Comparison of biportal endoscopic vs minimally invasive transforaminal lumbar interbody fusion for single-level low-grade lumbar spondylolisthesis: A retrospective cohort study

Abstract

BACKGROUND

Lumbar spondylolisthesis is a common spinal disorder that often leads to low back pain, radiating leg pain, and intermittent claudication. Minimally invasive transforaminal lumbar interbody fusion (MI-TLIF) has been widely adopted as an effective surgical treatment for this condition. In recent years, biportal endoscopic transforaminal lumbar interbody fusion (BE-TLIF) has emerged as a novel minimally invasive technique that offers enhanced visualization and may reduce paraspinal muscle injury. However, comparative studies with mid-term follow-up between BE-TLIF and MI-TLIF for single-level low-grade lumbar spondylolisthesis remain limited.

AIM

To compare the mid-term clinical and radiographic outcomes between BE-TLIF and MI-TLIF in the treatment of single-level lumbar spondylolisthesis.

METHODS

We conducted a retrospective comparative cohort study of 63 patients with single-level degree I or II lumbar spondylolisthesis who underwent surgery between December 2020 and December 2022. Thirty-two patients were treated with BE-TLIF and 31 with MI-TLIF. Patient demographics, operative parameters, complications, and clinical outcomes [assessed by Visual Analog Scale (VAS) and Oswestry Disability Index (ODI)] were recorded. Radiographic assessments included disc height (DH), lumbar lordosis (LL), segmental lordosis angle (SLA), slip percentage (SP), and fusion rate (FR) on X-ray and computed tomography scans.

RESULTS

The BE-TLIF group had a significantly longer operative time (189.36 ± 20.94 minutes vs 150.35 ± 21.16 minutes, P < 0.001) but a shorter hospital stay (5.14 ± 1.28 days vs 6.19 ± 1.33 days, P = 0.005) compared to the MI-TLIF group. The BE-TLIF group demonstrated significantly lower low back VAS scores at 3 days postoperatively (3.65 ± 0.78 vs 4.32 ± 0.69, Bonferroni-corrected P = 0.005) and lower postoperative creatine kinase (CK) levels (251.21 ± 29.48 IU/L vs 316.52 ± 33.37 IU/L, P < 0.001). There were no statistically significant differences in ODI scores, DH, LL, SLA, SP, or FR (90.6% vs 90.3%, P = 0.979) between the two groups at the final follow-up. The complication rates did not differ significantly between the two groups.

CONCLUSION

BE-TLIF demonstrated favorable clinical and radiographic outcomes in this cohort, with no statistically significant difference observed compared to MI-TLIF in most mid-term measures; however, these comparisons were exploratory and do not establish equivalence. BE-TLIF was associated with a shorter hospital stay, lower early postoperative pain, and a lower postoperative CK level whose clinical significance requires further study.

Key Words: Spondylolisthesis; Spinal fusion; Unilateral biportal endoscopy; Minimally invasive surgery; Transforaminal lumbar interbody fusion; Retrospective cohort study

Core Tip: This retrospective study compared biportal endoscopic transforaminal lumbar interbody fusion (BE-TLIF) and minimally invasive transforaminal lumbar interbody fusion for single-level lumbar spondylolisthesis. Despite longer operative time, BE-TLIF was associated with shorter hospital stay, less early postoperative pain, and lower creatine kinase levels, associations of uncertain clinical significance. At a mean follow-up of over 40 months, fusion rates, functional outcomes, radiographic parameters, and complication rates did not differ significantly between the groups. BE-TLIF is an effective minimally invasive alternative, with no statistically significant between-group differences in mid-term outcomes observed in this cohort.

INTRODUCTION

In recent years, minimally invasive transforaminal lumbar interbody fusion (MI-TLIF) has become a classic minimally invasive surgery for lumbar spondylolisthesis. It has been widely used in clinical practice due to its small trauma, fast postoperative recovery, short learning time and short hospital stay[1,2]. However, the deep surgical site and narrow working corridor often impair visualization. Excessive retraction from retractors may also induce paraspinal muscle ischemia, potentially compromising MI-TLIF outcomes[3,4]. With the popularization and application of spinal endoscopy, unilateral biportal endoscopy (UBE) has been gradually applied to clinical practice. Owing to its use of dual independent portals for the endoscope and instruments, UBE provides a wider operative field, thereby enhancing visualization and facilitating thorough spinal canal decompression and interbody fusion[5-7]. Heo et al[8]. first applied UBE technique to complete lumbar interbody fusion in 2017 and obtained good clinical results. An increasing number of articles report the application of this technique to patients with mild to moderate spondylolisthesis[9-11]. However, there are few comparative studies on the mid-term outcomes of lumbar spondylolisthesis with two surgical methods at present. The present study offers several distinctive contributions, including a longer follow-up duration, a more comprehensive multidimensional outcome assessment framework, and reporting of mid-term fusion and radiographic parameters in single-level low-grade lumbar spondylolisthesis. The prespecified primary outcome was the Oswestry Disability Index (ODI) score at the final follow-up. Secondary outcomes included Visual Analog Scale (VAS) scores, radiographic parameters [disc height (DH), lumbar lordosis (LL), segmental lordosis angle (SLA), slip percentage (SP)], fusion rate (FR), and complication rates. Therefore, this study retrospectively analyzed the clinical efficacy of biportal endoscopic transforaminal lumbar interbody fusion (BE-TLIF) and MI-TLIF in the treatment of single-level lumbar spondylolisthesis a mean follow-up of over 2 years. The results will provide a more suitable strategy for single-level lumbar minimally invasive fusion.

