Fusion surgery in robot-assisted esophagectomy: stepwise learning curves across the da Vinci™ and hinotori™ platforms

Abstract

Background

Robot-assisted minimally invasive esophagectomy (RAMIE) provides superior precision, three-dimensional visualization, and enhanced instrument dexterity compared to conventional approaches. The “Fusion Surgery” concept, which emphasizes dynamic cooperation between console and bedside surgeons, has the potential to improve operative efficiency and facilitate technical skill acquisition. This study used stepwise learning curves in RAMIE to explore the feasibility and initial experience with the da Vinci™ and hinotori™ robotic platforms.

Methods

We retrospectively analyzed 30 patients with esophageal cancer who underwent RAMIE using the Fusion Surgery approach between June 2024 and November 2025. The first 25 consecutive procedures were performed with the da Vinci system, followed by five with the hinotori system, all by the same surgical team. Thoracic lymphadenectomy was standardized into lower, middle, and upper mediastinal segments. Learning curves were assessed using cumulative sum (CUSUM) analysis of console operation time. Intraoperative blood loss and postoperative hospital stay were evaluated as secondary outcomes.

Results

The median total console operation time was 210 min. Middle mediastinal dissection time was significantly shorter with the hinotori system than with the da Vinci system (46 vs. 76 min, p = 0.010), whereas upper and lower mediastinal dissections showed favorable trends. CUSUM analysis demonstrated progressive improvement across all mediastinal levels. Upper mediastinal console time continued to decrease after transition to the hinotori system, middle mediastinal dissection showed marked improvement following the transition, and lower mediastinal dissection exhibited a steady downward trend from the initial cases. These findings indicate cumulative learning and effective transfer of technical skills across robotic platforms.

Conclusions

The Fusion Surgery approach may support gradual improvement in RAMIE performance across all mediastinal levels. Transitioning from the da Vinci to the hinotori robotic system did not interrupt the learning curve, supporting smooth continuation of surgical proficiency on a new platform. Fusion Surgery may provide a robust and adaptable framework for sustained technical advancement in RAMIE.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12957-026-04284-9.

Introduction

Robotic-assisted minimally invasive esophagectomy (RAMIE) has gained widespread adoption for esophageal cancer, offering enhanced precision, three-dimensional visualization, and instrument dexterity compared with conventional minimally invasive approaches [1–5]. The “Fusion Surgery” concept, in which console and bedside surgeons actively collaborate, improves operative efficiency and facilitates surgical education [6, 7]. In conventional robotic procedures, the bedside surgeon often assumes a limited role, primarily providing basic assistance, which may constrain both operative fluidity and learning opportunities. Fusion Surgery addresses these limitations by enabling dynamic cooperation between console and bedside surgeons during critical procedural steps, fostering real-time efficiency and mutual skill development [7, 8].

The hinotori™ Surgical Robot System (hinotori; Medicaroid Corporation, Kobe, Japan) is Japan’s first domestically developed surgical robot. Its flexible design and docking-free trocars allow precise dissection and radical lymphadenectomy during esophagectomy without arm collisions [9]. Noshiro et al. reported the safe introduction of the hinotori in an institution already experienced with the da Vinci™ Surgical System (da Vinci; Intuitive Surgical, Sunnyvale, CA, USA) [10]. In their series of 22 McKeown esophagectomies performed with the hinotori, perioperative outcomes—including operative times, complication rates, and short-term recovery—were comparable to historical da Vinci cases, thus demonstrating the system’s feasibility and safety.

In this study, we aimed to present the technical concept of Fusion Surgery and to demonstrate how this standardized approach enables a consistent RAMIE workflow during the transition from the da Vinci to the hinotori system. We also evaluated the learning process using cumulative sum (CUSUM) analysis to determine whether continuity of the learning curve—reflecting operative performance and team coordination—was maintained across different robotic platforms when a unified team concept and surgical workflow were systematically applied.

