INTRODUCTION
Interbody fusion remains a cornerstone treatment for degenerative spine conditions, providing mechanical stability, disc height restoration, neural decompression, and bony arthrodesis [1-3]. Among various approaches, transforaminal lumbar interbody fusion (TLIF) balances surgical access with biomechanical stability while attempting to maximize vertebral endplate contact [4,5]. Static (nonexpandable) TLIF implants or TLIF implants that expand in only one dimension to provide added disc height or lordosis allow for interbody placement through constrained approaches, but face inherent limitations in their ability to conform to the anatomical contours of vertebral endplates [3,4,6,7]. These limitations may result in suboptimal endplate coverage, potentially leading to stress points and inadequate load distribution, which may further increase subsidence risk and compromise fusion rates [8-10].
Recent development of 3D-expandable TLIF implants capable of multidimensional expansion in situ has aimed to address the limited endplate coverage attainable with static TLIF devices. Unlike static or single-axis expandable implants that increase only in height while maintaining a fixed footprint, these devices aim to optimize both disc height restoration and endplate coverage [11-14]. While preliminary studies suggest 3D-expandable implants may reduce subsidence complications compared to static alternatives [15-17], quantitative analysis of their endplate coverage remains unknown. Furthermore, no published literature has directly compared endplate coverage achieved between various TLIF implants, as well as against implants that are traditionally larger such as anterior lumbar interbody fusion (ALIF) [18].
Maximizing endplate coverage has emerged as a critical factor in successful interbody fusion, with direct implications for load distribution, subsidence prevention, and fusion potential. This study quantitatively analyzes the vertebral endplate coverage achieved by a novel 3D-expandable TLIF implant compared to static TLIF and ALIF implants. Using standardized radiographic measurements of implant-to-endplate contact area ratios, we test the hypothesis that 3D-expandable TLIF technology achieves superior endplate coverage compared to static TLIF implants, potentially approaching coverage rates historically available only through ALIF procedures. These findings have direct clinical relevance for surgical decision making and may challenge current paradigms regarding approach selection for optimal biomechanical and clinical outcomes in lumbar fusion procedures.
MATERIALS AND METHODS
1. Patients and Data Collection
We retrospectively analyzed patients undergoing TLIF or ALIF procedures by the senior author between 2014 and 2022. Patient demographic information and radiographic data were collected via review of the electronic medical record and postoperative radiographic studies. All procedures were performed at a single academic medical center. Inclusion criteria consisted of patients who underwent either single or multilevel, elective lumbar interbody fusion using either a static TLIF, 3D-expandable TLIF or ALIF implant. Patients without postoperative computed tomography (CT) imaging and patients where the area of the implant did not fully overlap with the vertebral body surface were excluded. From the complete list of patients in this period, random number generation was utilized to choose equal sample sizes between implant type. All patients had postoperative CT images.
2. Operative Technique
Anterior approach surgeries were performed with an abdominal access general surgeon. Patients were positioned supine on a flat Jackson table (Mizhuo OSI, USA) with arms 90° at the shoulder. The anterior approach was completed by the access surgeon, and once the disc space of interest was marked and confirmed with intraoperative fluoroscopy, the disc space preparation was performed. Once complete, sequential trials were placed under fluoroscopic guidance to confirm ideal implant size and geometry. The permanent implant was then placed under fluoroscopic guidance, with the use of biologics and/or autograft when appropriate. Robust hemostasis and standard closure of the abdominal incision were then carried out.
Posterior approach surgeries using static and 3D-expandable implants were performed in similar fashion. Patients were placed prone on either a regular table with a Wilson frame or an open Jackson table (Mizhuo OSI) with arms forward and 90° at the shoulder and elbow (i.e., “superman” pose). The surgical level was identified using intraoperative fluoroscopy and tubular dilation was performed to the facet/lamina in line with disc space of the level desired. Unilateral laminectomy/facetectomy were then performed and once the disc space was identified, discectomy and disc space preparation were carried out. Using fluoroscopic guidance, sequential trials were then placed into the disc space to determine final implant size and geometry. A static or 3D-expandable TLIF implant was then placed based on specific patient clinical characteristics or anatomy. Biologics and autograft were used when appropriate. Following implant placement, bilateral pedicle screw fixation was performed in percutaneous fashion. Robust hemostasis and standard closure of all incisions were then completed.
