Interbody fusion is a well-established treatment for degenerative disorders of the lumbar spine. After discectomy, access to the intervertebral disc space for graft placement may be achieved through an anterior, posterior, or transforaminal approach, each with its respective risks and benefits. The anterior approach restores lordosis with a wide footprint cage but risks vascular injury and damage to the abdominal viscera, sympathetic trunk, and superior hypogastric plexus, leading to retrograde ejaculation and male infertility.^1,2 Both the posterior and transforaminal lumbar approaches risk neural injury, durotomy, and trauma to the paraspinal muscles.^2,3

The minimally invasive lateral transpsoas approach—also known as the extreme lateral interbody fusion (XLIF, ELIF; NuVasive, Inc.), direct lateral interbody fusion (DLIF; Medtronic), or generally as the lateral lumbar interbody fusion (LLIF)—has become increasingly used to avoid such complications.⁴ Since its first published description in 2006, this approach has become an alternative to anterior lumbar interbody fusion (ALIF) for the lumbar segments above L5–S1, providing the wide interbody footprint without the high risks of bowel and vascular injury related to the anterior approach.⁴ Moreover, LLIF is characterized by a classically small incision, minimal blood loss, decreased postoperative pain, and short hospital stay.^5–7 While the lateral approach is generally well tolerated, debilitating injuries to the lumbar plexus, iliac vessels, and ureter are known risks, likely related to the lack of direct visualization of these nearby structures.^7–9 Traditionally, localizing the lumbar plexus is done indirectly and relies on intraoperative directional electromyography (EMG), fluoroscopy, and palpation of the retroperitoneal anatomy.⁵ Femoral, genitofemoral, ilioinguinal, iliohypogastric, and subcostal nerve palsies are especially debilitating and lead to unacceptable postoperative side effects, namely, quadriceps weakness, groin pain and anterior thigh numbness, genital neuropathic pain, and abdominal pseudohernia.^8,9 A meta-analysis of 18 studies including 2310 patients showed that 13.1% of patients experienced a lumbar plexus–related injury in a traditional LLIF approach.⁸ Even in experienced hands and stable neuromonitoring consisting of directional EMG, somatosensory evoked potentials, and motor evoked potentials, clinically significant nerve injury can still develop, leading to debilitating iliopsoas weakness, along with permanent groin and thigh pain.^9,10 Furthermore, new approach-related complications continue to be reported, such as pneumomediastinum, pneumopericardium, and delayed incisional hernia.^11–13 Direct visualization during LLIF could lead to decreased rates of these approach-related complications.^14–16

Given the risk profile of the traditional transpsoas approach, we have developed and investigated the efficacy of performing the technique with endoscopic assistance for direct visualization of the retroperitoneal space. With the assistance of general surgery, a safe transpsoas corridor is identified along with the lumbosacral plexus, iliac vessels, and ureter. Clearance along the psoas is given after medial rotation of the viscera to avoid bowel or vascular injury, reducing the chance of inadvertent medial placement. Here, we present a series of patients who underwent this procedure and discuss their outcomes.

Methods

Seven consecutive LLIF procedures performed by the senior authors (D.C.L. and D.C.C.) from 2013 to 2020 were reviewed. Based on surgeon preference and experience, it is customary practice to perform the LLIF procedure with endoscopic assistance. The same minimally invasive general surgeon (D.C.C.) and spine neurosurgeon (D.C.L.) performed each case together. Electronic health records were used to obtain information regarding patient demographics, preoperative symptoms and diagnoses, medical comorbidities, operative notes, and postoperative courses. The patients’ pain and function were evaluated by patient-reported outcome measures, including the visual analog scale (VAS), and by the Oswestry Disability Index (ODI). Preoperative VAS and ODI scores were compared with the patient’s most recent postoperative ODI and VAS. The data were analyzed using a paired t-test (IBM SPSS, version 26, IBM Corp.). A p value < 0.05 was considered statistically significant.

