Surgical resection is the standard for curative treatment of colorectal cancer. Unfortunately, anastomotic leakage (AL) remains a major complication after resection, but its origin is still mainly unknown. The incidence of AL in the literature varies from 3% to 28%, and one-third of all postoperative mortality is related to AL.1 Delay in diagnosing AL results in increased postoperative mortality.2
In general, AL is diagnosed within the first 2 weeks after surgery.3–5 In previous studies, late AL was defined as AL diagnosed 21 or 30 days after surgery or as AL diagnosed after hospital discharge.6–11 However, a recent study advocated that redefinition of early and late AL with a proper cutoff point of a specific day is necessary for precise discrimination and they determined the cutoff at postoperative day 6.12 This demonstrates that there is no consensus in the literature regarding the definition of late AL.
Most previous studies suggested that early and late AL are different entities, although these studies were based on relatively small sample sizes.6–11 These previous studies showed that the postoperative course differs for patients with early AL and late AL. Patients with early AL are more likely to undergo relaparotomy as intervention.10 , 12 However, the long-term stoma retention rate in patients with late AL is higher than in patients with early AL.11 These differences in postoperative course emphasize that, in clinical practice, more attention should be paid to the distinction between early AL and late AL. In addition, better insight in the nature of AL could also contribute to early detection of AL, especially of late AL, because one-third of AL is diagnosed after 30 days after surgery.13
A technically not well-constructed anastomosis might result in immediate anastomotic dehiscence with subsequent clinical symptoms, whereas a well-constructed anastomosis will develop anastomotic dehiscence more slowly in case wound healing is impaired. The aim of this study was to assess the differences in risk factors for early and late AL to demonstrate whether early AL is related to technical failure of the anastomosis and late AL is related to healing deficiencies.
MATERIALS AND METHODS
Data were derived from the Dutch ColoRectal Audit (DCRA), a nationwide quality improvement project that collects information on all Dutch patients undergoing surgical resection for primary colorectal cancer. Data registered were patient characteristics, tumor and treatment characteristics, as well as patient outcomes. For this study, no ethical approval or informed consent was required under Dutch law. Further details of this data set regarding collection and methodology have been published previously.14
Inclusion and Exclusion Criteria
All patients undergoing surgical resection for primary colorectal cancer in the Netherlands between January 2011 and December 2015, and registered in DCRA before March 31, 2016, were included in this study. Patients without a primary anastomosis and patients for whom the day of diagnosis of AL was unknown were excluded from analysis. Patients in whom AL occurred later than 90 days after surgery were excluded. Data are usually registered at 30 days after surgery unless the initial hospital stay takes longer. Therefore, we considered data registered about AL later than 90 days after surgery as unreliable. We excluded patients with multiple synchronous tumors because of differences in prognosis.15
Early Versus Late AL
Anastomotic leakage was defined as clinically relevant AL that requires radiological or surgical reintervention.16 We defined early AL as AL leading to reintervention until day 6 postoperatively and late AL as AL leading to reintervention after day 6 postoperatively. In previous literature, there is no consensus on the definition of late AL. To test our hypothesis, it was not sufficient to base our definition on the day of discharge, which is highly sensitive to institutes and other postoperative complications. Although it might be a fluent transition for early to late AL, for precise discrimination, we think it is important to use a definition based on a specific day. Besides, the transition in origin of AL from technical failure to healing deficiencies should be captured during the first postoperative days.
Early and late ALs were primary outcome measures. Potential risk factors for early and late AL were retrieved from the DCRA database including patient characteristics (sex, age, BMI, Charlson Comorbidity Index,17 , 18 ASA score,19 and previous abdominal surgery), tumor characteristics (tumor location, tumor stage, metastasis, and preoperative tumor complication), and treatment characteristics (surgical technique, urgency of surgery, diverting ileostomy, and additional resection of adjacent organs because of tumor growth or because of metastasis and preoperative radiotherapy). In the DCRA database, preoperative tumor characteristics are specified as anemia, ileus, abscess, and perforation.
