Determination of Risk Factors for Pseudoarthrosis in Minimally Invasive Spine Surgery in Transforaminal Lumbar Interbody Fusion Procedure

Thursday, June 1, 2023

8:30 AM – 6:15 PM

Location: Salons 7-8

Co-Author(s)

PD

Pooja R. Dave

Miami, Florida, United States
PT

Peter S. Tretiakov, BS

United States
JM

Jamshaid M. Mir, MD

Research Fellow
NYU Langone Health | NY Spine Institute, United States
OO

Oluwatobi Onafowokan

United Kingdom

Presenting Author (Person doing the Podium Present(s)

PP

Peter G. Passias, M.D.

Spine Surgeon
NYU Langone Orthopedic Hospital
New York, New York, United States

View E-Poster

Submitter(s)

PP

Peter G. Passias, M.D.

Spine Surgeon
NYU Langone Orthopedic Hospital
New York, New York, United States

View E-Poster

Disclosure(s):

Peter G. Passias, MD: No relevant disclosure to display

Background: Recent improvements with adult spinal deformity (ASD) surgery techniques have expanded to include minimally invasive surgery (MIS). Single positioning through combined approach to ASD has been studied on its benefits including reduced operative time and improved postoperative complications.

Purpose: To determine risk factors associated with the development of pseudoarthrosis in MIS TLIF procedures, as well as develop risk index

Methods: Operative adult spinal deformity patients (scoliosis≥20°, SVA≥5cm, PT≥25°, or TK≥60°) undergoing minimally invasive surgery with 2 year data were isolated. Patients were stratified into groups based on undergoing TLIF (T) or not (NT). Descriptive analyses and means comparison tests identified differences in baseline parameters, surgical characteristics, radiographic and deformity measurements, rates of complications, and patient-reported outcome measures. Multivariate logistic regression analyses determined significant predictors for development of pseudoarthrosis. Established weights were generated for predictive variables via back-step logistic regression for a risk score to predict development of pseudoarthrosis. Risk score was then validated via Receiver Operating Characteristic (ROC) curve method analysis. Categories via conditional inference tree (CIT) analysis-derived thresholds were tested for thresholds correlating to high risk.

Results: 465 spine patients included (59.9±14.7yrs, 27.3±5.3kg/m2, CCI: 1.63±1.6, FI: 3.12±1.65), with 34.4% (n=160) underwent TLIF. Cohort mean baseline SVA was 15.45±34.40mm, PI-LL 1.41±12.17°, and PT 16.24±8.80°. There were no differences in baseline age, gender, BMI, or baseline deformity between groups. A predictive risk score for pseudoarthrosis was developed with the following preoperative variables: age, frailty, history of diabetes, osteoporosis, depression, ASA grade, baseline L4-S1, and baseline T1PA. Using ROC, this predictive risk score generated an AUC of 0.87. Following CIT machine learning, thresholds for the Risk Score were derived: >5 No Risk (NoR), 3-5 Low Risk (LowR), 2-3 Moderate Risk (ModR), and < 2 High Risk (HighR). The rates of pseudoarthrosis for each category were: NoR – 0%; LowR - 1.6%; ModR – 9.3%; HighR – 24.3%. TLIF patients had overall the following pseudoarthrosis rates according to Risk Score: 5.9% NoR, 9% LowR, 2.7% ModR, 1.9% HighR. TLIF patients in HighR had 2.3 times lower risk with surgical prophylaxis (OR: -2.27 [.79, 2.31] p=.004). Additionally, the use of BMP-2 lowered the risk significantly in those in HighR (p=.032).

Conclusions: Development of a risk score can better inform surgeons on individualized patient factors associated with worsening outcomes. Our study developed a risk score for those undergoing TLIF procedure as it relates to the development of pseudoarthrosis. In those with high risk, surgical prophylaxis and BMP-2 demonstrated positive effects in patient outcomes, thus further enhancing justification for their use.