IN WHICH PATIENTS IS IT POSSIBLE TO PERFORM STANDALONE LATERAL LUMBAR INTERBODY FUSION WITHOUT CAGE SUBSIDENCE ?

Objective: To identify the factors related to the non-occurrence of cage subsidence in standalone lateral lumbar interbody fusion procedures. Methods: Case-control study of single level standalone lateral lumbar interbody fusion (LLIF) including 86 cases. Patients without cage subsidence composed the control group (C), while those in the subsidence group (S) developed cage subsidence. Preoperative data were examined to create a risk score based on correlation factors with S group. The proven risk factors were part of an evaluation score. Results: Of the 86 cases included, 72 were in group C and 14 in group S. The following risk factors were more prevalent in group S compared to C group: spondylolisthesis (93% vs 18%; p<0.001); scoliosis (31% vs 12%; p=0.033); women (79% vs 38%; p=0.007); older patients (average 57.0 vs 68.4 years; p=0.001). These risk factors were used in a score (0-4) to evaluate the risk in each case. The patients with higher risk scores had greater subsidence (p<0.001). Scores ≥2 were predictive of subsidence with 92% sensitivity and 72% specificity. Conclusions: It was possible to correlate the degree of subsidence in standalone LLIF procedures using demographic (age and gender) and pathological (spondylolisthesis and scoliosis) data. With a score based on risk factors and considering any score <2, the probability of non-occurrence of subsidence following standalone LLIF (negative predictive value) was 98%.


INTRODUCTION
2][3] Biomechanical features of LLIF constructions provide better rigidity than other constructions. 4,5It uses large diameter, hollow spacers that can engage the peripheral margins of the endplate, the design and positioning of which seems to be superior to others. 6In conventional lumbar fusion techniques, the interbody cage is always supplemented by internal fixation.In LLIF, the anterior and posterior longitudinal ligaments (ALL and PLL, respectively) are preserved.For this reason, a stand-alone construction can provide good stabilization, comparable to a TLIF supplemented with pedicle screws. 4Good results and complications from non-supplemented LLIF have been reported; [7][8][9][10][11] however, the ideal indication is not clear.
Cage settling and vertebral body fracture are potential and significant complications.When using LLIF standalone constructions, this risk may be even higher.The severity of cage subsidence is correlated with its extension. 123][14][15][16] Prevention of cage subsidence is, therefore, an issue of great research interest.
Some study groups have recently discovered that the use of wider LLIF cages results in more stable biomechanical constructions 4 and decreased subsidence rates. 12,14,17Based on this data, the use of 22mm or even wider cages is now recommended.Even more recent evidence has revealed that impaired bone quality 18 and inadvertent intraoperative endplate damage 19,20 can contribute to a poorer outcome.Despite the lack of research on cage settling issues, the case selection for standalone LLIFs that do not evolve to subsidence needs elucidation; the objective of this work was to identify the factors correlated with lack of cage subsidence in standalone LLIF procedures.

