Transanal Endoscopic Surgery: Who Should Be Doing This Procedure?

Abstract

Transanal endoscopic surgery (TES) was introduced in the 1980s, but more widely adopted in the late 2000s with innovations in instrumentation and training. Moreover, the global adoption of minimally invasive approaches to abdominal procedures has led to translatable skills for TES among colorectal and general surgeons. While there are similarities to laparoscopic surgery, TES has unique challenges related to the narrow confines of intraluminal surgery, angled instrumentation, and relatively uncommon indications limiting the opportunity to practice. The following review discusses the current evidence on TES learning curves, including potential limitations related to the broad adoption of TES by general surgeons. This article aims to provide general recommendations for the safe expansion of TES.

Minimally invasive surgery (MIS) has become the standard of care for many surgical procedures, reducing perioperative pain and hospital length of stay. Laparoscopic colorectal surgery gained popularity in the early 2000s with the publication of several trials collectively demonstrating short-term benefits and oncologic equivalence to the conventional open approach for the surgical management of colon cancer. ^{1 2 3 4}

Transanal endoscopic surgery (TES) was introduced in the 1980s, but not widely adopted until the late 2000s with advances in instrumentation and training. TES is technically challenging and is usually performed by trained colorectal surgeons. While there are similarities to laparoscopic surgery, TES has unique challenges related to the narrow confines of intraluminal surgery, uniquely angled instrumentation, and relatively uncommon indications limiting the opportunity to practice. In 2010, TES technique was first described as part of a rendezvous approach with laparoscopic abdominal surgery to facilitate rectal cancer surgery, now known as transanal total mesorectal excision (TaTME). ⁵ Thereafter, enthusiasm to learn TES techniques has expanded to include many surgeons who do rectal cancer surgery.

While TES has gained popularity among colorectal and general surgeons, it is not clear that the broad application of these techniques is being applied either appropriately or skillfully to ensure acceptable outcomes. Recent national registry data in Norway suggest unacceptable local recurrence rates after TaTME, likely attributable to poor performance of the TES component of the procedure. ⁶ Thus, many questions remain regarding the present and future of TES. Who should be doing this procedure? How should it be taught? The following review will discuss the current evidence on TES learning curves, along with potential limitations related to the broad adoption of TES for general surgeons.

Learning Curve Principles

When a surgeon or a trainee is learning a new procedure, surgical performance tends to improve with experience. Learning curves can be defined as graphical representations of performance against experience. ⁷ A learning curve illustrates the proficiency gain, which is usually plotted over time with increasing case number. A surgeon starting with a new procedure will be described as being in the early phase of his or her learning curve. One frequent misnomer is the use of the term “steep learning curve” to describe a procedure that is challenging or difficult to learn. This is an inaccurate representation of the learning process, since a steep curve literally means that one would ascend the curve rapidly over a small number of cases. In fact, complex procedures are more likely to be illustrated by a gradual learning curve, with small improvements in the outcome measures associated with each completed case.

Measures used to evaluate the learning process in the surgical field are derived from two categories: surgical process and patient outcomes. Surgical process includes operative variables such as operative time, blood loss, margin involvement, and lymph node yield. Patient outcomes can be related to both short-term and long-term outcome measures, such as complication rates, length of stay, mortality, recurrence, and cumulative survival. Surgical process variables are usually easier to analyze and are more commonly used. However, many will argue that surgical process are merely indirect measures of patient outcomes, the latter being far more relevant. ⁸ In the present article, various measures of TES learning curves will be discussed, along with their advantages and disadvantages.

Literature Review of TES Learning Curves

Over the last decade, many authors investigated the possible impacts associated with a learning curve for TES. As discussed in the previous article (Transanal Endoscopic Platforms. TAMIS versus rigid platforms: Pros and Cons by Dr Devane et al.), TES is usually performed using one of the main two platforms: the rigid platform for transanal endoscopic microsurgery (TEM) and the transanal minimally invasive surgery (TAMIS) soft platform. We reviewed the available evidence on the learning curves with both platforms.

