Predictive model for distal junctional kyphosis after cervical deformity surgery

BACKGROUND CONTEXT: Distal junctional kyphosis (DJK) is a primary concern of surgeons correcting cervical deformity. Identifying patients and procedures at higher risk of developing this condition is paramount in improving patient selection and care. PURPOSE: The present study aimed to develop a risk index for DJK development in the first year after surgery. STUDY DESIGN/SETTING: This is a retrospective review of a prospective multicenter cervical deformity database. PATIENT SAMPLE: Patients over the age of 18 meeting one of the following deformities were included in the study: cervical kyphosis (C2-7 Cobb angle>10 degrees), cervical scoliosis (coronal Cobb angle>10 degrees), positive cervical sagittal imbalance (C2-C7 sagittal vertical axis (SVA)>4 cm or T1-C6>10 degrees), or horizontal gaze impairment (chin-brow vertical angle>25 degrees). OUTCOME MEASURES: Development of DJK at any time before 1 year. METHODS: Distal junctional kyphosis was defined by both clinical diagnosis (by enrolling surgeon) and post hoc identification of development of an angle<-10 degrees from the end of fusion construct to the second distal vertebra, as well as a change in this angle by <-10 degrees from baseline. Conditional Inference Decision Trees were used to identify factors predictive of DJK incidence and the cut-off points at which they have an effect. A conditional Variable-Importance table was constructed based on a non-replacement sampling set of 2,000 Conditional Inference Trees. Twelve influencing factors were found; binary logistic regression for each variable at significant cutoffs indicated their effect size. RESULTS: Statistical analysis included 101 surgical patients (average age: 60.1 years, 58.3% female, body mass index: 30.2) undergoing long cervical deformity correction (mean levels fused: 7.1, osteotomy used: 49.5%, approach: 46.5% posterior, 17.8% anterior, 35.7% combined). In 2 years after surgery, 6% of patients were diagnosed with clinical DJK; however, 23.8% of patients met radiographic definition for DJK. Patients with neurologic symptoms were at risk of DJK (odds ratio [OR]: 3.71, confidence interval [CI]: 0.11-0.63). However, no significant relationship was found between osteoporosis, age, and ambulatory status with DJK incidence. Baseline radiographic malalignments were the most numerous and strong predictors for DJK: (1) C2-T1 tilt>5.33 (OR: 6.94, CI: 2.99-16.14); (2) kyphosis<-50.6 degrees (OR: 5.89, CI: 0.07-0.43); (3) C2-C7 lordosis<-12 degrees (OR: 5.7, CI: 0.08-0.41); (4) T1 slope minus cervical lordosis>36.4 (OR: 5.6, CI: 2.28-13.57); (5) C2-C7 SVA>56.3 degrees (OR: 5.4, CI: 2.20-13.23); and (6) C4_Tilt>56.7 (OR: 5.0, CI: 1.90-13.1). Clinically, combined approaches (OR: 2.67, CI: 1.21-5.89) and usage of Smith-Petersen osteotomy (OR: 2.55, CI: 1.02-6.34) were the most important predictors of DJK. CONCLUSIONS: In a surgical cohort of patients with cervical deformity, we found a 23.8% incidence of DJK. Different procedures and patient malalignment predicted incidence of DJK up to 1 year. Preoperative T1 slope-cervical lordosis, cervical kyphosis, SVA, and cervical lordosis all strongly predicted DJK at specific cut-off points. Knowledge of these factors will potentially help direct future study and strategy aimed at minimizing this potentially dramatic occurrence. (C) 2018 Elsevier Inc. All rights reserved.