Secondary Logo

Journal Logo

The Spine Blog

Sunday, November 27, 2016

Predicting Proximal Junctional Kyphosis

Proximal junctional kyphosis (PJK) and failure (PJF) are common complications following corrective surgery for adult spinal deformity (ASD). These complications are associated with worse patient reported outcomes (PROs) and are a frequent indication for reoperation. While prior studies have looked at specific risk factors for PJK, none have created a comprehensive model that predicts the risk of PJK or have evaluated the role of different risk factors in the same model. The International Spine Study Group (ISSG) sought to fill this gap in the literature by using a large database (510 ASD patients) that included patient and radiographic characteristics as well as follow-up out to 2 years post-operatively. They defined PJK as an increase in kyphosis at the proximal junctional level greater than 20 degrees compared to the initial post-operative radiograph along with an increase in the SRS-Schwab sagittal modifier grade. Proximal junctional failure was defined as a reoperation for PJK. In this population, the average age was 57, 78% were women, and the average fusion length was 12 levels. Twenty-seven percent of patients were classified as PJK/PJF (102 PJK, 37 PJF) within 2 years of their initial surgery. The most important variables predicting PJK/PJF were increasing age, lowest instrumented vertebra at the sacrum/ilium, baseline SVA > 10 cm, screws (vs. hooks) at the upper instrumented vertebra (UIV), UIV between T10 and L3, baseline PT > 30 degrees, and baseline PI-LL > 20 degrees. They used 70% of the patients to build their model, and 30% were used to test the model, which had an accuracy of 86.3%.


The ISSG should be congratulated on gathering data on over 500 ASD patients. The results provided no surprises, with increasing age, greater baseline sagittal imbalance, stopping proximally in the lower thoracic spine or distally at the sacrum/ilium, and using screws at the UIV all being known risk factors for PJK. Age is a likely surrogate for lower bone mineral density and less robust ligamentous and soft-tissue structures, and the authors noted it would have been helpful to include bone mineral density as a risk factor (it was not recorded in the database). The authors also chose to include only variables that could be considered pre-operatively. Post-operative factors such as degree of correction and residual imbalance could also play a role, though these were intentionally left out as they could not be considered during the pre-operative planning phase. This modeling strategy is novel and is different from the traditional regression modeling. In this case, many of the variables likely covaried with each other and would not have emerged as independent predictive variables in a regression model. As such, it is difficult to know how much each variable contributes to the risk of PJK, though the model's accuracy was maximized by including all of them. While the actual model could be considered for use clinically, the overall gestalt created by considering the important variables may be equally as helpful. Future models looking at other complications (i.e. infection, pseudarthrosis, hardware failure, medical complications, reoperation, etc.) and PROs would be very helpful to patients in the shared decision making process. The decision to proceed with a high morbidity surgery in a high risk population is a hard one, and having a stronger ability to predict outcomes would help patients facing this choice.


Please read Mr. Scheer's article on this topic in the November 15 issue. Does this change how you consider risk factors for PJK? Let us know by leaving a comment on The Spine Blog.

Adam Pearson, MD, MS

Associate Web Editor