Cortical bone trajectory (CBT) is a technique performed to enhance screw–bone purchase in the lumbar spine [1,2]. The screws follow a caudocranial trajectory in the sagittal plane and are directed more laterally in the transverse plane [1]. It is considered less invasive, necessitating a smaller incision, less lateral tissue dissection, and minimal bleeding [3,4]. In clinical studies, the notable pullout strength in CBT is thought to result from an increased screw–cortical bone contact, resulting in lower rates of screw loosening and less frequent loss of correction [5]. Biomechanical studies revealed that CBT has a greater range of the stiff zone during axial rotation and a higher resistance to toggle testing and resistance force than traditional pedicle screws [6,7].

A consensus Delphi study of 74 spine surgeons presented the major advantages of CBT, including a smaller exposure range, minimized intraoperative traction of the paravertebral soft tissue, and fewer neurological complications [8]. However, the major disadvantages include the difficulty in determining the path of screw placement during surgery and the lack of three-column fixation [8]. Although crossing the neurocentral junction with 80% penetration depth of the vertebral body is recommended in vertebral instrumentation, CBT fixation is restricted to the posterior half to the posterior one-third of the vertebral body [9,10]. Krag [11] noted that pedicle screw penetration of 80% is 32.5% stronger than with 50% penetration of the vertebral body.

Therefore, a modification was made to the entry point of the CBT trajectory in Dhulikhel Hospital/Kathmandu University Hospital. The modified cortical bone trajectory (MCBT) was hypothesized to have the advantage of longer screw placement, higher cortical purchase than the original pedicle trajectory (OPT), and a near-equivalent purchase compared with conventional CBT. The bone density along the four different points of OPT, MCBT, and CBT was assessed using the computed tomography Hounsfield unit (CTHU), providing a different insight into the CTHUs in each of the three trajectories.

This cross-sectional study was conducted in Dhulikhel Hospital/Kathmandu University Hospital. Ethical clearance was obtained from the Institutional Review Committee (IRC) of the Kathmandu University School of Medical Sciences (KUSMS) (IRC-KUSMS approval number: 184/2024). A thousand vertebrae of 200 consecutive individuals aged 18–75 years presenting with nonspinal pathologies who underwent abdominal computed tomography (CT) (SOMATOM Perspective 128 eco by Siemens with 5-mm transverse slice, reconstruction into 1-mm slice using “Syngovia” software; Siemens Healthineers, Forchheim, Germany) between March 2024 and July 2024 were selected for the study. Individuals with spinal pathologies such as vertebral fractures, tumors, metastasis, and insufficiency fractures, previous spine surgery, and infections were excluded.

Informed consent was not obtained because the CT scans were retrospectively selected from the hospital’s electronic data recordings without revealing any individual identification. Given the lack of variations in the CTHU of CT scans windowed for examinations of the abdomen, pelvis, or spine, an abdominal CT scan was selected for this study [12]. Similarly, CTHU was used as a surrogate marker of bone density because it has shown a stronger correlation in dual-energy X-ray absorptiometry as a better predictor of screw loosening and pullout strength [13–15].

For each individual, the CT transverse section of the L1–L5 vertebrae was cut into three planes. (1) For the OPT, the cut plane was horizontal to the pedicle, representing the plane of the pedicle screw insertion (Fig. 1). (2) For CBT, the cut was in a more caudocranial plane starting from a point tangential to the inferior part of the pedicle (5 o’clock for the right and 7 o’clock for the left pedicles in the coronal plane) and targeting the midpoint of the superior endplate representing the plane for CBT (Fig. 2). (3) For MCBT, the cut was in a caudocranial plane starting from a point tangential to the inferior part of the pedicle (6 o’clock of the pedicle in the coronal plane) targeting the anterosuperior junction of the vertebral body representing the plane at which the pedicle screws are inserted using MCBT (Fig. 3).

For each cut, a linear line was plotted crossing the midpoint of the narrowest zone of the trajectory at the pedicle. The CTHU of four points on this linear line—posterior cortex (PC), mid-pedicle (MP), midbody (MB), and anterior cortex (AC)—were selected to compare the bone density within the area of screw insertion (Figs. 1C, 2C, and 3C). Contrary to other studies where CTHU is measured in the region of interest (ROI) along different trajectories, the present study measures CTHU in four points, allowing individual point comparison and cumulative CTHUs along the trajectories. The CTHU ratio between MCBT and OPT, measured as specific value 1 (SV1), and the CTHU ratio between CBT and OPT, measured as SV2, were calculated and compared with age.

The collected data were analyzed using IBM SPSS ver. 25.0 for Windows (IBM Corp., Armonk, NY, USA). One-way analysis of variance (ANOVA) was performed to compare the means of CTHU in the three groups for each vertebral level. Univariate linear regression was performed between the average CTHU of each vertebra in all three trajectories and patient age. Pearson correlation was done to evaluate the correlation between age and specific values (SV1 and SV2). A value of p<0.05 was considered significant.

