The present study was the first of its kind to investigate the impact of CS on fusion rate and clinical results following OLIF combined with anterolateral fixation. The overall incidence of CS and fusion in our study were 32.9% and 87.2%, respectively, which were within the reported ranges of 10%-40% and 87.2%-97.9% following OLIF combined with bilateral pedicle fixation or stand-alone [3,4,5].
CS is a progressive development that manifests as cages sinking into vertebrae through adjacent endplates prior to complete fusion . Currently, large variations were reported in the development of CS after LIF technique [9, 14]. Chen et al.  presented CS as a late event which was identified at 3 months postoperatively and continuously progressed until 2 years after lateral LIF (LLIF) technique. In contrast, Marchi et al.  argued that CS occurred mainly within 6 weeks and without significant progress after 3 months following LLIF. A similar trend was found in this present study, our results suggested that CS following OLIF should be classified as an early complication that occurred primarily at 1 month postoperatively and did not progress significantly after 3 months postoperatively. Therefore, the early postoperative stage should be considered a vital period to address CS after OLIF surgery.
CS has great significance for LIF technique, as it means some surgical goals, such as intervertebral distraction, indirect decompression or alignment correction, may not be met [8, 9]. Various studies have compared CS to surgical results following LIF technique, and a clear relationship was not found [6,7,8,9]. Our results indicated that CS were related to surgical results following OLIF. Higher magnitudes of CS were associated with worse surgical improvements. Marchi et al.  proposed that low grade CS (DH reduction less than 25%) was the result of endplate remodelling due to the natural curvature of endplate, and does not interfere with subsequent fusion. Similarly, we noted that mild CS yielded a comparable fusion rate compared to no CS group. But it caused a transient poor clinical improvement. We speculated that this poor clinical improvement may be the result of transient local bony changes such as endplate inflammation, and may abate over time after CS stabilizes and correct to similarly improved clinical outcomes. In contrast, we found that severe CS caused not only poor clinical achievements, but also significantly reduced the fusion rate. On the one hand, we inferred that severe CS may aggravate and prolong this bone change, thus causing aggravated and constant poor clinical improvements. On the other hand, with respect to intervertebral fusion, it requires a stable biomechanical environment to promote trabecular connections. We considered that the remarkable reduction in the height of the intervertebral space due to severe CS may result in the re-relaxation of the ligamentous structure at the index level and thus fail to provide a stable biomechanical environment necessary for the fusion process, eventually leading to reduced fusion rate . Based on our aforementioned results, as CS following OLIF was associated with poor surgical improvements, it was helpful to identify the related risk factors so that CS can be prevented.
Risk factors related to CS are multifactorial. Generally speaking, endplate stiffness and the interfacial load between the implant and the endplate are the basic factors affecting the occurrence of CS [16, 17]. Oxland et al.  proposed that the endplate stiffness decreased by approximately 33% after injury, thus inducing CS. In our study, we found that endplate injury significantly increased the occurrence of CS, therefore, we suggest that attention should be given to avoiding endplate injury intraoperatively, which may be beneficial in reducing CS. Gentle manipulation, BMD examination and a cage with appropriate height should be recommended to avoid endplate damage. Endplate stiffness also varies with the anatomic region . Hou et al.  demonstrated that the failure loading required for CS was maximum when the cage was placed posterolaterally on the endplate with the strongest stiffness. Kim et al.  also reported that anterior cage position was a risk factor for CS following transforaminal LIF (TLIF) surgery. However, we failed to find a clear relationship between CS and cage position. We speculate that the position of the cage placed through the oblique channel was overall anterior and the range of anteroposterior position was narrow , so it is may not enough to reflect the stiffness discrepancy at different endplate regions. In addition, an early study concluded that the cranial endplate is 40% stiffer than the caudal endplate , we confirmed this conclusion in the present study as we found that CS occurred more frequently in the caudal endplate.
BMD was also considered to be a vital factor affecting endplate stiffness. Hou et al.  found that decreased BMD resulted in lower endplate stiffness and lower failure loading for CS. Tempel et al.  presented that the sensitivity and specificity of a DEXA T score of -1.0 or less for predicting CS following LLIF were 78.3% and 63.2%, respectively. Park et al.  calculated that osteoporosis increased the CS risk following TLIF by 4.8-fold. We calculated that the osteoporosis increased the risk by 6.0-fold following OLIF, which was slightly higher than the risk coefficient of CS following TLIF. This finding indicated that stringent constraints may be required for bone condition, to prevent CS following OLIF.
Increased compressive forces is another basic mechanism which result in endplate fracture and CS . DH over-distraction is widely accepted as risk factor for CS in spinal interbody fusion surgery . In our study, we found that the increase in DH distraction was significantly correlated with CS following OLIF. Therefore, selecting appropriate cage height and avoiding excessive DH distraction may be beneficial for CS prevention. At present, the correct methods to select the appropriate cage height are still controversial. Some studies have suggested that the height should be determined according to the DH measured preoperatively , while others recommend should be determined by the compressive and distractive force generated by cage implantation .
The impacts of disc space morphology on CS have been preliminarily mentioned [23, 27]. Park et al.  presented that pear-shaped disc space was more likely to induce CS, interpreted as the smaller contact between the endplate and cage causes a stress concentration, thus inducing CS. Similarly, we found the CS risk significantly increased in cases with flat disc space compared to concave space. Therefore, customizing a specific shape cage according to the disc space shape to increase the match between cage and endplate may be helpful to reduce CS.
We acknowledge that the retrospective study design with short-term follow up is a limitation. Secondly, subsidence is a multidimensional parameter and that its evaluation in a single plane, as in this study, may not be sufficient to evaluate its characteristics. Thirdly, there are biomechanical differences between anterolateral fixation and pedicle screw fixation, thus our results are only applicable to OLIF combined with anterolateral fixation. Also, as a reflection of the effectiveness of anterolateral fixation, the information of screw displacement was missing. Finally, other factors that may affect CS, such as cage length and lumbar lordosis, were not evaluated in this study and should be further investigated.