Department of Surgery, Division of Gastrointestinal Surgery, Massachusetts General Hospital, Boston, Massachusetts, USA
Corresponding: Patricia Sylla, Department of Surgery, Division of Gastrointestinal Surgery, Harvard Medical School, Massachusetts General Hospital, 15 Parkman Street, Wang 460, Boston, MA 02114, USA. Email: firstname.lastname@example.org
The surgical management of rectal cancer has evolved over the past century, with total mesorectal excision (TME) emerging as standard of care. As a result of the morbidity associated with open TME, minimally invasive techniques have become popular. Natural orifice translumenal endoscopic surgery (NOTES) has been held as the next revolution in surgical techniques, offering the possibility of ‘incisionless’ TME. Early clinical series of transanal TME with laparoscopic assistance (n = 72) are promising, with overall intraoperative and postoperative complication rates of 8.3% and 27.8%, respectively, similar to laparoscopic TME. The mesorectal specimen was intact in all patients, and 94.4% had negative margins. There was no oncological recurrence in average-risk patients at short-term follow up, and 2-year survivalrates in high-risk patients were comparable to that after laparoscopic TME. These preliminary studies demonstrate transanal NOTES TME with laparoscopic assistance to be clinically feasible and safe given careful patient selection, surgical expertise, and appropriate procedural training. We are hopeful that with optimization of transanal instruments and surgical techniques, pure transanal NOTES TME will become a viable alternative to open and laparoscopic TME in the future.
The American Cancer Society estimates there will be 40 340 new cases of rectal cancer in the USA in 2013. While radical surgery with or without neoadjuvant treatment has been the mainstay of resectable rectal cancer treatment, interest in less invasive approaches is increasing. Among these evolving surgical approaches, natural orifice ‘incisionless’ surgery holds particular promise. Here we aim to review the evolution of surgical techniques for rectal cancer, focusing on the current status of transanal natural orifice translumenal endoscopic surgery (NOTES) total mesorectal excision (TME).
Evolution of the Surgical Management of Rectal Cancer
Radical rectal cancer resection
Resection for adenocarcinoma of the rectum has evolved significantly over the last century. Local resection by various techniques was standard of care until Miles postulated that radical resection of the anus and mesenteric lymphatics would control the spread of cancer and reduce the high recurrence rates associated with local excision, and proposed the abdominoperineal resection in 1908. In 1931, Abel described a sphincter-sparing alternative, total mesorectal excision (TME) of the rectum. TME was not widely carried out until 1979, when Heald demonstrated safety in limiting mural clearance and preserving the anal sphincters if the mesorectum was excised intact. The widespread adoption of the TME principles of sharp dissection of tissue planes and achievement of an intact mesorectum, negative distal and circumferential radial margins (CRM), and harvest of at least 10–14 lymph nodes[5, 6] resulted in improved survival rates and functional outcomes.[7, 8] The introduction of surgical staplers allowed for low colorectal anastomoses while intersphincteric resection (ISR) allowed for even lower coloanal anastomoses, achieving sphincter preservation in very low rectal cancer with acceptable oncological and functional outcomes.[9, 10]
Open TME is associated with significant morbidity including sexual and urinary dysfunction, fecal incontinence, and wound-related complications such as infection and incisional hernias.[11-13] Thus, when laparoscopy was introduced, it was quickly embraced in an attempt to reduce overall morbidity.[14-19] Randomized controlled trials demonstrated reduced length of hospital stay (LOS) and recovery time and equivalent oncological outcomes with laparoscopic versus open colon cancer resection; similar data are emerging from preliminary randomized trials comparing laparoscopic versus open TME for rectal cancer.[20-22]
Laparoscopy, however, has not been demonstrated to reduce the morbidity of colorectal resections, likely because of the need to extend abdominal incisions for specimen extraction. Wound infection rates after laparoscopic colorectal surgery range from 2.7% to 12.8%,[23-25] with the extraction site accounting for up to 50% of those infections. Additionally, hernia formation occurs in 3.6–24.3% of port sites after laparoscopic colorectal resection.[27-33] Thus, ongoing efforts have focused on developing strategies to minimize the need for sizeable abdominal incisions.
