Comparison of Minimally Invasive Colectomy Between Intra‐ Versus Extra‐Corporeal Anastomosis Technique (COMMIT Study): Short‐Term Analysis Authors: Takuya Takami, Takehito Yamamoto, Yoshiro Itatani, Ryosuke Mizuno, Kae Okoshi, Yasunori Deguchi, Keiji Nagata, Tomoaki Okada, Rei Mizuno, Kazutaka Obama Published: 2026 ABSTRACT ABSTRACTAimTo compare the operative and short‐term outcomes between intracorporeal anastomosis (IA) and extracorporeal anastomosis (EA) in minimally invasive colectomy.MethodsThis multicenter cohort study used data from the Kyoto Colorectal Surgery Study Group database. Patients who underwent minimally invasive colectomy between 2018 and 2025 were included. Primary outcome was the incidence of intraoperative blood loss ≥ 200 mL. Secondary outcomes included continuous intraoperative blood loss, the incidence of intraoperative blood loss ≥ 100 and ≥ 300 mL, intraoperative blood transfusion, operative time, resection margins, harvested lymph node yield, and postoperative complications. Propensity score matching was performed to balance baseline characteristics.ResultsOf 3213 eligible patients, 928 and 2285 underwent IA and EA, respectively. After the propensity score matching, 650 matched pairs were identified. The incidence of intraoperative blood loss ≥ 200 mL was significantly lower in the IA group than that in the EA group (2.5% vs. 4.9%; risk difference, −2.46%; 95% confidence interval, −4.65% to −0.43%), whereas continuous intraoperative blood loss was comparable. Although operative time was generally longer for IA, subgroup analysis demonstrated significant effect modification according to surgical approach, with operative time tending to favor IA. The incidence of adequate resection margins tended to be higher in IA, although the difference was not statistically significant. Harvested lymph nodes and postoperative complications were generally comparable.ConclusionIA was associated with a lower incidence of intraoperative blood loss ≥ 200 mL during minimally invasive colectomy. These findings indicate that IA may reduce the occurrence of relatively high blood loss events.Keywords: anastomosis, colon cancer, minimally invasive surgery, multicenter study, short‐term outcomes Introduction Colorectal cancer is one of the most common malignancies globally, and its incidence is increasing [1, 2]. Minimally invasive surgical techniques, including laparoscopic and robot‐assisted surgery, have become standard treatment modalities for colon cancer [3]. Initially, during the early phase of minimally invasive colectomy, extracorporeal anastomosis (EA) is the conventional approach, whereby bowel resection and reconstruction are performed outside the body. However, since 2010, the adoption of intracorporeal anastomosis (IA) has been increasing, wherein these procedures are completed within the abdominal cavity [4]. Numerous studies have examined operative and short‐term outcomes when comparing IA and EA in laparoscopic colectomy. Regarding operative outcomes, IA has been associated with longer operative time, reduced intraoperative blood loss, smaller incision size, extended resection margins, and increased lymph node yield [5, 6, 7]. In terms of short‐term outcomes, IA has been associated with reduced postoperative pain, faster recovery of bowel function, and decreased duration of postoperative hospital stay. Most studies have reported comparable incidences of postoperative complications between IA and EA [8, 9, 10]. Although there is a growing consensus regarding operative and short‐term outcomes when comparing IA and EA particularly in laparoscopic colectomy, only a few studies have assessed these outcomes within large multicenter cohorts that include robot‐assisted surgery [11]. Moreover, the supposed benefits of IA, such as decreased intraoperative blood loss and extended resection margins, appear to be modest, and their clinical importance remains uncertain [6, 10]. Additionally, with increased experience, IA may be associated with reduced operative time compared with EA. By utilizing a large multicenter cohort that included robot‐assisted surgery, we assessed the incidence of clinically meaningful intraoperative bleeding, proportion of adequate resection margins (both proximal and distal margins ≥ 100 mm) as recommended by the Japanese guidelines [12], and operative time in relation to institutional experience with IA. We hypothesized that IA is associated with a reduced incidence of clinically meaningful intraoperative bleeding and an increased proportion of adequate resection margins. Furthermore, we anticipated that operative time in IA would eventually become shorter than that in EA, as institutional experience accumulated. Materials and Methods Study Design and Setting This retrospective cohort study utilized the Kyoto Colorectal Surgery