Percutaneous endoscopic lumbar discectomy (PELD) is an emerging minimally invasive alternative to open and tubular microdiscectomy for treating symptomatic lumbar disc herniation. Generally, PELD is performed using either the transforaminal or interlaminar route, with each offering distinct advantages. Transforaminal techniques exploit Kambin’s triangle, a natural working corridor that provides direct access to the disc space with minimal soft tissue dissection [1]. By avoiding subperiosteal dissection of the paraspinal musculature and minimizing disruption to the stabilizing ligamentous structures, transforaminal PELD may reduce postoperative muscle denervation [2], atrophy, and subsequent segmental instability [3] compared with conventional posterior midline techniques. Interlaminar PELD is particularly valuable at the L5–S1 segment, which has a notably wide interlaminar space and where anatomical variants such as a high iliac crest can impede the more lateral transforaminal approach [4]. Through direct endoscopic visualization via the interlaminar window, effective decompression is similarly achieved without extensive tissue disruption.

Early results from randomized and observational studies suggest that PELD provides symptom relief and functional recovery comparable to those of microdiscectomy, with the added benefits of reduced perioperative morbidity and a shorter hospital stay [5–8]. Furthermore, PELD procedures are frequently conducted under local or regional anesthesia, allowing for same-day patient discharge and increased surgical efficiency [9]. These early advantages are largely attributed to the preservation of native anatomy, which may also lead to favorable outcomes over the longer term. However, the technical challenges inherent to endoscopic discectomy, primarily altered visualization [10] and confined working channels, contribute to an extended learning curve. Such limitations also raise concerns regarding the adequacy of decompression, the possibility of residual or recurrent pathology, and the possibility of delayed treatment failure. Therefore, further research explicitly evaluating long-term clinical and functional outcomes after PELD is necessary; however, such studies are few.

The relative absence of long-term data poses a significant challenge for surgical decision-making. While randomized controlled trials (RCTs) remain the gold standard, they are often limited by short follow-up durations, small sample sizes, and difficulty in retaining participants over time. As a result, RCTs are often underpowered to detect late complications or capture gradual changes in function [11]. These limitations are particularly pertinent in spine surgery, where symptom recurrence and reoperation may occur several years after the index procedure. For these reasons, well-designed observational trials will likely remain a valuable and pragmatic complement to randomized evidence when assessing long-term real-world outcomes and complication profiles [12].

This systematic review and meta-analysis aimed to address the current evidence gap by synthesizing the available evidence on long-term outcomes after PELD, specifically focusing on functional improvement, complication rates, recurrent disc herniation, reoperation rates, and radiographic changes after transforaminal or interlaminar endoscopic lumbar discectomy.

Two independent reviewers (D.K. and O.S.) searched the literature according to the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) [13]. The PubMed, Embase, and Scopus databases were queried on June 16, 2024, without a publication date restriction. The following search string was used for all the databases: (“endoscopic lumbar dis*ectomy” OR “percutaneous endoscopic lumbar dis*ectomy” OR PELD OR TELD OR IELD OR “full endoscopic” OR “full-endoscopic” OR “transforaminal endoscopic lumbar dis*ectomy” OR “interlaminar endoscopic lumbar dis*ectomy”) AND (“cohort*” OR “observational” OR “RCT” OR “randomi*ed controlled trial*” OR “outcome*” OR “complication*” OR “reherniation*” OR “reherniation*” OR “recurrence*” OR “revision*”) NOT (“pediatric end-stage liver disease” OR “pediatric end-stage liver disease”).

Duplicate studies were manually identified through digital object identifier (DOI)/PubMed identifier (PMID) and title checks and then removed. The two reviewers independently screened the results based on the title and abstract, applying exclusion criteria (Table 1). The remaining full texts were then independently assessed according to the inclusion criteria (Table 1). In cases of disagreement, one author (J.M.) acted as the final decision maker regarding study inclusion. Corresponding authors were emailed to obtain inaccessible full texts, and studies were excluded if no copy was received.

The literature search identified 4,264 studies (PubMed: 1,169, Embase: 1,732, Scopus: 1,363). After removing 2,304 duplicates, the remaining 1,960 studies were screened according to exclusion criteria, yielding 88 reports for full-text assessment, of which 11 records were inaccessible. The screening procedure is illustrated in Fig. 1.

Full endoscopic lumbar discectomy offers effective nerve root decompression while minimizing disruption to the surrounding soft tissues. This technique may facilitate quicker recovery, preservation of spinal stability, and reduced scar formation, which could collectively enhance long-term outcomes. While the short-term benefits of PELD are increasingly supported by evidence, data describing long-term outcomes remain limited.

Functional outcomes after PELD, assessed using standardized measures such as the VAS and ODI, showed significant improvement from baseline and sustained improvement over time. At 5 or more years postoperatively, the patients consistently reported minimal back and leg pain and low levels of disability. Among the included studies, 87.7% of patients rated their outcomes as good or excellent, suggesting that durable patient satisfaction was achieved. These findings are largely comparable to those reported after minimally invasive microdiscectomy at similar follow-up intervals [26,27], although direct comparisons should be approached with caution because of differences in study design and patient populations. Longitudinal data from open and microdiscectomy cohorts suggest that patient satisfaction may wane over time, particularly as the initial postoperative improvement plateaus and degeneration progresses [28–30]. While a similar trend might be expected after PELD, this review finds no evidence to support this, although high-quality data remains scarce. Additionally, much of the existing literature describing long-term outcomes after discectomy was published more than two decades ago, and any apparent decline in patient-reported outcomes may reflect outdated techniques and perioperative care rather than intrinsic patient progression after discectomy in current practice [31].