Natural orifice specimen extraction
The drive to maintain the advantages of laparoscopy and reduce wound-related complications led to natural orifice specimen extraction (NOSE). Multiple groups have successfully removed colorectal specimens transvaginally and transanally.[34-41] Park et al. compared laparoscopic right colectomy with transabdominal versus transvaginal extraction (n = 68), finding significantly decreased postoperative pain and LOS, as well as fewer wound infections and intra-abdominal abscesses, although these differences were not statistically significant. Franklin et al. reported a series of 432 patients who underwent laparoscopic low anterior resection (LAR) with or without transanal extraction for rectal cancer. Two-year local recurrence rates, intraoperative complications, postoperative complications, and LOS were comparable.[43, 44]
Additional studies have confirmed infection rates following transanal NOSE versus transabdominal colorectal resections to be equivalent.[45, 46] One prospective study found no infection-related complications after transanal specimen extraction for diverticulitis despite positive polymicrobial peritoneal cultures. Furthermore, anal sphincter function was not negatively impacted by transanal NOSE,[48, 49] with patients reporting significant improvement in their quality of life at 6-month follow up.
Overall, based on these few non-randomized NOSE studies, transvaginal and transanal extraction of colorectal resection specimens appears to be safe, associated with reduced pain and shorter LOS, and offering at least the cosmetic benefit of eliminating extraction site incisions.
In 1977, Morson et al. harkened back to local excision and described a series of full-thickness local excisions of rectal cancer by transanal excision (TAE) in which only three of 91 patients recurred. Subsequent studies failed to reproduce such good results, however. Failure rates after TAE of 7–40% and 25–62% for T1 and T2 tumors, respectively, were reported, much higher than those after TME (0–7% for T1 tumors, 3–15% for T2 tumors).[52-59] For those patients who did recur after TAE, 5-year survival of only 30–53% after salvage surgery has been reported.
High failure rates following local excision for high-risk T1 and more advanced rectal cancers is attributed to the increasing rate of nodal involvement with more advanced T stage.[53, 60] One study showed that for T1 and T2 tumors with favorable histological features, 7% and 14% had nodal metastases, respectively. T1 and T2 tumors with high-risk features had 33% and 30% risk of nodal involvement, respectively. Furthermore, current staging modalities, including endorectal ultrasound (ERUS) and pelvic magnetic resonance imaging (MRI), are still not entirely predictive of nodal involvement.[62, 63]
Thus, TAE is traditionally reserved for patients with carefully selected tumors, including preoperatively staged T1 tumors that are mobile, occupy <40% of the lumen, well differentiated with no evidence of lymphovascular invasion, and are confined within the superficial submucosa (sm1 or sm2). TAE is also routinely used for palliation in patients not suitable for radical surgery.
Transanal endoscopic surgery
In 1983, Buess et al. introduced transanal endoscopic microsurgery (TEM) as a minimally invasive approach for full-thickness resection of benign rectal lesions.[64, 65] When compared to traditional TAE, transanal endoscopic surgery (TES) offers superior visualization, improved access to proximal lesions, increased rates of negative margins, decreased fragmentation of specimens, and lower recurrence rates.
TES was quickly applied to T1 rectal tumors, as well as to advanced lesions in patients unable to tolerate radical resection.[67-71] TES has the same limitations as TAE, including high recurrence rates for more advanced tumors and tumors with high-risk histological features.[72, 73] Nevertheless, in carefully selected T1 cancers, local recurrence rates for TES are similar to those for TME; 0–5% in T1 sm1/sm2 cancers, but up to 10–12% in T1 sm3 cancers.[68, 72-82] Furthermore, TES is associated with lower mortality (<0.5%) and morbidity (3–12%) than radical resection.[83-86] Thus, data support the use of TES for benign lesions, low-risk T1 cancers, or palliation.
Recently, in addition to traditional rigid TES platforms, more versatile and disposable transanal minimally invasive surgery (TAMIS) platforms have been introduced and have popularized TES worldwide. These platforms consist of multiport transanal devices that can be combined with standard laparoscopic equipment to carry out endolumenal excision. Preliminary studies have demonstrated both procedural and oncological safety of TES carried out with these new platforms for benign and malignant disease.[88, 89]
Transanal natural orifice translumenal endoscopic surgery
Since the description of a completely endoscopic transgastric appendectomy in 2005, NOTES has been held as the next revolution in surgical technique. NOTES proposes to take advantage of natural openings including the mouth, vagina, and anus to access the peritoneal cavity endoscopically and carry out complex surgical procedures without the need for transabdominal incisions.[91, 92] Theoretical advantages of NOTES include the ability to minimize wound-related complications and recovery time. Transanal NOTES for colorectal resection is intuitive, as it involves a viscerotomy (colotomy or proctotomy) created in the organ that is ultimately removed and incorporated in the colorectal anastomosis. From an oncological standpoint, the concept of transanal TME is not entirely novel, as it represents an extension of traditional intersphincteric resection (ISR). NOTES TME proposes to complete the entire rectal and mesorectal dissection using a primarily transanal endoscopic approach.