Study Group (KCRG) database, which includes data on patients with colorectal cancer who underwent surgery since 2018 at the Kyoto University Hospital and 24 affiliated institutions. This database has been used in several colorectal surgery studies [13, 14]. The study protocol was approved by the Ethics Committee of the Kyoto University Graduate School and Faculty of Medicine (approval number: R0286). Informed consent was obtained using an opt‐out form available on the study website. Patients who declined participation were excluded. Selection of Participants We included patients diagnosed with colon cancer who underwent minimally invasive colectomy (laparoscopic or robot‐assisted) between January 2018 and December 2025. Inclusion criteria were as follows: (i) patients diagnosed with appendiceal, cecal, ascending, transverse, descending, or sigmoid colon cancer who underwent resection with D2 or D3 lymph node dissection (LD), as defined by the Japanese Classification of Colorectal, appendiceal, and anal carcinoma [15]; (ii) patients from institutions performing ≥ 20 IA procedures during the study period; and (iii) availability of data specifying the anastomotic approach used (IA or EA). Exclusion criteria were as follows: (i) emergency surgery; (ii) construction of a diverting (covering) stoma; and (iii) transanal double‐ or single‐stapling technique anastomosis. Measurements Baseline covariates were selected as follows: (i) time‐ and institution‐related factors included the study period and the institutional IA volume at the treating institution; (ii) patient‐related factors included age, sex, body mass index (BMI), American Society of Anesthesiologists Physical Status Classification System, diabetes mellitus, respiratory disease, hypertension, heart failure, dialysis, and neoadjuvant chemotherapy; (iii) tumor‐related factors encompassed tumor location, presence and type of preoperative decompression as a treatment for colonic obstruction, clinical depth of tumor invasion (cT), clinical lymph node metastasis (cN), and clinical distant metastasis (cM); and (iv) surgery‐related factors comprised surgical approach (laparoscopic or robot‐assisted surgery), surgical procedure, combined resection of other organs, and extent of LD. The study period was categorized into an early period (2018–2021) and a late period (2022–2025). Institutional IA volume was defined according to the cumulative number of IA procedures performed at each institution during the study period before application of study‐specific exclusion criteria. Procedures involving double‐ or single‐stapling technique anastomosis were not included in the calculation of institutional IA volume. Institutions were dichotomized at the median institutional IA volume. Definitions of tumor location, cT, cN, cM, surgical procedure, and extent of LD were derived from the Japanese Classification of Colorectal, appendiceal, and anal carcinoma [15]. Outcomes Primary outcome was the incidence of intraoperative blood loss of ≥ 200 mL, consistent with a previous report [16]. Secondary outcomes included continuous intraoperative blood loss, the incidences of intraoperative blood loss ≥ 100 and ≥ 300 mL [14], intraoperative blood transfusion, operative time, proximal margin (PM), distal margin (DM), adequate resection margins, harvested lymph node yield, and postoperative complications. Adequate resection margins were defined according to the Japanese guidelines as both PM and DM being ≥ 100 mm [12]. Postoperative complications were defined as Clavien–Dindo grade ≥II events occurring within 1 month after surgery [17]. Statistical Analysis A complete case analysis was conducted, excluding observations with missing data. To achieve a balance in baseline characteristics between the IA and EA groups, propensity scores for IA were estimated using a logistic regression model. Subsequently, one‐to‐one propensity score matching was performed with a caliper width of < 0.05 of the standard deviation of the propensity score logit. Matching was performed without replacement. The variables incorporated into the propensity score model included all time‐ and institution‐related, patient‐related, tumor‐related, and surgery‐related factors as previously described. Baseline characteristics before and after the matching were evaluated using standardized mean differences (SMDs), with values < 0.1 considered indicative of adequate covariate balance [18]. The incidence of intraoperative blood loss ≥ 200 mL, intraoperative blood loss ≥ 100 mL, intraoperative blood loss ≥ 300 mL, intraoperative blood transfusion, adequate resection margins, and postoperative complications were evaluated by calculating risk differences (RDs) with corresponding 95% confidence intervals (CIs). Continuous intraoperative blood loss, operative time, PM, DM, and