Experience with H anastomosis and its perioperative management protocol in minimally invasive esophagectomy: a propensity score-matched comparison with stapled non-enhanced recovery after surgery pathway

Abstract

Background

H anastomosis is an emerging esophagogastric reconstruction technique that has shown promising perioperative results. However, real-world comparative evidence, particularly from rigorously matched cohorts, remains limited. Therefore, the present study aimed to compare short-term postoperative recovery metrics between the H anastomosis pathway and a stapled anastomosis pathway using propensity score matching (PSM).

Methods

This single-center retrospective study compared patients who underwent H anastomosis pathway between October 2024 and July 2025 (n=100) with those who received stapled anastomosis pathway between October 2023 and July 2024 (n=102). PSM was applied in a 1:1 ratio (75 matched pairs) based on age, sex, body mass index (BMI), and neoadjuvant therapy (NAT). Postoperative hospital stay (PHS) and postoperative fasting duration (PFD), modeled as overdispersed count variables, were analyzed using negative binomial regression. For low-frequency events—including severe pulmonary infection (SPI), reinsertion, and anastomotic leakage (AL)—Firth’s penalized logistic regression was used to reduce small-sample bias. Binary postoperative outcomes—including hypoalbuminemia, fever, high-grade fever, tachycardia, and atrial fibrillation (AF)—were evaluated using standard logistic regression.

Results

The H anastomosis pathway group demonstrated significantly shorter PHS and PFD (PHS: 10.27±4.19 vs. 17.29±6.95 days, P<0.001; PFD: 1.64±1.29 vs. 9.99±6.02 days, P<0.001) and lower incidences of postoperative fever, high-grade fever, AF, and hypoalbuminemia compared with the stapled anastomosis pathway group. These associations remained robust after adjustment for smoking and alcohol history and major comorbidities.

Conclusions

Although superiority cannot yet be claimed, our findings indicate that H anastomosis, when implemented together with its coordinated perioperative management protocol, provides a stable and feasible enhanced recovery after surgery (ERAS) oriented pathway for minimally invasive esophagectomy (MIE).

Highlight box.

Key findings

• In this propensity score matching analysis comparing H anastomosis pathway with a stapled anastomosis pathway, H anastomosis pathway showed smoother postoperative recovery and reduced tube-related burden.

• No definitive superiority can be claimed, but the integrated H anastomosis + enhanced recovery after surgery (ERAS) approach demonstrated feasibility and clinical promise.

What is known and what is new?

• H anastomosis has gained increasing interest, yet many centers remain cautious about applying ERAS elements such as tube-free and early feeding. ERAS after esophagectomy remains challenging due to concerns about reflux, leakage, and the inability to truly eliminate dependence on enteral nutrition.

• This study reports a structured, reproducible perioperative protocol tailored for H anastomosis. Real-world data show that the protocol can be safely implemented and may facilitate a more stable ERAS pathway.

What is the implication, and what should change now?

• These preliminary findings suggest that H anastomosis pathway may serve as a feasible platform for ERAS-oriented recovery after minimally invasive esophagectomy.

• Centers exploring ERAS in esophagectomy may consider this integrated approach, while further multi-center studies are needed to validate generalizability.

Introduction

Minimally invasive esophagectomy (MIE) with cervical esophagogastrostomy is the most widely adopted procedure for resectable esophageal cancer (1). Circular stapled side-to-end anastomosis with stapled gastric stump closure is popular for its simplicity and reproducibility (2), but offers less flexibility than hand-sewn techniques. Since late 2024, our center has introduced an emerging hand-sewn method—H anastomosis—developed at the originating center and subsequently adapted in our practice. This technique employs a layered, intussuscepted, end-to-end configuration (Figure 1). Prior technical publications have proposed that this intussuscepted design may reduce anastomotic leakage (AL), stricture, and reflux (3).

Figure 1.

Structure of H anastomosis. (A) Venous valve-like structure of the H anastomosis on CT scans. (B) Illustration of the venous valve-like structure of the H anastomosis. The arrows indicate the esophagogastric anastomosis. CT, computed tomography.

During the initial adoption of H anastomosis pathway at our institution, we learned from the originating center that thoracic gastric distention (TGD) can occur in the early postoperative phase. As clinical experience expanded, we further identified a second, previously uncharacterized event—delayed gastric emptying (DGE)—which presents a distinct pathophysiologic mechanism and requires a different management strategy. Recognition of these complications allowed refinement of perioperative monitoring and targeted interventions, enhancing the safety profile of the H anastomosis pathway.

