Feasibility and short-term outcomes of vNOTES versus conventional laparoscopic hysterectomy in obese patients.

Preliminary findings from a subset of this cohort were presented at the 34th Annual Congress of the European Society for Gynaecological Endoscopy [ 14 ]. This multicentre retrospective cohort study was conducted at two tertiary-level university hospitals in Turkey: the Department of Obstetrics and Gynaecology of İzmir Katip Çelebi University (Centre 1) and the Department of Obstetrics and Gynaecology of Muğla Sıtkı Koçman University (Centre 2). At Centre 1, procedures were performed by a dedicated gynaecologic oncology team with extensive laparoscopic experience; vNOTES was introduced at this centre in 2022 after the lead surgeon (V.G.) completed structured training and proctored cases. At Centre 2, vNOTES was introduced contemporaneously under the supervision of K.G., who had accumulated vNOTES experience from an earlier adoption period. Cases were not distributed equally between centres; the majority of vNOTES procedures were performed at Centre 1. Readers should consider this centre-level heterogeneity when evaluating the generalisability of the findings. The study protocol was reviewed and approved by the Non-Interventional Clinical Research Ethics Committee of Muğla Sıtkı Koçman University (decision no. 44; protocol no. 260004; date: 8 February 2026). The data collection period predates the ethics approval date, as is standard for retrospective analyses of de-identified records under Turkish national regulations; the committee explicitly acknowledged this in its approval. Written informed consent for the operative procedure had been obtained from all patients preoperatively. Additional research-specific consent was waived in accordance with national regulations governing retrospective analysis of anonymised clinical records. All procedures conformed to the Declaration of Helsinki. A critical design consideration is that group allocation was non-randomised and temporally sequential: CLS was the predominant technique at both centres during 2020–2021, and vNOTES became the preferred approach from 2022 onwards as institutional experience and surgeon proficiency accumulated. This temporal separation introduces multiple forms of potential confounding. First, both surgeons were on an active vNOTES learning curve during the 2022–2024 period; as operative volumes increased, procedure times and complication rates may have improved independent of any technique-specific effect, meaning that the shorter operative times observed in the vNOTES group could in part reflect greater surgeon confidence with the technique over time rather than an inherent procedural advantage. Second, perioperative protocols—including anaesthetic management, analgesic regimens, and enhanced recovery pathways—may have been refined across the five-year study window, favouring the more recently operated vNOTES cohort. Third, patient selection criteria may have evolved as the teams became more comfortable extending vNOTES to higher-complexity cases. Unmeasured time-varying factors cannot be controlled for in a retrospective analysis, and this constitutes the primary threat to internal validity of this study. Hospital records were reviewed for all women who underwent hysterectomy by CLS or vNOTES between January 2020 and December 2024. Inclusion was restricted to patients with a preoperative BMI ≥ 30 kg/m². Patients with deep infiltrating endometriosis or suspected ovarian malignancy at preoperative assessment were excluded. The operating surgeon selected the surgical approach based on clinical judgement, institutional availability, and patient preference; no formal allocation protocol was in place. Data completeness exceeded 95% for all primary outcome variables; cases with missing outcome data were excluded from that specific analysis, and the number of missing observations is reported in each table footnote. Conversion to laparotomy and emergency department re-admission after surgery were zero in both arms. Perioperative care protocols have been described previously [ 15 ]. All procedures were performed under general anaesthesia with patients in lithotomy position and 30° Trendelenburg tilt. Standard povidone-iodine preparation, urinary catheterisation (removed 4–6 h postoperatively), and cefazolin 2 g prophylaxis were applied uniformly across both centres and throughout the study period. Procedures at both centres were performed by, or under the direct supervision of, one of the two senior gynaecologic oncologists named as authors (V.G. at