Comparison of Load-Sharing Responses Between Graded Posterior Cervical Foraminotomy and Conventional Fusion Using Finite Element Modeling

According to the 2016 Global Burden of Diseases, neck pain is ranked among the top ten causes of disability worldwide, accounting for 301 diseases and injuries in 188 countries [1]. Cervical radiculopathy is a common chronic degenerative disease of the cervical spine [2–4]. Diagnosis of radiculopathy is based on clinical symptoms and confirmed by magnetic resonance imaging of the cervical spine. In cases of unilateral cervical radiculopathy, a lateral herniated disk fragment or facet hypertrophy with nerve root compression is often noted. The surgical treatment includes, posterior cervical foraminotomy, or anterior cervical discectomy and fusion with bone graft and plates. The United States Food and Drug Administration has approved artificial disks as a replacement for conventional fusion [5,6]. Artificial disks have different designs, different materials, and some are approved for more than one segment fixation. Posterior cervical foraminotomy is indicated for those patients with symptoms attributed to lateral disk herniation that tends to narrow the cervical foramen [7,8]. This procedure allows more space for the nerve root to decrease radiculopathy. Conventional fusion procedure involves decompressing the nerve root by directly removing the herniated disk fragment and can be performed for more centrally located disk herniation. The disk space is then replaced with an implant (autograft/allograft/synthetic cage or device) and often supplemented with a plate and screws to induce bony fusion across the disk space.

No significant differences in outcomes between foraminotomy and conventional fusion have been reported; however, the foraminotomy approach may avoid the risk of accelerated degeneration of adjacent segments caused by conventional fusion [9–13]. Healthcare costs are greater with conventional fusion due to placement of the implant [13]. In addition, conventional fusion may not be acceptable to younger patients, athletes, and military personnel. Foraminotomy is a minimally invasive surgical option that avoids the need for fusion; however, one clinical concern is the need to resect a portion of the facet joint that can may lead to micro-instability and worsened neck pain. While it is known that the amount of facet joint resected has a bearing on spine stability after foraminotomy, studies have not quantified the biomechanical effects of graded facet resection in foraminotomy.

The objectives of this study were to compare the external and internal biomechanical responses of the effects of conventional fusion and foraminotomy with different grades of facet resection using clinical metrics of range of motion, disk pressure, and facet loads at the index segment and adjacent caudal and cranial segments.

A three-dimensional osteoligamentous finite element model of the intact human cervical spinal column (C2-T1) was used in this study [14–19]. The intact spine was altered to simulate minimally invasive left-sided foraminotomy: 50%, 75%, and 100% facet resections (termed graded foraminotomies); and conventional fusion with anterior plating and screws to the vertebrae. All surgical simulations were performed at the C5–C6 segment, the commonly operated segment in civilian and military populations [20–22]. The conventional fusion procedure was simulated by inserting a bone graft centrally between the C5 and C6 vertebral bodies. The process included fixing the vertebral bodies in space during the surgical operation. A titanium plate was simulated with variable angle screws into the vertebral bodies. The solid models of the anterior cervical plate with variable screw system were developed using catiav6 software. The anterior screws were 18 mm long with a diameter of 3 mm. The interface between the bone graft and adjacent vertebral bodies had bonded contact. The material properties of the instrumentation were obtained from literature [23]. The foraminotomy procedure was simulated by removing the bone from the left facet region of the C5–C6 segment and part of the articular capsule ligament, and three different grades of 50%, 75%, and 100% of bone removal were simulated. For all graded models, resection was done at the midportion of the facet joints, and a portion of the C5 inferior facet and the C6 superior facet region was removed. Figure 1 shows the intact and surgically altered spines.