Transanal Endoscopic Microsurgery

One of the first article investigating learning curves for TES was published in 2009 by a group from the Netherlands. ⁹ The authors prospectively included all patients who underwent TEM between 2002 and 2007. A total of 105 patients were included and the procedures were performed by a single surgeon. The authors compared the outcomes of the first half of patients (2002–2004, 52 patients) to the following half (2005–2007, 53 patients). They found that the second half of patients had a significantly shorter mean operating time (97 vs. 123 minutes, p  = 0.018), a decrease in the complication rate (5.7 vs. 9.6%, p  = 0.001), along with a shorter length of hospital stay (4.4 vs. 6.1 days, p  = 0.046). A major limitation of this article is the method used for statistical representation of the TEM learning process. The “before/after” approach does not provide a proper learning curve assessment. This method raises concern for bias by data-dependent splitting for the cutoff choice. Even when authors are arbitrarily splitting the data into two halves, it is not possible to describe the underlying curve or to identify where on that curve the surgical proficiency is reached, since it takes a minimum of three points on a curve before a trend can be characterized. ¹⁰ While this study was more of a reflection on a single surgeon experience with TEM, it did suggest that increasing case volume tends to improve surgical efficiency, opening the way for further investigation regarding TES learning curves.

A second group for the Netherlands investigated the learning curve of colorectal surgeons beginning their TEM practice after a formal training program. ¹¹ The study included 4 surgeons and 693 rectal lesions. Findings suggested a learning curve effect for a decrease in the conversion rate to traditional transanal excision or abdominal approach with an odds ratio (OR) of 0.991 (95% confidence interval [CI]: 0.984–0.998, p  = 0.007) per each additional case. While a learning curve effect could not be demonstrated for postoperative complication rates in benign lesion, malignant cases were subject to a learning curve effect with an OR of 0.99 (95% CI: 0.988–0.998, p  = 0.006) per each additional procedure. Operative time per surgeon showed a decrease over time with a regression coefficient ( B ) of −0.11 (95% CI: −0.14 to −0.09, p  < 0.001). Tumor recurrence did not seem to be subject to a learning curve effect. The main strength of this study is the large number of cases included. However, around two-thirds of the procedures were performed by the first surgeon. Furthermore, they included retrospective data from multiple centers, and it is likely that tumor descriptions, complication definitions, and other relevant data were not standardized. The authors also excluded TEM cases in which the primary surgeon was supervising the procedure, and it is likely that mentoring contributes to the surgeon's learning curve to some extent.

A group from Cleveland Clinic Florida investigated the TEM learning curve for a single colorectal surgeon over 23 consecutive cases between 2005 and 2008. ¹² The colorectal surgeon was board certified and learned the technique from prof. Buess, the pioneer of TEM. ¹³ The cumulative sum (CUSUM) method was performed to plot a learning curve for the rates of excision in minutes/cm ² . The CUSUM approach has been advocated as an effective way to illustrate learning or to monitor surgical performance. ¹⁰ CUSUM charts are a graphical method for looking for trends or changes over time—or case number—using dichotomous proxies for learning. In this article, the authors reported a mean average rate of excision (ARE) of 8.9 minutes/cm ² . A stabilization of the learning curve was observed after the first 4 cases, showing an ARE of 13.8 minutes/cm ² compared with 7.9 minutes/cm ² for the last 19 cases ( p  = 0.001). The authors concluded that the ascent of the learning curve for TEM may be easily achieved. We believe that this conclusion is too simplistic, as it is unlikely that a specialized procedure such as TES could be mastered after only four cases, especially since the only outcome studied in this article was the ARE. On the contrary, the authors did point out that an additional rising and leveling of the ARE was found after the first 10 cases when more challenging proximal lesions were resected.

