Unilateral biportal endoscopy versus percutaneous endoscopic lumbar discectomy for far lateral disc herniation: a comparative study of clinical efficacy and learning curves Authors: Yazhou Li, Haoran Gao, Honggang Zhou, Xiao Sun, Yupeng Wang, Xiaohu Wang, Mingkui Shen, Haijun Ma Published: 2026 ABSTRACT AbstractPurposeTo compare the clinical outcomes and learning curves of unilateral biportal endoscopy (UBE) and percutaneous endoscopic lumbar discectomy (PELD) in the treatment of far lateral lumbar disc herniation (FLLDH).MethodsA retrospective analysis was conducted on 122 patients with FLLDH who underwent spinal endoscopic surgery between January 2022 and July 2024, including 62 cases in the PELD group and 60 cases in the UBE group. All procedures were performed by the same minimally invasive spine surgery team. Perioperative parameters (operative time, intraoperative blood loss, foraminal area, postoperative hospital stay, total cost), visual analogue scale (VAS) scores, Oswestry Disability Index (ODI), and modified Macnab criteria were compared. Learning curves were analyzed using cumulative sum (CUSUM) with operative time as the core variable. All patients were followed for at least 12 months.ResultsBoth groups achieved significant improvement in pain and functional outcomes after surgery. Operative time was significantly longer in the UBE group than in the PELD group (132.77 ± 27.55 min vs. 86.61 ± 24.52 min), and total blood loss was also higher (270.43 ± 119.44 ml vs. 37.32 ± 13.69 ml). The PELD group showed lower postoperative day-1 back pain scores, whereas long-term VAS, ODI, and modified Macnab outcomes were comparable between groups. The excellent/good rates were 91.9% in the PELD group and 93.3% in the UBE group. CUSUM analysis demonstrated learning curve inflection points at 24 cases for UBE and 38 cases for PELD.ConclusionBoth UBE and PELD are safe and effective for treating FLLDH. PELD offers advantages in perioperative minimally invasive characteristics and cost-efficacy, whereas UBE features a shorter learning curve, facilitating more rapid surgical mastery.Keywords: Far lateral lumbar disc herniation, Unilateral biportal endoscopy, Percutaneous endoscopic lumbar discectomy, Learning curve, Minimally invasive spine surgery Introduction Far lateral lumbar disc herniation (FLLDH) refers to disc herniation located in the extraforaminal zone or lateral intraforaminal zone, directly compressing the exiting nerve root or dorsal root ganglion. It accounts for 6.5%–12% of all lumbar disc herniations, with the highest incidence at L4–5 and L5–S1 [1]. Due to the deep anatomical location adjacent to the psoas major muscle, iliolumbar vessels, and the unique trajectory of the exiting nerve root, FLLDH often presents with severe radicular pain and may involve sensory deficits or motor weakness [2, 3]. Conservative treatment is significantly less effective compared with central or paracentral herniations [4, 5]. Traditional open surgery provides direct exposure but is associated with extensive muscle dissection, iatrogenic instability, and prolonged recovery [6]. With the development of minimally invasive spine surgery, unilateral biportal endoscopy (UBE) and percutaneous endoscopic lumbar discectomy (PELD) have become mainstream approaches for FLLDH [7]. PELD offers advantages of percutaneous access and reduced soft tissue trauma [8, 9], whereas UBE provides a panoramic operative field and bimanual instrument handling through its biportal configuration [10]. Although existing clinical studies have confirmed the safety and efficacy of both techniques, there remains controversy regarding their efficacy advantages—some studies suggest that PELD provides more significant early postoperative pain relief and shorter hospital stay [11–13], while UBE may have potential in reducing intraoperative fluoroscopy times and improving dural sac volume [14]. This study retrospectively compares the perioperative outcomes, clinical efficacy, safety, and learning curves of UBE and PELD in the treatment of FLLDH. Materials and methods Sample inclusion A retrospective review was conducted on 122 patients who underwent spinal endoscopic surgery for FLLDH between January 2022 and July 2024. Patients were divided into the PELD group (n = 62) and the UBE group (n = 60) according to surgical method. Inclusion criteria ① Clinical presentation of low back pain or radiating leg pain, with a confirmed diagnosis of FLLDH based on lumbar CT and MRI. ② Failure