Risk Factors of Allogenous Bone Graft Collapse in Two-Level Anterior Cervical Discectomy and Fusion

OBJECTIVE: Anterior cervical discectomy and fusion (ACDF) is commonly used surgical procedure for cervical degenerative disease. Among the various intervertebral spacers, the use of allografts is increasing due to its advantages such as no harvest site complications and low rate of subsidence. Although subsidence is a rare complication, graft collapse is often observed in the follow-up period. Graft collapse is defined as a significant graft height loss without subsidence, which can lead to clinical deterioration due to foraminal re-stenosis or segmental kyphosis. However, studies about the collapse of allografts are very limited. In this study, we evaluated risk factors associated with graft collapse. METHODS: We retrospectively reviewed 33 patients who underwent two level ACDF with anterior plating using allogenous bone graft from January 2013 to June 2017. Various factors related to cervical sagittal alignment were measured preoperatively (PRE), postoperatively (POST), and last follow-up. The collapse was defined as the ratio of decrement from POST disc height to follow-up disc height. We also defined significant collapses as disc heights that were decreased by 30% or more after surgery. The intraoperative distraction was defined as the ratio of increment from PRE disc height to POST disc height. RESULTS: The subsidence rate was 4.5% and graft collapse rate was 28.8%. The pseudarthrosis rate was 16.7% and there was no association between pseudarthrosis and graft collapse. Among the collapse-related risk factors, pre-operative segmental angle (p=0.047) and intra-operative distraction (p=0.003) were significantly related to allograft collapse. The cut-off value of intraoperative distraction ≥37.3% was significantly associated with collapse (p=0.009; odds ratio, 4.622; 95% confidence interval, 1.470-14.531). The average time of events were as follows: collapse, 5.8±5.7 months; subsidence, 0.99±0.50 months; and instrument failure, 9.13±0.50 months. CONCLUSION: We experienced a higher frequency rate of collapse than subsidence in ACDF using an allograft. Of the various preoperative factors, intra-operative distraction was the most predictable factor of the allograft collapse. This was especially true when the intraoperative distraction was more than 37%, in which case the occurrence of graft collapse increased 4.6 times. We also found that instrument failure occurs only after the allograft collapse.