Differences in radiographic and clinical outcomes of oblique lateral interbody fusion and lateral lumbar interbody fusion for degenerative lumbar disease: a meta-analysis

Background In the current surgical therapeutic regimen for the degenerative lumbar disease, both oblique lateral interbody fusion (OLIF) and lateral lumbar interbody fusion (LLIF) are gradually accepted. Thus, the objective of this study is to compare the radiographic and clinical outcomes of OLIF and LLIF for the degenerative lumbar disease. Methods We conducted an exhaustive literature search of MEDLINE, EMBASE, and the Cochrane Library to find the relevant studies about OLIF and LLIF for the degenerative lumbar disease. Random-effects model was performed to pool the outcomes about disc height (DH), fusion, operative blood loss, operative time, length of hospital stays, complications, visual analog scale (VAS), and Oswestry disability index (ODI). Results 56 studies were included in this study. The two groups of patients had similar changes in terms of DH, operative blood loss, operative time, hospital stay and the fusion rate (over 90%). The OLIF group showed slightly better VAS and ODI scores improvement. The incidence of perioperative complications of OLIF and LLIF was 26.7 and 27.8% respectively. Higher rates of nerve injury and psoas weakness (21.2%) were reported for LLIF, while higher rates of cage subsidence (5.1%), endplate damage (5.2%) and vascular injury (1.7%) were reported for OLIF. Conclusions The two groups are similar in terms of radiographic outcomes, operative blood loss, operative time and the length of hospital stay. The OLIF group shows advantages in VAS and ODI scores improvement. Though the incidence of perioperative complications of OLIF and LLIF is similar, the incidence of main complications is significantly different.


Background
Lumbar interbody fusion has been recognized as a powerful surgical tool for lumbar degenerative disease, including degenerative disc disease, spondylolisthesis, disc herniation, and deformity [1,2]. Traditional techniques, such as anterior lumbar interbody fusion and posterior/transforaminal lumbar interbody fusion, have been successful with high patient satisfaction and fusion rates [1]. However, complications, such as excessive blood loss, iatrogenic muscle and soft tissue injury, muscular denervation, cannot be avoided [3]. To decrease surgical trauma, reduce operative bleeding and reduce hospital stay, minimally invasive spine (MIS) techniques such as oblique lateral interbody fusion (OLIF), lateral lumbar interbody fusion (LLIF), also known as extreme/ direct lumbar interbody fusion (XLIF / DLIF), is progressively gaining popularity [3][4][5][6][7]. The advantages of both XLIF and OLIF are potential to restore both segmental and global lordosis, the latter often possible with targeting multiple levels through a single incision with minimally invasive surgery. Another advantage is the possibility of indirect decompression, avoiding the need in many cases to open spinal canal or foramina directly. Although radiographic and clinical outcomes of LLIF or OLIF have been assessed in many studies, however, few studies have compared the results of OLIF and LLIF. In the present study, considering the increasing interest in these techniques, our meta-analysis is performed to find differences in the radiographic and clinical outcomes of OLIF and LLIF for degenerative lumbar disease and, thus, provide vital evidence-based guidance for clinicians. criteria: studies reported outcomes at least one of the following outcomes: disc height (DH), visual analog scale (VAS), Oswestry disability index (ODI), operative blood loss, operative time, length of hospital stay, fusion, and complications; and had sufficient data to extract and pool. The following search terms were used: "oblique lumbar interbody fusion" or "OLIF" or "lateral lumbar interbody fusion" or "LLIF" or "extreme lateral interbody fusion" or "XLIF" or "direct lateral interbody fusion" or "DLIF". We have found 2097 results through the above databases. Two of the reviewers (H.M. L. and R.J. Z.) independently examined the data extraction using standardized data extraction forms. There were altogether 1337 potentially relevant studies identified from the electronic search, the abstracts, and full manuscripts were also reviewed. The reference list of all relevant retrieved manuscripts was searched manually to identify additional studies that might have been missed. A summary of the study selection process in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines was showed in Fig. 1.

