Procedural Complications and Implant Cutout of the SpineJack®-System in Thoracolumbar Vertebral Fractures: A Retrospective Analysis of 204 Implants from a Level-1 Trauma Center

Abstract

Study Design

Retrospective cohort study.

Objectives

Implant-related complications of SpineJack® stent-kyphoplasty, particularly endplate cutout, remain poorly characterized. This study determined the incidence of procedural complications, with a focus on cutout, and identified associated factors.

Methods

All patients treated with stand-alone SpineJack® stent-kyphoplasty at a level-1 trauma center (November 2014–December 2022) were retrospectively reviewed, excluding malignant fractures. In total, 204 implants were analyzed (129 OVFs, 75 VCFs). Intraoperative complications and postoperative endplate cutout, assessed on imaging after mobilization, were analyzed by fracture type, AO/OF classification, region and vertebral level.

Results

Mean age was 64.9 years (SD 17.5); OVF patients were significantly older than VCF patients (74.3 vs 48.8 years, p<0.001). Vascular cement leakage occurred in 2 cases (1.0%); no technical failures or conversions were recorded. Cement leakage occurred in 18 implants (8.8%) and cutout in 16 (7.8%), neither differing by fracture type. Cutout incidence differed significantly across OF grades (p=0.018), though not after excluding the small OF1 subgroup (p=0.211) or in multivariable analysis (OR 1.03, p=0.918), suggesting limited power; cutout was not correlated with cement leakage or vertebral level.

Conclusions

Stand-alone SpineJack® stent-kyphoplasty is associated with infrequent vascular and cement-related complications. Endplate cutout (7.8%) was numerically more frequent in severely collapsed osteoporotic (OF4) fractures, but this association was not robust to sensitivity or multivariable analysis, consistent with limited power rather than a true absence of effect; no cutout event in this cohort had a demonstrated clinical consequence. Prospective, adequately powered study is needed to confirm this association.

Introduction

Vertebral fractures are among the most common osteoporotic fragility fractures worldwide, with a rising absolute burden owing to population aging. ¹ Percutaneous vertebral augmentation — balloon kyphoplasty and stand-alone stent-kyphoplasty — is widely used for both traumatic vertebral compression fractures (VCFs) and osteoporotic vertebral fractures (OVFs). ² The landmark FREE trial established balloon kyphoplasty’s efficacy over non-surgical care, but also documented cement leakage in 27% of treated vertebrae. ³ Iatrogenic fracture and endplate perforation during balloon inflation are otherwise well described, with procedural complications reported in up to 17% of treated levels, ⁴ and endplate disruption has been linked to subsequent height loss, kyphotic deformity and cement leakage. ⁵

The SpineJack®-system is a mechanical, titanium implant-based stent-kyphoplasty device proposed as an alternative to balloon kyphoplasty, achieving controlled radial expansion of two titanium implants to distract the fractured endplate under direct mechanical control before cement augmentation — a mechanism proposed to reduce the risk of uncontrolled endplate injury. Pilot and comparative studies have suggested a comparable or favorable safety profile relative to balloon kyphoplasty, both short-term and over 3-year follow-up,^6,7 but neither specifically adjudicated endplate cutout as a distinct complication category. Implant-specific failure modes of the SpineJack® system — in particular, cutout of the implant through the vertebral endplate — therefore remain poorly characterized, and to our knowledge no dedicated analysis of cutout incidence and risk factors has previously been reported.

At our institution, we previously reported a low overall complication rate in a cohort of 310 SpineJack® implants used for both osteoporotic and non-osteoporotic vertebral fractures, without a dedicated analysis of cutout. ⁸ We therefore performed a focused, case-by-case review of procedural complications and postoperative imaging — with particular attention to endplate cutout — in patients treated with stand-alone SpineJack® stent-kyphoplasty at our level-1 trauma center, restricted to stand-alone procedures without additional posterior instrumentation and excluding pathological fractures due to malignancy. The aim of this study was to determine the incidence of procedural complications, in particular endplate cutout, and to identify fracture- and procedure-related factors associated with its occurrence.

Material and Methods

Patients

All patients with vertebral fractures who were treated with stand-alone percutaneous SpineJack® stent-kyphoplasty at the University Hospital of Zurich between November 2014 and December 2022 were retrospectively identified. Pathological fractures due to malignancy were excluded given their distinct underlying bone pathology and treatment goals. In total, 204 implants met these criteria and were included in this analysis, comprising 129 osteoporotic vertebral fractures (OVFs) and 75 traumatic vertebral compression fractures (VCFs) (Figure 1).

