PLIF was first proposed by Cloward in the 1940s [9], and it has now become the standard treatment for lumbar degenerative diseases. However, cage subsidence is a common complication after PLIF, induced by low bone density, endplate damage, too small and pre-placed cage, and excessive distraction of the intervertebral space [10]. Except for low BMD, which can be measured preoperatively, other factors are closely related to surgical technique and intraoperative selection. Therefore, BMD testing should be routinely performed before lumbar fusion surgery, and active antiosteoporosis treatment should be performed for patients with osteoporosis to reduce the occurrence of postoperative cage subsidence.
At present, the definition of cage subsidence has not been fully standardized. Marchi et al. [4] graded cage subsidence according to the ratio of the reduction in the height of the intervertebral space on lateral lumbar radiographs. However, after lumbar fusion, a certain reduction in the height of the intervertebral space is considered a normal process of endplate remodeling due to aggressive discectomy [11]. Therefore, the grade 0 description in this classification method does not clearly distinguish between subsidence immediately after surgery and cage subsidence due to low bone density. Considering the possible subsidence of the cage that is not parallel to the intervertebral space, cage subsidence has been defined as the displacement of the cage toward the rostral or caudal endplate by > 2 mm on CT sagittal images [4, 12, 13]. It may be more accurate to measure the difference in distance between cages entering the endplate on postoperative and late CT scans, but postoperative CT was not routinely obtained in the present study. Therefore, the measurements derived by CT sagittal images in this study were consistent with those of previously published methods [4, 12, 14].
The correlation between cage subsidence and postoperative clinical efficacy has been controversial. Cho et al. [15] conducted a 2-year follow-up of 55 osteoporotic and nonosteoporotic patients who received PLIF and found that, compared with the nonosteoporotic group, the osteoporosis group had a higher incidence of cage subsidence, but without significant difference in the improvement of clinical symptoms. Similarly, Oh [5] and Satake [16] also found no significant difference in postoperative clinical efficacy between the cage subsidence and nonsubsidence groups. However, Tohmeh et al. [17] believed that cage subsidence would significantly affect clinical efficacy after surgery. When the subsidence was > 4 mm, the postoperative Oswestry Disability Index, quality of life assessment, and VAS scores for low back pain were significantly worse. Similarly, Marchi et al. [18] suggested that early postoperative cage subsidence would induce transient low back pain. Despite the high incidence of cage subsidence rate at 23.9% in the present study, clinical efficacy between groups was not significantly different. Because cage subsidence usually occurs within 3 months after surgery, it is recommended that patients wear a lumbar brace for more than 3 months.
HU value can be used to selectively measure the bone density of cancellous bone to avoid the degeneration area, thereby improving the diagnostic accuracy. Zhou et al. [19] compared the accuracy of T-score and lumbar HU values in predicting cage subsidence and found that the preoperative HU value of the lumbar spine was more accurate, but the T-score in the present study was obtained by measuring the lumbar spine using DXA. However, because of lumbar spondylolisthesis, intervertebral space stenosis, osteophyte formation, osteosclerosis, and abdominal wall vascular calcification, BMD was overestimated by the T-score obtained using lumbar DXA [20]. Rey et al. [21] and Bina et al. [7] compared the forearm DXA measurements with lumbar spine DXA measurements and reported a significant linear correlation between forearm BMD and lumbar spine BMD and that forearm DXA was useful in diagnosing osteoporosis in postmenopausal women, exhibiting better accuracy than lumbar DXA. In addition, Pouillès et al. [8] found that DXA measurement of the forearm is an effective tool for OP screening and can directly identify approximately 50% of patients without central OP. Moreover, forearm DXA is a fast, inexpensive, and low-radiation skeletal state assessment method [22], which is widely used in clinical practice. In this study, BMD of the distal forearm was measured using DXA, and the results showed that both forearm T-score and HU values could predict the cage subsidence following PLIF, but forearm T-score was a more accurate predictor.
In our study, logistic regression analysis revealed the forearm T-score and mean global HU value as independent risk factors for cage subsidence after PLIF (P = 0.016 and 0.031, respectively). Compared with the post-PLIF subsidence rate (30.2%) reported by Cho et al. [15], the subsidence rate (23.9%) in this study was relatively low, which may be related to our use of the pedicle screw system to improve posterior stability. The forearm T-score and mean global HU value of the 17 patients in the subsidence group were significantly lower than those of the 54 patients in the nonsubsidence group. Both forearm T-score and mean global HU value can predict fusion subsidence. Compared with the mean global HU value (AUC, 0.744; 95% CI, 0.544–0.943), the forearm T-score (AUC, 0.840; 95% CI, 0.672–1.000) was more predictive of cage subsidence. In our study, using a forearm T-score of − 2.6 as the threshold yielded a sensitivity of 84.6% and a specificity of 83.3%; with the mean global HU value threshold of 104.2 HU had a sensitivity of 84.6% and specificity of 66.4%. Therefore, patients with forearm T-score < − 2.6 were at a greater risk of cage subsidence after PLIF. Therefore, patients with forearm T-score > − 2.6 should be selected as candidates if spine surgeons avoid cage subsidence after PLIF.
Our study has some limitations. First, the single-center study design was retrospective in nature, the sample size was small, and follow-up time was short; thus, a large-sample long-term prospective study is warranted to validate our findings. Second, the study population included only patients with single-segment PLIF combined with bilateral pedicle screw fixation. Therefore, further investigation of patients undergoing other surgical modalities is required. Third, we did not discuss the correlation between cage size, cage position, intervertebral-space-correction height, and cage subsidence, many of which may lead to a mismatch between the vertebral endplate and cage, resulting in cage subsidence.