harvested lymph node yield were evaluated by estimating median differences with corresponding 95% CIs. The 95% CIs were estimated using bootstrap resampling for 2000 iterations. As a sensitivity analysis, mixed‐effects logistic regression models with institution as a random effect were fitted for the primary outcome to account for potential clustering by institution, and odds ratios (ORs) with corresponding 95% CIs were estimated. Two models were constructed: Model 1 adjusted for surgical approach and study period, whereas Model 2 additionally adjusted for patient‐, tumor‐, and surgery‐related covariates. Prespecified subgroup analyses for intraoperative blood loss ≥ 200 mL, operative time, and adequate resection margin were conducted based on study period (early vs. late), institutional IA volume (low vs. high), BMI (< 25 vs. ≥ 25 kg/m2), clinical tumor depth (≤ cT3 vs. cT4), clinical lymph node status (cN− vs. cN+), surgical approach (laparoscopic vs. robot‐assisted), and surgical procedure (right vs. left colectomy). To further evaluate the influence of institutional IA experience on operative time, a separate learning‐curve analysis was performed. Institutional IA experience was defined as the cumulative number of IA procedures performed at each institution before the index operation. Both IA and EA procedures were assigned to one of four categories according to the cumulative institutional IA experience at the time of surgery (0–49, 50–99, 100–149, and ≥ 150 prior IA procedures), and operative time differences between IA and EA were evaluated within each category. All statistical analyses were performed using R version 4.6.0 (R Foundation for Statistical Computing, Vienna, Austria). Results Characteristics of Study Participants In total, 4239 patients underwent minimally invasive colectomy with D2 or D3 LD. Patients who underwent emergency surgery (n = 32), diverting (covering) stoma construction (n = 13), or double‐ or single‐stapling technique anastomosis (n = 660) were excluded. Additionally, 335 patients with missing data were excluded. Consequently, 3213 patients were included in the final analysis, with 928 and 2285 patients in the IA and EA groups, respectively. After a 1:1 propensity score matching, 650 matched pairs were identified (Figure 1). Flow chart of the study participants. DST, double‐stapling technique; EA, extracorporeal anastomosis; IA, intracorporeal anastomosis; SST, single‐stapling technique. Table 1 summarizes the baseline characteristics of the two groups before and after matching. After matching, all baseline covariates were well balanced, with SMDs of < 0.1. Covariate balance before and after matching is further illustrated in the Love plot (Figure S1). TABLE 1Baseline characteristics before and after matching.Before matchingAfter matchingIA (n = 928)EA (n = 2285)SMDIA (n = 650)EA (n = 650)SMDSurgery period, n (%)Early (2018–2021)144 (15.5)1253 (54.8)0.904143 (22.0)148 (22.8)0.018Late (2022–2025)784 (84.5)1032 (45.2)507 (78.0)502 (77.2)Institutional IA volume, n (%)Low‐volume institutions204 (22.0)1115 (48.8)0.584174 (26.8)162 (24.9)0.042High‐volume institutions724 (78.0)1170 (51.2)476 (73.2)488 (75.1)Age (years), median [IQR]75 [69–81]75 [67–80]0.06475.5 [67–80]75 [68–80]0.001Sex, n (%)Male460 (49.6)1085 (47.5)0.042323 (49.7)326 (50.2)0.009Female468 (50.4)1200 (52.5)327 (50.3)324 (49.8)BMI (kg/m2), n (%)< 18.5117 (12.6)285 (12.5)0.08591 (14.0)75 (11.5)0.07418.5–25578 (62.3)1506 (65.9)411 (63.2)422 (64.9)≥ 25233 (25.1)494 (21.6)148 (22.8)153 (23.5)ASA‐PS, n (%)1106 (11.4)363 (15.9)0.16080 (12.3)73 (11.2)0.0562684 (73.7)1531 (67.0)464 (71.4)466 (71.7)3136 (14.7)380 (16.6)104 (16.0)107 (16.5)42 (0.2)11 (0.5)2 (0.3)4 (0.6)Diabetes mellitus, n (%)196 (21.1)434 (19.0)0.053135 (20.8)143 (22.0)0.030Respiratory disease, n (%)47 (5.1)88 (3.9)0.05935 (5.4)37 (5.7)0.013Hypertension, n (%)442 (47.6)1019 (44.6)0.061299 (46.0)309 (47.5)0.031Heart failure, n (%)8 (0.9)28 (1.2)0.0367 (1.1)5 (0.8)0.032Dialysis, n (%)5 (0.5)29 (1.3)0.0775 (0.8)5 (0.8)< 0.001Neoadjuvant chemotherapy, n (%)17 (1.8)11 (0.5)0.1279 (1.4)8 (1.2)0.014Location, n (%)Appendix18 (1.9)28 (1.2)0.14912 (1.8)14 (2.2)0.068Cecum153 (16.5)464 (20.3)113 (17.4)128 (19.7)Ascending354 (38.1)921 (40.3)243 (37.4)232 (35.7)Transverse209 (22.5)490 (21.4)140 (21.5)135 (20.8)Descending125 (13.5)246 (10.8)93 (14.3)90 (13.8)Sigmoid69 (7.4)136 (6.0)49 (7.5)51 (7.8)Preoperative decompression, n (%)Not performed837 (90.2)1997 (87.4)0.148580 (989.2)572 (88.0)0.050Nasogastric decompression tube0 (0.0)2 (0.1)0 (0.0)0 (0.0)Nasointestinal decompression tube5 (0.5)40 (1.8)5 (0.8)5 (0.8)Transanal decompression tube5 (0.5)11 (0.5)5 (0.8)6 (0.9)Stent64 (6.9)203 (8.9)49 (7.5)57 (8.8)Others17 (1.8)32 (1.4)11 (1.7)10 (1.5)cT, n (%)Tis4 (0.4)17 (0.7)0.1874 (0.6)3 (0.5)0.060T1150 (16.2)297 (13.0)102 (15.7)95 (14.6)T2157 (16.9)318 (13.9)104 (16.0)99 (15.2)T3377 (40.6)888 (38.9)265 (40.8)283 (43.5)T4240 (25.9)765 (33.5)175 (26.9)170 (26.2)cN, n (%)N0475 (51.2)1209 (52.9)0.100332 (51.1)327 (50.3)0.053N1307 (33.1)721 (31.6)212 (32.6)226 (34.8)N2135 (14.5)302 (13.2)98 (15.1)89 (13.7)N311 (1.2)53 (2.3)8 (1.2)8 (1.2)cM, n (%)M0835 (90.0)2051 (89.8)0.007588 (90.5)585 (90.0)0.016M193 (10.0)234 (10.2)62 (9.5)65 (10.0)Surgical approach, n (%)Laparoscopic563 (60.7)2185 (95.6)0.933535 (82.3)550 (84.6)0.062Robot‐assisted365 (39.3)100 (4.4)115 (17.7)100 (15.4)Surgical procedure, n (%)Partial colectomy92 (9.9)179 (7.8)0.17359 (9.1)53 (8.2)0.043Ileocecal resection258 (27.8)613 (26.8)185 (28.5)195 (30.0)Right hemicolectomy359 (38.7)1058 (46.3)247 (38.0)244 (37.5)Left hemicolectomy152 (16.4)315 (13.8)110 (16.9)109 (16.8)Sigmoid colectomy67 (7.2)120 (5.3)49 (7.5)49 (7.5)Combined resection of other organs, n (%)79 (8.5)228 (10.0)0.05163 (9.7)58 (8.9)0.026Lymph node dissection, n (%)D271 (7.7)312 (13.7)0.19662 (9.5)66 (10.2)0.021D3857 (92.3)1973 (86.3)588 (90.5)584 (89.8) Note: Institutional IA volume was dichotomized at the median cumulative number of IA procedures performed per institution during the study period (low‐volume: 21–52 cases; high‐volume: 54–222 cases).Abbreviations: ASA‐PS, American Society of Anesthesiologists Physical Status Classification System; BMI, body mass index; EA, extracorporeal anastomosis; IA, intracorporeal anastomosis; IQR, interquartile range; SMD, standardized mean difference. Baseline characteristics before and after matching. Note: Institutional IA volume was dichotomized at the median cumulative number of IA procedures performed per institution during the study period (low‐volume: 21–52 cases; high‐volume: 54–222 cases). Abbreviations: ASA‐PS, American Society of Anesthesiologists Physical Status Classification System; BMI, body mass index; EA, extracorporeal anastomosis; IA, intracorporeal anastomosis; IQR, interquartile range; SMD, standardized mean difference. Characteristics of patients excluded because of missing data and unmatched patients are presented in Tables S1 and S2, respectively. Institution‐level distributions are shown in Figure S2. Temporal changes in surgical volume, IA proportion, and robot‐assisted surgery proportion according to institutional IA volume group and study period are shown in Figure S3. Primary Outcome The incidence of intraoperative blood loss ≥ 200 mL was consistently low in the IA group, both before and after the propensity score matching. Before the matching, the incidence of intraoperative blood loss ≥ 200 mL was 2.2% in the IA group and 4.9% in the EA group (RD, −2.75%; 95% CI, −3.99% to −1.36%). Similar results were observed after the propensity score matching, with incidences of 2.5% and 4.9% in the IA and EA groups, respectively (RD, −2.46%; 95% CI, −4.65% to −0.43%) (Table 2). TABLE 2Intraoperative blood loss and transfusion before and after matching.OutcomeBefore matchingAfter matchingBlood loss ≥ 200 mL, n (%)IA20/928 (2.2)16/650 (2.5)EA112/2285 (4.9)32/650 (4.9)RD (%) [95% CI] a −2.75 [−3.99 to −1.36]−2.46 [−4.65 to −0.43]Continuous blood loss (mL), median [IQR]IA0 [0–10]0 [0–10]EA0 [0–37]0 [0–32]Median difference (n) [95% CI] b 0 [−5 to 0]0 [0–0]Blood loss ≥ 100 mL, n (%)IA57/928 (6.1)40/650 (6.2)EA264/2285 (11.6)64/650 (9.9)RD (%) [95% CI] a −5.41 [−7.37 to −3.29]−3.69 [−6.74 to −0.75]Blood loss ≥ 300 mL, n (%)IA10/928 (1.1)8/650 (1.2)EA61/2285 (2.7)14/650 (2.2)RD (%) [95% CI] a −1.59 [−2.51 to −0.55]−0.92 [−2.49 to 0.52]Transfusion, n (%)IA18/928 (1.9)11/650 (1.7)EA74/2285 (3.2)20/650 (3.1)RD (%) [95% CI] a −1.30 [−2.39 to −0.03]−1.38 [−3.19 to 0.29]Abbreviations: CI, confidence interval; EA, extracorporeal anastomosis; IA, intracorporeal anastomosis; IQR, interquartile range; RD, risk difference. a RDs were calculated with EA as the reference group (IA–EA). b Median differences were calculated using EA as the reference group (IA–EA). Intraoperative blood loss and transfusion before and after matching. Abbreviations: CI, confidence interval; EA, extracorporeal anastomosis; IA, intracorporeal anastomosis; IQR, interquartile range; RD, risk difference. RDs were calculated with EA as the reference group (IA–EA). Median differences were calculated using EA as the reference group (IA–EA). Mixed‐effects logistic regression models incorporating institution as a random effect yielded consistent results. In Model 1, IA was associated with a lower incidence of intraoperative blood loss ≥ 200 mL (OR, 0.46; 95% CI, 0.26–0.81). Similar findings were observed in Model 2 (OR, 0.39; 95% CI, 0.22–0.67). Secondary Outcomes After the propensity score matching, continuous intraoperative blood loss was comparable between the two groups (median difference, 0 mL; 95% CI, 0–0 mL). The incidence of intraoperative blood loss ≥ 