Figure 1.

Patient and implant flow diagram for the study cohort, from all SpineJack® implants placed during the study period to the 204 implants (180 patients) included in this analysis

Inclusion and Exclusion Criteria

Inclusion criteria: patients treated with stand-alone percutaneous SpineJack® stent-kyphoplasty for a painful thoracolumbar vertebral fracture at our institution during the study period; no upper or lower age limit was applied.

Exclusion criteria: fractures with a degree of instability requiring additional posterior instrumentation beyond stand-alone stent-kyphoplasty; pathological fractures due to malignancy (primary bone tumors or metastases); and pediatric fractures.

Endpoints

The primary endpoints were the incidence of intraoperative complications — technical implant failure, intraoperative conversion to an alternative procedure, vascular cement leakage, and non-vascular cement leakage — and, as a separate postoperative endpoint, endplate cutout, assessed on postoperative imaging obtained after mobilization; for cutout specifically, we further examined the fracture- and procedure-related factors associated with its occurrence. Vascular cement leakage was recorded as a binary complication (present/absent) rather than by quantified blood loss, consistent with departmental registry practice. No patient in this cohort underwent revision surgery during the study period; all procedures were primary, stand-alone interventions.

The present study is restricted to a focused, implant-level analysis of intraoperative complications and endplate cutout, based on case-by-case review of postoperative imaging not performed in our earlier 310-implant safety analysis, ⁸ and represents the first systematic characterization of cutout for this device at our institution.

Fracture Classification

Traumatic VCFs were classified according to the AOSpine thoracolumbar injury classification system, ⁹ and OVFs according to the AO Spine-DGOU Osteoporotic Fracture (OF) classification system. ¹⁰ Both classification systems have demonstrated moderate to substantial inter- and intra-rater reliability in multicenter validation studies.^11,12 Fracture classification was performed independently by two of the authors (M.K. and J.S.), with any disagreement resolved by consensus. Fracture etiology (osteoporotic vs. traumatic) was assigned based on injury mechanism, consistent with standard definitions of fragility fracture: fractures resulting from low-energy trauma (a fall from standing height or less) or no identifiable trauma were classified as osteoporotic OVFs, while those resulting from higher-energy trauma were classified as traumatic VCFs. ¹³

Procedure and Complication Definitions

All procedures were performed as stand-alone percutaneous stent-kyphoplasty without additional posterior instrumentation. Intraoperative complications — technical implant failure (e.g., failure of implant expansion), intraoperative conversion to an alternative procedure, vascular cement leakage, and non-vascular cement leakage — were recorded in the departmental procedure registry and confirmed by retrospective review of intraoperative fluoroscopic imaging. Vascular cement leakage — extravasation of cement into the venous vasculature (basivertebral or segmental vein) — was distinguished from non-vascular cement leakage (into the disc space, paravertebral soft tissue, or epidural space), consistent with established leakage-type definitions. ¹⁴ Endplate cutout was assessed on postoperative imaging obtained after mobilization, rather than intraoperatively, in order to also capture cutout developing after mobilization under physiological load, typically on postoperative day 2, supplemented by CT where clinically indicated; all cutout events in this cohort were identified at this timepoint. Cutout was defined as radiographic perforation of the superior or inferior vertebral endplate by the implant (Figure 2). Because ascertainment was postoperative, it could not be determined whether the perforation was already present intraoperatively but unrecognized at the time, or developed once the patient was mobilized and the spine bore physiological load; this case definition therefore captures cutout however it originated, without distinguishing between these two possibilities. The treated vertebral level was also recorded for each implant.

Figure 2.

Postoperative radiograph showing endplate cutout of the SpineJack® implant, obtained after mobilization. The implant struts extend beyond the vertebral body contour, perforating the endplate margin into the disc space. The image is an anonymized postoperative imaging frame without any patient-identifying information