Multiple imputations were performed to deal with missing values assuming data were missing at random.20 Five imputed data sets have been created based on AL, hospital, sex, Charlson Comorbidity Index, diverting ileostomy, metastasis, and preoperative tumor complications. Multivariate multinominal logistic regression analyses were performed to test independent associations between patient, tumor, and treatment characteristics and the occurrence of no, early, and late AL. A multinominal logistic regression model is applicable when an outcome variable has more than 2 categories, but no ordering in these categories can be assumed. All clinically relevant variables were added to the model as independent variables (full model). Covariate selection was driven by available knowledge and biological plausibility of potential confounders. Tests for interactions between covariates were not implemented. More details concerning the relevant predictors of AL were described elsewhere.21 , 22 Results were reported as ORs with 95% CIs. Significance was considered as a p value of <0.05. All statistical analyses were performed in SPSS version 22.
A total of 49,941 patients underwent surgery for primary colorectal cancer in the Netherlands between 2011 and 2015. After exclusion of patients without a primary anastomosis (n = 11,246), 38,695 patients were eligible for inclusion. Patients in whom AL occurred later than 90 days after surgery and patients for whom the day of diagnosis of AL was unknown (n = 558) were excluded from analysis. Because of differences in prognosis, we excluded patients with multiple synchronous tumors (n = 1208). In total, 36,929 patients were included.
Of these 36,929 patients, 80.9% underwent surgery for a colon tumor and 63.1% underwent laparoscopic surgery (Table 1).
AL leading to reintervention occurred in 1537 (4.2%) patients. Early AL occurred in 863 (2.3%) patients and late AL occurred in 674 (1.8%) patients. The median interval between colorectal resection and intervention for AL was 6 days. The median interval between colorectal resection and intervention for early AL was 4 days, and for late AL, the median interval between colorectal resection and intervention was 10 days. In 18%, AL was diagnosed after hospital discharge. In patients with early AL, 3.1% were diagnosed after hospital discharge, and in patients with late AL, 37.4% were diagnosed after hospital discharge. The incidence of early AL in patients with a colon tumor was 2.3% and, in patients with a rectum tumor, incidence of early AL was 2.4%, whereas the incidence of late AL was 1.6% in patients with a colon tumor and 3.0% in patients with a rectum tumor.
From a multivariable multinomial logistic regression model, independent predictors of early AL relative to no AL and late AL relative to no AL included male sex (OR, 1.8; 95% CI, 1.6–2.1; p < 0.001 and OR, 1.2; 95% CI, 1.0–1.4; p = 0.013) and rectal cancer (OR, 2.1; 95% CI, 1.6–2.8; p < 0.001 and OR, 1.6; 95% CI, 1.0–2.4; p = 0.046). Additional independent predictors of early AL relative to no AL included BMI (OR, 1.1; 95% CI, 1.0–1.2; p = 0.001), laparoscopic surgery (OR, 1.2; 95% CI, 1.0–1.4; p = 0.019), emergency surgery (OR, 1.8; 95% CI, 1.4–2.2; p < 0.001), and no diverting ileostomy (OR, 0.3; 95% CI, 0.2–0.4; p < 0.001). Independent predictors of late AL relative to no AL were Charlson Comorbidity Index of ≥II (OR, 1.3; 95% CI, 1.1–1.6; p = 0.003), ASA score III to V (OR, 1.2; 95% CI, 1.0–1.5; p = 0.030), preoperative tumor complications (OR, 1.1; 95% CI, 1.0–1.4; p = 0.048), extensive additional resection because of tumor growth (OR, 1.7; 95% CI, 1.2–2.5; p = 0.003), and preoperative radiation (OR, 2.0; 95% CI, 1.2–3.4; p = 0.010) (Table 2).
Independent predictors for early AL relative to late AL were male sex (OR, 1.5; 95% CI, 1.2–1.9), laparoscopic surgery (OR, 1.3; 95% CI, 1.0–1.6), emergency surgery (OR, 1.9; 95% CI, 1.3–2.7), no diverting ileostomy (OR, 0.4; 95% CI, 0.2–0.6), and no preoperative radiotherapy (OR, 0.4; 95% CI, 0.2–0.8). These variables had a different effect on the occurrence of early AL compared with late AL (Table 2).
In addition, stratification for colon and rectum showed that diverting ileostomy and preoperative radiotherapy were independent risk factors for late AL in rectum tumors but not for colon tumors. Furthermore, in the stratified analysis, open surgery was an independent risk factor for early AL in colon tumors, whereas laparoscopic surgery was an independent risk factor for early AL in rectum tumors.
