Colorectal cancer (CRC) is a major cause of cancer mortality and morbidity, and it has been reported that this cancer contributes to approximately 10% of the cancer mortality rate [
Many attempts have been made to decrease the rate of AL after rectal cancer surgery. A diverting stoma has been reported to reduce the rate of anastomotic failure; however, this remains controversial [
Various strategies have been sequentially used at our institution to reduce the incidence of AL after laparoscopic rectal cancer surgery. A direct comparison of the strategies might result in serious selection bias and failure to obtain a high clinical significance. Therefore, the development of a prediction model of AL after surgery for rectal cancer and the determination of the risk-reducing factors in controllable strategies are very important. In this regard, the present study aimed to construct a prediction model and identify the most effective strategy for reducing AL in patients treated with laparoscopic rectal cancer surgery.
The present study enrolled 736 consecutive patients with rectal adenocarcinoma who underwent laparoscopic resection performed by a single surgeon (KHR) between August 2004 and February 2015. All included patients had histologically confirmed rectal adenocarcinoma and primary anastomosis. Conventionally, the rectum is divided into three parts based on the anatomic distance from the anal verge: the upper rectum (8–12 cm), mid rectum (4–8 cm), and lower rectum (0–4 cm). The exclusion criteria were the presence of a tumor location above 12 cm from the anal verge, anastomosis performed using a hand-sewn method, and the use of a diverting stoma. This study was reviewed and approved by the institutional review board of our hospital. The surgical technique of laparoscopic surgery for rectal cancer has been described previously [
AL was investigated at the surgeon’s discretion on the basis of clinical symptoms of sepsis, including abdominal pain, tenderness, rebound tenderness, fever, and leukocytosis. It was suspected clinically if pus or fecal discharge was noted from the pelvic drain. All ALs were confirmed by using rigid sigmoidoscopy, abdominopelvic computed tomography, or operative findings.
Each of the strategies has been implemented since its initial use throughout the duration of the study period.
The application of fibrin glue over a stapled anastomosis site was routinely performed since August 2007. In this study, it was applied from the 155th consecutive patient. In the patients, 1-2 mL of Tissuecol (Baxter, Vienna, Austria) or Greenplast (Green Cross Corporation, Yongin, Korea) was used over the extraluminal anastomosis surface [
Reinforcing sutures were used since January 2011. In this study, the sutures were used from the 397th consecutive patient. After anastomosis was performed, reinforcing 4-0 PDS (Ethicon Inc., Summerville, NJ) was used intraorally. At least two interrupted sutures were performed, and the sutures always included the point at which the circular and linear stapling line met.
Extended SFM was performed since December 2011. In this study, it was performed from the 480th consecutive patient. After ligation of the inferior mesenteric vein at the lower border of the pancreas, dissection was continued over the anterior surface of the pancreas to the splenic hilum until the lesser sac entered. Then, the splenic flexure was easily mobilized, the lateral ligament was divided, and the omentum was subsequently dissected from the colon [
A transanal drainage tube was used since January 2013. In this study, it was used from the 584th consecutive patient. Following anastomosis, a 10-Fr rubber catheter with two or three holes near the proximal tip was placed in the neorectum.
The
The nomogram was validated internally with 240 bootstrap resamples. The validated function in the rms package was used to calculate the bias-corrected C-index, which was calculated by using Somers’ Dxy rank correlation as follows:
Because the estimated risk of AL varies significantly among patients, an adjustment for risk was performed. We used RA-CUSUM analysis on the basis of individual risk derived from the logistic regression model. The statistical principles were adapted from the tutorial by Steiner et al. [
The present study included 736 patients. Among these patients, clinical AL occurred in 65 patients (8.8%) and relaparotomy was performed in 53 patients (7.2%). The detailed characteristics of the patients are presented in Table
Univariate analysis of risk factors for anastomotic leakage in patients treated with laparoscopic rectal cancer surgery without diverting stoma (
Variables | Number of anastomotic leakage/total patients (%) |
|
---|---|---|
Sex | 0.002 | |
Female | 12/272 (4.4) | |
Male | 53/464 (11.4) | |
Age, yr | 0.999 | |
≥70 | 29/330 (8.8) | |
<70 | 36/406 (8.9) | |
BMI (kg/m2) | 0.839 | |
<25 | 46/507 (9.1) | |
≥25 | 19/229 (8.3) | |
ASA score | <0.001 | |
1 | 14/280 (7.8) | |
2 | 41/517 (7.9) | |
3 | 10/39 (25.6) | |
AJCC stage | 0.163 | |
0-II | 33/439 (7.5) | |
II/IV | 32/297 (10.8) | |
Maximum tumor size (cm) | 0.896 | |
<4 | 24/320 (7.5) | |
≥4 | 29/416 (7.0) | |
Location of tumor | 0.005 | |
Upper | 25/447 (5.6) | |
Mid | 16/215 (7.4) | |
Low | 12/74 (16.2) | |
Operative time (min) | 0.003 | |
<240 | 37/624 (5.9) | |
≥240 | 16/112 (14.3) | |
Transfusion | <0.001 | |
No | 41/674 (6.1) | |
Yes | 12/62 (19.4) | |
Neoadjuvant chemoradiation | 0.051 | |
No | 42/651 (6.5) | |
Yes | 11/85 (12.9) | |
Number of linear stapler firing | 0.061 | |
<2 | 20/376 (5.3) | |
≥2 | 33/360 (9.2) |
AJCC: American Joint Committee on Cancer; ASA: American Society of Anesthesiologists, BMI: body mass index.