MATERIALS AND METHODS

General information

This retrospective comparative cohort study included patients with lumbar spondylolisthesis who underwent BE-TLIF or MI-TLIF in our department between December 2020 and December 2022. All cases were consecutively enrolled during the study period, and all surgical procedures were performed by the same team of spine surgeons to minimize operator-related bias. Lumbar X-ray, computed tomography (CT) and magnetic resonance imaging (MRI) examinations were performed before operation. The surgical technique was determined by the same team based on a comprehensive evaluation, taking into account patient preference, economic considerations (BE-TLIF consumables are slightly more expensive), and operating room scheduling constraints related to endoscopic system availability.

Inclusion criteria: (1) Degree I or II lumbar spondylolisthesis (isthmic or degenerative). The diagnosis of lumbar spondylolisthesis was confirmed on standing lateral radiographs. Degenerative spondylolisthesis was defined as anterior vertebral displacement without evidence of a pars interarticularis defect on CT, whereas isthmic spondylolisthesis was diagnosed when a pars interarticularis defect was identified on CT or plain radiographs; (2) With/without lumbar disc herniation or lumbar spinal stenosis; and (3) Low back pain with/without unilateral radicular pain or motor deficit after conservative treatment for more than 6 weeks.

Exclusion criteria: (1) Multi-level spondylolisthesis (≥ 2 levels); (2) Associated kyphosis; and (3) Associated cauda equina syndrome or spinal trauma, tumor, or infection.

A total of 63 patients were included in this study. The protocol received approval from the Ethics Committee of Jinling Hospital (No. 2020DZSKTZX-019). Clinical and radiographic data were extracted from electronic medical records and the PACS system by two independent researchers, with cross-verification performed to ensure accuracy. Any discrepancies were resolved through discussion with a third senior researcher.

Surgical methods

All patients were placed in prone position with sponge pads on both sides of the abdomen to suspend the abdomen.

BE-TLIF procedure: Under C-arm fluoroscopy, bilateral pedicle positions of the target vertebra were identified and marked. Four guide needles were inserted from the superolateral aspect of the pedicles into the vertebral body, followed by guidewires advanced to the anterior third of the vertebral body and secured. Two left-sided incisions were made to establish a caudal working portal and a cranial viewing portal. Soft tissue over the interlaminar space was cleared to fully expose the inferior third of the upper lamina and the superior third of the lower lamina. Partial laminectomy, facetectomy, and ligamentum flavum resection were performed on the left side, with harvested bone preserved for grafting. The dura sac and nerve roots were carefully dissected. The disc was exposed with protection of the dural sac and left nerve root. Discectomy and endplate cartilage removal were performed, with endoscopic confirmation of complete subchondral bone preparation. The harvested bone fragments mixed with allograft were implanted into the disc space, followed by placement of an appropriately sized cage under endoscopic guidance. Four pedicle screws were inserted bilaterally. After guidewire removal, the dural sac and nerve roots were confirmed to be free of compression. Rods were inserted, and screws were alternately tightened to reduce the spondylolisthesis, confirmed by fluoroscopy. Finally hemostasis was achieved with radiofrequency coagulation, a negative-pressure drain was inserted, and the incision was closed in layers (Figure 1).

Figure 1 Key steps of biportal endoscopic transforaminal lumbar interbody fusion procedure, shown in chronological order. A: Exposure of the ligamentum flavum after soft tissue clearance using radiofrequency plasma; B: Resection of the inferior edge of the lamina with a Kerrison rongeur; C: Resection of the medial aspect of the facet joint (white arrows indicate the resected facet margin), with exposure of the dural sac; D: Protection of the traversing nerve root, with the underlying disc exposed; E: Endplate preparation after discectomy (white arrows indicate the prepared endplate surface); F: Insertion of a trial implant to determine the appropriate cage size; G: Placement of the interbody cage under endoscopic guidance; H: Final pedicle screw and rod fixation confirmed by intraoperative fluoroscopy.