Final implant size was determined intraoperatively at the discretion of the surgeon. Visual inspection, intraoperative fluoroscopy, and tactile feedback were used to determine the chosen implant size.
3. Imaging Evaluation and Area Calculation
Radiographic assessment was conducted using postoperative CT imaging to obtain precise measurements of implant and vertebral surface areas by an experienced author. The operative level referenced in the operative report was confirmed on the postoperative CT. Endplate dimensions were measured in the anteroposterior (AP) dimension on a midline sagittal bone window CT, and the mediolateral (ML) dimension on either disc-space specific axial sequences or a coronal bone window CT (Fig. 1). The area of the superior and inferior endplates was then calculated using an ellipse approximation (Fig. 2).
For static TLIF and ALIF implants, the dimensions in the AP and ML axes were measured on postoperative axial bone window CT (Fig. 3). Given the inherent error of measurements on CT imaging with metallic artifact present, if the measured AP or ML axes were larger than the vendor specified dimensions of the implant, the maximum allowed value for the dimension was that of the vendor specification.
For the 3D-expandable TLIF implant, a circular approximation (determined in consultation with a lead designer of the device) was used by measuring the diameter of the implant on the axial bone window CT and adding 2 mL in order to account for radiolucent components of the implant (Figs. 4 and 5).
The coverage ratio was defined as the ratio of twice the implant area to the sum of the superior and inferior vertebral endplate areas. All length and width measurements were measured in units of millimeters (mm) and all area measurements were calculated in units of mm2.
4. Statistical Analysis
Data was collected and stored in a secure database using Excel (Microsoft, USA). Continuous variables were reported with mean and standard deviation. Categorical variables were reported with frequencies and percentages. Continuous variables were analyzed using 1-way analysis of variance (ANOVA; normally distributed) or Kruskal-Wallis (nonnormally distributed). Where continuous variable analysis revealed statistical significance, post hoc pairwise comparisons were performed using Tukey honestly significant difference (HSD) test. Categorical variables were analyzed using chi-square or Fisher exact tests. Subgroup analyses were conducted for each operative level (L3–4, L4–5, and L5–S1) to evaluate differences in coverage ratios among the 3 cohorts. Analysis of covariance (ANCOVA) and least-squares means were used in analysis when patient anatomic variations may have confounding influence. Statistical significance was defined as p<0.05 for all tests. Analyses were performed using Python.
5. Reliability
To assess measurement reliability, 10 cases were randomly chosen from each of the 3 cohorts for a total of 30 cases (20%) and 170 individual measurements (including implant dimension, endplate size, etc.). A second, experienced observer then performed measurements on this subset in a blinded fashion. Reliability for continuous variables was quantified using the intraclass correlation coefficient (ICC [2,1], 2-way effects, absolute agreement, single-measure model) for interobserver reliability. Agreement was further examined with Bland-Altman analysis (mean bias and 95% limits of agreement). Excellent reliability was predetermined as >0.90.
RESULTS
A total of 159 patients were included in the study. The static TLIF, 3D-expandable TLIF, and ALIF cohorts each had 53 patients. The baseline characteristics of the 3 groups are summarized in Table 1. There were no statistically significant differences in age (p=0.934) or sex distribution (p=0.283) between the cohorts. However, a significant difference was noted in the distribution of operative levels (p=0.03), with ALIF procedures more commonly performed at L5–S1, whereas both static and 3D-expandable TLIF implants were most frequently placed at L4–5. Two-level surgeries were higher in the ALIF group (p<0.001).
