Abstract
Background
Transanal TME (taTME) combines abdominal and transanal dissection to facilitate sphincter preservation in patients with low rectal tumors. Few phase II/III trials report long-term oncologic and functional results. We report early results from a North American prospective multicenter phase II trial of taTME (NCT03144765).
Methods
100 patients with stage I–III rectal adenocarcinoma located ≤ 10 cm from the anal verge (AV) were enrolled across 11 centers. Primary and secondary endpoints were TME quality, pathologic outcomes, 30-day and 90-day outcomes, and stoma closure rate. Univariable regression analysis was performed to assess risk factors for incomplete TME and anastomotic complications.
Results
Between September 2017 and April 2022, 70 males and 30 females with median age of 58 (IQR 49–62) years and BMI 27.8 (IQR 23.9–31.8) kg/m2 underwent 2-team taTME for tumors located a median 5.8 (IQR 4.5–7.0) cm from the AV. Neoadjuvant radiotherapy was completed in 69%. Intersphincteric resection was performed in 36% and all patients were diverted. Intraoperative complications occurred in 8% including 3 organ injuries, 2 abdominal and 1 transanal conversion. The 30-day and 90-day morbidity rates were 49% (Clavien–Dindo (CD) ≥ 3 in 28.6%) and 56% (CD ≥ 3 in 30.4% including 1 mortality), respectively. Anastomotic complications were reported in 18% including 10% diagnosed within 30 days. Higher anastomotic risk was noted among males (p = 0.05). At a median follow-up of 5 (IQR 3.1–7.4) months, 98% of stomas were closed. TME grade was complete or near complete in 90%, with positive margins in 2 cases (3%). Risk factors for incomplete TME were ASA ≥ 3 (p = 0.01), increased time between NRT and surgery (p = 0.03), and higher operative blood loss (p = 0.003).
Conclusion
When performed at expert centers, 2-team taTME in patients with low rectal tumors is safe with low conversion rates and high stoma closure rate. Mid-term results will further evaluate oncologic and functional outcomes.
Graphical abstract
Similar content being viewed by others
Avoid common mistakes on your manuscript.
In conjunction with major advancements in neoadjuvant treatment strategies, worldwide adoption of total mesorectal excision (TME) technique in the curative resection of locally advanced rectal cancer over the past 3 decades has reduced 5-year local recurrence (LR) rates to 3.3–5.7% in large contemporary series [1,2,3,4,5]. Laparoscopy was a major advance in innovation for the surgical treatment of rectal cancer, although acceptance and adoption have been slowed by long learning curves and lingering concerns regarding inferior rate of circumferential resection margins (CRM) relative to open TME. By providing minimally invasive access to the pelvis, laparoscopic and robotic-assisted TME have improved short-term postoperative recovery without significantly impacting oncologic or functional outcomes relative to open proctectomy. As reflected in recent comparative trials, persistently high abdominoperineal resection (APR), CRM positive and conversional rates reported in male and obese patients relate to tumor location ≤ 6 cm from the anal verge [1, 6, 7]. Transanal TME (taTME) evolved from NOTES (Natural Orifice Transluminal Endoscopic Surgery) and is most commonly performed in a hybrid fashion in combination with abdominal laparoscopic assistance. Since the report of the first case in 2010, rapid adoption of taTME worldwide reflects the perceived benefits of direct in-line access to the distal rectum, augmented exposure, and navigation provided by pneumodistention and image guidance through multiport transanal endoscopic platforms in facilitating these complex procedures [8,9,10,11]. Large retrospective institutional and multicenter cohort studies have reported procedural and short-term oncologic results commensurate with those from laparoscopic trials, with notably lower conversion rates and higher rates of sphincter preservations [12, 13], but also a non-negligible incidence of procedure-specific adverse events including