MI-TLIF procedure: Following C-arm fluoroscopic localization of the target vertebra, a 5-cm longitudinal paramedian incision was made on the symptomatic side. The Quadrant retractor was inserted to expose the interlaminar space and facet joint. Partial laminectomy, isthmus resection, and facetectomy were performed for decompression, with the harvested bone preserved for grafting. The ligamentum flavum was excised to expose the disc, with careful protection of the dural sac and nerve roots. After standard discectomy and endplate preparation, the previously collected bone fragments mixed with allograft were implanted into the disc space. An appropriately sized interbody cage was then inserted under fluoroscopic guidance. Pedicle screws were inserted on the symptomatic side under direct vision, while a contralateral Wiltse approach was used for screws placement. Rods were inserted and screws were tightened to reduce the spondylolisthesis, with final fluoroscopic confirmation of satisfactory reduction. The wound was irrigated with saline, and the dural sac and nerve roots were confirmed to be free of compression. Two negative-pressure drains were placed (one on each side), and the incision was closed in layers.

Materials and perioperative protocol: The same standardized materials and perioperative protocol were applied to both groups. All patients received polyetheretherketone cages, with sizes ranging from 8 mm × 10 mm × 22 mm to 10 mm × 12 mm × 26 mm, selected based on DL and endplate dimensions. For interbody grafting, a mixture of autograft (harvested during decompression) and demineralized bone matrix allograft was used in all cases. Pedicle screw fixation was performed using the same screw system (Sanyou Medical Co., Ltd., Shanghai, China) in all patients. The perioperative protocol, including prophylactic antibiotics (cefazolin 1 g, 30 minutes before incision), standardized multimodal analgesia, routine imaging follow-up, and standardized rehabilitation, was identical between the two groups. We confirm that all these variables were standardized.

Outcome measures

The operation time, hemoglobin change before and after operation, postoperative hospital stay and operation related complications were recorded. Perioperative complications were defined as any adverse event occurring within 30 days postoperatively. VAS was used to evaluate low back pain before operation, at 3 days, 3 months, 12 months, and at the last follow-up. ODI (version 2.0, validated Chinese version) and radiographic assessments (lumbar spine X-ray and CT) were taken preoperatively, postoperative 3 months, postoperative 12 months, and at the last follow-up. All VAS and ODI assessments were performed by an independent researcher who was blinded to the surgical approach. VAS referred to low back pain at all time points.

X-ray measurements include: (1) DH: At the operative level: Distance between the lower endplate of spondylolisthesis and the upper endplate of lower vertebral body on lateral X-ray; (2) LL: The angle between the superior endplate of L1 and the superior endplate of S1 on lateral X-ray; (3) SLA: The angle between the inferior endplate of the cranial vertebra and the superior endplate of the caudal vertebra within the surgical segment on lateral radiographs; (4) SP: The slip distance of the upper vertebra on the lower vertebra divided by the anteroposterior diameter of the lower vertebra multiplied by 100%. Reduction rate: (preoperative SP-postoperative SP)/preoperative SP × 100%; and (5) FR: Interbody fusion was assessed using Bridwell interbody fusion grading system on CT at the last follow-up[12]. According to this system, Grade I is defined as fused with remodelling and trabeculae present; Grade II as an intact graft with incomplete remodeling; Grade III as an intact graft with lucency at the cranial and/or caudal end; and Grade IV as absence of fusion with graft collapse or resorption. Grades I and II were considered indicative of successful fusion. The FR was calculated as: (number of segments with successful fusion/total number of segments) × 100%. The two spine surgeons were masked to the surgical approach (BE-TLIF vs MI-TLIF) during all radiological measurements to minimize potential bias. The data were recorded as the average of the two measurements. If there were discrepancies in the conclusions, the evaluators reached consensus after discussion. Intraclass correlation coefficients (ICC) with 95%CI were calculated to assess inter-observer reliability for all radiological measurements: DH (ICC = 0.87, 95%CI: 0.78-0.93), LL (ICC = 0.89, 95%CI: 0.81-0.94), SLA (ICC = 0.86, 95%CI: 0.76-0.92), and SP (ICC = 0.92, 95%CI: 0.86-0.96), indicating excellent inter-observer reliability. For the categorical Bridwell fusion grading, inter-observer agreement was assessed using weighted kappa with quadratic weights (weighted κ = 0.77; 95%CI: 0.61-0.88, by nonparametric bootstrap), indicating substantial agreement.