100 mL was lower in the IA group than in the EA group (RD, −3.69%; 95% CI, −6.74% to −0.75%). The incidence of intraoperative blood loss ≥ 300 mL also tended to be lower in the IA group, although the difference did not reach statistical significance (RD, −0.92%; 95% CI, −2.49% to 0.52%). Intraoperative transfusion tended to be less frequent in the IA group than in the EA group, although the difference did not reach statistical significance (RD, −1.38%; 95% CI, −3.19% to 0.29%) (Table 2). Operative time was longer in the IA group than that in the EA group (median difference, 11.5 min; 95% CI, 2–22 min). The PM was comparable between the two groups (median difference, 0 mm; 95% CI, 0–10 mm). In contrast, DM was longer in the IA group than that in the EA group (median difference, 6.5 mm; 95% CI, 0–10 mm). Although the incidence of adequate resection margins was numerically higher in the IA group, the difference was not statistically significant (RD, 5.08%; 95% CI, −0.28% to 10.41%). Harvested lymph node yield was comparable between the two groups (median difference, −1; 95% CI, −2 to 1.5) (Table 3). TABLE 3Operative time, proximal margin, distal margin, adequate resection margin, and harvested lymph nodes before and after matching.OutcomeBefore matchingAfter matchingOperative time (min), median [IQR]IA281 [228–345.2]272.5 [222–339]EA250 [206–305]261 [214–315]Median difference (min) [95% CI] a 31 [23–41]11.5 [2–22]Proximal margin (mm), median [IQR]IA110 [80–140]100 [80–140]EA100 [75–150]100 [78.2–140]Median difference (mm) [95% CI] a 10 [0–10]0 [0–10]Distal margin (mm), median [IQR]IA110 [90–132.8]106.5 [90–130]EA100 [70–120]100 [80–120]Median difference (mm) [95% CI] a 10 [10–15]6.5 [0–10]Adequate resection margin, n (%)IA450/928 (48.5)287/650 (44.2)EA798/2285 (34.9)254/650 (39.1)RD (%) [95% CI] b 13.6 [9.8–17.3]5.08 [−0.28 to 10.41]Harvested lymph nodes (n), median [IQR]IA25 [17–37]24 [17–35]EA25 [16–36]25 [17–34.8]Median difference (n) [95% CI] a 0 [−1 to 1]−1 [−2 to 1.5] Note: Adequate resection margin was defined as both proximal and distal margins ≥ 100 mm.Abbreviations: CI, confidence interval; EA, extracorporeal anastomosis; IA, intracorporeal anastomosis; IQR, interquartile range; RD, risk difference. a Median differences were calculated using EA as the reference group (IA–EA). b RD was calculated with EA as the reference group (IA–EA). Operative time, proximal margin, distal margin, adequate resection margin, and harvested lymph nodes before and after matching. Note: Adequate resection margin was defined as both proximal and distal margins ≥ 100 mm. Abbreviations: CI, confidence interval; EA, extracorporeal anastomosis; IA, intracorporeal anastomosis; IQR, interquartile range; RD, risk difference. Median differences were calculated using EA as the reference group (IA–EA). RD was calculated with EA as the reference group (IA–EA). Regarding postoperative complications, no substantial differences were observed between the two groups after matching for most complications. However, organ/space surgical site infection (SSI) tended to occur more frequently in the IA group than in the EA group, although the difference did not reach statistical significance (RD, 1.38%; 95% CI, −0.05% to 3.00%) (Table 4). Other complications comprised pneumonia, deep vein thrombosis, pulmonary embolism, renal dysfunction, urinary tract infection, neurological complications, and sepsis. TABLE 4Postoperative complications before and after matching.OutcomeBefore matchingAfter matchingSuperficial SSI, n (%)IA12/928 (1.3)9/650 (1.4)EA19/2285 (0.8)7/650 (1.1)RD (%) [95% CI] a 0.46 [−0.26 to 1.46]0.31 [−0.99 to 1.66]Deep SSI, n (%)IA1/928 (0.1)1/650 (0.2)EA8/2285 (0.4)3/650 (0.5)RD (%) [95% CI] a −0.24 [−0.60 to 0.27]−0.31 [−1.21 to 0.44]Organ/space SSI, n (%)IA19/928 (2.1)16/650 (2.5)EA23/2285 (1.0)7/650 (1.1)RD (%) [95% CI] a 1.04 [0.16–2.22]1.38 [−0.05 to 3.00]Anastomotic leakage, n (%)IA10/928 (1.1)7/650 (1.1)EA15/2285 (0.7)7/650 (1.1)RD (%) [95% CI] a 0.42 [−0.23 to 1.35]0.00 [−1.26 to 1.26]Ileus, n (%)IA15/928 (1.6)8/650 (1.2)EA21/2285 (0.9)9/650 (1.4)RD (%) [95% CI] a 0.70 [−0.10 to 1.78]−0.15 [−1.53 to 1.19]Bowel obstruction, n (%)IA4/928 (0.4)2/650 (0.3)EA6/2285 (0.3)4/650 (0.6)RD (%) [95% CI] a 0.17 [−0.24 to 0.86]−0.31 [−1.30 to 0.57]Others, n (%)IA99/928 (10.7)67/650 (10.3)EA242/2285 (10.6)68/650 (10.5)RD (%) [95% CI] a 0.08 [−2.19 to 2.54]0.15 [−3.19 to 3.50]All, n (%)IA129/928 (13.9)90/650 (13.9)EA284/2285 (12.4)82/650 (12.6)RD (%) [95% CI] a 1.47 [−1.05 to 4.17]1.23 [−2.47 to 4.94] Note: Other complications comprised pneumonia, deep vein thrombosis, pulmonary embolism, renal dysfunction, urinary tract infection, neurological complications, and sepsis.Abbreviations: CI, confidence interval; EA, extracorporeal anastomosis; IA, intracorporeal anastomosis; RD, risk