The feasibility, safety, and reproducibility of pure and hybrid transanal NOTES rectosigmoid resection were demonstrated first in swine survival experiments and subsequently in human cadaver series.[64, 93-103] In the largest cadaver series to date (n = 32), our group demonstrated the feasibility of pure and hybrid transanal NOTES rectosigmoid resection, with an intact mesorectum demonstrated in all cadavers (Fig. 1). The study also highlighted the shortcomings of currently available instrumentation for transanal endoscopic work and the importance of laparoscopic assistance. The laparoscopic transabdominal view helps in identifying critical pelvic structures such as the ureters and pelvic nerves, especially early during the learning curve for the procedure. Laparoscopic assistance with retraction of the colon and rectum during transanal TME is also helpful, especially in the setting of a bulky uterus and redundant sigmoid colon. Finally, laparoscopic assistance allows for the takedown of the splenic flexure, mobilization of the entire left colon, as well as dissection and division of the inferior mesenteric artery to be carried out simultaneously with transanal TME, which has the potential to reduce operative time.
In 2009, our group reported the first clinical case of rectosigmoid resection for rectal cancer using transanal NOTES with laparoscopic assistance[104-106] (http://www.youtube.com/watch?feature=player_embedded&v=9t1pDmslnuk). The patient was a 76-year-old woman with a preoperatively staged T2N2 tumor located 8 cm from the anal verge (AV) and treated with full-course preoperative chemoradiation. The mesorectal specimen was intact, with negative margins and 23 negative lymph nodes. Her ileostomy has since been reversed with good functional outcomes and no evidence of recurrence to date. Since this initial case, several small clinical series of TME for rectal cancer using a transanal NOTES approach with laparoscopic assistance have been published,[107-113] highlighting variations in patient selection, surgical technique, complications, and outcomes.
Ten series of transanal TME with laparoscopic assistance have been published since 2010 (total n = 72, range 1–30 patients per study). Mean patient age was 64.2 years (range 36–82), mean body mass index (BMI) was 25.2 (range 16–33), and 75% were male. Mean tumor location was 6.4 cm from the AV (range 2–15, exact location reported in 41 cases). Forty four percent of tumors were located <5 cm from the AV, 52% were 5–10 cm, and 4% were >10 cm. Preoperative node status was reported in 23 patients; three were node positive (13%). On final pathology, 37.5% of resected specimens had positive nodes. Neoadjuvant treatment was given to 81% of patients for locally advanced tumors (Table 1)., [107-115]
Table 1. Patient characteristics of published clinical series on transanal TME with laparoscopic assistance
Mean operative time was 272.5 min (range 120–460), with a mean of 2.8 abdominal trocars used (range 1–5), mean estimated blood loss of 77.8 mL (range 10–300), and mean LOS of 10.2 days (range 4–25). The overall intraoperative complication rate was 8.3%, including a 2.8% incidence of conversion to open surgery. The overall postoperative complication rate was 27.8%. All TME specimens were complete, and 94% had negative margins (Tables 2, 3)., [107-115]
Table 2. Intraoperative and postoperative outcomes of published clinical series on transanal TME with laparoscopic assistance
No recurrence (13), treated for recurrence (12), cancer-related deaths (4) at 21-month follow up
No recurrence at 5.4-month follow up
In most clinical series, eligibility criteria included rectal lesions with high-grade dysplasia or preoperatively staged T1, T2, or T3 rectal adenocarcinoma by ERUS or pelvic MRI, and no evidence of metastases on computed tomography (CT) scans. The series by Sylla et al. also required that tumors be located 4–12 cm from the AV, with a predicted mesorectal margin >5 mm on MRI. Velthuis et al. included only mid-rectal cancers, whereas Zorron et al. and Dumont et al. included low- and mid-rectal tumors.[108, 112] Sylla et al. was the only study to specifically exclude node-positive tumors; of the patients in the remaining trials with recorded preoperative node status, 13% were node positive. The majority of patients with T3 tumors, and two patients with T2 tumors,[110, 113] received neoadjuvant chemoradiation.
In most series, patients were highly selected, with exclusion criteria including BMI >35, T4 or recurrent tumors,[108, 114] and prior extensive abdominal or pelvic surgery.[110, 115] However, two studies specifically included patients with a hostile pelvis, in whom TME was expected to be difficult. Dumont et al. included patients with at least one difficult anatomical characteristic (BMI >30, narrow pelvis, voluminous prostate). Rouanet et al. not only included patients with difficult anatomy (high BMI, narrow pelvis, large prostate, fatty mesorectum, fibrosis), but also included patients with high-risk tumors (T4, recurrent, large anterior tumor, predicted CRM ≤1 mm on MRI). Eighty-three percent of patients had a predicted CRM ≤1 mm and three patients had distant metastases. Tumors in this series were much more advanced than in other published series. These results underscore the critical importance of careful patient selection for the transanal TME approach as we continue to evaluate the short and long-term oncological safety of this approach.