Recently, a group from Ottawa published their experience in a newly established TEM program in a single institution. ¹⁴ A similar method was used to review the ARE (minute/cm ² ) as a measure of surgical efficiency. The authors plotted the results using restricted cubic splines and reviewed predictors of higher ARE using multivariate regression. Between 2009 and 2014, 108 patients underwent TEM performed by two MIS and colorectal surgery fellowship-trained surgeons. ARE was available for 95 patients. Mean ARE was 18.6 minutes/cm ³ . Adjusting for significant covariates, the ARE improved with each additional case until 16 procedures were completed. Significant predictors of higher ARE found with multivariate analysis were younger age (< 50 years), experience of less than five cases, and a tumor histology of carcinoid, gastrointestinal stromal tumor, or scar lesion. One limitation is that the method used to calculate the ARE assumed that all specimens have a cuboid shape. Nevertheless, the study's aim was not to accurately estimate the ARE but to demonstrate a learning curve effect related to a faster rate of excision over time.

Transanal Minimally Invasive Surgery

TAMIS has recently gained popularity in the general surgery community, especially since it involves the use of standard laparoscopic instruments. While there are differences in the layout and instrumentation, these differences are nuanced when compared with the differences between video-assisted transanal surgery and laparoscopic abdominal surgery.

A group from Korea published their single-center experience on short-term outcomes and learning curve analysis in 46 consecutive patients undergoing TAMIS from 2012 to 2014. ¹⁵ Learning curve from a single surgeon was analyzed with CUSUM charts of operative time. Mean surgery duration was 85 minutes. The authors found that operative time significantly decreased after 17 consecutive cases. Comparative analysis for patient and tumor characteristics between the first 17 patients and the following 29 patients did not show significant differences. Total operative time decreased with time, with a mean operative time of 125 minutes for the first 17 patients and 62 minutes for the following 29 patients ( p  < 0.001). Mean hospital stay was also significantly longer for the first 17 patients (8.3 vs. 2.3 days, p  = 0.030). Of note, this study showed a concerning high number of reoperations and major complications during the learning curve period of the first 17 patients, with 2 patients requiring a laparoscopic anterior resection and loop ileostomy due to leakage and abscess, and 1 patient requiring endoscopic management of a significant postoperative bleeding. This emphasizes the necessity of safety monitoring when implementing a new procedure.

In 2018, authors from a single tertiary institution in China reported their experience with TAMIS. ¹⁶ The study consisted of a retrospective review of 25 patients who underwent TAMIS from 2013 to 2016. All procedures were performed by a group of trained laparoscopic colorectal surgeons. Mean operative duration was 61.3 minutes. For the learning curve evaluation, the authors used a subjective graphical representation of the correlation between case number and operative time. This curve showed a decreasing trend in surgery duration for the first 10 cases, with the curve reaching a plateau afterward. An important limitation of this study is the limited number of cases, especially considering that they were performed by a group of several colorectal surgeons.

A group from Orlando recently published a large retrospective cohort study reviewing all TAMIS cases performed between 2009 and 2016 at a single tertiary center. ¹⁷ A CUSUM analysis was performed to determine the number of cases required to reach surgical proficiency, with the main outcome being the margin positivity rate (R1 resection). The acceptable and unacceptable margin positivity rates were defined as the R1 resection rates for TEM (10%) and traditional transanal excision (26%). A total of 254 TAMIS cases were included with an overall R1 resection rate of 7%. All procedures were performed by colorectal surgeons from a single institution. Patients were divided into three groups according to their operating surgeon, with two of these groups including procedures performed by two surgeons together. One group included 17 patients who underwent robotic TAMIS (7% of the entire cohort), with the first case being performed on the 37th cumulative case of this group. CUSUM analysis showed that surgeons performing TAMIS reached an acceptable R1 resection rate between 14 and 24 cases. The groups involving procedures performed by two surgeons together had a lower number of cases required to reach proficiency (14 and 20 cases) compared with the group with a single surgeon (24 cases). The authors also showed that the mean operative time decreased and stabilized along with proficiency gain. Interestingly, the mean lesion size was larger after proficiency gain (3.0 vs. 2.3 cm, p  = 0.008), suggesting that more challenging cases are attempted as experience increases over time. Other patient and tumor characteristics were similar before and after proficiency was reached.