difference; SSI, surgical site infection. a RDs were calculated with EA as the reference group (IA–EA). Postoperative complications before and after matching. Note: Other complications comprised pneumonia, deep vein thrombosis, pulmonary embolism, renal dysfunction, urinary tract infection, neurological complications, and sepsis. Abbreviations: CI, confidence interval; EA, extracorporeal anastomosis; IA, intracorporeal anastomosis; RD, risk difference; SSI, surgical site infection. RDs were calculated with EA as the reference group (IA–EA). Subgroup Analysis In the subgroup analyses, the incidence of blood loss ≥ 200 mL was consistently low in the IA group across most subgroups, without evidence of significant effect modification (Figure 2). Operative time was generally longer in the IA group than that in the EA group. However, a significant interaction was observed according to surgical approach. In robot‐assisted surgery, the operative time was shorter in the IA group than that in the EA group, whereas IA was associated with longer operative time in laparoscopic surgery (Figure 3). The direction of effect for adequate resection margins generally favored IA across most subgroups, without evidence of significant effect modification (Figure 4). Subgroup analysis of blood loss ≥ 200 mL. BMI, body mass index; CI, confidence interval; EA, extracorporeal anastomosis; IA, intracorporeal anastomosis; RD, risk difference. Early and late represent patients who underwent surgery during 2018–2021 and 2022–2025, respectively. Institutional IA volume was dichotomized at the median cumulative number of IA procedures performed per institution during the study period (low‐volume: 21–52 cases; high‐volume: 54–222 cases). The procedure subgroup analysis included only patients who underwent right or left colectomy; patients who underwent partial colectomy were excluded. p values represent tests for interaction between subgroup variables and anastomotic approach. RDs were calculated with EA as the reference group (IA–EA). Error bars indicate 95% CIs. Subgroup analysis of operative time. BMI, body mass index; CI, confidence interval; EA, extracorporeal anastomosis; IA, intracorporeal anastomosis. Early and late represent patients who underwent surgery during 2018–2021 and 2022–2025, respectively. Institutional IA volume was dichotomized at the median cumulative number of IA procedures performed per institution during the study period (low‐volume: 21–52 cases; high‐volume: 54–222 cases). The procedure subgroup analysis included only patients who underwent right or left colectomy; patients who underwent partial colectomy were excluded. p values represent tests for interaction between subgroup variables and anastomotic approach. Median differences were calculated with EA as the reference group (IA–EA). Error bars indicate 95% CIs. Subgroup analysis of adequate resection margin. BMI, body mass index; CI, confidence interval; EA, extracorporeal anastomosis; IA, intracorporeal anastomosis; RD, risk difference. Adequate resection margins were defined as both proximal and distal margins ≥ 100 mm. Early and late represent patients who underwent surgery during 2018–2021 and 2022–2025, respectively. Institutional IA volume was dichotomized at the median cumulative number of IA procedures performed per institution during the study period (low‐volume: 21–52 cases; high‐volume: 54–222 cases). The procedure subgroup analysis included only patients who underwent right or left colectomy; patients who underwent partial colectomy were excluded. p values represent tests for interaction between subgroup variables and anastomotic approach. RDs were calculated with EA as the reference group (IA–EA). Error bars indicate 95% CIs. Learning Curve Analysis The median operative time differences between IA and EA were 14 min (95% CI, 0.5–27 min), 28 min (95% CI, −7 to 61.5 min), −12 min (95% CI, −47 to 40 min), and −6 min (95% CI, −39.5 to 31 min) in the groups with 0–49, 50–99, 100–149, and ≥ 150 prior institutional IA procedures, respectively (Figure S4). Discussion To the best of our knowledge, to date, this is the largest study to evaluate the operative and short‐term outcomes of IA and EA in a cohort that included robot‐assisted surgery. In this study, IA was associated with lower incidences of intraoperative blood loss ≥ 200 and ≥ 100 mL. Similar trends were observed for blood loss ≥ 300 mL and intraoperative transfusion, although these differences did not reach statistical significance. In contrast, continuous intraoperative blood loss was comparable between the two groups. Given that the median blood loss was 0 mL in both groups, the majority of patients experienced little intraoperative bleeding regardless of anastomotic technique. Nevertheless, the lower incidences of blood loss ≥ 200 mL observed in the IA group