Dimensional Changes of Lumbar Intervertebral Foramen in Direct Anterior Approach‐Specific Hyperextension Supine Position

Objective To investigate the changes in the lumbar intervertebral foramen (LIVF) dimensions from neutral supine to direct anterior approach (DAA)‐specific hyperextension supine position through a standardized three‐dimensional (3D) reconstruction computerized tomography (CT) method. Methods A total of 35 healthy volunteers (18 men and 17 women) were included in this retrospective study. The mean age of enrolled subjects was 28.9 ± 5.0 years. In September 2018, all the individuals underwent a 3D CT scan of the lumbar spine in neutral and 30° hyperextension supine positions, which mimicked the patient’s position in DAA total hip arthroplasty (THA). The dimensions of the LIVF, including foraminal area, height, and width, were measured on 3D reconstructed CT models at all lumbar foraminal levels. Foraminal area was defined as the area bounded by the adjacent superior and inferior vertebral pedicles, the posterosuperior boundary of the inferior vertebral body, the surface of the intervertebral disc posteriorly, the posteroinferior boundary of the superior vertebral body, and the surface of the ligamentum flavum anteriorly. Foraminal height was defined as the longest distance between the border of the superior and the inferior pedicle. Foraminal width was defined as the shortest distance between the posteroinferior edge of the superior vertebrae and the opposing boundary. Subgroup analysis and multiple linear regression were used to evaluate the relationship between percentage changes of the LIVF dimensions and side, sex, and age. Results The LIVF dimensions varied significantly between the two positions at all levels (P < 0.05). From neutral to hyperextension supine position, the foraminal area reduced by 20.1% at lumbar 1–2 (L1–2), 22.6% at L2–3, 19.9% at L3–4, 18.1% at L4–5, and 12.0% at lumbar 5–sacral 1 (L5–S1) level, respectively; the foraminal height reduced by 9.5% at L1–2, 10.5% at L2–3, 9.5% at L3–4, 9.6% at L4–5, and 6.1% at L5–S1 level, respectively; the foraminal width reduced by 12.8% at L1–2, 14.5% at L2–3, 13.0% at L3–4, 10.4% at L4–5, and 8.4% at L5–S1 level, respectively. The changes in LIVF dimensions were biggest at L2–3 level and smallest at L5–S1 level. Subgroup analysis showed that there were no significant differences in the percentage changes of LIVF dimensions between the sexes and sides (P > 0.05). Multiple linear analysis showed that the percentage changes of LIVF dimensions were not related to side, sex, and age (P > 0.05). Conclusion The dimensions of the LIVF showed significant decrease at all levels in the DAA‐specific hyperextension supine position compared with the neutral supine position, and the percentage changes of LIVF dimensions were not influenced by side, sex, and age.


