Learning curve and perioperative outcomes: a retrospective cohort study between synchronous and combined inflatable mediastinoscope-laparoscopic radical esophagectomy

Abstract

Objective

To characterize the learning curve of inflatable mediastinoscope-assisted synchronous laparoscopic esophagectomy and to compare perioperative outcomes between synchronous and combined approaches.

Methods

A retrospective analysis was performed on the clinical data of 229 consecutive patients who underwent minimally invasive radical esophagectomy in the Department of Thoracic Surgery between January 2018 and December 2020. All patients were divided into the synchronous group (n = 118) and the combined group (n = 111), and after 1:1 propensity score matching, 106 patients were successfully matched in each group for final analysis. The cumulative sum (CUSUM) method was used to construct learning curves for both the synchronous and combined groups based on operative time, and learning phases were identified according to inflection points, dividing it into initial, plateau, and proficiency phases based on inflection points. Baseline characteristics, perioperative indicators (operative time, intraoperative blood loss, etc.), postoperative complications, and short-term outcomes were compared between the overall groups and across learning phases within the synchronous group.

Results

Baseline characteristics were balanced and comparable between the two groups (P > 0.05). Overall, the synchronous group demonstrated significantly superior outcomes compared to the combined group in terms of operative time, intraoperative blood loss, postoperative drainage volume at 3 days, total lymph node yield, time to first flatus, time to first ambulation, and postoperative hospital stay (P < 0.05). The CUSUM learning curve equation for the synchronous group was Y = − 0.001x³−0.146x² + 30.865x−80.301 (R² = 0.987), with inflection points at the 40^th and 70^th cases. Dividing the learning curve into the initial phase (cases 1–40), stabilization phase (cases 41–70), and proficiency phase (cases 71–106). As learning progressed, the synchronous group demonstrated significant improvements in operative time, blood loss, drainage volume, lymph node dissection rate, time to ambulation, and length of hospital stay (P < 0.05). Furthermore, the overall complication rate during the proficiency phase was significantly lower than that during the initial phase (P < 0.05).

Conclusion

Compared with the sequential combined inflatable mediastinoscope-assisted laparoscopic approach, the synchronous approach was associated with improved perioperative outcomes, fewer pulmonary complications, and a favorable learning trajectory in this retrospective cohort. These findings support its feasibility as an alternative minimally invasive strategy, although further prospective multicenter studies are required for validation.

Esophageal cancer (EC) poses a significant threat to digestive health [1, 2], and surgery remains the cornerstone of treatment for resectable cases within a multidisciplinary framework [3]. Driven by advances in minimally invasive techniques, video-assisted thoracoscopic esophagectomy (VATS) has emerged as the mainstream approach, offering distinct advantages of minimal invasiveness, enhanced safety, and accelerated postoperative recovery compared to traditional open surgery [4]. Building on this progress, the refinement of thoracoscopic and laparoscopic technologies has further elevated pneumoperitoneum-assisted mediastinoscopy combined with laparoscopy as a preferred option for EC resection, thanks to its minimal incisions, rapid recovery, and reduced postoperative complications [5]. Mediastinoscope-assisted esophagectomy has evolved from conventional mediastinoscope-assisted transhiatal esophagectomy to more refined approaches using pneumomediastinum or inflatable mediastinoscopy combined with laparoscopy [6–10]. Previous studies have reported that these non-transthoracic approaches may reduce the need for thoracic access and one-lung ventilation while allowing mediastinal esophageal mobilization and lymph node dissection under direct visualization [7, 8]. More recently, single-port inflatable mediastinoscopy combined with laparoscopic esophagectomy and inflatable video-assisted mediastinoscopic transhiatal esophagectomy have been described as feasible minimally invasive options for selected patients with esophageal cancer [9, 10]. However, this combined approach still harbors limitations, including complex procedural steps and prolonged single-lung ventilation—factors that may exacerbate pulmonary ischemia–reperfusion injury. To address these issues, our team modified the procedure by synchronizing mediastinoscopy with laparoscopy, aiming to reduce mechanical lung injury and inflammatory response to accelerate pulmonary function recovery and shorten operative time to alleviate systemic stress. The combined approach evolved from the conventional McKeown minimally invasive esophagectomy workflow and represented an important transitional stage in the development of the synchronous technique. The combined approach adopted mediastinoscopic and laparoscopic procedures in a sequential manner based on accumulated experience with conventional minimally invasive esophagectomy and mediastinoscope-assisted esophageal surgery. The synchronous approach was subsequently developed as a technical modification and optimization of the combined approach, aiming to improve operative efficiency and perioperative recovery while maintaining the same surgical principles and dissection fields.

Originally developed to detect trend shifts, cumulative sum (CUSUM) analysis has been widely used as a quantitative statistical tool for learning curve assessment since the 1970s [11, 12]. CUSUM analysis is particularly suitable for evaluating the acquisition of technically demanding surgical procedures by identifying changes in operative performance and proficiency thresholds. Previous studies have applied CUSUM or risk-adjusted CUSUM methods to assess learning curves in minimally invasive esophagectomy and robot-assisted minimally invasive esophagectomy, demonstrating its value in evaluating technical maturation and perioperative outcomes.

