XLIF and OLIF are the two most commonly used LLIF for the treatment of lumbar spine disorders, which have been popular among spine surgeons since their introduction due to their minimally invasive nature, high fusion rate, and precise efficacy. XLIF splits the psoas major muscle to operate, and how to avoid lumbar plexus nerve injury is a constant concern among scholars [4, 5]. Intraoperative neurophysiological monitoring can effectively reduce lumbar plexus nerve injury, but it is a passive measure with low specificity and high cost, which is not suitable for primary hospitals. Based on the anatomical study of the lumbar plexus, the traditional XLIF was modified in our center in advance, and it was not necessary to be routinely equipped with neurophysiological monitoring intraoperatively . Previously, we predominantly made modifications to the surgical entry point. However, the cage was still implanted vertically with the assistance of a 3-blade retractor, which required making the break of the table between the iliac crest and greater trochanter for side bending to eliminate iliac crest occlusion in the L45 segment with high iliac crest. During the later application, we found that series of problems including the tension of the psoas major muscle due to the break of the table, the mutual constraint of the forces between the three blades of the retractor and the inability to place the retractor vertically due to the high iliac crest could lead to the inability to implant the cage vertically.
OLIF was first reported by Mayer in 1997 and was officially named and designed with a dedicated two-blade retractor by Silvestre in 2012 [11, 12]. OLIF chooses an oblique approach between the psoas major muscle and the large artery without splitting the psoas major muscle, which is less likely to interfere with lumbar plexus, but is likely to cause injury to the sympathetic trunk, ureter, vessels, and peritoneum around the approach. Based on the narrower space between psoas major muscle and large artery in Asians compared to Europeans and Americans, Fan et al.  proposed a modified OLIF approach by separating the anterior border of the psoas major muscle dorsally from the disc and/or vertebral body under direct vision to reveal sufficient disc area. This modified OLIF approach reduced complications to some extent, and this study reported an access-related complication rate of 15.6% (13/83). Even so, this modified OLIF is mostly unsuitable in cases with extremely narrow space between psoas major muscle and artery, elevated psoas major muscle, and right-sided approach.
Based on the pre-modified XLIF, we proposed XOLIF by combining the advantages of OLIF tilted placement of 2-blade retractor. XOLIF differs from XLIF in the following ways: (1) the break of the table between the iliac crest and greater trochanter for side bending is not needed in the L45 segment with high iliac crest; (2) a 2-bladed retractor is used; (3) the retractor is placed at an oblique angle; and (4) the psoas major muscle is retracted dorsally using Kirschner wires. XOLIF has many technical advantages. (1) The 2-blade retractor is placed at an oblique angle, suitable for the high iliac crest L45 segment. (2) Splitting the psoas major muscle at the anterior edge of the psoas major muscle on direct vision for access reduces the risk of vascular, nerve, and organ injury, as well as lumbar plexus injury. (3) In the past, the OLIF procedure often required an assistant to use tissue pulling hooks to pull the psoas major dorsally, which was laborious and unstable, and the result was unsatisfactory when the psoas major was thickened. The dorsal muscle can easily herniate into the operating area and affect the operation with 3-blade retractor of XLIF. The use of 2 Kirschner wires as a retractor is easy and has the best effect in retracting psoas major muscle, solving the problem of blade occlusion during intraoperative fluoroscopy as well. (4) A 1.5 mm-diameter Kirschner wire with a guide rod is inserting into the center of the intervertebral space. The position of the apex of the guide rod is confirmed by lateral fluoroscopy to precisely guide the lateral disc incision position and to avoid implanting cage extremely ventrally or dorsally.
The present study showed that there was no statistical difference among the three groups in terms of operation time, intraoperative blood loss, postoperative VAS and ODI. Besides, the interbody fusion rate evaluated by CT scan at 12 months after surgery reached more than 90%, and the fusion rate of the three groups was not statistically different. Factors such as fusion cage material, the type of filler, bone graft material, internal fixation method, smoking, obesity and osteoporosis can affect interbody fusion rate [14,15,16]. A study by Nourian et al.  showed that BMP can improve LLIF interbody fusion rate. Internal fixation can create a good mechanical stability for interbody fusion. Adjunctive internal fixation is recommended for patients with osteoporosis, spondylolisthesis, and intraoperative endplate injury, but there are no specific criteria for the way of internal fixation [18, 19].