Statistical Analysis

Statistical analysis was performed using SPSS® (IBM Corp., Armonk, NY). Data are presented as frequencies (n, %) for categorical variables and means with standard deviation (SD) for continuous variables. Normality of continuous variables was assessed using the Shapiro-Wilk test; as age and admission pain score were non-normally distributed in at least one comparison group (p<0.05), the Mann-Whitney U test was used for between-group comparisons of continuous variables. Differences between OVFs and VCFs in categorical variables were assessed using the Pearson chi-square test (uncorrected). Because several cells in the cross-tabulation of OF grade by cutout occurrence had expected counts below 5, the asymptotic chi-square approximation was considered unreliable for that specific comparison; an exact permutation test (20,000 Monte Carlo resamples) was used instead, as was Fisher’s exact test for the thoracolumbar-junction vs cutout comparison. The association between cutout occurrence and cement leakage was assessed using Pearson’s correlation coefficient. The level of statistical significance was set at p<0.05. To assess the robustness of the OF grade–cutout association given the small size of individual OF subgroups, two post hoc sensitivity analyses were performed at reviewer request: the exact permutation test was repeated after excluding the OF1 subgroup (n=3), and multivariable logistic regression was used to model cutout as a function of age, vertebral region, and fracture type (OVF vs. VCF) across the full cohort, and, separately, age, region, and OF grade (continuous) within the OVF subgroup.

Ethical Considerations

This study was carried out in accordance with the terms of reference of the local institutional ethics committee (Kantonale Ethikkommission Zürich, Switzerland; BASEC-Nr. 2017-00408).

Results

Demographics and Fracture Characteristics

Of the 204 implants analyzed, 129 (63.2%) were placed for OVFs and 75 (36.8%) for VCFs. The mean age of the overall cohort was 64.9 years (SD 17.5, range 19–93), with a male majority (51.5%). Patients with OVFs were significantly older than those with VCFs (74.3 years, SD 10.4 vs 48.8 years, SD 15.3; Mann-Whitney U, p<0.001); mean pain scores at admission were also higher in OVFs than VCFs (5.1 vs 4.3; Mann-Whitney U, p=0.035). Most fractures occurred in the lumbar spine (n=118, 57.8%), with the remainder in the thoracic spine (n=86, 42.2%); there was no significant difference in regional distribution between OVFs and VCFs (p=0.634) (Table 1).

Table 1.

Baseline demographic and fracture characteristics of the non-malignant cohort (n=204)

Characteristic	OVF (n=129)	VCF (n=75)	p-value
Mean age, years (SD)	74.3 (10.4)	48.8 (15.3)	<0.001
Lumbar region, n (%)	73 (56.6%)	45 (60.0%)	0.634
Cement leakage, n (%)	13 (10.1%)	5 (6.7%)	0.408
Endplate cutout, n (%)	13 (10.1%)	3 (4.0%)	0.120

Among the 75 VCFs, the most frequent AOSpine subtype was A3 (n=34, 45.3%), followed by A1 (n=25, 33.3%), A4 (n=11, 14.7%) and A2 (n=5, 6.7%). Among the 129 OVFs, the most frequent OF grade was OF2 (n=50, 38.8%), followed by OF4 (n=42, 32.6%), OF3 (n=32, 24.8%), OF1 (n=3, 2.3%) and OF5 (n=2, 1.6%).

Intraoperative and Postoperative Complications

Vascular cement leakage occurred in 2 cases (1.0%), both in the OVF group and managed conservatively with observation alone, without hemodynamic compromise or need for further intervention; no technical implant failures (e.g., failure of implant expansion) and no intraoperative conversions to an alternative procedure were recorded in this cohort. Intraoperative cement leakage was observed in 18 implants (8.8%), with no significant difference between OVFs (10.1%, 13/129) and VCFs (6.7%, 5/75; p=0.408). On postoperative imaging, endplate cutout was identified in 16 implants (7.8%), again with no significant difference between OVFs (10.1%, 13/129) and VCFs (4.0%, 3/75; p=0.120; p=0.177, given the small number of events in the VCF group).

Cutout occurred numerically more often in the thoracic spine (11.6%, 10/86) than in the lumbar spine (5.1%, 6/118), although this difference did not reach statistical significance (p=0.086). Cutout incidence differed significantly across OF grades (p=0.018; Table 2). The highest crude rates were observed in the single OF1 subgroup (66.7%, 2/3) and in OF4 fractures (16.7%, 7/42). Given the very small OF1 subgroup, this association is considered to be driven primarily by OF4 — the most severely collapsed common osteoporotic fracture pattern in this cohort — rather than by OF1. Among VCFs, cutout incidence did not differ significantly across AOSpine subtypes (p=0.273).

Table 2.

Endplate cutout by fracture classification

Classification grade	Cutout n/N (%)	p-value
OF1 (OVF)	2/3 (66.7%)	​
OF2 (OVF)	2/50 (4.0%)	​
OF3 (OVF)	2/32 (6.2%)	​
OF4 (OVF)	7/42 (16.7%)	​
OF5 (OVF)	0/2 (0.0%)	0.018*
AO1 (VCF)	1/25 (4.0%)	​
AO2 (VCF)	1/5 (20.0%)	​
AO3 (VCF)	1/34 (2.9%)	​
AO4 (VCF)	0/11 (0.0%)	0.273

Exact permutation test (20,000 resamples); the asymptotic Pearson chi-square for this table was 15.4 (df=4, p=0.004), but is reported as secondary given 6/10 cells with expected counts <5.

