Background:

Cage subsidence may occur following transforaminal lumbar interbody fusion (TLIF) and lead to nonunion, foraminal height loss and other complications. Low bone quality may be a risk factor for cage subsidence. Assessing bone quality through Hounsfield units (HU) from computed tomography has been proposed in recent years. However, there is a lack of literature evaluating the correlation between HU and cage subsidence after TLIF.

Methods:

Two hundred and seventy-nine patients suffering from lumbar degenerative diseases from April, 2016 to August, 2018 were enrolled. All underwent one-level TLIF with a minimum of 1-year follow-up. Cage subsidence was defined as > 2 mm loss of disc height at the fusion level. The participants were divided into 2 groups: cage subsidence group (CS) and non-cage subsidence group (non-CS). Bone quality was determined by HU, bone mineral density of lumbar (BMD-l) and femoral (BMD-f) from dual-emission X-ray absorptiometry (DXA). HU of each vertebra from L1 to L4 (e.g., HU1 for HU of L1) and mean value of the four vertebrae (HUm) were calculated. Visual analog scale (VAS) of back/leg pain and Oswestry disability index (ODI) were used to report clinical outcomes.

Results:

Cage subsidence occurred in 82 (29.4%) cases at follow-ups. Mean age was 50.8 ± 9.0 years with a median follow-up of 18 months (range from 12 to 40 months). A total of 90.3% patients presented fusion with similar fusion rate between the two groups. ODI and VAS in leg were better in non-CS group at last follow-ups. Using receiver operating characteristic curves (ROCs) to predict cage subsidence, HUm provided a larger area under the curve (AUC) than BMD-l (Z = 3.83, P < 0.01) and BMD-f (Z = 2.01, P = 0.02). AUC for HU4 was larger than BMD-f and close to HUm (Z = 0.22, P = 0.481).

Conclusions:

Cage subsidence may indicate worse clinical outcomes. HU value could be a more effective predictor of lumbar cage subsidence compared with T-score of DXA after TLIF.