Introduction

The use of minimally invasive surgical techniques is spreading worldwide [1–3]. Natural orifice transluminal endoscopic surgery (NOTES) may lead to theoretical advantages including reduced abdominal wall complications, decreased postoperative pain, shorter hospital stay, faster return to work, and improved cosmesis [4, 5]. NOTES-inspired techniques have been increasingly reported in many surgical fields, including colorectal surgery [2, 3, 6–8].

However, the diffusion of these techniques is still hindered by their poor reproducibility and fear of generating new complications due to the access. Moreover, more light needs to be shed on their oncological appropriateness in patients with colorectal cancer.

Laparoscopic rectal resections, including more specifically “up-to-down” total mesorectal excision (TME) for low rectal cancer [9–11], are particularly challenging. Our preliminary experience with NOTES rectal procedures has already been published [3]. Preservation of the anatomical integrity of the mesorectum is a major predictive factor of long-term survival and recurrence-free survival [2, 3, 7] in patients with distal rectal cancer. NOTES-inspired techniques with a “down-to-up” transanal approach to TME may offer better preservation of the lowest part of the mesorectum [12–14].

We report the results of a prospective study in patients with low rectal cancer comparing “down-to-up” transanal TME (NOTESTME) to laparoscopic transabdominal TME (LAPTME). The primary endpoint was the macroscopic quality of the operative TME specimen. Secondary endpoints included intraoperative visualization of the sacral nerves, operating time, status of resection margins, number of retrieved nodes, and inadvertent tumor perforation.

Materials and methods

Patients

From January 2011 to December 2014, adult patients with rectal adenocarcinoma within 7 cm from the dentate line were considered for the study. Exclusion criteria comprised body mass index (BMI) > 40 kg/m2, an American Society of Anesthesiologists (ASA) score > III, prior open abdominal surgery, major organ insufficiency, peritoneal carcinomatosis, metastatic disease, T4 on ultrasound or MRI, or bowel obstruction.

Preoperative assessment included digital pelvic examination, endorectal ultrasound (ERUS), computed tomography (CT) scan of the chest and abdomen, and pelvic magnetic resonance imaging (MRI) scan. In patients with T stage higher than T2 on ultrasound/MRI or in case of suspected nodal involvement, neoadjuvant chemoradiation was administered consisting of 45–50 Gray radiation in 5–6 weeks, with continuous fluorouracil (200–225 mg/m2 daily). Surgery was scheduled 8–12 weeks after the last session of radiation.

The study protocol was approved by the local ethics committee. All patients signed an informed consent form. Data collection was prospective. Assignment to either group was sequential and based on the rank of inclusion in the study. Odd-number patients were assigned to the NOTESTME group, and even-number patients were assigned to the LAPNOTES group.

Technique

The NOTESTME consisted of a transanal minimally invasive surgery (TAMIS)-based technique using a single-port access for “down-to-up” TME. Under direct endorectal vision, a 2–0 Vicryl purse-string closure was performed in order to avoid potential spillage of tumor cells and maintain retropneumoperitoneum. The circumferential full-thickness incision was realized 10–15 mm above the dentate line unless lower tumor extension was suspected. In such cases, the incision line was made at the level of the dentate line itself followed by a frozen section of the distal margin. In the case of positive margins, an abdominoperineal resection (APR) was indicated. Such patients were excluded from the study, along with those having locally advanced tumors (i.e., UT4 or MRIT4).

Starting from the anatomical intersphincteric plane, dissection was carried cephalad in order to make enough room for the introduction of the single-port platform (GelPOINT path, Applied Medical, California, USA, or the SILSport, Covidien, Minneapolis, USA). Using an articulating video laparoscope (Endoeye, Olympus, Japan), a sealing device, and conventional laparoscopic instruments, the “down-to-up” TME begins by posterior dissection of the rectosacral ligament which leads to the presacral space, “the holy plane.” After extending the dissection posterolaterally, the dissection of the anterior mesorectum is achieved precisely on the midline. Careful anterolateral dissection at the level of the prostato-seminals junction facilitates the identification of the neurovascular bundles of Walsh because of the exposure of small vessels penetrating the prostate (the prostatic capsular artery and veins). The plane of dissection intimately follows the anterolateral aspects of the mesorectum, running posterior to the bundles of Walsh and anteriorly to Denonvilliers’ fascia. In women, the bundles of Walsh are identified at the level of junction of the lower and middle thirds of the vagina. After division of the lateral rectal ligaments, TME is performed circumferentially toward the peritoneal reflection. The procedure can then be completed either exclusively transanally (i.e., pure NOTES) or with additional laparoscopic assistance for the intraabdominal dissection.