Multiple transanal platforms were used for NOTES TME; five trials used rigid platforms and five trials used TAMIS platforms. Our group used a rigid metal proctoscope, which provides platform stability (Fig. 2). Other groups used flexible single-incision platforms, noting excellent maneuverability and triangulation of instruments.
Purse-string occlusion of the rectum was carried out at the start of the transanal dissection in 90% of cases; in seven cases, it was done after full-thickness circumferential rectal resection.[108-110] Monopolar or bipolar energy was used for the transanal dissection. The splenic flexure was partially or fully mobilized in 55% of patients.
Coloanal anastomoses were either handsewn (64%) or stapled with a circular stapler (36%).[108, 110, 111] No apparent difference in rates of anastomotic leakage was observed, although sample size was limited. A diverting ileostomy was created in 92% of patients.[114, 116]
For the clinical series that included rectal tumors <5 cm from the AV, various sphincter-sparing procedures were used in order to ensure negative distal margins. Dumont et al. carried out anal mucosectomy just above the dentate line, whereas Rouanet et al. and Sylla et al. carried out partial ISR (pISR).
Intraoperative and postoperative complications
There were six intraoperative complications (8.3%), including two conversions to open surgery (2.8%). One patient developed pneumatosis of the small bowel mesentery without adverse consequences. Rouanet et al. reported two urethral injuries (one was early in the learning curve and involved a bulky anterior tumor, the other was in the setting of a large prostatic carcinoma) and a possible air embolism. He also reported two conversions that occurred early in the learning curve and involved posterior fixed tumors. Excluding the high-risk patients selected by Rouanet et al. there was one intraoperative complication (2.4%) and no conversions to open surgery.
The mean LOS was 10.2 days (range 4–25) and the postoperative complication rate was 27.8%. Six patients experienced temporary urinary dysfunction (8.3%) likely secondary to pelvic nerve injury. Two of these patients had low rectal tumors and underwent pISR. This rate of urinary dysfunction after transanal TME is comparable to that after laparoscopic TME (up to 20%).[117, 118] There were two cases of postoperative ileus and one bowel obstruction (overall rate 7%). Two patients developed anastomotic complications (2.8%); one with a handsewn anastomosis developed an anastomotic fistula (treated with antibiotics and transanal abscess drainage) and one had an anastomotic leakage (type of anastomosis and management not reported). Two patients developed dehydration as a result of high ileostomy output (2.8%), with one patient requiring re-admission.
Six patients experienced infectious complications (8.3%), including two with peritonitis, one with a presacral abscess requiring laparoscopic drainage, one with an anastomotic fistula treated with antibiotics, and one with severe chronic obstructive pulmonary disease (COPD) who developed pneumonia. One patient with significant comorbidities developed sepsis, requiring critical care. This is comparable to published complication rates of approximately 27% following laparoscopic TME. There were no postoperative mortalities.
The mesorectal specimen was complete in all 72 cases of transanal TME with laparoscopic assistance, with a mean 15.8 nodes (range 8–53) collected. Margins were negative in 94% of cases. Rouanet et al. reported four patients with positive CRM; one had a tumor that was adherent to surrounding structures (converted to open), two had aggressive tumors and margins showing microscopic invasion, and one had regional recurrence located across the sacral plane. In this high-risk patient population, Rouanet et al. achieved negative margins in 87% of cases compared to 95% negative margins in the 84 low-risk patients who underwent laparoscopic TME during the same study period, again underscoring the importance of careful patient selection as the short and long-term oncological safety of transanal TME is under evaluation.
Most clinical series reported only short-term oncological outcomes. Dumont, de Lacy, and Sylla reported no recurrence at an average follow up of 4.3 months, 30 days, and 5.4 months, respectively.[112, 114, 115] In the Rouanet series of high-risk patients, 43% had no recurrence after a median follow up of 21 months (range 10–40), 40% were treated for local or distant recurrence, and 13% died of cancer-related causes. Of those who recurred, four patients had solely locoregional recurrence; two of them had had a positive CRM. Overall survival rates were 96.6% and 80.5% and relapse-free survival rates were 93.3% and 88.9% at 12 and 24 months follow up, respectively. Even in this high-risk population, these results are comparable to short-term oncological outcomes after standard TME. Longer-term oncological outcomes after transanal TME with laparoscopic assistance are being collected. While promising, the current short-term oncological outcomes highlight the need for careful patient selection for transanal NOTES TME.
