Could the Type of Allograft Used for Anterior Cervical Discectomy and Fusion Affect Surgical Outcome? A Comparison Between Cortical Ring Allograft and Cortico-Cancellous Allograft

The study protocol was approved by the Institutional Review Board of Asan Medical Center (IRB No. 2022-0863). Informed consent was waived due to retrospective nature of the study.

Study Design

This study retrospectively analyzed the data of patients who underwent ACDF between July 2016 and August 2019 for cervical myelopathy or radiculopathy and were regularly followed up for over 12 months. Patients who had previously undergone cervical spine surgery or combined posterior fixation and had been diagnosed with tumor, infection, or fracture were excluded. Patients were categorized into 2 groups based on the type of allograft: cortical ring and cortico-cancellous allograft groups. Subsequently, subgroup analysis was performed for both single- and multi-level operations (Fig. 1). All surgeries were performed by 1 orthopedic surgeon (DHL) in a single institution.

Fig. 1
Patient selection process. ACDF: anterior cervical discectomy and fusion.

• Click for larger image
• Download as PowerPoint slide

Method of Evaluation

For radiological evaluation, we reviewed the pre- and postoperative radiographs of the patients for cervical parameters including C2-7 lordosis, segmental lordosis of interbody fusion, and allograft complications including resorption, subsidence, and fracture. To assess the fusion rate, flexion and extension lateral views were obtained at 6 and 12 months following the surgery. The images were enlarged by 150% to facilitate the evaluation of interspinous motion (ISM) (Fig. 2). The criteria for fusion based on plain radiographic ISM is < 1 mm at the arthrodesis level and > 4 mm at a non-arthrodesed superjacent level (the level above the fused area).16) Additionally, we examined the multi-axial reconstructed computed tomography (CT) scans obtained at 2 days and 1 year after surgery. Four peripheral margins were evaluated: anterior and posterior margins on sagittal views and left and right margins on coronal views. A bridging bone with cortical density, which was present in at least 1 of these 4 peripheral margins, was categorized as fused.17) Extra-graft bone bridging (ExGBB) was defined as any bridging bone outside the graft or cage, without a transverse lucent line crossing the peripheral margins of the operated disc space. Meanwhile, intra-graft bone bridging (InGBB) was defined as any cortical or trabecular bridging bone without a transverse, lucent line within the confines of the graft or cage, and that was not outside the graft or cage (Fig. 3).

Fig. 2
Interspinous motion (ISM) in cervical lateral radiographic views. The images were enlarged by 150% to facilitate the evaluation of ISM. The ISM was defined as the difference in the interspinous distance between lateral flexion and extension radiographic views on an operative segment (ISM = A – B). The criteria for fusion based on plain radiographic ISM was < 1 mm at the arthrodesis level and > 4 mm at the non-arthrodesed superjacent level (the level above the fused area, a – b).

• Click for larger image
• Download as PowerPoint slide

Fig. 3
Extra-graft bone bridging (ExGBB) and intra-graft bone bridging (InGBB) in the multi-axial computed tomography scans obtained postoperatively. (A) ExGBB was defined as any bridging bone outside the graft or cage without a transverse lucent line crossing the peripheral margins of the operated disc space (arrows). (B) InGBB was defined as any cortical or trabecular bridging bone without a transverse, lucent line within the confines of the graft or cage and that was not outside the graft or cage (arrows).

• Click for larger image
• Download as PowerPoint slide

Significant allograft resorption was defined as a height difference of ≥ 1 mm in the 1-year postoperative CT scan.15) The amount of graft resorption was calculated by the difference in allograft height measured on CT obtained 2 days and 1 year postoperatively. Significant graft subsidence was defined as a graft migration of > 3 mm into the adjacent vertebral body, as calculated from the immediate postoperative radiographs compared to those obtained at 1-year follow-up.18) Graft fracture was assessed when the fracture line at the cortical portion of the allograft was detected on the 1-year postoperative CT scan (Fig. 4). Additionally, we assessed the occurrences of allograft complications based on the position of the cage. The position of the cage was considered to be in the middle if the cage center was within ± 5 mm of the center of the vertebral body, anterior if it was positioned more towards the front, and posterior if it was positioned more towards the back. Patient-reported outcome measures, including neck pain visual analog scale (VAS), arm pain VAS, and neck disability index were assessed. These measures were evaluated both before the surgery and at the final follow-up.