urethral injury and CO2 embolism [14,15,16,17]. Several national cancer audits, cohort studies, and the first recently published randomized trial of laparoscopic vs taTME have reported 3-year local recurrence rates ranging from 1.9% to 6.2% [18,19,20,21,22,23,24,25,Sample size and statistical analysis The primary objective of this study was to determine whether taTME was non-inferior to standard LAR in terms of the proportion of subjects that achieve complete or near-complete mesorectal excision. With a sample size of 100, the one-sided binomial test will reject the null hypothesis that the success rate is ≤ 80% if the study procedure leads to efficacy of the total mesorectal excision for 87 or more subjects. This design achieves a power of 87% using one-sided binomial test for non-inferiority with 5% type 1 error assuming the true success rate is 90%. Patient demographic, disease-related, treatment-related, operative characteristics as well as pathologic outcomes and 90-day surgical complications were summarized for continuous variables as median and first and third quartiles, Q1–Q3, and for categorical variables as counts and percentages. Distributions of categorical variables were compared among group using the χ2 or Fisher’s exact test when appropriate. Univariable log-binomial regression was used to investigate associations between some of the aforementioned characteristics, identified through an extensive literature search as potential risk factors, and incomplete pathologic TME grade, reporting relative risks (RRs), corresponding 95% confidence intervals (CIs), and p-values. RRs were selected instead of odds ratios via logistic regression because the latter tends to overestimate the strength of the association when the incidence of the outcome is 10% or more, as it is in this study. Times from surgery to stoma closure were analyzed using the Kaplan–Meier (KM) method. Comparison of stoma closure KM distributions was made between groups with the log-rank test. Eighty-two patients were censored for anastomotic leak at the time of their ileostomy reversal or permanent colostomy creation. Two patients were censored for stoma closure at their permanent colostomy creation. Duration of follow-up for postoperative complications was calculated as the maximum of 90 days and days to ileostomy reversal or permanent colostomy creation, except for in one patient whose follow-up was only 70 days due to death. Univariable Cox proportional hazards regression was used to investigate associations between characteristics identified through literature search as potential risk factors for anastomotic leak and/or stoma closure, reporting hazard ratios (HRs), corresponding 95% confidence intervals (CIs), and p values. All statistical analyses were performed using SAS statistical software (version 9.4; SAS Institute). Hypothesis testing was conducted at the 5% level of significance. However, p values less than 10% were considered borderline significant in exploratory regression analyses.
Results
Accrual and Demographics
Between September 2017 and April 2022, 137 patients were enrolled across 11 sites in the United States and Canada. There were 37 screen failures due to exclusion criteria, withdrawal of consent or loss to follow-up. In 8 cases where clinical complete response was achieved post-neoadjuvant therapy, patients withdrew from the study to enter watch and wait protocols (Fig. 1). In total, 100 patients underwent taTME procedures with total accrual ranging from 2 to 17 subjects per study site (Fig. 2). Below target patient accrual at 5 study sites was due to site PI relocation, screen failures, and COVID-related interruptions. Between March 2020 and February 2021, COVID-related trial disruptions resulted in delays in accrual, in-person follow-up intervals, surveillance, and ileostomy reversal (Fig. 2). Among 100 taTME patients, 70 were male vs 30 female, 22% self-identified as other than Caucasian, median age was 58 (IQR 49–62), and median BMI was 27.8 (IQR 23.9–31.8) with 33% of patients with BMI ≥ 30 (Table 1).