Introduction
T otal hip arthroplasty (THA) is a reliable and effective surgical intervention for patients with end-stage hip joint disease 1,2 . The posterolateral approach, the lateral approach, and the direct anterior approach (DAA) are the most commonly used approaches during the operation 3,4 . DAA is a muscle-sparing method that provides a truly intranervous and intermuscular approach 5,6 . It has gained increasing popularity due to less soft tissue damage and shorter recovery time 3,4,7 . The exposure of the proximal femur is the key step in the DAA operation. Typically, the patient is placed in supine position and the hip is centered over the hinge of the operating table at the level of the anterior superior iliac spine (ASIS) 3 . The table can be extended at 30 during the operation to simplify the exposure of the femur 3,7 .
However, we questioned whether this procedure is harmless to the patients. In our experience, some patients have complained about lumbar pain after the DAA THA, although it was relieved later. A recent publication reported that the hyperextension impact could lead to a chance fracture. This indicates that position-related complications do exist in DAA THA 8 . We also wanted to find out whether this hyperextension position impacts the lumbar intervertebral foramen (LIVF), the nerve root outlet area. The boundary of the LIVF consists of adjacent superior and inferior vertebral pedicles, the posterosuperior margin of the inferior vertebral body, the intervertebral disc, the posteroinferior margin of the superior vertebral body, the superior and inferior articular facets and the ligamentum flavum [9][10][11] . Theoretically, the dimensions of the LIVF decrease from flexion to extension positions [11][12][13][14][15] . Extension can increase the facet joint movement and the bulging of the intervertebral disc and ligamentum flavum, resulting in more contact with the nerve root and a higher chance of compression or irritation of the nerve root 16 . Therefore, the specific position in the DAA THA might be a possible risk factor for nerve root injury, especially with prolonged operation time in a continuous hyperextension position.
Previous studies have elucidated the dimensional changes of LIVF from flexion to extension in standing or sitting position. Ren et al. 12 discovered that the foraminal area, height, and width decreased significantly at lumbar 1-5 (L 1-5 ) levels from neutral to extension standing position in patients with low back pain, while they did not find changes in LIVF dimensions at the lumbar 5-sacral 1 (L 5 -S 1 ) level. In another in vivo dynamic study, Zhong et al. 13 measured the LIVF dimensions from a flexion position of 45 to a maximal extension standing position with weight-lifting activity in asymptomatic volunteers. They found that the foraminal area and width decreased significantly at all lumbar foraminal levels except L 5 -S 1 , and that the foraminal height remained constant throughout the activity. However, another study by Singh et al. 11 compared the foraminal area under flexion standing, upright sitting, and extension standing positions in patients with low back pain. They found that the foraminal area increased significantly from flexion standing to upright sitting position at all lumbar foraminal levels except L 5 -S 1 but decreased significantly from upright sitting to extension standing position at all levels, including L 5 -S 1 .
In addition, they found that the decrease in the foraminal area was biggest at the L 2-3 level and smallest at the L 5 -S 1 level. Similarly, Schmid et al. 14 reported that the foraminal area decreased significantly from upright sitting to extension supine position in asymptomatic volunteers at all lumbar foraminal levels, including L 5 -S 1 . Furthermore, Schmid et al. 14 and Zamani et al. 15 both found a decrease in the foraminal area in extension sitting position and an increase in flexion sitting position at all levels. However, detailed information regarding the morphological changes of the LIVF A B from conventional neutral supine position to DAA-specific hyperextension supine position has not been reported.
Knowing the dimensional changes of LIVF from neutral to hyperextension supine position may provide valuable information to understand the potential nerve root complications after DAA THA. In this study, the LIVF dimensions from L 1-2 to L 5 -S 1 on both sides were measured with the three-dimensional (3D) reconstruction computed tomography (CT) images at a standardized mid-pedicle plane, which usually corresponds to the narrowest cross-section area of the LIVF 17,18 . The purpose of current study was: (i) to investigate the changes in the LIVF dimensions, including foraminal area, height, and width, from neutral to 30 hyperextension position; (ii) to investigate at which level the LIVF dimensions have the biggest and smallest changes; and (iii) to investigate the relation between the changes in LIVF dimensions and side, sex, and age.
We hypothesized that the LIVF dimensions would decrease with the hyperextension of the lumbar spine in supine position, and the changes in LIVF dimensions would not be influenced by side, sex, and age. principle: (i) Healthy subjects who has no previous spinal diseases and surgical history; (ii) underwent lumbar spine 3D CT in hyperextension position; (iii) underwent lumbar spine 3D CT in neutral supine position; (iv) foraminal area, height, and width at all lumbar foraminal levels; and (v) a retrospective study. Exclusion criteria were: (i) current or prior lower back pain or radiculopathy; (ii) history of spinal surgery; and (iii) anatomic abnormalities or other spinal disorders.

Ethical Approval
Our study was approved by the institutional review board of our institute, and informed consent was obtained from all participants.