METHODS
A single center retrospective case-control study with a database spanning the years 2008 to 2015.This observational study was reviewed by the research ethics review board of our institution.The informed consent was waived because it was a retrospective research involving medical records.Inclusion criteria: single-level stand-alone lumbar LLIF.Exclusion criteria: cages with 18mm anteroposterior dimension (due to their demonstrated inferiority; 12,14,17 any previous lumbar arthrodesis/ arthroplasty surgery; any kind of supplementation (posterior/lateral/anterior).The interbody cages were packed with synthetic bone substitute.
The patients were grouped into Control group (C) and Subsidence group (S), according to the degree of interbody cage settling.Subsidence grading was evaluated in 3-month lateral orthostatic radiographs and classified on a 4-point scale (0-III) described previously 12 (Grade 0, 0%-24%; Grade I, 25%-49%; Grade II, 50%-74%; and Grade III, 75%-100% collapse of the level).The choice of the 3-month time point was based on a previous 12-month follow-up study in which patients who presented subsidence at the 12-month time point already presented a detectable subsidence in the 3-month evaluation.The C group comprised grade 0 and I cases (low-grade), while the S group had the grades II and III cases (high-grade), this division was based on the clinical impairment (pain outcome) that had already been demonstrated 12 in the S group, in comparison with the C group.
The preoperative diagnosis and images were reviewed, to identify the presence of spondylolisthesis and scoliosis.The intraoperative fluoroscopy images and surgical reports were reviewed aiming to identify the occurrence of unintentional anterior longitudinal ligament (ALL) rupture or cortical bone breaks during the preparation of the intervertebral space.These cases were excluded from the analysis.The anteroposterior dimension and height of the cage were measured.The influence of each factor in each subsidence grade subgroup was analyzed.A risk score was established based on preoperative factors with significant correlation with cage subsidence.
Statistical analyses were performed using SPSS software (SPSS, Version 10, SPSS, Chicago, Ill., USA).The Mann-Whitney U test was used for unpaired data, analysis of variance (ANOVA), Fischer's exact test, and the Chi-square test with Pearson's correlation were used with an alpha of 0.05.
Surgical factors were also compared between groups.Non-intentional ALL rupture occurrence was too low in this sample, and no significance was seen (1 vs 1 case; p=0.301).Another factor that did not show a statistical trend for either the S or C group was cage AP dimension (p=0.351;22 vs 24 vs 26mm; Table 2).The cage height was observed to be higher in the S group than in the C subset (9.8 vs 10.9mm; p=0.035), while in the S group 50%, of the cases had 12mm cages and only 22% had 12mm cages in the C group.
To investigate whether there was a cumulative effect of the individual risk factors found above, the cases were scored by adding one unit for each risk factor found.The SCORE was comprised of the following criteria: female gender; spondylolisthesis; scoliosis; age > 61 years (confidence interval upper bound value from C group = 60.5 years).Minimum SCORE = 0 and maximum SCORE = 4. Accordingly, it was demonstrated that higher SCORES were correlated with higher severity of subsidence.(Table 3; p<0.001)Table 4 shows the association between exposure (SCORE ≥ 2) and outcome (subsidence).The specificity and sensitivity of a SCORE < 2 for predicting which case would not develop high-grade cage subsidence are 75% and 93%, respectively.So, for any particular positive SCORE result (≥2) in our sample, the probability that it is true positive was 42% (positive predictive value; PPV).For any particular negative SCORE result (<2), the probability that it is true negative was 98% (negative predictive value; NPV).

