Paraspinal sarcopenia at the upper instrumented vertebra is a predictor of discoligamentous but not bony proximal junctional kyphosis

Introduction: Stresses at the upper instrumented vertebra (UIV) following long-segment thoracolumbar fusion can result in proximal junctional kyphosis (PJK). Both poor bone quality and paraspinal sarcopenia have been suggested as risk factors, but their relative contributions to failure mode is unclear.

Methods: Adults ≥50 undergoing thoracolumbosacral fusion with UIV in the upper thoracic spine (T1-5) at a single tertiary care center were identified. Data were gathered on pre- and postoperative spinopelvic parameters, bone quality (using Hounsfield units and vertebral bone quality score), and paraspinal cross-sectional area at L3 and the UIV. PJK was defined by ≥10° increase in the proximal junctional angle inscribed by the inferior endplate of the UIV and superior endplate of the suprajacent vertebra. Risk factors for PJK in time-to-event analysis were performed using Cox regression. PJK was subdivided into types 1-3 based on the Yagi-Boachie classification.

Results: Fifteen of 76 included patients (median age 66; 72.4% female) experienced PJK; 10 experienced type 1, 4 experienced type 2, and 1 experienced type 3. Univariable Cox proportional hazards modeling demonstrated that the occurrence of PJK was negatively correlated with total paraspinal muscle CSA at the UIV (HR 0.74/100cm²; 95% CI [0.57, 0.6]; p=0.02). Lower total paraspinal CSA at L3 (HR 0.94/100cm²; p=0.07) and higher postoperative global tilt (HR 1.03; p=0.09) also trended towards significance. Similarly, type 1 PJK was predicted by decreased total paraspinal CSA at the UIV (HR 0.64/100cm²; [0.45, 0.92]; p=0.02); L3 total paraspinal CSA again only trended towards significance (HR 0.94/100cm²; p=0.12). Paraspinal CSA was not predictive of type 2 PJK; lower HU at the UIV and UIV+1 (p=0.16) trended towards being a significant predictor of type 2 PJK but did not meet threshold for significant (HR 0.98/unit; p=0.16). Direct comparison of patients experiencing type 1 and type 2 PJK showed higher average paraspinal CSA and lower average HU at the UIV, though this was not significant.

Conclusion : Global alignment and paraspinal sarcopenia appear most predictive of PJK; however, when subanalyzing by PJK type, paraspinal sarcopenia appears only predictive of type 1 PJK. Type 2 PJK may be better predicted by bone quality. Further investigation using larger cohorts is merited.