Evaluation Method
The lumbar spine was scanned using multislice spiral CT with a slice thickness of 0.6 mm (Siemens, Germany) in September 2018. In all subjects, five foramina from L 1-2 to L 5 -S 1 were evaluated on both sides. The subjects were first scanned in conventional neutral supine position. Then we placed a position mat under the pelvis at the level of ASIS so that the hip was hyperextended at 30 during CT scanning, to mimic the hyperextension supine position in the DAA THA (Fig. 1). The 30 angle was determined by protractor.
All image data were saved in DICOM format, and were subsequently imported to Mimics 19.0 (Materialize, Leuven, Belgium) for preprocessing. Several previous studies have used 3D CT scans for the measurement of LIVF dimensions 9, 13, 17 . Among them, Rao et al. 17 proposed a pedicleto-pedicle method that could obtain a standardized snapshot of the foramen, and, thus, we adopted their method in our study. According to Rao's description, the special sagittal slice was aligned along the midline of superior and inferior pedicles and perpendicular to the disc space by the "reslice" function of mimics ( Fig. 2A,B). Then the dimensions of LIVF were obtained by the "measure" function of mimics.

Foraminal Area
The foraminal area was defined as the outline of the LIVF 10 . It is measured by the boundary of the adjacent superior and inferior vertebral pedicles, the posterosuperior portion of the inferior vertebral body, the posterior portion of the intervertebral disc, the posteroinferior portion of the superior vertebral body, and the anterior portion of ligamentum flavum 10 (Fig. 2C,D). Previous studies report that foraminal area narrowing can result in nerve root compression in the lumbar region 10,19 . We measured the foraminal area in neutral and hyperextension supine position, respectively, and the percentage changes of the foraminal area from neutral to hyperextension supine position were also calculated.

Foraminal Height
Foraminal height was defined as the longest distance between the boundary of the superior and inferior pedicle 10,20 . It was measured from the most inferior aspect of the upper pedicle to the most superior aspect of the lower pedicle 10,20 (Fig. 2C,  D). Because the anatomical morphology of the pedicle is fixed, the upper and lower movability range of the nerve root is determined by the distance between the adjacent pedicles 9 . We measured the foraminal height in neutral and hyperextension supine position, respectively, and the percentage changes of foraminal height from neutral to hyperextension supine position were also calculated.

Foraminal Width
Foraminal height was defined as the distance between the posteroinferior edge of the superior vertebrae and the anterior boundary of the superior articular process 13 . It was measured on the line through the posteroinferior corner of the superior vertebrae and vertical to the anterior surface of the opposing facet 13 (Fig. 2C,D). Because the anatomical morphology of the pedicle is fixed, the anterior and posterior movability range of the nerve root is determined by the distance between the posteroinferior edge of the superior vertebrae and the anterior boundary of the superior articular process 9 . We measured the foraminal width in neutral and hyperextension supine position, respectively, and the percentage changes of foraminal width from neutral to hyperextension supine position were also calculated.

Statistical Analysis
All parameters were expressed with mean ± SD. All statistical analysis was performed with SPSS 21.0 software (IBM, Armonk, USA). Two of the authors performed blinded measurements. One of the authors repeated measurements with 4-week intervals. The reliability of intrarater and interrater measurements were assessed using intraclass correlation coefficients (ICCs), which can be interpreted as: <0.40 poor; 0.40-0.59 fair; 0.60-0.74 good; 0.75-1.00 excellent 21 . The Wilcoxon signed-rank test was used to compare the LIVF dimensions measured in two positions. The paired t-test was used to compare the percentage changes of LIVF dimensions between right and left sides, and the independent t test was used to compare the percentage changes between sexes. Multiple linear regression was used to evaluate the relationship between the percentage changes of LIVF dimensions and the subjects' age and sexes. A P-value less than 0.05 was considered significant difference.

Reliability
The intraclass correlation was found to be 0.92 for foraminal area, 0.94 for foraminal height and 0.97 for foraminal width.
The interclass correlation was 0.88 for foraminal area, 0.90 for foraminal height, and 0.91 for foraminal width. All measurements showed excellent ICCs.
Relationship between Percentage Changes of the Lumbar Intervertebral Foramen Dimensions and Side, Sex, and Age No significant difference was found between the changes in LIVF dimensions (foraminal area, height, and width) and the side and sex (P > 0.05) (Tables 2 and 3). The overall regression analyses for all three measurements were found to be not statistically significant (P > 0.05) ( Table 4).

