Endoscopic Versus Minimally Invasive Transforaminal Lumbar Interbody Fusion In 1-Segment Lumbar Spondylolisthesis: A Prospective Randomised Pilot Study

Background: There are currently several minimally invasive techniques for lumbar spine interbody fusion that have been shown to minimize surgical and perioperative morbidity. We aimed to evaluate the curative ecacy of endoscopic transforaminal lumbar interbody fusion (Endo-TLIF) by comparing perioperative characteristics and 1.5-year observational outcomes in 1-segment lumbar spondylolisthesis between the minimally invasive TLIF (MIS-TLIF) technique and the optimized Endo-TLIF technique. Methods: One hundred and two patients treated by MIS-TLIF (48 cases) or Endo-TLIF (54 cases) were included from March 2018 to April 2019. Perioperative parameters and clinical outcomes were evaluated. Inammatory biomarkers were measured for postoperative traumatic stress and muscle injury. Fusion rates were determined at 18 months after surgery by CT. Results: The Endo-TLIF group had similar incision length, return to work time and rate (p>0.05). Blood loss, left bed time and analgesic ratio were signicantly less in Endo-TLIF group (p<0.05). The Endo-TLIF group had a signicantly longer operative and radiation exposure time compared with MIS-TLIF group (p<0.05). The CRP and ESR levels were lower in the Endo-TLIF group compared with the MIS-TLIF group (p<0.05). The VAS and ODI improved signicantly in both groups after surgery. Signicant decreases in low back VAS in the Endo-TLIF group were found at postoperative day 1 and 3 months after surgery (p<0.05). The fusion rate in the Endo-TLIF group (90.7%) was similar to that in the MIS-TLIF group (95.8%). Conclusions: Endo-TLIF has comparable perioperative analysis and clinical outcomes in comparison to MIS-TLIF and manifests a greater improvement in less multidus muscle damage, inammation and faster patient recovery. Thus, Endo-TLIF should be considered as a feasible and effective technique


Background
Spinal fusion has been shown to be the preferred surgical option to reduce chronic back pain, back-related leg pain, and increase quality of life in the treatment of lumbar degenerative disorders (LDD) [1]. Although conventional posterior open fusion surgeries can achieve wide decompression of neural structures and provide stabilization for surgically treated segments, they could also result in extensive destruction of posterior anatomical structures, requiring long periods of recovery. Meanwhile, the psychological, economic and social costs of the postoperative period should not be underestimated. Therefore, various types of minimally invasive (MIS) spine surgeries have been attempted to treat LDD [2][3][4][5]. These MIS techniques have resulted in less pain, reduced blood loss and shorter hospital stays for patients, enabling spinal surgeons to offer fusion surgery to those who may otherwise be wary of it [6]. Recent studies [7], including our previous research [8], have shown that MIS techniques, such as minimally invasive transforaminal interbody fusion (MIS-TLIF), might offer comparable clinical results with traditional open TLIF, such as a shorter hospital stay, less blood loss and shorter recovery time. However, MIS-TLIF still requires an open incision, including partial paravertebral musculature separation, partial laminotomy and facetectomy to achieve a successful discectomy and place the cage.
The transforaminal endoscopic posterolateral approach is a well-known standard in endoscopic spine surgery that allows direct access to the disc with minimal invasive techniques. With the advancement of endoscopic technique, several percutaneous endoscopic fusion techniques have been reported for treating LDD [9,10]. Since 2017, we have explored the endoscopic transforaminal lumbar interbody fusion (Endo-TLIF), and tried to achieve improved recovery rates. This approach utilizes a combination of endoscopic visualization, discectomy, cages, percutaneous xation and osteobiologics. However, several studies reported that the endoscopic interbody fusion techniques had shown limitations, especially in the insertion of the rigid bulletshaped cage [10]. The standard cage is too large and rigid to pass through the conventional endoscopic working channel. Thus, a smaller but expandable cage was inserted into the intervertebral space [11]. In addition, some clinicians used more allogeneic bone to increase the amount of bone graft, which may reduce the rate of intervertebral fusion. Thus, bone morphogenetic protein (BMP) was combined with the application [10][11][12]. In this study, we used a novel C-shaped working cannula system, which not only allowed the insertion of a larger cage, but also allowed the channel to be inserted into the posterior margin of the intervertebral space. In this way, the nerve root is protected on the lateral side. BMP was not applied in this study, though allogeneic bone was the main bone grafting material in Endo-TLIF group. Furthermore, the use of cages without size reduction could be expected to reduce side effects such as the collapse of the endplate and increase the success rate of interbody fusion.
Typically, investigations of MIS assess outcomes from a variety of pathologies, such as lumbar spinal stenosis, lumbar instability syndrome and spondylolisthesis, and different diagnoses may lead to bias in assessing clinical outcomes. Therefore, we aimed to compare the perioperative characteristics and 1.5-years observational outcomes for the treatment of patients with one-segment spondylolisthesis using the Endo-TLIF or MIS-TLIF technique.