Fig. 4
Allograft resorption and fracture in anterior cervical discectomy and fusion. (A) Allograft resorption was assessed when there was a height difference of ≥ 1 mm in 1-year postoperative computed tomography (CT) scan. (B) Graft fracture was assessed when the fracture line at the cortical portion of the allograft was detected on the 1-year postoperative CT scan (arrows).

• Click for larger image
• Download as PowerPoint slide

Surgical Technique and Allograft Materials

The position of the patient was adjusted to achieve a physiologic or slightly lordotic alignment by placing horizontal padding behind the scapula. After positioning, the intraoperative c-arm was used to confirm the true view of the anteroposterior aspect of the uncovertebral joint. Surgeries were performed through a left-side transverse incision using the Smith-Robinson approach, with the operated segment localized. A box-shaped endplate preparation technique was used to enhance interbody fusion. Anterior compressive pathologies, including bone spurs, discs, and ossification of the posterior longitudinal ligament (PLL), were removed. In cases where radiculopathy symptoms and an MRI indicated the need for root decompression, uncovertebral joint resection was performed. To preserve cervical stability, uncinate process resection was performed minimally to achieve root decompression.19) Meticulous control of soft-tissue and bone bleeding was implemented to prevent heterotopic ossification. The PLL was resected in cases of preexisting damage due to severe herniation of disc material crossing over the PLL, particularly when the disc fragments were located behind the ligament.

Two types of allograft were utilized: a freeze-dried cortical ring type allograft (SteriGraft Cervical ACF implant, Bone Bank Allografts) and a freeze-dried cortico-cancellous allograft (Cornerstone ASR, Medtronic) (Fig. 5). Allograft size was selected based on intraoperative assessment, and allograft height that tightly fits into the discspace after releasing distraction without significant resistance was selected. An anterior cervical plate was applied for stabilization (Venture, Medtronic). A local autobone graft obtained during osteophyte or endplate preparation was embedded outside of the cage after applying the cervical plate in all cases.20) To prevent adjacent segment ossification development or adjacent segment disease, we used a short-length plate that slightly overlapped the upper and lower margins of the cranial and caudal level endplates.21, 22) The anterior plate (Venture, Medtronic) allows variable-angle screws up to 18° and fixed-angle screws at 10°. We inserted variable screws at the cranial and caudal levels at the largest angle possible.21, 22) We positioned the plate more than 5 mm away from the adjacent disc. Additionally, to increase the fusion rate, we inserted screws that were at least 75% of the vertebral body length.23)

Fig. 5
Allograft materials: cortical ring and cortico-cancellous types. (A) Plain cervical lateral radiograph of a patient operated using cortical ring allograft and the structure of the cortical ring allograft. (B) Plain cervical lateral radiograph of a patient operated using cortico-cancellous allograft and the structure of the cortico-cancellous allograft.

• Click for larger image
• Download as PowerPoint slide

The same postoperative management was employed for both groups. Following the surgery, a drain was inserted, and the wound was closed in layers. The suction drain was typically removed 1–2 days after the operation. Patients were instructed to wear a postoperative rigid cervical collar (Miami J, Ossur) for 4 weeks after a single-level operation and for 3 months after a multi-level operation.19, 24, 25)

Of the 257 individuals, 227 were enrolled in the study. Among them, 134 (59.0%) underwent ACDF with cortical ring allograft, while 93 (41.0%) received cortico-cancellous allograft. No significant differences were observed in demographics between the 2 groups. However, the 2 groups significantly differed in the number of levels operated on, with the cortical ring and cortico-cancellous groups having an average of 2.03 and 1.31 levels, respectively (Table 1).

Table 1
Patients Demographics

• Click for larger image
• Click for full table
• Download as Excel file

We conducted a subgroup analysis by dividing the patients into single- and multi-level operation groups. In the single-level operation, no significant differences were observed in patient demographics and complications (Table 2). However, in the multi-level operation, the average number of levels operated was significantly higher in the cortical ring allograft group (2.3 vs. 2.1). No other significant differences were observed in the demographics or complications (Table 3). When comparing the results of single-level operation, the cortico-cancellous allograft exhibited a significantly frequent allograft resorption (24 / 66, 36.4%) compared to the cortical ring allograft (1, 3.7%; p = 0.001). Moreover, the cortico-cancellous allograft group demonstrated significantly greater subsidence (p = 0.038). However, the fusion rates did not significantly differ between the 2 groups (Table 4). In multi-level operations, cortico-cancellous allografts (5 / 27, 18.5%) resulted in a significantly higher fracture rate than did the cortical ring allograft (5 / 105, 4.7%, p = 0.030). Although the fusion rate at postoperative 1 year was higher in the cortical ring group assessed by both ISM (63.2% vs. 55.5%) and InGBB (66.7% vs. 40.7%), the difference was not statistically significant (p = 0.510 and p = 0.500, respectively) (Table 5).

