Introduction

Robotic surgery has expanded the potential of minimally invasive surgery through articulated instrumentation giving the freedom of movement to replicate open surgery. These advantages are best demonstrated in pelvic surgery where space is limited allowing for precise minimally invasive surgery under direct vision [1]. The use of the da Vinci robot (Intuitive Surgical, Inc., Santa Clara, CA, USA) in colorectal surgery is gaining acceptance and its adaptation has grown exponentially [2,3,4]. Various studies have highlighted that robotic colorectal surgery offers comparable short-term and oncological outcomes to those of laparoscopic surgery, with lower conversion rates and fewer complication rates. Robotic surgery in the pelvis, especially for carcinoma of the rectum, provides 3D visualization with increased depth perception. The tips of the instruments have seven degrees of freedom, and wrist action is controlled by the surgeon’s hand control at the console to allow better ergonomics even in the narrow anatomy of the pelvis. This better freedom of movement with improved visualization of the surgical field enables surgeons to perform oncological resections without doing any significant damage to important pelvic structures. The term ‘learning curve’ was first used in 1909 by Bryan and Harter, referring to a diagram plotting the acquisition of a telegraphic language over time [5]. The surgical learning curve is the time and number of cases an average surgeon takes to attain surgical proficiency [6]. This curve generally starts from the top, then dips in the middle, and later follows a more flattened plateau path. Time to reach this plateau mainly depends upon the technical challenges of the procedure, the surgeon's prior surgical experience, and how common a procedure is being performed that warrants a new acquisition of skills.

With the rapid adoption of robotics in colorectal surgery, there has been growing interest in the pace at which surgeons gain competency, as it may aid in self-assessment or credentialing [7]. However, this is challenged by the lack of established training pathways and a defined robotic surgery curriculum for new surgeons. The learning curve in robotic colorectal surgery can be measured with the surgical process (such as time or adequacy of resection) or patient outcome (such as morbidity or quality of life) [8]. Therefore, we sought to present the serial experience of a recently qualified laparoscopic colorectal surgeon in practice that performed robotic pelvic surgeries including recto-sigmoid and anal malignancies. This experience was evaluated based on both surgical process and patient outcome giving an individual perspective related to the learning curve which can be applied to surgeons in a similar role.

Methods

Inclusion criteria

A serial retrospective review of the robotic colorectal surgery database at University Hospital Coventry and Warwickshire (UHCW), United Kingdom (UK) from July 2017 till February 2020 was performed. All elective robotic cases performed by one colorectal surgeon (AB) were included in the study. These cases included all pelvic surgeries for malignancies, including Robotic high anterior resections (rAR), Robotic low anterior resections (rLAR), and robotic extra levator abdominoperineal excision (rELAPE), of all age groups and gender.

Exclusion criteria

Procedures related to right and transverse colon either benign or malignant; left colon resections (e.g., diverticular disease) or rectal procedures for benign conditions (e.g., prolapse) and hybrid procedures were excluded from the study.

Surgeon experience

The first case was performed by AB within the first year of being appointed a consultant. Before robotic surgery, the surgeon had completed the standard colorectal training program in the UK, including a short post-CCT fellowship in advanced rectal cancer surgery. The study included all 48 robotic colorectal resections performed in this period by AB without the assistance of another skilled robotic surgeon. Before this independence, the surgeon had completed basic robotic simulation modules, a one-day cadaveric robotic course, and was proctored in 10 cases by another experienced robotic colorectal surgeon (3 cases externally and 7 internally proctored by the robotic colorectal surgeon (CE)).

Robotic surgical system

All surgeries were performed using the da Vinci® Si™ Surgical System except for the last five done on the da Vinci® Xi™ Surgical System following a departmental upgrade. Both systems are developed and have copyright under Intuitive Surgical based in Sunnyvale, California, USA.