suggest that IA may reduce the occurrence of relatively high‐blood‐loss cases rather than decrease blood loss uniformly across the entire cohort. One possible explanation is that IA avoids bowel exteriorization and therefore minimizes mesenteric traction. During EA, bowel exteriorization may place tension on mesenteric vessels and fragile mesenteric tissue, potentially increasing the risk of vessel injury and bleeding. In contrast, IA allows mesenteric manipulation, bowel resection, and anastomosis to be completed intracorporeally, which may reduce mechanical stress on mesenteric structures. Previous studies have also demonstrated substantial anatomical variation in the venous drainage of the colon, particularly in the right colon [19, 20, 21, 22, 23]. Although such anatomical complexity may increase the risk of inadvertent vessel injury during colectomy, the present study was not designed to evaluate the mechanisms underlying the observed association. Therefore, these explanations remain speculative. Previous studies have suggested that excessive intraoperative blood loss is associated with worse disease‐free survival and a higher incidence of distant metastasis in colorectal cancer surgery [24, 25]. However, the present study did not evaluate long‐term oncologic outcomes, and therefore no conclusions regarding the oncologic implications of IA can be drawn. Although our previous analysis of the KCRG database demonstrated favorable recurrence‐related outcomes among patients undergoing IA [26], that study was also retrospective in nature. Accordingly, prospective studies, including ongoing randomized trials evaluating long‐term oncologic outcomes of IA and EA, are required to determine whether IA has any impact on recurrence or survival [27]. Operative time was generally longer in the IA group than that in the EA group. However, subgroup analysis demonstrated significant effect modification according to surgical approach, and operative time was shorter in the IA group during the robot‐assisted surgery, although this finding should be interpreted cautiously given the exploratory nature of subgroup analyses. Robot‐assisted surgery provides several technical advantages, including a three‐dimensional operative view, stable camera control, and highly articulated instruments, which facilitate meticulous and precise anastomosis even within the confined space of the abdominal cavity [28, 29]. Furthermore, unlike conventional laparoscopic surgery, an assistant is not required to perform counterintuitive mirror‐image manipulation during anastomosis. These advantages may contribute to a reduction in the time required for IA. Additionally, because IA reduces the need for extensive bowel mobilization and re‐insufflation, the operative time may eventually become shorter than that of EA. However, robot‐assisted procedures may be preferentially performed by surgeons with greater experience in minimally invasive colorectal surgery, and therefore the influence of surgeon expertise on the observed findings cannot be excluded. The current study suggests a potential institutional learning curve effect for IA. The difference in operative time between IA and EA diminished with increasing institutional IA experience. Although operative time remained longer in the IA group during the early phase of IA adoption, the magnitude of this difference decreased as institutional experience accumulated. However, because individual surgeon experience was not available, the relative contributions of institutional experience and surgeon experience could not be determined. Regarding resection margins, PM was comparable between the two groups, whereas DM was significantly longer in the IA group. Although the direction of effect for adequate resection margins favored IA, the difference did not reach statistical significance. Furthermore, harvested lymph node yield was comparable between the two groups. Although IA was associated with a longer DM, no corresponding increase in harvested lymph node yield was observed, suggesting that the clinical significance of the observed difference in DM may be limited. In this study, the incidence of postoperative complications was generally comparable between the two groups, which is consistent with previous findings [8, 9, 10]. However, although the difference was not statistically significant, organ/space SSI tended to occur more frequently in the IA group. One possible explanation is that intracorporeal bowel opening during IA may increase the risk of bacterial contamination within the abdominal cavity. Nevertheless, given the small number of events and the lack of statistical significance, this finding should be interpreted with caution. Previous studies have demonstrated that preoperative oral antibiotic preparation can reduce the occurrence of SSI [30]. In