Transforaminal lumbar interbody fusion via a fusion cage is widely performed to treat degenerative lumbar spinal disease. The cage acts as a spacer between vertebral endplates and provides mechanical support for the anterior column, leading to restoration of disc height and indirect decompression of the neuroforamen. Previous studies reported favorable clinical outcomes and fusion rates following TLIF via cages [14,15,16]. However, implantation of a single cage on one side can result in certain side effects. Despite the advantages, nerve root injury, dura mater injury, and PCM are possible complications that would result in spinal canal or exiting root compression, leading to low back pain, leg pain, or pseudoarthrosis [2, 7,8,9].
In the current study, we found that 24 out of 953 patients developed PCM in 4.92 ± 5.98 months (3 days - 22 months) after surgery. The prevalence of PCM was 2.52%, and one fourth of the patients (6 of 24) received revision surgery due to severe symptoms. The incidence was compatible to prior studies that reported the prevalence of PCM after TLIF or PLIF ranging from 1.17 to 14.7%, and occurred within 7 months after operation [3, 4, 6, 10]. Most cases of PCM occurred in the early postoperative period before achieving solid fusion. This complication could lead to devastating results and require revision operation for symptomatic patients.
Our hypothesis that cage position would affect PCM incidence was verified. The incidence of PCM increased as cages were placed more posteriorly among 211 cages in the PCM and control groups in the present study (Fig. 6). Posteriorly located cages were shown to be a significant risk factor for PCM in univariate and multivariate analyses. However, the cage coronal position showed no significant impact on PCM. To the best of our knowledge, this study is the first study that demonstrated a significant relationship between sagittal cage position and PCM.
A prior study from Aoki et al followed 125 patients with 4 PCM cases after receiving TLIF with radiopaque cages [3]. They measured the distance between the cage posterior margin and posterior margin of the cranial (or caudal) endplate. The migrated cages were found to have more posterior initial position than non-migrated cages. However, the difference was not significant. The authors failed to demonstrate the relationship between PCM and posteriorly located cages, possibly due to the small sample size. This current study had more patients compared to Aoki et al (24 patients versus 4 patients) and other existing literatures [3, 4, 6,7,8, 10]. Because the cages used in this study were all radiolucent, it was difficult to define cage margins by radiograph as Aoki et al did. Instead of measuring definite distance between the cage margin and the endplate margin, we plotted the cage center from the radiopaque markers, and calculated the depth ratio and the coronal ratio to demonstrate relative distance between the cage center and the disc center. As endplate length and width varies between individuals, we supposed the depth ratio and the coronal ratio were theoretically better methods for evaluating the influence of cage position on PCM.
With regard to biomechanical stabilization when comparing anteriorly located cages versus posteriorly located cages, a study of a synthetic spine model performed by Polly et al demonstrated constructs with anterior cage placement were significantly stiffer than constructs with central or posterior cage placement. The constructs with anterior cage placement had significantly reduced rod strain and increased cage strain in axial compression compared to constructs with posterior cage placement, which reinforced the concept of load sharing between anterior and posterior columns [5]. Following studies with cadaveric models by Faundez et al and Tallarico et al reported different results. The authors found no significant difference in flexion-extension ROM between anterior cage placement and posterior cage placement constructs [17, 18]. An in vivo kinematic study demonstrated a non-significant correlation (p = 0.055) between more anteriorly located cages and greater standing flexion-extension stability [19]. The results regarding construct stability were inconclusive. The remnant thick anterior annulus fibrous after TLIF could support the anterior column and influence overall stability, which might make the impact of sagittal cage positioning less distinguishable. However, in regard to pressure between the endplate and cage, the cage shares more load when placed more anteriorly in the disc space. Compared to posteriorly located cages, we supposed that anteriorly located cages bear more pressure as gravity transmits and generates greater friction to resist posterior migration.
To investigate the influence of coronal cage position on construct stability, Quigley et al utilized a synthetic spine fusion model and compared constructs with anterolateral, central, or anterior cage placement [20]. The authors found no significant difference in strain, load-to-failure, or cage displacement after long-term cycling. Comer et al also compared a cadaveric model with centrally placed and laterally placed cage constructs [21]. Despite significantly lower stiffness under compression, constructs with lateral cage placement demonstrated similar stiffness to centrally positioned ones in flexion, extension, lateral bending, and torsion. Thus, it is theorized that the coronal position of the cage has less impact on PCM.