A similar analysis was performed by a group from the Netherlands. ¹⁸ The authors aimed to investigate surgeon-specific learning curves for TAMIS and analyze the effects of proctoring on operating time and surgical outcomes. They prospectively collected the data on all TAMIS cases performed by two colorectal surgeons between 2010 and 2017. Margin positivity, specimen fragmentation, adverse events, and operative time were evaluated with a CUSUM analysis to determine the number of cases required to reach proficiency. The first surgeon performed 103 procedures without proctoring and developed the institutional program. The second surgeon performed 26 cases with proctorship from the first surgeon in the established institutional program. CUSUM curve for operative time suggested that proficiency was reached after 36 cases for the first surgeon and after 10 cases for the second. For margin positivity, proficiency was reached after 31 cases and 6 cases for the first and second surgeons, respectively. The complications curve for the first surgeon showed a three-phase learning curve, with an acceptable adverse events rate (16%) reached at the 18th case, followed by a slight increase in complications before finally trending down after 26 cases. The second surgeon had only one complication in the learning period with no change point in the CUSUM analysis. For both surgeons, the preproficiency periods were associated with a longer operative time. Overall complications were also higher in the preproficiency period (21.7 vs. 13.0%, p  = 0.02). The authors concluded that approximately 18 to 31 procedures are required to reach TAMIS proficiency for complications and margin status. Institutional programs and proctorship may be associated with a shorter learning curve (6–10 cases). Table 1 represents an overview of the studies assessing TES learning curves.

Table 1. Overview of TES learning curve studies.

Authors	Year	Number of patients	Platform	Methods	Summary of findings	Number of cases suggested to reach proficiency
Koebrugge et al	2009	105	TEM	Surgical outcomes comparison between the first half of patients and the following half over time Single surgeon	Length of hospital stay, operating time, and complication rate reduced for the second half of patients	N/A
Barendse et al	2013	693	TEM	Multivariate regression analysis for learning curve effects in four surgeons beginning a TEM practice	Learning curve effect found on conversion rate, procedure time, and complication rate (malignant cases) No effect found for tumor recurrence	N/A
Maya et al	2014	23	TEM	Learning curve for the average rates of excision (min/cm 2 ) using the CUSUM method Single surgeon	Stabilization of the rates of excision after the first four cases Additional late learning curve when more challenging proximal lesions were resected	4
Helewa et al	2016	95	TEM	Learning curve for the average rates of excision (min/cm 3 ) using cubic splines method and multivariate regression Two surgeons	Rates of excision improved with each additional case until the 16th case Significant predictors of slower rates of excision included experience of less than five cases	16
Hur et al	2016	46	TAMIS	Learning curve analysis using CUSUM charts of the operative time Single surgeon	Operative time significantly decreased after 17 cases Longer hospital stay for the first 17 cases compared with the following 29 patients	17
Chen et al	2018	25	TAMIS	Learning curve analysis by graphical representation of the correlation between cases and operative time Multiple surgeons	Decreasing trend in operative time for the first 10 cases	10
Lee et al	2018	254	TAMIS	CUSUM analysis to determine the number of cases required to attain proficiency for acceptable margin positivity rate Five surgeons	Proficiency for an acceptable R1 resection rate was reached after 14–24 cases Mean operative time stabilization correlated with proficiency gain	14–24
Clermonts et al	2018	129	TAMIS	CUSUM analysis to determine proficiency for operative time, complications, and margin positivity Comparison between surgeons A and B, the latter having access to proctoring in the established institutional program	Operative time proficiency: 36 cases (10 for surgeon B) R1 resection rate proficiency: 31 cases (6 for surgeon B) Complications proficiency: 18–26 cases Potentially shorter learning curve for surgeons with proctoring (6–10 cases)	18–31

Abbreviations: CUSUM, cumulative sum; N/A, not available; TAMIS, transanal minimally invasive surgery; TEM, transanal endoscopic microsurgery; TES, transanal endoscopic surgery.