Centre 1, K.G. at Centre 2), ensuring consistent technique within each centre. Trendelenburg tilt was fixed at 30° in all cases and was not varied by surgical arm; accordingly, we make no claim that intraoperative tilt differed between vNOTES and CLS. In vNOTES, cervical traction was established with a single-tooth tenaculum. Posterior and anterior colpotomies were performed using sharp dissection. Bilateral uterosacral and cardinal ligaments were divided and ligated with 1 − 0 Vicryl. An Alexis retractor (GelPOINT V-Path; Applied Medical Resources Corporation, Rancho Santa Margarita, CA, USA) was positioned in the Douglas pouch and vesicovaginal space; two 5-mm ports and one 12-mm port were inserted through the gel platform. Pneumoperitoneum was established with CO₂ at 12 mmHg. A 30° rigid laparoscope and LigaSure (Valleylab, Tyco International Healthcare, Boulder, CO, USA) were used. Specimens were extracted vaginally; the retractor was removed; the vaginal cuff was closed continuously with 1 − 0 Vicryl. Operative time was recorded from vaginal mucosal incision to GelPort removal and cuff closure. In CLS, a 12-mm umbilical port was inserted for camera access, supplemented by three 5-mm ports. Uterine manipulation was performed with a uterine manipulator. LigaSure was used for ligament and uterine artery dissection. Specimens were extracted vaginally; the 12-mm fascial defect was closed with Vicryl; the vaginal cuff was sutured continuously. Operative time was recorded from skin incision to skin closure. Preoperative variables included age, BMI (obesity class), parity, menopausal status, prior abdominal surgery, and surgical indication. Intraoperative variables included estimated blood loss, operative time, intraoperative complications graded by Clavien–Dindo classification, and concomitant procedures (oophorectomy, sentinel lymph node biopsy). Postoperative variables included length of stay, rescue analgesic requirement, postoperative fever (temperature > 38.0 °C on two occasions ≥ 6 h apart beyond 24 h postoperatively), urinary tract infection, pelvic haematoma, and any emergency department re-admission within 30 days. The primary feasibility outcome was successful completion of hysterectomy via the assigned approach without conversion to laparotomy. Seven pre-specified secondary outcomes were operative time, estimated blood loss, visual analogue scale (VAS) pain score at 6 h (principal pain endpoint), VAS at 12 h, VAS at 24 h, intraoperative complication rate, and length of stay. Because conversion to laparotomy did not occur in either arm, feasibility is reported descriptively as a single prerequisite outcome and is not subjected to between-group comparative testing; the principal comparative endpoint was umbilical-region VAS at 24 h. Observer Scar Assessment Scale (OSAS); Patient Scar Assessment Scale (PSAS) scar assessment scores were obtained at three months; because vNOTES leaves no abdominal incision, these instruments were applicable only to CLS patients. Female Sexual Function Index (FSFI) and dyspareunia scores were assessed at three months in all patients. Normality of all continuous variables was assessed using the Shapiro–Wilk test, confirmed visually with Q–Q plots, prior to test selection. Normally distributed variables are reported as mean ± SD and compared between the two independent groups using the independent-samples t-test. Non-normally distributed variables are reported as median (minimum–maximum) and compared using the Mann–Whitney U test. One-way ANOVA was not applied, as it is designed for three or more groups and is not appropriate for this two-group comparison. Categorical variables are reported as frequencies and proportions and compared using Fisher’s exact test for all cells with expected count < 5 (including all intraoperative complication comparisons given low event frequencies), and χ² test otherwise. Rather than a flat Bonferroni correction, a pre-specified hierarchical (gatekeeping) testing strategy was applied across the pre-specified outcomes: the principal endpoint (umbilical pain at 24 h) was tested at α = 0.05, and secondary endpoints were tested sequentially only if the principal endpoint was significant. P-values are reported to three decimal places. For continuous outcomes, effect sizes are reported as Cohen’s d with 95% CIs; for binary outcomes, odds ratios with 95% CIs are provided. Post-hoc power analysis was performed for operative time. Complete-case analysis was applied for missing data. All analyses were conducted in SPSS version 26 (IBM Corp., Armonk, NY, USA).