of standard conservative treatment for 3 months. ③ Underwent UBE or PELD performed by the same minimally invasive spine team. ④ Complete perioperative and imaging data with at least 12 months follow‑up. Exclusion criteria ① Lumbar spondylolisthesis on imaging. ② History of lumbar surgery or spinal trauma. ③ Severe cardiopulmonary, hepatic, renal dysfunction or psychiatric/cognitive disorders affecting follow-up. ④ Coagulopathy or active infection. ⑤ Requirement for repeated PELD or UBE procedures. Surgical techniques UBE procedure Under general anaesthesia, patients were positioned prone. The target level was localized using C-arm fluoroscopy. A vertical line was drawn 2 cm lateral to the pedicle’s lateral border, and a horizontal line was drawn at the SAP tip; two skin entry points were marked 2 cm above and 1 cm below their intersection. After sterilization and draping, spinal needles were inserted under fluoroscopic guidance, and serial dilators were used to establish an endoscopic viewing portal and an independent working portal (Fig. 1) a, b: Preoperative MRI and CT demonstrated a right-sided far-lateral disc herniation at the L4/5 level; c: Preoperative CT showing the cross-sectional area of the intervertebral foramen; d, e: Intraoperative positioning and establishment of the endoscopic viewing and working portals were performed; f, g: The exiting nerve root before and after decompression; h: Postoperative three-dimensional CT reconstruction; i: Postoperative CT showing the cross-sectional area of the intervertebral foramen; j, k, l: Postoperative MRI and CT indicated that the right lateral nerve root compression was relieved A 30° arthroscope was introduced through the viewing portal. Soft tissues were dissected via the working channel, and hemostasis was achieved with a plasma radiofrequency electrode( Model: MC301A, Product: Spine-o-UBE RadioFrequency Plasma Surgical Electrodes, Manufacturer: Jiangsu BONSS Medical Technology Co., Ltd.). A high-speed burr(Model: DGB40UA110, Product: UBE High-speed Drill Burs, Manufacturer: Jiangsu BONSS Medical Technology Co., Ltd.) was used to remove the lateral lamina, part of the SAP tip, and a portion of the transverse process to enlarge the operative corridor. Following resection of the ligamentum flavum, the exiting nerve root and dural sac were visualized. After protecting the nerve root with a nerve retractor, the herniated disc fragment was excised using pituitary forceps. Intradiscal decompression and radiofrequency coagulation were performed. Adequate root relaxation and dural pulsation were verified. A drain was placed, and the incisions were closed (Fig. 2). a, b, c: Preoperative MRI and CT demonstrated a left-sided far-lateral disc herniation at the L5/S1 level; d: Preoperative CT showing the cross-sectional area of the intervertebral foramen; e, f: Intraoperative positioning and establishment of the endoscopic viewing and working portals were performed, ① and ② indicate the expected target points of the portals. The working channel has reached the approximate region for initial soft-tissue dissection, and the dashed line indicates the planned trajectory of the burr.; g: Remove the protruding nucleus pulposus compressing the nerve root; h: Postoperative three-dimensional CT reconstruction; i: Postoperative CT showing the cross-sectional area of the intervertebral foramen; j, k, l: Postoperative MRI and CT indicated that the left lateral nerve root compression was relieved At the end of each procedure, adequate neural decompression was confirmed intraoperatively based on predefined criteria, including visualization of a freely mobile exiting nerve root, restoration of dural pulsation, and absence of residual compression under endoscopic inspection. PELD procedure Patients were placed in the lateral decubitus position with a soft cushion under the waist to optimize foraminal access. Under C‑arm fluoroscopic guidance, and the target disc level was localized using standard bony landmarks. The puncture point was 10 cm lateral to the midline of the spinous process. After routine sterilization and draping, local anaesthesia was administered with 1% lidocaine. An 18‑gauge spinal needle was introduced at an angle of approximately 30° to the sagittal plane, passing through the skin, subcutaneous tissue, quadratus lumborum, and psoas muscle, advancing toward the ventral aspect of the superior articular process (SAP). A guidewire was then inserted, and sequential dilation was performed