**3. Results**

This study included 91 patients with full preoperative and postoperative radiographic data with a minimum 1 year follow-up. The cohort included 76 women and 15 men (mean age, 73.2 years; mean follow-up period, 24.2 months). Of the patients, 22 underwent LLIF and 69 underwent PLIF/TLIF. Propensity score matching resulted in 21 pairs of patients undergoing LLIF and PLIF/TLIF. Consequently, biases between the treatment groups diminished after the propensity score matching. The patients' ages and sexes, BMIs, number of intervertebral fusion levels, and radiographic parameters before operation were adjusted (Table 1).


**Table 1.** Patient characteristics in the lateral lumbar interbody fusion (LLIF) group and the posterior lumbar interbody fusion (PLIF)/transforaminal lumbar interbody fusion (TLIF) group after the matching.

Mean ± standard deviation; LLIF, lateral lumbar interbody fusion; PLIF, posterior lumbar interbody fusion; TLIF, transforaminal lumbar interbody fusion; BMI, body mass index; SVA, sagittal vertical axis; LL, lumbar lordosis; PI, pelvic incidence; PT, pelvic tilt; TK, thoracic kyphosis; deg., degree; No., number.

The LLIF group had significantly lower intraoperative blood loss (849 vs. 2359 mL). However, total surgical time was significantly longer in the LLIF group (536 vs. 421 min; Table 2). All parameters were significantly improved at 4 weeks after surgery and at the final follow-up compared with those at the preoperative period in both groups (Table 3). Consequently, most of the radiographic parameters were not significantly different between the two groups postoperatively. However, SVA tended to be smaller in the LLIF group at the final follow-up (Table 3).

**Table 2.** Surgical invasion in the LLIF group and the PLIF/TLIF group.


 interbody posterior interbody interbody fusion; ± standard deviation; LLIF, lateral lumbar interbody fusion; PLIF, posterior lumbar interbody fusion; TLIF, transforaminal lumbar inter-body fusion; \*, *p* < 0.05.

**Table 3.** The comparison of the LLIF group and the PLIF/TLIF group for radiographic parameters at each follow-up time.



**Table 3.** *Cont*.

Mean ± standard deviation; LLIF, lateral lumbar interbody fusion; PLIF, posterior lumbar interbody fusion; TLIF, transforaminal lumbar inter-body fusion; SVA, sagittal vertical axis; LL, lumbar lordosis; PI, pelvic incidence; PT, pelvic tilt; TK, thoracic kyphosis; deg., degrees; \**, p* < 0.05.

Table 4 shows the incidence of surgical complications in both groups. No significant difference in the incidence of either overall local or systemic complications was observed in terms of perioperative complications. However, the incidence of epidural hematoma was significantly lower in the LLIF than in the PLIF/TLIF group (0.0% vs. 19.0%; *p* = 0.035). Moreover, the incidence of late complications was not significantly different between the two groups.

**Table 4.** Surgical complications of the LLIF group and the PLIF/TLIF group.


Mean \*: *p* < 0.05.

#### **4. Discussion**

The surgical invasiveness and the associated risks of complications in ASD surgeries remain problematic. Recently, the minimally invasive LLIF technique has been increasingly performed to potentially reduce the surgical risks of ASD surgeries [12,19–21]. However, insufficient information exists on whether LLIF indeed decreases the surgical invasiveness and the incidence of surgical compilations in ASD patients. Therefore, the use of PLIF/TLIF and the recently increased use of LLIF were compared, both of which are intervertebral corrections for the treatment of ASD. Furthermore, the propensity matching method was used to compare the surgical results in the two procedures. As surgical treatment for

ASD is generally a high-risk surgery [8,9], conducting a randomized trial is difficult. The propensity score is a balancing score calculated by logistic regression analysis, which makes the distribution of measured baseline covariates similar between the two treatment groups [22]. In the current study, all of the covariates were successfully adjusted after one-to-one propensity score matching (Table 1), indicating no selection bias in the baseline characteristics between the LLIF and PLIF/TLIF groups.