Endoscopic Approach

After undergoing general endotracheal anesthesia and administration of prophylactic antibiotics, the patient is secured in the lateral decubitus position with the aid of a bean bag and an axillary roll. Bladder decompression is achieved with placement of a urinary catheter. The table is gently flexed to open the operative field between the costal margin and the iliac crest. An approximately 2-cm transverse incision is marked under fluoroscopic guidance over the disc space approximately 3 cm superior to the iliac crest in the midaxillary line. Once the skin and external oblique fascia are incised, the external and internal oblique and transversalis muscles are spread until the retroperitoneum is encountered. The retroperitoneal potential space is dissected using the Spacemaker balloon dissector (Covidien), which is inflated under direct endoscopic visualization (Video 1).

VIDEO 1. Operative video of endoscope-assisted LLIF with narrated audio. Copyright Irene Say. Published with permission. Click here to view.

Once the space is opened, the balloon is deflated and replaced with a 12-mm balloon trocar. The space is insufflated with dry carbon dioxide gas to a pressure of 15 mm Hg. One 5-mm port is placed medial under direct visualization medial to the endoscope. The retroperitoneal fat pad is dissected from the psoas using an endoscopic Kittner dissector (Medline Industries, Inc.) and LigaSure (Medtronic), a bipolar vessel sealing technology. The retroperitoneal contents are then rotated medially, allowing direct identification of the genitofemoral nerve along the psoas, the ureter, and iliac and gonadal vessels (Fig. 1).

Intraoperative endoscopic photograph of retroperitoneal anatomy. The ureter, iliac artery, and genitofemoral nerve (GFN) along the psoas muscle are labeled. Figure is available in color online only.

Instrumentation and Fusion

Under direct endoscopic visualization, the initial dilator is placed through the psoas muscle posterior to the genitofemoral nerve and overlying the desired disc space (Fig. 2). Fluoroscopy is used to confirm positioning at the correct vertebral level. The psoas is serially dilated under direct endoscopic visualization to protect the lumbar plexus, ureter, and iliac vessels (Fig. 3). Directional EMG monitoring confirms proximity to the nerves of the lumbar plexus. After docking of the retractor, the microscope is brought in to visualize the disc space. The disc is entered with a disc knife, and discectomy is performed in a standard fashion, using the pituitary rongeur, Kerrison rongeurs, and the box cutter. After endplate preparation, the interbody cage (Timberline MPF, Zimmer Biomet, or Regatta, SeaSpine Inc.) with allograft is placed into the disc space using fluoroscopic localization. Figure 4 demonstrates lateral screw and interbody spacer positioning at L2–3 in an illustrative radiograph.

Photograph of operating room setup demonstrating endoscope-assisted placement of the initial dilator through the psoas and K-wire through the intervertebral disc. Figure is available in color online only.

Endoscopic photograph showing initial tubular dilation through the psoas muscle, with the identified genitofemoral nerve (GFN) at the L3–4 level, and anterior/posterior, rostral/caudal orientation marked. Figure is available in color online only.

Standing anteroposterior lumbar spine radiograph obtained in a patient 9 months postoperatively, demonstrating proper placement of the lateral plate and screws at L2–3.

Results

Seven consecutive patients with endoscopic LLIF were reviewed for clinical feasibility. Three patients were male and 4 were female. The average age was 64.1 years (range 58–78 years). All patients presented with back pain and radiculopathy, which was bilateral in one patient and unilateral in the remaining patients. The indications for operation and levels fused are shown in Table 1. There were no intraoperative complications, and the mean postoperative length of stay was 3.14 ± 1.55 days (± SD). The mean operative time per level was 105 ± 25 minutes. The average postoperative follow-up time was 23.1 ± 14.2 months. Patient-reported outcome measures, including ODI and VAS scores, improved in all patients and are described in Table 1. The mean preoperative ODI score was 39.5% ± 15.2%, and the mean postoperative ODI score was 17.8% ± 9.6%, with a statistically significant decrease in the ODI score by 21.7 (p = 0.006). The mean preoperative VAS score was 6.5 ± 2.0, and the mean postoperative VAS score was 0.6 ± 0.5, with a statistically significant decrease in VAS score by 6.0 (p < 0.001). No patient experienced injury to the viscera, aorta or iliac vessels, or ureter. There were no permanent neurological complications.