Although TGD and DGE have been reported in the literature, we have not observed these events in patients undergoing conventional stapled anastomosis with routine postoperative gastric decompression. This likely reflects that gastric decompression masks the early manifestations of TGD and DGE. Their occurrence under the tube-free, early-feeding H anastomosis pathway highlights the distinct physiologic environment created by this integrated strategy.

Most comparative studies of anastomotic techniques have focused on AL incidence, while overlooking postoperative fasting duration (PFD) and the role of continuous gastric decompression, which may promote spontaneous closure of minor leaks (4,5). Thus, evaluating techniques that allow oral intake within 24 hours presents a fundamentally different challenge from those requiring a 7-day fasting regimen, rendering direct comparisons inherently inequitable (6,7). These considerations highlight the need for real-world comparative data evaluating outcomes of this enhanced recovery after surgery (ERAS) strategy relative to traditional stapled anastomosis pathway, acknowledging that the two approaches represent distinct therapeutic philosophies rather than simple technical variations. We present this article in accordance with the STROBE reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2589/rc).

Methods

Ethical approval for this retrospective study was obtained from the Ethics Committee of the Affiliated Huaian No. 1 People’s Hospital of Nanjing Medical University (No. KY-2025-185-01), individual consent for this analysis was waived due to the retrospective nature. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. We reviewed 102 consecutive patients undergoing MIE with stapled anastomosis pathway (Oct 2023–Jul 2024) and 100 patients receiving H anastomosis pathway (Oct 2024–Jul 2025). All procedures were performed by the same surgical team at a single center to minimize confounding factors. The study periods were selected to cover corresponding calendar months in consecutive years to ensure temporal consistency between cohorts. All postoperative outcomes were obtained during the index hospitalization. Patients with AL were followed until they resumed oral intake. Data collection was performed after patient discharge as part of the initial study setup. We then conducted a retrospective comparison of the two cohorts using PSM and appropriate regression models for postoperative outcomes.

Participants and groupings

Stapled anastomosis pathway (non-ERAS) group

A total of 102 consecutive patients who underwent MIE with circular stapled esophagogastric anastomosis via the posterior mediastinal (esophageal bed) route between October 2023 and July 2024 were included. A disposable circular stapler (CDH21A or CDH25A; Johnson & Johnson) was used in all operations. In our practice, CDH25A is preferentially selected whenever feasible, whereas CDH21A is used in patients with a relatively small esophageal diameter based on intraoperative assessment. In all patients, reinforcing sutures were routinely placed between the muscular layers of the esophagus and the stomach after completion of the circular stapled anastomosis. Intraoperatively, an 18-Fr nasogastric tube (NGT) and a 10Fr nasojejunal feeding tube (NJT) were placed for gastric decompression and enteral nutrition.

Oral intake was initiated when all of the following conditions were met: fasting duration >7 days, no leukocytosis, serum albumin >30 g/L, absence of fever, and a normal upper gastrointestinal contrast study (UGICS) (8). Once oral intake was resumed, patients followed a standardized progression of water for 1 day, clear soup for 3 days, soft porridge for approximately 20 days, and then gradual advancement to a regular diet. The NJT remained in place after discharge and was removed only when enteral nutrition was no longer required.

Patients were discharged when they had no pneumonia, pneumothorax, pleural effusion, or fever. For patients permitted to resume oral intake, discharge was allowed after tolerating soft porridge for 2 days, with the NJT retained as a precaution. Those with suspected AL remained fasting and were discharged with the NJT in place, with oral intake postponed until the leakage risk resolved after discharge.

H anastomosis pathway (ERAS) group

After completion of team training, H anastomosis was adopted as the default reconstructive technique for consecutive eligible patients during the study period, and no technique-specific patient selection criteria were applied. A total of 100 participants who underwent MIE with circular H anastomosis via the posterior mediastinal (esophageal bed) route between October 2024 and July 2025 were included. The procedure was combined with pyloroplasty, and no NGT or NJT was routinely placed during surgery.