The mean age of the patients was 40.57±13.33 years, with predominance of the patients aged 31–45 years (36.1%) followed by those aged 46–60 years (29.4%). The majority of the patients were male (54%). The mean CTHUs of OPT, MCBT, and CBT were 354.2±70 HU, 457.3±90 HU, and 529.9±75 HU, respectively (Fig. 4), and the difference was significant (p<0.01).

The mean CTHU of MCBT was 29% higher than that of OPT and 13.7% lower than that of CBT. The comparison of the mean CTHUs in PC, MP, MB, and AC in L1–L5 vertebra is shown in Fig. 5. The difference in the CTHUs for the PC of all three trajectories (OPT, 573.9±169 HU; MCBT, 948.8±199 HU; and CBT, 1,116.6±149 HU) was highly significant compared with the other three points (MP, MB, and AC) (p<0.001). The CTHU in the PC in MCBT was 65.6% higher than that in OPT and 14.9% lower than that in CBT. The difference in CTHUs in the MP between MCBT and CBT was not significantly different (average CTHU of MCBT, 305±145 HU; CBT, 302±125 HU). The mean CTHUs of the upper lumbar vertebra (L1 and L2) were 343.6±62.1 HU, 433.47±85.4 HU, and 512.12±72.3 HU for OPT, MCBT, and CBT, respectively. Similarly, the mean CTHUs of the lower lumbar vertebra (L3–L5) were 361.3±74.7 HU, 473.12±89.1 HU, and 541.73±75.6 HU for OPT, MCBT, and CBT, respectively.

The ANOVA showed a significant difference between the mean CTHU of OPT when compared with sex (male, 350.1±60.6 HU; female, 359.1±80.3 HU; p<0.05) but not significant with MCBT (male, 457.6±85.1 HU; female, 456.9±95 HU; p=0.9) and CBT (male, 530.6±71.3 HU; female, 529±80.5 HU; p=0.74). Similarly, a significant difference in the mean CTHU of all three trajectories was found when compared with patients aged <60 years and >60 years (p<0.01). However, the percentage drop in the mean CTHU in OPT in patients aged >60 years was 19%, which was higher than those in MCBT (7.4%) and CBT (4.5%). The univariate regression of the CTHUs in the three trajectories with age showed a comparable decline of the CTHU with age in CBT and MCBT (Pearson’s r: −0.20 versus −0.22; adjusted R²: 0.040 versus 0.047). However, a higher decline in CTHU with age was noted in OPT (Pearson’s r=−0.38, adjusted R²=0.14) (Fig. 6).

The average specific value between MCBT and OPT (SV1) was 1.31±0.25. Similarly, the average specific value between CBT and OPT (SV2) was 1.53±0.29. SV1 and SV2 showed a significant positive correlation with age (Pearson coefficient=0.2, p<0.01 for SV1 and age; Pearson coefficient=0.3, p<0.01 for SV2 and age).

A significant mean difference in CTHUs in the L1–L5 vertebrae (p<0.01) was noted, except for the CTHUs in the MP in MCBT (p=0.057) and AC in CBT (p=0.422). A significant negative correlation of CTHU at each measured point was noted in all three trajectories (except for the CTHU of the AC in MCBT and PC in CBT) with the age of the patients (Table 1).

In this study, the mean CTHUs were 354.2±70 HU, 457.3±90 HU, and 529.9±7 HU for OPT, MCBT, and CBT, respectively. The mean CTHU in MCBT was 29% higher than that in OPT and 13.7% lower than that in CBT. Among the four points, PC had the highest mean CTHU, followed by AC in all three trajectories. Although piercing the AC in OPT is not commonly practiced, AC was included in our measurements with the assumption of bicortical screw purchase. Despite including the AC (which is the second dense point in the trajectory), the mean CTHU in OPT was still significantly lower than that in MCBT, which further explains the superiority of MCBT over OPT. MCBT is comparable with CBT in terms of age, sex, and individual points along the trajectories. Among the four points, the CTHU in PC had a huge contribution to the average CTHUs in all trajectories, followed by the CTHU in AC. However, the mean CTHU in MCBT (480.04 HU), excluding AC, was still higher than the mean CTHU in OPT (354.2 HU) with the inclusion of AC, showing the superiority of MCBT over OPT.

A significant linear correlation exists between increasing age and decreasing BMD in the lumbar mid-vertebral body [12]. Although the CTHUs of OPT, MCBT, and CBT were inversely related to age, SV1 (MCBT:OPT) and SV2 (CBT:OPT) increased with age. Similarly, the univariate regression showed that the CTHUs of CBT and MCBT had a comparable decline with age (Pearson’s coefficient of approximately −0.2). However, a higher decline in the CTHU (Pearson’s coefficient of approximately −0.4) with age was noted in OPT. This could be because of a more disproportionate involvement of the trabecular bone than the cortical bone with increasing age [16,17]. Thus, CBT and MCBT, which have more contact with the cortical bone than OPT, are less affected by age and, therefore, better suitable options for older people [18].