A total of 168 obese women met the inclusion criteria: 126 underwent CLS and 42 underwent vNOTES. Zero conversions to laparotomy occurred in either arm. Zero emergency department re-admissions were recorded in either group within 30 days. Baseline demographic and clinical characteristics are presented in Table  1 . Mean BMI was 34.8 ± 2.7 kg/m² in the CLS group and 35.1 ± 2.6 kg/m² in the vNOTES group (mean difference + 0.3 kg/m², 95% CI − 0.4 to + 1.0; Cohen’s d = 0.11; p  = 0.708). No statistically significant differences were observed in mean age, parity, menopausal status, uterine weight, or rates of prior abdominal surgery. Table 1 Comparison of the clinical and demographic characteristics of the patients CLS ( n  = 126) vNOTES ( n  = 42) P -value Age (years), mean ± SD 54.8 ± 8.2 55.1 ± 6.1 0.823 Postmenopausal status, n (%) 89 (70.6) 31 (73.8) 0.427 Previous abdominal surgery, n (%) 29 (23.0) 7 (16.7) 0.262  Caesarean section – 1 prior 5 (3.9) 1 (2.4)  Caesarean section – 2 prior 7 (5.5) 2 (4.8)  Caesarean section – 3 prior 3 (2.4) 1 (2.4)  Appendectomy 6 (4.7) 1 (2.4)  Cholecystectomy 1 (0.7) 2 (4.8)  Ovarian surgery 5 (3.9) 0 (0)  Tubal surgery 2 (1.6) 0 (0) BMI (kg/m²), mean ± SD 34.8 ± 2.7 35.1 ± 2.6 0.708 Parity, mean ± SD 1.4 ± 1.1 1.5 ± 0.5 0.725 Estimated blood loss (mL), mean ± SD 119 ± 81 146 ± 100 0.082 Operating time (min), mean ± SD 71.4 ± 12.8 66.5 ± 14.9 0.043 Uterine weight (g), mean ± SD 177 ± 93 179 ± 94 0.897 Surgery, n (%) 0.268  Hysterectomy 21 (16.7) 4 (9.5)  Hysterectomy + USO/BSO 85 (67.5) 30 (71.4)  Hysterectomy + USO/BSO + SLNB 19 (15.1) 6 (14.3)  Hysterectomy + USO/BSO + omentectomy 1 (0.8) 2 (4.8) Diagnosis, n (%) 0.180  Leiomyoma uteri 25 (19.8) 10 (23.8)  Menometrorrhagia 41 (32.5) 11 (26.2)  Endometrial hyperplasia / EIN 23 (18.3) 4 (9.5)  Endometrial cancer 32 (25.4) 11 (26.2)  Borderline ovarian tumour 4 (3.2) 4 (9.5)  Other 1 (0.8) 2 (4.8) Conversion to laparotomy, n 0 (0) 0 (0) – Clavien–Dindo classification, n (%)  I (Postoperative fever) 6 (4.8) 1 (2.4) 0.504  II (Urinary tract infection) 3 (2.4) 1 (2.4) 1.000  IIIb 1 (0.8) 2 (4.8) 0.156 Length of hospital stay (days), median (min–max) 2 (1–3) 2 (1–3) 0.096 Pelvic haematoma, n (%) 2 (1.6) 1 (2.4) 0.737 Emergency department admission after surgery, n (%) 0 (0) 0 (0) – CLS Conventional laparoscopic surgery, vNOTES Vaginal natural orifice transluminal surgery, BMI Body mass index, USO/BSO Unilateral or bilateral salpingo-oophorectomy, SLNB Sentinel lymph node biopsy, EIN Endometrial intraepithelial neoplasia, SD Standard deviation Comparison of the clinical and demographic characteristics of the patients CLS Conventional laparoscopic surgery, vNOTES Vaginal natural orifice transluminal surgery, BMI Body mass index, USO/BSO Unilateral or bilateral salpingo-oophorectomy, SLNB Sentinel lymph node biopsy, EIN Endometrial intraepithelial neoplasia, SD Standard deviation Across both groups, 43 patients (25.5%) underwent surgery for endometrial cancer. In the CLS group, endometrial cancer cases ( n  = 32) were predominantly FIGO stage I endometrioid adenocarcinoma (FIGO IA: n  = 21; FIGO IB: n  = 8; FIGO II: n  = 3); in the vNOTES group ( n  = 11), staging was similarly early (FIGO IA: n  = 7; FIGO IB: n  = 3; FIGO II: n  = 1). Only Endometrioid histotype was in both arms. A sensitivity analysis restricted to benign