Trial Accrual, bar graph with cumulative number of patients accrued on the taTME trial at 2-month intervals and line graph demonstrating the rate of accrual per 2 months
Tumor characteristics
Median tumor height was 5.8 cm from the anal verge (IQR 4.5–7.0). Tumors were preoperatively staged as clinical stage I (35%), stage II (22%), or stage III (43%). Among stage I tumors, 8 cT1N0 or cT1/2N0 patients underwent completion taTME after local excision and uncovered high-risk histopathological features (poor differentiation, lymphovascular invasion, high tumor budding, and positive margin) warranting radical resection based on MDT consensus recommendations (Table 1). Neoadjuvant treatment was completed in 71% of patients, including 64 of the 65 stage II/III tumors and 7 stage I (T2N0) tumors. A significantly higher proportion of tumors located ≤ 6 cm from the AV were treated with neoadjuvant therapy than tumors > 6 cm (76.4% vs 25.4%, p = 0.011). Neoadjuvant regimens included long-course CRT in 51/71 (71.8%), TNT in 17/71 (23.9%), chemotherapy only in 2/71 (2.8%), and SCRT in 1/71 (1.4%). TNT strategy was employed by 6 of 11 sites during the study period, with 5 patients between 2017 and 2019 compared with 12 patients between 2021 and 2022 receiving TNT. Among 11 stage III, 5 stage II, and 1 stage I treated with TNT, 64.7% vs 35.3% received induction vs consolidation TNT. Overall, 80.3% (57/71) of tumors were re-staged post-neoadjuvant treatment. Median time between the end of neoadjuvant radiation and taTME was 76 (IQR 68–82) days (Table 1).
Surgical procedures
All procedures were completed using a 2-team approach with abdominal dissection performed by multiport (94%), robotic (4%), single port (1%), or hand-assisted laparoscopy (1%) with splenic flexure mobilization performed in 94 cases. Concurrent surgical procedures were performed in 8 patients (Table 2). Transanal platforms used included rigid (26%) or disposable endoscopic platforms (74%). In 38 cases where tumors were located a median of 4 cm from the AV, transanal mucosectomy (2) or partial intersphincteric resection (ISR, 36) was performed in combination with taTME. There were 11 intraoperative complications reported among 8 patients including 2 abdominal conversions to open surgery (ischemic colonic conduit, bleeding), 1 transanal conversion to abdominal robotic dissection (failure to progress), 1 suspected case of CO2 embolus, 2 vaginal injuries during transanal dissection, and 1 colotomy, 1 enterotomy, and 1 ureteral transection during abdominal dissection, all identified and repaired. Perfusion assessment of the colonic conduit and/or colorectal anastomosis was performed by fluorescence angiography using indocyanine green (ICG) in 55% (Table 2). Inadequate perfusion was observed in 9 cases and prompted intervention in 8 cases (change in proximal resection margin with additional mobilization, suture removal). TME specimens were extracted transanally in 52%, with handsewn coloanal anastomosis performed in 54%. Anastomotic reconstruction was completed end-end (89%), side-to-end (6%), or with coloanal J pouch (5%). All patients were diverted with a loop ileostomy and pelvic drains were placed in 80%. Median total operative time was 311.5 (IQR 265.5–380.5) minutes and median transanal dissection time was 101 (IQR 74.5–174) minutes. Median estimated blood loss was 100 (IQR 50–250) ml and blood transfusion was not required.
30- and 90-day postoperative outcomes
Median length of hospital stay was 5 (IQR 4–8) days. In total, 71 complications were reported in 49 patients within 30 days of surgery (CD3/4 in 28.6%) with a 17% readmission rate (Table 3, Fig. 3). Postoperative ileus with dehydration with or without high stoma output accounted for 35.2% (25/71) of 30-day complications and 47.1% (8/17) of readmissions. Ileus and high stoma output were associated with transient acute kidney injury in 4 (16%) patients. Urinary retention was reported in 19% with a 21.4 vs 13.3% incidence in male vs female patients (p = 0.344). Urinary retention was managed with medication in 68% of patients and resulted in prolonged bladder catheterization beyond discharge in 21% of patients. Other morbidities included anastomotic (10%), infectious (6%), and thrombotic (3%) complications, small bowel obstruction (SBO, 4%), pancreatic fistula (1%), bleeding duodenal ulcer (1%), myocardial infarction (1%), and intractable rectal pain (1%). Reoperation was required in 11/49 patients (22.4%), for anastomotic complications (n = 8), SBO (n = 2) and refractory rectal pain (n = 1). Between postoperative day 30 and 90, 11 new complications were recorded in 11 patients (CD1/2 in 72.7%, CD3/5 in 27.3%) including 1 death from suicide at POD70, 5 new anastomotic complications, 1 ileus with dehydration, 1 SBO, 1 DVT, and 1 stomal and 2 mucosal prolapse. Reoperation was required in 2/11 (18.2%) patients for 1 anastomotic complication and 1 stoma prolapse (Fig. 3).
