (Fig. 2A) is effectively removed (Fig. 2A’).  This is evident because both images are similarly windowed for image analysis, and a soft tissue CT signal (gray color) is observable in only the preprocessed specimen (Fig. 2A).

Quantitative exploration of the de-marrowing process also demonstrates that the structural and mechanical properties of the trabecular network are preserved. Quantified trabecular area network (Tb.NA) results from the µCT analysis across specimens demonstrates a very strong Pearson correlation coefficient (r = 0.998) between the pre-processed and de-marrowed states (Fig. 3).  Total bone volume fraction (BVF) comparisons between these conditions also demonstrated a very high correlation (r = 0.987) and is illustrated in Figure 4.  Single factor analysis of variance (ANOVA) also demonstrated that both Tb.NA (p = 0.96) and BVF (p = 0.68) were not statistically different between the two states. Together, the qualitative and quantitative results provide evidence suggesting that specimens are adequately cleared through the de-marrowing process, and the trabecular network remains intact.

 Impregnated Specimens – Qualitative:
Qualitative inspection of registered axial MDCT scans from the impregnated state illustrates how well the impregnation material filled the vacant marrow space (Fig. 5).  The silicone polymer (n= 3) appeared to fill the marrow space and adhere effectively to the trabecular network, yet it slightly altered the signal of both the trabecular and cortical regions (Fig. 5A cf. A’).  Moreover, cavitations were more evident with increased areas of void (Fig. 5A’), suggesting that the marrow space is unlikely to be fully filled which would ultimately affect further analysis when these specimens are rescanned at different institutions or with different MDCT scanners.  Although the trabecular integrity may be preserved, the marrow replacement was not adequate.

Preparing an adequate gelatin matrix solution that effectively filled the trabecular space and provided a stable specimen for analysis proved challenging.  Specimens (n=1, repeated) were qualitatively analyzed based on the MDCT scans.  Given that the trabecular spaces were not appropriately filled and demonstrated increased voided space (Fig. 5 B’’), µCT scanning was not completed on these specimens.  These attempts were on the same bone specimen, where after the first attempt the gelatin matrix was removed and subsequently replaced with a different gelatin composition.  Attempts to further alter the constituent composition of the gelatin matrix were abandoned.

Paraffin wax impregnated specimens (n=8) were qualitatively superior in both filling and preservation of the trabecular network.  MDCT scans demonstrated appropriate trabecular network visualization between the preprocessed (Fig. 5C) and paraffin impregnated states (Fig. 5C’’).  A subset of these specimens (n=4) was re-scanned with µCT to demonstrate the state of the specimen throughout each phase of the process (Figure 6).

Impregnated Specimens – Quantitative:
Strong correlations in quantitative measures of paraffin impregnated specimens were observed in both µCT and MDCT, with higher correlations being observed in the µCT scans.  Correlation of Tb.NA between the preprocessed and paraffin impregnated states was very strong in the µCT scans (Fig. 7A, r = 0.99).  MDCT scans of the same specimens demonstrate a similar positive correlation (Fig. 7B, r = 0.94) that is slightly lower as compared with the µCT scans.  Measures of BVF were also highly correlated in the µCT scans of the preprocessed and paraffin impregnated specimens (Fig. 7C, r = 0.97), however the correlation from the MDCT scans was, as observed other measures, slightly lower (Fig. 7D, r = 0.92).  Although ANOVA results demonstrated no difference between the preprocessed and impregnated measures for Tb.NA (p = 0.92) and BVF (p = 0.79) measures in the µCT scans, differences were evident in the MDCT scans (p < 0.05).

DISCUSSION

When using preserved bone models as phantoms for studying osteoporosis, it is critical to establish that the internal trabecular architecture is appropriately fixed to ensure reliability and validity across multiple scans and scanning conditions.  Further, the method of fixation must ensure the internal structure appropriately represents the native construct of the marrow cavity.  Data from the current study provides strong evidence that the de-marrowing process did not significantly alter either the trabecular architecture, Tb.NA, or the BVF (Fig. 4 and 5).  Filling the marrow cavity with paraffin after de-marrowing provided strong correlations between the preprocessed and impregnated specimens, suggesting that an appropriate representation of the marrow had been achieved (Fig. 6 and 7).