Methods

Study design and patients

This retrospective, single-center study included 30 patients with esophageal cancer who underwent RAMIE using the Fusion Surgery concept between June 2024 and November 2025. Specifically, 25 consecutive patients initially underwent curative esophagectomy with the da Vinci, followed by five consecutive patients who underwent the same procedure using the hinotori. In this study, all consecutive patients who underwent RAMIE during the study period were included, regardless of tumor stage, tumor location, age, or the presence and type of neoadjuvant therapy, and no cases were deliberately excluded for anatomical or procedural considerations. All procedures were performed by a single, consistent surgical team consisting of the same console surgeon and bedside assistant, thereby ensuring uniformity in surgical technique, team coordination, and strict adherence to the principles of Fusion Surgery throughout the study period. To maintain a well-defined and homogeneous study population and to ensure reliable evaluation of perioperative outcomes, patients who underwent surgery for benign esophageal disease, those requiring non-curative resections, emergent or palliative operations, or those with incomplete or missing perioperative data were systematically excluded. All patients were admitted three days before surgery and were deemed ready for discharge once oral intake was tolerated, pain was controlled, vital signs and labs were stable, and postoperative rehabilitation was completed.

The study protocol was approved by the Institutional Review Board of Gifu Graduate School of Medicine (approval number: 2024–323). The requirement for written informed consent was waived due to the retrospective nature of the study.

Surgical procedure: port placement and instruments

Patient positioning, cart positioning, and port placement were identical for both robotic systems, as illustrated in Fig. 1. Four robotic ports and one assistant port were arranged in the same configuration: the camera port was positioned at the seventh intercostal space along the midaxillary line; arm 1 at the tenth intercostal space along the posterior axillary line; and arms 3 and 4 at the fifth and third intercostal spaces, respectively.

Fig. 1.

Robotic port placement and patient positioning. Patient positioning and port placement were identical for both the da Vinci and hinotori systems. Four robotic ports and one assistant port were used. Standard instruments included bipolar forceps on arm 1 and monopolar curved scissors on arm 3

The instruments used for both the da Vinci and hinotori systems were nearly identical, with arm 1 typically equipped with bipolar forceps and arm 3 with monopolar curved scissors as the standard configuration. Instrument or arm exchanges, including camera cleaning, were performed as needed within a reasonable range, without disrupting the smooth progression of the operation. In this cohort of 30 patients, the number of instrument exchanges and camera cleanings was 2.5 ± 1.8 and 4.2 ± 2.7, respectively. Intraoperative blood loss was calculated as the sum of the suctioned fluid volume and the estimated blood absorbed by surgical sponges, with irrigation fluid volume subtracted from the total.

Fusion surgery concept

The surgical team consisted of a board-certified esophageal surgeon as the console surgeon and a nationally certified laparoscopic surgeon as the bedside assistant, both with substantial experience in upper gastrointestinal minimally invasive surgery. In both robotic systems, regardless of tumor location or the site of lymph node metastasis, the thoracic phase was performed in a standardized, reproducible sequence consisting of lower, middle, and upper mediastinal lymph node dissections (Fig. 2, indicated by yellow, green, and red lines, respectively). This stepwise approach was designed to ensure procedural consistency and safety across all cases.

Fig. 2.

Stepwise thoracic mediastinal lymph node dissection. The thoracic phase was performed in a standardized sequence of lower (yellow), middle (green), and upper (red) mediastinal dissections. The bedside surgeon used a vessel sealer (V) and surgical clips (C) at predefined sites (V1–V5 and C)

During the thoracic phase, the bedside surgeon primarily utilized three instruments: a vessel sealer, surgical clips, and a linear stapler. In this study, Fusion Surgery explicitly defined the specific application sites of these instruments as V1–V5 and C in Fig. 2. The vessel sealer was employed in five predefined situations (V1–V5 in Fig. 2), based on anatomical considerations and our standard retrospective practice, with an average of five to eight activations per case. These five sites are areas rich in vessels draining into lymph nodes, and the sealer was mainly used to coagulate and divide vessels draining into lymph nodes (Fig. 3, V1 and V5) and to dissect connective tissue planes demarcating lymph node boundaries (Fig. 3, V2–V4). Before the bedside surgeon uses the vessel sealer, the console surgeon always ensures adequate exposure of the operative field; therefore, the time required for each activation is less than one minute. The linear stapler was applied twice in each operation—once for division of the azygos vein and once for transection of the esophagus—to provide secure hemostasis and enhance procedural efficiency. Surgical clips were used exclusively during dissection of the left recurrent laryngeal nerve lymph nodes (C in Fig. 2), and the number of clips applied varied according to anatomical complexity and the degree of nodal adhesion, with a mean of 4.5 ± 3.8 clips used per case. Like the vessel sealer, the small vessels clipped by the bedside surgeon are fully exposed by the console surgeon, so the total time required for clipping is approximately five minutes.