This study showed that male sex and rectal cancer were independent risk factors for both early and late AL. Younger age, increased BMI, laparoscopic surgery, emergency surgery, and no diverting ileostomy were independent risk factors for early AL. In addition, high Charlson Comorbidity Index, high ASA score, preoperative complications, additional resection because of tumor growth, and preoperative radiotherapy were independent risk factors for late AL. Male sex, laparoscopic surgery, emergency surgery, construction of diverting ileostomy, and preoperative radiotherapy had a different effect on the occurrence of early, compared with late AL. Our results demonstrated that most risk factors for early AL were surgery-related factors, representing surgical difficulty, which might lead to technical failure of the anastomosis. Most risk factors for late AL were patient-related factors, representing the frailty of patients and tissues that influences the healing capacity of bowel tissue.
The results of this study are in accordance to previous literature. A previous study showed that prolonged duration of surgery and blood loss during surgery, both representing surgical difficulty, were related to early AL.6 Another study indicated that preoperative radiotherapy or chemoradiation was a risk factor for late AL.7 , 10 , 11 Although it has been demonstrated that advanced tumor stage (American Joint Committee on Cancer stage III–IV) and a histological finding of poorly differentiated or mucinous adenocarcinoma were independent risk factors for early AL, this study did not find this.11 One study found a low incidence of late AL (0.04%), and the authors attributed this to the extended period of pelvic drainage, which may have shortened the interval of diagnosis.8 On the contrary, another study reported an incidence of AL after 30 days postoperatively of 31.6%.23 It should be taken into account that these previous studies applied different definitions for late AL. Besides, these previous studies were based on relatively small sample sizes.
Male sex was an independent risk factor for both early AL and late AL, and thus for AL in general. However, male sex seemed to be a greater risk factor for early AL in comparison with late AL. This could be attributable to the smaller pelvis and stronger muscular wall in men, which impedes surgery. Furthermore, rectal cancer was an independent risk factor for both early and late AL, which can be explained by the fact that the risk of AL, in general, is increased for anastomoses situated closer to the anal verge.24
Younger age and increased BMI were independent risk factors for early AL, possibly because younger patients are less prone to healing deficiencies. Also, increased BMI is associated with AL in colorectal surgery.25 The increased mesocolon thickness and abdominal pressure in obese patients may complicate the construction of the anastomosis. However, increased BMI is related to impaired microcirculation, which is considered to decrease the healing capacity at the anastomotic site, which may also play a role for late AL, although this study did not demonstrate this.
Furthermore, we found that laparoscopic surgery was an independent risk factor only for early AL. The COLOR study indicated that the incidence of AL does not differ between laparoscopic and open surgery.26 , 27 Nevertheless, it has been shown recently that risk factors for AL are different between laparoscopic and open surgery. Risk factors for AL after laparoscopic surgery were related to surgical difficulty.28 This is in accordance to our findings and hypothesis. Furthermore, in the early years of laparoscopic surgery, some comorbidities were considered as contraindications for laparoscopic surgery. Therefore, in this observational study, we should take into consideration that selection bias might have affected our results even though we have adjusted for comorbidities in the multivariate analysis.
Emergency surgery was also identified as an independent risk factor for early AL. Emergency surgery is often performed during evening and night shifts because of acute indications. Surgery at these hours is associated with worse postoperative outcomes.29 , 30 Colorectal surgery performed during evening and night shifts is related to AL.31 Surgery at these times might be performed by less specialized surgeons implying surgical difficulty due to less experience is highly suggestive for technical failure of the anastomosis.
Preoperative tumor complications were heterogeneous in influencing our hypothesis, because these not only represented surgical difficulty, but also frailty of patients’ tissue at the anastomotic site. Nevertheless, our results proposed that preoperative tumor complications were an independent risk factor for late AL. Table 1 showed that almost 20% of the preoperative tumor complications were anemia, which may lead to reduced healing capacity at the anastomotic site. Furthermore, ileus could also strongly affect the quality of bowel tissue, but this also represents surgical difficulties in constructing the anastomosis.
