Collection Radiographic Data

All patients underwent lumbar computational tomography (CT) three times in the imaging center of our hospital, including 1 week before, 1 week after, and 1 year after OLIF surgery (tube voltage: 120 kV) (Mikula et al., 2019; Xi et al., 2020; Zou et al., 2020). The CT scan settings were uniform in all enrolled patients. An experienced spine surgeon independently measured the following radiographic parameters. The interobserver and intraobserver reliability of these measured parameters was verified in 10 randomly selected patients. One week after the imaging measurement, the spine surgeon and a senior radiologist independently remeasured the imaging parameters of these selected patients.

The screw loosening status of the cranial and caudal vertebral bodies was identified separately. In the postoperative 1-year CT imaging data, vertebral bodies with ≥1 mm width radiolucent zones around the screw were defined as screw loosening (Bredow et al., 2016; Bokov et al., 2019; Zou et al., 2020). The BMD of these patients was identified by measuring their Hounsfield unit (HU) values. During HU measurement in vertebral bodies, the region of interest (ROI) was expanded to the largest within the cancellous bone but excluded other bony structures, such as cortical, BEP, and osteophytes (Schreiber et al., 2014; Xi et al., 2020; Zou et al., 2020). Values of HU were measured at the midsagittal plane, central transverse plane, transverse planes close to the superior and the inferior endplate separately, and the average value of these planes was set as the HU of the vertebral body (Pickhardt et al., 2013; Mikula et al., 2019; Xi et al., 2020; Zou et al., 2020). The sufficiency of contact between BEP and GB was quantified by calculating the grafted bony occupancy rate (GBOR) (Kim et al., 2012; Ushirozako et al., 2020). GBOR was measured in the cage’s central sagittal and coronal planes (rather than the vertebral body) in the postoperative CT imaging data (Figure 1).

The schematic of the definition and surgical simulations of GBOR, and typical cases for different contact sufficiency and resulting screw fixation status.