Fig. 3.

Intraoperative photos and illustrations showing bedside assistant instrument use in mediastinal lymphadenectomy. Photos and illustrations depict the bedside assistant using a vessel sealer and clips at six predefined sites (V1–V5, and C)

Beyond these defined tasks, the bedside surgeon assisted with organ traction, retraction, suction, and coagulative hemostasis under the direction of the console surgeon to maintain optimal exposure. The frequency of such supportive interventions increased, particularly in obese patients or those with advanced-stage disease, but they did not affect the overall progress of the surgery.

Segmentation of surgical time

For analytical purposes, console time (from the end of docking to roll-out) during the thoracic phase, excluding the abdominal phase, was subdivided into three segments corresponding to lower, middle, and upper mediastinal dissections. Lower mediastinal dissection time was defined as the interval from the start of console operation to completion of No. 110 lymph node dissection (console start to V2), including mobilization of the lower esophagus and dissection around the diaphragmatic crura. Middle mediastinal dissection was defined as the period between V2 and completion of subcarinal lymph node dissection (V2 to V4), encompassing dissection around the bronchial bifurcation and adjacent vessels. Upper mediastinal dissection was defined as the interval from division of the azygos vein to console end, including exposure and dissection along both recurrent laryngeal nerves.

Outcomes

The primary outcome was the continuity of the learning curve during the transition from the da Vinci to the hinotori robotic esophagectomy, assessed using CUSUM analysis of console time for each mediastinal dissection stage (lower, middle, upper). Given the small sample size of the hinotori group, this analysis is presented as an exploratory assessment of feasibility and initial experience rather than as definitive evidence of a learning curve. CUSUM analysis was performed sequentially according to case order to identify trends in procedural efficiency and to evaluate whether skills acquired with the da Vinci were effectively maintained following the transition to the hinotori system.

Secondary outcomes included intraoperative blood loss, postoperative complications (graded according to the Clavien-Dindo classification [11]), and length of hospital stay, thus providing a comprehensive assessment of safety and procedural efficiency across both robotic platforms.

Statistical analysis

Continuous variables were assessed for normality using the Shapiro–Wilk test. Variables with a normal distribution are presented as mean ± standard deviation (SD) and were compared using the Student t-test for two groups or one-way ANOVA for three groups. Variables not normally distributed are presented as median [interquartile range, IQR] and were compared using the Wilcoxon rank-sum test for two groups or the Kruskal–Wallis test for three groups. Categorical variables are summarized as counts and percentages and were compared using Fisher’s exact test. All p-values are two-sided, and statistical significance was defined as p < 0.05. All statistical analyses were performed with R software (version 4.5.1; R Foundation for Statistical Computing, Vienna, Austria), using the “stats” package for standard tests and custom scripts for CUSUM analyses.

Results

Baseline characteristics are summarized in Table 1. In total, 30 patients who underwent RAMIE were included. Their mean age was 73 ± 6.96 years, and 24 patients were male. The mean body mass index (BMI) was 22.2 ± 2.30 kg/m². According to the American Society of Anesthesiologists physical status classification, 22 patients were class I, seven were class II, and one was class III. Tumors were predominantly located in the middle thoracic esophagus in 22 patients, with five located in the cervical to upper thoracic region and three in the lower thoracic or abdominal esophagus. Clinical T stages (UICC 8th edition [12]) were 1a in three patients, 1b in eight, 2 in four, 3 in seven, 4a in six, and 4b in two patients. Clinical N stages were N0 in 14 patients, N1 in nine, N2 in six, and N3 in one patient, whereas distant metastasis (M1) was present in two patients, both suspected to have solitary supraclavicular lymph node metastases. Neoadjuvant chemotherapy (NAC) was administered in 20 patients (66.7%). The two patients with M1 both showed a remarkable response to NAC and were assessed as achieving a complete response according to the Response Evaluation Criteria in Solid Tumors version 1.1 [13], after which standard curative esophagectomy was performed. The median total console time was 210 min [IQR 186.75–219.5], with median durations of 68.5 min [56.5–86.5] for upper mediastinal, 69 min [56.5–86.5] for middle mediastinal, and 61.5 min [54.3–74.8] for lower mediastinal lymph node dissections. Median intraoperative blood loss was 20 mL [10–30], and median length of hospital stay was 21.5 days [17–27.8]. In analyses stratified by tumor location and NAC status, no significant differences were observed in patient characteristics or surgical outcomes (Supplementally Table 1 and 2).