Statistical methods
When pooling the data from the included studies, the Stata software (version 14.0; TX 77845, USA) was used. The z-test was applied to the mantel-haenszel analysis, and standardized mean difference (SMD) and 95% confidence interval (CI) was estimated by the random-effects model [58]. The heterogeneity was assessed using the chi-square test. The potential effects of mean age and follow-up on radiological and clinical outcomes were assessed by performing meta-regression. A p-value of 0.05 or less was considered statistically significant.

Description of study
We have included fifty-six studies involving 2852 patients in this analysis. A total of 1304 patients aged 54.1-69 years (62.1 years on average) were included in the OLIF group. Follow-up was reported from 1 month to 24 months (10.5 months on average). Fusion levels have been reported to involve T10-T11 to L5-S1 (1 level: 154 patients; 2 levels: 192 patients, and ≥ 3 levels: 80 patients). A total of 1548 patients aged 51-74 years (62.2 years on average) were included in the LLIF group. Follow-up duration has been reported to range from 3 months to 35.4 months (17.7 months on average). Fusion levels have been reported to involve T12-L1 to L4-L5 (1 level: 278 patients, 2 levels: 139 patients, and ≥ 3 levels: 142 patients).
Operative blood loss, operative time, and length of hospital stay Operative blood loss, operative time, and length of hospital stay have been reported in eleven studies of OLIF [2, 8-11, 16, 17, 21, 23-25] including 787 patients. OLIF combined with posterior fixation resulted in an average blood loss of 136 mL (49.7 for 1 operated level; 112.5 for 2 operated levels, 148.7 for 3 operated levels or more), an average operative time of 114 min (54.5 for 1 operated level; 67.2 for 2 operated levels, 119.5 for 3 operated levels or more), and an average length of hospital stay of 6.8 d (5.1 for 1 operated level; 6.4 for 2 operated levels, 7.9 for 3 operated levels or more) [58]. OLIF stand-alone [8,11] resulted in an average operative blood loss of 83 mL, an average operative time of 104.2 min, and an average length of hospital stay of 6 d. Operative blood loss, operative time, and length of hospital stay have been reported in 29 studies of LLIF [6, 10, 26-32, 34, 35, 37-42, 44-46, 49, 51, 52, 54-57] including 1698 patients. LLIF combined with posterior fixation resulted in an average operative blood loss of 176 mL (71.6 for 1 operated level; 115 for 2 operated levels, 176 for 3 operated levels or more), an average operative time of 188 min (134.8 for 1 operated level; 174 for 2 operated levels, 253 for 3 operated levels or more), and an average length of hospital stay of 5.9 d (2 for 1 operated level; 3 for 2 operated levels, 4 for 3 operated levels or more). LLIF stand-alone [6,30,34,40,42,46,49,52] resulted in an average operative blood loss of 76.4 mL, an average operative time of 126.2 min, and an average length of hospital stay of 5 d (Table 1).

Meta-regression analyses and publication bias
Results of meta-regression demonstrated that the followup period had no effect on DH, VAS, or ODI, and mean age had no effect on DH or VAS in both surgical groups, However, the mean age was negatively correlated with decreasing ODI score in the OLIF group (Table 3).
No publication bias for DH or ODI was observed in either group, but publication bias for VAS was observed for the LLIF group (Fig. 5).