However, studies comparing the learning curves of synchronous and combined pneumoperitoneum-assisted mediastinoscopic radical esophagectomy remain limited [13]. Therefore, the present study employed CUSUM analysis to construct and compare learning curves for both approaches [14], while systematically evaluating perioperative outcomes across different learning phases. The findings of this study may provide practical guidance for centers and surgeons adopting this technique and may help optimize surgical training strategies.

Materials and methods

General information

This study was a retrospective cohort study. A total of 229 patients who underwent minimally invasive laparoscopic radical esophagectomy by the same surgical team were enrolled retrospectively from the Department of Thoracic Surgery, Huaihe Hospital of Henan University, between January 2018 and December 2020. Patients were categorized based on surgical approach into the pneumoperitoneum-assisted mediastinoscopy with simultaneous laparoscopy group (synchronous group, 118 cases) and the combined group (111 cases). After propensity score matching, 106 patients were finally included in the synchronous group and 106 in the combined group. Seventeen patients were not included in the final matched cohort because they could not be successfully matched within the prespecified caliper during propensity score matching (Fig. 1). The selection principle of surgical approaches balances technical background and clinical practice; after evaluating the patient’s condition, the attending surgeon discusses the advantages and disadvantages of each approach with the patient and their family, and jointly determines the surgical method while respecting their wishes.

Fig. 1.

Flow diagram of patient selection and propensity score matching. Note: Seventeen patients (12 in the synchronous group and 5 in the combined group) were excluded after propensity score matching because no appropriate match could be identified within the prespecified caliper

Inclusion criteria: (1) All patients had preoperative endoscopic examination combined with pathological diagnosis confirming esophageal squamous cell carcinoma, meeting surgical indications for T1-2N0-1M0, non-bulky T3N1M0 (Stage IIIa), or resectable T1-2/Stage IVa (regional lymph node metastasis only); (2) All patients successfully underwent pneumoperitoneum-assisted mediastinoscopy with simultaneous laparoscopy or combined laparoscopic procedures; (3) Complete clinical data, including baseline characteristics, surgical parameters, and follow-up data, with follow-up continued through 2025 to ensure a minimum follow-up duration of 5 years.

Exclusion Criteria: (1) Coagulation disorders, severe immunodeficiency, or allergy to anesthetic agents; (2) High sensitivity to anesthetic agents. (3) History of prior thoracic surgery, gastrointestinal tumor surgery, or dense pleural adhesions (impeding endoscopic manipulation); (4) Tumor maximum diameter ≥ 5 cm (giant T3), invasion through the esophageal fibrous membrane, or concomitant distant organ metastasis (M1b).

Surgical methods

Operative team and surgeon experience

All procedures were performed by a fixed surgical team consisting of two primary surgeons. One surgeon was responsible for the cervical mediastinoscopic phase, while the other performed the abdominal laparoscopic phase. Both surgeons had more than 10 years of experience in minimally invasive esophagectomy prior to the initiation of this study.

The operative roles were strictly maintained throughout the study period with each surgeon consistently performing the same part of the procedure. In addition, all operations followed a standardized surgical protocol. Although the assisting surgeons remained relatively stable, the scrub nurses and operating room nursing staff were not completely fixed throughout the study period. However, all nursing staff were from the same thoracic surgery operating unit and were familiar with minimally invasive esophagectomy procedures and standardized perioperative workflows. Standardized operative coordination and perioperative management protocols were consistently applied to minimize variability related to surgeon performance and team coordination.

Preoperative preparation

Fasting is required after 11:00 PM the night before surgery. Following the final meal, administer one dose of Compound Calming Decoction (to improve preoperative sleep and reduce anxiety). On the morning of surgery, insert a nasogastric tube, prepare all surgical instruments including plasma and staplers, and complete bowel preparation Fig. 1.

Inflatable mediastinoscope-assisted synchronous laparoscopic surgery group

In the synchronous group, the patient was placed in a supine split-leg position under general anesthesia (Fig. 2A), and the procedure was carried out simultaneously by two attending surgeons working in coordination (Fig. 2B). One surgeon was responsible for the cervical mediastinoscopic phase. A cervical incision was made along the anterior border of the sternocleidomastoid muscle, and CO₂ insufflation was established to create an artificial mediastinal space. The esophagus was mobilized under mediastinoscopic guidance, with careful dissection of the upper and lower mediastinum and systematic lymphadenectomy, including the recurrent laryngeal nerve lymph nodes. At the same time, the second surgeon performed the abdominal laparoscopic phase using standard trocar placement (Fig. 2C). Gastric mobilization and abdominal lymph node dissection were completed laparoscopically, followed by construction of the gastric conduit (Fig. 2D). Indocyanine green (ICG) fluorescence angiography was routinely used intraoperatively to assess gastric conduit perfusion before esophagogastric anastomosis. Anastomotic site selection was based on fluorescence perfusion assessment combined with conventional surgical judgment. The prepared conduit was then delivered through the posterior mediastinal route to the cervical incision. Finally, cervical esophagogastric anastomosis was performed after completion of both phases.