Table 2
Patient Characteristics (Single-Level Operation)

• Click for larger image
• Click for full table
• Download as Excel file

Table 3
Patient Characteristics (Multi-Level Operation)

• Click for larger image
• Click for full table
• Download as Excel file

Table 4
Radiographic Results (Single-Level Operation)

• Click for larger image
• Click for full table
• Download as Excel file

Table 5
Radiographic Results (Multi-Level Operation)

• Click for larger image
• Click for full table
• Download as Excel file

In the analysis of cage position and allograft complications, the posterior position showed a significantly high rate of subsidence (2 / 10, 20%) and fracture (3 / 10, 30%) in single-level operations, whereas the occurrence of fractures was significantly higher in the middle cage position (8 / 55, 14%) in multi-level operations (Table 6). The sagittal alignment between the groups with allograft complications and without complications showed no significant differences in cervical parameters in single-level fusions. However, in multi-level fusions, the allograft complication group showed a significant kyphotic change in the segmental lordosis of fusion levels, decreasing from 7.8° (± 7.0°) 1 month postoperatively to 4.4° (± 7.5°) 1 year postoperatively. In comparison, the no complication group showed a change from 7.3° (± 4.9°) to 7.2° (± 5.5°) over the same period (Fig. 6). The patient-reported outcome measures at 1-year postoperative mark did not reveal a significant intergroup difference between the 2 groups, both in single- and multi-level operations (Table 7).

Fig. 6
Changes over time in segmental lordosis and C2-7 lordosis in groups with allograft complications and no complications. (A, B) Single-level operations. (C, D) Multi-level operations. ^*Significant difference from preoperative assessment. ^†Significant difference between 2 groups. ^‡Significant difference from the prior assessment. p < 0.05.

• Click for larger image
• Download as PowerPoint slide

Table 6
Analysis of Allograft Complications Based on the Location of the Allograft

• Click for larger image
• Click for full table
• Download as Excel file

Table 7
Patient-Reported Outcome Measures

• Click for larger image
• Click for full table
• Download as Excel file

The selection of an appropriate graft material is crucial for achieving successful bone fusion and optimal clinical outcomes in ACDF.12) Tricortical iliac crest autograft, known for its superior biological activity, was traditionally considered the gold standard interbody spacer for ACDF.26) However, its use has decreased due to drawbacks such as donor site morbidity and increased surgical time.2, 3) Recent studies on ACDF have shifted towards alternative graft materials that offer comparable outcomes while bypassing the issues associated with autografts. Additionally, as anterior plating gains popularity, an increasing number of studies advocate the use of allografts in ACDF with plate augmentation, demonstrating improved postoperative clinical and radiological outcomes.27, 28, 29) Allografts are recognized as osteoconductive and may offer a certain degree of osteoinductivity, albeit lacking the osteogenic potential of autografts.30) Their availability and prevention of donor site morbidity are notable advantages. Moreover, the fusion rates of plated allografts appear comparable to those of autografts.5) However, there are some drawbacks to allografts, including cost, variation in osteoinductive potential between different manufacturers based on sterilization techniques and donors, as well as the potential risks of disease transmission and host rejection. Recently, various types of allografts, including cortical ring, cortico-cancellous, and dense cancellous, are utilized as interbody materials in ACDF.10, 11, 12)

Cortical ring allografts are known to be mechanically stronger than cancellous bone and have demonstrated successful outcomes in ACDF procedures. However, they require prolonged time for remodeling and complete incorporation. Lee et al.13) reported that a cortical ring type allograft filled with demineralized bone matrix in a single- or double-level plate-supplemented ACDF surgery demonstrated comparable clinical and radiological outcomes, along with reduced bleeding, when compared to the standard autograft. Other studies have shown that the use of cervical plates significantly reduces graft-related complication rates in 1- and 2-level ACDF with cortical allograft, leading to markedly enhanced arthrodesis.31)

A cortico-cancellous allograft is a freeze-dried bone cage that consists of cortical lateral walls and a cancellous center. The cortical portion provides structural support for the disc space, while the cancellous portion acts as a scaffold for bone in-growth, reducing graft subsidence and enhancing fusion rates. Park et al.12) suggested that patients treated with cortico-cancellous allograft or autograft for ACDF showed similar clinical and radiologic outcomes, but with a decreased subsidence rate in the cortico-cancellous allograft group. Therefore, cortico-cancellous allograft can be a potential alternative graft material for ACDF.