Postoperative complications 30- and 90-day complications with highest CD grade per patient
Anastomotic complications and stoma reversal
At a median follow-up of 5 (IQR 3–7, range 2.3–23) months, 18 anastomotic complications were reported among 17 males and 1 female (p = 0.012), and included 10 early (diagnosed POD0-POD29), 5 late (POD30-POD90), and 3 very late anastomotic complications (> 90 days). Among 10 early anastomotic complications, 8 underwent early re-intervention and 2 were managed with percutaneous drainage and/or antibiotics (Tables 4 and 5, Fig. 4a). Among the 8 that underwent early re-intervention, 5 presented with pelvic sepsis requiring urgent pelvic washout with either transanal repair of anastomotic dehiscence (n = 3), or anastomotic takedown with end colostomy (n = 2), while 3 relatively asymptomatic patients underwent early transanal closure of small anastomotic defects. Overall, resolution of 6 anastomotic complications was achieved with 0–1 reinterventions within 30 days followed by ileostomy reversal between POD86 and POD372 (Table 5). Delayed anastomotic revision with redo-coloanal anastomosis was required in 1 patient, and Hartman’s reversal was performed in another patient who underwent early anastomotic takedown with end colostomy. Both patients underwent delayed ileostomy closure between POD271 and POD304. Two patients were permanently diverted including one who underwent anastomotic takedown due to conduit ischemia, and another who was converted to a permanent colostomy due to due to a refractory anastomotic defect later complicated by peritoneal carcinomatosis (Fig. 4a and b, Table 5).
A. Management and outcomes of early anastomotic complications (< 30 days); B. Management and outcomes of late anastomotic complications (> 30 days). See Table 5 for details regarding each case. POD postoperative day; CD Clavien-Dindo; IR interventional radiology; IV intravenous; CAA coloanal anastomosis
Among late anastomotic complications diagnosed between POD30 and POD349, 3 were described as small defects or sinuses on endoscopy and resolved without any intervention, one was a pelvic abscess identified on POD37 and resolved with oral antibiotics, while one was an anastomotic stricture that required a single operative dilatation. In all 5 cases, stoma reversal was achieved between POD103 and POD250. In 3 patients with asymptomatic late anastomotic dehiscence and stricture identified on radiology and/or endoscopy, one resolved with serial dilatations with stoma closure on POD249, and 2 eventually required anastomotic revision with transanal sleeve advancement or coloanal pull-through, with delayed ileostomy closure between POD427 and POD695 (Tables 4 and 5, Fig. 4b).
Overall, 72.2% (13/18) of all anastomotic complications identified over the course of this trial required ≥ 1 surgical re-intervention including 90% of early complications, diagnosed within 30 days, and 50% of late complications, diagnosed after 30 days. Stoma closure was achieved in 80% of patients with early complications vs 100% of patients with late complications. Among interventions performed for anastomotic failure, anastomotic takedown was performed in 6 patients followed by immediate (4) or delayed reconstruction (1). Anastomotic failure resulted in permanent colostomy in 2 patients.
Anastomotic complications were reported among 21.8% (12/55) of patients who underwent intraoperative FA perfusion assessment vs 13.3% (6/45) among those who did not, and among 17.4% (8/46) vs 44% (4/9) of patients with adequate vs inadequate perfusion (Table 6). By univariable cox proportional hazards regression analysis, male sex (HR 7.52 [0.99, 56.75]; p = 0.0502) and inadequate bowel perfusion by FA (HR 3.3 [0.96, 11.32]; p = 0.0572) were associated with higher risk of anastomotic complications, but neither variable reached statistical significance (Table 6).