No diverting ileostomy was an independent risk factor for early AL. From DCRA, it was previously shown that stoma construction in rectal surgery does not affect the incidence of AL or mortality rates.32 In addition, it was recently shown that, when AL occurred in patients with a diverting ileostomy, fewer reinterventions were required, which could be suggestive for less severe clinical presentation of AL.33 It might be possible that a diverting ileostomy delays the diagnosis of AL because of less severe presentation of AL.
Also, additional resection because of tumor growth was an independent risk factor for late AL. These major surgical procedures are demanding for patients, among others because of blood loss and hypotension during surgery. Hypotension compromises local tissue perfusion and leads to reduced tissue oxygenation, causing healing deficiencies. Surgery with additional resections because of tumor growth is also technically demanding but might not specifically complicate the construction of the anastomosis.
Preoperative radiotherapy is indicated in most cases of rectal cancer.34 Preoperative radiotherapy reduces the incidence of local recurrence but is also related to higher postoperative morbidity.35 Our results showed that preoperative radiotherapy was an independent risk factor for late AL. Preoperative radiotherapy not only affects tumor tissue, but also the surrounding healthy tissue, including the adjacent bowel wall and its vascularization. This could imply decreased healing capacity at the anastomotic site and therefore be related to late AL.
In addition, stratification for colon and rectum showed comparable results. As expected, diverting ileostomy and preoperative radiotherapy were not a risk factor for late AL in colon resections possibly due to the small numbers, because these strategies are usually not applied in the treatment of colon tumors. Furthermore, in the stratified analysis, open surgery was an independent risk factor for early AL in colon tumors, whereas laparoscopic surgery was an independent risk factor for early AL in rectum tumors. However, because the laparoscopic approach reflects a technically challenging procedure, it is possible that, in colon surgery, the open approach was used more often for difficult cases, resulting in selection bias.
In this study, the cutoff between early and late AL was set on 6 days based on the median. However, the transition from early AL, hypothesized to be related to surgical difficulty, to late AL, hypothesized to be related to the frailty of tissue and patients, might not be captured at this exact day, and the transition might very well be a more fluent process. Therefore, we could not state that there are 2 separate populations of AL, but our findings indicate that, within the group of AL, there might be different entities.
This distinction in origin between early and late AL also has implications for fair comparison of the quality of hospitals. In early AL, the technical skills of the surgeon have more influence, and, hence, the surgeon could be more accountable, whereas, for late AL, the patient characteristics might be of more influence.
There were some limitations in our study. First, the definition of late AL was arbitrary. This study only evaluated clinically relevant AL that required reintervention, and, therefore, the definition of late AL was based on the day of reintervention. For this retrospective study, registration of the day of intervention was more reliable than the day of clinical symptoms. In previous studies, late AL was defined as AL diagnosed after hospital discharge, after 6, 21, or 30 days postoperatively. Most previous studies aimed to determine whether there are 2 entities of colorectal AL. However, we hypothesized that the time of occurrence of AL reflects the origin of leakage. Therefore, we defined late AL as AL leading to reintervention after day 6 postoperatively, which was the median (postoperative day 6). Because data were available, we have also performed the analysis with the cutoff point of late AL at first quartile of discharge (day 5) and third quartile of discharge (day 10). These analyses did not fundamentally change the results presented in our study and the conclusion was similar.
Second, the DCRA data are usually registered until 30 days after surgery unless the initial hospital stay is longer. Therefore, extreme late AL is not included. Besides, he underregistration of AL, in general, might be a problem in nationwide databases. Last, the analysis of observational data could be affected by confounding, and this might lead to bias. Although we performed a multivariate analysis to adjust for patient, tumor, and treatment characteristics, still unknown confounding factors could be present that were not registered in the DCRA, such as medication use, smoking, criteria for diverting ileostomy, mobilization of splenic flexure, blood loss, and operative time.
However, the strength of this study was that results were based on a nationwide cohort representing a large sample size that induces statistical power to detect differences between risk factors for early and late AL. Furthermore, previous studies were only concerned with AL of the sigmoid and rectum; we have now shown that our hypothesis may be applicable to AL of all colorectal cancers.