Table 1.

Baseline patient characteristics in 30 robot-assisted esophagectomy cases

Characteristics	Data (n = 30)
Age (mean ± SD), years	73 ± 6.96
Sex (Male/Female)	24/6
Body mass index (mean ± SD), kg/m2	22.2 ± 2.30
ASA physical status (I/II/III)	22/7/1
Tumor location (Ce, Ut/Mt/Lt, Ae)	5/22/3
Clinical T (UICC ver.8)
1a	3
1b	8
2	4
3	7
4a	6
4b	2
Clinical N (UICC ver.8)
0	14
1	9
2	6
3	1
Clinical M (UICC ver.8)
0	28
1	2
Clinical Stage (UICC ver.8)
I/II/III/IVA/IVB	10/4/8/6/2
Neoadjuvant chemotherapy, n (%)	20 (66.7)
RECIST (ver.1.1)	(n = 20)
PD/SD/PR/CR	0/1/17/2
Pathological Stage (UICC ver.8)
IA/IB/IIA/IIB/IIIA/IIIB	8/4/4/6/5/3
Console time (median [IQR]), min	210 [186.75–219.5]
Upper mediastinal lymph node dissection	68.5 [56.5–86.5]
Middle mediastinal lymph node dissection	69 [56.5–86.5]
Lower mediastinal lymph node dissection	61.5 [54.3–74.8]
Blood loss (median [IQR]), mL	20 [10–30]
Length of hospital stay (median [IQR]), days	21.5 [17–27.8]

Values are presented as mean ± standard deviation (SD) or median [interquartile range, IQR] as appropriate

Console time includes robotic operation for upper, middle, and lower mediastinal lymph node dissections, excluding the abdominal phase

ASA American Society of Anesthesiologists physical status, Ce cervical, Ut upper thoracic, Mt middle thoracic, Lt lower thoracic, Ae abdominal esophagus, RECIST Response Evaluation Criteria in Solid Tumors, PD progressive disease, SD stable disease, PR partial response, CR complete response

Patient characteristics and surgical outcomes for the da Vinci and hinotori groups are summarized in Table 2. There were no significant differences in age between the da Vinci and hinotori groups (mean ± SD: 73.1 ± 7.35 vs 72.1 ± 5.18 years, p = 0.866). BMI was also similar between groups (22.4 ± 2.35 vs 21.2 ± 1.98 kg/m², p = 0.289). The median total console time was significantly shorter in the hinotori group (180 min [155–185]) than in the da Vinci group (212 min [201–212], p = 0.005). Analysis by mediastinal dissection stage revealed comparable lower mediastinal dissection times between the da Vinci (60 min [46–73]) and hinotori (65 min [58–72]) groups (p = 0.933). For upper mediastinal dissection, the hinotori group showed a non-significant trend toward shorter times (da Vinci: 73 min [55–85], hinotori: 62 min [58–66], p = 0.148). Middle mediastinal dissection times tended to be shorter in the hinotori group (da Vinci: 76 min [44–76], hinotori: 46 min [42–50], p = 0.010); however, given the small sample size, this finding should be interpreted cautiously as a preliminary observation rather than definitive evidence of superiority. Postoperative complications occurred in five of 25 patients (20%) in the da Vinci group and in none in the hinotori group (p = 0.556, Fisher’s exact test; OR = 0, 95% CI 0–6.06). NAC was administered in 16 of 25 patients (64%) in the da Vinci group and four of five patients (80%) in the hinotori group (p = 0.640; OR = 2.20, 95% CI 0.18–122.85). The distribution of tumor stage did not differ significantly between groups (p = 0.306). The length of hospital stay was comparable between groups (median 22 [17–28] vs 18 [17–19] days, p = 0.486), and blood loss did not differ significantly (median 20 [10–40] vs 15 [12–17] mL, p = 0.756).