using 3.5‑mm, 4.5‑mm, and 5.5‑mm dilators, followed by placement of a 7.5‑mm working cannula. After confirming appropriate cannula placement under fluoroscopy, a spinal endoscope was inserted. Soft tissues surrounding the SAP were cleared, progressively exposing the tip of the SAP, the SAP–transverse process junction, and the target disc. A ring‑type osteotome was used to remove the tip and middle portion of the SAP. A high‑speed diamond burr further enlarged the foramen. The caudal margin of the ligamentum flavum was exposed and removed to visualize the exiting nerve root and the displaced disc fragment. Bipolar radiofrequency was used for hemostasis and adhesiolysis. Herniated disc fragments were removed, followed by intradiscal decompression and radiofrequency annuloplasty. After confirming adequate nerve root mobility and absence of residual compression, the cannula was withdrawn and the incision was sutured. Observation indicators Perioperative Parameters: Operative time, fluoroscopy frequency, preoperative and postoperative foraminal area, postoperative hospital stay, total hospitalization cost, hemoglobin (Hb), hematocrit (Hct), estimated blood volume (EBV) [15] total blood loss (TBL), intraoperative blood loss (IBL), hidden blood loss (HBL), and perioperative complications were recorded. TBL was calculated using the Gross formula [16]. The Gross formula for estimating TBL is as follows: . Hidden blood loss (HBL) was calculated as HBL = TBL – IBL. The cross-sectional area of the intervertebral foramen (CSA-IVF) was measured on sagittal reconstructed CT images at the level of maximal foraminal narrowing. Bony boundaries were manually delineated using the hospital imaging software, with the enclosed area calculated automatically. Measurements were performed independently by two experienced spine surgeons blinded to clinical outcomes, and the mean value was used for analysis. Preoperative and postoperative CSA-IVF values, as well as the absolute change (postoperative minus preoperative), were recorded and compared between groups. Clinical outcomes included VAS scores for back and leg pain, ODI, and modified Macnab criteria. VAS and ODI were assessed preoperatively and at postoperative day 1, 3 months, and 12 months. Modified Macnab criteria were evaluated at final follow-up. Complications, including nerve injury, dural tear, wound hematoma, and disc space infection, were recorded throughout the perioperative period and at each follow-up visit. Learning curve evaluation Cumulative sum (CUSUM) analysis was employed to evaluate the learning curves of unilateral biportal endoscopy (UBE) and percutaneous endoscopic lumbar discectomy (PELD), with operative time selected as the primary performance indicator. Patients in the UBE and PELD groups were consecutively numbered from 1 to N₁ and 1 to N₂, respectively, according to the chronological order of surgery. The operative times for individual cases were denoted as T₁, T₂, …, Tₙ, and the mean operative time for each group was defined as T_mean. The initial CUSUM value (CUSUM₁) was calculated as the difference between the operative time of the first case and the group mean (CUSUM₁ = T₁−T_mean). For subsequent cases, CUSUM values were calculated recursively according to the formula CUSUMₙ = (Tₙ−T_mean) + CUSUMₙ₋₁, with calculations continued until the final case. Theoretically, CUSUM values tend to approach zero as operative performance stabilizes [17]. To minimise the potential influence of case complexity on operative time, all patients were enrolled consecutively, and no intentional case selection or staging was performed throughout the study period. Learning curves were visualized using scatter plots, and CUSUM data were fitted to regression models using PyCharm 2025.2.4-based statistical analysis. The goodness of fit was evaluated using the coefficient of determination (R²), with values closer to 1 indicating a superior fit. Based on these results, the optimal model was selected for each technique. The learning curve was subsequently divided into a learning phase and a proficiency phase, defined by the first sustained downward inflection point of the CUSUM curve, indicating that operative time consistently fell below the group mean. This transition point was considered to represent the minimum number of cases required to achieve procedural proficiency. All procedures were performed by the same minimally invasive spine surgery team, and the CUSUM analysis reflects the collective learning