In the LAPTME group, the primary approach was abdominal with transumbilical, single port or standard laparoscopic “up-to-down” dissection.

In the NOTESTME group, conversion was defined by the need of an abdominal approach in order to complete the procedure. In such cases, the subsequent colonic mobilization was always performed using the abdominal approach. In the LAPTME group, conversion was defined by the addition of laparotomy or transanal laparoscopic assistance in order to complete rectal dissection.

All coloanal anastomoses were hand-sewn, favoring side-to-end over end-to-end anastomosis. All patients had a diverting loop ileostomy, except some of those with UT1 N0 or UT2 N0 tumors who did not receive preoperative chemoradiation.

All the procedures were performed by a single senior surgeon (EC) who has extensive expertise in both minimally invasive and colorectal surgeries.

The primary endpoint was the quality of the resected TME specimen as assessed by the pathologist (AR), based on the classification of Quirke et al. [15]. Specimens were classified into three categories, depending on whether they were considered as good, moderate, or poor (Table 1).

Table 1 Grading of the TME specimen as assessed by the pathologist (AR) based on the classification of Quirke et al. [15]
Full size table

Secondary endpoints included operative, bilateral visualization of the neurovascular bundles of Walsh, operating time, and oncological parameters including gross or microscopic tumor perforation, status of distal and circumferential margins, and the number of harvested lymph nodes.

Data were collected prospectively. All the procedures were entirely video-recorded.

The study was conducted on an intention-to-treat basis. Statistical analysis was performed with the SPSS software package. Data were analyzed by using Fisher’s exact test for categorical values and Student’s t test for continuous variables. All reported p values were two-tailed, and a p value of <0.05 was considered statistically significant.

Results

Initially, we enrolled 18 patients in the NOTESTME group and 17 patients in the LAPTME group. However, 2 patients were dropped from the LAPTME group because of the preoperative death due to myocardial infarction of one patient and a intraoperative diagnosis of peritoneal carcinomatosis in the other.

The study included 18 patients in the NOTESTME group (12 women, 6 men) and 15 patients (8 women, 7 men) in the LAPTME group.

Mean age was 55.4 ± 11.9 years (range 31–79 years) in the NOTESTME group and 57.8 ± 7.4 years (range 39–66 years) in the LAPTME group (p = 0.48). Mean BMI was 27.1 ± 4.5 kg/m2 (range 21–38 kg/m2) in the NOTESTME group and 29.0 ± 4.2 kg/m2 (range 23–36 kg/m2) in the LAPTME group (p = 0.22). Preoperative patients’ characteristics are summarized in Table 2.

Table 2 Summary of preoperative patients’ characteristics
Full size table

No conversion to open surgery occurred. In the NOTESTME group, transabdominal assistance for rectal mobilization was necessary in 8 patients (44.4 %) including 4 with standard, multiport laparoscopy and 4 with single-port laparoscopy. The causes of the abdominal assistance in NOTESTME were difficult dissection, poor exposure, and/or difficult colon mobilization but never bleeding. This was mostly encountered at the level of the mid-portion of the mesorectum. Six of these conversions occurred during the first 10 NOTESTME procedures. In the remaining 10 patients of the NOTESTME group, the procedure was concluded exclusively transanally (i.e., pure NOTES).

In the LAPTME group, transanal assistance to complete the TME under the peritoneal reflection was necessary in 4 patients (26.7 %). The cause of additional assistance was always poor abdominal exposure of the lowest part of the mesorectum.

Mean operative time was 245 ± 66 min (range 155–440 min) in the NOTESTME group and 275 ± 58 min (range 180–400 min) in the LAPTME group, respectively (p = 0.18). Gross tumor perforation was noticed during dissection in one patient in each group (p = 1). No intraoperative transfusion or injury of surrounding organs occurred in any patient in either group. Operative details are summarized in Table 3.

Table 3 Summary of operative parameters
Full size table

The quality of the TME specimen as evaluated by the pathologist based on the classification of Quirke et al. (Table 1) was comparable between the two groups (p = 0.35). The quality of the resected specimen was considered as good or moderate in 16 patients with NOTESTME (88.9 %) and in 11 patients who had LAPTME (73.3 %). In intergroup comparisons, the quality of the specimen was comparable whether or not a conversion was needed to complete the procedure (p = 0.41 and p = 0.24 in intergroup analyses in NOTESTME and LAPTME, respectively). Intergroup, NOTESTME intragroup and LAPTME intragroup comparisons of the macroscopic and microscopic anatomical parameters are summarized in Tables 4, 5 and 6, respectively.