Table 2.

Patient characteristics and surgical outcomes by robotic system

	da Vinci group	hinotori group	p-Value
	n = 25	n = 5
Age (mean ± SD), years	73.08 ± 7.35	72.06 ± 5.18	0.8656
Male, n (%)	21 (84)	3 (60)	0.2543
Female, n (%)	4 (16)	2 (40)
Body mass index (mean ± SD), kg/m2	22.36 ± 2.35	21.20 ± 1.98	0.2893
Clinical stage (1/2/3/4a/4b) (UICC ver.8)	8/7/4/5/1	1/0/2/1/1	0.306
Neoadjuvant chemotherapy, n (%)	16 (64)	4 (80)	0.640
Console time (median [IQR]), min	212 [201–212]	180 [155–185]	0.005
Upper mediastinal lymph node dissection	73 [55–85]	62 [58–66]	0.148
Middle mediastinal lymph node dissection	76 [44–76]	46 [42–50]	0.010
Lower mediastinal lymph node dissection	60 [46–73]	65 [58–72]	0.933
Blood loss (median [IQR]), mL	20 [10–40]	15 [12–17]	0.756
Postoperative complications (Clavien-Dindo ≥ II), n (%)	5 (20)	0 (0)	0.556
Length of hospital stay (median [IQR]), days	22 [17–28]	18 [17–19]	0.486

Values are presented as mean ± standard deviation (SD) or median [interquartile range, IQR] as appropriate

Console time includes robotic operation for upper, middle, and lower mediastinal lymph node dissections, excluding the abdominal phase

Postoperative complications are classified according to Clavien-Dindo grade ≥ II

Operative times for upper, middle, and lower mediastinal dissections, as well as blood loss, were further analyzed across three surgical phases (phase 1: da Vinci cases 1–12; phase 2: da Vinci cases 13–25; phase 3: hinotori cases 26–30) (Table 3). Median operative times decreased from 82 [63–90] min in phase 1 to 62 [55–63] min in phase 3 for upper mediastinal dissection and from 77 [69–94] min to 46 [45–63] min for middle mediastinal dissection, whereas lower mediastinal dissection times remained stable. Blood loss was minimal and consistent across phases (20 [0–30] to 15 [15–20] mL). The Kruskal–Wallis test revealed no statistically significant differences between the three phases regarding upper, middle, or lower mediastinal dissection duration or intraoperative blood loss (all p > 0.05). Although statistical significance was not reached, a trend toward decreased upper mediastinal dissection time and blood loss was observed from early da Vinci to hinotori cases; given the small sample size, these should be considered preliminary observations rather than definitive evidence of superiority. These results suggest that operative performance and procedural stability were maintained throughout all RAMIE phases. Total console time decreased progressively across the phases, especially for middle mediastinal dissection, as illustrated in Fig. 4. Changes in operative time were further evaluated using CUSUM analysis. For upper mediastinal dissection, operative time remained relatively stable during the latter da Vinci cases but decreased sharply after transition to hinotori, indicating the onset of a learning effect and steady advancement. Middle mediastinal dissection was relatively stable initially, but a marked reduction in operative time was observed following hinotori adoption, demonstrating ongoing skill acquisition. Lower mediastinal dissection showed a consistent downward trend from the first cases, indicating that the learning effect was present from the beginning (Fig. 5). Additionally, it was confirmed that the CUSUM curves continued to decline regardless of the presence or absence of NAC (Supplementary Fig. 1).

Table 3.

Operative outcomes of mediastinal lymphadenectomy across surgical phases

	Phase 1	Phase 2	Phase 3	p-Value
	(da Vinci cases 1–12) n = 12	(da Vinci cases 13–25) n = 13	(hinotori cases 26–30) n = 5
Upper mediastinal dissection (median [IQR]), min	82 [63–90]	69 [57.3–77]	62 [55–63]	0.174
Middle mediastinal dissection (median [IQR]), min	77 [69–94]	70.5 [59.8–86.5]	46 [45–63]	0.099
Lower mediastinal dissection (median [IQR]), min	59 [52–105]	63 [57.5–76]	65 [55–69]	0.994
Estimated blood loss (median [IQR]), mL	20 [0–30]	25 [11.3–47.5]	15 [15–20]	0.627

Values are presented as median [interquartile range, IQR] as appropriate

P-values were calculated using the Kruskal–Wallis test

Fig. 4.