The intact, conventional fusion, and foraminotomy surgical spines were constrained at the inferior surface of the T1, and physiological moments (2 N·m) combined with a follower load (75 N) were applied to the superior vertebra [24]. The follower load was implemented using follower cables [25]. First, the intact spine was exercised in flexion and extension and lateral bending modes, and the overall range of motion was determined under each loading mode. The next step was to determine the magnitude of bending moments to the conventional fusion and graded foraminotomy spines that matched the range of motion obtained above for the intact spine and was done following the hybrid loading method [26]. This was accomplished by altering the externally applied moment, in each mode, until the overall column range of motion of the spine with the conventional fusion and graded foraminotomies matched with the magnitude of the range of motion of the intact spine. The range of motion the operated (index), and adjacent caudal and cranial spinal segments representing clinical measures, facet loads at these segments representing the resisted forces due to surgery in the posterior of the spine, and disk pressures in the adjacent segments representing force due to surgery in the anterior of the spinal column were obtained from all foraminotomy, conventional fusion, and intact spines. The range of motion, disk pressure, and facet forces at the index and adjacent segments for all cases were normalized with respect to the intact spine, in flexion, extension, and lateral bending. These data from the conventional fusion and foraminotomy spines were used to compare changes in the external and internal biomechanical responses of the spine by normalizing with respect to the intact spine.

The model was validated with human cadaver experimental data including those from our laboratory and others for the range of motion, disk pressure, and facet loads, the three metrics used to comparatively evaluate different surgical options. The model-predicted metrics were within mean±1 standard deviation data from experimental studies [16,27–32]. Moments required to achieve the overall range of motion of the intact spine for the conventional fusion and foraminotomies in flexion were 3.5 N·m, 1.92 N·m, 1.8 N·m, and 1.7 N·m, respectively. In extension, these data were 4.0 N·m, 1.95 N·m, 1.74 N·m, and 1.69 N·m, respectively. In the left lateral bending mode, they were 3.0 N·m, 1.8 N·m, 1.75 N·m, and 1.65 N·m, respectively. In the right lateral bending mode, they were 3.0 N·m, 1.89 N·m, 1.8 N·m, and 1.73 N·m, respectively.

Figure 2 shows the summary of range of motion results. In flexion, conventional fusion decreased motions up to 99% at the index and increased up to 27% at adjacent segments. Foraminotomy with 50%, 75%, and 100% facet resection responded to increased motions at the index (21%, 31%, and 38%, respectively) and decreased at adjacent segments (up to 11%, 14%, and 17%, respectively). In extension, conventional fusion decreased motions up to 96% at index and increased up to 35% at adjacent segments. Foraminotomy with 50%, 75%, and 100% facet resection responded with increased motions at the index (21%, 31%, and 38%, respectively) and decreased at adjacent segments (up to 13%, 17%, and 21%, respectively).

In the right lateral bending mode, conventional fusion decreased motions up to 98% at the index and increased up to 17% at adjacent segments, and foraminotomy with 50%, 75%, and 100% facet resections responded with increases of 20%, 35%, and 47%, respectively, at the index, and decreased up to 9%, 16%, and 20%, respectively, at adjacent segments. In the left lateral bending mode, conventional fusion decreased motions up to 99% at the index and increased up to 21% at adjacent segments, and foraminotomy with 50%, 75%, and 100% facet resections responded with increases of 24%, 56%, and 65%, respectively, at the index, and decreased up to 10%, 18%, and 21%, respectively, at adjacent segments.

Figure 3 shows changes in the intradiscal pressures of the conventional fusion and all three graded foraminotomy models comparing with the intact spine at the superior, index, and inferior segments. In flexion, compared to the intact spine, conventional fusion increased pressures up to 22% at adjacent segments. Foraminotomy with 50%, 75%, and 100% facet resection responded to increased pressures at the index (7%, 21%, and 38%, respectively) and decreased at adjacent segments (up to 5%, 16%, and 26%, respectively). In extension, conventional fusion increased pressures up to 34% at adjacent segments. Foraminotomy with 50%, 75%, and 100% facet resection responded increased motions at the index (16%, 25%, and 41%, respectively) and decreased at adjacent segments (up to 5%, 12%, and 25%, respectively).