Briefly, the H anastomosis is constructed under direct visualization. After resection of the cervical esophageal muscular layer and the gastric seromuscular layer to expose the mucosa, a continuous end-to-end mucosa-to-mucosa anastomosis is performed, forming a mucosal bridge. This mucosal bridge is then gently advanced into the gastric lumen to achieve an intussuscepted configuration, followed by interrupted suturing of the remaining muscular and seromuscular layers to complete the anastomosis (9). Detailed step-by-step technical descriptions have been reported previously and are therefore not reiterated here. Owing to its layered and intussuscepted design, the procedure typically requires an additional 20–30 minutes compared with a stapled technique.

On postoperative day 1, all participants underwent UGICS and chest computed tomography (CT). If the pylorus failed to open on the initial fluoroscopic assessment, the patient was instructed to remain standing for 10 minutes before a repeat evaluation. DGE was diagnosed when iodixanol did not pass into the duodenum on both assessments. TGD was diagnosed when the CT scan demonstrated a maximal cross-sectional diameter of the conduit exceeding twice its intraoperative width (approximately two fingers), as assessed visually by the operating surgeon rather than by quantitative intraoperative measurement. Representative imaging findings are presented in Figures 2,3.

Figure 2.

Delayed gastric emptying. (A) Delayed gastric emptying observed on upper gastrointestinal contrast study. (B) Absence of delayed gastric emptying on upper gastrointestinal contrast study. The arrows indicate the location of the pylorus.

Figure 3.

Thoracic gastric distention. (A,B) Thoracic gastric distention observed on computed tomography scans. (C,D) Absence of thoracic gastric distention on computed tomography scans. The arrows indicate the gastric conduit.

Oral intake was initiated when UGICS showed no DGE and CT revealed no TGD. Participants with DGE were managed with an 18Fr NGT for gastric decompression, and UGICS was repeated 3–5 days later (10,11). Participants with TGD were managed using a “chimney” gas drainage technique: a 12-Fr NGT was inserted 30 cm and left open to air, after which oral intake was resumed. The management flowcharts are presented in Figure 4. The 12-Fr NGT was used solely for shallow gas drainage rather than fluid aspiration, so the risk of clogging was negligible, and patient discomfort was minimized, consistent with ERAS principles. For participants without DGE or TGD, oral intake followed a structured regimen: clear soup for 3 days, soft porridge for approximately 20 days, and then gradual advancement to a regular diet.

Figure 4.

Flowchart of management and “chimney” gas drainage method. (A) Flowchart of perioperative management. (B) “Chimney” gas drainage method for thoracic gastric distention.

Participants were discharged after chest CT on postoperative days 5–7 confirmed absence of pulmonary infection, pneumothorax, pleural effusion, and fever, and if they could tolerate soft porridge for 2 consecutive days with oral intake sufficient to meet daily nutritional requirements. No tubes were required after discharge.

The perioperative management of H anastomosis was implemented as an integrated safety pathway rather than as a set of separable interventions. Components including pyloroplasty, tube strategy, postoperative day-one imaging, dietary progression, and discharge criteria were systematically applied to all patients in the cohort, based on the unique physiological and biomechanical characteristics of H anastomosis. These measures were defined a priori to mitigate specific risks inherent to the technique and to enable its safe implementation within an ERAS framework.

Because these components were uniformly applied for patient safety, no exposure variability existed for these factors within the cohort. Consequently, formal staged adjustment, mediation analysis, or sensitivity analyses targeting individual protocol elements were not statistically feasible.

For clarity and consistency, throughout the manuscript, the term “H anastomosis pathway” is used to denote the combination of the H anastomosis technique and its standardized perioperative management protocol, including pyloroplasty, tube strategy, postoperative imaging, dietary management, and discharge criteria. When referring exclusively to the surgical construct itself, we explicitly use the term “H anastomosis”.

Comparison between the two groups

A comparison organized according to the treatment timeline was conducted, and all perioperative management details for the two groups are summarized in Table 1. Descriptive statistics before PSM are shown in Table 2.

Table 1. Comparison between the stapled anastomosis group and the H anastomosis group.