Endplate coverage metrics for each implant type are detailed in Table 2. Implant area differed significantly between groups, with the largest implant area observed in the ALIF cohort (882±188 mm²), followed by the 3D-expandable implant (640±92 mm²), and the smallest area recorded in the static TLIF implant (268±51 mm²). Average endplate area was significantly smaller in the static TLIF group compared to the 3D-expandable and ALIF groups. The 3D-expandable and ALIF cohorts did not significantly differ in endplate area. The coverage ratio, which relates the area of the vertebral body covered by the implant, was highest in the ALIF group (0.46±0.13), followed by the 3D-expandable implant (0.35±0.06), and lowest in the static TLIF cohort (0.18±0.04). ANOVA revealed statistically significant differences for all coverage metrics (p<0.001).
Subgroup analyses further explored coverage ratios at specific operative levels. Across all operative levels, the trend in the coverage ratio remained consistent with ALIF being greatest, followed by the 3D-expandable TLIF implant and then the static TLIF implant. At L3–4, results for the comparison between coverage ratios between implant type were not significant given the small sample size, but they were significant for both L4–5 and L5–S1 (Table 3). Post hoc analysis using Tukey HSD test confirmed that all pairwise comparisons for the coverage ratio were statistically significant (Table 4), and showed ALIF superior to both TLIF implants, and 3D-expandable TLIF superior to static TLIF.
Multivariate analysis was carried out to account for the statistically significant difference in average endplate area amongst the static TLIF cohort. ANCOVA testing showed that after accounting for average endplate area, the implant type remains the driving factor for the coverage ratio (R2=0.80; Table 5). Among covariates, inferior endplate area was weakly, independently associated with coverage (p=0.002). After controlling for endplate area, adjusted mean coverage ratios (Table 6) for static TLIF, 3D-expandable TLIF and ALIF were 0.13, 0.36, and 0.49, as compared to the observed 0.18, 0.35, and 0.46. Adjusted pairwise comparisons (Table 7) showed that all coverage ratio relationships persisted. These results highlight the independence that implant type has from anatomic variability (i.e., average endplate area) on the endplate coverage ratio.
Among 170 individual measurements made, interobserver reliability was excellent. Comparing 170 measurements between the primary reviewer and a blinded, experienced observer (total 340 measurements), the ICC was equal to 0.97, indicating an excellent interobserver reliability. Bland-Altman analysis showed a mean bias of -0.64 and 95% limits of agreement spanning -9.4 to 8.1. Overall, systematic bias was minimal and random variation between observers compact.
DISCUSSION
This study presents the first quantitative radiographic comparison of vertebral endplate coverage of a novel 3D-expandable TLIF implant as compared to static TLIF and ALIF implants. The patient cohorts were evenly matched across key demographic factors of age, sex, and multilevel surgery. However, a statistically significant difference in operative level distribution was noted, with ALIF procedures more frequently performed at L5–S1, and TLIF procedures more commonly at L4–5. This reflects common surgical decision making, where ALIF is generally preferred at L5–S1 due to anatomical favorability allowing safe and efficient access through an anterior corridor [4,19]. TLIF is more routinely favored at L4–5 due to ease of the posterolateral approach and avoidance of midline vascular structures at higher lumbar levels. Given the difference in operative level distribution between the cohorts, subgroup analysis by operative level helped account for these differences and validated the core findings of the study; the 3D-expandable TLIF implant remained superior to static TLIF at all 3 levels. Lack of significance at the L3–4 level is explained by the low sample size leading to high variability between observed measurements at that level.
Significant differences were also seen in the anatomic measurements of patient endplate areas. The static TLIF cohort had significantly smaller endplate area measurements for both the superior and inferior endplates as compared to the 3D-expandable and ALIF cohorts (which were similar between each other). Multivariate analysis accounting for average endplate area showed that the mean coverage ratios between static TLIF, 3D-expandable TLIF, and ALIF continued to be statistically significant, as well as their differences when compared directly. It follows logic that, given fixed observed values of implant area for static TLIF, if the average endplate area overall for the cohort, compared to the other cohorts, is smaller, the representative coverage ratio calculated would be unrepresentatively higher. We see this in the ANCOVA analysis accounting for average endplate size in that the adjusted coverage ratio drops to 0.13 from 0.18 for static TLIF.