Dense cancellous allografts exhibit a compressive strength similar to that of cervical endplates, thereby diminishing the likelihood of graft pistoning and settling. They possess a trabeculated and porous structure, which provides a better scaffold for osteoconductive bone ingrowth compared to that of cortical allografts. However, dense cancellous allografts have been associated with high rates of resorption when used in ACDF. Although most segments fused and no patients required revision for symptomatic pseudarthrosis, the resorption of the graft resulted in voids within the graft, with only 53% of levels showing bridging trabeculations covering more than half of the disc space. Despite the theoretical advantages, caution should be exercised when considering the use of dense cancellous allografts in ACDF due to their propensity for resorption.11)

In our study, we observed a higher incidence of allograft resorption or fracture in cortico-cancellous allografts than in cortical ring allografts. Several studies highlighted the differences in surgical outcomes between single- and multi-level ACDF.32, 33, 34) Veeravagu et al.33) studied the rates of adverse events and the need for revision surgery in patients undergoing single- versus multi-level ACDF, and they found that cervical reoperations were more common in the multi-level ACDF group. Based on the established differences in surgical outcomes between single- and multi-level operations, we conducted our study by dividing groups into single- and multi-level for evaluation. Cortico-cancellous allografts resulted in significantly greater subsidence and resorption than cortical ring allografts in a single-level operation. These findings are likely attributed to the mechanical properties of the cortico-cancellous type, which may undergo dissociation or deformation under compressive forces, making them mechanically weaker than cortical ring type allografts. According to Kwon et al.'s study on the biomechanical analysis of allograft spacer failure,35) when comparing 3 types of allografts used in ACDF (cortical only, cortical cancellous, and cortical lateral walls with a cancellous center), the cortical lateral walls with a cancellous center exhibited the highest stress on the cortical bone of the spacers and the endplate around the posterior margin of the spacers. They concluded that cervical spacers with a smaller cortical component could be a risk factor for allograft spacer failure and subsidence, especially during flexion and extension.

On the other hand, no statistically significant differences were observed in subsidence and resorption between the 2 types of allografts in multi-level operations (1.7 mm vs. 2.1 mm, 49.1% vs. 48%). Interestingly, fracture complications of the allograft were significantly higher in the cortico-cancellous type group in multi-level operations (4.7% vs. 18.5%). The incidence of resorption was higher in multi-level than in single-level operations. This increase may be attributed to the higher number of allografts used in multi-level operations. It is likely that in multi-level operations, the manifestation of allograft complications occurred in the form of fractures. The difference in the pattern of allograft complications between a single- and multi-level operations may be due to the varying biomechanical characteristics depending on the number of surgical levels involved. Yang et al.15) suggested a high rate of graft morphologic changes (54.3%) in cortico-cancellous allografts. These morphologic changes, such as graft resorption or fracture, were associated with increased pseudarthrosis, subsidence, and decreased postoperative segmental lordosis. Notably, these morphologic changes did not significantly impact the clinical outcomes. The higher rate of morphologic changes observed in cortico-cancellous allografts may be attributed to their mechanical properties.

In multi-level operations, allograft complications, including resorption, subsidence, and fracture, resulted in kyphotic changes in segmental lordosis. According to Yamagata et al.,36) the operated segment lordosis significantly decreased in the group with cage subsidence compared to the no subsidence group. Additionally, Yang et al.15) reported a significant difference of -3.1° in segmental lordosis between the postoperative and final state in the group with allograft morphologic changes and a difference of -1° in the group with unchanged segments, indicating more kyphotic changes in the group with morphologic changes. This could be due to the decrease in segment height caused by cage complications.

In this study, in the multi-level group, the subsidence rates were 13.2% (14 / 106) for cortical ring allograft and 18.5% (5 / 27) for cortical cancellous allograft. Subsidence is the most commonly reported cage-related complication, with an incidence of 21% (according to a systematic review by Noordhoek et al.).37) That review included various types of cages, such as PEEK, titanium, cage-screw-combination, and polymethyl-methacrylate (PMMA), and showed that the use of PEEK or titanium cages or cages with integrated screws can lower the risk of subsidence. However, the type of allograft that we used in our study was not investigated in the review by Noordhoek et al.37) A study by Yson et al.38) compared subsidence rates between allograft and PEEK cages and reported no statistically significant difference between the PEEK group (29%, 25 / 85) and the allograft group (28%, 9 / 32). That study concluded that the choice between PEEK and allograft as spacers in ACDF did not significantly affect the subsidence rate or clinical outcomes. In the current study, we observed a lower subsidence rate compared to previous studies. We implemented the use of the Miami J brace, a rigid type of neck brace, for 12 weeks. According to Guo et al.,19) excessive resection of the uncovertebral joint can compromise cervical spine stability, particularly during lateral flexion and rotation. The uncovertebral joints play a crucial role in limiting these movements. Therefore, to mitigate the cage stress concentration after ACDF, the use of a rigid neck brace that restricts lateral flexion and rotation is recommended.