Statistical analysis

Statistical analysis was performed using IBM SPSS Statistics (version 23.0; IBM Corp., Armonk, NY, United States). Continuous variables were expressed as mean ± SD. Repeated-measures ANOVA with a compound symmetry covariance structure was used to compare VAS, ODI, and radiographic parameters (DH, LL, SLA, SP) between the two groups across multiple time points, with time as the within-subject factor, group as the between-subject factor, and baseline values included as covariates in the model. The compound symmetry assumption was assessed using Mauchly’s test; when the assumption was violated, Greenhouse-Geisser-corrected results were reported. Normality of residuals was assessed with the Shapiro-Wilk test and Q-Q plots, and homogeneity of variance with Levene’s test. Given the multiple time points and secondary outcomes, secondary analyses were interpreted as exploratory; post-hoc pairwise comparisons were performed using Bonferroni correction. Fisher’s exact test was used to compare FR and complications. Effect estimates are reported as mean differences or risk differences with 95%CIs for all primary and secondary outcomes. P < 0.05 was statistically significant. The prespecified primary outcome was the ODI score at final follow-up. Statistical power was estimated with reference to a clinically meaningful between-group difference in the final ODI, using the between-group ODI minimal clinically important difference (MCID) of 4 points reported in the literature[13]. With 32 and 31 patients in the BE-TLIF and MI-TLIF groups, respectively, and an observed pooled standard deviation of 3.12 (α = 0.05, two-tailed), the study had approximately 99% power to detect a between-group difference of 4 points in the final ODI. This study was not designed or powered to demonstrate equivalence or noninferiority; therefore, the absence of a statistically significant difference should not be interpreted as evidence of comparable efficacy.

RESULTS

Demographics and operative data

All patients in BE-TLIF and MI-TLIF groups successfully completed the surgery without changing the surgical method during the surgery. All 63 patients completed postoperative follow-up. No missing data were present for any variable of interest (Figure 2). The follow-up duration ranged from 36 to 48 months, with a mean of 41.23 ± 3.50 months in BE-TLIF group and 40.52 ± 3.50 months in MI-TLIF group. There was no statistically significant difference between the two groups in terms of sex composition, average age, BMI, course of disease, and smoking status, level segment, spondylolisthesis type, Meyerding grade, and follow-up time (P > 0.05) (Table 1).

Figure 2 Participant flow diagram. A total of 81 patients with lumbar spondylolisthesis who underwent surgery between December 2020 and December 2022 were screened. Eighteen patients were excluded according to the prespecified exclusion criteria: 11 with spondylolisthesis involving ≥ 2 levels, 3 with associated kyphosis, and 4 with associated cauda equina syndrome or spinal trauma, tumor, or infection. The remaining 63 patients were enrolled and allocated to the biportal endoscopic transforaminal lumbar interbody fusion group (n = 32) or the minimally invasive transforaminal lumbar interbody fusion group (n = 31). All patients completed the follow‑up and were included in the final analysis. BE-TLIF: Biportal endoscopic transforaminal lumbar interbody fusion; MI-TLIF: Minimally invasive transforaminal lumbar interbody fusion.

Table 1 Demographics and operative data, n (%)/mean ± SD.

Variable	BE-TLIF (n = 32)	MI-TLIF (n = 31)	t/χ2	P value
Sex (male/female)	14/18	12/19	0.026	0.872
Age (year)	51.95 ± 9.20	54.29 ± 12.59	-0.758	0.451
BMI (kg/m2)	23.87 ± 1.92	23.81 ± 2.02	0.12	0.905
Course of disease (months)	12.30 ± 3.22	13.10 ± 2.68	1.07	0.289
Smoking (yes/no)	16/16	15/16	0.016	0.900
Level
L3/4	3 (9.4)	5 (16.1)		0.7691
L4/5	23 (71.9)	20 (64.5)
L5/S1	6 (18.8)	6 (19.4)
Spondylolisthesis type
Degenerative	24 (75.0)	25 (80.6)	0.440	0.507
Isthmic	8 (25.0)	6 (19.4)
Meyerding grade
I	26 (81.3)	24 (77.4)	0.152	0.697
II	6 (18.8)	7 (22.6)

¹Fisher-Freeman-Halton exact test for overall distribution of levels.

BMI: Body mass index; BE-TLIF: Biportal endoscopic transforaminal lumbar interbody fusion; MI-TLIF: Minimally invasive transforaminal lumbar interbody fusion.

Perioperative parameters and complications

There was no statistically significant difference in hemoglobin levels, C-reactive protein (CRP) and surgical complications between the two groups (P > 0.05). The operative time was significantly longer in the BE-TLIF group compared to the MI-TLIF group (P < 0.001). However, the length of hospital stay in the BE-TLIF group was significantly shorter than that in MI-TLIF group (P = 0.005). In addition, postoperative creatine kinase (CK) levels were significantly higher in the MI-TLIF group compared to the BE-TLIF group (P < 0.001). The difference in complications was not statistically significant between the two groups (Table 2).

Table 2 Comparison of perioperative parameters and complications, n (%)/mean ± SD.