In OPT, a significant difference in the CTHU was found between male and female patients. Still, the difference was not significant for MCBT and CBT despite a negative correlation of CTHU with age in both trajectories. The difference in CTHU between patients aged <60 years and >60 years was significant for OPT, MCBT, and CBT in the present study, which was similar to the findings by Kojima et al. [1] and Zhang et al. [17]. However, the decline rate in the mean CTHU in OPT was much higher than those in MCBT and CBT in patients aged >60 years, indicating that CBT and MCBT are less influenced by age than OPT. In the study by Kojima et al. [1], the subgroup of the age category was >70 years and <70 years, where women aged >70 years presented the lowest mean CTHUs for CBT and OPT. However, in the study by Zhang et al. [17], the age category was 10 years, and the difference in the CTHU across all age groups significantly decreased with age in both CBT and OPT. Mai et al. [19] concluded that the osteoporosis cohort demonstrated a significantly greater difference in CTHUs between the fixation points in CBT and OPT than that of the nonosteoporotic cohort at all lumbar levels. A study also determined that the relative differences in CTHUs significantly increased with each decade of age at each lumbar spinal level, which is similar to the findings of the present study. Furthermore, the average CTHUs of the lower lumbar vertebrae (L3–L5) were larger than those of the upper lumbar vertebrae (L1 and L2) in all trajectories, which might be because the caudal lumbar vertebrae mostly bear the weight of activities, resulting in higher CTHUs.

Some studies have evaluated CTHU in different trajectories for pedicle screw insertion. In the study by Matsukawa et al. [20], three-dimensional finite-element (FE) models of 20 L4 vertebrae were constructed from CT data and five different trajectories; that is, the traditional, vertical, and three lateral trajectories with different sagittal directions (caudal, parallel, and cranial) were evaluated for the pullout strength with nonlinear FE analyses. They concluded that regional variations in the vertebral bone density and the amount of the denser bone–screw interface contribute to the differences in stiffness among different screw trajectories. Similarly, Liu et al. [21] showed that the modification in the trajectory of conventional CBT had better CTHUs in the ROI and could accommodate longer screws, which could contribute to superior biomechanical properties than conventional CBT. This study also demonstrated a regional variation in bone density with the change in screw trajectory, with the CBT trajectory being the densest, followed by the MCBT and OPT trajectories.

At our center, the above morphometric findings were preliminarily tested in >50 patients with fractures of the lumbar vertebra using MCBT trajectory. The entry point in MCBT is just beneath the superior facet on the pars interarticularis (the intersection point of a horizontal line passing through the inferior border of the pedicle and a vertical line bisecting the pedicle in true anteroposterior view of the vertebra), and the trajectory passes tangential to the inferior border of the pedicle, which is also rich in cortical bone, aiming at the junction between the superior endplate and the anterior vertebral body (Fig. 3). The average length of the screws in our series for the lumbar vertebra in MCBT was 45 mm, which was longer than that in CBT (38.48 mm) measured in the CT scan by Matsukawa et al. [22]. Although further morphological study is warranted, our series showed that MCBT can accommodate longer screws than CBT. In their biomechanical evaluation of the fixation strength among different sizes of pedicle screws using the CBT, Matsukawa et al. [23] revealed that longer screws significantly increased the pullout strength and vertebral fixation strength in transverse rotation. These results signify the advantage of MCBT for the lumbar spine, allowing longer screw fixation. To further support our findings, we plan to evaluate the morphological parameters of MCBT and the insertional torque while placing screws in these trajectories in our next study.

The strengths and limitations of this study are as follows. This study used a modified technique for comparing the CTHUs at four points of three trajectories for the quantitative analysis of a large sample (1,000 vertebrae from 200 patients), which enhanced the reliability of the results. MCBT showed higher CTHUs than OPT, indicating improved screw–bone contact and fixation strength, which offers a practical alternative for lumbar fixation with the potential for longer screw insertion. However, this study has a single-center setting that potentially limits the generalizability of the findings; thus, clinical implication studies with follow-ups, including complications, screw loosening, and facet joint violation, are needed.

In this study, MCBT shows a significantly higher CTHU than OPT, indicating better bone density. The CTHU in the PC, which serves as an entry point and is critical in the screw trajectory for both MCBT and CBT, is comparable in both techniques. Therefore, MCBT is not only superior to OPT but also can be an alternative to CBT for lumbar spine fixation, with the added advantage of longer pedicle screw insertion without compromising the advantages of minimal soft tissue dissection and reduced bleeding.