The findings in this study support that 3D-expandable TLIF implants may maximize the implant-endplate interface over conventional static TLIF implants and approach comparable footprints to ALIF implants. 3D-expandable TLIF implants may accomplish this while obviating the need for anterior approaches; in other words, they may allow for added surgical confidence in performing posterior-only approaches by offering large implant sizes through the same surgical corridor. This becomes relevant in select cases, as surgeons may be able to avoid multistage, multiposition procedures. For example, a multilevel construct to correct deformity can be entirely from a posterior approach with the confidence that a 3D-expandable TLIF implant can achieve comparable endplate coverage to that of an ALIF implant; or for single-level degenerative pathology, a planned ALIF with posterior instrumentation to reduce a spondylolisthesis and provide additional fixation may be substituted for an all-posterior approach utilizing a 3D-expandable TLIF implant. This technology does not eliminate the time and place for multistage, multiposition surgeries, but it may enable surgeons to avoid these more involved procedures in carefully selected cases with improved efficiency, safety and comparable outcomes.
1. Implant Biomechanics and Implant Design
The biomechanical interface between the implant and vertebral endplate is pivotal in achieving solid arthrodesis and maintaining spinal alignment. Bone quality, influenced by density and biophysical composition, and the interaction of implants with bony surface are paramount for structural stability [20,21]. Endplate integrity after surgical preparation is essential for resisting subsidence and promoting uniform load sharing across the fusion construct [22]. Undersized or poorly fitting implants can concentrate stress at focal points, predisposing to implant migration, cage subsidence, or pseudarthrosis [20]; these findings have been shown to occur in both the lumbar and cervical spine [23,24]. Various finite element analyses have shown that implants with larger area led to decreased internal stress or fewer stress peaks (i.e., focal points) within the vertebral endplate [20,25,26]. Since the 3D-expandable TLIF implant showed significantly larger coverage of the endplate as compared to static TLIF implants, it would indicate improved load distribution and decreased risk of focal stress injury to the endplate as it nears the coverage achieved by ALIF.
Placement of the implant within the disc space and where it rests on the endplate is also of significance. The epiphyseal ring is considered the strongest aspect of the cortical endplate surface and lies around its perimeter. Various biomechanical studies have shown that implants placed along this ring, as opposed to the center of the endplate, provide higher stress tolerance and lower subsidence risk [27,28]. The surgical approach in ALIF lends itself to placement of the implant along the anterior epiphyseal ring, while posterior approaches rely heavily on disc space preparation and intraoperative fluoroscopy or other imaging modalities to drive specific implant placement. Since the 3D-expandable TLIF implant in this study provides greater coverage of the endplate as compared to static TLIF, and is expanded in situ within the disc space, it may provide greater ease and likelihood of placement along some portion of the epiphyseal ring, thus providing greater stability and stress tolerance.
2. Approaches to Lumbar Interbody Fusion and Surgical Outcomes
Multiple surgical approaches exist for achieving lumbar interbody fusion, including posterior lumbar interbody fusion, TLIF, ALIF, and lateral lumbar interbody fusion. Each technique offers distinct advantages and limitations based on anatomic accessibility, risk profile, and ability to restore sagittal alignment and disc height [4,29,30]. TLIF is versatile as it can be performed open or in a minimally-invasive fashion, unilateral or bilateral, and with new technology allowing for different implants [31-33]. Previous studies have shown that expandable TLIF cages are associated with reduced intraoperative neural retraction, greater postoperative disc height, and improvements in patient-reported outcomes when compared to static devices [9,15,34]. ALIF remains a strong favorite, given the broad disc space and endplate exposure, potential for significant correction of lordosis, and the avoidance of posterior soft-tissue disruption.