Patient 1 developed mild, transient ipsilateral lateral thigh numbness postoperatively, which spontaneously resolved by the 3-month follow-up. Patient 2 developed moderate, transient ipsilateral psoas weakness on postoperative day 1, which also spontaneously resolved by the 3-month follow-up.

Patient 4 had an uncomplicated immediate postoperative course. However, on a surveillance radiograph obtained 6 months postoperatively, the lateral plate backed out of the lateral screws. The patient was taken back to surgery for hardware failure and pseudarthrosis, where endoscope-assisted lateral revision instrumentation was performed uneventfully.

All patients endorsed improvement in their symptoms at their most recent follow-up (average follow-up time 23 months). Blood loss was minimal in all cases.

Discussion

The lateral transpsoas approach to the spine has evolved as an invaluable alternative to the traditional anterior lumbar approach, with reduced risks of vascular and bowel injury.^17,18 However, the lateral approach for interbody fusion is not without morbidity or even mortality.¹⁹ Clinically significant complication rates range between 6% and 52% for LLIF,^6,15 aside from the transient, known mild hip flexion weakness or anterior thigh pain related to dilating and retracting within the psoas muscle fibers.²⁰ New and potentially fatal complications associated with the lateral approach, such as bowel perforation, pneumomediastinum, pneumopericardium, and contralateral psoas hematoma, continue to be reported.^11,21,22 Lumbar plexopathy is a known complication of the direct lateral approach that surgeons have attempted to minimize using anatomical landmarks, neuromonitoring, and fluoroscopy.²³ The original LLIF description involves blindly splitting the psoas muscle fibers with serial dilator tubes, posing stretch injury to the nearby nerves with retraction or compromising the adjacent iliac vein and ureter.⁴ By contrast, our technique employs direct visualization offered by the endoscope and ensures that the instruments are passing through a safe corridor. We investigated the approach-related morbidity and efficacy of the endoscope-assisted LLIF. To our knowledge, this technique has not previously been described in detail or directly compared with nonendoscopic lateral approaches.

We hypothesized that the endoscope-assisted lateral approach to the spine would be a clinically feasible technique to reduce visceral, vascular, and lumbar plexus injury because it allows for mobilization of the viscera off the psoas muscle and affords excellent visualization of the iliac vessels and lumbar plexus. An endoscopic retroperitoneal approach was first described by McAfee et al.²⁴ Although this technique avoided complications associated with the transperitoneal approach, such as retrograde ejaculation and great vessel injury, it required posterior retraction of the psoas that frequently resulted in muscle weakness and swelling.²⁴ Bergey et al. avoided this by using a more posterior trajectory through the psoas muscle but encountered a 30% incidence of transient genitofemoral nerve palsy.²⁵ Recently, Schonauer et al. described an endoscope-assisted lateral interbody fusion in the immediate operative field of the psoas but did not apply the retroperitoneal endoscopic techniques of gas and balloon insufflation to rotate the visceral contents away from the psoas prior to docking, as provided by our access general surgeon (D.C.C.)²⁶ In contrast, our technique directly identifies the genitofemoral nerve, iliac vessels, and ureter, visualizing and ensuring the integrity of these structures throughout every case.

Patient 1 in our case series experienced anterior lateral thigh numbness in the pattern of a lateral femoral cutaneous nerve palsy, likely due to transpsoas dilation at the L2–3 level. At this level, the lateral femoral cutaneous and genitofemoral nerves were visualized, and direct trauma was avoided. Temporary postoperative ipsilateral anterior thigh numbness and pain have been reported to be as high as 60% and 43%, respectively, in the literature, due to nerve stretch injury during retractor placement in the psoas.²⁷ While safe placement is ensured with direct visualization, the risk of stretch injury to either the genitofemoral or lateral femoral cutaneous nerve from retraction is not eliminated.^27,28 The genitofemoral nerve typically overlies the anterior to middle third of the psoas, but the anatomical variability of the lumbar plexus makes it difficult to consistently identify a safe working zone.^28–30 Furthermore, Banagan et al. found wide anatomical variation of the lumbar plexus in relationship to the psoas muscle and concluded that there are no safe zones for psoas dilation.²⁸ This underscores the importance of direct visualization to minimize the risk to each patient and his or her individual anatomy.