Variable	BE-TLIF (n = 32)	MI-TLIF (n = 31)	MD/RD (95%CI)	t/χ2	P value
Operation time (minute)	189.36 ± 20.94	150.35 ± 21.16	39.01 (28.40 to 9.62)	6.65	< 0.001
Hospital stay (days)	5.14 ± 1.28	6.19 ± 1.33	-1.05 (-1.71 to 0.39)	-2.93	< 0.0051
Preop Hb (g/L)	136.77 ± 12.59	138.10 ± 11.24	-1.33 (-7.34 to 4.68)	-0.41	0.681
Postop 24 hours Hb (g/L)	126.23 ± 13.50	125.84 ± 14.57	0.39 (-6.69 to 7.47)	0.10	0.919
Preop CRP (mg/L)	5.68 ± 2.34	6.03 ± 2.71	-0.35 (-1.63 to 0.93)	-0.50	0.621
Postop 24 hours CRP (mg/L)	20.90 ± 6.03	21.00 ± 6.15	-0.10 (-3.17 to 2.97)	-0.06	0.949
Preop CK (IU/L)	82.34 ± 35.74	84.26 ± 34.36	-1.92 (-19.58 to 15.74)	-0.20	0.844
Postop 24 hours CK (IU/L)	251.21 ± 29.48	316.52 ± 33.37	-65.31(-81.20 to -49.42)	-7.35	< 0.001
Complications	2 (6.3)	3 (9.7)	-3.4% (-16.8% to 9.9%)		0.672
Dural tear	1	2			0.610
Numbness	1	1			1.000

¹Fisher’s exact test.

BE-TLIF: Biportal endoscopic transforaminal lumbar interbody fusion; MI-TLIF: Minimally invasive transforaminal lumbar interbody fusion; Hb: Hemoglobin; CK: Creatine kinase; CRP: C-reactive protein.

In the BE-TLIF group, one case of dural tear occurred intraoperatively. Given the small size of the defect, no direct repair was performed. Postoperatively, the patient was placed in a Trendelenburg position (head-down tilt). The drain was removed after the drainage fluid was observed to be without significant sanguineous content. No delayed complications (e.g., cerebrospinal fluid leak, headache, or meningitis) were observed in any of the patients with dural tears during the follow-up period. One case of transient lower limb numbness was reported postoperatively, which alleviated after three weeks of neurotrophic medication. Numbness was assessed based on patient-reported symptoms and physical examination; no formal dermatomal mapping was performed. In the MI-TLIF group, two cases of dural tear and one case of transient lower limb numbness occurred. These were managed in the same manner as described above (Table 2).

Clinical outcomes

The clinical outcome scores of the both groups improved at all postoperative follow-up points (Table 3). There was no statistical difference in VAS score and ODI index between the two groups at preoperatively, 3 months postoperatively, 12 months postoperatively or the final follow-up (P > 0.05). However, the BE-TLIF group demonstrated significantly lower low back VAS scores at 3 days postoperatively compared with the MI-TLIF group (P = 0.005).

Table 3 Comparison of clinical outcomes, n (%)/mean ± SD.

Variable	BE-TLIF (n = 32)	MI-TLIF (n = 31)	MD/RD (95%CI)	Group × time interaction
VAS				F = 3.82, P = 0.007
Preoperative	7.86 ± 1.08	7.45 ± 1.03	0.41 (-0.17 to 0.99)
Postoperative 3 days	3.65 ± 0.78	4.32 ± 0.69	-0.67 (-1.08 to -0.26)	P = 0.0052
Postoperative 3 months	3.48 ± 0.51	3.64 ± 0.49	-0.16 (-0.43 to 0.11)
Postoperative 12 months	1.95 ± 0.84	2.06 ± 0.81	-0.11 (-0.57 to 0.35)
Final follow-up	0.91 ± 0.82	0.86 ± 0.80	0.05 (-0.40 to 0.50)
ODI				F = 0.52, P = 0.721
Preoperative	69.68 ± 5.65	71.48 ± 5.49	-1.80 (-5.49 to 1.89)
Postoperative 3 months	24.09 ± 2.78	24.23 ± 2.63	-0.14 (-1.56 to 1.28)
Postoperative 12 months	22.18 ± 4.91	21.84 ± 4.08	0.34 (-2.07 to 2.75)
Final follow-up	14.91 ± 3.22	15.77 ± 3.02	-0.86 (-2.42 to 0.70)
MCID achievement	25 (78.1)	23 (74.2)	3.9 (-17.1 to 24.9)	P = 0.7241

¹Bonferroni-corrected P values (adjusted significance threshold α = 0.01).

²Fisher’s exact test.

VAS: Visual analogue scale; ODI: Oswestry Disability Index; MD: Mean difference; RD: Risk difference; BE-TLIF: Biportal endoscopic transforaminal lumbar interbody fusion; MI-TLIF: Minimally invasive transforaminal lumbar interbody fusion.