In the left lateral bending mode, conventional fusion increased pressures up to 28% at adjacent segments, and foraminotomy with 50%, 75%, and 100% facet resections responded with increases of 23%, 38%, and 63%, respectively, at the index, and decreased up to 5%, 12%, and 23%, respectively, at adjacent segments. In the right lateral bending mode, conventional fusion increased pressures up to 28% at adjacent segments, and foraminotomy with 50%, 75%, and 100% facet resections responded with increases of 5%, 13%, and 23%, respectively, at the index, and decreased up to 16%, 21%, and 23%, respectively, at adjacent segments.

Figure 4 shows changes in the facet loading responses of the conventional fusion and all three graded foraminotomy models comparing the intact spine at the superior, index, and inferior segments. In extension, conventional fusion decreased left side facet loads up to 85% at the index and increased up to 26% at adjacent segments. Foraminotomy with 50%, 75%, and 100% facet resection decreased loads at the index (15%, 23%, and zero, respectively) and increased at adjacent segments (up to 13%, 26%, and 45%, respectively). Right side facet load had up to 85% decrease in conventional fusion at the index level and up to 27% increase at adjacent segments. Foraminotomy with 50%, 75%, and 100% facet resection increased facet loads at the index (20%, 34%, and 60%, respectively) and adjacent segments (up to 14%, 23%, and 41%, respectively). In the left lateral bending mode, conventional fusion decreased facet loads up to 90% at the index and increased up to 28% at adjacent segments, and foraminotomy with 50%, 75%, and 100% facet resections responded with increases of 15%, 29%, and zero, respectively, at the index, and adjacent segments up to 20%, 18%, and 13%, respectively. In the right lateral bending mode, conventional fusion decreased loads up to 89% at the index and increased up to 27% at adjacent segments, and foraminotomy with 50%, 75%, and 100% facet resections responded with increases of 10%, 19%, and 23%, respectively, at the index and adjacent segments up to 7%, 14%, and 23%, respectively.

Conventional fusion leads to adjacent level disease [33]. This observation served as a basis to focus this study on adjacent level kinetics. Increased range of motion and load-sharing in both anterior and posterior columns with conventional fusion offer support to clinical observations. Other procedures such as those simulated in this study may offer an alternative, as they decrease some of these metrics. As full facetectomy considerably increased disk pressure at the index and stress at the index and adjacent level facets more than conventional fusion, it may be appropriate to maintain facetectomy to below the 75% level. Understanding load-sharing between spinal components may assist in the decision-making process and patient counseling, as treatments are based on factors such as individual anatomy, disease status, and occupation of the patient. For example, if the occupation involves the use of head supported mass, and if the patient has facet arthrosis, the surgeon may select a surgical procedure that induces less demand on the facet joint (facet stress/load from the procedure). The generic model used to simulate different types of surgeries focused only on the surgical effects and eliminated effects of other alterations, commonly present in surgical patients. Patient-specific models generated using morphing techniques using our model can incorporate individual anatomical variations, and their outputs would be more applicable to individual patients.

Using a validated finite element model, this study compared conventional fusion and graded foraminotomies with respect to the intact spine using range of motion, a clinically relevant metric, and intrinsic parameters of facet loading, representing posterior column load-sharing, and disk pressures, representing anterior column load-sharing within the spinal column. Such metrics can only be studied using finite element models, as cadaver tests are not feasible. Segmental level range of motion, pressure, facet loads were compared to the intact spine under physiological loading conditions: flexion, extension, and lateral bending (both left and right). Under all physiological loading modes, conventional fusion almost completely restricted the motion at the index segment whereas it increased motion at both adjacent superior and inferior segments, up to 35%. The posterior load-sharing decreased at the index segment by up to 90% but it increased by up to 26% at the adjacent segments along with similar increases for the intradiscal pressure. These increases may contribute to adjacent segment degeneration. This finding is consistent with the literature data [34,35].