process of the team rather than an individual surgeon. Statistical analysis Data analyses were performed using IBM SPSS Statistics 27.0, and all continuous data were tested for normality and are presented as mean ± standard deviation (x̄±s). Normality was assessed using the Shapiro–Wilk test. Depending on distribution, independent-samples t-tests, variance-adjusted t-tests, or Mann–Whitney U tests were used. Categorical variables were compared using χ² or Fisher’s exact test. A two-sided P < 0.05 was considered statistically significant. Results Baseline Data Baseline demographic and clinical characteristics were well balanced between the two groups, with no significant differences in age, sex distribution, BMI, operated level, comorbidities, or operative side.(Table 1). Table 1Comparison of baseline characteristics between the two groupsPELDUBE P Age(years)62.84 ± 7.4160.67 ± 7.270.105Gender(M/F)37: 2533: 270.601BMI(m/Kg² )26.74 ± 2.8127.24 ± 2.310.284Disc level (n, %)0.737L3/45(8%)3(5%)L4/525(40%)23(38%)L5/S132(52%)34(57%)Comorbidities, n (%)32(52%)27(45%)0.465Hypertension2321Diabetes1413Coronary heart disease22cerebral infarction20Operation side, n (%)0.601Left34(55%)36(60%)Right28(45%)24(40%) Comparison of baseline characteristics between the two groups Perioperative conditions Table 2 compares the surgical-related parameters, imaging indicators, and hematological indices between the UBE and PELD groups. In terms of surgical-related parameters, significant differences were found between the two groups: the UBE group had a significantly longer operative time (132.77 ± 27.55 min) compared to the PELD group (86.61 ± 24.52 min) (P < 0.001). However, the number of intraoperative fluoroscopic exposures was similar between the groups (5.45 ± 1.68 for the PELD group, 4.97 ± 1.5 for the UBE group) with no significant difference (P > 0.05). Table 2Comparison of perioperative outcomes between the two groupsPELDUBE P Operation time (min)86.61 ± 24.52132.77 ± 27.55< 0.001Number of fluoroscopy5.45 ± 1.684.97 ± 1.50.096CSA-IVF (mm² )Preoperative106.16 ± 17.52108.03 ± 18.080.562Postoperative150.58 ± 32.31154.50 ± 28.830.482Preoperative Hb (g/L)127.08 ± 12.30126.86 ± 10.390.915Postoperative Hb (g/L)123.85 ± 10.62110.29 ± 9.91< 0.001Preoperative Hct (%)38.23 ± 2.3037.87 ± 2.060.358Postoperative Hct (%)37.02 ± 2.0933.88 ± 1.84< 0.001TBL (ml)37.32 ± 13.69270.43 ± 119.44< 0.001IBL (ml)13.17 ± 4.0695.20 ± 33.99< 0.001HBL (ml)24.15 ± 12.34175.67 ± 83.47< 0.001Postoperative Hospital Stay (day)3.98 ± 1.594.48 ± 1.610.088Total hospitalization costs (¥)20863.98 ± 3802.2437950.36 ± 3405.69< 0.001 Comparison of perioperative outcomes between the two groups Regarding imaging evaluations, preoperative foraminal area (PELD group: 106.16 ± 17.52 mm2, UBE group: 108.03 ± 18.08 mm2) and postoperative foraminal area (PELD group: 150.58 ± 32.31 mm2, UBE group: 154.50 ± 28.83 mm2) showed no significant differences between the groups (P > 0.05), suggesting that both techniques were similarly effective in improving foraminal volume. Among perioperative indicators, the postoperative hospital stay was 3.98 ± 1.59 days in the PELD group and 4.48 ± 1.61 days in the UBE group, with no statistically significant difference (P > 0.05). However, the total surgical cost was significantly higher in the UBE group (37950.36 ± 3405.69 ¥) than in the PELD group (20863.98 ± 3802.24 ¥), with a statistically significant difference (P < 0.001). In terms of blood loss and hematological indices, preoperative hemoglobin (Hb) and hematocrit (Hct) levels showed no significant difference between the two groups (P > 0.05). However, postoperative Hb (UBE group: 110.29 ± 9.91 g/L, PELD group: 123.85 ± 10.62 g/L) and Hct (UBE group: 33.88 ± 1.84%, PELD group: 37.02 ± 2.09%) were significantly lower in the UBE group (P < 0.001). Total blood loss (TBL), intraoperative blood loss (IBL), and hidden blood loss (HBL) were all significantly higher in the UBE group (TBL: 270.43 ± 119.44 ml, IBL: 95.20 ± 33.99 ml, HBL: 175.67 ± 83.47 ml) compared to the PELD group (TBL: 37.32 ± 13.69 ml, IBL: 13.17 ± 4.06 ml, HBL: 24.15 ± 12.34 ml), with all differences being statistically significant (P < 0.001). These findings suggest that PELD is more advantageous in controlling intraoperative and perioperative blood loss. Clinical Outcomes Both procedures resulted in significant improvements in pain and function. Preoperative VAS and ODI values were comparable between groups (P > 0.05). Postoperative VAS and ODI