Fig. 2.

Operative setup for inflatable mediastinoscope-assisted synchronous laparoscopic esophagectomy. A Patient positioning; B Surgical team arrangement; C Placement of abdominal trocars; D Gastric conduit construction

Inflatable mediastinoscope combined with laparoscopic surgery group

In the combined group, the operation was performed sequentially by one attending surgeon. The cervical mediastinoscopic phase was completed first, including cervical incision, esophageal mobilization, recurrent laryngeal nerve protection, and mediastinal lymph node dissection. After completion of the cervical phase, the same surgeon proceeded to the abdominal laparoscopic phase, including gastric mobilization, abdominal lymphadenectomy, gastric conduit construction, and cervical esophagogastric anastomosis.

Postoperative management

Following surgery, in line with the principles of enhanced recovery, patients are encouraged to mobilize early to prevent pulmonary embolism. Both groups received the same standardized postoperative rehabilitation and early mobilization protocol. Postoperative nursing assessment, pain management, respiratory training, and ambulation guidance were performed according to the same institutional enhanced recovery protocol and at the same postoperative intervals. Early ambulation was generally encouraged beginning on postoperative day 1 after assessment of hemodynamic stability, respiratory condition, drainage status, and pain control. No additional or preferential physical therapy intervention was provided to either group. Concurrently, nutritional support via nasogastric feeding, anti-infective therapy, and fluid replacement are administered. Hematological parameters were dynamically monitored during the postoperative period. Upper gastrointestinal contrast examination using oral iohexol was routinely performed approximately 5–7 days after surgery to assess anastomotic integrity, including contrast leakage, delayed contrast passage, or other anastomotic abnormalities. Chest CT was additionally performed when clinically indicated to evaluate thoracic complications such as pleural effusion, mediastinal collections, pulmonary infection, or abscess formation.

Postoperatively, oral intake was withheld initially, and enteral nutrition was provided through the feeding tube. Enteral tube feeding was generally initiated on postoperative day 1–2 after hemodynamic stabilization and in the absence of severe gastrointestinal intolerance or procedure-related complications. The feeding rate and nutritional volume were gradually increased according to patient tolerance.

Anastomotic integrity and swallowing tolerance were evaluated during the early postoperative period. In patients without clinical evidence of anastomotic leakage, aspiration risk, or severe dysphagia, small amounts of clear-liquid oral intake were generally initiated approximately 14 days after surgery. Oral intake was then advanced stepwise from clear liquids to full liquids and subsequently to a soft diet according to individual tolerance and outpatient reassessment. The nasogastric tube was not routinely retained for prolonged gastrointestinal decompression. Instead, in selected patients, it was temporarily maintained during the early postoperative transition period to provide supplemental enteral nutritional support when oral intake remained insufficient. Tube removal was individualized according to oral intake tolerance, nutritional status, anastomotic condition, and overall clinical recovery during follow-up.

Clear liquids were used as the initial stage of oral intake, followed by gradual dietary advancement according to patient tolerance and postoperative recovery.

Observation indicators

Baseline characteristics

Baseline characteristics included age, sex, tumor location, tumor stage, tumor diameter, preoperative neoadjuvant therapy, and detailed comorbidity and lifestyle-related risk factors, including diabetes mellitus, hypertension, coronary heart disease, chronic pulmonary disease, smoking history, and alcohol use.

Perioperative indicators

Operative time was defined as the interval from initial skin incision to complete closure of all surgical incisions and did not include anesthesia induction, intubation, awakening, extubation, or anesthesia recovery time. Perioperative indicators included operative time, intraoperative blood loss, postoperative drainage volume at 3 days, total number of harvested lymph nodes confirmed by postoperative pathological examination, including both mediastinal and abdominal lymph nodes, time to first flatus postoperatively, time to first ambulation postoperatively, and postoperative length of hospital stay.

Postoperative complications

All postoperative complications were systematically assessed and categorized according to the definitions of the Esophageal Complications Consensus Group (ECCG). Complication severity was further graded using the Clavien–Dindo classification system [16].

Learning curve analysis

Learning curve analysis was performed for both the synchronous and combined groups. Operative time was arranged chronologically according to case sequence, and the cumulative sum (CUSUM) of deviations from the mean operative time was calculated to construct learning curves. Inflection points were identified based on fitted curves, and the learning process was divided into three phases: initial, stabilization, and proficiency phases.