Of the groups that used rigid transanal platforms, only Rouanet et al. assessed functional outcomes. Of the 12 patients who underwent stoma closure, 40% were fully continent, 15% were incontinent to liquids, 35% were incontinent to gas, and 25% had stool fragmentation, with a median Wexner score of 11 at 12-month follow up. Of note, all patients underwent pISR; nevertheless, functional outcomes were comparable to prior studies of coloanal anastomosis with ISR for rectal cancer.
Dumont et al. used a TAMIS platform and carried out anal mucosectomy in all four patients, and also reported encouraging short-term functional outcomes. Although mucosectomy has been associated with fecal incontinence, the median Wexner score was 5 at the 3-month follow up, similar to that reported by prior trials of TME with coloanal anastomosis.[123, 124]
Pure NOTES transanal TME
Leroy et al. described one of the only two cases of pure NOTES transanal TME done for a preoperatively staged T2 rectal cancer, with final pathology demonstrating a tubulovillous adenoma. Bottom-up mesorectal excision was completed using a rigid metal TES platform, and coloanal handsewn anastomosis was done with no diverting ileostomy. The patient developed a small pelvic hematoma requiring CT-guided drainage. The mesorectum was intact with 16 harvested lymph nodes. Zhang et al. published the other case report of pure NOTES transanal TME for a preoperatively staged T3N1 rectal cancer located 8 cm from the AV. Transanal TME was done using a flexible single-incision platform with stapled coloanal anastomosis and no diverting ileostomy. The mesorectum was intact with negative margins and 12 harvested nodes.
Preliminary results from small series published to date by our group and others around the world suggest that transanal NOTES rectal cancer resection is feasible, reproducible, and preliminarily safe with high-quality TME and excellent lymph node retrieval in carefully selected patients. In most cases, patients tolerated the procedure well; most complications were minor and transient. Nevertheless, preliminary data on complications and short-term oncological outcomes emphasize the importance of careful patient selection and the need for large-scale trials on long-term outcomes and oncological safety of NOTES transanal TME before widespread adoption can be recommended.
Beyond the ability to remove the specimen transanally, NOTES may offer technical advantages over laparoscopic and open approaches. In-line visualization of pelvic structures makes dissection up and out of the pelvis by the transanal technique safe and efficient. The transanal approach with CO2 insufflation and direct endoscopic visualization facilitates precise dissection of the distal resection margin, presacral plane, and perirectal planes, particularly advantageous when operating in the narrow male pelvis. Furthermore, sphincter preservation can be achieved even for very low tumors by adding ISR prior to initiating transanal TME.
With regards to training for transanal NOTES TME, surgeons must be experienced with laparoscopic TME and TES. Our work demonstrates a steep learning curve of five cadavers, underscoring the need for pre-procedural training. Based on our prior work in swine and cadavers, we have concluded that the ideal training model for this procedure is human cadavers. In the hands of surgeons with expertise in laparoscopic TME and TES as well as periprocedural training in human cadavers, groups who have published on this procedure anticipate that outcomes following transanal TME in carefully selected patients should be equivalent to those following laparoscopic TME.
Ongoing studies are needed to further characterize perioperative and functional outcomes, as well as long-term oncological outcomes. These studies will also explore the safety of transanal NOTES in a broader patient population, including obese patients and patients with node-positive rectal cancer. Future studies will also compare laparoscopic or robotic versus transanal NOTES TME.
Improvements in transanal platforms and instruments may eventually enable the entire TME procedure to be safely and effectively carried out using a pure transanal endoscopic approach. Laparoscopic assistance is currently necessary for full splenic flexure mobilization, assistance with pelvic dissection, avoidance of organ injury, and decreased operative time. Adaptation to longer and more flexible scopes, platforms, and instruments will surely improve, if not eliminate, the issues of navigating the sacral promontory and allow for more proximal colon dissection.
Short-term outcomes of small clinical series suggest that transanal endoscopic TME is not only feasible but safe in carefully selected patients when carried out by surgeons with appropriate expertise and training. These early data are promising and, as our experience develops and outcomes mature, we anticipate improved characterization of the patient population best suited for this approach, development of a pure transanal NOTES approach, and expansion of the procedural applications of NOTES.
Conflicts of Interest
Dr Sylla received honoraria from Covidien and Novatract Surgical for consulting, and an honorarium from Applied Medical for teaching. Dr Emhoff and Ms Lee have no conflicts of interest to disclose.