To evaluate fusion in ACDF, we used 3 widely used methods. One method used 150% magnified ISM, and the other 2 methods used CT-based ExGBB and InGBB. Among these, ExGBB has been proven to have the highest diagnostic sensitivity and specificity for detecting pseudarthrosis compared to all the other radiographic criteria.39) According to Song et al.,16) the method of measuring 150% magnified ISM demonstrated accuracy that was comparable to CT. Park et al.20) reported a difference between 6-month ISM and 1-year ISM, with a fusion rate of 40% at 6 months, 80% at 12 months, and 88% at 24 months, indicating a significant difference between the early 6-month and 12-month periods. These findings are consistent with those seen in our study.

The fusion rate of ACDF is known to be high; however, recent reports have suggested that pseudarthrosis is a common complication after ACDF, particularly for multi-level operations. In our study, the fusion rate was 88% for single-level cases and 66% for multi-level cases. Other studies have shown the fusion rate for single-level cases to be higher than that for multi-level cases, as multi-level cases are diagnosed as pseudarthrosis, if even a single segment does not meet the fusion criteria.16, 20, 39) In the study by Park et al.,20) the fusion rate at 12 months in the group using allografts with an average fusion level of 1.5 was reported to be 80% when assessed using the ISM method, 88% by the InGBB method, and 74% by the ExGBB method. Wewel et al.40) reported that the incidence of pseudarthrosis was higher in patients with 4-level ACDF (56%, 9 / 16) compared to those with 3-level ACDF (42%, 24 / 56). However, pseudarthrosis did not affect clinical outcomes.15, 20)

In single-level fusions, subsidence and fractures occurred significantly when the allograft was inserted in the posterior position. In multi-level fusions, allograft fractures occurred significantly when the allograft was positioned in the middle or posterior locations. The study by Yamagata et al.36) on cage subsidence and cervical alignment showed that on the fusion level, the cage size and cage position were significantly related to cage subsidence. Cage positions were analyzed by categorizing them into grade 0 when the cage was within 2 mm of the anterior margin of the plate and grade 1 when the cage was more than 2 mm away. A grade 1 cage position resulted in a significantly higher risk of cage subsidence compared to a grade 0 position and they concluded that excessive distraction at the fusion level should be avoided and the cage position should be adjusted to the anterior vertical line. Additionally, a study on posterior lumbar interbody fusion reported that posterior cage positioning on the vertebral endplate was associated with a higher risk of subsidence.41) These findings are similar to those of our study, where cage subsidence or fractures occurred significantly when positioned more towards the back.

Our retrospective study has several limitations. First, the allograft was randomly selected without consistent criteria, resulting in an uneven distribution of cortical ring and cortico-cancellous allografts between the single- and multi-level operations. This could have introduced a potential bias when comparing the 2 groups. Second, the number of surgical levels differed between the cortical ring and cortico-cancellous allograft groups in the multi-level operations. Although the difference in levels (2.3 vs. 2.1) was not substantial, it could be a limitation of this study as it may have impacted the results. Third, the lack of reliability assessment for radiographic assessment is another limitation of this study. Finally, there is inconsistency in allograft complications such as subsidence, resorption, and fracture between the single- and multi-level operations. This might be attributed to differences in mechanical characteristics between the single- and multi-level operations; however, further evaluation through biomechanical studies is necessary to better understand these differences.

The present study demonstrated that allograft resorption or fracture occurs more frequently with cortico-cancellous than with cortical ring allograft. Cortico-cancellous allografts resulted in significantly greater subsidence in a single-level operation. Furthermore, although not statistically significant, a large difference was observed in the fusion rate in a multi-level operation (66.7% vs. 40.7%), with the cortical ring allograft showing a higher fusion rate, warranting further verification. Due to the higher rate of allograft resorption or fracture, and significantly greater subsidence with cortico-cancellous allograft, cortical ring allografts may yield more stable results for ACDF.