The overall stoma closure rate was 93% at 12 months and 98% at 23 months follow-up and median time to reversal was 154 (IQR 94–224) days (Table 4, Fig. 5). Stoma reversal rate was 89 vs 100% in patients with and without anastomotic complications, and median time to closure of 226 (IQR 140–304) days vs 140 (IQR 91–213) days, respectively (p = 0.0015). Factors contributing to delays in stoma reversal beyond 90 days were related to adjuvant treatment, anastomotic complications, COVID-related delays in elective surgery, patient preference, or other unclear reasons (Table 4).
Kaplan–Meier curve for time from taTME to stoma closure in patients with and without anastomotic complications (days)
Oncologic outcomes
Final pathologic staging demonstrated no residual disease (pT0N0) in 8 of 10 cases where completion taTME had been performed after polypectomy or local excision of a cT2 or cT1 tumor with high-risk histopathologic features. Complete pathologic response (pCR) was demonstrated in 25/71 (35.2%) of patients who received neoadjuvant treatment. The pCR rate was 41% (7/17) post TNT, 33% (17/51) post CRT, and 50% (1/2) post chemotherapy only (Table 7). The median number of harvested lymph nodes was 19 (IQR 14.5–26). Following reconciliation between site pathologists and central blinded reviewers, final TME grading was complete in 69%, near complete in 21%, and incomplete in 10% of specimens. There was no specimen perforation, but 1 case had positive CRM and 1 case had positive distal resection margin (DRM). After adjusting for the number of TME specimens with residual tumor, the positive margin rate was 3% (2/67). Adjuvant chemotherapy was completed in 46 patients, as recommended by MDT based on preoperative clinical stage or final pathology stage.
By univariable log-binomial regression analysis, ASA ≥ 3 (RR 13.50 [1.77, 102.47]; p = 0.0118), blood loss per 100 ml increase (RR 1.46 [1.14–1.87]; p = 0.0027), and longer time interval between end of neoadjuvant RT (NRT) to surgery (RR 1.07 [1.01, 1.14]; p = 0.0297, Table 8) were associated with a significant increase in the risk of incomplete TME (Table 8).
Discussion
This phase II multicenter North American trial is the most recent large prospective trial to confirm the procedural safety, favorable pathologic outcomes, and low rates of conversion achieved with taTME when performed by experienced surgeons using a 2-team hybrid approach. The 2% abdominal open conversion rate is consistent with the 0–5.1% conversion rates reported in large taTME series of patients, and was achieved despite a predominantly male cohort with the highest proportion of patients with BMI ≥ 30 reported to date (33%) and in whom 67% of tumors were located ≤ 6 cm from the AV [14, 19, 22, 45,46,47,48]. The current trial results validate prior findings that combining taTME with a laparoscopic or robotic abdominal approach enables completion of minimally invasive sphincter-preserving TME, particularly in obese patients with low rectal tumors. Intraoperative complications related to transanal dissection are rare and included 2 vaginal injuries but no rectal perforation or urethral injury, which underscores the experience of participating surgeons [15].
The primary endpoint of the trial was mesorectal TME grade, an important pathologic endpoint and strong predictor of local recurrence and disease-free survival that also serves as an accurate marker of surgical quality [34, 49,50,51]. Complete and near-complete TME grade was achieved in 90% with a 10% rate of incomplete TME and 3% rate of positive margins. These results are in line with the 1.5–11% rates of incomplete TME reported in large retrospective taTME series [22, 46, 47, 52], and the 8% and 9.3% rates reported in the laparoscopic arm of the ACOSOG RCT and the robotic arm of the ROLAAR RCT, respectively [53, 54]. But this rate is significantly higher than the recently reported 0% rate of incomplete TME in the taTME arms of the 2 laparoscopic vs taTME RCT in China and Spain [55, 56]. Relative to the RCT by Liu et al. in which tumors were similarly located (5.0 vs 5.8 cm from the AV) in patients with higher BMI (22.9 vs 27.8), our higher rate of incomplete TME may reflect a more challenging dissection in a previously radiated field, given that 69% of our taTME cohort was treated with NRT vs only 14% of the Chinese cohort. Among 10 incomplete TME specimens in our trial, 8 occurred in patients treated with neoadjuvant long-course CRT or TNT, and for every 1 week increase in time from end of NRT to surgery, the risk of incomplete TME increased by 7%.