Because this association rested on two small OF subgroups relative to the modest total number of cutout events (n=16), its robustness was assessed in two ways. First, excluding the OF1 subgroup, the association between OF grade and cutout was no longer statistically significant (Monte Carlo exact test, χ²=5.16, p=0.211). Second, in multivariable logistic regression across the full cohort, fracture type was not an independent predictor of cutout (OR 0.79, 95% CI 0.14–4.32, p=0.786), nor was age (OR 1.05, 95% CI 1.00–1.11, p=0.056) or region (OR 0.43, 95% CI 0.15–1.24, p=0.118); within the OVF subgroup, OF grade modeled as a continuous variable was likewise not independently predictive (OR 1.03, 95% CI 0.55–1.95, p=0.918). Given the small number of cutout events, these findings are consistent with an underpowered rather than absent effect and should be interpreted as hypothesis-generating.

There was no significant correlation between endplate cutout and intraoperative cement leakage (r=0.038, p=0.591; only 2 of the 13 implants with cement leakage in the OVF group also experienced cutout, despite both complications occurring in exactly 13 OVF implants each — an apparent numerical coincidence rather than an overlap). Cutout events occurred across a range of thoracic and lumbar levels (most frequently T12, n=6; T9, n=3; and L4, n=3), without significant clustering at the thoracolumbar junction (T11–L2) compared with other levels (7.7%, 10/130 vs 8.1%, 6/74; p=1.000).

Discussion

In this retrospective analysis of 204 SpineJack® implants, non-cutout procedural complications (vascular and cement-related) were infrequent, consistent with previous reports of the safety of this device.^6,8 Endplate cutout, the focus of this analysis, occurred in 7.8% of implants and, while not significantly associated with fracture type or vertebral level, its incidence varied numerically across OF grades, with the highest crude rate among common fracture patterns occurring in OF4 fractures — an association that did not survive sensitivity or multivariable analysis (see below). It is important to emphasize that cutout in this study is a radiographic finding: no cutout event in this cohort was associated with pain, loss of correction, neurological compromise, implant migration, or revision surgery during the available follow-up, and cutout should not be equated with clinical implant failure in the absence of confirmatory long-term outcome data (see Limitations).

Our overall complication profile falls within the range reported for conventional balloon kyphoplasty, where cortical or endplate fracture has been reported in approximately 5% of treated levels and combined procedural complications or failures in up to 17% of levels ⁴ ; this is a descriptive comparison across non-randomized, non-contemporaneous series with differing case definitions, rather than a matched or head-to-head comparison. A recent systematic review and meta-analysis similarly found no significant difference in device-related complications between mechanical vertebral body augmentation systems, such as the SpineJack®, and conventional balloon kyphoplasty, although reduced cement volume was noted with mechanical systems. ¹⁵ Neither the original pilot comparison of kyphoplasty and SpineJack® nor its 3-year follow-up specifically reported cutout as an outcome, focusing instead on clinical and radiographic efficacy measures.^6,7 Our cutout rate of 7.8% therefore adds implant-specific granularity that has not previously been available for this device, in a field where cutout of mechanical stent-kyphoplasty devices has not previously been quantified in a dedicated analysis.

Notably, our own group’s earlier analysis of a larger, substantially overlapping cohort of 310 SpineJack® implants reported a low overall complication rate without a specific finding of endplate cutout. ⁸ This is not evidence that cutout did not occur in that series; it reflects that the earlier analysis did not define or systematically screen for cutout as its own complication category, and so was not designed to detect it. Its absence from that report is therefore a limitation of that study’s case definition, not a finding that contradicts the present one. The present study addresses this directly: it applies, for the first time in this registry, a specific, postoperative imaging-based case definition for cutout, and represents a distinct, focused research question rather than a re-analysis or duplicate publication of the same institutional cohort.

Cutout was numerically most frequent in OF4 fractures, which may reflect a biomechanical link between fracture severity and endplate integrity: more comminuted, structurally compromised endplates may be inherently more susceptible to perforation — whether during implant deployment or subsequently under physiological loading — independent of the level treated; however, as detailed above, this association did not survive sensitivity or multivariable analysis and should be interpreted cautiously. We found no clustering of cutout at the thoracolumbar junction (T11–L2) — a zone otherwise associated with greater fracture risk during backward falls ¹⁶ — suggesting that implant-related endplate failure is governed more by local fracture morphology than by regional biomechanical vulnerability. Given the small number of cutout events (n=16) across 14 vertebral levels, this finding is descriptive and hypothesis-generating; larger, pooled series are needed to confirm it.

