Risk Factors for Proximal Junctional Kyphosis in Osteoporotic Versus Non-Osteoporotic Patients After Adult Spinal Deformity Surgery

Background/Objectives: Proximal junctional kyphosis (PJK) is a frequent mechanical complication following adult spinal deformity (ASD) surgery, and osteoporosis is a well-recognized risk factor. However, whether the underlying risk factor profiles for PJK differ according to bone mineral density status remains unclear. This study aimed to identify and compare independent risk factors for PJK between osteoporotic and non-osteoporotic patients following long-segment posterior spinal fusion for ASD. Methods: This retrospective cohort study included 356 patients who underwent ≥5-level posterior fusion to the sacrum or pelvis for ASD with a 2-year follow-up. Patients were stratified into a non-osteoporotic (non-OP; n = 284) and an osteoporotic (OP; n = 72) group based on preoperative dual-energy X-ray absorptiometry (T-score ≤ −2.5). PJK was defined as a proximal junctional angle (PJA) ≥20° with an increase of ≥10° from the preoperative value. Demographic, surgical, and radiographic variables were analyzed separately for each group using univariate and stepwise multivariate logistic regression analyses. Predictive performance was assessed using receiver operating characteristic curve analysis. Results: The overall PJK incidence tended to be higher in the OP group than the non-OP group (36.1% vs. 24.6%, p = 0.055), and fracture-type PJK occurred significantly more frequently in osteoporotic patients (29.2% vs. 15.8%, p = 0.016). The variables that reached statistical significance differed between the two groups. In the non-OP group, lower Hounsfield units at the upper instrumented vertebra (UIV; OR = 0.990, p = 0.011), higher preoperative PJA (OR = 1.098, p < 0.001), and greater L1 pelvic angle (L1PA) offset indicating relative overcorrection (OR = 1.136, p < 0.001) were independent predictors of PJK. In the OP group, higher preoperative PJA (OR = 1.149, p = 0.007) and greater age-adjusted pelvic incidence–lumbar lordosis offset (OR = 1.052, p = 0.039) were identified as independent risk factors. Multivariable models demonstrated acceptable discriminative ability in both groups (area under the curve [AUC] = 0.741, 95% confidence interval [CI] = 0.671–0.804 for the non-OP group; AUC = 0.753, 95% CI = 0.632–0.867 for the OP group; optimism-corrected AUC 0.729 and 0.735 after bootstrap internal validation). Conclusions: The variables associated with PJK differed according to osteoporosis status following ASD surgery. Careful UIV selection avoiding kyphotic junctional alignment may be relevant for both groups. Avoiding overcorrection relative to the age-adjusted target may warrant particular attention in osteoporotic patients, while UIV bone quality and L1PA overcorrection may be relevant to surgical planning in non-osteoporotic patients.