In addition, we calculated the proportion of patients achieving MCID for ODI scores. Using the commonly accepted MCID threshold of 12.8 points for ODI in lumbar degenerative disease[14], the MCID achievement rate was 78.1% (25/32) in the BE-TLIF group and 74.2% (23/31) in the MI-TLIF group, with no statistically significant difference between the two groups (χ² = 0.125, P = 0.724). MCID estimates vary by measurement context; the 12.8-point ODI threshold used here applies to within-group change, whereas the 4-point ODI threshold used for the a priori power calculation applies to between-group comparisons.

Radiological parameters

Radiological parameters of the two groups are shown in Table 4. Both groups showed improvements in SLA from baseline (within-group comparisons), but no statistically significant between-group differences were observed (Group × Time interaction P > 0.05). Both groups showed significant improvements in DH, LL, and SP from baseline to the final follow-up (P < 0.05). At all evaluated time points-preoperative, 3 months postoperative, 12 months postoperative, and final follow-up-there were no statistically significant differences between two groups in DH, LL, or SP (P > 0.05). At final follow-up, the difference in FR between BE-TLIF group (90.6%) and MI-TLIF group (90.3%) was not statistically significant. A typical case in BE-TLIF group is shown in Figure 3, and one from MI-TLIF group is shown in Figure 4.

Figure 3 A 54-year-old female patient with L4 lumbar spondylolisthesis (Meyerding grade I) from the biportal endoscopic transforaminal lumbar interbody fusion group. A and B: Preoperative anteroposterior and lateral radiographs; C and D: Postoperative 3-month anteroposterior and lateral radiographs; E and F: Anteroposterior and lateral X-ray at the final follow-up after operation showed good reduction; G and H: Computed tomography scans at the final follow-up after operation showed that the reconstruction of the trabeculae between the upper and lower endplates, and the position of the Cage was good.

Figure 4 A 56-year-old female patient with L4 lumbar spondylolisthesis (Meyerding grade I) in minimally invasive transforaminal lumbar interbody fusion group. A and B: Preoperative antero-posterior and lateral radiographs; C and D: Postoperative 3-month antero-posterior and lateral radiographs; E and F: Antero-posterior and lateral X-ray at the final follow-up after operation showed good reduction; G and H: Computed tomography scans at the final follow-up after operation show good trabecular reconstruction between the upper and lower endplates and a good Cage position.

Table 4 Radiological parameters, n (%)/mean ± SD.

Variable	BE-TLIF (n = 32)	MI-TLIF (n = 31)	MD/RD 95%CI	Group × time interaction
DH (mm)				F = 0.58, P = 0.630
Preoperative	9.06 ± 0.78	8.86 ± 0.54	0.20 (-0.21 to 0.61)
Postoperative 3 months	10.33 ± 0.67	10.60 ± 0.56	-0.27 (-0.61 to 0.07)
Postoperative 12 months	10.35 ± 0.68	10.62 ± 0.63	-0.27 (-0.62 to 0.08)
Final follow-up	10.40 ± 0.60	10.62 ± 0.58	-0.22 (-0.54 to 0.10)
LL (°)				F = 0.45, P = 0.718
Preoperative	31.31 ± 0.61	31.15 ± 0.58	0.16 (-0.21 to 0.53)
Postoperative 3 months	32.04 ± 0.55	31.92 ± 0.52	0.12 (-0.20 to 0.44)
Postoperative 12 months	34.00 ± 0.49	34.14 ± 0.59	-0.14 (-0.38 to 0.10)
Final follow-up	33.89 ± 0.53	33.95 ± 0.54	-0.06 (-0.28 to 0.16)
SLA (°)				F = 0.21, P = 0.889
Preoperative	9.49 ± 1.12	9.76 ± 1.11	-0.27 (-0.84 to 0.30)
Postoperative 3 months	11.65 ± 0.90	11.49 ± 0.95	0.16 (-0.31 to 0.63)
Postoperative 12 months	11.88 ± 0.62	11.92 ± 0.59	-0.04 (-0.29 to 0.21)
Final follow-up	11.63 ± 0.74	11.68 ± 0.82	-0.05 (-0.39 to 0.29)
SP (%)				F = 0.36, P = 0.782
Preoperative	17.65 ± 3.70	18.59 ± 4.33	-0.94 (-3.42 to 1.54)
Postoperative 3 months	3.42 ± 1.12	3.62 ± 1.44	-0.20 (-0.92 to 0.52)
Postoperative 12 months	3.59 ± 0.71	3.70 ± 1.06	-0.11 (-0.51 to 0.29)
Final follow-up	3.66 ± 0.77	3.74 ± 0.97	-0.08 (-0.44 to 0.28)
FR (%)	29 (90.6)	28 (90.3)	0.3 (-14.2 to 14.8)	P = 0.9791

¹Fisher’s exact test.