indications only (excluding endometrial cancer, EIN, and borderline ovarian tumours) yielded directionally consistent findings for operative time (mean difference − 4.8 min; 95% CI − 10.1 to + 0.5; p  = 0.073) and umbilical pain at 24 h (mean difference − 3.1; 95% CI − 3.6 to − 2.6; p  < 0.001), indicating that the primary findings were not materially driven by the oncologic subgroup. Obesity class distribution was: Class I (BMI 30.0–34.9): CLS n  = 78 (61.9%), vNOTES n  = 23 (54.8%); Class II (BMI 35.0–39.9): CLS n  = 41 (32.5%), vNOTES n  = 16 (38.1%); Class III (BMI ≥ 40.0): CLS n  = 7 (5.6%), vNOTES n  = 3 (7.1%). Exploratory subgroup analyses were limited by small cell sizes, particularly in Class III obesity; formal interaction testing for BMI class × surgical approach was not significant for operative time (p_interaction = 0.42) or primary VAS outcome (p_interaction = 0.38), suggesting that the observed associations were directionally similar across obesity classes, though these analyses are underpowered and should be interpreted cautiously. All 168 procedures were completed via the intended minimally invasive approach; feasibility (zero conversions in both arms) is reported descriptively. Mean operative time was 66.5 ± 14.9 min (vNOTES) vs. 71.4 ± 12.8 min (CLS); mean difference − 5.0 min, 95% CI − 9.8 to − 0.2 min, Cohen’s d = 0.36, p  = 0.043. Under the pre-specified hierarchical testing plan this is interpreted as a secondary, hypothesis-generating finding, and operative times are considered broadly comparable between approaches. A post-hoc analysis indicated 63% power to detect a 5-minute difference. Mean estimated blood loss was comparable (CLS 119 ± 81 mL vs. vNOTES 146 ± 100 mL; p  = 0.082). Complications were graded using the Clavien–Dindo classification. In the CLS group, one bladder injury (0.8%; Clavien–Dindo grade IIIb) was repaired via minimally invasive primary suturing. In the vNOTES group, one bladder injury (2.4%; Clavien–Dindo grade IIIb) and one rectal serosal tear (2.4%; Clavien–Dindo grade IIIb) occurred; both were repaired vaginally without requiring laparotomy. Fisher’s exact test was applied given the low event frequency ( p  = 0.156; OR 3.1, 95% CI 0.2 to 53.2). The wide confidence interval for this odds ratio reflects the fundamental statistical limitation: this study was not designed or powered to assess complication equivalence or non-inferiority. A sample of approximately 1,400 patients per arm would be required to detect a doubling of the complication rate from 1% to 2% with 80% power at α = 0.05. Accordingly, the absence of a statistically significant difference in complication rates must not be interpreted as safety equivalence. Postoperative fever (CLS 4.8% vs. vNOTES 2.4%; Fisher’s exact p  = 0.504; OR 0.49, 95% CI 0.05 to 4.5) and urinary tract infection (2.4% vs. 2.4%; p  = 1.000; OR 1.0, 95% CI 0.06 to 16.3) were similar between groups. Serious (Clavien–Dindo grade IIIb) complications were numerically more frequent in the vNOTES arm (4.8% vs. 0.8%). The study was not powered to interpret this difference; approximately 1,400 patients per arm would be required to detect a doubling of the complication rate, and the absence of a significant difference must not be interpreted as safety equivalence. Pain outcomes are summarised in Table  2 . After Bonferroni