scores showed significant reductions from baseline at all follow-up time points in both groups (P < 0.05). On postoperative day-1, back-pain VAS was slightly lower in the PELD group compared with UBE (2.82 ± 1.45 vs. 3.33 ± 1.34; P = 0.046). This early difference did not persist. Leg-pain VAS remained similar between groups at day one, three months, and 12 months (P > 0.05). Back-pain VAS at 3 and 12 months also showed no significant differences. Functional recovery followed the same pattern. ODI values were comparable at baseline and remained similar at day one, three months, and 12 months (P > 0.05). At 12 months, both groups demonstrated high rates of favourable outcomes based on the modified Macnab criteria (PELD 91.9%, UBE 93.3%; P = 1.000). Overall, PELD and UBE achieved equivalent short-term and long-term clinical outcomes. The small day-1 advantage observed with PELD did not translate into a sustained clinical difference.(Table 3; Figs. 3 and 4). Table 3Comparison of clinical outcomes between the two groups during follow-up time pointsPELDUBE P VAS BPPreoperative6.11 ± 1.515.93 ± 1.870.5591-day postoperation2.82 ± 1.453.33 ± 1.340.0463-month postoperation1.81 ± 1.251.78 ± 1.320.92112-month postoperation0.98 ± 0.880.85 ± 0.710.357VAS LPPreoperative7.45 ± 1.267.20 ± 1.400.2991-day postoperation2.76 ± 1.392.63 ± 1.410.6243-month postoperation1.21 ± 1.131.35 ± 1.190.50612-month postoperation0.76 ± 0.760.80 ± 0.780.764ODI scorePreoperative59.73 ± 12.5357.22 ± 14.050.3001-day postoperation33.26 ± 12.1132.60 ± 11.700.7613-month postoperation21.50 ± 7.7022.22 ± 8.460.62512-month postoperation10.68 ± 4.189.67 ± 4.590.206MacNab (n, %)1.000Excellent/good/fair/poor35/22/5/035/21/4/0Excellent/good rate91.9%93.3% Comparison of clinical outcomes between the two groups during follow-up time points Comparison of clinical outcomes between two groups Comparison of clinical satisfaction rates between two groups Learning curve results To determine the learning period required for surgeons to achieve proficiency with each technique, CUSUM scatter plots for the UBE and PELD groups were subjected to curve-fitting analysis (Fig. 5). The optimal model for the UBE cohort was a cubic polynomial, expressed as y = 0.0108x³ − 1.61x² + 58.71x − 54.37, with an R² value of 0.984, indicating excellent goodness of fit. The PELD cohort was likewise best fitted by a cubic polynomial, y = − 0.0072x³ + 0.36x² + 4.06x + 91.98,R²=0.972,with an R² of 0.972, suggesting a similarly robust model fit. Learning curve of CUSUM analysis In CUSUM analysis, the peak of the fitted curve represents the transition point at which the surgeon moves from the learning phase into the proficiency phase. For the UBE group, this inflection point occurred at case 24, whereas the corresponding peak for the PELD group was observed at case 38. The learning curve morphology was characterized by an upward trend before the peak, indicating greater variability and longer operative times, followed by a downward and stabilized trend reflecting improved efficiency. Based on this pattern, the minimum number of cases required to achieve procedural proficiency was 24 for UBE and 38 for PELD. These findings suggest that surgeons tend to reach a stable and consistent level of performance with UBE earlier than with PELD. (Fig. 5) Table 4 compares the operative times between the learning phase and the post‑proficiency phase for both techniques. In the learning phase, the mean operative time was 95.87 ± 21.53 min for PELD and 157.63 ± 20.11 min for UBE (P < 0.001). After reaching proficiency, the operative time decreased to 71.96 ± 21.98 min for PELD and 116.20 ± 17.55 min for UBE, with the intergroup difference remaining statistically significant (P < 0.001). Table 4Operative time in the learning phase and post-proficiency phasePELDUBE P Learning phase (min)95.87 ± 21.53157.63 ± 20.11< 0.001Post-proficiency phase (min)71.96 ± 21.98116.20 ± 17.55< 0.001 Operative time in the learning phase and post-proficiency phase Complications Both groups exhibited low complication rates, with no severe adverse events such as dural tears, CSF leak, wrong level, disc space infections, or major hemorrhage. All complications resolved with conservative management. In the PELD group (n = 62), there were 2 complications (3.23%), both involving transient neuropathic pain in the distribution of the affected nerve root, without accompanying motor weakness or sensory deficits. In the UBE group (n = 60), there were also 2 complications (3.33%): one case of transient neuropathic pain and one case of focal sensory disturbance characterized by mild hypoesthesia in the ipsilateral lower limb, without motor involvement. All patients received conservative treatment including mecobalamin (0.5 mg, three times daily), celecoxib (200 mg, twice daily), rest, and activity restriction. Symptoms progressively improved within one week and fully resolved by one month postoperatively. There was no significant difference in overall complication rates between the two groups (P > 0.05). The underlying mechanisms were presumed to be associated with transient nerve traction, local inflammatory changes, or mild irritation during manipulation. Both techniques demonstrated excellent safety profiles, consistent with previously reported complication rates in the literature (5.37%–10.80%) [18–20]. Discussion FLLDH represents a distinct and clinically important subtype of lumbar disc herniation, characterised by disc extrusion in the extraforaminal region with preferential compression of the exiting nerve root. Since its first systematic description by Abdullah [5], the understanding of FLLDH has evolved substantially with advances in spinal anatomy and imaging techniques. With increasing clinical experience, it has been recognized that FLLDH accounts for approximately 7.0%–12.0% of all LDH cases [1], occurs most frequently at the L4–5 level [21], and typically presents in patients around the age of sixty [22]. In contrast, in the present cohort, the L5–S1 level was the most commonly affected segment, accounting for 52% of cases in the PELD group and 57% in the UBE group. The higher proportion of L5 - S1 cases in our cohort compared with previous reports may reflect regional referral patterns or anatomical characteristics of the study population; however, a definitive explanation cannot be established from the present retrospective study. In the early stages of clinical recognition, the diagnosis of FLLDH posed substantial challenges. Owing to its deep extraforaminal location and frequent obstruction by the superior articular process(SAP), plain radiographs and early CT-based imaging techniques often failed to adequately visualise the pathological region. With the introduction of high-resolution computed tomography and magnetic resonance imaging, the extraforaminal anatomy could be delineated more clearly, leading to a marked improvement in diagnostic accuracy from below 50% to over 85% [5, 23]. Concurrently, accumulating clinical evidence has demonstrated that conservative management of FLLDH yields limited effectiveness, with a reported success rate of approximately 10% [24], which is substantially lower than that observed in other subtypes of lumbar disc herniation. Consequently, surgical intervention has become the primary therapeutic option for achieving meaningful symptom relief in most affected patients. The surgical treatment of FLLDH has historically been challenging because of its deep anatomical location and proximity to stabilising posterior elements. Conventional open midline approaches provide reliable exposure but often require extensive bony resection, which may compromise segmental stability and contribute to persistent low back pain [25, 26]. Subsequent paramedian approaches improved direct access to the extraforaminal region while preserving posterior midline structures; however, they were still associated with considerable paraspinal muscle dissection and increased surgical morbidity [27]. The advent of minimally invasive spinal techniques has therefore been pivotal in refining surgical strategies for FLLDH by enabling targeted neural decompression while minimising collateral tissue injury. PELD and UBE are two representative minimally invasive techniques currently applied in the management of FLLDH. Both techniques aim to achieve adequate decompression of the exiting nerve root through limited surgical corridors, yet they differ fundamentally in operative configuration, visualisation, and instrument handling. In the present study, both PELD and UBE resulted in significant and sustained improvements in pain relief and functional outcomes at mid-term follow-up, with no clinically meaningful differences between the two groups. These findings are consistent with Wang et al. [28], who similarly reported no significant differences in VAS or ODI outcomes between UBE and PELD for FLLDH during follow-up. The present study demonstrated that despite comparable clinical efficacy, notable