Data collection

Demographics including age and gender, indication for surgery, types of operations, intra-operative time (both console time and total operative time), intra-operative complications, and 30-day postoperative outcomes (return to theatre, readmission, mortality) were recorded for all cases along with tumour location, resection margins status, the number of lymph nodes harvested and final pathological stage. For analysis, we divided the 48 cases into 4 chronological periods, each cohort having 12 cases with comparable demographics and case complexity factors.

Statistical analysis

A single reviewer collected all data through predesigned Performa, and data were analysed using SPSS version 20. Categorical variables were expressed as frequency and percentages and compared using the Chi-square test. Continuous variables expressed as means and standard deviations and means between different periods were compared using the one-way Anova test. A p value of less than 0.05 was taken as statistically significant.

We used every 6th case to plot a graph against the operative time in minutes (both console and total operative time) to find out the plateau phase of the learning curve, suggestive of the gain of maximum proficiency of skills required for robotic colorectal resections.

Results

Forty eight recto-sigmoid resections were performed during the study period; 43 cases were for rectal and anal malignancies, 4 for left colon malignancies and 1 had a resection for a large rectal polyp which was found to be benign on final histopathology. This case was sequentially included as a rectal resection as per inclusion criteria and was the only benign case included in the series in the final analysis. All cases were subsequently divided into 4 different periods/groups (12 cases in each period), and this grouping was done in chronological order based on time for comparison of different variables.

Out of 48 patients, 38 (79%) were males, and 10 (21%) were females. The mean age of our population was 65.1 ± 10.2 years. All 4 groups were similar in terms of age and gender distribution (Table 1).

Table 1 Patients’ demographics and surgical complications in four different periods
Full size table

Anterior resection of the rectum with or without anastomosis was the most commonly performed procedure in 27 (56.2%) cases, followed by extra-levator abdominoperineal resections in 17 (35.4%) cases.

Time-based metrics

Our experience showed that mean console time from docking to undocking of robot decreased from 192 min in the first period to 146 min in the last period. Similarly, the mean total operative time reduced from 363 to 328 min from the first to the fourth period (Table 2). Although with time and gain of experience, surgical time reduced but this difference was not statistically significant.

Table 2 Console time and total operative time in minutes in four different periods
Full size table

Adequacy of resection

There was no difference in the pathological staging of cancer in each period. Assessment of pathological specimen showed more than 12 lymph nodes yield from each oncological resection performed in all four periods (Table 3). All patients except one had an oncologically complete R0 resection. The single involved margin (R1) was for a tumour that had invaded beyond the anterior circumferential margin (CRM) that was missed on preoperative staging (in this case the mesorectal excision was complete).

Table 3 Comparison of effective oncological resections in four different periods
Full size table

Learning curve

When we plotted the console and total operative time of every 6th robotic resection on a linear graph, we found crests and troughs in the curve until the 30th case. From case no 30 onwards, the curve started attaining a plateau which corresponds to the surgeon’s ability to constantly perform robotic resection with steadiness and more comfort (Fig. 1).

Fig. 1
figure 1

Shows every 6th case graph, which shows console time and total operative time required in performing every 6th case

Full size image

Patient-based metrics

Intra-operative complications (minor bleeding and minor serosal injury to bowel or vaginal wall) were more common in the first period. Their number reduced as the surgeon gained more expertise in later periods. This difference was statistically significant (Table 1).

Post-operative complications were measured using the Clavien-Dindo classification of surgical complications. One patient from each first and second period required 30 days to return to theatre because of perineal wound complications (Grade III morbidity). No thirty-day mortality (Grade V) was observed in our study population.

The mean length of hospital stay in 48 cases was 9.2 days. Interestingly our study results indicated that the mean length of hospital stay decreased with time, although this was not statistically significant. This suggests that as operative time reduced, so did the length of in-patient stay (Table 4).