Statistical analysis

Data analysis was performed using SPSS 26.0 software with patients grouped according to different surgical approaches for clinical data analysis. To minimize bias between groups, propensity score matching (PSM) was employed. Propensity score matching was performed to reduce selection bias between the synchronous and combined groups. Covariates were selected based on their potential clinical relevance to surgical approach selection and postoperative outcomes. The propensity score model included age, sex, tumor location, tumor stage, tumor diameter, preoperative neoadjuvant therapy, diabetes mellitus, hypertension, coronary heart disease, chronic pulmonary disease, smoking history, alcohol use, and year of surgery. These variables were included because they may influence operative complexity, postoperative complications, recovery, and oncological outcomes. A 1:1 nearest-neighbor matching method was applied with a caliper width of 0.2. Quantitative data are expressed as $\overline{x}\pm s$. For comparisons of continuous variables between the synchronous and combined groups after propensity score matching, independent-samples t-tests were used for normally distributed variables with homogeneous variance, while the Mann–Whitney U test was used for non-normally distributed variables. For comparisons among the three learning phases within the synchronous group, one-way analysis of variance (ANOVA) was used for normally distributed variables with homogeneous variance, whereas the Kruskal–Wallis test was applied for non-normally distributed variables or variables with unequal variance. When an overall difference was detected among the three phases, post hoc comparisons were performed using the initial phase as the reference group. Categorical data are presented as “number (percentage) [n (%)]”. Before intergroup comparisons, sample size (n) and expected frequency (T) were assessed: Pearson’s chi-square test was applied when n ≥ 40 and all expected frequencies T ≥ 5, reporting χ² values and P values. When T < 5, the Fisher’s exact test was used, reporting only the P-value without the χ² value. The CUSUM method was employed to analyze the learning curve of surgical duration, fitting the curve equation and calculating R². Survival curves were generated using the Kaplan–Meier method. Differences in survival between the synchronous and combined groups, as well as across learning phases within each group, were compared using the log-rank test. In this study, P < 0.05 was considered statistically significant.

Ethical review

The ethical review results of this study have been confirmed by the Medical Ethics Committee of Henan University, with approval number: HUSOM2023-348. All patients have completed the written informed consent process.

Results

Comparison of baseline characteristics between patient groups

A total of 229 patients were initially screened, including 118 patients in the synchronous group and 111 patients in the combined group. After 1:1 propensity score matching, 106 patients from each group were successfully matched and included in the final analysis, resulting in a matched cohort of 212 patients. Seventeen patients were excluded because no appropriate match could be identified within the predefined caliper width. The patient selection and matching process is illustrated in Fig. 1. After matching, baseline clinicopathological characteristics, comorbidity-related variables, and year of surgery were well balanced between the two groups (all P > 0.05; Table 1).

Table 1.

Comparison of baseline characteristics between the two patient groups

Variable	Before matching			After matching
	Synchronization group (n = 118)	Joint group (n = 111)	P	Synchronization group (n = 106)	Joint group (n = 106)	P
Gender (male/female)	78/40	70/41	0.631	76/30	68/38	0.239
Age (years,χ ± s, range)	61.68 ± 8.92	63.19 ± 3.14	0.524	60.45 ± 7.96	61.56 ± 9.42	0.412
Tumor location			0.956			0.730
Upper thoracic	12	10		12	9
Middle thoracic	59	56		54	53
Lower thoracic	47	45		40	44
Tumor stage			0.999			0.821
I	35	33		33	31
II	40	37		37	35
III	30	29		28	28
IVa	13	12		8	12
Tumor diameter (cm)			0.866			0.780
< 3	47	43		45	43
3–5	71	68		61	63
Preoperative neoadjuvant therapy			0.483			0.680
Yes	53	55		50	53
No	65	56		56	53
Diabetes mellitus			0.132			0.271
Yes	48	56		45	53
No	70	55		61	53
Hypertension			0.179			0.386
Yes	35	42		32	38
No	83	69		74	68
Coronary heart disease			0.095			0.191
Yes	27	37		25	34
No	91	74		81	72
Chronic pulmonary disease			0.176			0.394
Yes	24	31		21	26
No	94	80		85	80
Smoking history			0.169			0.240
Yes	73	78		67	75
No	45	33		39	31
Alcohol use			0.275			0.186
Yes	79	67		74	65
No	39	44		32	41
Year of surgery			0.083			0.568
2018	40	52		38	45
2019	45	39		40	38
2020	33	20		28	23

PSM covariates were selected based on clinical relevance to surgical approach selection, operative complexity, postoperative complications, and oncological outcomes. Caliper = 0.2 standard deviations; the standardized mean difference (SMD) of all covariates is < 0.1

Comparison of perioperative indicators, recurrence and metastasis between the two groups

The synchronous group demonstrated significantly superior outcomes compared to the combined group in terms of operative time, intraoperative blood loss, drainage volume at 3 days postoperatively, time to first flatus postoperatively, time to first ambulation postoperatively, and postoperative hospital stay, with all differences being statistically significant (P < 0.001). The synchronous group demonstrated a significantly higher total lymph node yield than the combined group (P < 0.001). Specifically, the number of harvested mediastinal lymph nodes was significantly greater in the synchronous group (P < 0.001), while the number of abdominal lymph nodes harvested was also slightly higher (P = 0.035), as shown in Table 2. Time to first flatus was also statistically shorter in the synchronous group; however, this variable was interpreted as an auxiliary indicator of gastrointestinal recovery rather than a standalone determinant of postoperative management. Long-term follow-up revealed that the synchronous group demonstrated significantly lower 5-year local recurrence rates (P = 0.013), 3-year distant metastasis rates (P = 0.032), and 5-year distant metastasis rates (P = 0.034) compared to the combined group (Table 3). Survival curve analysis further indicated a favorable survival trend in the synchronous group (P = 0.0138), as shown in Fig. 3.

