Patient-specific mechanical analysis of pedicle screw insertion in simulated osteoporotic spinal bone models derived from medical images
Article information
Abstract
Study Design
Biomechanical study.
Purpose
To investigate the mechanical characteristics of bone models created from medical images.
Overview of Literature
Recent advancements in three-dimensional (3D) printing technology have affected its application in surgery. However, a notable gap exists in the analyses of how patient’s dimorphism and variations in vertebral body anatomy influence the maximum insertional torque (MIT) and pullout strength (POS) of pedicle screws (PS) in osteoporotic vertebral bone models derived from medical images.
Methods
Male and female patients with computed tomography data were selected. Dimensions of the first thoracic (T1), fourth lumbar (L4), and fifth lumbar (L5) vertebrae were measured, and bone models consisting of the cancellous and cortical bones made from polyurethane foam were created. PS with diameters of 4.5 mm, 5.5 mm, and 6.5 mm were used. T1 PS were 25 mm long, and L4 and L5 PS were 40 mm long. The bone models were secured with cement, and the MIT was measured using a calibrated torque wrench. After MIT testing, the PS head was attached to the machine’s crosshead. POS was then calculated at a crosshead speed of 5 mm/min until failure.
Results
The L4 and L5 were notably larger in female bone models, whereas the T1 vertebra was larger in male bone models. Consequently, the MIT and POS for L4 and L5 were higher in female bone models across all PS diameters than in male bone models. Conversely, the MIT for T1 was higher in male bone models across all PS; however, no significant differences were observed in the POS values for T1 between sexes.
Conclusions
The mechanical properties of the proposed bone models can vary based on the vertebral structure and size. For accurate 3D surgical and mechanical simulations in the creation of custom-made medical devices, bone models must be constructed from patient-specific medical images.
Introduction
Recent advancements in three-dimensional (3D) printing technology have revolutionized its application in spine surgery [1]. Advanced 3D bone models have transitioned from educational aids to critical tools in preoperative planning for spinal deformities; however, their utility extends beyond mere anatomical replication [2–4]. These models are derived from patient-specific medical images and offer surgeons and trainees a tangible method to refine surgical techniques, particularly in cadavers and scarce-space environments [5–8]. To effectively utilize these models for pre- and postoperative surgical simulations, their mechanical properties must align closely with those of patients. Current rigid bone models fall short in simulating the nuanced mechanical challenges of surgery on patients with osteoporosis.
Despite in vitro studies of bone models examining mechanical aspects, such as maximum insertional torque (MIT) and pullout strength (POS), in relation to variables, such as pedicle screw (PS) length and insertion angle [9], no studies have focused on simulated osteoporotic bone models based on patient-specific imaging. Consequently, this limits our understanding of how PS performance varies among osteoporotic vertebral anatomy.
In this study, we hypothesized that even using consistent manufacturing methods, the torque and POS of PS will vary based on the shape and size of the vertebral origin of the bone models. By assessing these mechanical discrepancies in osteoporotic spine bone models created by 3D printing, this study aimed to aid in the future design of personalized treatment approaches and medical devices.
Materials and Methods
Study design
This study was approved by the ethics committee of the Yamaguchi University Hospital, and written informed consent was obtained from the patients (H28-054).
This study was conducted in collaboration with TANAC Corporation (Motomachi, Gifu, Japan). Computed tomography (CT) images of male and female patients, reflecting average heights in Japan and without deformities, were selected.
The pedicle width was measured in axial and sagittal CT of the first thoracic (T1) to the fifth lumbar (L5) vertebrae in two patients, and the vertebrae in which a 6.5-mm PS could be inserted in either the left or right pedicle width were selected. As a result, T1, fourth lumbar (L4), and L5 vertebrae were selected. Measurements included anterior, middle, and posterior vertebral heights, anterior–posterior and transverse diameters at the vertebral center, and maximum pedicle width in axial and sagittal CT images (Fig. 1A–C, Table 1). The maximum PS length was established from the CT images, indicating that the PS can be inserted up to 25 mm in T1 and 40 mm in both L4 and L5.
The vertebrae were measured: (A) the anterior, middle, and posterior vertebral height, (B) the anterior–posterior diameter and transverse width of vertebrae, and pedicle width of axial and (C) sagittal view of computer tomography. (D) The bone model of the first thoracic (T1), fourth lumbar (L4), and fifth lumbar (L5) vertebrae.
Data from the first thoracic vertebra and fourth/fifth lumbar vertebrae
Stereolithography data construction
Stereolithography was performed using Simpleware ScanIP (Synopsys Inc., Sunnyvale, CA, USA). After the extraction of the spine, the vertebrae were mapped as cancellous and cortical bones. The 3D bone models of T1, L4, and L5 were constructed. The bone models (TANAC BONE, TANAC Corp.) were created by casting in a mold using a master model from a 3D printer based on CT [10]. The model consisted of two layers, representing cancellous and cortical bones made of polyurethane foam, with varying compositions and densities to simulate osteoporosis (Fig. 1D).