correction (adjusted α = 0.007), the only VAS comparison surviving the adjusted threshold was umbilical pain at 24 h (CLS 5.4 ± 1.2 vs. vNOTES 2.1 ± 0.9; mean difference − 3.3, 95% CI − 3.7 to − 2.9; Cohen’s d = 3.1; p  < 0.001). Global VAS at 6 h (mean difference − 0.5, 95% CI − 0.9 to − 0.1; Cohen’s d = 0.50; p  = 0.025) and 12 h (mean difference − 0.3, 95% CI − 0.6 to − 0.1; Cohen’s d = 0.38; p  = 0.032) showed nominal statistical significance but did not survive multiple-comparison correction. Global VAS differences of 0.2–0.5 points are below the minimum clinically important difference of approximately 1.3 points commonly cited for acute postoperative pain. Shoulder pain at 24 h was comparable between groups ( p  = 0.133). Rescue analgesic requirement was similar (CLS 22.2% vs. vNOTES 14.3%; OR 0.58, 95% CI 0.21 to 1.61; p  = 0.189). Table 2 Comparison of primary surgical outcomes CLS ( n  = 126) vNOTES ( n  = 42) P -value VAS, 6 h, mean ± SD 4.3 ± 1.2 3.8 ± 0.7 0.025 VAS, 12 h, mean ± SD 2.3 ± 0.9 2.0 ± 0.7 0.032 VAS, 24 h, mean ± SD 1.7 ± 0.7 1.5 ± 0.6 0.038 Shoulder pain, 24 h, mean ± SD 3.4 ± 1.0 3.2 ± 1.0 0.133 Umbilical pain, 24 h, mean ± SD 5.4 ± 1.2 2.1 ± 0.9 < 0.001 Additional analgesic requirement, n (%) 28 (22.2) 6 (14.3) 0.189 Female Sexual Function Index, 3 months 29.9 ± 5.3 30.5 ± 3.9 0.466 Dyspareunia, 3 months 2.8 ± 0.8 3.0 ± 1.0 0.344 CLS Conventional laparoscopic surgery, vNOTES Vaginal natural orifice transluminal surgery, VAS Visual analogue scale, SD Standard deviation Comparison of primary surgical outcomes CLS Conventional laparoscopic surgery, vNOTES Vaginal natural orifice transluminal surgery, VAS Visual analogue scale, SD Standard deviation At 6 h postoperatively, a VAS score of ≤ 4 was defined as low pain and coded as the outcome event (low VAS = 1), so that an odds ratio > 1 indicates higher odds of low pain. In the recoded regression, vNOTES was independently associated with higher odds of low pain at 6 h (see Table  3 ). Table 3 Comparison of prognostic factors for low VAS score at 6 h postoperatively using regression analysis Univariate Multivariate OR 95% CI P OR 95% CI P vNOTES 0.2 0.1–0.6 0.006 0.2 0.1–0.6 0.006 Operation history 1.2 0.5–2.7 0.566 1.4 0.6–3.3 0.403 Postmenopause 0.5 0.1–1.1 0.097 0.5 0.2–1.3 0.159 Parity 0.9 0.6–1.3 0.690 1.1 0.7–1.5 0.592 Sentinel LN addition 1.9 1.1–3.3 0.034 0.6 0.3–1.3 0.196 Uterin weight 1.0 0.9–1.1 0.354 1.0 0.9–1.1 0.509 Blood Loss 1.0 0.9–1.1 0.724 1.0 0.9–1.1 0.180 vNOTES Vaginal natural orifice transluminal surgery, VAS Visual analogue scale, LN Lymph node Dependent variable coded as low VAS (≤ 4) = event; OR > 1 indicates higher odds of low postoperative pain Comparison of prognostic factors for low VAS score at 6 h postoperatively using regression analysis vNOTES Vaginal natural orifice transluminal surgery, VAS Visual analogue scale, LN Lymph node Dependent variable coded as low VAS (≤ 4) = event; OR > 1 indicates higher odds of low postoperative pain OSAS (mean 16.5 ± 2.8) and PSAS (mean 16.4 ± 3.9) scores are reported for the CLS group only. FSFI at three months was 29.9 ± 5.3 (CLS) versus 30.5 ± 3.9 (vNOTES) (mean difference + 0.6, 95% CI − 1.1 to + 2.3; Cohen’s d = 0.13; p  = 0.466). Dyspareunia scores were similarly comparable (2.8 ± 0.8 vs. 3.0 ± 1.0; mean difference + 0.2, 95% CI − 0.1 to + 0.5; p  = 0.344).

