Early operative morbidity in 184 cases of anterior vertebral body tethering

Fusion is the current standard of care for AIS. Anterior vertebral body tethering (AVBT) is a motion-sparing alternative gaining interest. As a novel procedure, there is a paucity of literature on safety. Here, we report 90-day complication rates in 184 patients who underwent AVBT by a single surgeon. Patients were retrospectively reviewed. Approaches included 71 thoracic, 45 thoracolumbar, 68 double. Major complications were those requiring readmittance or reoperation, prolonged use of invasive materials such as chest tubes, or resulted in spinal cord or nerve root injury. Minor complications resolved without invasive intervention. Mean operative time and blood loss were 186.5 ± 60.3 min and 167.2 ± 105.0 ml, respectively. No patient required allogenic blood transfusion. 6 patients experienced major (3.3%), and 6 had minor complications (3.3%). Major complications included 3 chylothoracies, 2 hemothoracies, and 1 lumbar radiculopathy secondary to screw placement requiring re-operation. Minor complications included 1 patient with respiratory distress requiring supplementary oxygen, 1 superficial wound infection, 2 cases of prolonged nausea, and 1 Raynaud phenomenon. In 184 patients who underwent AVBT for AIS, major and minor complication rates were both 3.3%.


Materials and methods
All methods were carried out in accordance with relevant guidelines and regulations.
Patients. All patients were counseled on their surgery options and concomitant risks and benefits, as well as what remains unknown regarding AVBT including timing of surgery in relation to peak height velocity, and uncertain variables such as time until possible cord breakage. More skeletally mature patients who were offered AVBT were counseled about the lack of data for skeletally mature patients and the potential for future revision surgery, including PSF. Only after it was determined that the patient's priorities and expectations aligned with the potential advantages and drawbacks of AVBT was the surgery deemed appropriate. General indications included diagnosis of AIS, no prior spinal surgery, and no chronic co-morbidities.
Chart review. Approval for a retrospective chart review and waiver of informed consent was obtained from the Mount Sinai School of Medicine Institutional Review Board (IRB). 184 consecutive AVBT cases, performed by a single surgeon, were included. Patient medical records were reviewed for the following demographic and surgical characteristics: age, sex, estimated blood loss (EBL), rib resection, pre-operative Cobb angle, operative time, tether location, number of tether cords, number of levels tethered, and complications. Rib resection was analyzed as a yes/no variable, inclusive of individual ribs removed (1or 2) to facilitate screw placement and also multiple (3+) ribs (thoracoplasty) removed to address severe rib deformity. Data were collected by two trained research associates and validated by an attending orthopedic spine surgeon.
Complications. Complications were divided into eight domains consistent with previous studies: Pulmonary, Neurological, Gastrointestinal, Cardiovascular, Instrumentation, Pain, Wound/Infection, and Other [28][29][30] . Medical records in an EPIC Hyperspace database were analyzed for complications. Searchable keywords were identified for each complication domain and were used as a post-hoc verification tool. After complications were identified, they were further divided by time and severity. For descriptive purposes, perioperative complications were defined as occurring within 90 days of surgery. Major complications were defined as those that required readmittance or reoperation, prolonged the use of invasive materials such as chest tubes, or caused injury to a nerve root or the spinal cord. Minor complications were those that resolved without invasive intervention.
Statistical analyses. For continuous variables, normality was evaluated with a Kolmogorov-Smirnov test.
Parametric tests were used where distributions were normal and non-parametric tests were used when distributions were not normal (p < 0.05). First, demographic and clinical features were compared among patients who experienced any complications and those who did not, and then between patients who experienced major complications and those who did not (Table 1). A binomial logistic model was created with pre-operative Cobb angle, rib resection, number of levels tethered, and whether one or two curves were instrumented as independent variables, while complication outcome (dichotomous yes/no) was the dependent variable. This was done twice to produce an all-complication model and a major-complication model. These analyses generated odds ratios (OR) with 95% confidence intervals (CI). Analyses were executed with SPSS software, IBM Corp., Armonk, NY.   (Table 2). There were 22 patients (12.0%) who received a rib resection, and this was associated with a greater total complication rate, (4/22 = 18.2%), than patients who did not receive a rib resection (8/162 = 4.9%, p = 0.04, Table 1); these complications included 2 chylothoraces, 1 hemothorax, and 1 pulmonary insufficiency (Table 2). When parsed by tether location, a greater proportion of complications occurred in patients with 2 curves tethered (9/68 = 13.2%), as opposed to either only thoracic (1/71 = 2.2%) or only thoracolumbar (2/45 = 4.4%) curves (p = 0.01, Table 1). These complications included 2 hemothoracies, 2 pulmonary insufficiency, 2 prolonged nausea, 1 chylothorax, 1 superficial wound infection, and 1 instrumentation ( Table 2).