This study demonstrated that early and late ALs have different risk factors. Our findings suggest that risk factors for early AL are related to surgical difficulty that may lead to technical failure of the anastomosis, resulting in immediate anastomotic dehiscence, whereas risk factors for late AL are related to the frailty of patients and tissues, which may imply healing deficiencies at the anastomotic site leading to delayed anastomotic dehiscence in a possibly technically well-constructed anastomosis. In our opinion, especially in patients with high risk for late AL, it is important for surgeons to inform patients about the possible occurrence of AL in the late postoperative period, especially because 18% of AL occurs after hospital discharge. Furthermore, in early AL, the quality of the surgery seems to have more influence than in late AL, so hospital comparisons should consider the different entities separately, with different case-mix adjustments.
The authors would like to thank all surgeons and other health care providers who are involved in registering patients in the Dutch Surgical Colorectal Audit.
1. Snijders HS, Wouters MW, van Leersum NJ, et alMeta-analysis of the risk for anastomotic leakage, the postoperative mortality caused by leakage in relation to the overall postoperative mortality. Eur J Surg Oncol. 2012;38:1013–1019.
2. Macarthur DC, Nixon SJ, Aitken RJAvoidable deaths still occur after large bowel surgery. Scottish Audit of Surgical Mortality, Royal College of Surgeons of Edinburgh. Br J Surg. 1998;85:80–83.
3. Alves A, Panis Y, Pocard M, Regimbeau JM, Valleur PManagement of anastomotic leakage after nondiverted large bowel resection. J Am Coll Surg. 1999;189:554–559.
4. Hyman N, Manchester TL, Osler T, Burns B, Cataldo PAAnastomotic leaks after intestinal anastomosis: it’s later than you think. Ann Surg. 2007;245:254–258.
5. Kanellos I, Vasiliadis K, Angelopoulos S, et alAnastomotic leakage following anterior resection for rectal cancer. Tech Coloproctol. 2004;8(suppl 1):s79–s81.
6. Floodeen H, Hallböök O, Rutegård J, Sjödahl R, Matthiessen PEarly and late symptomatic anastomotic leakage following low anterior resection of the rectum for cancer: are they different entities? Colorectal Dis. 2013;15:334–340.
7. Lim SB, Yu CS, Kim CW, Yoon YS, Park IJ, Kim JCLate anastomotic leakage after low anterior resection in rectal cancer patients: clinical characteristics and predisposing factors. Colorectal Dis. 2016;18:O135–O140.
8. Maeda H, Okamoto K, Namikawa T, et alRarity of late anastomotic leakage
after low anterior resection of the rectum. Int J Colorectal Dis. 2015;30:831–834.
9. Matthiessen P, Lindgren R, Hallböök O, Rutegård J, Sjödahl RRectal Cancer Trial on Defunctioning Stoma Study Group. Symptomatic anastomotic leakage diagnosed after hospital discharge following low anterior resection for rectal cancer. Colorectal Dis. 2010;12(7 online):e82–e87.
10. Morks AN, Ploeg RJ, Sijbrand Hofker H, Wiggers T, Havenga KLate anastomotic leakage in colorectal surgery
: a significant problem. Colorectal Dis. 2013;15:e271–e275.
11. Shin US, Kim CW, Yu CS, Kim JCDelayed anastomotic leakage following sphincter-preserving surgery for rectal cancer. Int J Colorectal Dis. 2010;25:843–849.
12. Li YW, Lian P, Huang B, et alvery early colorectal anastomotic leakage within 5 post-operative days: a more severe subtype needs relaparatomy. Sci Rep. 2017;7:39936.
13. Borstlap WAA, Westerduin E, Aukema TS, Bemelman WA, Tanis PJDutch Snapshot Research Group. Anastomotic leakage and chronic presacral sinus formation after low anterior resection: results from a large cross-sectional study. Ann Surg. 2017;266:870–877.
14. Van Leersum NJ, Snijders HS, Henneman D, et alDutch Surgical Colorectal Cancer
Audit Group. The Dutch surgical colorectal audit. Eur J Surg Oncol. 2013;39:1063–1070.
15. van Leersum NJ, Aalbers AG, Snijders HS, et alSynchronous colorectal carcinoma: a risk factor in colorectal cancer
surgery. Dis Colon Rectum. 2014;57:460–466.