Maximum insertional torque and pullout strength testing
The PS (ARMADA) used in this study were developed by NuVasive (San Diego, CA, USA), and PS with diameters of 4.5 mm, 5.5 mm, and 6.5 mm were used. PS insertion was planned from the base of the transverse process through the center of the pedicle. Initially, a puncture point was made using a needle, and a 25-mm bone hole for T1 and 40-mm holes for L4 and L5 were drilled with an awl. Before PS insertion, tapping was performed using a drill size 1 mm smaller than the PS diameter (for the 4.5-mm PS, the thinnest 4.0-mm tap was used) (Fig. 2).
(A) The needle to create a puncture point. (B) Bone hole was created with an awl. (C) The tapping.
So that the longitudinal axis of the PS was in the vertical direction, the vertebral portion of the bone models was fixed in a polyvinyl chloride pipe by pouring cement into the gap resin. A 3D-CAD (computer-aided design) was leveraged to create an outer frame on the polyvinyl chloride pipe so that the bone hole was perpendicular to the floor. The bone model was fixed through the guide (Fig. 3). Because the transverse process of the vertebral spine might obstruct the screw head, leading to high torque values, bone rongeurs were employed to resect the area where the screw head was anticipated to make contact.
(A, B) The vertebral part was fixed in a polyvinyl chloride pipe. Cement was poured into the resin such that the longitudinal axis of the pedicle screws was in the vertical direction.
The screwdriver lacked torque measurement capabilities, so a strain gauge was utilized. Because of space constraints, an adapter was fabricated to mount the strain gauge at a 45° angle to the axial direction. Torque tests were conducted using a sensor interface (PCD-320A; Kyowa Electronic Instruments Co., Tokyo, Japan) with measurements taken on a dynamic strain meter (DPM-713B, Kyowa Electronic Instruments Co.) (Fig. 4). The insertional torque was measured using a calibrated torque wrench with a specially designed connector.
(A) The adapter attached a strain cage and a lead for measurement. (B) The screw insertion and (C) the screw insertion location (L4 and L5 cross-section).
The MIT was calculated using strain measured by the dynamic strain meter with the following equation:
d: adapter diameter of 5.4 mm
E: Young’s modulus of 200 GPa
ν: Poisson’s ratio of 0.3
Pullout strength
The polyvinyl chloride pipe fixed the bone model with the cement and was set on the load tester (EZ-LX; Shimadzu Manufacturing Co., Kyoto, Japan), and the multiaxial head of the PS was connected to the machine’s crosshead through a titanium alloy rod. A vertical pullout load was applied to the bone models at a crosshead speed of 5 mm/min until failure at which the load reached the highest value. The POS was defined as the highest pullout load for evaluating the mechanical integrity at the bone–screw interface (Fig. 5). Five of each PS type was tested in each model, and the MIT and POS were calculated on a total of 90 bone models.
Pullout strength test. (A, B) The polyvinyl chloride pipe apparatus was set at the load tester, and pedicle screw was connected to the machine’s crosshead. (C) Vertical pull-out load was applied at a crosshead speed of 5 mm/min.
Statistical analysis
A one-way analysis of variance was conducted to assess the effects of torque and pullout force on the PS diameter and individual variances across the three vertebrae in the male and female bone models. StatFlex software ver. 6.0 (Artec Co. Ltd., Osaka, Japan) was used for statistical analysis. All p-values <0.05 were considered significant.
Results
Maximum insertional torque
L4
The MIT (Ncm) of the male L4 vertebral bone models were 78.4 (standard deviation [SD]=10.9) for the 4.5-mm PS, 109.7 (SD=32.4) for the 5.5-mm PS, and 125.9 (SD=29.5) for the 6.5-mm PS, whereas those of the female L4 vertebral bone models were 64.0 (SD=8.4) for the 4.5-mm PS, 81.1 (SD=17.9) for the 5.5-mm PS, and 94.9 (SD=16.0) for the 6.5-mm PS (Fig. 6A).
(A) The maximum insertional torque (MIT) of L4. Bars represent the mean±standard deviation. (B) MIT of L5. (C) MIT of T1. The vertical axis represents torque (Ncm) and the horizontal axis indicates sex and pedicle screw (PS) diameter for all graphs. *p<0.05. **p<0.01. ***p<0.001.