Discussion
Our meta-analysis was designed to explore differences in radiographic and clinical outcomes of OLIF and LLIF. The importance of restoring sagittal balance and lumbar lordosis is increasingly recognized in any spinal reconstruction surgery. The most important factor affecting lumbar lordosis recovery is DH restoration [59]. In the  [8,43,48], and some researchers have found that the OLIF group showed a significant increase of posterior disk space compared with the LLIF group (40.2% vs. 31.6%, respectively), which may be explained by the fact that the orthogonal maneuver provides the controllability of cage position, and earlier change of direction from oblique to direct lateral makes the anterior position of the cage to obtain a large segmental angle for the kyphosis correction [10]. However, other study demonstrated that OLIF might make no difference in restoring the DH, as more than half of the patients without applying posterior supplemental fixation in their study [11]. And considering the heterogeneity of this meta-analysis, meta-regression was applied for exploring the cause of heterogeneity of the potential Oswestry disability index improvement and loss, respectively, after surgery. SMD, standard mean difference effect of mean age and follow-up on DH restoration, but we found that the follow-up period and mean age did not affect DH in both groups.
According to previous studies, both OLIF and LLIF presented lower postoperative VAS and ODI scores compared with preoperative values [8,9,38,39], which is consistent with our results. Furthermore, we have found the postoperative VAS and ODI scores in the OLIF group were better than those in the LLIF group (− 3.25 versus − 2.18; − 3.06 versus − 1.76, respectively), probably due to complications related to the psoas muscle injury in the LLIF group. Besides, meta-regression has shown that the mean age was negatively correlated with decreasing ODI score in the OLIF group.
This meta-analysis suggests that OLIF showed similar operative blood loss, operative time, and hospital stay than those compared with LLIF. This may be because the OLIF and LLIF techniques are performed through the retroperitoneal space and by blunt dissection. Besides, the number of manipulations during surgery is decreased especially when radiographs are required, as no microscope or other specific complex ancillary is used in both groups.
The meta-analysis also suggests that OLIF achieves a similar fusion rate (> 90%) compared with that of LLIF.   However, a high rate of perioperative complications was observed related to OLIF and LLIF, and there were differences between the two groups. Higher rates of nerve injury and psoas weakness were reported for LLIF, while higher rates of cage subsidence, endplate damage, and vascular injury were reported for OLIF. Direct injury to the lumbar plexus is the greatest risk from LLIF, as it has the potential to result in sensory and motor deficits. According to previous studies, the incidence of thigh dysfunction and numbness after LLIF has been reported to range from 0.7 to 30%, and the incidence of motor weakness ranged from 3.4 to 23.7% [6,47,60,61]. In our study, we have found the incidence of anterior thigh numbness and leg weakness was high at 21.2%. Although most of those symptoms are transient and completely recovered by time in most studies, there are still some studies reported that some patients presented with the permanent damages [10,62]. The possible reason for No publication bias for DH or ODI was found in either group. Publication bias was found for VAS for the LLIF group OLIF oblique lateral interbody fusion; LLIF lateral lumbar interbody fusion; * P < 0.05 is considered as the factor that contributes to the heterogeneity of effect. If coefficient interval goes across both negative and positive values, it means that no evidence supports that the factor contributes significantly to the heterogeneity of effect this phenomenon may be ascribed to direct injury to lumbar plexus, ilioinguinal nerve or hypogastric nerve that are situated within the abdominal wall, and injury to the genitofemoral nerve on the psoas muscle during the surgical approach. Although thigh pain/numbness and psoas weakness were observed in OLIF group, the incidence of these complications was lower compared to LLIF. However, higher rates of cage subsidence (5.1%), endplate damage (5.2%) and vascular injury (1.7%) were reported for OLIF. The presence of a damaged endplate, an embedded cage, cage sedimentation, and shifting correlated highly, mainly because of endplate lesions [58]. The high risk of cage subsidence after OLIF may be also due to the smaller interbody grafts limited by the discectomy corridor [61]. Besides, vascular injury was a noteworthy complication of OLIF, and the incidence of vascular injury has been reported to range from 1.1 to 2.8% [3,25,61]. Although the incidence of major vascular injury was low, careful preoperative radiological evaluation of vascular anatomy should be conducted to avoid these complications [62]. One limitation of this review is that most of the studies included were retrospective studies with incomplete outcome reporting. Secondly, we did not compare segmental lordosis, global lordosis as more studies need to be included. Thirdly, the fusion result of both techniques may be influenced by the kind of graft material. Another limitation is that all the included studies are searched through an online database but not included unpublished studies, which might have led to publication bias in our meta-analysis. Finally, the interpretation of thigh symptoms was controversial, as surgeons have gradually realized that thigh symptoms disappeared spontaneously 2-3 weeks after surgery, so the definition of this symptom was changed to recognize it as an access-related symptom but not as a complication. However, any metaanalysis has the risk of publication, and we believe that our final result is convincing.

Conclusions
The two groups are similar in terms of radiographic outcomes, operative blood loss, operative time and the length of hospital stay. The OLIF group shows advantages in VAS and ODI scores improvement. Though the incidence of perioperative complications of OLIF and LLIF is similar, the incidence of main complications is significantly different.