differences were observed in peri-operative parameters: PELD offered advantages in terms of shorter operative duration, lower intra-operative and hidden blood loss, and lower total hospital costs, which is consistent with the findings of Qiao et al. [29], who reported that the UBE group had higher total blood loss, longer operative time, longer hospital stay, and higher total hospital costs. These findings are consistent with the percutaneous transforaminal nature of PELD, which limits paraspinal muscle disruption and restricts the potential space for occult bleeding. In contrast, UBE requires a wider operative corridor and more extensive muscle detachment, which may contribute to increased diffuse oozing from muscular and cancellous bone surfaces. Moreover, prolonged continuous saline irrigation during UBE, while essential for maintaining a clear endoscopic view, may obscure active bleeding and facilitate blood accumulation within paraspinal tissue compartments. Prolonged irrigation may also lead to dilution and washout of local coagulation factors at the surgical site, thereby impairing effective haemostasis and cumulatively increasing blood loss. It should be noted that the estimation of HBL based on hematocrit changes may be influenced by perioperative fluid shifts. In UBE procedures, continuous saline irrigation is required to maintain a clear operative field, which may result in interstitial fluid extravasation and partial systemic absorption. This fluid shift can dilute postoperative hematocrit values and potentially lead to an overestimation of calculated blood loss. Therefore, the significantly higher HBL observed in the UBE group should be interpreted with caution, as it may partially reflect this methodological limitation rather than true hemorrhagic volume. However, UBE offers distinct procedural advantages that may be particularly relevant in anatomically complex cases. The biportal configuration provides a broader operative field and greater instrument manoeuvrability, allowing surgeons to employ bimanual techniques analogous to those used in open surgery. This enhanced flexibility may be advantageous when managing challenging anatomical conditions, such as hypertrophic transverse processes, high iliac crests, or morphological variation of the facet joints. This advantage also reduces the risk of UBE when dealing with cases such as L5-S1 segment or facet joint instability [30]. These characteristics may also contribute to the shorter learning curve observed for UBE in this study. CSA-IVF was used as an imaging parameter to quantify the extent of foraminoplasty and bony decompression during surgery. Adequate enlargement of the intervertebral foramen is particularly relevant in FLLDH because sufficient surgical access is required to expose and decompress the exiting nerve root. In the present study, both techniques achieved comparable postoperative CSA-IVF enlargement. Nevertheless, CSA-IVF should be considered complementary to clinical outcome measures rather than a direct substitute for functional or symptomatic improvement. Learning curve analysis revealed that fewer cases were required to achieve procedural proficiency with UBE compared with PELD. This difference may be attributed to several factors. First, the visual orientation in UBE closely resembles that of conventional open surgery, facilitating intuitive anatomical recognition. Second, UBE permits the use of standard spinal instruments, reducing the need for adaptation to specialised endoscopic tools. Finally, surgeons with prior arthroscopic experience may more readily transfer existing skills to the biportal endoscopic environment. By contrast, PELD necessitates precise coordination of visualisation and instrumentation within a single working channel, placing higher demands on endoscopic spatial awareness and fine motor control, particularly during the early learning phase. Although operative time was used as the primary variable in CUSUM analysis, it does not inherently reflect the quality of decompression. In this study, surgical adequacy was ensured through strict intraoperative criteria, including direct visualization of nerve root decompression and confirmation of neural mobility. Moreover, operative time during the learning phase was longer than that in the post-proficiency phase and the full cohort. In the present study, the learning phase was associated with longer operative times for both UBE (157.63 ± 20.11 min, cases 1–24) and PELD (95.87 ± 21.53 min, cases 1–38). After the proficiency cutoff, operative times decreased to 116.20 ± 17.55 min for UBE and 71.96 ± 21.98 min for PELD. Notably, the intergroup difference in operative time remained highly significant both during the learning phase (61.76 min) and after proficiency was achieved (44.24 min). Excluding learning phase cases reduced the intergroup difference by only approximately 2 min (from 46.16 min to 44.24 min). This consistency indicates that the longer operative time observed for UBE is not an artifact of the learning curve but rather reflects inherent differences between the two endoscopic systems. Moreover, operative time comparisons may be influenced by the distribution of surgical levels. In the present study, L5–S1 cases accounted for the largest proportion in both groups. This level is known to present greater technical challenges due to anatomical constraints such as a high iliac crest, narrow working corridor, and variability in extraforaminal access. In clinical practice, UBE may be preferentially selected for anatomically complex cases, which could further contribute to longer operative times. Therefore, pooling cases from different levels (L3-4, L4-5, and L5-S1) may introduce confounding bias and potentially exaggerate the observed difference in operative duration. Accordingly, the finding that UBE requires a longer operative time should be interpreted with caution, as it may partially reflect case complexity rather than intrinsic procedural inefficiency. Limitations Several limitations should be acknowledged. First, this was a retrospective single-center study with a relatively limited sample size, which may restrict the generalizability of the findings. Second, the learning curve analysis reflected the experience of a single surgical team and did not account for inter-surgeon variability. Third, operative time comparisons may have been influenced by the high proportion of L5–S1 cases and the absence of level-stratified analysis. In addition, MRI-based assessment of neural decompression was not performed. Future multicenter prospective studies with larger cohorts, longer follow-up, and level-specific analyses are needed to further validate these findings. Conclusion Both UBE and PELD are safe and effective techniques for the treatment of FLLDH, providing comparable mid-term clinical outcomes. PELD demonstrated advantages in operative time, blood loss, and hospitalization cost, whereas UBE showed a shorter learning curve. CUSUM analysis indicated that procedural proficiency was achieved after approximately 24 cases for UBE and 38 cases for PELD. These findings may help guide surgical decision-making and endoscopic training strategies in the management of FLLDH. Acknowledgements Not applicable. Abbreviations Unilateral biportal endoscopy Percutaneous endoscopic lumbar discectomy Far lateral lumbar disc herniation Visual analogue scale for back pain Visual analogue scale for leg pain Oswestry Disability Index Body mass index Total blood loss Intraoperative blood loss Hidden blood loss Cross-sectional area of the intervertebral foramen Superior articular process Computed tomography Magnetic resonance imaging Authors’ contributions All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Haoran Gao, Xiao Sun. The first draft of the manuscript was written by Yazhou Li. Honggang Zhou and Yupeng Wang contributed to statistical analysis and interpretation of the data. Xiaohu Wang assisted with data collection and literature review. Haijun Ma and Mingkui Shen supervised the study and critically revised the manuscript for important intellectual content. All authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Funding This study was supported by National key clinical specialty construction Project (Yu Wei Medical Letter [2023] No.30), and Henan Charity Federation Daojian Fund General Project (No. szsyky24001, No. szsyky24021). Data availability The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate This study was approved by the Ethics Committee of The Third People’s Hospital of Henan Province, and all patients provided informed consent. All procedures were performed in accordance with the Declaration of Helsinki and relevant ethical guidelines and regulations. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Mingkui Shen, Email: smk0422@163.com. Haijun Ma, Email: mhjspine1234@163.com. References Associated Data Data Availability Statement The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.