Surgical complications following PELD primarily arise in the early postoperative period, with transient neurological deficits, most commonly dysesthesia, being the most prevalent. Several potential causes have been proposed by the authors of the reviewed studies, including inadequate foraminoplasty, reactive nerve root edema, thermal injury, and surgeon inexperience [25]. Although dysesthesia after microdiscectomy is uncommon, it is a well-recognized complication of PELD, particularly after resection of lateral disc herniations, with a reported incidence of 1%–5% [32]. It is typically attributed to dorsal root ganglion irritation during insertion of the transforaminal working cannula or traction on the exiting nerve root [33]. Various technical modifications have been proposed to reduce this risk, although none have fully eliminated it [32–34]. Despite these concerns, dysesthesia is generally mild [33] and self-limited, with only eight instances of persistent deficit identified in this review of >2,000 patients after over 5 years of follow-up. Given this apparent absence of late adverse events, previous meta-analyses suggesting equivalent early complications of PELD and microdiscectomy can likely be relied upon to guide long-term decision-making [7].

Reported recurrence rates following discectomy increase with longer follow-up. Rates of 7.5% at 5 years [35], 8.5% at 8.6 years, and 9.7% at 10 years have been published [36]. A recent nationwide cohort study in the United States confirmed that recurrence is a persistent risk after the early postoperative period, finding a 5-year recurrence rate of 14.4% after primary discectomy and even higher recurrence rates after revision [37]. The pooled recurrence rate after PELD in this series compares favorably with these estimates, being 6.7% at 5 or more years postoperatively. Conventional wisdom states that extensive discectomy reduces the risk of recurrent herniation; therefore, this comparison may be confounded by varying surgeon practice. However, this interpretation has been challenged by numerous investigations reporting the absence of a relationship between the extent of dissection and the risk of recurrent herniation [38]. Elderly patients, smokers, and diabetics have been credibly associated with a greater risk of recurrent herniation, possibly due to the acceleration of the degenerative cascade unrelated to surgery [37,39]. However, the average age of participants included in this review was just 47 years, which likely contributed to a flattering estimate of the long-term recurrence rate. Robust, long-term comparative studies controlling for competing factors are necessary before definitive conclusions can be drawn regarding the comparative risk of recurrence between PELD and conventional techniques.

Long-term radiological outcomes following PELD were infrequently reported, and heterogeneous outcome measures were used, thus preventing a meta-analysis. Qualitatively, current evidence suggests that full endoscopic decompression generally results in minimal disc height loss (approximately 20%). In the two studies reporting segmental ROM, slight reductions were noted at long-term follow-up, suggesting no evidence of long-term instability after PELD based on this metric. Notably, vertebral translations were not described, yet instability could manifest in this area. Minor reductions in disc height are likely an inevitable outcome after discectomy [27]; however, randomized data indicate that greater disc height collapse may occur following more extensive disc resection [40], while lower-quality studies imply that microendoscopic and full endoscopic techniques may better preserve disc height than conventional microdiscectomy in the long term [22,41,42]. Such associations between disc height and surgical technique are at least somewhat confounded by variations in surgeon philosophy and surgical goals. Disc height collapse has been cited as a cause of instability after surgery [3] and has been linked with recurrent disc herniation [43], thus raising the possibility that the modest recurrence rate observed in this synthesis may reflect a relative preservation of intervertebral function.

This study had several limitations. Most included studies were retrospective, introducing potential biases from selective loss to follow-up and incomplete baseline data. Heterogeneity in surgical techniques, outcome definitions, and follow-up durations limits the generalizability of pooled estimates. The observational nature of the data introduced confounding, particularly from the imposition of minimum follow-up periods. If patients lost to follow-up differ meaningfully from those retained, selection bias may occur. In studies that reported loss to follow-up or exclusions for insufficient follow-up, this proportion was consistently under 15%; therefore, these studies may be considered a faithful representation of long-term outcomes. However, four studies did not report this metric, representing a residual source of bias. Small sample sizes in many studies reduce statistical power, although pooling partially mitigates this. Recurrent herniation and reoperation are longitudinal outcomes best assessed using survival analysis; however, the included publications did not provide Kaplan-Meier charts or similar data, which would have allowed for this type of analysis. Eleven articles identified during the literature search were inaccessible, and their exclusion may have affected the results of this study. Subgroup analyses assessing learning curve effects, differences in regional and general anesthesia, transforaminal versus interlaminar techniques, and effects of surgical level are important but were infeasible because of the small number of studies and aggregate level reporting. Perhaps the most relevant unanswered question is how PELD compares to conventional techniques in the long term. However, the near absence of long-term comparative trials limits such assessments to qualitative comparisons, with a high risk of bias and confounding.

This meta-analysis suggests that PELD provides durable symptom relief and a low risk of long-term recurrence, along with favorable functional outcomes. Radiological outcomes also appear acceptable, with modest disc height reduction and no evidence of long-term instability in the limited studies reporting ROM. However, most of the included studies were retrospective and heterogeneous in methodology, introducing a significant risk of bias and limiting the generalizability of the findings. The lack of long-term comparative trials prevents definitive conclusions regarding the relative merits of PELD over other discectomy techniques. Future research should prioritize prospective comparative trials with long-term follow-up to validate these conclusions.