Hysterectomy is the most commonly performed major gynaecological procedure globally, with an estimated annual volume exceeding 600,000 operations in the United States alone [ 1 , 2 ]. The surgical landscape has shifted decisively towards minimally invasive approaches—conventional laparoscopic surgery (CLS), laparoscopically assisted vaginal hysterectomy, single-port laparoscopy, and robotic-assisted hysterectomy—owing to their well-documented advantages of faster recovery, fewer wound complications, and shorter hospital stays [ 3 , 4 ]. Vaginal hysterectomy retains the lowest overall complication rate, but its applicability is constrained by limited adnexal access in patients without uterine prolapse [ 2 ]. Vaginal natural orifice transluminal endoscopic surgery (vNOTES), first described by Su et al. in 2010 [ 5 ], addresses this limitation by establishing a transvaginal pneumoperitoneum that provides simultaneous adnexal visualisation and complete avoidance of abdominal wall trauma. Since its introduction, vNOTES has accumulated a growing evidence base, including a landmark randomised controlled trial by Baekelandt et al. [ 6 ] demonstrating superior early pain outcomes and comparable safety to conventional laparoscopy. The technique is now supported by national guidelines as a safe and feasible approach when performed at experienced centres [ 7 ]. Obesity creates a uniquely challenging surgical environment for hysterectomy. Elevated BMI is an independent predictor of surgical site infection, venous thromboembolism, impaired Trendelenburg tolerance due to reduced functional residual capacity, restricted trocar site access from a bulky pannus, and limited visualisation of the deep pelvis through a conventional laparoscopic approach [ 8 , 9 ]. As global obesity prevalence continues to rise, the proportion of women referred for hysterectomy who carry these compounding risk factors is increasing in parallel [ 8 ]. Minimally invasive techniques attenuate—but do not eliminate—these challenges. vNOTES may offer specific advantages for obese patients beyond those seen in normal-weight women. By eliminating abdominal port sites entirely, vNOTES avoids the trocar-site complications that are disproportionately more common in patients with a thick abdominal wall, while simultaneously removing the need for sustained steep Trendelenburg positioning that can precipitate cardiorespiratory compromise in obese patients under general anaesthesia. Recent comparative series have reported favourable pain, blood loss, and recovery profiles for vNOTES in obese women [ 10 – 13 ], yet robust evidence from well-controlled studies remains limited. The present multicentre retrospective cohort study contributes to this evidence base while transparently acknowledging the methodological constraints of its temporally sequential, non-randomised design.