Category	Stapled anastomosis	H anastomosis
1. Intraoperative handling
Pyloroplasty	Not required	Required
Nasogastric tube	Required	Not required
Nasojejunal feeding tubes	Required	Not required
2. Postoperative resumption of oral intake
Fasting	7 days	1 day
Serum albumin >30 g/L	Required	Not required
No fever	Required	Not required
White blood cell counts	Normal	Not required
UGICS shows no AL	Required	Required
DGE	Not involved	Gastrointestinal decompression
TGD	Not involved	“Chimney” gas drainage method
3. Oral intake progression
Water	1 day	Not required
Clear soup	3 days	3 days
Soft porridge	20 days	20 days
Transition to regular diet	Yes	Yes
4. Discharge criteria
No pneumonia	Required	Required
No pneumothorax	Required	Required
No pleural effusion	Required	Required
Oral intake	Not mandatory	Meet daily nutritional requirements
Nasojejunal feeding tubes	Maintained	No need
5. Management of AL
Fasting	Required	Required
Gastrointestinal decompression	Required	Required
Nasojejunal feeding tubes	Required	Required

AL, anastomotic leakage; DGE, delayed gastric emptying; TGD, thoracic gastric distention; UGICS, upper gastrointestinal contrast study.

Table 2. Descriptive characteristics and postoperative events of all participants before PSM.

Variables	H anastomosis (n=100)	Stapled anastomosis (n=102)	P value
Age (years)	70.53±6.86	68.67±6.85	0.055
Sex (male: female)	70:30	64:38	0.35
BMI (kg/m2)	23.32±3.10	23.65±3.01	0.45
NAT (y: n)	8:92	11:91	0.66
Smoking history (y: n)	20:80	29:73	0.22
Alcohol history (y: n)	18:82	24:78	0.43
COPD (y: n)	14:86	16:86	0.89
Hypertension (y: n)	18:82	27:75	0.20
Diabetes (y: n)	8:92	11:91	0.66
CHD (y: n)	19:81	15:87	0.53
IPT (y: n)	8:92	5:97	0.54
PHS (days)	10.90±6.16	17.48±7.76	<0.001
PFD (days)	1.56±1.17	10.26±6.56	<0.001
Hypoalbuminemia (y: n)	40:60	50:52	0.25
Fever (y: n)	30:70	75:27	<0.001
High-grade fever (y: n)	12:88	47:55	<0.001
SPI (y: n)	3:97	6:96	0.52
Reinsertion (y: n)	1:99	4:98	0.38
AL (y: n)	2:98	1:101	0.99
AF (y: n)	5:95	13:89	0.09
Tachycardia (y: n)	10:90	20:82	0.09
DGE (y: n)	6:94	–	–
TGD (y: n)	27:73	–	–

Data are presented as mean ± standard deviation or n. P values in this table are provided for descriptive purposes only and do not represent effect estimates. AF, atrial fibrillation; AL, anastomotic leakage; BMI, body mass index; CHD, coronary heart disease; COPD, chronic obstructive pulmonary disease; DGE, delayed gastric emptying; IPT, iatrogenic pneumothorax; n, no; NAT, neoadjuvant therapy; PFD, postoperative fasting duration; PHS, postoperative hospital stay; PSM, propensity score matching; SPI, severe pulmonary infection; TGD, thoracic gastric distention; y, yes.

Although the two cohorts differed substantially in their perioperative management strategies, the approach to AL was identical in both groups. For participants with confirmed AL, an 18-Fr NGT and a 10-Fr NJT were placed for gastric decompression and enteral nutrition, oral intake was withheld. Once the leak-related local infection was controlled, participants were discharged. Oral intake was resumed after discharge only when the risk of leakage was confirmed to be absent.

Variable definitions

The independent variable was group assignment (H vs. stapled anastomosis pathway). Covariates included age, sex, body mass index (BMI), neoadjuvant therapy (NAT), smoking, alcohol use, chronic obstructive pulmonary disease (COPD), hypertension, diabetes, coronary heart disease (CHD), and iatrogenic pneumothorax (IPT). We provided the postoperative pathological stage for each patient in table online https://cdn.amegroups.cn/static/public/jtd-2025-1-2589-1.zip. However, pathological stage was not included as a covariate in the PSM because it reflects the biological characteristics of the tumor and is not directly associated with the choice or performance of the anastomotic technique.

Dependent variables were postoperative hospital stay (PHS), PFD, hypoalbuminemia, fever (>37.2 ℃), high-grade fever (>38.5 ℃), sinus tachycardia, new-onset atrial fibrillation (AF), severe pulmonary infection (SPI), reinsertion, and AL. The only Clavien-Dindo ≥ III complications were reinsertion due to sputum retention causing respiratory failure, occurring in 1/98 patients in the H anastomosis pathway group and 4/99 patients in the stapled anastomosis pathway group. All ≥ III complications were appropriately managed. Other postoperative complications, all classified as Clavien-Dindo I–II, included AL (2 patients in the H anastomosis pathway group and 1 patient in the stapled anastomosis pathway group), hypoalbuminemia, fever, high-grade fever, SPI, AF and tachycardia. All I–II complications were managed conservatively with medications or supportive therapy. No patients in either group experienced 30-day readmission or mortality.

