What Factors Predict the Risk of Proximal Junctional Failure in the Long Term, Demographic, Surgical, or Radiographic?

Results From a Time-dependent ROC Curve

Abstract

Study Design. 

Retrospective review of prospective multicenter database.

Objective. 

To identify an optimal set of factors predicting the risk of proximal junctional failure (PJF) while taking the time dependency of PJF and those factors into account.

Summary of Background Data. 

Surgical correction of adult spinal deformity (ASD) can be complex and therefore, may come with high revision rates due to PJF.

Methods. 

Seven hundred sixty-three operative ASD patients with a minimum of 1-year follow-up were included. PJF was defined as any type of proximal junctional kyphosis (PJK) requiring revision surgery. Time-dependent ROC curves were estimated with corresponding Cox proportional hazard models. The predictive abilities of demographic, surgical, radiographic parameters, and their possible combinations were assessed sequentially. The area under the curve (AUC) was used to evaluate models’ performance.

Results. 

PJF occurred in 42 patients (6%), with a median time to revision of approximately 1 year. Larger preoperative pelvic tilt (PT) (hazard ratio [HR]=1.044, P = 0.034) significantly increased the risk of PJF. With respect to changes in the radiographic parameters at 6-week postsurgery, larger differences in pelvic incidence-lumbar lordosis (PI-LL) mismatch (HR = 0.924, P = 0.002) decreased risk of PJF. The combination of demographic, surgical, and radiographic parameters has the best predictive ability for the occurrence of PJF (AUC = 0.863), followed by demographic along with radiographic parameters (AUC = 0.859). Both models’ predictive ability was preserved over time.

Conclusions. 

Over correction increased the risk of PJF. Radiographic along with demographic parameters have shown the approximately equivalent predictive ability for PJF over time as with the addition of surgical parameters. Radiographic rather than surgical factors may be of particular importance in predicting the development of PJF over time. These results set the groundwork for risk stratification and corresponding prophylactic interventions for patients undergoing ASD surgery.

Level of Evidence: 4

In Brief

Time-dependent ROC analysis allows us to identify an optimal set of factors that predicts the risk of PJF while taking the time dependency of PJF and those factors into account. Our results show that radiographic parameters have superior predictive ability over surgical parameters for the development of PJF over time.