Table 2.

Comparison of intraoperative conditions between the synchronous group and the combined group

Variable	Synchronization group (n = 106)	Joint group (n = 106)	t	P
Operation time (min)	115.28 ± 23.70	141.53 ± 18.34	−9.018	< 0.001
Blood loss (ml)	62.26 ± 11.27	84.34 ± 13.45	−12.955	< 0.001
Drainage volume at 3d (ml)	304.26 ± 14.23	340.45 ± 18.35	−16.046	< 0.001
Total lymph node yield	19.19 ± 3.46	15.95 ± 2.83	7.463	< 0.001
Mediastinal lymph nodes harvested	8.37 ± 2.11	5.91 ± 1.23	10.370	< 0.001
Abdominal lymph nodes harvested	10.82 ± 2.75	10.04 ± 2.61	2.118	0.035
Time to pass gas after surgery (h)	35.27 ± 4.98	40.18 ± 5.76	−6.639	< 0.001
Time to first ambulation (d)	2.62 ± 0.29	3.34 ± 1.15	−6.250	< 0.001
Postoperative hospital stays (d)	9.21 ± 1.78	11.56 ± 0.92	−12.075	< 0.001

Table 3.

Comparison of local recurrence and distant metastasis between the synchronous group and the combined group at 1–5 years post-surgery

Variable	Synchronization group (n = 106)	Joint group (n = 106)	χ2	P
Local recurrence (1y)	5	7	0.353	0.552
Local recurrence (3y)	13	19	1.325	0.250
Local recurrence (5y)	16	31	6.151	0.013
Distant metastasis (1y)	9	11	0.221	0.638
Distant metastasis (3y)	13	25	4.617	0.032
Distant metastasis (5y)	18	31	4.486	0.034

Fig. 3.

Kaplan–Meier curves for overall survival between the synchronous group and the combined group. Overall survival was compared using the log-rank test

Comparison of postoperative complications

Postoperative complications were systematically classified according to ECCG-defined categories, including anastomotic, pulmonary, cardiac, gastrointestinal, and chyle-related complications (Table 4).

Table 4.

Comparison of postoperative complications classified according to ECCG definitions

ECCG category	Specific complication	Synchronization group (n = 106)	Combined group (n = 106)	χ2	P
Anastomotic complications	Anastomotic leak	3	10	–	0.073
	Anastomotic stricture	6	14	–	0.096
Pulmonary complications	Pneumonia	17	31	5.278	0.022
	Pneumothorax	4	13	–	0.041
	Pleural effusion	9	17	2.806	0.094
Recurrent laryngeal nerve injury	Hoarseness	16	21	0.819	0.366
Cardiac complications	Arrhythmia	20	26	0.999	0.317
Gastrointestinal complications	Gastric emptying disorder	9	11	0.221	0.638
Chyle-related complications	Chylothorax	7	9	0.270	0.603
Total complication events		58	70	2.839	0.092
Patient-level Clavien–Dindo classification	None	7	16	3.950	0.047
	Clavien–Dindo I/II	86	73	4.252	0.039
	Clavien–Dindo III/IV/V	13	17	0.621	0.431

ECCG-defined complications are reported as complication events because one patient may have experienced more than one complication. Clavien–Dindo classification is reported at the patient level according to the highest severity grade for each patient

There was no significant difference in the total number of complication events between the synchronous and combined groups (P = 0.092). Anastomotic complications, including anastomotic leak and stricture, were infrequent in both groups, with no statistically significant differences observed (leak: P = 0.073; stricture: P = 0.096).

Pulmonary complications were significantly less frequent in the synchronous group compared with the combined group, particularly for pneumonia (P = 0.022) and pneumothorax (P = 0.041). No significant differences were observed in other complication categories, including recurrent laryngeal nerve injury, cardiac complications, gastrointestinal complications, and chyle-related complications (all P > 0.05).

At the patient level, complication severity was further evaluated using the Clavien–Dindo classification. The proportion of patients without complications was significantly lower in the synchronous group (P = 0.047), whereas the proportion of Clavien–Dindo grade I–II complications was higher (P = 0.039). However, there was no significant difference in severe complications (Clavien–Dindo grade III–V) between the two groups (P = 0.431).

Synchronous group learning curve analysis

The CUSUM learning curve plotted based on the operative time sequence showed excellent goodness-of-fit (R² = 0.987). The fitted equation was: Y = − 0.001X³ − 0.146X² + 30.865X − 80.301 (where X represents the case number). The curve exhibited two distinct inflection points at the 40th and 70th cases. Based on this, the learning process was divided into three phases: the initial phase (cases 1–40), the stabilization phase (cases 41–70), and the proficiency phase (cases 71–106). See Fig. 4A.

Fig. 4.