Both the LLIF and PLIF/TLIF groups in this study resulted in favorable sagittal alignment correction, with significant improvements in SVA, LL, and PT, as well as PI−LL mismatch, compared with the preoperative parameters. In the comparison between the LLIF and PLIF/TLIF groups, SVA was smaller and the improvement in SVA from the preoperative value tended to be larger at the final follow-up in the LLIF group. However, we did not find marked differences between the two groups in terms of postoperative LL and PI−LL at 4 weeks after surgery and at the final follow-up. Previous studies have reported that LLIF is better able to correct sagittal imbalance than posterior corrective fusion in ASD patients. However, the surgical procedure of the posterior approach in these studies was the posterior spinal fusion with interbody fusion only at L5/S1, not multilevel PLIF/TLIF [11,23,24]. In this study, four-level interbody fusion (L2/3–5/S) was conducted in both the LLIF and PLIF/TLIF groups. In the LLIF procedure, interbody space can be lifted up using large cages, and may potentially have greater ability to restore segmental alignment. However, if the procedure is performed at multiple segments, the tightness of the anterior longitudinal ligament (ALL) limits the degree of lift-up and the correction of lumbar lordosis. Thus, the correction angle of LL in the LLIF group was slightly higher but not significant compared with that of the PLIF/TLIF group (LLIF, 44.1◦ ± 15.1◦; PLIF/TLIF, 39.4◦ ± 15.5◦). The release of ALL in addition to the LLIF procedure (anterior column realignment technique) would make a more radical correction [25,26], although the potential risk of vascular injury exists.

Since ASD mostly affects the elderly population, surgical invasiveness for ASD patients is a major problem [23,24]. Intraoperative blood loss is one of the major factors related to surgical invasiveness. Thus, this study demonstrated that the intraoperative blood loss of the LLIF group was reduced to one-third compared with the PLIF/TLIF group. In the PLIF/TLIF group, access to the intervertebral disc space was performed through the epidural space, and therefore, bleeding from the epidural venous plexus was inevitable at multilevel intervertebral discs. However, the LLIF group showed minimized epidural bleeding because access to the intervertebral disc space was performed through the retroperitoneal approach, except access to the L5/S level. Previous studies also reported the benefits of utilizing the lateral approach in reducing intraoperative blood loss [23,24]. Consequently, the total operation time in the LLIF group averaged 100 min longer than that in the PLIF/TLIF group. The PLIF/TLIF group was performed in this study through a single posterior approach, whereas the LLIF group was performed through the combined approach: the lateral oblique approach for anterior interbody fusion, and the subsequent posterior approach for pedicle screw fixation. Thus, the combined approach makes the total operating time much longer in the LLIF group. Previous reports have shown that blood loss is a risk factor for perioperative complications in lumbar fusion surgery; however, surgical time was not a significant factor [27–29]. Despite the longer operating time, LLIF procedures may have the benefit of reducing the bleeding and surgical risks through the combined approach. Additionally, two-stage surgery (performing anterior and posterior approaches on a separate day) can be selected for high-risk patients in LLIF procedures. Thus, the LLIF procedure is considered to be a good option for safe, corrective surgery for elderly ASD patients because patients with ASD are commonly older in age. Percutaneous posterior fixation, which is applicable for flexible patients, would further reduce surgical invasiveness [30–34].

The overall incidence of perioperative complications, including local and systemic complications, was not significantly different between the two groups in the present study. However, the incidence of epidural hematoma was significantly higher in the PLIF/TLIF

group than in the LLIF group. The possible reasons for the reduced hematoma incidence are that the LLIF group in this study required minimal epidural manipulation, and that surgery-related bleeding occurred significantly less in the LLIF group compared to the PLIF/TLIF group. Previous studies reported that the abundant intraoperative blood loss was a significant risk factor for early perioperative complications in the fusion and instrumentation of degenerative lumbar scoliosis [27–29]. Therefore, the LLIF procedure may potentially reduce surgical risks for ASD patients. In terms of late complications, no differences were noted in overall incidence, implant failure, PJK, or newly occurred vertebral fractures between the two groups. This result is similar to previous reports of comparisons between LLIF and single posterior approaches in surgical treatments for ASD [11,23,24,35]. We generally use hard braces after the surgery for 3–6 months. However, the long-term use of hard braces is reported to cause complications in elderly patients [36]. It was also reported that bracing after deformity-correction surgery did not effectively reduce PJK [37]. At the final stages, shifting from a hard brace, to a soft brace, or a dynamic brace [38] may be important for avoiding brace-related complications, preserving back muscle strength and improving QOL.

This study has several limitations. The main limitations are the small number of patients and their short-term follow-up. Further studies should involve a larger sample size and a longer follow-up period. In addition, HR-QOL was not investigated in this study. Previous studies have reported that lumbar spinal fusions and corrective surgeries are effective in improving QOL [6,24,39–41]. Further studies should be conducted using HR-QOL evaluations, such as the Oswestry Disability Index, SRS-22, and SF-36. Lastly, the choice of surgical approaches was not randomized in this study. A propensity score matching analysis was conducted to minimize the selection bias of the surgical procedures.

Despite these limitations, this is the first study that compared multilevel LLIF and PLIF/TLIF procedures for ASD patients using propensity score matching methods, clearly showing that intraoperative bleeding and postoperative hematoma were reduced in the LLIF procedures.