Patient Collection
Our clinical study proposal was approved by the medical ethics committee of Lianyungang Clinical College of Nanjing Medical University. Written informed consents were obtained from 102 consecutive patients prior to inclusion in the study. The inclusion criteria of this study included complaints of unilateral or bilateral radicular leg pain; single-level lumbar disc herniation or stenosis with one-grade or two-grade lumbar spondylolisthesis demonstrated by anteroposterior, lateral and exion extension plain radiographs, computed tomography (CT) scans and magnetic resonance imaging (MRI); and a lack of response to extensive conservative therapy for at least 3 months before surgery. Patients with osteoporosis, drug abuse, neoplasm, bone infection, or systemic diseases were excluded. Patients were assigned by a single-blind quasi-randomization within the spine department. Brie y, after the patients met the inclusion criteria and gave consent for the study, they were numbered serially at the spine department and were randomly assigned to MIS-TLIF or Endo-TLIF group. The surgeons were not informed of the group to which a patient was assigned until immediately before surgery. Finally, 102 consecutive patients with new onset spondylolisthesis were treated between March 2018 and April 2019, in which 48 patients received MIS-TLIF and 54 cases received Endo-TLIF.

Operative Techniques
All operations in Endo-TLIF group were performed under general anesthesia. With the patient placed in the prone position on a radiolucent operating table, percutaneous pedicle screws were placed using an anteroposterior uoroscopic technique with Jamshidi needles and an intervertebral spreader was applied (Fig. 1A). The endoscopic procedure was begun by accessing Kambin's triangle on the side. Successive dilation of the access site was performed to allow for the introduction of a 7.5-mmdiameter working channel (joimax). The traversing and exiting nerve roots were visualized with the working channel endoscope and decompressed directly by removing any compressive bone or cartilaginous tissues using electrocautery, pituitary rongeurs, microosteotomes and automatic power drills. The disc space was cleared, and the intervertebral spreader was opened (Fig. 1A).
Discectomy and endplate preparation for interbody fusion were performed after adequate transforaminal decompression. Epidural bleeding was controlled by radiofrequency probes. A 9-mm-diameter working channel was used and various kinds of reamer, curettes, endplate shavers and dissectors were used for complete preparation of the endplate under clear endoscopic view. Speci cally, the remaining cartilaginous endplate could be detached from the osseous endplate with the tongue of the channel under full endoscopic guidance without endplate injury (Fig. 1E). After fusion site preparation, removing the endoscope and absorbing endoscopic water ow, the intervertebral spreader was expanded and the working cannula was then replaced with a larger C-shaped working channel ( Fig. 1A and D) that protected the nerve root and facilitated the delivery and placement of the polyether-etherketone interbody cage (Capstone, Medtronic Sofamor Danek, USA) under uoroscopic guidance (Fig. 1A,B and C).
Meanwhile, the anterior disc space was lled with allograft bone chips (Fig. 1A). Finally, the intervertebral spreader was removed and percutaneous pedicle screws or facet screws were inserted and reset lumbar spondylolisthesis. Upon completing the instrumentation procedure, direct closure of the skin incision was performed (Fig. 1F) and the patient was monitored for complications.
The technique of our previous study for unilateral MIS-TLIF included bilateral screw xation and was performed on the more symptomatic side [8]. Brie y, uoroscopic guidance was used to determine the disc space and mark the lateral pedicle line and the lateral view was checked for tubular retractor system insertion. After a vertical skin incision (25 mm for the segment) at the lateral pedicle line, a tubular retractor (diameter: 22 mm, Medtronic Sofamor Danek, Memphis, TN) was introduced to the facet joint under uoroscopic guidance and unilateral total facetectomy and laminectomy were performed ( Fig. 2A). After a complete facetectomy, the ligamentum avum was removed to expose the lateral border of the ipsilateral exiting and traversing the nerve root.
Decompression of the central or contralateral stenosis was then performed. A discectomy was performed, and a polyetheretherketone interbody cage (Capstone, Medtronic Sofamor Danek, USA) lled with autologous local bone was inserted. Then, the percutaneous pedicle screw system (Sextant; Medtronic Sofamor Danek, USA) was placed through the minute incision under uoroscopic guidance ( Fig. 2B and C).