In the present study, the ALIF group demonstrated the greatest endplate coverage, consistent with the technique’s anterior access and ability to accommodate larger implants. Notably, the 3D-expandable TLIF implant approached ALIF-level coverage at all levels (though not reaching significance), despite using a posterior approach. This suggests that newer expandable interbody designs may help bridge the biomechanical advantages of ALIF with the safety and familiarity of TLIF, especially when anterior access is contraindicated. Understanding the effect of operative level on clinical outcomes may shed light on surgical decision making. One study looking at patient-reported and radiographic outcomes between TLIF and ALIF compared to operative level (L4–5 or L5–S1), found that TLIF outperformed ALIF in Oswestry Disability Index at L4–5, and ALIF outperformed TLIF in radiographic outcomes at L5–S1 [35]. Tye et al. [36] and Lightsey et al. [37] showed that for isthmic spondylolisthesis at L5–S1, when comparing ALIF with or without posterior instrumentation versus TLIF, ALIF was superior in patient-reported outcomes at long-term follow up, as well as various radiographic outcomes. These studies show that operative level likely involves multiple factors in regard to both patient and radiographic outcomes. Having comparable implants in terms of endplate coverage may lend more weight to these addition considerations.
3. Subsidence and Complications
The issue of implant subsidence remains one of the feared complications of interbody fusion. While clinical outcomes were not directly assessed in this study, the established correlation between increased implant-endplate contact and reduced cage subsidence with better maintenance of alignment lends clinical relevance to these findings [38,39]. In a systematic review of various lumbar interbody techniques, ALIF has been shown to potentially have the lowest rate of subsidence, with TLIF having the highest reported rate – upwards of 50% [40,41]. By increasing load sharing and minimizing focal stress points with a larger implant and endplate coverage, it stands to reason that ALIF subsidence would be lowest as it again provides the greatest endplate coverage and often can be directly placed along the epiphyseal ring. One study looking at factors affecting implant subsidence in posterior lumbar fusions found that larger vertebral body area was significantly associated with subsidence [41]. Larger vertebral body area would equate to a lower coverage ratio assuming the implant is unchanged. Similarly, You et al. [42] found that larger implant size in endoscopic TLIF procedures led to decreased rates of subsidence. The results of the presented study suggest that the use of 3D-expandable TLIF implants may help achieve subsidence rates like those of ALIF, rather than static TLIF implants, given the coverage ratios are similar. It should be noted that various studies have shown comparisons of unidimensional expandable TLIF versus static TLIF implants with rates of subsidence higher in the expandable groups [34,43]. These studies cite endplate violation as a potential primary cause, but importantly the rates of revision surgery or nonunion remained less in the expandable TLIF groups. Overall, the topic of implant subsidence continues to be contested. One systematic review looking at banana-shaped versus straight bullet-shaped TLIF implants found no difference in subsidence rates, even though the design and structural contact of the 2 implants vary [44]. The problem of implant subsidence is a multifactorial issue which will continue to warrant clinical investigation and study.
4. Study Limitations and Conclusions
This study is limited by its retrospective design and reliance on radiographic analysis rather than direct clinical outcomes. Prior studies have specifically compared clinical outcomes, as well as segmental radiographic measurements, of the 3D-expandable TLIF implant used in this study [15]. While endplate coverage is a meaningful biomechanical metric and the focus of this study, complementary prospective studies are needed to correlate these findings with long-term fusion rates, functional scores, and complication rates. Additionally, implant sizing and expansion protocols may vary between surgeons, so this single series may not reflect the implant area and coverage ratio achieved by other surgeons, using any of the implants in the present study. While interobserver reliability was excellent, it does not distract from the fact that measurements made from CT imaging have inherent limitations on the resolution and tolerance of the measurement tools. This introduces a degree of error that may not be consistent across imaging platforms or institutional imaging capabilities.
By combining a posterior approach with near-anterior-level endplate coverage, 3D-expandable TLIF implants present a compelling benefit to conventional static or unidimensional expandable TLIF implants, as well as a compelling alternative in patients where ALIF or is not feasible. Future comparative studies integrating clinical outcomes, cost analysis, and patient-reported metrics will be essential to guide evidence-based decision making. As implant technology continues to evolve, optimizing implant-endplate conformity without increasing surgical morbidity will remain a key goal in lumbar interbody fusion surgery.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Full text via PMC
Download Citation
Print