Operative time trends for console and mediastinal lymph node dissection across the 30 consecutive cases. The graph shows the console time (blue) and upper (red), middle (green), and lower (yellow) mediastinal lymph node dissection times across 30 consecutive cases. Cases are divided into three phases: phase 1 (cases 1–12, initial da Vinci cases), phase 2 (cases 13–25, latter da Vinci cases), and phase 3 (cases 26–30, hinotori cases)

Fig. 5.

Cumulative sum (CUSUM) curves for mediastinal dissection and total console time. CUSUM curves for upper (red), middle (green), and lower (yellow) mediastinal dissection, along with total console time (blue), are shown for the 30 consecutive cases

Discussion

In complex surgeries, such as those in hepatobiliary and pancreatic fields, which require advanced technical skills and the use of multiple instruments, the effectiveness of combining robotic and laparoscopic or thoracoscopic approaches, referred to as Fusion Surgery, has been increasingly reported [14, 15]. The CUSUM method offers a robust quantitative tool to evaluate not only when surgical proficiency is achieved but also the extent of improvement. It is widely used for analyzing learning curves, designing stepwise training programs, and ensuring patient safety in robotic procedures, including gastric and colorectal surgery [16–22]. In the context of esophageal cancer surgery, Zhao et al. assessed the learning curve for minimally invasive esophagectomy using CUSUM analysis and demonstrated its impact on operative time, intraoperative blood loss, and postoperative mortality [23]. Similarly, Yuan et al. showed that in robot-assisted McKeown esophagectomy for thoracic esophageal cancer, operative times were significantly reduced after the 27th case, indicating that cumulative surgical experience directly enhances technical proficiency [24]. Zhuo et al. analyzed thoracic lymphadenectomy during the transition from McKeown to Ivor Lewis approaches, finding that 23 McKeown cases were required to achieve significant improvement in upper mediastinal lymph node dissection and that a new learning phase emerged upon switching to Ivor Lewis [25]. Collectively, these existing studies on RAMIE highlight that both cumulative experience and stepwise adaptation are important not only for mastering advanced surgical skills but also for modifying the surgical approach.

To date, no existing study has evaluated the acquisition of surgical proficiency during a transition involving a change of robotic system —from the da Vinci to the hinotori—particularly using step-specific learning curves. Our study addresses this gap by examining segmental learning curves, providing a detailed perspective on adopting and integrating a new robotic platform in esophageal cancer surgery. In our series of 30 robotic cases, performance stabilized in the latter da Vinci cases, although minor fluctuations related to case complexity and anesthesia remained. Thus, the learning curve had entered a mature phase but had not completely plateaued. After transitioning from the da Vinci to the hinotori system at case 25, no upward inflection in the CUSUM curve was observed, and the downward trend persisted. Although the hinotori cases were performed by a team already experienced with the da Vinci system, visualization of the learning curve via CUSUM remains meaningful, demonstrating that the standardized Fusion Surgery concept maintained operative consistency and team coordination across the transition, and that the change in robotic platform did not impede the natural progression of technical skill acquisition, with competencies gained using the da Vinci system effectively retained. Comparisons between the two-group analysis (phase 1 + 2 vs phase 3) and the three-group analysis revealed that middle mediastinal dissection time was significantly shorter only after introduction of the hinotori system. Particularly in the middle mediastinum, numerous deeply located lymph nodes present a challenging anatomical region prone to bleeding due to membrane injury. This result likely reflects that, with cumulative experience, the bedside surgeon’s coagulation of lymph node inflow vessels (V3 to V4) became more stable, leading to a substantial reduction in the time required for hemostatic maneuvers. For the upper mediastinum, although the three-group analysis demonstrated only a non-significant trend toward improvement—likely reflecting its higher inherent technical complexity—the CUSUM curve showed a continuous reduction in operative time from da Vinci to hinotori phases. This pattern suggests that experience with the hinotori system was effectively integrated with prior da Vinci experience, enabling seamless and progressive improvement, even in technically demanding regions. Overall, these findings indicate that the transition to the hinotori system did not compromise surgical performance and that ongoing skill acquisition and procedural efficiency continued across all mediastinal regions, including the most challenging.