To minimize the risk of PCM, we should avoid placing the cage with its center located posterior to the disc center. However, due to limited discectomy by TLIF approach [22], the remnant disc and the bone graft packed in the disc space may hinder surgeons from placing the cage center anteriorly enough to reach or pass the disc center. The technique of disc preparation plays an important role in avoiding PCM. Before cage insertion, the trial implant should reach adequate depth after disc preparation and bone graft packing. This process would allow surgeons to insert the cage along the path created by trial implant and achieved an ideal cage position. It is always necessary to monitor cage position during surgery by fluorescent imaging. The depth ratio in lateral radiographs is a useful tool to detect mal-positioned cages before wound closure.
In the current study, PCM was found in discs with significantly increased preoperative disc height. Height variance, which was defined as ‘cage height - preoperative disc height’, represented the relative size of the cage in the disc space. Less height variance was found to be a significant risk factor in the univariate and multivariate analyses, which verified the hypothesis that undersized cages are a risk factor for PCM. Multiple studies reported similar results. Kimura et al analyzed 1070 cases (9 with PCM) retrospectively and found significantly larger preoperative disc height in migrated cages [6]. Li et al and Aoki et al reported undersized cages to be a significant risk factor for PCM [3, 10]. Larger cage height creates more tension in peri-vertebral soft tissue, resulting in higher contact pressure at endplate-cage interfaces. The interfaces are able to provide more friction during motion to resist posterior migration. We recommend using cages with heights equal to or larger than preoperative disc heights.
Cage geometry was not verified to influence PCM in the present study. However, kidney-shaped cages had higher PCM incidence than bullet-shaped cages, even though this difference was not statistically significant (p = 0.063). Seventy-five percent of migrated cages (18 of 24) were kidney-shaped, while only 56.32% of inserted cages were kidney-shaped in the current study. We hypothesized that the non-significant difference resulted from the cage insertion technique. Kidney-shaped cages are designed to rotate approximately 90 degrees within the disc space during insertion. Therefore, the cage moves in a curved trajectory during insertion, making it relatively difficult to drive the whole cage anteriorly when the bone graft is packed in the anterior disc space. For bullet-shaped cage, the cage moves anteriorly in a straight trajectory during insertion and takes less effort to reach central or anterior disc space. Furthermore, less effort is required to withdraw the bullet-shaped trial implant from within the disc space, which may make surgeons choose larger implants to achieve enough tension. It is theorized that these differences during insertion make it less possible to have undersized bullet-shaped cages located in posterior disc space.
Prior biomechanical studies showed no difference in stability between kidney-shaped and bullet-shaped cage constructs [21]. Reports from Aoki et al mentioned that all of their migrated cages were bullet-shaped cages [3]. Pan et al and Zhao et al reported bullet-shaped cages had significantly higher PCM incidence [4, 7]. However, there was no comparison to kidney-shaped cages from Aoki et al [3], and all these studies were analyses of less than 10 patients with PCM. Our results suggest cage geometry has less influence on PCM, and both kidney-shaped and bullet-shaped cage designs were viable option when surgeons inserted an implant with appropriate size and enough depth.
Segmental instability of the lumbar spine resulting from degenerative or spondylolytic spondylolisthesis are common indications for spinal fusion surgery. We examined the influence of segmental stability on PCM and measured several radiographic parameters in this series. There was significantly more sagittal translation in discs developing PCM compared to the control group in the univariate model. However, there was no significant difference between discs with migrated and non-migrated cages in the PCM group. The multivariate logistic model showed non-significant results for sagittal translation. Previous studies also reported conflicting results. Aoki et al found no significant difference in translation, slippage or ROM, while Kimura et al reported more ROM in discs with PCM [3, 6]. We speculated that it was not slippage percentage in standing lateral radiographs, but sagittal translation in dynamic images that might influence PCM. Further studies with larger sample sizes are needed to verify this hypothesis.
Significantly longer fusion levels were found for patients with PCM in univariate analysis, while fusion level was a non-significant factor in the multivariate logistic model. In the present study, we found that, for patients receiving multilevel fusion in the PCM group, PCM occurred mostly (15 of 21, 71.43%) at the end level of the fusion. As gravity transmits vertically, there is large cantilever bending torque applied at the end levels, possibly resulting in a higher failure rate. Kimura et al reported significantly longer fusion levels in all 9 cases with PCM. All of them were multilevel fusions, and PCM occurred at the end disc [6]. Based on these results, more caution is required for multilevel fusions, and additional effort should be put into end level stability.
There are several limitations in this study. First, although this study had more patients with PCM than any other study, the number of patients was still relatively small. Second, the study was conducted retrospectively without randomization. Third, large proportions of patients (11.82%) were excluded due to short follow-up, even though all of their latest radiographs showed stable cages without signs of posterior migration. Fourth, in addition to cage position and cage height, the spine sagittal alignment parameters are possible risk factors that yet to be determined. These parameters were not included in this study because whole spine radiographs were not available during the period of case collection. Further study is needed to investigate the role of sagittal alignment parameters in development of PCM.