Clinical Outcome Assessment and Radiological Evaluation
Perioperative parameters such as operation time, blood loss, analgesic ratio, incision length, return to work time and rate, complications, postoperative hospitalization days and radiation exposure time were recorded during the operations. Postoperative complications and symptom recurrence requiring reoperation were assessed through review of medical record documentation and/or telephone interviews with patients. Visual Analog Scale (VAS) and Oswestry Disability Index (ODI) scores were evaluated for patients' symptoms at preoperation and postoperative days 1 and 3 and 18 months after surgery. In ammatory responses such as the concentration of white blood cells (WBC), c-reactive protein (CRP), tumor necrosis factor alpha (TNF-α) and erythrocyte sedimentation rate (ESR) of the enrolled patients were measured at preoperation and postoperative day 1 and 4. The in ammatory markers were tested by the clinical laboratory of our institution. Segmental lordosis was measured by Cobb's angle between the upper and lower endplates of the vertebrae that formed the operative segments. Fusion rates were determined by CT and ranked according to ve grades based on the anterior fusion criteria described by Brantigan (Fig. 3E) [13]. Two radiologists who were blinded to the patients' data performed all evaluations independently.

Statistical Analyses
Statistical analysis was performed using SPSS 20.0 software (SPSS Inc, Chicago, IL, USA). All data are presented as the mean ± S.D. (standard deviation). Unpaired t-test was used to compare the operation parameters between the two groups, and a paired ttest was used to compare preoperation and postoperation parameters in each group. The differences of the fusion rates among groups were compared using c 2 test for comparison. Statistically signi cant differences were considered when P values were less than 0.05.

Patients Characteristics
The analysis identi ed 102 patients meeting the study criteria, with a follow-up of 18 months. During the study, no patients within the series were excluded or lost to follow-up. There were no signi cant differences among the 2 groups regarding sample size, age, sex, symptom duration, operation level and body mass index (p > 0.05, Table 1).  Table 2). However, blood loss, left bed time, postoperative hospitalization days and analgesic ratio were signi cantly less in the Endo-TLIF group in comparison to the MIS-TLIF group during the hospitalization (p < 0.05, Table 2). In addition, there was one case of cage migration without any symptoms in the Endo-TLIF group and revisions were not performed. One super cial infection occurred in a single patient in the MIS-TLIF group, which was treated conservatively. The concentration level of CRP was lower in the Endo-TLIF group compared with the MIS-TLIF group, and the CRP levels rose gradually on days 1 and 4 after surgery (p < 0.05, Table 3). The ESR level also increased gradually after surgery, and the level in the Endo-TLIF group was signi cantly lower than that in the MIS-TLIF group on day 4 (p < 0.05, Table 3). There were no signi cant differences with respect to WBC and TNF-α between the 2 groups (p > 0.05, Table 3).

Comparison of VAS and ODI between Endo-TLIF and MIS-TLIF groups
The VAS was measured on a scale of 0 for no pain to 10 for the worst pain imaginable. The mean VAS for lower back pain and leg pain indicated that the 1.5 year postoperative pain levels in the two groups were both signi cantly lower than the preoperative levels and relief of pain was consistent throughout the entire observation time (p < 0.05, Table 4). Signi cant differences for low back pain were found between the two groups at postoperative day 1 and 3 months after surgery (p < 0.05, Table 4). The ODI demonstrated comparable results with the VAS, with patients experiencing consistent symptom reduction throughout the 18 months (p < 0.05, Table 4) and there was no signi cant difference between the two groups during the observation time (p > 0.05, Table 4).

Segmental lordosis and Fusion status of patients in Endo-TLIF and MIS-TLIF groups
The intragroup comparison was done according to operation methods. Segmental lordosis revealed signi cant changes 1.5 years postoperatively compared with the preoperative state and no differences were found between the two groups (Table 4). Fusion status at the review time was assessed by CT scan at 1.5 years postoperatively. According to the anterior fusion criteria described by Brantigan. grades 4 and 5 were considered fused [13]. Forty-six out of 48 patients (95.8 %) in MIS-TLIF group and 49 out of 54 patients (90.7%) in Endo-TLIF group (Fig. 3E) met the fusion criteria, respectively. And the fusion rate in Endo-TLIF (90.7%) was similar to that in the MIS-TLIF group (95.8 %) (p = 0.202).