Limitations in Defining the Learning Curve

While the learning curve for TES appears to be relatively short in many of the previously discussed studies, other important factors should be considered before extrapolating these results. An important limitation related to the assessment of learning curves is the choice of study end points. The majority of the previously discussed studies investigated operative time or rates of excision as a surrogate marker for surgical proficiency. Surgical flow and efficiency of motion are part of various rating scales used to assess surgical skill and can be evaluated indirectly with average rates of excision. ^{19 20} Operative time may reflect the technical difficulty level of a procedure and has also been used in the past as an indicator for learning curves in laparoscopic colorectal resections. ²¹ However, shorter operative time does not always correlate with surgery quality. A group from Boston demonstrated that operative time can be a flawed surrogate marker for investigating learning curves in laparoscopic colorectal resections. ²² In this study, comparing surgeons with a high volume of cases, surgery duration did not decrease over time, and the surgeon with the shorter procedure time had a significantly higher morbidity as defined by the complication rate, readmission rate, and longer length of stay. Furthermore, operative time may be a biased estimate of the learning process since it is believed that surgeons will perform more challenging and longer procedures as experience grows over time. This trend was observed by Barendse et al, illustrating a significant correlation between progressing experience and tumor size. ¹¹ The authors also noted that increasing experience in the first 200 cases significantly correlated with procedures performed for malignant lesions. Maya et al also noted a second “late” learning curve effect on the CUSUM curve for resection rate when more challenging proximal lesions were approached over time. ¹² Given these considerations, it is likely that operative time leads to an underestimation of the true underlying learning curve, especially for more complex cases.

Regardless of the learning curve study end points, one of the most important limitations in the generalizability of these results is related to the variability in preliminary training of the surgeons involved. All previously discussed studies included laparoscopic colorectal surgeons in high-volume referral centers, many of whom performed TES during fellowship programs. Several authors stated that they followed some form of TES-oriented ex vivo training or course prior to the beginning of their study. ^{11 14 18} Some authors also included procedures performed with more than one surgeons or with access to proctoring. ^{17 18} Moreover, as pointed out by Lee et al, the experience with TES grew during the same period in which TaTME was introduced in many tertiary institutions. ¹⁷ Surgeons performing both TES and TaTME unquestionably had a faster learning curve, since both procedures use the same platforms for surgical access. Additionally, many authors reported higher rates of adverse events during the initial learning curve period, with some cases even requiring salvage laparoscopic low anterior resection and loop ileostomy for postoperative septic complications. ^{15 18} These findings emphasize that advanced colorectal surgery abilities are required to manage potential adverse events with TES. Therefore, care should be taken before extrapolating these relatively short learning curves to surgeons practicing in smaller, lower volume institutions, especially those with limited experience in minimally invasive colorectal surgery.

How Can We Improve the Safe Expansion of TES?

Is there really a specific number of cases that could be used as a cutoff point for establishing TES proficiency? While it would appear that measurable plateaus in some aspects of case performance occurs around 20 TES procedures, it is not clear that the metrics measured reflect true surgical proficiency. Given many limitations of the available literature for TES learning curves, using a single arbitrary number of procedures to allow any surgeon to perform TES is inadequate. Similar to all quality improvement strategies, clinical outcomes following TES implementation should be audited to ensure patients' safety at an institutional level. While there is no formal assessment or regulations for the implementation of new surgical procedures in North America, the Colorectal Surgical Society of Australia and New Zealand recommended a credentialing process for TEM in 2009. ²³ The training process required the surgeon to complete a formal TEM course, to assist in the completion of five cases and then to perform their first five cases under the direct supervision of a credentialed TEM surgeon. Additionally, the surgeon was required to maintain an annual surgical volume of five cases, with auditing of certain key performance indicators.

Clermonts et al provided a schematic overview of their suggested training pathway for surgeons wishing to implement a TAMIS practice. ¹⁸ The first step of this credentialing process is to obtain an extensive experience in laparoscopic colorectal resections, which the authors consider to be around 200 independent laparoscopic colon resections and 50 rectal resections. The second step is to acquire some form of standardized training with either a fellowship in a center where TAMIS procedures are performed, or to follow a dedicated TAMIS hands-on course. Standardized procedure-specific protocol should then be introduced in the surgeon's own institute. It is also suggested that the first five cases should be proctored by a surgeon with extensive experience with TAMIS and that case selection should be emphasized in the first 10 cases, minimally. Finally, the authors recommend that outcome data should be monitored after implementation. While these recommendations are subjective in nature, it supports our concerns about the limitations of the previously discussed studies suggesting short and modest learning curves in TES.