Discussion
I n the current study, we investigated the LIVF dimensions including foraminal area, height, and width on the standardized mid-pedicle slice and compared the parameters measured in conventional neutral supine and hyperextension supine position. The results confirmed our initial hypothesis that the LIVF dimensions were significantly decreased at all intervertebral foraminal levels with the hyperextension of the lumbar part. The changes in LIVF dimensions were largest at L 2-3 level, and smallest at L 5 -S 1 level. In addition, the percentage changes were not influenced by side, sex, and age.
Several in vivo studies have reported position-dependent morphologic changes of LIVF. Generally, the LIVF dimensions increase in flexion and decrease in extension [11][12][13][14][15] . The reduction in LIVF dimensions could result in increased compression on the nerve root due to facet joint motion and ligamentum flavum bulging 16 . In some patients, the nerve root impingement caused by foraminal stenosis occurs in extension but is relieved in flexion 22,23 . Hyperextending the hip joint is recommended to attain appropriate exposure of the proximal femur during the femoral preparation in patients undergoing DAA THA 3, 5, 6 ; therefore, a special position might change the morphology of the LIVF and lead to  This study first investigated the changes in LIVF dimensions when subjects were placed in DAA-specific hyperextension supine position. Our results showed similar decrease of the foraminal area to the previous studies at L 1 -L 5 levels [11][12][13][14] . However, the change in the foraminal area reached significant difference at the L 5 -S 1 level, which was similar to the results in Singh's and Schmid's study 11,14 but different from those in Ren's and Zhong's study 12,13 . The foraminal height and width also reached significant difference at the L 5 -S 1 level in our study. These differences might be attributed to the different imaging modalities and measurement methods. Moreover, putting the mat under the subjects could place them in passive hyperextension position, and the kinematics in such a position might differ from the active extension position reported in previous studies.
In our study, the changes in LIVF dimensions were largest at L 2-3 level, which was coincident with Ren's study 12 .
The changes were smallest at L 5 -S 1 level, which was possibly due to different biomechanics at L 5 -S 1 level compared to the upper levels. The sacrum is markedly curved and tilted backwards, and the first sacral vertebra articulates with the fifth lumbar vertebra at specific lumbosacral joint angle 19 . Thus, the L 5 -S 1 intervertebral foramen is thought to be less mobile compared to upper levels in extension of the lumbar spine 11,19 . However, the incidence of lumbar nerve root compression was found to be rare in L 2 nerve root but common in L 5 nerve root due to the increased ratio of the nerve root diameter to the foraminal area being lower compared to upper levels 16 . Based on the above reasons, attention should be paid to the decrease of LIVF dimensions at L 4-5 and L 5 -S 1 levels.
Although the percentage changes of foraminal width were slightly higher for the male subjects from L 2 -L 5 levels, we did not detect a significant difference in LIVF dimensions between the sexes and sides at all levels. The results of the regression analysis showed no influence of side, sex, and age on the percentage changes in LIVF dimensions.

Limitations
Some limitations to our study should be considered. First, the subjects involved in our study were mainly young and healthy volunteers; therefore, the changes might be different in older people and patients with lumbar spine diseases. Second, the patients were under general anesthesia during the operation, and the muscle relaxation effect possibly influenced the LIVF dimensions. The subjects in the current study were scanned in a conscious state, which might not accurately reflect the intraoperative changes of LIVF dimensions. The exact changes during the operation should be further investigated in future studies. Despite these limitations, our study still demonstrates a decrease in LIVF dimensions with the hyperextension of the lumbar spine in supine position.

Conclusion
In this study, we observed that the LIVF dimensions, including area, height, and width, decreased significantly with extension of the spine in supine position. The biggest decrease of foraminal dimensions was at L 2-3 level and the smallest at L 5 -S 1 level. The changes in dimensions were not influenced by side, sex, and age. This study demonstrated the position-dependent change of LIVF comparing the neutral supine position with the DAA-specific hyperextension supine position. This is of great value for understanding and preventing the potential risk of nerve root injury during DAA THA.