An experimental study used six C2–C7 human cadaveric spine specimens to demonstrate the distribution of facet load at the adjacent segments after conventional fusion surgery. In extension, facet joints experienced higher pressures than the intact spine [36]. A recent review reported that reoperations involve an adjacent level after initial conventional fusion [37]. Studies have also reported that conventional fusion accelerates the degeneration of adjacent segments due to severely limited motions (up to 90%) at the index level [33,38,39]. These findings appear to be well supported with the present finite element modeling results.

Posterior cervical foraminotomy on the other hand showed biomechanical changes at the index and adjacent segments opposite to the conventional fusion. Foraminotomy increased range of motion at the index level, while decreasing motion at the adjacent levels. Simulated foraminotomy models produced the greatest increase in motion in lateral bending with 50–100% facet resections, implying that the facet joint has the greatest stabilization role in lateral bending. Due to the matching overall motion between intact, conventional fusion, and foraminotomy spines, the range of motion of adjacent segments had a compensatory decrease, which may be related to the increase at the index level segment. Reduced motion at the adjacent segment suggests that the risk of adjacent segment degeneration may decrease with foraminotomy compared to conventional fusion, an advantage of the foraminotomy procedure.

The results of this finite element simulation support prior cadaver studies showing that resection of over 50% of the facet joints contributes to increased mobility and instability at the index level in the cervical spine [40–42]. This may be a contributor to increased neck pain in foraminotomy patients with more than 50% of the joint resection. Based on these findings, if more than one-half of the facet joint needs to be resected to adequately decompress the nerve root during a foraminotomy, the surgeon may need to consider supplemental fusion or an conventional fusion as an alternate approach. Future study is needed to determine if one-half facet joint resection at two adjacent levels during a bi-level foraminotomy will yield a similar biomechanical result.

The decision to perform a foraminotomy or conventional fusion for the individual patient depends on patient and surgeon-related factors. Patient anatomy is often a major reason to choose a procedure, in particular, the location of the nerve root compression, size, and orientation of the facet joints as well as sagittal alignment. These factors highlight the importance of including patient-specific anatomy in future finite element models, and the authors of this study are pursuing such modeling approaches for personalized medicine [42–45]. This would provide a more personalized biomechanical assessment to help counsel the individual patient. Additional factors that could contribute to biomechanical instability include the presence of spondylolisthesis or any instability at baseline, which would favor the use of fusion procedure instead of foraminotomy.

These comparative assessments of minimally invasive foraminotomies with various resections and conventional fusion may allow the surgeon to decide on the optimal option for treatment from a biomechanical perspective, in addition to surgical experience, and it may also assist in patient counseling. Finite element modeling with patient specific spine anatomies will be more effective in the surgical decision-making process, and the present morphable validated model can be used on a patient-by-patient basis.

Using finite element models, the present investigation quantified internal load-sharing responses and external range of motion responses at the surgical and adjacent segments for conventional fusion and graded facet resections for the foraminotomy procedures. Biomechanical metrics were dependent on the magnitude of facet resection and type of the metric, i.e., anterior versus posterior column load-sharing and spinal segment, i.e., index and adjacent segments. Recognizing that foraminotomy serves as a motion preserving alternative to conventional fusion, the present modeling study highlights various intrinsic biomechanical factors and potential instability issues with more than one-half facet resection.

The research was supported by the Office of the Assistant Secretary of Defense for Health Affairs, through the Broad Agency Announcement, the National Center for Advancing Translational Sciences, National Institutes of Health, and the Department of Veterans Affairs Medical Research. This material is the result of work supported with resources and use of facilities at the Zablocki VA Medical Center, Milwaukee, WI. Dr. Yoganandan is an employee of the VA Medical Center. The opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the sponsors.

• Office of the Assistant Secretary of Defense for Health Affairs, through the Broad Agency Announcement (Award No. W81XWH-16-1-0010; Funder ID: 10.13039/100000005).

• The National Center for Advancing Translational Sciences, National Institutes of Health, Award No. UL1TR001436, and the Department of Veterans Affairs Medical Research (Funder ID: 10.13039/100000002).

The authors attest that all data for this study are included in the paper.