On univariate analysis, male sex, a high American Society of Anesthesiologists (ASA) score, low rectal cancer, perioperative blood transfusion, and a long operation time were identified as significant risk factors for AL (Table
Multivariate analysis of risk factors associated with anastomotic leakage.
Variables | Relative risk | 95% CI |
|
---|---|---|---|
Sex | |||
Male | 1 | ||
Female | 0.272 | 0.129–0.526 | <0.001 |
ASA score | |||
1/2 | 1 | ||
3 | 3.818 | 1.587–8.622 | 0.002 |
Location of tumor | |||
Upper | 1 | ||
Mid | 1.757 | 0.913–3.331 | 0.085 |
Low | 3.721 | 1.761–7.635 | <0.001 |
Operative time (min) | 1.343 | 1.082–1.668 | 0.008 |
Transfusion | |||
No | 1 | ||
Yes | 3.495 | 1.624–7.172 | <0.001 |
ASA: American Society of Anesthesiologists; CI: confidence interval.
A nomogram with significant risk factors was developed (Figure
A nomogram for predicting postoperative anastomotic leakage after laparoscopic rectal cancer surgery. To use the nomogram, we first drew a vertical line to the top “Points” row to assign points for each variable. Then, we summed the total points and drew vertical line from the “Total points” row to obtain the probability of anastomotic leak.
The calibration plots showed that the model was very close to the ideal (Figure
A calibration plot of the predicted and observed probabilities of anastomotic leakage after laparoscopic rectal cancer surgery. The
Based on the individual probability for AL after laparoscopic rectal cancer surgery, the RA-CUSUM graphical slope was determined and plotted according to the final surgical outcome. The RA-CUSUM graph for AL is presented in Figure
Risk-adjusted cumulative sum curve analysis for anastomotic leakage after laparoscopic rectal cancer surgery. The cut-off points were at the 70th case and the 500th case. Each of the strategies has been implemented since its initial use throughout the duration of the study period. A: application of fibrin glue, B: use of reinforcing sutures, C: extended medial-to-lateral splenic flexure mobilization, and D: use of a transanal drainage tube.
We constructed a nomogram for predicting AL and found that extended SFM and the surgeon’s learning curve played important roles in the reduction of AL after laparoscopic rectal cancer surgery. These findings may have clinical implications in the careful selection of candidates for diverting ileostomy based on our nomogram. Additionally, we suggest that extended SFM could be considered to reduce anastomosis tension when a more extended rectal resection is needed, for example, in patients with low rectal cancer and those preoperatively treated with CRT. To our knowledge, this is the first study that has used a nomogram and RA-CUSUM analysis to evaluate whether a surgical strategy can reduce AL after laparoscopic rectal cancer surgery.
The overall clinical AL rate in this study was 8.8%, and 81.5% of the patients underwent surgical intervention. The AL rate in this study is comparable to the rates of previous studies (3–21%) and the rates after laparoscopic surgery for rectal cancer [
Our study found that male sex, a high ASA score (≥3), low rectal cancer, perioperative blood transfusion, and a long operation time were risk factors for AL. These risk factors can be categorized as patient-related factors (sex and ASA score), tumor-related factors (tumor location), and surgery-related factors (operation time and blood transfusion). The risk of AL was 3.7-fold higher in male patients than that in female patients in this study. This was also noted in previous studies [
The combination of risk factors is very important, and the use of a nomogram is simple. For example, patients with the lowest risk were estimated to have an AL risk of only 1.6% after surgery, while patients with the highest risk were estimated to have an AL risk of 68.0%, implying that risk adjustment is the most critical step for sequential statistical analysis. For the RA-CUSUM curve constructed in this study, the following possible scores were plotted in the graphical presentation; if AL occurred, the score is 0.984 for patients with the lowest risk and 0.32 for patients with the highest risk after surgery. The rising curve was approximately three times steeper for patients with the lowest risk than for those with the highest risk. We demonstrated two change points at the 70th case and the 505th case. We speculate that the former might be attributed to the surgeon’s learning curve and the latter might be attributed to the protective effect of extended SFM for AL. Previous studies have assessed whether fibrin glue application [
To determine the strategy that is the most appropriate for the reduction of AL after laparoscopic rectal cancer surgery, many factors have to be controlled and an expert surgeon is required. In the present study, we adjusted many nonmodifiable variables (patient and tumor factors) known to be associated with AL after rectal cancer surgery, using RA-CUSUM analysis. The surgeon could control tension at the anastomosis site, vascular supply, stapler use, and diverting stoma creation. We believe that reduction of tension at the anastomosis site is the most critical factor, and this can be assessed in the operating room. In our study, after more than 500 cases of laparoscopic rectal cancer surgery, analyses to determine whether the use of a transanal drainage tube could reduce the AL rate were performed. We did not find a significant association, even using a propensity score analysis [
The present study has several limitations. First, preoperative CRT has been shown to be a potential risk factor for AL [
Our nomogram for predicting AL after laparoscopic rectal cancer surgery might be helpful to identify the individual risk of AL. Furthermore, extended SFM might be the most appropriate strategy for reducing AL in patients treated with laparoscopic cancer surgery.
Drs. Chang Hyun Kim, Soo Young Lee, Hyeong Rok Kim, and Young Jin Kim have no conflicts of interest or financial ties to disclose.