Institutional Factors

In the recent decades, several authors investigated the association between surgical volume and hospital setting on clinical outcomes in various subspecialized surgery fields. ^{24 25 26 27} While there are no current publications on this association specifically for TES, many authors analyzed this relationship for colorectal surgery. A Cochrane review published in 2012 investigated the effects of hospital volume, surgeon caseload, and specialization on the outcomes of colorectal cancer surgery. ²⁸ Overall 5-year survival was significantly improved for patients treated in high-volume hospitals (hazard ratio [HR] = 0.90, 95% CI: 0.85–0.96), by high-volume surgeons (HR = 0.88, 95% CI: 0.83–0.93) and colorectal specialists (HR = 0.81, 95% CI: 0.71–0.94). Operative mortality was significantly better for high-volume surgeons and specialists but did not differ in relation to the hospital caseload. Another recent meta-analysis investigated the association between hospital and surgeon volume on outcomes for colorectal surgery. ²⁹ Pooled results from 47 publications demonstrated a volume–outcome relationship that favors high-volume facilities and high-volume surgeons. Higher hospital and surgeon volume resulted in reduced 30-day and intraoperative mortalities. Postoperative complication rates depended on hospital volume (HR = 0.89, 95% CI: 0.81–0.98, p  < 0.05) but not on surgeon volume except for anastomotic leak (HR = 0.59, 95% CI: 0.37–0.94, p  < 0.01). The authors also found a relation between high-volume surgeons and greater 5-year overall survival and lymph node retrieval while reducing recurrence rates, operative time, length of stay, and costs.

These studies suggest better patient outcomes with increasing surgeon and institutional volume in colorectal cancer surgery. However, this causal relationship may be biased by other factors, such as multidisciplinary team management in high-volume center. There is increasing evidence that the use of multidisciplinary team is associated with improved decision-making and outcomes in many types of neoplasia, including rectal cancer. ^{30 31 32 33} This should also be considered in the TES setting, as patient selection is essential to achieve favorable outcomes. Furthermore, postoperative malignant pathology results from TES should be discussed in multidisciplinary teams to ensure that the proper treatment and follow-up is provided. In 2019, the Canadian Partnership Against Cancer—in partnership with the Canadian Society of Colon and Rectal Surgeons—published a document outlining Pan-Canadian standards for rectal cancer care. ³⁴ One recommendation is that appropriate early-stage rectal cancer patients should be assessed at a “TES center,” which is defined by a center providing the following services:

1. At least one rectal cancer surgeon with advanced training/expertise in one of the TES platforms.

2. Access to a pathologist with experience and expertise in evaluating local excision specimens, including documentation of all factors known to influence the need for immediate radical resection (e.g., depth of cancer invasion, lymphovascular invasion, tumor budding, margin status, etc.).

3. Access and experience with rigorous follow-up not typical of rectal cancer treated by radical resection.

Another factor that may play a role in the relationship between institutional volume and surgical outcomes is cooperation and mentoring from colleagues. This is especially important when more challenging cases are encountered, such as circumferential lesions, large defects, and managing intraperitoneal entry. As suggested by learning curves publications on TES, surgeons working as a team or having access to proctoring appear to reach surgical proficiency end points earlier. ^{17 18}

Conclusion

The widespread adoption of laparoscopic colorectal surgery created a favorable environment for the development and implementation of TES in the general surgery community. Several publications investigating the learning curves of TEM and TAMIS reported a wide range of suggested case number required to reach surgical proficiency, suggesting that the learning process is highly variable under different practice settings. The various choice of study end points and the baseline level of experience and training of included surgeons are important limitations in the generalization of the results from learning curve studies. While there are no clear universal guidelines for the safe implementation of a TES practice, surgeons should acquire an adequate level of training prior to engaging in TEM or TAMIS procedures. Centralization of care allows for increased surgical volume, multidisciplinary care, and mentoring from colleagues, which is likely to be associated with better patient outcomes.