Synchronous group learning curve analysis. Note: A Scatter plot of operative times for synchronous laparoscopic esophageal cancer surgery with inflatable mediastinoscope, CUSUM learning curve, and fitted curve; B CUSUM learning curve and fitted curve during the initial phase (Phase I); C CUSUM curve and fitted curve during the steady-state phase (Phase II); D CUSUM curve and fitted curve for the proficient stage (Stage III)

Initial phase synchronization group learning curve analysis

During the initial phase, the CUSUM curve exhibited a significant upward trend, with the fitted equation being Y = 22.845X − 21.609 (where X represents the case number). The curve demonstrated excellent fit (R² = 0.997). The steep slope indicates that during this phase, the duration of each procedure was significantly higher than the group average, with cumulative time deviation continuously increasing. See Fig. 4B.

Analysis of learning curves for synchronized groups in the steady-state phase

During the stabilization y phase, the upward trend of the CUSUM curve markedly slowed and flattened, with the fitted equation being Y = 2.629X + 772.472 (where X represents case number). The curve exhibited excellent fit (R² = 0.996). The diminishing slope indicates that the surgical duration gradually approached and stabilized near the average level, with a significant reduction in the growth rate of cumulative time deviation. See Fig. 4C.

Analysis of learning curves for synchronized groups in the proficiency phase

The CUSUM curve during the proficiency phase shifted to a stable downward trend, with the fitted equation being Y = − 28.688X + 3013.373 (where X represents case number). This curve exhibited excellent fit (R² = 0.995). The negative slope indicates that the operative time per case in this phase has systematically fallen below the group average, with cumulative time deviation gradually narrowing. See Fig. 4D.

Combined group learning curve analysis

The CUSUM learning curve for the combined group was constructed using the same analytical approach as for the synchronous group (Fig. 5). The curve demonstrated a three-phase pattern with identifiable inflection points, indicating progressive improvement in operative efficiency over time. Compared with the synchronous group, the combined approach exhibited a relatively prolonged learning phase and a slower transition to proficiency.

Fig. 5.

Learning curve analysis of the combined inflatable mediastinoscope-assisted laparoscopic esophagectomy. A Scatter plot of operative time together with the cumulative sum (CUSUM) learning curve and fitted polynomial curve in the combined group; B CUSUM curve with linear regression during the initial phase (Phase I); C CUSUM curve with linear regression during the stabilization phase (Phase II); D CUSUM curve with linear regression during the proficiency phase (Phase III)

Comparison of outcomes across learning phases

Synchronous group

As the learning phase progressed in the synchronous group, operative time, intraoperative blood loss, postoperative 3-day drainage volume, time to first flatus, time to first ambulation, and postoperative hospital stay showed significant improvements (P < 0.05). The number of lymph nodes removed also increased with technical maturation (Table 5). In addition, the overall complication rate decreased from the initial phase to the proficiency phase, indicating improved procedural safety with accumulated surgical experience (Table 6). However, Kaplan–Meier survival analysis showed no statistically significant difference among the initial, stabilization, and proficiency phases in the synchronous group (P = 0.2886, Fig. 6A), suggesting that although technical proficiency improved perioperative outcomes, long-term survival remained relatively stable across different learning phases.

Table 5.

Comparison of intraoperative conditions among patients in three stages

Variable	Initial phase (n = 40)	Stable phase (n = 30)	Proficient phase (= 36)
Operation time (min)	137.27 ± 13.07	117.82 ± 1.54*	88.72 ± 14.43*
Blood loss (ml)	120.32 ± 35.28	92.48 ± 30.13*	75.89 ± 22.54*
Drainage volume at 3d (ml)	450.81 ± 125.78	345.29 ± 110.31*	285.37 ± 95.78*
Mediastinal lymph nodes harvested	6.28 ± 2.19	8.11 ± 2.31*	8.98 ± 2.45*
Time to pass gas after surgery (h)	38.12 ± 4.14	37.92 ± 4.09	36.05 ± 4.25*
Time to first ambulation (d)	3.27 ± 1.03	2.34 ± 0.97*	1.86 ± 0.63*
Postoperative hospital stays (d)	12.56 ± 3.05	10.12 ± 2.57*	8.24 ± 2.01*

Compared with the initial stage, *P < 0.05

Table 6.

Comparison of postoperative complications across learning phases classified according to ECCG definitions

ECCG Category	Specific complication	Initial phase (n = 40)	Stable phase (n = 30)	Proficient phase (n = 36)
Anastomotic complications	Anastomotic leak	2	1	0
	Anastomotic stricture	2	3	1
Pulmonary complications	Pneumonia	10	5	2*
	Pneumothorax	2	1	1
	Pleural effusion	5	4	1
Recurrent laryngeal nerve injury	Hoarseness	9	4	3
Cardiac complications	Arrhythmia	11	7	2*
Gastrointestinal complications	Gastric emptying disorder	5	3	1
Chyle-related complications	Chylothorax	4	2	1
Overall complication rate		29	20	9*
Clavien–Dindo classification	None	3	2	2
	Clavien–Dindo I/II	31	24	31
	Clavien–Dindo III/IV/V	6	4	3

Compared with the initial stage, *P < 0.05

Fig. 6.