Discussion
Over the past decade, minimally invasive lumbar interbody fusion has become very popular for treating a variety of lumbar spinal disorders. Our previous study [8] had indicated that MIS-TLIF allowed for decreased soft-tissue manipulation, which may have the bene ts of reducing blood loss and facilitating expeditious postoperative recovery. Meanwhile, it could lead to a reduction in nerve root traction while being able to address central and neural foraminal stenosis [14]. However, surgical techniques based on the Mast Quadrant still require articulectomy and partial paravertebral musculature separation. Recently, new minimally invasive spine surgeries have been applied to further reduce the trauma and enhance postoperative recovery. Speci cally, endoscopic spine and lumbar interbody fusion procedures minimize the skin incision and traumatization of posterior muscles and ligaments [15,16]. However, endoscopic fusion techniques remain a challenging and technically complex procedure associated with the transforaminal endoscopic approach [17]. In the present study, we tested TLIF using the endoscopic approach to achieve maximal preservation of normal musculo-ligamentous structures versus using tubular retractors.
Anatomically, the size of the intervertebral foramen is the main factor affecting the puncture and catheterization in the process of Endo-TLIF. Because the superior articular process of the lower vertebral body is directly opposite the upper intervertebral disc, it is di cult to accommodate the conventional 7.5mm diameter or bigger channels [18]. A recent study reported the rate of dysesthesia for Endo-TLIF was higher than for regular endoscopic procedures. This could be due to a bigger outer diameter of the Endo-TLIF cannula (14-16 mm) compared to regular endoscopic cannulas (7.5 mm) and strong vibrations when placing the cannula and inserting the expandable cage [19]. In addition, several studies showed that when performing the endoscopy-guided interbody cage placement, the smaller fusion cage had to be used because of the limitation that the conventional cage was too large to pass through the working channel. This may result in an increased risk of nonunion, delayed cage subsidence, or migration, especially in patients with severe osteoporosis [17][18][19]. Therefore, novel techniques to enlarge the intervertebral foramen, place the endoscopic channel in the intervertebral space smoothly and safely, obtain more effective decompression and implant a conventional cage are crucial.
In this study, we demonstrated how to directly enlarge the intervertebral foramen by improving the design of the intervertebral spreader while avoiding extensive damage to the surrounding tissue and superior facet (Fig. 1A). Meanwhile, the novel C-shaped working cannula system, which aims to place the cage into the intervertebral space safely and quickly, is able to achieve minimally invasive endoscopic decompression and protect the nerves ( Fig. 1A and D). Although the diameter of the channel is 12mm (Fig. 1D), its C-shaped construction, especially the forward structure occupies about half space of the circular passage, which can be safely placed into the intervertebral space. This greatly reduces the risk of nerve damage and tissue trauma. Fusion devices less than 11mm in width and less than 14mm in height can be implanted through this channel easily, although the height of a conventional interbody cage is usually no more than 13 mm. With a converter, we can still perform intraspinal processing and bone graft under this channel (Fig. 1D).
Our study showed that the Endo-TLIF group had comparative incision length, mean return to work time, and rate of return to work compared with MIS-TLIF group. However, Endo-TLIF reported less blood loss, decreased postoperative hospitalization days, left bed time and lower analgesic ratio compared with the MIS-TLIF group ( Table 2). The endoscopic spine approach minimizes the skin incision and traumatization of posterior muscles and ligaments such as multi dus muscles. Moreover, early studies show percutaneous endoscopic surgery can be performed under local anesthesia or epidural anesthesia [11,12]. In addition, the VAS and ODI in the MIS-TLIF and Endo-TLIF groups were both signi cantly lower than the preoperative levels, and the decrease was consistent throughout the entire observation time. The long-term clinical outcomes demonstrate durable and meaningful functional status improvements following these MIS procedures. The low back VAS in the Endo-TLIF group was signi cantly lower than in the MIS-TLIF group at day 1 and 3 months after surgery (Table 2). Thus, our results suggest that Endo-TLIF manifests more improvements in postoperative quicker recovery.
Several studies consider that hypertrophied facet joints and thickened transverse processes, especially high iliac crests (HIC) are considered a relative contraindication for transforaminal percutaneous disc decompression at L5-S1 [22]. While they provide a de nition and classi cation of HIC, the highest point of the iliac crest used in the classi cation is not actually the position of the puncture needle or working cannula contacting the ilium. Yang et al [23] studied one hundred patients and introduce a lumboiliac triangular technique based on biplane oblique uoroscopy which facilitated transforaminal percutaneous disc decompression for patients with L5-S1 lumbar disc herniation (LDH) combined with HIC. Thus, this technique has been used to complete some Endo-TLIF with L5 spondylolisthesis combined with HIC in our study.
Previous studies have documented the harmful effects of extensive paraspinal muscular dissection and retraction lumbar surgery [8]. Indeed, retractor blades have been shown to increase intramuscular pressure that can ultimately lead to postoperative LBP and ischemia. We took in ammatory markers and white blood cells (WBC) as the observation indices. The CRP level rose gradually over the rst and fourth day after surgery and was lower in the Endo-TLIF group compared with the MIS-TLIF group after surgery. The ESR level also rose after surgery, and the level in Endo-TLIF group was signi cantly lower than that in MIS-TLIF group on day 4 (Table 3). Thus, all of these results indicate that Endo-TLIF technology has a more positive in uence in improving postoperative LBP and analgesic ratio compared with MIS-TLIF. However, the operative time and radiation exposure time in the Endo-TLIF group was signi cantly longer than that in the MIS-TLIF group and suggests the limitation of Endo-TLIF lies in the long learning curve required to ensure each step of the procedure is safe and effective. Several steps of endoscopic TLIF require uoroscopic guidance and it goes without question that methods for reducing radiation exposure in the setting of Endo-TLIF will have signi cant long-term health bene ts.
Endo-TLIF reached a similar fusion rate to that in MIS-TLIF (90.7% vs 95.8%) in our study. This indicates that Endo-TLIF could have a good fusion rate. Before the experiment, we were concerned about the possibility of low fusion rate due to more allografts and very little amount of autografts were implanted in the Endo-TLIF group and the osteogenic capacity of allograft may weaker than autologous bone in the MIS-TLIF group [24]. Studies have shown high fusion rates were appeared in MIS-TLIF, such as 88.9% at postoperative 1 year and 96.0% at postoperative 3 years [8, 25]. Dong Hwa Heo[26] presented the fusion rate of the Endo-TLIF group was 73.9% at postoperative 1 year. Thus, the similar fusion rate between two groups indicating that BMP may not be necessary in Endo-TLIF. Theoretically, trauma of paraspinal soft tissue in open posterior procedures has an adverse effect on adjacent segments and reduces the blood supply for fusion [27]. Our result was in accord with the theory and minimal muscle injury in Endo-TLIF is conducive to an improvement in fusion rate. Meanwhile, the conventional large cage with large amount of allografts was allowed to pass through the novel C-shaped working cannula system. This may also help to improve the fusion rate.
Limitations of this study include a small sample size and limited follow-up period. Randomized controlled trials with long-term follow-up and a larger number of patients are needed to obtain more accurate clinical results.