There is a paucity of data regarding the impact of delayed time to TME following completion of NRT on TME grade, and this has not been described in prior taTME series or trials. The 2015 TIMING phase II trial assigned patients with locally advanced tumors into 4 groups treated with CRT: CRT followed by 2, 4, or 6 cycles of mFOLFOX6, effectively increasing the interval to TME by 6 up to 20 weeks [57]. While surgeons reported an increase in pelvic fibrosis dissection with longer delays to surgery, they reported no differences in technical difficulty, surgical complications, R1 resection rate, or sphincter preservation. However, the authors did not analyze the impact of longer delays on TME grade. The UNICANCER-PRODIGE23 RCT comparing standard CRT to FOLFIRINOX followed by CRT for patients with locally advanced cancer, reported no difference in mesorectal grade between cohorts (5 vs 8% rate of incomplete TME). However, given that TNT was administered as induction treatment, there was no difference in the time interval between the end of NRT and surgery, which was relatively short in both groups (54.5 and 55 days, respectively) [58]. Given the increasing adoption of TNT worldwide in an effort to increase the rate of non-operative management, the impact of this delay on perioperative morbidity as well as TME quality will need to be further examined.
In the Spanish laparoscopic vs taTME RCT where a 0% rate of incomplete TME was reported in the taTME arm, a similar proportion of patients were treated with NRT (69.1% vs 69%) but the NRT-to-TME interval was not specified. The difference in rate of incomplete TME may reflect a higher level of technical difficulty during transanal dissection based on differences in median tumor height (8.0 cm vs 5.8 cm), and in the proportion of cases where ISR was performed (0% vs 36%) in the Spanish vs North American taTME cohorts, respectively [56]. In this phase II trial, ASA ≥ 3 and increased blood loss were identified as significant risk factors for incomplete TME, with increased blood loss generally expected during more difficult pelvic dissections. Lastly, higher rates of incomplete TME grading in this trial likely reflect more stringent quality control measures used for pathological specimen assessment. Concordance in TME grading between site pathologists and blinded central reviewers was evaluated, and major discrepancies in TME grading were resolved with final grade assigned based on consensus agreement. Neither of the recent taTME RCTs specifically incorporated quality control measures to validate TME grading. Importantly, Liu et al. in reporting a 0% rate of incomplete TME acknowledged that “an overestimation bias for successful resection still remains a certain possibility” [55].
Regarding other pathologic endpoints, our 1.5% positive CRM and 1.5% positive DRM rates compare favorably with the 1.4–12.7% CRM and 0–2.5% DRM-positive rates published in large taTME series [19, 22, 23, 59] and with the 0.9%/0%, and 0.4%/12.7% DRM-positive rates in the Liu et al. and Serra et al. taTME vs lap TME RCTs, respectively [55, 56]. TaTME was combined with ISR in 36% of patients whose tumors were located at a median of 4 cm from the AV in our North American trial, compared with 14.9% and 0% of the Chinese and Spanish taTME cohorts where median tumor height was 5 and 8 cm, from the AV, respectively [55, 56]. These results highlight the role that ISR will continue to play in facilitating sphincter preservation as well as R0 resection for low rectal tumors, without adverse impact on oncologic outcomes [59,60,61,62,63,64].