All variables were complete. PHS and PFD were non-normally distributed; thus, negative binomial regression was applied.

Statistical analysis

The study size was determined by including all consecutive patients who met the eligibility criteria during the specified study periods. Given the retrospective design and exploratory aim of comparing postoperative recovery patterns between the two anastomotic techniques, a formal sample size calculation was not required; instead, we used the full available cohort to maximize statistical power and generalizability.

Univariate analyses informed covariate selection for PSM, guided by clinical relevance. Stapled anastomosis pathway (102 cases) were 1:1 matched with H anastomosis pathway (100 cases) using nearest-neighbor matching (caliper 0.1) based on age, sex, BMI, and NAT. Covariate balance was assessed with a Love plot. Post-matching, variance inflation factors (VIFs) were calculated. Negative binomial regression was used for continuous non-normal outcomes; Firth’s logistic regression for rare binary outcomes (12); and standard logistic regression for common binary outcomes. Similar results were observed when using an alternative caliper width of 0.2 in PSM (data not shown).

Variables that were systematically applied to all patients as part of the predefined H anastomosis pathway (e.g., pyloroplasty and other pathway-mandated components) were not modeled as covariates, as the absence of exposure variation results in structural non-positivity and renders effect estimation statistically undefined.

Analyses were performed in RStudio (Version 1.4.1106) with packages MatchIt, cobalt, car, MASS, ggplot2, ggpubr, logistf, and broom. A two-sided P<0.05 was considered significant. The R code and original dataset are provided in table online https://cdn.amegroups.cn/static/public/jtd-2025-1-2589-1.zip.

Results

PSM matching results

To identify appropriate covariates for matching, univariate analyses were initially conducted. None of the covariates demonstrated a statistically significant association with the grouping variable. Nevertheless, based on clinical importance and their potential influence on postoperative outcomes, we selected four key covariates—age, sex, BMI, and NAT—for inclusion in the PSM process (13,14). The propensity scores were calculated by setting group as the dependent variable and age, sex, BMI, and NAT as covariates.

Prior to matching, notable imbalances were observed between groups. The standardized mean difference (SMD) for age was −0.2722, indicating that participants in the stapled anastomosis pathway group were generally younger than those in the H anastomosis pathway group (mean age: 68.7 vs. 70.5 years). For sex, the SMD was −0.1501, reflecting a slightly lower proportion of males in the stapled group (62.75% vs. 70.00%). SMDs for BMI and NAT were 0.1081 and 0.0898, respectively, indicating smaller but non-negligible imbalances. The overall propensity score SMD was 0.3505, with a maximum empirical cumulative distribution difference of 0.1933.

After matching, 75 matched pairs were retained, with no unmatched cases excluded. Post-matching assessments showed a marked improvement in covariate balance. All covariates achieved excellent balance: age (−0.0136), sex (−0.0276), BMI (0.0186), NAT (0.0000); propensity score SMD decreased to 0.0326, and the maximum cumulative difference dropped to 0.0667.

These results indicate successful balancing of baseline characteristics, supporting robust comparison of postoperative outcomes. Covariate balance before and after matching is illustrated in the Love plot (Figure 5A).

Figure 5.

Results after propensity score matching. (A) Love plot showing covariate balance before and after propensity score matching. (B) Violin plots comparing postoperative hospital stay and postoperative fasting duration between groups. (C) Forest plot comparing clinical outcomes between H anastomosis pathway and stapled anastomosis pathway. *, continuous variables. BMI, body mass index; OR, odds ratio; RR, risk ratio.

Multicollinearity assessment of covariates

All 11 covariates were assessed for multicollinearity using a generalized linear model with group assignment as the outcome to calculate VIFs. Moderate collinearity was observed between smoking and alcohol history (VIFs 8.05 and 7.99), but these variables were retained, as the collinearity was not considered sufficient to compromise model performance (15). This collinearity is clinically interpretable, as individuals who tend to smoke also tend to consume alcohol (16). The remaining covariates showed VIFs <2 (age and sex 1.25; NAT 1.10; BMI 1.14; COPD 1.14; hypertension 1.11; diabetes 1.09; CHD 1.13; IPT 1.05), indicating no substantial multicollinearity.