DH: Disc height; LL: Lumbar lordosis; SLA: Segmental lordosis angle; SP: Slip percentage; FR: Fusion rate; MD: Mean difference; RD: Risk difference; BE-TLIF: Biportal endoscopic transforaminal lumbar interbody fusion; MI-TLIF: Minimally invasive transforaminal lumbar interbody fusion.

DISCUSSION

Lumbar spondylolisthesis is a common spinal disorder that often leads to varying degrees of low back pain, radiating leg pain, lower limb numbness, and intermittent claudication. Non-surgical management often yields unsatisfactory results. Lumbar decompression and interbody fusion has been proven to be one of the most effective surgical treatments for lumbar spondylolisthesis[15].

Since Foley et al[16] first applied the retractor-based MI-TLIF technique for lumbar spondylolisthesis, it has been widely adopted in clinical practice, gradually replacing traditional TLIF procedures[16,17]. Several studies have reported that compared to TLIF, MI-TLIF offers benefits such as reduced blood loss, earlier recovery, and decreased rates of deep wound infections[18,19]. However, a limitation to the development of MI-TLIF is the constrained surgical field and working space imposed by the retractor system[20]. With technological advancements, full endoscopic lumbar interbody fusion is currently one of the most prominent techniques in the field of minimally invasive spine surgery. Compared to uniportal techniques, biportal endoscopic interbody fusion is garnering increasing attention[6]. Several studies reporting outcomes of BE-TLIF for lumbar spondylolisthesis demonstrate advantages including minimal trauma, less bleeding, and rapid early postoperative recovery[9,21]. Nevertheless, comparative studies with mid-term follow-up between BE-TLIF and MI-TLIF remain scarce. Therefore, this study compared the clinical efficacy and radiographic outcomes of these two techniques for treating lumbar spondylolisthesis with a minimum follow-up of two years.

Consistent with findings in the literature[20,22], our results showed that the BE-TLIF group had a shorter hospital stay compared to the MI-TLIF group (5.14 ± 1.28 days vs 6.19 ± 1.33 days), but a longer operative time (189.36 ± 20.94 minutes vs 150.35 ± 21.16 minutes). We attribute the prolonged operative time in the BE-TLIF group to the necessity for meticulous hemostasis to maintain a clear endoscopic visual field and the influence of the learning curve associated with the UBE technique. A multicenter study reported that spine surgeons require approximately 39 cases to overcome the learning curve for BE-TLIF, defined as achieving an operative time of less than 160 minutes[23]. The senior surgeon in our team had performed approximately 30 cases of UBE decompression procedures prior to the introduction of BE-TLIF; therefore, we observed that surgeons with prior UBE experience might require fewer cases to reach the proficiency threshold.

It is important to acknowledge that treatment allocation in this retrospective cohort was influenced by several non‑clinical factors beyond surgeon assessment. Patient preference, cost considerations (BE‑TLIF consumables are slightly more expensive), the availability of the endoscopic system, and the calendar time of surgery all played a role in determining whether a patient underwent BE‑TLIF or MI‑TLIF. These factors may have introduced selection bias and confounding, and their potential impact on the observed outcomes should be considered when interpreting the results.

In our study, compared to preoperative scores, both the VAS and ODI scores significantly decreased at all postoperative follow-up time points in both groups. This indicates that both BE-TLIF and MI-TLIF effectively improved symptoms in patients with lumbar spondylolisthesis. Regarding the primary outcome, the 95%CI for the between-group difference in the final ODI (-0.86 points; 95%CI: -2.42-0.70) lay entirely within the prespecified ± 4-point clinical margin, indicating that any true between-group difference, if present, is unlikely to be of clinical importance. Post-hoc power based on the observed effect size (approximately 21%) is not used to interpret the results, because observed power is a one-to-one function of the p-value and conveys no information beyond the significance test itself. Notably, the BE-TLIF group demonstrated significantly lower VAS scores at 3 days postoperatively (3.65 ± 0.78) than the MI-TLIF group (4.32 ± 0.69). This finding aligns with previous studies[21,24,25] and is consistent with the hypothesis that BE-TLIF may be associated with less soft tissue disruption, although this remains speculative and requires direct confirmation. Although the difference in VAS scores at 3 days postoperatively was statistically significant after Bonferroni correction, this isolated finding was observed at a single early time point and did not persist at subsequent follow-ups. Moreover, the observed difference (0.67 points) fell below the generally accepted MCID threshold of 1.5 points for VAS in lumbar spine surgery[26]. Given the multiple time points and outcomes analyzed, the possibility of a type I error (chance finding) cannot be excluded. Therefore, this early VAS finding should be interpreted with caution. It is well-established that MI-TLIF represents a significant improvement over traditional TLIF in terms of muscle damage and bleeding[27,28]. However, sustained retraction from the retractors can lead to muscle ischemia, potentially exacerbating postoperative pain and impairing muscle function[29]. In contrast, the UBE procedure does not require retractors; the surgical corridor is established through the intermuscular plane of the multifidus, theoretically minimizing soft tissue disruption. Postoperative levels of CRP and CK can reflect the degree of muscle injury and inflammation at the surgical site[30]. In our study, the lower postoperative CK level was observed in the BE-TLIF group (251.21 ± 29.48 IU/L vs 316.52 ± 33.37 IU/L, P < 0.001); whether this association reflects a true reduction in paraspinal muscle injury remains uncertain. Because the postoperative CRP level did not differ significantly between the two groups (20.90 ± 6.03 mg/L vs 21.00 ± 6.15 mg/L, P = 0.949), indicating no significant between-group difference in the systemic inflammatory response. Therefore, while CK levels may reflect local muscle damage, the absence of a corresponding CRP difference suggests that the clinical significance of this CK reduction requires further validation through direct muscleinjury measurements (e.g., MRI muscle atrophy assessment) in future studies. In this study, hemoglobin change was used as a reference indicator of perioperative blood loss. However, we acknowledge that hemoglobin change alone may be influenced by fluid administration and does not fully represent intraoperative blood loss.