L5
The MIT (Ncm) of the male L5 vertebral bone models were 81.9 (SD=18.1) for the 4.5-mm PS, 100.7 (SD=22.5) for the 5.5-mm PS, and 132.7 (SD=29.0) for the 6.5-mm PS, whereas those of the female L5 vertebral bone models were 54.7 (SD=11.2) for the 4.5-mm PS, 96.7 (SD=20.5) for the 5.5-mm PS, and 124.1 (SD=24.4) for the 6.5-mm PS (Fig. 6B).
T1
The MIT (Ncm) of the male T1 vertebral bone models were 88.3 (SD=10.7) for the 4.5-mm PS, 99.3 (SD=18.7) for the 5.5-mm PS, and 136.3 (SD=20.5) the for 6.5-mm PS, whereas those of female T1 vertebral bone models were 102.1 (SD=20.6) for the 4.5-mm PS, 140.3 (SD=45.4) for the 5.5-mm PS, and 169.8 (SD=44.2) for the 6.5-mm PS (Fig. 6C).
Pullout strength
L4
The POS (N) of male L4 vertebral bone models were 626.8 (SD=95.7) for the 4.5-mm PS, 1,039.5 (SD=113.3) for the 5.5-mm PS, and 1,066.6 (SD=80.8) for the 6.5-mm PS, whereas those of female L4 vertebral bone models were 548.1 (SD=107.1) for the 4.5-mm PS, 846.5 (SD=147.2) for the 5.5-mm PS, and 829.5 (SD=174.2) for the 6.5-mm PS (Fig. 7A).
(A) Pullout strength (POS) of L4. Bars represent the mean±standard deviation. (B) POS of L5. (C) POS of T1. The vertical axis represents load (N) and the horizontal axis indicates sex and pedicle screw (PS) diameter for all graphs. *p<0.05. **p<0.01. ***p<0.001.
L5
The POS (N) of male L5 vertebral bone models were 552.2 (SD: 51.8) for the 4.5-mm PS, 844.7 (SD: 83.1) for the 5.5-mm PS, and 993.9 (SD: 223.8) for the 6.5-mm PS, whereas those of female L5 vertebral bone models were 462.7 (SD: 68.9) for the 4.5-mm PS, 827.1 (SD: 137.6) for the 5.5-mm PS, and 938.6 (SD: 93.4) for the 6.5-mm PS (Fig. 7B).
T1
The POS (N) of male T1 vertebral bone models were 789.7 (SD=117.5) for the 4.5-mm PS, 685.8 (SD=103.6) for the 5.5-mm PS, and 803.4 (SD=97.2) for the 6.5-mm PS, whereas those of female T1 vertebral bone models were 799.2 (SD=103.4) for the 4.5-mm PS, 886.1 (SD=74.5) for the 5.5-mm PS, and 896.1 (SD=220.2) for the 6.5-mm PS. No significant differences were found between bone models or in PS widths (Fig. 7C).
Discussion
PS fixation is a widely utilized surgical technique for restoring spine stability. In designing PS, a study showed that the MIT and POS are interrelated [11]. The optimal PS design should maximize the diameter and length while ensuring that they do not breach the cortical layer of the pedicle or vertebral body [12]. This is crucial because fixation heavily relies on the pedicle’s dense structure, with the pedicle region contributing to 60% of the resistance against PS pullout and 80% of torsional stability [13,14]. Accordingly, the use of larger-diameter PS can achieve contact with the subcortical bone within the pedicle [15]. Matsukawa found that better PS fit and deeper engagement with the peripheral bone in the pedicle improved the fixation of interbody fusion [16]. Conversely, longer PS could achieve improved fixation by increasing the contact with the cancellous bone in the vertebral body, which is an anatomically lower-density region [14]. In this study, PS with larger diameters yielded similar results for both the MIT and POS. Regarding the POS of T1, PS fitness exceeded 70% even with a 4.5-mm PS relative to the pedicle diameter, potentially risking pedicle damage.
Although few biomechanics studies are measuring the MIT and POS, some clinical studies offer insights. Okuyama et al. [17] reported an average MIT of 128±37 Ncm in cases with PS loosening versus 150±40 Ncm in cases without, although the PS diameter and length were not specified. Inoue et al. [18] divided patients into two groups based on preoperative teriparatide treatment. The MIT of the teriparatide group was 128±42 Ncm, which was significantly higher than that of the control group. Our results align with the findings by Inoue et al. [18] that thicker PS diameter yields higher MIT. However, this is an actual patient, and the vertebral body size and other factors (height) differ. The MIT values in our osteoporotic bone models were like these clinical data, suggesting that the model’s quality was comparable and suitable for simulating clinical conditions.