Regression analysis

All models were adjusted for the same set of covariates, including age, sex, BMI, NAT, smoking history, alcohol history, COPD, hypertension, diabetes, CHD, and IPT. Multivariable regression analyses of postoperative outcomes, as shown in Table 3, indicate that H anastomosis pathway is associated with faster recovery across multiple dimensions, even after adjusting for relevant clinical covariates.

Table 3. Multivariable regression analyses of postoperative outcomes after PSM.

Outcome	Model type	Predictor	OR/RR (95% CI)	P value
PHS	Negative binomial	Stapled anastomosis	1.72 (1.53–1.93)	<0.001
	Negative binomial	Age (per year)	1.01 (1.00–1.02)	0.007
	Negative binomial	IPT	1.51 (1.19–1.86)	<0.001
PFD	Negative binomial	Stapled anastomosis	5.95 (4.58–7.73)	<0.001
	Negative binomial	IPT	1.66 (1.18–2.33)	0.005
SPI	Firth’s logistic	Stapled anastomosis	1.61 (0.20–18.45)	0.65
Reinsertion	Firth’s logistic	Stapled anastomosis	4.87 (0.47–568.8)	0.20
AL	Firth’s logistic	Stapled anastomosis	1.08 (0.10–11.93)	0.94
Hypoalbuminemia	Standard logistic	Stapled anastomosis	2.20 (1.02–4.78)	0.045
	Standard logistic	Age (per year)	1.12 (1.05–1.20)	<0.001
	Standard logistic	BMI (per kg/m²)	0.78 (0.67–0.90)	<0.001
	Standard logistic	Sex (male)	3.58 (1.45–9.16)	0.006
	Standard logistic	Diabetes	5.70 (1.32–24.6)	0.02
Fever	Standard logistic	Stapled anastomosis	7.00 (3.21–15.25)	<0.001
High-grade fever	Standard logistic	Stapled anastomosis	8.17 (3.18–20.94)	<0.001
AF	Standard logistic	Stapled anastomosis	6.65 (1.36–32.55)	0.03
	Standard logistic	BMI (per kg/m²)	1.75 (1.24–2.46)	0.002
Tachycardia	Standard logistic	Sex (male)	0.30 (0.10–0.89)	0.03

Anastomosis type was coded as 0= H anastomosis, 1= stapled anastomosis. All ORs/RRs represent adjusted estimates performed in multivariable regression analyses in the PSM cohort. AF, atrial fibrillation; AL, anastomotic leakage; BMI, body mass index; IPT, iatrogenic pneumothorax; OR, odds ratio; PFD, postoperative fasting duration; PHS, postoperative hospital stay; PSM, propensity score matching; RR, risk ratio; SPI, severe pulmonary infection.

Negative binomial regression

Negative binomial regression models were used to assess the association between anastomotic technique and two continuous postoperative outcomes: PHS and PFD.

The anastomotic technique (group) was a significant independent predictor in both models. Compared to the H anastomosis pathway group, participants in the stapled anastomosis pathway group had significantly longer PHS and PFD, as illustrated in the violin plots (Figure 5B). Marked differences in PHS and PFD were observed between the two groups (PHS: 10.27±4.19 vs. 17.29±6.95 days; PFD: 1.64±1.29 vs. 9.99±6.02 days), based on crude estimates after PSM (detailed data for all primary endpoints are provided in Table S1). IPT was significantly associated with delayed recovery, corresponding to a 51% increase in PHS and a 66% increase in PFD. Age also showed a modest but statistically significant association with prolonged PHS (P=0.007) and a borderline association with PFD (P=0.06).

The models demonstrated good fit, with dispersion parameters (θ) of 25.44 for PHS and 16.05 for PFD, and AIC values of 876.13 and 659.77, respectively.

Firth’s logistic regression

Firth’s penalized logistic regression was applied to assess the associations between anastomotic technique and three rare postoperative complications: SPI, reinsertion due to mucus plug-induced asphyxia, and AL, adjusting for the same covariates. No statistically significant association was observed between groups.

Standard logistic regression

Standard logistic regression was performed to evaluate the impact of anastomotic technique on a range of binary postoperative outcomes, adjusting for the same covariates.