Conclusions
Endo-TLIF surgery technique seems to be associated with less multi dus muscle damage, in ammation and faster patient recovery compared with MIS-TLIF group. Meanwhile both two technologies can be used to effectively treat 1-segment lumbar spondylolisthesis with similar fusion rate. Intraoperative and postoperative view for endoscopic TLIF underwent bilateral screw xation. A:The C-shaped working channel was introduced to the intervertebral space under full-endoscopic guidance after the pedicle screws were inserted and the homemade intervertebral spreader was applied. B: Intraoperative C-arm lm in anteroposterior view: the percutaneous pedicle screw system was inserted accurately and the interbody fusion cage was implanted under uoroscopic guidance. C: Intraoperative C-arm lm in lateral view: the percutaneous pedicle screw system and cage were inserted accurately. D: The C-shaped working channel and a converter. E: Endoscopic view of fully decompressed roots, the vertebral endplate after preparation and the fusion cage after implantation. F:Postoperative view of minute incision, the bar represents 10mm.

Figure 2
Intraoperative and postoperative view for unilateral MIS-TLIF underwent bilateral screw xation. A:The tubular retractor was introduced to the facet joint under uoroscopic guidance after the pedicle screws were inserted. B:Intraoperative C-arm lm in anteroposterior view. The percutaneous pedicle screw system was inserted accurately. C:Intraoperative C-arm lm in lateral view. The percutaneous pedicle screw system was inserted accurately. D:Postoperative view of minute incision, bar = 10 mm.