Survival curves stratified by learning phases. A Synchronous group; B Combined group

Combined group

Survival analysis across different learning phases in the combined group is presented in Fig. 6B. Although the survival curve of the proficiency phase showed a trend toward improved outcomes compared with the initial phase, no statistically significant differences were observed among the three phases (P = 0.1598). These findings suggest that, unlike the synchronous approach, the impact of the learning process on long-term survival outcomes in the combined group may be less pronounced.

Discussion

In this retrospective cohort study, we compared perioperative outcomes, postoperative complications, survival outcomes, and learning curves between synchronous and combined inflatable mediastinoscope-assisted laparoscopic radical esophagectomy. After propensity score matching, baseline characteristics were well balanced between the two groups. The synchronous approach demonstrated shorter operative time, less intraoperative blood loss, reduced postoperative drainage volume, earlier postoperative recovery, a greater number of dissected lymph nodes, and shorter postoperative hospital stay compared with the combined approach. The increased lymph node yield was mainly attributable to improved mediastinal lymph node harvest rather than differences in abdominal lymphadenectomy, suggesting that the synchronous approach may provide better mediastinal exposure and operative coordination during dissection. Because both groups received the same standardized postoperative rehabilitation and early mobilization protocol, the earlier time to first ambulation in the synchronous group was unlikely to be attributable to differences in physical therapy management. Instead, it may reflect reduced perioperative physiological stress and faster early recovery associated with the synchronous approach. These findings suggest that the synchronous technique may improve operative efficiency and perioperative recovery while maintaining oncological adequacy, which is consistent with the general advantages reported for minimally invasive esophagectomy [3–5].

The major technical feature of the synchronous approach is its parallel operative workflow. Although the synchronous approach showed more favorable perioperative outcomes, the combined approach remains clinically relevant. It represents a mature sequential procedure and requires only one attending surgeon to complete the cervical and abdominal phases in sequence. Therefore, it may be more feasible in centers where two experienced surgeons or a stable dual-team simultaneous workflow are not consistently available. In this context, the combined approach may serve as a practical option during the early adoption of inflatable mediastinoscope-assisted laparoscopic esophagectomy or in institutions with limited personnel resources. In this procedure, the cervical mediastinoscopic team and abdominal laparoscopic team operate simultaneously, allowing mediastinal esophageal mobilization and abdominal gastric mobilization to proceed in parallel. This design reduces time loss caused by sequential operative steps, team repositioning, and repeated adjustment of instruments or operative fields. In contrast, the combined approach requires sequential completion of cervical and abdominal procedures, which may prolong total operative time and increase procedure-related physiological stress. Previous studies have demonstrated that prolonged operative duration and surgical complexity are associated with increased perioperative morbidity [16, 17], supporting the observed efficiency advantage of the synchronous approach in our study. Although the difference in time to first flatus between the two groups reached statistical significance, the absolute difference was approximately 5 h and should be interpreted cautiously in terms of clinical relevance. In our clinical practice, postoperative decision-making and discharge planning are based on comprehensive recovery parameters, including ambulation status, drainage condition, nutritional tolerance, imaging findings, absence of major complications, and overall clinical stability, rather than time to first flatus alone. Therefore, this variable was interpreted primarily as a supportive indicator of postoperative gastrointestinal recovery and part of the overall enhanced recovery trend observed in the synchronous group.

Postoperative complications were systematically assessed and categorized according to ECCG definitions, with Clavien–Dindo grading used to evaluate complication severity. The overall complication rate did not differ significantly between the two groups. However, pulmonary complications, especially pneumonia and pneumothorax, were significantly less frequent in the synchronous group. This finding is clinically meaningful because pulmonary morbidity is one of the most important determinants of postoperative recovery after esophagectomy [18]. The reduced pulmonary complication burden may be related to shorter operative time, reduced intraoperative stress, and decreased respiratory interference during the synchronous procedure. In contrast, anastomotic complications, including anastomotic leak and stricture, were infrequent in both groups and showed no statistically significant difference, which is consistent with previous reports indicating that anastomotic outcomes are primarily influenced by technical precision rather than operative sequence [19, 20]. These findings suggest that the perioperative advantage of the synchronous approach is mainly driven by improved procedural efficiency and reduced pulmonary morbidity rather than differences in anastomotic safety.

CUSUM analysis has been widely used to evaluate surgical learning curves and quantify technical proficiency. In this study, CUSUM analysis demonstrated a three-phase learning curve for the synchronous technique, with inflection points at the 40th and 70th cases [11–14]. Similar multi-phase learning patterns have been reported in minimally invasive esophagectomy and robotic-assisted procedures, supporting the validity of our findings. The initial phase reflected the period of technical exploration and team adaptation, while the stabilization and proficiency phases indicated progressive procedural standardization and improved team coordination.