The anastomotic technique (group) was a significant independent predictor for several outcomes. Compared to participants in the H anastomosis pathway group, those in the stapled anastomosis pathway group had higher odds of hypoalbuminemia, fever, high-grade fever, and AF.

Additional findings: older age, lower BMI, male sex, and diabetes were independently associated with hypoalbuminemia. Female sex was associated with an increased risk of postoperative tachycardia. Higher BMI was independently associated with increased odds of postoperative AF.

A forest plot summarizing effect estimates (logOR or logRR) for different postoperative outcomes relative to the anastomotic technique is shown in Figure 5C.

Discussion

In the context of MIE, tube-free and early-feeding management have traditionally been viewed as a relatively aggressive ERAS strategy. We previously attempted to implement ERAS in patients undergoing conventional stapled anastomosis without pyloroplasty. Although both the NGT and NJT were omitted, a transabdominal feeding jejunostomy was additionally placed.

The attempt proved unsuccessful, as the risk of AL and reflux-related aspiration increased. Oral intake could not meet the patients’ nutritional requirements, resulting in a pseudo-ERAS pathway. Moreover, the transabdominal feeding jejunostomy itself caused patient discomfort and introduced additional potential risks, such as bowel obstruction and intra-abdominal infection. As a result, we were compelled to revert to the non-ERAS strategy that retained both the NGT and NJT.

H anastomosis is a recently developed hand-sewn technique with both advantages and inherent risks. Its accompanying management pathway is specifically tailored to this procedure. To ensure proper execution of ERAS, it is essential to thoroughly understand and strictly adhere to these protocols. In this context, we adopted a patient-centered and individualized management strategy. Although a tube-free approach was applied in the majority of patients undergoing H anastomosis, NGT placement was not withheld when clear clinical indications were present, in order to ensure postoperative safety.

Inherent risks and management strategies

DGE and TGD are generally considered transient postoperative gastrointestinal dysfunctions, associated with vagal nerve disruption and gastrointestinal reconstruction (10,17). In addition, the anti-reflux design of H anastomosis—which is absent in stapled anastomosis—limits belching and may thereby increase the severity of TGD. Such events are uncommon in patients undergoing stapled anastomosis with routine gastric decompression, which has been shown to reduce the risk of AL (5). Nonetheless, NGT-free management with early oral feeding inherently increases the risk of DGE and TGD, necessitating proactive monitoring.

In this study, we consider DGE and TGD as partially independent yet interrelated dysfunctions: patients with TGD do not necessarily develop DGE, whereas most patients with DGE exhibit some degree of TGD; a few patients present with DGE without TGD initially, but subsequent oral intake may trigger TGD.

Management strategies differ between the two dysfunctions. DGE, regardless of coexisting TGD, requires fasting with gastric decompression to prevent AL or gastric rupture. To reduce DGE incidence, all participants underwent pyloroplasty (18), and early postoperative UGICS was performed to assess suitability for oral intake. Patients diagnosed with DGE received repeat UGICS to guide the timing of early feeding (11). In our cohort, approximately 6% of H anastomosis pathway patients experienced DGE, which all resolved within 3–7 days.

TGD occurred in approximately 27% of patients. Isolated TGD could often be relieved with a simple “chimney” gas drainage method, allowing continued oral intake without fasting. Although TGD may persist for more than a week, it generally resolves within one month, and removal of the chimney after 5 days has been shown to be safe.

During the initial adoption phase of H anastomosis pathway, two cases of AL were observed, both associated with concurrent DGE and TGD that were not managed according to the later-established protocol. This suggests that these complications may directly contribute to anastomotic failure (6). Following implementation of targeted management—including early imaging assessment and tailored interventions—no further leaks occurred.

With proactive identification and management of DGE and TGD, H anastomosis pathway demonstrated favorable anastomotic integrity compared to stapled anastomosis pathway. Its layered, intussuscepted, end-to-end design, achieved through meticulous hand-sewing, may offer greater structural reliability to support early oral feeding—something not readily achievable with stapled anastomosis (3,19).

Advantages of H anastomosis pathway

Using rigorous PSM, we compared the postoperative outcomes of H anastomosis pathway and stapled anastomosis pathway. Our preliminary results suggest that H anastomosis pathway may shorten PHS and PFD. While no statistically significant differences were observed in the rates of SPI, reinsertion, or AL, H anastomosis pathway was associated with a significantly lower incidence of hypoalbuminemia, fever, high-grade fever, and AF. Several findings warrant further explanation:

1. IPT was identified as a factor associated with prolonged PHS and PFD (20). This finding is clinically plausible, as the persistence of pneumothorax—also a characteristic feature of AL—and the resulting imbalance in pleural pressure may reduce clinicians’ confidence in initiating early oral intake (21).