Table 4 Intergroup comparison of macroscopic and microscopic anatomical parameters
Full size table
Table 5 NOTESTME intragroup comparison of macroscopic and microscopic anatomical parameters
Full size table
Table 6 LAPTME intragroup comparison of macroscopic and microscopic anatomical parameters
Full size table

Tumor perforation was microscopically confirmed in 2 patients in the NOTESTME group and in 1 patient in the LAPTME group (p = 1). Histological analysis of the specimen revealed negative distal margins in all patients. The mean distal margin in NOTESTME group was 32.5 ± 16 mm (range 10–68 mm) in comparison with 36.1 ± 13.9 mm (range 11–59 mm) in the LAPTME group (p = 0.50). Differences did not reach statistical significance in either of the intragroup comparisons. The mean radial margin was 11.4 ± 6 mm (range 0–19 mm) in the NOTESTME group and 13.7 ± 8.3 mm (range 0–27 mm) in the LAPTME group (p = 0.38). Intergroup and intragroup comparisons did not reach statistical significance for either the rate of tumor perforation on microscopic examination or the status of the distal margin on the specimen.

Similarly, median number of retrieved lymph nodes and overall specimen length were comparable between and within the NOTESTME and LAPTME groups.

During surgery, the neurovascular bundles of Walsh were visualized in 14 patients in the NOTESTME group (77.8 %) and in 5 patients in the LAPTME group (33.3 %). This difference was statistically significant (p = 0.02). These structures were identified in 9 out of 10 patients (90 %) who had “pure” NOTESTME compared to 5 out of 8 patients (62.5 %) in converted NOTESTME (p = 0.27). Nerves were identified in 2 out of 11 patients who had “pure” LAPTME (18.2 %) compared to 3 out of 4 patients (75 %) who had LAPTME with additional transanal assistance (p = 0.08). The mortality was nil in both groups. The overall morbidity was 16.7 % in the NOTESTME group and 20 % in the LAPTME group (p > 0.05). One anastomotic leak was found in each group and was treated conservatively. Both patients with an anastomotic leak had had loop ileostomy during the initial TME procedure.

In the NOTESTME group, 1 patient developed urinary infection and 1 patient developed small bowel obstruction at the ileostomy site. Both were treated conservatively. In the LAPTME group, 1 patient developed transient renal insufficiency.

The mean length of hospital stay was 10.4 days in the NOTESTME group (range 4–29 days) and 9.4 days in the LAPTME group (range 6–19 days), (p > 0.05).

Discussion

Our study showed that NOTESTME performed with readily available, conventional laparoscopic instruments, with or without additional abdominal laparoscopic assistance, is a reliable surgical alternative for LAPTME in selected patients with rectal tumors. Moreover, as compared to standard laparoscopy, this NOTES-based approach could allow a better definition of the neurovascular bundles of Walsh’s anatomy. Overall, nerves were significantly more often identified in the NOTESTME group as compared to the LAPTME group (77.8 and 33.3 %, respectively, p = 0.02). Even in the pure LAPTME group which represents the optimal conditions (otherwise transanal assistance would have been added), the rate of nerve identification remains low (18 %). However, the sacral nerves were identified in 90 % of pure NOTESTME. This difference was statistically significant (p = 0.002). When the transanal approach was added for difficult up-to-down lower mesorectal dissection, the nerves could be identified in 75 % of cases versus 18.2 % with “pure” LAPTME (p = 0.08). Conversely, a lower rate of nerve identification was seen in laparoscopically assisted NOTESTME (62.5 %) as compared to pure NOTESTME (90 %), p = 0.27. Although they did not reach statistical significance, most probably due to the small sample sizes, these findings suggest that in all local conditions (easy dissection or poor exposure) and both primary approaches (abdominal or transanal), NOTESTME seemed to enhance the identification of the pelvic nerves.

Another indicator of better operative definition of anatomy with NOTESTME is the quality of the resected specimen in nearly 90 % of patients (of which 87.5 % had good specimen grading). Less than 73 % of patients who had pure LAPTME had such results. Additionally, the transanal approach seems to have made it easier to obtain an appropriate specimen quality in difficult up-to-down LAPTME. However, this difference did not reach statistical significance probably because of the limited number of patients in each group. Microscopic analysis demonstrated negative distal margins in all patients. These findings may encourage large-scale studies in order to validate the role of the NOTESTME in reducing distal margin positivity in patients with distal cancer.