16. Rahbari NN, Weitz J, Hohenberger W, et alDefinition and grading of anastomotic leakage following anterior resection of the rectum: a proposal by the International Study Group of Rectal Cancer. Surgery. 2010;147:339–351.
17. Charlson M, Szatrowski TP, Peterson J, Gold JValidation of a combined comorbidity index. J Clin Epidemiol. 1994;47:1245–1251.
18. Charlson ME, Pompei P, Ales KL, MacKenzie CRA new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–383.
19. Soliani PNew classification of physical status. Anesthesiology. 1963;24:111.
20. Sterne JA, White IR, Carlin JB, et alMultiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ. 2009;338:b2393.
21. Kolfschoten NE, Marang van de Mheen PJ, Gooiker GA, et alDutch Surgical Colorectal Audit group. Variation in case-mix between hospitals treating colorectal cancer
patients in the Netherlands. Eur J Surg Oncol. 2011;37:956–963.
22. Kolfschoten NE, Wouters MW, Gooiker GA, et alDutch Surgical Colorectal Audit group. Nonelective colon cancer resections in elderly patients: results from the Dutch Surgical Colorectal Audit. Dig Surg. 2012;29:412–419.
23. Tan WP, Hong EY, Phillips B, Isenberg GA, Goldstein SDAnastomotic leaks after colorectal anastomosis occurring more than 30 days postoperatively: a single-institution evaluation. Am Surg. 2014;80:868–872.
24. Rullier E, Laurent C, Garrelon JL, Michel P, Saric J, Parneix MRisk factors for anastomotic leakage after resection of rectal cancer. Br J Surg. 1998;85:355–358.
25. Frasson M, Flor-Lorente B, Rodríguez JL, et alANACO Study Group. Risk factors for anastomotic leak after colon resection for cancer: multivariate analysis and nomogram from a multicentric, prospective, national study with 3193 patients. Ann Surg. 2015;262:321–330.
26. Bonjer HJ, Deijen CL, Abis GA, et alCOLOR II Study Group. A randomized trial of laparoscopic versus open surgery for rectal cancer. N Engl J Med. 2015;372:1324–1332.
27. Veldkamp R, Kuhry E, Hop WC, et alCOlon cancer Laparoscopic or Open Resection Study Group (COLOR). Laparoscopic surgery versus open surgery for colon cancer: short-term outcomes of a randomised trial. Lancet Oncol. 2005;6:477–484.
28. Kim CW, Baek SJ, Hur H, Min BS, Baik SH, Kim NKAnastomotic leakage after low anterior resection for rectal cancer is different between minimally invasive surgery and open surgery. Ann Surg. 2016;263:130–137.
29. Chacko AT, Ramirez MA, Ramappa AJ, Richardson LC, Appleton PT, Rodriguez EKDoes late night hip surgery affect outcome? J Trauma. 2011;71:447–453.
30. Coumbe A, John R, Kuskowski M, Agirbasli M, McFalls EO, Adabag SVariation of mortality after coronary artery bypass surgery in relation to hour, day and month of the procedure. BMC Cardiovasc Disord. 2011;11:63.
31. Komen N, Dijk JW, Lalmahomed Z, et alAfter-hours colorectal surgery
: a risk factor for anastomotic leakage. Int J Colorectal Dis. 2009;24:789–795.
32. Snijders HS, van Leersum NJ, Henneman D, et alOptimal treatment strategy in rectal cancer surgery: should we be cowboys or chickens? Ann Surg Oncol. 2015;22:3582–3589.
33. Boyce SA, Harris C, Stevenson A, Lumley J, Clark DManagement of low colorectal anastomotic leakage in the laparoscopic era: more than a decade of experience. Dis Colon Rectum. 2017;60:807–814.
34. Peeters KC, Marijnen CA, Nagtegaal ID, et alDutch Colorectal Cancer
Group. The TME trial after a median follow-up of 6 years: increased local control but no survival benefit in irradiated patients with resectable rectal carcinoma. Ann Surg. 2007;246:693–701.
35. Cedermark B, Johansson H, Rutqvist LE, Wilking NThe Stockholm I trial of preoperative short term radiotherapy in operable rectal carcinoma. A prospective randomized trial. Stockholm Colorectal Cancer
Study Group. Cancer. 1995;75:2269–2275.