Study design: Retrospective comparative cohort study.
Objective: Assess the incidence, risk factors, and outcomes of recurrent proximal junctional kyphosis (r-PJK) in PJK revision patients.
Summary of background data: Several studies have identified the incidence and risk factors for PJK after primary surgery. However, few studies have reported on PJK recurrence after revision for PJK.
Methods: A multicenter database of patients who underwent PJK revision surgery with minimum 2-year follow-up was analyzed. Demographic, operative, and radiographic outcomes were compared in patients with r-PJK and patients without recurrence no-Proximal Junctional Kyphosis (n-PJK). Postoperative Scoliosis Research Society-22r, Short Form-36, and Oswestry Disability Index were compared. Preoperative and most recent spinopelvic, cervical, and cervicothoracic radiographic parameters were compared. Univariate and multivariate analyses were used to determine r-PJK risk factors. A predictive model was formulated based on our logistic regression analysis.
Results: A total of 70 patients met the inclusion criteria with an average follow-up of 21.8 months. The mean age was 66.3 ± 9.4 and 80% of patients were women. Before revision, patients had a proximal junctional angle angle of -31.7° ± 15.9°. The rate of recurrent PJK was 44.3%. Logistic regression showed that pre-revision thoracic pelvic angle (odds ratio [OR]: 1.060 95% confidence interval [CI] 1.002; 1.121; P = 0.042) and prerevision C2-T3 sagittal vertical axis (SVA; OR: 1.040 95% CI [1.007; 1.073] P = 0.016) were independent predictors of r-PJK. Classification with these parameters yielded an accuracy of 72.7%, precision of 80.6%, and recall of 73.5%. When examining correction, or change in alignment with revision surgery, we found that change in SVA (OR: 0.981 95% CI [0.968; 0.994] P = 0.005) was the only predictor of r-PJK with accuracy of 66.7%, precision of 74.2%, and recall of 69.7%.
Conclusion: Patients after PJK revision surgery had a recurrence rate of 44%. Logistic regression based on the prerevision variables showed that prerevision thoracic pelvic angle and prerevision C2-T3 SVA were independent predictors of r-PJK.
Level of evidence: 4.