To strengthen the comparative design, we additionally constructed a parallel CUSUM learning curve for the combined group using the same analytical approach. Compared with the combined approach, the synchronous technique showed a more efficient and stable learning trajectory. This finding suggests that the advantages of the synchronous approach are not solely attributable to accumulated surgical experience, but may also reflect intrinsic technical benefits related to workflow optimization. Because all procedures were performed by a fixed surgical team with clearly defined operative roles, the observed learning curve primarily reflects procedural adaptation rather than inter-surgeon variability, which is consistent with previous studies emphasizing the importance of standardized training pathways in complex esophageal surgery. After re-evaluation of the pathological records, total lymph node yield was recalculated to include both mediastinal and abdominal lymph nodes. The difference between groups was primarily observed in mediastinal lymph node harvest, whereas abdominal lymph node yield was relatively comparable. Because both groups underwent the same pathological processing and lymph node assessment protocol at our institution, the observed difference is unlikely to be attributable to specimen handling or pathological counting bias.

Learning-phase analysis further demonstrated that as surgical experience accumulated, perioperative outcomes improved significantly, including reductions in operative time, intraoperative blood loss, and postoperative recovery time, along with increased lymph node yield. These findings are in line with previous reports indicating that technical maturation is associated with improved surgical efficiency and oncological quality. In addition, the overall complication rate decreased from the initial phase to the proficiency phase, highlighting the importance of structured learning and team coordination in optimizing patient safety.

Regarding long-term outcomes, the synchronous group demonstrated favorable survival trends and lower rates of recurrence and distant metastasis. However, when stratified by learning phase, no statistically significant differences in survival were observed within either group. This suggests that while technical proficiency significantly improves perioperative outcomes, its impact on long-term survival may be limited. Previous studies have shown that long-term outcomes after esophagectomy are influenced by multiple factors, including tumor stage, lymph node status, and adjuvant therapy, which may explain the lack of significant differences across learning phases in this study.

This study has several limitations. First, this was a single-center retrospective study, and although propensity score matching was applied, residual selection bias cannot be completely excluded. In addition, although year of surgery was included in the adjustment model, residual temporal confounding may still persist, as improvements in anesthesia protocols, perioperative management, and institutional experience over time could have influenced postoperative outcomes. Second, although the primary surgeons and operative roles were fixed throughout the study period to minimize surgeon-related variability, the assisting surgeons and operating room nursing staff were not completely fixed. Therefore, the observed learning curve and perioperative outcomes may better reflect the standardized implementation of this technique within a specialized thoracic surgical unit rather than the performance of an entirely fixed operative team. This may also suggest a certain degree of reproducibility and practical applicability of the procedure under routine clinical conditions. However, further validation from multicenter studies is still warranted. Third, although complications were classified according to ECCG standards [15], retrospective data collection may still introduce reporting bias. Finally, long-term oncological outcomes require further validation in multicenter prospective studies.

In conclusion, within the context of this retrospective comparison with the sequential combined inflatable mediastinoscope-assisted laparoscopic approach, the synchronous technique was associated with shorter operative time, reduced intraoperative blood loss, faster postoperative recovery, fewer pulmonary complications, and a more favorable learning trajectory. These findings suggest that the synchronous approach is a feasible alternative minimally invasive strategy in appropriately selected patients. However, this study does not establish superiority over open esophagectomy or conventional VATS, and further multicenter prospective studies are needed to validate its broader safety and effectiveness.

Funding

This research was supported by the Key Scientific Research Project Guidance Program of Higher Education Institutions in Henan Province (26A320002; 26A320004). Henan Provincial Department of Science and Technology Research Project (262102310195; 262102310196; 262102310203).

Data availability

The data underlying this paper will be made available upon reasonable request to the corresponding author.

Declarations

Disclosures

Dr Haitao Wei, Nuo Xu, Zu-yuan Zhao, Xiapeng Li, Bao-li Hu, Jiao Wang, Li Li, and Qian Long have no conflicts of interest or financial ties to disclose.

Ethical approval

All procedures involving human participants were performed in adherence to the ethical standards of the Declaration of Helsinki and the approval of the Medical Ethics Committee of Henan University. This study was approved by the Medical Ethics Committee of Henan University (Approval Number: HUSOM2023-348) on June 25, 2025. All patients provided written informed consent prior to participation in the study, in compliance with the Declaration of Helsinki. The study was registered with the Chinese Clinical Trial Registry (Registration Number: ChiCTR2500104922).

Consent to participate

All participants included in this study provided written informed consent prior to their enrollment. The informed consent process included a detailed explanation of the study purpose, surgical procedures (synchronous or combined inflatable mediastinoscope-laparoscopic radical esophagectomy), potential risks, benefits, data collection, and follow-up requirements. Participants were informed of their right to withdraw from the study at any time without affecting their clinical care. For participants who were unable to provide consent independently due to physical or cognitive limitations, written informed consent was obtained from their legally authorized representatives. All consent procedures were conducted in accordance with the ethical principles outlined in the Declaration of Helsinki.

Artificial intelligence and AI-assisted technology

No AI-assisted tools were utilized by the authors for research, writing, or editing during the preparation of this manuscript.