DISCUSSION
Standalone LLIF procedures are feasible, but carry the risk of cage subsidence.Cage selection to prevent subsidence is only now beginning to be understood.The present report shows subsidence risk factors in a retrospective case-control study in a LLIF cohort with only single-level standalone cases.Preoperative risk factors were identified: elderly and female patients, and also diagnoses of scoliosis and spondylolisthesis at the index level.A score based on those factors was applied, to determine in which case a standalone indication would evolve without subsidence.The test showed great sensitivity, with extremely low false negative test results (2%).
Lumbar interbody fusion performed with lateral implants has been shown to be biomechanically superior to other constructions, 4,5 as the larger diameter hollow spacers appear to be favorable for engaging the peripheral margins of the end plate. 6The good stiffness achieved with lateral wide implants enables standalone constructions to be performed in some cases.However, it remains an off label indication to date.The selection of the appropriate instrumentation option is still a topic of debate.Recently, Tohmeh 21 published a video lecture with a preliminary scoring system to guide the choice of internal spinal fixation modality.Based on a literature review and author's opinion, this algorithm attributes scores according to the severity of the spondylolisthesis, degenerative disc, facet disease and adjacent segment disease.
The occurrence of cage settling generates radiological and clinical consequences.In the radiological field, Tohmeh et al. 17 and Marchi et al. 12 demonstrated loss of both disc and foramen height, decreased lordosis gain, but no negative effects on bone fusion.Inadvertent endplate fracture following lateral cage placement is a fracture that can impair the biomechanical stability. 19Although there is no congruent data regarding the clinical impact of subsidence, 7,9,16 some groups have found poorer clinical outcomes in short, 12,15 medium 8 and long-term 17 follow-up for subsidence in LLIF.The short-term results were attributed to transient painful micromotion, in addition to pain from the fracture itself.Moreover, along with the loss of correction, restenosis may occur in high-grade subsidence, and there is the need of further direct decompression.Including the results from the current work, the rate of re-intervention due to subsidence varies from 0% to 50%, depending on the supplementation option chosen.Vertebral fracture is rare, but possible. 15Considering the possible consequences of cage settling, the topic must be better understood in order to avoid subsidence.The first strong evidence of subsidence prevention in LLIF came from retrospective comparative studies 12,14 which demonstrated that cages with a larger anteroposterior diameter (22 x 18mm) could decrease the rate of cage settling.It is now known that other intraoperative features also influence the occurrence of subsidence: endplate damage during preparation and overdistraction of the disc space.Santoni et al. 19 reported the harmful role of intraoperative endplate damage in a cadaveric model.Accordingly, a clinical study 17 demonstrated that the magnitude of subsidence is lower if the cage does not sink into the endplates during surgery.Taller interbody grafts are correlated with higher distractive and compressive forces in vitro. 22In clinical practice, we observed more cases with 12mm-height cages in subsidence versus the control group, as was also seen by other reports on LLIF. 17,20Other possible issues include insufficient contralateral annular release, and improper cage sizing in the lateral dimension, which may result in failure to cover the apophyseal ring.
A significant trend of increasing subsidence rates has been correlated with LLIF with longer constructions 14 and more instable pathologies. 10,11,23Another pathology-related criterion has been studied in LLIF subsidence: poor bone quality.Similarly to the present study, Tempel et al. 18 conducted a case-control study to analyze bone mineral density (BMD) in cases that evolved with subsidence following LLIF.They found that a DEXA T score of less than -1.0 predicts subsidence, with sensitivity and specificity of 78.3 and 63.2%, respectively.Female sex and age may be potential confounders of the association between osteoporosis and cage subsidence, as the incidence of osteoporosis and osteoporotic fractures increases markedly with age and female sex. 24Along with poorer bone quality, spinal muscular atrophy could negatively affect cage stabilization and spinal bone mineral density in the elderly postmenopausal population. 25ome limitations of this study must be pointed out.(1) The major limitation is inherent to the study design: a retrospective study.However, it is feasible as a preliminary investigation of a suspected risk factor, and sometimes a case-control study is the only ethical way to investigate an association.(2) The study only analyzed single-level constructions.Readers must, therefore, be cautious when extrapolating the conclusions for multi-level fusion.(3) The DEXA scan results were not available for all patients, so they could not be included in the analysis.(4) The results are based in a single-center experience.

CONCLUSION
The present study correlated cage subsidence in stand-alone LLIF procedures with the following preoperative risk factors: age, gender, spondylolisthesis and scoliosis.With a score based on those risk factors, in our sample, the probability of a particular stand-alone LLIF procedure with a score of <2 evolving without subsidence was 98%.This could be the first step towards the understanding of case selection for LLIF with no internal fixation.

Table 1 .
Preoperative risk factors analysis per study subgroup.

Table 2 .
Intraoperative risk factors analysis per study subgroup.Values are represented as mean ± standard deviation or in absolute incidence within each subsidence grade subgroup.Cage height (Cage HT) and cage anteroposterior dimension (Cage AP).* statistically significant

Table 3 .
Score results per subsidence grade and study subgroups.
Values are expressed as mean ± standard deviation for each subsidence grade subgroup.* statistically significant