In the analysis of the thoracic vertebrae, Tsai et al. [19] evaluated the difference between 4.5-mm and 5.5-mm diameter PS inserted into the upper thoracic vertebra cadavers. They reported that the MIT for the 5.5-mm diameter PS was 59% greater than for the 4.5-mm diameter PS. A similar trend was observed for T1, particularly in female bone models. The absence of significant MIT differences between the 4.5-mm and 5.5-mm PS in male bone models could be attributed to the size of the vertebral body, although this factor was not specified in the study by Tsai et al. [19].
In clinical practice, POS evaluation is challenging because it requires a pullout situation post-PS insertion. As with the MIT, POS data for various clinical cases are scarce. Karataglis et al. [20] used cadavers to evaluate the role of the dorsal vertebral cortex during PS use. They found that preserving the dorsal cortex increased the mean POS to 1,295 N, compared with 956 N when the cortex was removed. In our analysis, we preserved the dorsal cortex, observing similar values for the 5.5- and 6.5-mm PS but lower values for 4.5-mm PS. Heller et al. [21] measured the POS in the upper thoracic spine of cadavers and reported 775 N as the mean POS of T1, which was higher than those of T2–T4. The diameter used of the PS used in this study was 3.5 mm, which may have resulted in POS slightly higher than our results. Shibasaki et al. [9] conducted tests on osteoporotic lumbar bone models to measure the MIT and POS for PS inserted in various directions. The POS values ranged from 700 to 980 (N). Our 4.5-mm PS showed lower POS, whereas those of 5.5- and 6.5-mm PS were comparable to their data. This indicates that the PS diameter’s occupancy rate relative to the pedicle remains a crucial factor.
Previous studies have delved into how variations in PS diameter, length, and insertion direction affect spine mechanics. Notably, the work by Burkhard et al. [22] involving a 3D-printed spine model based on medical images demonstrated that MIT and related metrics are influenced by the intricate internal structures visible in CT. However, their analysis was limited to a single vertebra, without broader examinations across multiple cases and vertebrae [22]. Interestingly, this study examined patient-specific differences. The male bone models showed higher MIT at L4 and L5, whereas the female models had higher MIT at T1. This discrepancy can be attributed to the larger vertebral sizes in female L4 and L5 models and male T1 models. Surprisingly, actual measurements showed a difference contrary to the assumption that the vertebrae are uniformly larger in taller individuals. Consequently, the larger cancellous bone in L4 and L5 for women and T1 for men suggested that the PS placed in these larger vertebrae resulted in lower occupancy rates and smaller values. Variations in T1 POS data may be caused by manually creating a bone hole in the narrow pedicle and the oblique direction of the T1 PS affecting the pullout direction. Nevertheless, obtaining data from medical images under identical manufacturing conditions is insightful. Accordingly, differentiating between the cortical and cancellous bones in these models could yield more precise data.
This study has some limitations. First, the length and method of PS insertion were constant, which may not fully represent clinical variability. The bone models, which were designed to mimic osteoporosis, possessed uniform hardness; variations in internal structures such as density could alter the POS and MIT. Thus, stiffness variations of 3D models such as cadaver and clinical data for future use in clinically relevant surgical procedures and preoperative planning are necessary. The specific design of the inserted PS might influence the data. The study was also based on just one male and one female patient; future studies should include bone models derived from a broader range of medical images.
Despite these limitations, to the best of our knowledge, this is the first study to report mechanical properties using bone models derived from medical images of different patients. To enhance the clinical simulations or development of medical devices for specific pathological conditions, future studies should use bone models that consider patient-specific bone morphology and strength of the diseased bone for greater accuracy.
Conclusions
The mechanical properties of the bone models vary based on the structure and size of the vertebrae, even when created using similar methods. For accurate 3D surgical and mechanical simulations in the creation of custom-made medical devices, bone models must be constructed from patient-specific medical images.
Key Points
This study analyzed the mechanical properties of bone models constructed from medical images.
The mechanical properties of the bone models can vary based on the structure and size of their vertebral origin, even when created using similar methods.
For accurate three-dimensional surgical and mechanical simulations in the creation of custom-made medical devices, bone models must be constructed from patient-specific medical images.
Acknowledgments
We thank NuVasive for providing the surgical instruments necessary to conduct this study. TANAC Corporation provided $5,000 to cover joint 2022–2023 research expenses for this study.
Notes
Conflict of Interest
TANAC Corporation provided $5,000 to cover joint 2022–2023 research expenses for this study. Except for that, no potential conflict of interest relevant to this article was reported.
Author contributions
Conceptualization: NN, JO. Methodology: NN, HT, YM, YI, JO. Formal analysis: NN, YM. Data curation: HT, YM, FJ, MF, KF, JO. Writing–original draft: NN, JO. Writing–review & editing: HS, YK. Supervisor: TS. Final approval of the manuscript: all authors.