To our knowledge, this study was the first time to analyze the composition ratio of cases with psoas major and vascular space stenosis, psoas major muscle elevation, psoas major muscle hypertrophy, and high iliac crest in each group. The composition ratio of cases with psoas major and vascular space stenosis, psoas major muscle elevation was significantly higher in the XOLIF and XLIF groups than that in the OLIF group, because these two types are mostly unsuitable for OLIF and have a higher risk of vascular and nerve injury in the access . The risk can be predicted preoperatively by assessing the condition of the space between psoas major and artery with lumbar MRI . The L4-5 segment with high iliac crest cannot be performed with traditional XLIF because of the occlusion of the iliac crest. Incisions of XOLIF and OLIF are more ventral, and the 2-blade retractor is placed obliquely just to avoid the blocked iliac crest. As a result, the cage can be implanted vertically with a special angled handle . Although hypertrophy of psoas major muscle is not contraindication for traditional XLIF, it is relatively difficult to operate intraoperatively. XOLIF does not make the break of the table between the iliac crest and greater trochanter for side bending during the procedure, which keeps the muscles in a relaxed state. In addition, it is easy to implant cage vertically by 2 Kirschner wires retracting psoas major muscle dorsally. XOLIF combines advantages of OLIF and XLIF in treatment of the above cases, with wider indications.
Perioperative complication is the key factor to evaluate the technique. The OLIF approach through the space between the psoas major muscle and the vessels avoids interference with the lumbar plexus, but the risk of injury to the sympathetic trunk, ureter, vessels, and peritoneum is higher compared to XLIF . The XLIF approach with vertically splitting the psoas major muscle operates far away from the sympathetic trunk, ureter, vessels, and peritoneum, but the risk of injury to the lumbar plexus is higher. Differences of approaches between OLIF and XLIF lead to differences in access-related complications . XOLIF takes both advantages into account in terms of surgical safety.
Silvestre et al.  reported an overall complication rate of 11.2% in 179 cases of OLIF surgery. The access-related complications included 2 cases of rupture of iliac vein, 1 case of rupture of iliac-lumbar vein, 1 case of rupture of peritoneum, and 3 cases of sympathetic chain injury, 2 cases of neurological deficit, 2 cases of psoas muscle weakness or thigh numbness. Shunsuke et al.  retrospectively investigated a total of 2998 cases of minimally invasive LLIF performed from 2013 to 2015, including 1995 cases of XLIF and 1003 cases of OLIF, with an overall complication rate of 18% for XLIF 19.4% and OLIF 15.3%. However, complications such as intraoperative endplate injury and cage subsidence were not counted in the above studies, which may account for the low overall rate. Abe et al.  conducted a retrospective study of 155 cases of OLIF carried out by multiple centers. The overall incidence of complications was 48.3% (75/155). The three most common complications were endplate fractures or cage subsidence (18.7%), transient iliopsoas muscle weakness or thigh numbness (13.5%), and segmental artery injury (2.6%).
A total of 51 (32.7%) intraoperative and postoperative complications occurred in 156 patients in the present study, and most of these were minor complications including 7 cases (4.5%) of intraoperative endplate injury, 18 cases (11.5%) of postoperative cage subsidence, and 12 cases (7.7%) of transient iliopsoas muscle weakness or thigh pain/numbness. The incidence of endplate injury and cage subsidence was not higher in the three groups of this study than that of previous studies. The key to preventing endplate injury is to operate gently with parallel gaps, not to use reamers excessively, and select the appropriate cage. Cage subsidence is affected by many factors, including obesity, osteoporosis, intraoperative endplate injury, oversized cage, internal fixation method, etc. [27, 28]. Preoperative identification of risk factors for cage subsidence, intraoperative non-injury to the endplate and selection of an appropriate internal fixation method are the keys to preventing postoperative cage subsidence. In this study, 3 cases of vascular injury, 2 cases of sympathetic chain injury, and 1 case of peritoneal rupture occurred in the OLIF group, whereas, no such complications occurred intraoperatively in the XOLIF and XLIF groups because the access was far from great vessels, sympathetic chain, and peritoneum. Postoperative transient iliopsoas muscle weakness or thigh pain/numbness was often seen after LLIF, mostly due to intraoperative injury or irritation of psoas major muscle and psoas plexus [4, 29]. Because OLIF does not split the psoas major muscle, the incidence of postoperative iliopsoas muscle weakness and thigh paresthesia is lower than that of XLIF . XOLIF splits only a portion of the anterior border muscle fibers of the psoas major, which can take into account the advantage of OLIF not to interfere too much with the psoas major.
There were some limitations of this study. Firstly, the study just included L4-5 cases. Of course, XOLIF as well as XLIF and OLIF can be used in segments above L45. However, the upper lumbar segments do not have conditions of psoas major and vascular space stenosis, psoas major muscle elevation, psoas major muscle hypertrophy, and high iliac crest, which cannot highlight the advantages of XOLIF, so the study did not include segments above L45. Secondly, the study was a single-center retrospective cohort study, and future multicenter prospective studies are needed to further validate the superiority of XOLIF.