Introduction

Colorectal cancer (CRC) accounts for a substantial proportion of the global cancer burden, affecting approximately 1.8 million people and resulting in approximately 900,000 cancer-related deaths each year [1]. Surgery remains the primary treatment modality for the majority of patients with CRC, but despite recent achievements in minimally invasive cancer surgery, morbidity and mortality due to post-operative complications remains an immense challenge. Anastomotic leakage (AL) continues to be one of the most common and feared complications, with an incidence of 5 to 20% after low anterior resection for rectal cancer [2,3,4,5,6,7,8,9]. The consequences of AL can be severe, with subsequent mortality rates ranging from 6 to 22% [2,3,4,5,6,7,8,9]. Other potential sequelae include the need for re-operation, persistent stomas, and an increased risk of locoregional recurrence [10, 11]. Several risk factors for AL have been previously identified and include male gender, a low anastomosis, age, comorbidity, and peri-operative bleeding [12, 13].

The present study addresses the use of a diverting stoma in connection with low anterior resection in rectal cancer treatment. A diverting stoma may be constructed during the primary operation in order to protect the healing anastomosis, with the most common choices being a loop ileostomy or loop transverse colostomy. The use of a diverting stoma has been shown to mitigate the clinical impact, and in some studies also the incidence, of AL [14]. However, the use of diverting stoma is not without complications and their routine use remains a matter of international debate. Although initially intended as a temporary measure, such stomata are never reversed in up to 20% of patients [15]. In addition, diverting ileostomies may be complicated by high output and subsequent renal failure, whereas transverse colostomies tend to prolapse and lead to local cutaneous complications. In addition, stoma reversal has its own range of complications, with reported mortality rates as high as 3% [16].

The aim of the present study was to examine the associations between diverting stomata and the clinical consequences of AL, using the Danish colorectal cancer database.

Methods

This was a nationwide cohort study including all patients suffering from AL after low anterior resection from 2001 to 2010. The data was extracted from the Danish Colorectal Cancer Group database, which contains data from 98.8% of all CRC operations performed in Denmark from 2001 to 2010. All patients included in the study were recorded in the DCCG database between May 2001 and December 2010 with a primary diagnosis of rectal adenocarcinoma. All patient files were collected and reviewed by one observer (NS). A database collecting relevant information was established, including demography; primary surgery; post-operative care; methods of diagnosis of AL; treatment of AL; data regarding stoma and reversal, short-term, and long-term complications; cancer-specific mortality; and all-cause mortality. For the present study, data related to the subject of interest were extracted from the database and analyzed. The study was approved by the Danish Data Protection Agency.

The primary outcome of the present study was a 90-day mortality; the secondary outcomes were all-cause mortality, morbidity associated with AL, length of stay, complications related to stoma, and risk of permanent stoma.

Statistics

Survival analyses were performed using the Kaplan-Meier method. T test and chi-square were used to test null hypothesis. The results of the multivariable analyses are presented as hazard ratios (HR) with 95% confidence intervals (CI). All analyses were 2-sided and considered statistically significant if p < 0.05. Data were analyzed with SPSS Statistics version 22.0 (IBM Corp, Armonk, NY).

Results

In the 9-year period, a total of 4063 patients were operated for primary rectal cancer with the formation of an anastomosis. Five hundred eighty-one patients developed an AL (11.9%), although a total of 101 patients were lost to subsequent follow-up. As such, 480 patients were included in our analyses. The median follow-up of patients at the time of analysis was 3 years (range 2 days to 12.5 years). Patients were stratified into 2 groups, those with a diverting stoma (DS group) and those without (non-DS group). The non-DS group comprised 228 patients (47.5%). The DS group comprised 252 patients (52.5%). The overwhelming majority of patients in the DS group received a loop ileostomy (98%), with only 2% of patients receiving a diverting colostomy. Patient characteristics and surgical data are shown in Tables 1 and 2.

Patients in the non-DS group became symptomatic following AL at an earlier post-operative time point and, as such, were diagnosed significantly earlier when compared with the DS group. The median post-operative day of was day 6 in the non-diverted group compared with day 9 in patients with a diverting stoma (p < 0.001). The methods of diagnosis also differed between the two patient groups. While radiological assessment (CT scan, contrast enema, and X-ray) was predominant in the non-DS group, digital rectal exam and endoscopy were more likely to be used in the DS group (Table 3).