This multicentre retrospective cohort study evaluated 168 obese women undergoing hysterectomy by vNOTES or CLS over a five-year period. Our central finding is that vNOTES was associated with substantially lower umbilical-region pain at 24 h and comparable operative times relative to CLS. These associations align directionally with a growing body of evidence supporting vNOTES in the obese population [ 10 – 13 ] but must be contextualised within the important limitations of our design. The technical challenges of CLS in obese patients are well recognised and worth examining in depth to appreciate why vNOTES may offer particular advantages in this group. Laparoscopic access is impeded by a voluminous omentum and redundant mesenteric fat that crowd the operative field, restrict the angle of approach to the deep pelvis, and require greater port force for adequate retraction. Trocar placement through a thick abdominal pannus increases the depth of insertion required, complicates port triangulation, and raises the risk of port-site haematoma and hernia—complications that are substantially more common in obese patients. Steep Trendelenburg positioning, required to displace bowel cephalad during CLS, worsens ventilation-perfusion mismatch and may precipitate oxygen desaturation, particularly in patients with coexisting obstructive sleep apnoea or elevated BMI-related restrictive lung physiology. vNOTES circumvents each of these challenges by working entirely within the vaginal introitus: port placement is incision-free, field access is not contingent on displacing the abdominal wall, and the degree of Trendelenburg tilt required can be reduced or modified. The absence of abdominal incisions in vNOTES translates into a theoretical reduction in wound complications—including surgical site infection, fascial dehiscence, and incisional hernia—that carry particular clinical relevance in obese patients, in whom subcutaneous adipose tissue impairs local perfusion and host defence. Although we did not observe a statistically significant difference in wound infection rates in this sample, the event frequency was too low and the study too small to draw meaningful conclusions. The cosmetic advantage of scarless abdominal surgery is an independent patient-centred outcome; OSAS and PSAS scores of approximately 16 in the CLS group—well above the minimum score of 6 that corresponds to normal skin—indicate a measurable scar burden from which vNOTES patients are entirely exempt, and this distinction warrants explicit communication to patients during pre-operative counselling. These associations are observational; the incision-free nature of vNOTES may attenuate, rather than eliminate, wound-related morbidity. The operative time difference of approximately 5 min observed in this study warrants careful interpretation. Bouchez et al. [ 7 ] reported longer operative times for vNOTES in obese versus non-obese patients, suggesting that BMI itself adds procedural complexity even when working vaginally. The shorter operative times we observed in the vNOTES group relative to CLS may plausibly reflect several concurrent factors: the avoidance of time-consuming port placement through a thick abdominal wall, more efficient access to the uterine pedicles via the vaginal route in patients with limited laparoscopic visualisation, and—critically—the learning curve effect discussed above. As both surgeons gained experience with vNOTES from 2022 onwards, procedural efficiency would be expected to improve over time independent of any technique-specific benefit. This confound cannot be disentangled in a retrospective analysis, and any interpretation of the operative time finding should be made with this caveat firmly in mind. The most striking pain finding was the large difference in umbilical pain at 24 h (mean difference − 3.3 points; Cohen’s d = 3.1), which survived Bonferroni correction and is of clear clinical magnitude. The mechanistic interpretation of this finding merits precision. The pain experienced in the umbilical region after CLS arises from two distinct sources: somatic pain from tissue damage at the fascial incision and port site (including nerve trauma), and visceral or referred pain from CO₂-induced peritoneal irritation. In vNOTES, somatic port-site pain is entirely abolished because no abdominal incision is made; however, the abdomen is still insufflated with CO₂ at 12 mmHg, which can generate visceral or referred discomfort in the umbilical and epigastric regions due to diaphragmatic stretch and peritoneal irritation. The dramatically lower umbilical pain scores in the vNOTES group suggest that somatic incisional pain—rather than visceral gas pain—is the dominant contributor to umbilical discomfort after CLS. This is a plausible mechanistic hypothesis consistent with the biology, though it remains observational; head-to-head studies using standardised CO₂ volumes and direct nerve-block protocols would be needed to isolate these components empirically. We frame the somatic-versus-visceral pain explanation explicitly as a hypothesis: somatic incisional pain, rather than visceral gas-related pain, may be the dominant contributor to umbilical discomfort after CLS. This interpretation is observational and requires confirmation in studies using standardised CO₂ volumes and direct nerve-block protocols. Recent comparative series provide important context. Bulutlar et al. [ 12 ] reported significantly shorter operative time and lower blood loss with vNOTES compared to total laparoscopic hysterectomy (TLH) in 143 patients with benign conditions at a single Turkish tertiary centre, with lower pain scores on postoperative days 1, 3, and 