Table 4 Mean length of hospital stays in days of four different periods
Full size table

Discussion

In surgery, the most commonly used metrics for defining the learning curve are the number of cases in relation to operating time [9]. This is because these metrics are easier to analyze and also incorporate factors such as adequacy of resections for malignancies [10]. Patient outcomes on the other hand include operative morbidity, patient satisfaction, and quality of life. Our experience offers an assessment of robotic colorectal resections with a particular focus on pelvic resections for rectal and anal malignancies in respect of both time and patient-based metrics. These outcomes were plotted against a single surgeon, the time required to attain proficiency, and the total number of cases that were performed before sustaining a plateau on the learning curve.

Our results suggest that a newly appointed consultant surgeon requires a minimum number of 30 cases performed independently, before attaining a plateau on the learning curve. Furthermore, after achieving the learning curve, the complications are low and oncological outcomes are excellent, ensuring that patients are not adversely impacted during this learning period. Thus, the robotic platform can safely be used by newly appointed consultants even without any prior robotic experience. This has been depicted in several other studies, which have shown the learning curve for robot-assisted rectal cancer surgery to be in the region of 15–35 cases [11,12,13,14,15].

Furthermore, in our study, mean operative time for robotic colorectal resection was 293.3 ± 70 min, which is quite comparable to 253 ± 72 min found in large series of 502 cases of robotic colorectal resections presented by Parascandola et al. [9]. Interestingly on the literature review, the mean operative time for novice surgeons who had prior experience of less than 5 laparoscopic or open rectal resections was considerably higher, i.e., 397.2 ± 184.3 min[16], while the operative time in our study is 192.08 ± 62 min. This shows the importance of background experience of minimally invasive surgery before embarking on robotic procedures.

Robotic systems are designed to allow precise movements, careful dissection, and suturing in small confined spaces. Therefore, the da Vinci Surgical System has been widely used for complex procedures, such as rectal cancer surgery with a narrow pelvis undergoing total mesorectal excision (TME) [17, 18]. As demonstrated in a study done by Pigazzi et al. [19] that the operative time reduced after 20 cases while doing robotic total mesorectal excision (TME) our study also showed a reduction in mean operative time after 24 cases which dropped from 363 min in the first period to 279 min in the third period.

In their study, Shaw et al. [20] mentioned that the overall rate of complications was reduced after 15 cases. The similar observation we encountered in our study showed a significant reduction in the incidence of intra-operative complications after completing the initial twelve cases. Lymph nodes yield is one of the key determining factors for effective oncological resections. Abodeely et al. [21] published their data of a similar number of 48 patients, and their average number of lymph nodes in the specimen was 19.5, which is very much comparable to our cases of 20.7. The mean length of stay (LOS) in their study was 7 days, while it was about 9 days in our study.

Limitations

There are several limitations to our study. Selection Bias forms a major part of the cohort of cases, which is discussed as follows: only left-sided and rectal cancer patients were included; each case was sequentially added as surgeon preference and this provides no randomization; also, there is no evaluation of individual case complexity along with the exclusion of hybrid cases which precludes a non-uniform analysis of the learning curve. Most of the cases were performed on the Si system and the introduction of Xi could further decrease the operating time making this a significant outcome in future studies.

Besides, this experience needs to be reciprocated by other laparoscopic trained colorectal surgeons as this is a single surgeon experienced in a single center. The range of colorectal surgeries performed also needs to be broadened to evaluate cross differences in the learning curve. This could also be achieved using multi-dimensional analysis which can address multiple indicators of surgical performance and includes variables such as conversion rate, complications, oncological and functional outcome. Finally, multi-center experience is required for the integration of the data to set a defined robotic curriculum for training pathways in robotic colorectal surgery.

Conclusions

Robotic surgery is safe and effective to perform in recto-sigmoid malignancies by a novice laparoscopic trained colorectal surgeon. Mean console and total operative time both decreased with time in robotic surgery after a short learning curve along with low complication rates and adequate resections for cancer surgeries. This shows that with adequate training and proctoring, robotic surgical platforms can be safely adopted and utilized for good oncological and functional outcomes, and a formal robotic proficiency skills curriculum will help in this regard.