Risk factors of proximal junctional kyphosis after surgical correction of spinal deformities caused by Scheuermann’s disease

Objective. To identify risk factors for the development of Proximal Junctional Kyphosis (PJK) in patients with Scheuermann’s kyphosis operated on using segmental instrumentation. Material and Methods . The study group consisted of 43 patients (13 females, 30 males), mean age was 17 years, and mean postoperative follow-up was 6 years. Spondylograms with a patient in a standing position performed before surgery, a week after surgery, and at the end of the follow-up period were analyzed. Radiological parameters were studied: cervical lordosis, absolute rotation angle, thoracic entry angle, T1 vertebral body tilt, neck tilt, skull tilt, thoracic kyphosis, thoracolumbar kyphosis, lumbar lordosis, vertebral and pelvic parameters (Pelvic Incidence, Pelvic Tilt, Sacral Slope), sagittal vertical axis, proximal junctional angle (PJA), and length of the posterior spinal fusion. Results. The mean PJA before surgery was 7° [3°; 8°], immediately after surgery – 10° [8°; 13°], by the end of the follow-up period – 25°[19°; 32°]. The incidence of PJK by the end of the follow-up period was 79.1 % (in 34 out of 43 patients). The initial value of thoracic kyphosis was 77° [72°; 86°], after surgery – 41° [31°; 46°], at the last examination – 43° [35°; 53°]. The inclination of the T1 vertebral body in the sagittal plane before surgery was 39° [30°; 45°], at the stages of follow-up – 33° [22°; 37°] and 39° [27°; 45°]. Some significant predictors were identified. An increase in the inclination of the T1 vertebral body (p = 0.005) by k° is associated with an increase in the risk of PJK by 1.19k [1.08k; 1.37k] times, and an increase of thoracic kyphosis by k° (p = 0.023) – by 1.12k (1.03k; 1.27k) times. The formula for preoperative predicting the likelihood of this complication is: P (PJK) = 1 - 1/(1 + exp (-23.14 + 0.26 × T1 + 0.21 × TK)), where P(PJK) is the probability of proximal junctional kyphosis; exp(z) is the exponential function to the power of z; T1 (T1 vertebral body tilt) and TK (thoracic kyphosis) are preoperative values of variables. Using ROC analysis, the threshold value for predicting PJK was determined to be 74.2 %, that is, the development of PJK was predicted in patients with a PJK probability greater than the threshold value calculated by the model formula. The predictive ability of the multivariate model was tested on the basis of the available initial data with a known final result. The prediction was correct in 41 cases out of 43. Conclusion. Using the multivariate logistic regression method, two mutually independent multiplicative indicators were determined for predicting PJK with high accuracy (sensitivity 94.1 %, specificity 100.0 %) – inclination of the T1 vertebral body and thoracic kyphosis.