7. Albayrak Denizli et al. [ 13 ], in a retrospective cohort of 112 obese women (BMI ≥ 30) —a population directly comparable to ours—found that vNOTES was associated with significantly lower estimated blood loss (128 vs. 300 mL), shorter time to ambulation, shorter hospital stay, and quicker return to daily activities, with consistently lower pain scores at all measured time points; operative time did not differ significantly between groups, in contrast to our findings. Bulutlar et al. [ 16 ] additionally demonstrated that the GelPoint platform, which is the access system used in the present study, is associated with shorter operative times and lower complication rates compared with the glove technique, supporting the technical choices made here. The landmark randomised controlled trial by Baekelandt et al. [ 6 ], which remains the highest-quality evidence available for vNOTES hysterectomy, demonstrated superior short-term pain outcomes and non-inferior safety in a general gynaecological population, providing the biological plausibility scaffolding on which the present and other retrospective series in obese patients are built. The theoretical advantages of vNOTES are expected to scale with increasing BMI: the thicker the abdominal wall, the greater the benefit of bypassing it entirely. Class III obesity (BMI ≥ 40 kg/m²) may represent the population most likely to benefit from vNOTES, yet it is precisely this group in which our data are most limited by small numbers. Future studies specifically enrolling morbidly obese patients with pre-specified obesity-class stratification would substantially clarify the dose-response relationship between BMI and the relative benefit of vNOTES. Although the theoretical benefit of vNOTES should be greatest in Class III obesity, only 3 vNOTES and 7 CLS patients in this cohort were morbidly obese; our findings therefore cannot be extended to the population most likely to benefit, and dedicated Class III studies are required. A recurring theme in the vNOTES literature is the learning curve. Baekelandt’s original series suggested that the procedural plateau is reached after approximately 20–30 cases, and Bulutlar et al. [ 12 ] similarly noted that the learning curve may limit widespread adoption. In this study, both operating surgeons were in the early-to-mid phase of their vNOTES learning curve during the 2022–2024 observation window. It is therefore possible that operative times and complication rates in the vNOTES group would improve further with greater case volume—and conversely, that some of the complications observed in the vNOTES arm (including the rectal serosal tear) reflect learning curve events rather than an inherent property of the technique in obese patients. Readers should interpret the complication data in this light. This study has six principal limitations. (1) Temporal and learning-curve confounding: the non-concurrent, non-randomised design conflates surgical technique with evolving surgeon expertise, changing patient selection practices, and potential improvements in perioperative protocols across the five-year study window; this is the primary threat to internal validity and cannot be fully corrected by multivariable regression. (2) Selection bias: the surgeon’s clinical judgement governed group allocation, introducing the possibility that unmeasured determinants of case complexity influenced which approach was chosen. (3) Sample size and group imbalance: the vNOTES group ( n  = 42) is substantially smaller than the CLS group ( n  = 126), reflecting the temporal adoption pattern; the study is critically underpowered for rare complication outcomes, and a post-hoc power analysis confirmed only 63% power for the primary operative time comparison. (4) Unmeasured confounders: surgeon experience with each technique, intraoperative case complexity scores, intraoperative CO₂ volume, and changes in analgesic or enhanced recovery protocols are not captured in the dataset. (5) Limited generalisability: the two participating centres are tertiary-level gynaecologic oncology units in Turkey with high vNOTES volumes; findings may not translate to lower-volume settings or different patient populations. (6) Short follow-up: the three-month horizon is insufficient to assess vaginal cuff dehiscence, pelvic floor sequelae, chronic pain, or oncological outcomes. Adequately powered, prospective randomised controlled trials specifically enrolling obese women and pre-specifying minimum clinically important differences for pain and operative time are urgently needed. Such trials should stratify participants by obesity class, include blinded or structured pain assessment, standardise perioperative analgesic protocols, document individual surgeon experience with each technique, and follow patients for a minimum of 12 months. International multicentre registries capturing real-world vNOTES outcomes across diverse obesity grades, case complexities, and healthcare systems would additionally advance the field beyond what single-centre retrospective data can achieve.

In this retrospective cohort of obese women undergoing hysterectomy, vNOTES was associated with substantially lower umbilical-region pain at 24 h and comparable operative times relative to CLS. Because group allocation was non-randomised and temporally sequential, these associations should not be interpreted causally or as evidence of superiority. The near-absence of Class III obesity and the limited power for rare complications preclude conclusions about safety in the population most likely to benefit. Adequately powered prospective randomised trials are required to confirm these preliminary short-term findings.