This is important, as these models can be used as referenced specimens when working with those that are osteoporotic. The prepared models can be easily re-scanned at different intensities, across different modalities, and over extended periods of time to serve as standards for future work.  These specimens have other advantages when compared with using fresh tissue that would move between multiple freeze/thaw sessions to accomplish the same re-sampling procedures.  In addition, the freeze/thaw cycles required for using the specimens in the intended purpose may provide inferior information about the trabecular bone structure, particularly when assessed through nuclear magnetic resonance spectroscopy (Prantner et al., 2010).  Moreover, temperature differences in scanned specimens can influence bone mineral densitometry (Whitehouse et al., 1993), thus introducing a potential compromise in data reliability.  Temperature and storage concerns are largely negated with the current models stored at room temperature. In addition to superiority of these specimens in issues of storage, the current specimens are easily transportable, facilitating scanning identical anatomy across multiple machines and locations.  These observations, coupled with the ease of storage and transportability, suggest strong potential for using the specimens to investigate the trabecular structure using µCT and MDCT to begin assessing osteoporotic samples. Further studies on the implications for finite element analysis and assessment are required.

Unlike the paraffin specimens, those impregnated with silicone or gelatin were not effective in providing a stable matrix to protect the trabecular architecture.  The voids observed in the silicone specimens were likely due to incomplete penetrance of the curing agent, impingement, or silicone leakage.  Complete filling to observe the trabecular architecture can be accomplished through different polymers and approaches which include different resins and sectioning (Jiang et al., 2021; Sun et al., 2021).  However, these approaches would not be effective in expanding the goals of the project since full (non-sectioned) specimens more appropriately reflect the clinical context. CT measures were higher in the plastinated specimens, confirming observations in the work of others (Shanthi et al., 2015). Like the challenges associated with silicone, the large tibial samples in this work presented significant challenges in formulating a gelatin matrix that provided the requisite stability and representation of the marrow.  Previous studies have used bone phantoms with a gelatin matrix impregnated into small samples for magnetic resonance imaging with very positive results (Gee et al., 2015).  Those phantoms demonstrated an effective marrow matrix replica and were sustained through testing.  However, the phantom bone cores were considerably smaller in volume (45.8 cm³) as compared to the specimens used in the current study, estimated as a truncated cone (407.3 cm³).  This order of magnitude difference in volume likely magnified the issues addressed in Gee et al. (2015), particularly with void filling and matrix stability.

Overall, the paraffin impregnated specimens demonstrated agreement in MCDT bone volume fraction (r = 0.92) and trabecular network area (r = 0.94) measures with the native tissue.  These results were more robust in the µCT data (bone volume fraction, r = 0.97; and trabecular network area, r = 0.99).  Additionally, of the impregnating procedures, paraffin wax provided the best qualitative specimens for handling, storage, and processing time.  Moreover, these specimens may be used to explore the microstructure of trabecular rod and plate construct across multiple imaging modalities.

Limitations.  The major limitation to the present study is related to specimen age.  Since aged bone typically demonstrates decreased values in BVF and altered trabecular structure they may not represent perfectly healthy tissue.  Yet, even with this limitation the goal of demonstrating preserved trabecular structure was demonstrated.

CONCLUSION

These data suggest that paraffin demonstrates a space occupying matrix that effectively protects the trabecular network and maintains the ability to quantify important biomechanical measures for assessing bone integrity in the study of osteoporosis.

Ethics Statement

Specimens used in this study were obtained from the institution’s Deeded Body Program and were ethically used to advance the research mission.  The use of cadaveric specimens in research is exempt from requiring authorization from the Institutional Review Board.

Funding

Portions of this study were funded by NIH grant R01 HL142042

Data Sharing

The data are not publicly available due to privacy or ethical restrictions. Data are available by contacting the authors.

Acknowledgements

The authors sincerely thank those who have donated their bodies for education and research purposes.  Those individuals, and their families, make possible the training/research that impacts our overall knowledge and may influence patient care and humankind. We respectfully offer our most sincere gratitude.