Transanal TME for patients with rectal cancer is not totally new. Many reports in the literature [12, 14, 16, 17] have already highlighted this increasingly popular approach. Dozens of patients were enrolled in these studies assessing transanal TME with abdominal assistance, using either standard or single-port laparoscopy. TME, however, remains a very challenging procedure, especially in case of a narrow pelvis, previous irradiation or surgery, advanced tumors, and obesity. Besides the expertise of the surgeon, the main quality indicators of a TME include nerve identification, integrity of the rectum, quality of the mesorectum, and node harvesting. In this prospective study, the NOTESTME group was comparable to the LAPTME group in terms of length of specimen (26.2 ± 6.8 and 29.3 ± 16.5 cm, respectively p = 0.49), quality of the specimen (good or moderate in 88.9 and 73.3 %, respectively, p = 0.35), microscopically confirmed tumor perforation (11.1 and 6.7 %, respectively, p = 1), mean radial margin (11.4 ± 6 and 13.7 ± 8.3 mm, respectively, p = 0.38), and number of harvested lymph nodes (10.8 ± 4.3 and 12.3 ± 9.9, respectively, p = 0.61). It is worthy of note that conversion from one technique to the other did not lead to significant differences in these parameters. Finally, both techniques were associated with comparable operative mortality and morbidity. Based on these data, the NOTESTME appears to fulfill the requirements for appropriate oncological resection.

Moreover, with growing experience, the NOTESTME potentially offers several advantages for anatomical dissection. The LAPTME places the hands and the eyes of the surgeon dozens of centimeters away from the operative site. On the other hand, the NOTESTME offers a magnified view and shorter focal length between the operator’s eye and the lowest, critical part of the mesorectum and the sacral nerves.

The pelvic nerves consist of sympathetic and parasympathetic components, which together play a major role in urinary and sexual function. The risk of pelvic nerve injury during surgery depends on the surgical technique and the extent of the underlying disease. Neoadjuvant radiation increases the risk of nerve injury, as dissection planes are more difficult to identify. The damage to sympathetic nerves usually occurs during ligation of the inferior mesenteric artery at its origin. Conversely, the parasympathetic nerves are typically injured during the perineal dissection or lateral pelvic wall dissection when mistakenly driven too far behind the Waldeyer’s fascia. In men, the risk of nerve injury is higher during anterior rectal dissection at the base of prostate where the neurovascular bundles of Walsh run [18, 19]. In NOTESTME, the enhanced anatomical definition may lead to better preservation of the parasympathetic nerves, despite the reduced space. As for the sympathetic nerves, the sacral promontory often represents a “hedge” preventing excessively proximal ligation of the inferior mesenteric artery in “pure” NOTESTME [20].

In this series, 10 of the 18 NOTESTME procedures were performed entirely transanally. In addition to the difficulties encountered during rectal dissection, NOTESTME is associated with new technical challenges during intraabdominal mobilization of the colon. The use of an articulating laparoscope is of paramount importance in order to overcome the sacral promontory, the main midline obstacle. Increasing experience showed us that retropneumoperitoneum “predissected” the mesocolon from the Waldeyer’s fascia. Even if the procedure was not completed entirely transanally, this phenomenon rendered the final abdominal approach less cumbersome. The transanal colonic mobilization necessitates considerable tilting of the operating table to the right. With conventional laparoscopic instruments passing in the “groove” between the sacral promontory and left ala, the mesocolon is progressively mobilized until the splenic flexure is reached. The latter is mobilized along with the freeing of the root of left transverse mesocolon. According to our experience, there is no need to dissect the greater omentum from the colon in order to obtain a tension-free coloanal anastomosis.

Our study had several limitations including small sample sizes that underpowered the statistical analysis and the significance of our conclusions. There are heterogeneous subgroups and overlapping techniques. This prevented us from claiming the superiority of NOTESTME over LAPTME in terms of a clear definition of anatomy and better preservation of the specimen. Finally, our study was limited to the comparison of anatomical parameters between NOTESTME and LAPTME, without correlation with clinical or oncological outcomes.

Conclusions

Our results show that NOTESTME, though still evolving, is a promising, safe, and reversible technique for the treatment of selected patients with distal rectal cancer. Larger studies on a broader group of patients are warranted in order to confirm these findings and validate the presumed advantages of NOTESTME in terms of clinical and mid- to long-term oncological outcomes.