Restoring LL and reducing the degree of vertebral slippage are crucial indicators for ensuring successful outcomes in lumbar spondylolisthesis surgery[21,31]. The results of this study demonstrated significant improvements in LL, SLA, and SP in both groups compared to preoperative values, with no statistically significant differences observed between the groups. These results indicate that both approaches achieved similar improvements in these radiographic parameters; however, because this study was not designed to demonstrate equivalence, the absence of a statistically significant between-group difference should not be interpreted as evidence of comparable efficacy.

Compared to open surgery, the use of a drill for decompression under endoscopy may yield less autologous bone graft, and some surgeons express concern that continuous irrigation in endoscopic surgery might interfere with interbody fusion. While the quantity of autograft obtained during MI-TLIF decompression is similar to that in open surgery, the obstruction from the tube and bleeding within the spinal canal can sometimes hinder complete removal of the cartilaginous endplate or potentially damage the subchondral bone, particularly in osteoporotic patients[32]. Recent clinical studies on BE-TLIF suggest that using Kerrison rongeurs for decompression can provide sufficient autologous bone graft. Furthermore, the direct and magnified visualization under endoscopy allows for thorough preparation of the endplate to create an optimal graft bed while preserving the subchondral bone, theoretically reducing the risk of cage subsidence. Our findings regarding FR are consistent with other reports, showing no statistically significant difference between the BE-TLIF and MI-TLIF groups[9,11]. Given the limited power for FR comparisons, the observed similarity should be interpreted cautiously and likely reflects a ceiling effect rather than definitive evidence of equivalence. The FR in the BE-TLIF group (90.6%) was slightly lower than the 94.7%-100% rates reported in some recent literature[33-35]. We speculate this might be related to the non-use of bone morphogenetic protein-2 for interbody grafting in this subset of patients, although this remains to be confirmed in future studies.

Although we obtained satisfactory mid-term clinical outcomes, this study has several limitations. First, it was a retrospective, single-center study with a relatively small sample size comparing the two techniques. Although the study had high power (approximately 99%) to detect a clinically important between-group difference of 4 points (the ODI MCID) in the final ODI, it was not designed a priori as an equivalence or noninferiority trial; therefore, the absence of a statistically significant difference should be interpreted with caution and should not be interpreted as evidence of equivalence. Given the limited number of complications (only five events in total) and the fact that this study was not designed or powered to demonstrate equivalence in fusion or complication rates, the lack of a statistically significant difference should be interpreted with caution. Second, the spectrum of lumbar pathologies included was limited, encompassing only Grade I-II spondylolisthesis and excluding cases with severe disc degeneration. The results may not be applicable to patients with higher-grade slips, multilevel disease, severe disc degeneration, or revision surgeries, and they also require confirmation in other surgical settings with different levels of surgeon experience. Third, because of incomplete documentation in some retrospective records, certain baseline variables-including severity of spinal stenosis, disc degeneration grade, neurologic deficits, bone mineral density-could not be reliably retrieved, which limited a more comprehensive baseline characterization. To address these limitations, a prospective, multicenter, randomized controlled trial is planned to further explore the efficacy of BE-TLIF in managing complex lumbar degenerative conditions.

CONCLUSION

In this cohort, BE‑TLIF demonstrated favorable clinical and radiographic outcomes, with no statistically significant difference observed compared to MI‑TLIF in most mid‑term measures. BE-TLIF was associated with a shorter hospital stay, lower early postoperative pain, and a lower postoperative CK level. However, given the retrospective design and modest sample size of this study, these findings should be interpreted as hypothesis‑generating and warrant further validation through prospective, multicenter, randomized controlled trials.