The lack of correlation between cutout and cement leakage (r=0.038, p=0.591) suggests that these two complications arise through largely independent mechanisms — cement leakage relating principally to cortical wall integrity and cement viscosity/pressurization, whereas cutout relates to the mechanical interaction between the expanding implant and the endplate itself. This is broadly consistent with prior balloon kyphoplasty literature, in which the two complications have not been consistently linked. ⁴

These findings carry practical implications beyond the immediate procedural setting, which studies reporting only immediate procedural safety do not capture. A large endplate deficit is a recognized risk factor for construct instability after vertebral augmentation, alongside split-type fracture morphology and excessive segmental angular motion, and has been linked to late revision surgery in some series. ¹⁷ Should a cutout event translate into such a persistent endplate defect, it may carry downstream consequences for construct stability and adjacent segment loading; our exploratory findings raise the possibility that patients with OF4 fractures could represent a subgroup at elevated risk in this respect, although this requires confirmation in adequately powered studies, even though the present dataset does not include the longer-term follow-up needed to test this directly. In practice, this raises the question of whether preoperative assessment of endplate and cortical integrity — for example, on preoperative CT — could be used to identify patients in whom a modified deployment technique or an alternative augmentation strategy altogether might reduce cutout risk in the most severely collapsed fracture patterns.

Prospective work is needed to test these hypotheses directly. Preoperative CT-based Hounsfield unit measurements have recently been shown to predict non-infectious complications after posterior instrumentation for traumatic thoracolumbar fractures, ¹⁸ supporting a similar approach for cutout risk. Useful next steps include a prospective study correlating preoperative endplate and trabecular bone quality with cutout risk, systematic prospective screening for cutout across institutional SpineJack® series, and longer-term follow-up to determine whether cutout predicts subsequent height loss or adjacent segment fracture, as has been shown for endplate deficits more broadly. ¹⁷

Limitations

This study has several limitations. First, its retrospective design precludes causal inference. Endplate cutout was ascertained from postoperative imaging obtained after mobilization (typically postoperative day 2); ascertainment therefore depended on the availability of that early imaging, and patients without a retrievable study at this timepoint could in principle have been under-ascertained for this low-frequency event (n=16), which also limits the statistical power of subgroup analyses (e.g., the AOSpine subgroup comparison and the OF1 subgroup, n=3). Second, we do not have long-term clinical outcome data; most patients did not return for follow-up beyond the routine 6-week postoperative visit, in the absence of ongoing symptoms. Third, because cutout was ascertained on postoperative imaging rather than intraoperatively, our case definition cannot distinguish perforation occurring during implant deployment from delayed perforation developing after mobilization under physiological load, and the relative contribution of each mechanism remains unresolved. This also limits clinical actionability, since by the time cutout is identified on postoperative imaging, no corrective intraoperative maneuver is possible. Its principal value therefore lies in identifying fracture patterns at elevated a priori risk — such as OF4 — for consideration of a modified deployment technique or an alternative augmentation strategy before the index procedure. Finally, both the AOSpine and OF classification systems used for subgroup analysis have only moderate reported inter-rater reliability,^11,12 and this was a single-center study; both factors may limit the reliability and generalizability of our findings.

Conclusion

Stand-alone SpineJack® stent-kyphoplasty for thoracolumbar vertebral fractures is associated with infrequent vascular and cement-related complications. Endplate cutout, although uncommon (7.8%), was numerically more frequent in severely collapsed osteoporotic (OF4) fractures and showed no significant clustering by spinal region or vertebral level; however, this association did not survive sensitivity or multivariable analysis, consistent with limited statistical power rather than the absence of a true effect, and no cutout event in this cohort was associated with a demonstrated clinical consequence. Adequately powered, prospective study is warranted to determine whether fracture severity is a genuine risk factor for cutout, and to clarify the underlying biomechanics.

ORCID iDs

Julian Scherer https://orcid.org/0000-0001-8943-5817

Georg Osterhoff https://orcid.org/0000-0001-5051-0998

Ethical Considerations

This study was carried out in accordance with the terms of reference of the local institutional ethics committee (Kantonale Ethikkommission Zürich, Switzerland; BASEC-Nr. 2017-00408).

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.*