121. Identifying the utility of Hounsfield units (HU) to predict proximal junctional kyphosis in adult spinal deformity

Abstract

<h3>BACKGROUND CONTEXT</h3> Proximal junctional kyphosis (PJK) and proximal junctional failure (PJF) are two of the most challenging complications in adult spinal deformity (ASD) surgery. Bone quality has been identified as an important factor in outcomes of ASD patients. The association between Hounsfield units (HU) on CT scan at UIV and L1, and PJK occurrence remains poorly defined. Previous published literature has found HU<120 to indicate poor bone health. <h3>PURPOSE</h3> Hounsfield units at the UIV and L1 can predict the occurrence of PJK in ASD. <h3>STUDY DESIGN/SETTING</h3> Retrospective review of prospectively collected multicenter data. <h3>PATIENT SAMPLE</h3> A prospective database with inclusion criteria of age >18, PI-LL >10, PT>25, SVA>5, TK >60 and Cobb >20 was retrospectively reviewed and patients included only if complete: preop Hounsfield unit data at L1 and UIV, PJK prophylaxis data (PRPH), fusion to the pelvis and 2-year radiographic data. <h3>OUTCOME MEASURES</h3> Development of PJK or PJF. These were defined as per Glattes et al as: PJA <-10 & ΔPJA < -10 at 2 years postoperation and PJK as per Lafage et al as: PJA <-28 & ΔPJA < -22 at 2 years or revision surgery for PJK before 2 years postoperation. Health related quality of life measures (HRQL) included SF 36, ODI, SRS-22. <h3>Methods</h3> Retrospective review of prospectively collected multicenter data. Data analyzed with SPSS software. Multilinear regression models were run to identify independent variables associated with PJK and PJF. Odds ratios calculated. Decision tree analysis used to see if the effect of HU is not conditional of other parameters. <h3>Results</h3> This study included 240 patients, avg age 64 (31-84), BMI 28.9 (17.7-46), CCI 2.1 (0-8), and 188 (78.3%) females. Of the total patients, 17.1% (41) reported osteoporosis and 56.3% had previous spine surgery. There was significant in improvement in all HRQL from pre- to postsurgery (p<0.001). PJK occurred in 45% (108/240) and PJF in 13% (31/240). HU as an independent variable did not correlate with PJK magnitude. When controlling for PRPH, there was no correlation. Multilinear regression found that HU was not an independent predictor of PJK or PJF. PJK/F was subclassified as bony (bPJK) and analysis revealed that bPJK had larger angular deformity, higher rate with LT UIV, and associated with HU at UIV (p=0.011) as did bony PJF (p=0.037). UIV HU<120 was associated with bPJK (p=0.029) and bPJF (p=0.000). Odds ratio for UIV HU<120 to develop bPJK is 2.33 (CI: 1.07-5.08) and 4.67 for bPJF (CI: 1.87-11.7). Decision tree analysis, however, showed UIV HU was a conditional predictor of PJK in patients who have a PI-LL corrected >13 and with global realignment >19. <h3>Conclusions</h3> Hounsfield units alone are not an independent predictor of PJK or failure in ASD. Decision tree analysis, however, reveals that in the setting of patients requiring major global realignment and lordosis generation, HU are predictive of PJK. Furthermore, UIV HU<120 is 2.3 times more likely to result in bPJK and 4.6 times more likely in bPJF. <h3>FDA DEVICE/DRUG STATUS</h3> This abstract does not discuss or include any applicable devices or drugs.