Table 1 Clinicopathological demographics of patients developing an anastomotic leakage after low anterior resection for rectal cancer. BMI, body mass index; ASA, American Society of Anesthesiology; UICC, Union for International Cancer Control
Full size table

Significant differences in the treatment of AL were also noted between these patient groups. While more than half (133 patients, 53%) were managed non-surgically in the DS group, only 14 patients (6%) of patients in the non-DS group were managed in this fashion. Concordantly, the number of patients undergoing a Hartmann procedure was significantly higher in the non-DS group compared with those patients with the DS group (128 patients versus 60 patients, p < 0.001).

Mortality associated with AL

A total of 53 patients died within the first 90 days post-operatively, giving a 90-day mortality rate of 11.4%. The 90-day mortality rate was slightly higher in the non-DS group compared with the DS group, although this was not statistically significant (31 (13.5%) versus 22 (8.7%) patients, p = 0.085). Not surprisingly, multiorgan failure and sepsis represented the predominant etiology in early mortality in both patient groups, accounting for 12 (55%) deaths in the DS group and 26 (84%) deaths in the non-DS group. Other reasons for early mortality included pulmonary disease, heart disease, and a single case of cerebral infarction.

At the time of analysis, 133 (45%) patients in the non-DS group and 162 (55%) patients in the DS group were still alive (Fig. 1). Recurrent cancer was the most common cause of delayed mortality and was not found to differ significantly between the 2 groups (27 (11.8%) patients in the non-diverted group, 37 (14.7%) in the diverted group). All causes of death are outlined in Table 4.

Fig. 1
figure 1

Kaplan-Meier plot showing survival in days

Full size image
Table 2 Surgical variables of patients developing an anastomotic leakage after low anterior resection for rectal cancer. TME, total mesorectal excision; PME, partial mesorectal excision
Full size table
Table 3 The diagnosis and treatment of anastomotic leakage in patients with and without diverting stomata. CT, computed tomography; MR, magnetic resonance
Full size table
Table 4 Causes of mortality in patients suffering an anastomotic leakage with or without a diverting stoma. MOF, multi organ failure
Full size table
Table 5 Multivariable regression of factors associated with 90-day mortality. ASA, American Society of Anesthesiology; LAR, low anterior resection; AL, anastomotic leakage; DM, diabetes mellitus
Full size table
Table 6 Complications in patients suffering an anastomotic leakage with or without a diverting stoma. ACS, abdominal compartment syndrome; GI, gastrointestinal
Full size table
Table 7 Multivariable regression of factors associated with a permanent stoma. ASA, American Society of Anesthesiology; UICC, Union for International Cancer Control; LAR, low anterior resection; AL, anastomotic leakage
Full size table

Independent prognostic factors associated with a risk of 90-days mortality were investigated using multivariable Cox regression. Interestingly, the association between diverting stoma and 90-mortality was weak, while age (hazard ratio (HR) 1.07, 95% confidence interval (CI) 1.03–1.11), pre-operative morbidity (ASA) (HR 2.46, 95% CI 1.16–5.20), peri-operative complication as expressed by the need of blood transfusion (HR 3.52, 95% CI 1.56–7.97), and late-occurring leakage (HR 0.16, 95% CI 0.04–0.71) seemed to be strongly associated to the risk of 90-day mortality (Table 5).

Morbidity associated with AL

Patients in the non-DS group suffered more complications due to AL than patients in the DS group. This was true to both surgical (120 (53%) patients in the non-DS group versus 71 (28%) in the DS group, p < 0.001) and non-surgical complication (110 (48%) patients in the non-DS group versus 89 (35%) in the DS group, p = 0.004). Acute infections and wound dehiscence complications were abundant in the non-DS group along with complications related to single or multiorgan failures (respiratory 16%, renal 16%, and circulatory 29%). In contrast, post-operative fistulas (rectovaginal, rectoenteric, rectocutaneous) were more common in the DS group (8% versus 5%). The severity of complications, as measured by the Clavien-Dindo score, was significantly higher in the non-DS group (p < 0.001) (Table 6).

Length of stay

Difference in length of stay between the 2 groups was not statistically significant with a median stay of 27 days (range 1–266) in the non-DS group, compared with 22 (range 4–182) in the DS group. When looking at the subgroup that was discharged alive, the result was almost identical (Table 4).

Complications related to stoma and risk of permanent stoma

The complications related to a primary diverting stoma and a secondary stoma fashioned after the occurrence of an anastomotic leakage were similar, except for a higher incidence of high stoma output in patients with primary diversion (5% in the DS group versus 2 % in the non-DS group). Twenty percent of patients developed a stoma-related complication prior to stoma reversal. Apart from high output, these included parastomal hernias (10%), leakage from the stoma (2.7%), and stoma prolapse (2.5%). A total of 5% of patients underwent surgery for stoma-related complications during follow-up. Stoma reversal was performed in 212 patients, of whom 5 died during the operation (3 from the non-DS group, 2 from the DS group), giving a stoma reversal–related mortality rate of 2.4%.