This prospective trial is unique in that it provided extended capture of early and late postoperative complications following taTME, including outcomes of all anastomoses until primary stoma closure or permanent stoma creation. Overall, 71 complications among 49 patients and 82 complications among 56 patients were reported at 30 and 90 days, respectively, which is relatively high relative to other taTME publications; but the majority were minor, with severe (CD ≥ 3) complications only reported in 14 and 17 patients at 30 and 90 days, respectively. Ileus and stoma-related dehydration accounted for 35% of 30-day complications and 47% of readmissions, and were a direct consequence of loop ileostomy creation. Although published rates of diversion following sphincter-preserving taTME vary from 55.7 to 100% [14, 21, 22, 55, 65], our trial reflects standard practice in North American centers where high-risk colorectal and all coloanal anastomoses are routinely diverted [22, 53]. Fecal diversion is particularly prevalent when ISR and handsewn anastomoses are performed, and in patients receiving extended courses of radiotherapy [13, 14, 66]. Urinary retention contributed to 27% of 30-day complications and occurred at a higher rate than the reported 2–13% incidence in other taTME series [9, 21, 67], which may be related to the timing of catheter removal. The impact of taTME procedures on long-term urinary and sexual functional will be reported in a later publication.
Anastomotic complications accounted for 10% of the trial 30-day morbidity. Given the lack of agreement with standardized definitions and diagnostic criteria for anastomotic leaks, we opted to report all anastomotic complications identified and treated throughout the trial clinical follow-up period. This broader framework for reporting early and late anastomotic complications, which also incorporates stoma closure and non-closure rates, has been used by other groups when reporting outcomes of restorative proctectomy from large colorectal cancer registries, clinical trials, institutional series, and the international taTME registry [13, 14, 68,69,70,71,72]. While the reported 30-day anastomotic leak rates range between 2 and 10% in most large TME and taTME series [9, 21, 53, 55, 56], these rates are not inclusive of pelvic abscess or phlegmon adjacent to anastomoses, nor of leaks or sequelae of leaks (sinus, fistula, stricture) identified beyond 30 days of surgery. In contrast, the TME registries and trials that intentionally report early and late-occurring anastomotic complications up to 1 year post TME, report rates as high as 20% [13, 68,69,70,71]. Using this anastomotic complication reporting framework, at a median follow-up of 5 (IQR 3–7) months, anastomotic complications were identified in 10% by 30 days, 15% by 90 days, and 18% by 349 days in our cohort. The 10% early leak rate is consistent with the 6.3–11% 30-day leak rates reported in large taTME series [20, 22, 24, 28, 38, 55, 59] and the 18% overall anastomotic complication rate is similar to the 15.7%-17.7% rates reported at 5–12 months post-taTME in the international registry and other retrospective series [13, 14]. Although the incidence of anastomotic complications was higher in male patients (24.3 vs 3.3%, p = 0.0502), this did not reach statistical significance in this study, but it has been identified as a risk factor in other taTME studies [21,22,23, 26, 38, 52, 59]. The role of intraoperative perfusion assessment on anastomotic failure had not been previously reported in the setting of a prospective phase II TME trial. In the current study, intraoperative fluorescence angiography (FA) was used at 4 of the 11 study sites, with the decision to use FA made on a case-by-case basis. Perfusion assessment was reported to have altered the operative plan in 14.5% of cases, which is consistent with 2 prior retrospective taTME series demonstrating a change in planned proximal resection margin in 18 and 28.7% of patients when FA was used [45, 73]. Despite active intervention to correct inadequate perfusion identified on FA, the incidence of anastomotic complications was higher in the FA group, which likely reflects selection bias with preferential use of FA in cases where ischemia was suspected. Although documentation of adequate bowel perfusion by FA was not entirely protective against anastomotic complications, it was associated with a lower risk of anastomotic complication, although this did not reach statistical significance.