When the two robotic platforms are compared, both systems are equipped with four robotic arms; however, the da Vinci system utilizes seven-axis arms, whereas hinotori system incorporates eight-axis robotic arms. Furthermore, hinotori is distinguished from da Vinci by several notable features, including its docking-free design and a wide 16:9 high-definition three-dimensional endoscopic view [26–32]. These characteristics are particularly advantageous for esophagectomy, as the mediastinum—the operative field—is a confined anatomical space surrounded by the ribs and vital organs. In addition to the enhanced flexibility afforded by the eight-axis robotic arms, the risk of collision between the robotic arms is minimized, and the movements of the bedside surgeon are not restricted. Furthermore, because the mediastinum extends widely along the cranio-caudal axis, the wide 16:9 aspect ratio is particularly advantageous. The ergonomically optimized cockpit design, which reduces surgeon fatigue, represents another notable advantage of hinotori. Such system-specific features likely facilitated a smooth and efficient transition between robotic platforms, enabling surgeons to maintain and further enhance technical performance. Overall, our results suggest that Fusion Surgery is not confined to familiarity with a single robotic system but represents a universal framework applicable across different platforms. This approach allows the bedside surgeon to gain experience close to the patient within an optimal learning environment, facilitating the maturation of the surgical team. In this cohort, active assistance by the bedside surgeon under the console surgeon’s direction became smoother with experience, and the frequency of verbal instructions from the console surgeon decreased. As a result, even when adopting new robotic systems, technical proficiency can be improved in a continuous and stepwise manner, thereby ensuring both efficiency and safety in complex esophageal cancer surgeries.

As a pilot and exploratory study, our findings are subject to several inherent limitations. The small sample size, particularly in the hinotori group, limits statistical power and all p-values derived from comparisons involving this group should be interpreted with caution. Non-significant differences are presented as non-significant trends rather than definitive evidence of equivalence, reflecting the limited statistical power inherent to this study design. Furthermore, this is a single-center, retrospective analysis, which may introduce biases related to patient selection, perioperative management, and case complexity. In this study, the bedside surgeon was a nationally certified laparoscopic surgeon; whether similar outcomes could be achieved by an experienced trainee or resident requires further investigation.

Consequently, all conclusions are presented with appropriate caution, and the results primarily serve to inform feasibility and hypothesis generation for future prospective studies.

Conclusions

This study suggested that Fusion Surgery in RAMIE may contribute to the enhancement of surgical proficiency. CUSUM analysis indicated stepwise improvements in operative times across the upper, middle, and lower mediastinal segments. Notably, the transition from the da Vinci to the hinotori system did not appear to cause any re-elevation of CUSUM curves, suggesting that previously acquired proficiency was largely maintained. These observations indicate that, even in an era of increasingly diverse robotic platforms, the Fusion Surgery concept may support a smooth continuation of acquired skills without disrupting the learning curve, offering a potentially robust and adaptable framework for safe and efficient adoption across multiple systems. Integrating cumulative robotic experience may support gradual improvement in surgical performance, even with novel platforms, which can ultimately contribute to enhanced procedural efficiency and potentially improved patient outcomes in esophageal cancer surgery. The information obtained from the present study is limited; to more robustly demonstrate the efficacy of Fusion Surgery, future studies should adopt a prospective, multicenter design and evaluate both oncologic and functional outcomes.

Authors’ contributions

Conceptualization, Y.S. and N.M.; Writing – Original Draft Preparation, Y.S.; Writing – Review & Editing, Y.S. and N.M.; Supervision, Y.H., Y.T., S.F., N.M., R.Y., T.H., K.M., M.F., I.Y., R.A., J.Y.T., and N.M. All authors have read and agreed to the published version of the manuscript.

Funding

There are no sources of funding to declare.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Gifu University Graduate School of Medicine (approval number: 2024–323). Written informed consent was obtained from all participants prior to inclusion in the study.

Consent for publication

Written informed consent was obtained from the patients for publication of identifiable data and accompanying images.

Competing interests

The authors declare no competing interests.