majority of the institutions included in this cohort, preoperative oral antibiotic preparation, in conjunction with mechanical bowel preparation, was routinely administered before minimally invasive colectomy. However, because this was a retrospective cohort study and perioperative management was not standardized by protocol, differences in bowel preparation and other perioperative practices may have influenced postoperative complication profiles. Several previous studies have suggested that IA may reduce postoperative ileus by minimizing bowel traction and manipulation [5]. However, such an association was not observed in the present study. Differences in perioperative management and patient selection may partly explain the discrepancy between studies. This study has several limitations. First, the results may have been influenced by residual confounding owing to unmeasured variables, including surgeon‐related factors. Although institutional IA volume and study period were incorporated into the propensity score model and additional mixed‐effects analyses were performed, detailed information regarding surgeon experience and technical proficiency was unavailable. Given the retrospective design of this cohort study, no predefined criteria were established for selecting IA or EA, and the choice of procedure was at the discretion of the operating surgeon during routine clinical practice. Consequently, IA may have been performed more frequently by highly experienced surgeons, potentially affecting the outcomes. Second, the analysis was confined to complete cases without missing data, which may have introduced a selection bias, although the proportion of excluded cases due to missing data were less than 10%. Third, this study lacked detailed information regarding intraoperative bleeding including its cause, anatomical source, and timing. Therefore, the mechanisms underlying the association between IA and lower incidences of intraoperative blood loss ≥ 200 mL could not be directly evaluated. Finally, the study population was limited to patients treated at the institutions that participated in the database. Additionally, a large proportion of cases from early‐period and low‐IA‐volume institutions were excluded after propensity score matching; therefore, the overall results may mainly reflect outcomes in later‐period and higher‐IA‐volume settings. These factors may have limited the generalizability and transportability of the findings. In conclusion, IA was associated with a lower incidence of intraoperative blood loss ≥ 200 mL during minimally invasive colectomy without compromising resection quality or postoperative outcomes. These findings indicate that IA may reduce the occurrence of relatively high blood loss events. Further large‐scale prospective studies are warranted to validate these findings and clarify the long‐term outcomes associated with IA. Author Contributions Takuya Takami: conceptualization, methodology, data curation, validation, formal analysis, writing – original draft, writing – review and editing, investigation. Keiji Nagata: writing – review and editing. Kae Okoshi: writing – review and editing. Ryosuke Mizuno: writing – review and editing, resources. Kazutaka Obama: supervision, writing – review and editing, writing – original draft, conceptualization, project administration. Yasunori Deguchi: writing – review and editing. Rei Mizuno: writing – review and editing. Yoshiro Itatani: conceptualization, methodology, software, validation, supervision, writing – original draft, writing – review and editing, project administration, investigation. Tomoaki Okada: writing – review and editing. Takehito Yamamoto: conceptualization, methodology, software, validation, writing – review and editing, writing – original draft, supervision, investigation. Funding The authors have nothing to report. Ethics Statement The protocol for this study was approved by the Ethics Committee of the institution and conformed to the provisions of the Declaration of Helsinki. Ethics Committee of the Kyoto University Graduate School and Faculty of Medicine, Approval No. R0286. Consent Informed consent was obtained through an opt‐out form on the website. Patients who declined to participate were excluded. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Table S1: Characteristics of analysis patients and patients excluded because of missing data. Table S2: Characteristics of matched and unmatched patients. Figure S1: Covariate balance before and after propensity score matching. Figure S2: Institution‐level distribution of IA and robot‐assisted surgery proportions. Figure S3: Surgical volume, IA proportion, and robot‐assisted surgery proportion according to institutional IA volume and study period. Figure S4: Association between institutional IA experience and operative time.