Independent prognostic factors associated with a risk of a permanent stoma were investigated using multivariable Cox regression. Covariates found to be independently prognostic included age (hazard ratio (HR) 1.04, 95% confidence interval (CI) 1.01–1.08), blood transfusion during primary surgery (HR 2.35, 95% CI 1.16–4.78), an intact anastomosis after LA (HR 0.019, 95% CI 0.009–0.04), and primary diversion at the time of initial surgery (HR 0.50, 95% CI 0.26–0.97) (Table 7).

Discussion

In this nationwide study of the importance of a primary diverting stoma in rectal cancer surgery, we found, in accordance with earlier studies, that primary diversion does indeed mitigate the clinical impact of AL. Although not statistically significant, the rates of early mortality in patients with a diverting stoma were a third lower than in those patients without. Patients with a diverting stoma who suffered AL were also found to have a reduced risk of multiorgan failure, as well as lower rates of both surgical and non-surgical complications. In addition, a diverting stoma was associated with an increased likelihood of successful conservative treatment of AL and a decreased risk of a permanent stoma. This study also identified differences not only in the clinical presentation of AL in patients with or without diverting stomata but also in the diagnostic tools typically used to confirm the diagnosis of this complication. The authors acknowledge the limitations of this study, which due to its retrospective nature may be subject to both selection and confounding bias. These limitations do not affect the primary analysis, but certainly, the comparison between the various surgical techniques must be viewed in light of the selection bias.

Although these data make a strong case for the routine use of diverting stomata in surgery for rectal cancer, it should be noted that this study also identified substantial morbidity associated with such stomata, including a mortality rate of almost 3% during stoma reversal. Published rates of reversal-related mortality vary considerably, with reports ranging from 0.06 to 6.4% [17,18,19]. Although the mortality rate in this study is in keeping with published literature, it is recognized that this may be an overestimation of the risk, given that a reversal of a stoma is likely to be more technically challenging in patients who have suffered an AL after their primary surgery than those who have not. In addition to avoiding potential stoma-related morbidity, the omission of a diverting stoma may also reduce the incidence of low anterior resection syndrome (LARS) by facilitating the immediate use of the anal sphincter and avoiding disuse atrophy, resulting in a better functional outcome [20]. One may also make the economic argument for stoma omission, given that it avoids a second admission and second operation. The strongest argument may however be that, given the rate of AL in all-comers in this national cohort was less than 12%, the routine use of diverting stomata would expose almost 9 in every 10 patients to the potential complications of stoma formation and reversal for no clinical benefit.

Diverting stomata are already used selectively in Denmark. Approximately 59% of patients (14% of those undergoing partial mesorectal excision (PME) and 91% of those undergoing total mesorectal excision (TME)) undergoing restorative surgery for cancer within 15 cm from the anal verge are diverted [21]. The decision whether to divert is often made during surgery, based on objective factors such as the patient’s pre-operative performance and peri-operative factors, such as bleeding. However, it is well-recognized that this decision may also be influenced by subjective factors, with the surgeon’s own preferences often based on recent experiences and tradition or culture at their institution [22]. Peri-operatively, there may be a tendency to focus on the risk of AL at the expense of the bigger picture, a situation further compromised by the limited accuracy of the surgeon’s ability to predict AL [23].

The key issue in this complex decision-making process is to identify which patients are at greatest risk of AL and therefore stand to benefit most from a diverting stoma. Several prognostic factors of AL are known pre-operatively, including age, neoadjuvant treatment, male gender, and comorbidities and, to a certain extent, how low the anastomosis is likely to be. These factors could be used to make an evidence-based decision on whether a diverting stoma is indicated pre-operatively, a decision preferably made by a multidisciplinary team (MDT), with the use of a specific decision-making tool [24,25,26]. The use of such a selective strategy has been shown to be both feasible and effective, with a reduction in stoma formation from 90 to 8%, without increasing the rates of re-operation, re-admission, or peri-operative mortality [25]. Once made, this MDT decision should only be overturned by the occurrence of predefined peri-operative factors such as more than a certain amount of bleeding or a lower anastomosis than anticipated. This objective decision-making process should focus the use of diverting stomata to those patients likely to benefit the most.