Results
In the all-complication binomial logistic model, tether location and rib resection were both associated with complications (p = 0.001 and p = 0.037 respectively). The model was statistically significant, χ 2 (4) = 15.9, p = 0.003, and correctly classified 92.4% of cases. Patients who received a rib resection were 4.6 times more likely to experience a complication than those who did not (95% CI 1.10-19.13). Patients with tethers on both thoracic and thoracolumbar curves were 13.2 times more likely to have a complication (95% CI 2.70-64.45) than patients who received a tether on only one curve. The number of levels tethered, and pre-operative Cobb angle were not significantly associated with complications (p = 0.082 and p = 0.321 respectively).

Discussion
In the present study, we report the 90-day post-operative complications in 184 consecutive cases of AVBT. Rib resection and two-instrumented curves appear to predict the all-complication rate, but only rib resection was significant for major complications. The all-complication rate was 6.5% and was evenly split between major (3.3%) and minor (3.3%) complications. The major complications were predominately pulmonary in nature (5/6 = 66.7%) with 3 chylothoracies and 2 hemothoracies.
Currently, PSF is the standard of care for patients with AIS, but alternatives such as AVBT have been introduced to sidestep some of the drawbacks of fusion. In order for stakeholders to appropriately balance their needs and concerns when opting for surgery, they must understand the costs, benefits, and risk profiles of those options. Reported benefits of anterior thoracoscopic approaches over fusion include less blood loss 17,18 , less muscle dissection 31 , less post-operative pain 32 , fewer levels instrumented 33 , and improved self-image and mental health SRS-22 scores compared to standard open thoracotomy approaches 34 . However, those same studies reported drawbacks including increased operative time, a steep learning curve, and increased incidence of pulmonary complications. Moreover, the long-term risks with AVBT are unknown, however cord breakage and revision have been reported 35,36 .
Without a PSF control group, we cannot directly compare our AVBT cohort to PSF, however it is worth noting that the complication profiles appear to be different. The proportion of major pulmonary complications This accords with previous studies that reported thoracoscopic anterior approaches are associated with a greater incidence of pulmonary complications relative to posterior approaches [11][12][13]17 , and may be an important factor when deciding between various surgical options. Chylothoracies made up a disproportionate amount of the major complications [3/6 (50%)] reported here. Two of the three cases occurred in patients in whom double cords were placed on one side of the spine and in one patient a disc release was performed. The common thread is that more extensive dissection for double-corded tethers or disc release may have led to the inadvertent injury of the thoracic duct or tributaries thereof. The leaks occurred in both right and left sided approaches suggesting no predilection for one side or another. Avoidance of such injuries should be aided by vigilant adherence to standard surgical principles, i.e., dissection of tissues only with full visualization and retraction, and blunt dissection of structures anterior to the spine away from where the sharp dissection is performed which should be over the spine directly.
This study suffers from several limitations. Our sample sizes are too small to permit multivariate analysis of factors associated with complications. Future studies should continue to add to the data and parse out the types of complications seen, as they may differ in character and evolve as surgeons become more comfortable with AVBT. Nonetheless, it is noteworthy that even with a small sample size, a pulmonary theme emerged. Additionally, we were not able to statistically compare PSF to AVBT in terms of complication rates or unique complication profiles, but as data becomes available researchers should pursue those goals.
Finally, we have included in this cohort, patients who have been treated outside of the standard indications for AVBT based on skeletal maturity. There are 18 patients between the ages of 18-21 who have surpassed peak growth velocity and are the subject of ongoing study. Skeletal maturity implies little ability for AVBT to modulate growth and means that curve correction will be maintained by the implant, which is not expected to last indefinitely. There is potential for bone and soft tissue remodeling which would result in maintenance of curve correction although this has not yet been established. When cords do break, it is usually after 2-3 years 38 , although this does not imply clinical failure, which is usually defined by residual major Cobb angle and indication for PSF 35,36,39 . For these patients, intraoperative correction, the ability to delay PSF, and the potential for curve correction with additional flexibility outweighed potential downsides, and ultimately aligned with their values. However, the purpose of our present study is to evaluate early surgical morbidity of the AVBT procedure which should not be impacted by skeletal maturity.

Conclusions
AVBT has demonstrated some success, but as a novel procedure, there is still much to be learned. For select candidates, AVBT may be an appropriate treatment but decisionmakers should understand that the complications may be more pulmonary in nature than those seen in PSF. And while motion preservation is an important benefit of AVBT, its main goal is to maintain curve correction through growth modulation. Surgeons who perform AVBT should be cautious in their use of rib resection to optimize screw trajectory, as this was strongly associated with major complications.