Abstract: BACKGROUND: Lumbar interbody fusion has received remarkable results in lumbar degenerative illnesses; however, some of cage retropulsion has occurred. There is no unified understanding of the reasons why cage retropulsion has happened postoperatively, which may be related to age, sex, body mass index, preoperative diagnosis, multi-segmental fusion, operative segment, cage position, small cage, pear-shaped intervertebral disc, osteoporosis, and unilateral screw fixation. In clinical practice, the authors found that most of the patients with cage retropulsion could affect the injury of the posterior edge of the fusion segment on CT after operation. Therefore, this paper discusses the relationship between the end plate injury and cage retropulsion. 
OBJECTIVE: To investigate the causes of cage retropulsion after lumbar interbody fusion, and analyze the relationship between the intraoperative endplate injury and cage retropulsion. 
METHODS: The data of 1 823 patients undergoing lumbar interbody fusion from three clinical medical centers from January 2014 to December 2018 were retrospectively analyzed, including 21 cases in the cage retropulsion group and 1 802 cases in the non-cage retropulsion group. The t-test was used to compare the age and multiple fusion segments between the two groups. χ2 test was used to compare sex, preoperational diagnosis, and operational segments between the two groups. Totally 21 matched groups were organized; each had one cage retropulsion case and four non-cage retropulsion patients, who had the same sex and similar age (<±2 years); and the fusion segments contained the cage retropulsion segment in the cage retropulsion patient. Multifactor conditional logistic regression analysis was used to look over the risk factors of cage retropulsion in 21 matched groups. The potential risk factor involved the logistic analysis was the injury of endplate (≥ 5 mm), pear disk, BMI, small cage, and the posterior location of the cage.  
RESULTS AND CONCLUSION: (1) The average age of patients in the cage retropulsion group was 63.3 years, and that in the non-cage retropulsion group was 58.7 years; t-test showed significant difference (P < 0.05). However, the average number of fusion segments was 1.9 and 1.8 in the cage retropulsion group and the non-cage retropulsion group, respectively, with no significant difference by t-test (P=0.664); there was no significant difference in sex, preoperative diagnosis and surgical segments between the two groups by χ2 test (P > 0.05). (2) Multifactor conditional logistic regression analysis results indicated that the endplate injury (≥ 5 mm), pear-shape intervertebral disc and the posterior cage location were significant risk factors related to cage retropulsion (all P < 0.05; OR(95%CI)=54.9(23.8-126.7), 64.9(12.2-346.4) and 21.2(10.1-44.6). (3) These findings show that the endplate injury ((≥ 5 mm) is correlated with postoperative cage retropulsion, which is a risk factor for cage retropulsion. It is suggested that large reamer (12#) should not be used repeatedly to clean intervertebral disc, especially for patients with severe degeneration of nucleus pulposus, narrow intervertebral space and pear-shaped intervertebral disc.

Key words: posterior lumbar fusion, intraoperative endplate injury, cage retropulsion, cage, pear-shaped intervertebral disc, risk factors