2. Reduced incidence of hypoalbuminemia in the H anastomosis pathway group may be attributable to early oral feeding and pyloroplasty (7). These patients were able to transition to soft porridge earlier, which contains solid components of grains, vegetables, and meat, and therefore provides superior nutritional value compared with clear soup administered via NJT. Some clinicians have expressed concern that omitting the NJT may increase the risk of nutritional insufficiency after discharge (22). However, in our experience, routine pyloroplasty together with structured dietary counseling has reliably supported adequate oral intake, enabling patients to maintain satisfactory nutrition without NJT.

3. H anastomosis exhibits a favorable anti-reflux effect. In our previous Bama mini-pig study, its anti-reflux pressure was 9 cmH₂O (3). A postoperative UGICS video in H anastomosis patients (head-down, feet-up position) showed an anastomotic anti-reflux angulation of −30° (Video 1), indicating that the mechanical effect observed in Bama mini-pig translates to functional performance in patients. Lower rates of fever and high-grade fever may be attributed to the anti-reflux design of the H anastomosis. This approach likely reduces reflux of digestive fluids into the oropharynx and helps maintain a cleaner oral environment. It should be noted that this is currently a hypothesis, as no objective reflux measurements or patient-reported reflux scores were collected in this study. An ongoing prospective clinical study has been initiated to further investigate this hypothesis, in which oral and airway secretions are being systematically collected for pepsinogen measurement and microbial DNA sequencing. Previous research has shown that improved oral hygiene reduces the incidence of pulmonary infections, and the H anastomosis may offer a similar protective effect (23).

4. Stapled anastomosis pathway and high BMI were identified as risk factors for AF. This may be related to increased inflammatory responses, a greater tendency for fluid-electrolyte imbalance during fasting, sympathetic activation induced by indwelling tubes, and elevated cardiac workload in participants with higher BMI (24). It is noteworthy that postoperative AF is widely regarded as a marker of postoperative morbidity and mortality (25).

Video 1.

Download video file ^{(9.2MB, mp4)}

Postoperative upper gastrointestinal contrast study demonstrating the anti-reflux effect of H anastomosis.

In this PSM comparison, we evaluated H anastomosis pathway against conventional stapled anastomosis pathway. While the current evidence does not allow us to claim the superiority of either approach, our findings indicate that the H anastomosis pathway is a stable, reproducible, and practically implementable ERAS-oriented strategy for MIE. These results should be interpreted within the context of the integrated perioperative pathway specifically developed for H anastomosis, and may not be directly generalizable to settings in which individual components are modified or omitted.

Limitations and future directions

Several limitations should be acknowledged. Potential effects related to calendar time and learning curve cannot be fully excluded, such bias would be likely to favor stapled anastomosis pathway and underestimate the performance of the H anastomosis pathway. TGD was diagnosed visually, which may be subjective, while DGE was assessed by standardized imaging with minimal inter-observer variability. The sample size was insufficient to reliably assess rare complications. Benefits related to reflux reduction and oropharyngeal hygiene were not directly measured. PSM cannot fully eliminate confounding, and post-matching regression adjustment may not completely compensate for unmatched variables. Randomization was challenging because blinding was impossible and patients were reluctant to be allocated to the conventional stapled anastomosis pathway. Nevertheless, randomized trials remain the optimal design for future validation. In addition, long-term outcomes were not evaluated because this study focused on early postoperative recovery.

Conclusions

This preliminary PSM analysis suggests that H anastomosis pathway may offer a stable and feasible ERAS-oriented pathway for MIE. While these early findings are encouraging, they remain exploratory, and further prospective studies are needed to confirm the safety, effectiveness, and generalizability of this approach.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. Ethical approval for this retrospective study was obtained from the Ethics Committee of the Affiliated Huaian No. 1 People’s Hospital of Nanjing Medical University (No. KY-2025-185-01), individual consent for this analysis was waived due to the retrospective nature. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Data Sharing Statement

Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2589/dss

jtd-18-03-224-dss.pdf ^{(81.2KB, pdf)}

DOI: 10.21037/jtd-2025-1-2589