Patient 2 developed transient hip flexion weakness that spontaneously resolved by the 3-month follow-up. His surgery had been performed at L4–5, which is associated with a higher incidence of hip flexion weakness compared with other lumbar segments, as suggested by a cadaveric study of lumbar plexus anatomy over L4–5.^28–30 Of note, this patient was morbidly obese, with a BMI of 37.1 and a high-riding iliac crest, requiring an increased operative time of 160 minutes. Increased retraction time in the psoas likely contributed to this transient hip flexion weakness, which is a similar consideration with the traditional transpsoas lateral approach.^31–34

Patient 4 required a revision endoscope-assisted surgery for pseudarthrosis and hardware failure of the lateral plate and did not experience any approach-related complications. It is unlikely that the hardware failure was a consequence of endoscopic assistance.

No patients in our study experienced permanent neurological, urological, visceral, or vascular injury. While these serious complications are historically rare even with the traditional percutaneous approach, there is a growing number of reported complications, particularly in cases of deformity.³⁵ This is likely because patients with rotational scoliosis may have aberrant anatomy and be more vulnerable to injury without direct visualization of these structures during surgery. Because of the intrinsically minimally invasive nature of this procedure, recognizing and reversing any damage is challenging. These serious complications were avoided in our series precisely because of the endoscopic assistance. With endoscopy, rotating the viscera medially prior to psoas dilation and visualizing the iliac vasculature cleared the psoas as a safe working corridor. As the endoscope utilized the existing incision and a small (4-mm) anterior incision as the manipulation port, this approach compares favorably in terms of the number of incisions to the traditional nonvisualized LLIF approach, which often is performed with two incisions, with the second incision being used to release peritoneal adhesions and guide dilation.³⁶

The average length of hospitalization was 3.1 days, which is similar to published series of patients undergoing open and endoscopic transpsoas approaches.^25,33,37 The average operative duration was 105 minutes per level, which is also similar to reported durations for the mini-open and endoscopic approaches.^25,33,37 While our technique requires an approach surgeon, direct visualization of the retroperitoneum may be particularly advantageous for patients with difficult anatomy. Given its unfamiliarity to most spine surgeons, this approach may be most useful for patients with previous retroperitoneal, specifically kidney, or revision lateral surgery or surgery performed in the obese patient. Although limited to seven cases, our preliminary outcomes with endoscopic LLIF suggest its clinical feasibility to meaningfully improve patient outcomes for the LLIF, with statistically significant reductions in ODI and VAS scoring. Only one patient had a transient episode of spontaneously resolving psoas weakness, and only one patient had a transient episode of anterior thigh numbness.

Conclusions

Our preliminary experience with endoscopic LLIF suggests that it is a clinically feasible technique, providing excellent visualization of the lumbar plexus and iliac vessels. It is generally well tolerated and maintains the benefits of minimally invasive spine surgery, including a small incision, minimal blood loss, and rapid perioperative recovery. Further investigation is needed to determine its overall safety and cost-effectiveness.

Acknowledgments

This work was made possible through the generous support of J. Yang & Family Foundation. D.C.L. is a 1999 P.D. Soros Fellow. The views expressed herein are those of the authors and do not necessarily represent or reflect the views of funding agencies. J.A.T. had funding from an institutional NIH R25 grant during completion of this paper. We would like to thank Naomi Gonzalez for logistical support.

Disclosures

The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

Author Contributions

Conception and design: Lu, Niu, Archie, Chen. Acquisition of data: Lu, Say, Niu, Thum. Analysis and interpretation of data: Lu, Say, Niu, Thum, Chen. Drafting the article: Lu, Say, Niu, Archie, Chen. Critically revising the article: Lu, Say, Niu, Chen. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Lu. Statistical analysis: Say. Administrative/technical/material support: Say, Thum. Study supervision: Lu, Say, Archie, Chen.

Supplemental Information