This trial highlighted emerging trends in the management of anastomotic complications among North American rectal cancer practices, including the use of transanal rescue approaches to manage early symptomatic and asymptomatic leaks, and anastomotic revision with coloanal anastomosis for refractory anastomotic complications that failed initial interventions. Among early anastomotic complications, given that all anastomoses had been diverted, some leaks were diagnosed in patients who were otherwise asymptomatic. One leak was suspected based on CRP elevation and another diagnosed on CT scan with rectal contrast performed to evaluate for possible early ileostomy closure. Whether symptomatic or not, early leaks were managed within 30 days with transanal repair attempted in 75% of reoperations. Early re-intervention led to resolution and ileostomy reversal in 80% patients with ≤ 1 additional re-intervention. All anastomotic complications diagnosed > 30 days postoperatively were managed with 0–1 major surgical intervention, and stoma reversal was achieved in all cases. Overall, 2 patients with anastomotic complications were converted to permanent colostomy due to colonic conduit ischemia and refractory anastomotic dehiscence in the setting of disease progression, respectively. There was no mortality related to anastomotic complications and 89% stoma-free rate was achieved at 2 years in patients with anastomotic complications. The trial’s overall 93% 1-year and 98% 2-year stoma reversal rates are higher than those in other trials that report long-term outcomes of sphincter-preserving TME with or without a primary fecal diversion [74]. Among 595 patients in the 2015–2017 Dutch ColoRectal Audit (DCRA) that underwent sphincter-preserving TME, the 1-year stoma closure was 81.3 with 19.7% of patients requiring a permanent stoma [72]. In a meta-analysis that included 10 studies, the pooled rate of non-closure of diverting stomas following LAR for rectal cancer was 19% (95% CI 12–24%; p < 0.001) at follow-up ranging from 13 months to 7.1 years. Risks factors for non-closure included surgical complications, anastomotic leakage, stage IV tumors, and local recurrence [60]. Long-term stoma closure rates have been reported in 7 retrospective taTME series and large registries, and range from 56.5% at 4.73 months to 89.5% at 27 months [9, 21, 22, 28, 59, 61, 62, 65]. However, long-term stoma closure rates have not been previously reported in the context of a prospective rectal cancer trial. The high stoma closure rate in the current trial is most likely attributed to early management of symptomatic leaks, which helped minimize leak-related morbidity and mortality, as well as the aggressive management for refractory anastomotic defects and sequelae. Redo-coloanal anastomosis was performed in 3 cases with refractory anastomotic dehiscence or stricture 7–18 months post-taTME with successful ileostomy reversal in all 3 cases. Delayed anastomotic reconstruction for failed anastomoses has been reported in very few small TME series [63, 64, 75]. In a recent systematic review of 9 studies and 291 patients, redo-anastomosis was performed at a median 14–41 months post the index procedure, with a 79% rate of stoma closure and 16% morbidity rate [64, 75].
A limitation of the trial concerns strict inclusion and exclusion criteria for enrollment. Similar to the two recent RCT’s of lap vs taTME, cT3b tumors with predicted positive CRM, sphincter invasion or cT4 tumors on preoperative staging were excluded, even when downstaging was achieved with neoadjuvant treatment and restaging. While this was necessary in order to establish the feasibility and safety of taTME on a homogenous cohort of resectable cancer cases, our cohort does not entirely reflect current practice where taTME has become the preferred approach for tumors with incomplete response or tumor regrowth following neoadjuvant treatment. While ongoing RCTs will validate the long-term oncologic safety of taTME and evaluate functional results relative to laparoscopic TME, it will become important to understand the role that taTME plays in contemporary rectal cancer practices in the era of robotic surgery, TNT and an increased demand for sphincter preservation among young patients.
Conclusions
In this prospective phase II multicenter North American trial, the procedural and preliminary oncologic safety of taTME was demonstrated. In a predominantly male cohort with low rectal tumors and median BMI 27.8, a 2-team taTME approach achieved acceptable rates of complete and near-complete mesorectal grade with a low rate of open conversion. Continued monitoring and management of anastomotic complications resulted in high stoma closure rates.
