Impact of Anastomotic Leakage After Colorectal Cancer Surgery on Quality of Life: A Systematic Review.

This systematic review was conducted according to the latest edition of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. 13 The study protocol was developed a priori and registered at PROSPERO (ID 411065). The primary outcomes were QoL and health-related QoL (HRQoL). QoL was defined using the World Health Organization definition of “an individuals’ perception of their position in life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards, and concerns.” 14 HRQoL in cancer is often used interchangeably with QoL because there is no consensus on a standardized definition. We have applied the definition of Testa and Simonson 15 on HRQoL as the “physical, psychological and social domains of health, seen as distinct areas that are influenced by a person’s experiences, beliefs, expectations and perceptions.” AL was defined as a combination of clinical signs and symptoms (eg, abdominal pain or tenderness, peritonitis, fever, tachycardia, purulent or fecal discharge from an abdominal drain or the vagina, purulent discharge per anus), biochemical elements (elevated white blood cell count and/or C-reactive protein), and radiological confirmation of an interruption of the anastomosis and/or a perianastomotic collection on CT. 6 , 16 The literature search was performed on March 13, 2023, and repeated on August 14, 2023. PubMed, Cochrane Library, and Embase were searched with the use of MeSH-, Emtree-, and free terms, including “colorectal neoplasms,” “(adeno)carcinoma,” “colorectal surgery,” “anastomotic leak,” “complications,” “quality of life (QoL),” and “health-related quality of life (HRQoL)” (see Supplemental S1 at http://links.lww.com/DCR/C410 ). Reference lists of relevant publications were cross-checked to identify additional studies. This hand-search method was continued until no further relevant studies were identified. All English or Dutch articles published in peer-reviewed index journals reporting on QoL in patients older than 18 years with AL after oncological colorectal resections were considered eligible for inclusion. Analysis of QoL after AL had to be identified as a predetermined aim in the Methods section of the study to be eligible for inclusion. Trials were included irrespective of blinding. Systematic reviews and secondary sources such as letters to the editor, technical descriptions, conference proceedings, and commentaries were excluded. Articles reporting on fewer than 10 patients or solely reporting on outcomes after colorectal resections for benign indications were excluded. Since the first systematic review on the definition of AL was published in the year 2001, all articles published before January 1, 2000, were excluded. 8 Furthermore, articles were excluded when no validated HRQoL instrument had been applied. All search results were imported into an online tool designed for systematic reviews (Rayyan). 17 After the removal of duplicates, articles were screened for eligibility by 2 independent researchers (A.H.C.G. and D.J.I.H.) according to the predefined criteria. First, articles were screened on the basis of titles and abstracts. Definitive article inclusion followed if the eligibility criteria were met after full-text screening by both reviewers. Disagreements were resolved through discussion. All references were stored in the Endnote Reference Management Tool (version 20.4, Clarivate, Chandler, United States). Two independent researchers (A.H.C.G. and D.J.I.H.) performed a qualitative analysis and extracted data from the main text, tables, and figures using a predefined and standardized data extraction table. Extracted data included first author, year of publication, country, study design, study period, inclusion and exclusion criteria, aim of the study, number of patients, general patient characteristics, indication for surgery, surgical procedures performed, the applied (validated) QoL questionnaires, time of assessment, and secondary outcomes. Furthermore, definitions, time frame, and criteria for diagnosis of AL were collected. Data acquired via the outlined search strategy were summarized in tables. Findings were described using a narrative approach (ie, primarily words and text were used to summarize and explain the findings). Because of the heterogeneity among included studies in terms of the definition of AL and questionnaires used to assess QoL, pooling in a meta-analysis was impossible. To ascertain the validity of the included studies, the risk of bias in each study was assessed by 2 reviewers (A.H.C.G. and D.J.I.H.) with a revised Risk of Bias in Non-randomized Studies of Interventions tool to assess the risk of bias in nonrandomized studies. 18 All types of bias were evaluated for every study and judged to be “low risk,” “moderate risk,” or “high risk.” Possible confounding domains were a priori defined as active smoking, malnutrition, male sex, BMI, comorbidities or higher ASA classification, emergency surgery, and longer operative time.

The electronic literature search generated 1323 articles and 980 unique articles after removing duplicates. Of these, 866 were excluded after title and abstract screening. A full-text screening of the resulting 114 articles was performed and another 101 were excluded. Cross-reference checking generated 1 additional article, and 1 more additional publication was identified after repeating the search before submission. Ultimately, 13 articles were included in the analysis (Fig. 1 A). Study selection. A, PRISMA flow chart outlining study selection strategy. B, Adapted ROBINS-I tool Risk of Bias from included studies. CRC = colorectal cancer; PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-Analyses; QoL = quality of life; ROBINS-I = Risk of Bias in Non-randomized Studies of Interventions. All 13 articles included were cohort or case-matched studies and comprised 4618 individual patients, with study sample sizes ranging from 32 to 1207 patients (Table 1 ). 19 – 26 , 28 – 31 Four studies reported on colorectal resections, 19 , 21 , 24 , 29 and all other studies focused on rectal resections. All studies included patients diagnosed with CRC, with only 2 studies also including patients with benign indications for colorectal resections (eg, diverticulitis, IBD). Because benign indications were presented separately, the outcomes of these patients were excluded from this review. 24 , 29 The final population consisted of 4618 patients, of whom 2946 (64%) were men and 1672 (36%) women, with a mean age of 61.9 years. Among these patients, 527 (11%) developed AL and 4091 (89%) recovered without clinical, radiological, or biochemical signs of AL (Table 2 ). 19 – 26 , 28 – 31 The median time of follow-up was 4.3 years (4.8 months–14.4 years). Additional study information on the perioperative care of patients in each study is provided in Supplemental S2 at http://links.lww.com/DCR/C410 . Characteristics of included studies AL = anastomotic leakage; CRC = colorectal cancer; HRQoL = health-related quality of life; NA = not applicable; TME = total mesorectal excision. Patient characteristics AL = anastomotic leakage; APR = abdominoperineal resection; IQR = interquartile range; ISR = intersphincteric rectal resection; LAR = low anterior resection; NA = not applicable; PME = partial mesorectal excision; TME = total mesorectal excision. AL rates do not reflect the incidence of AL due to the study design. Group A: no complications, group B: surgical complications, group C: nonsurgical complications, and group D: surgical and nonsurgical complications. Benign indications for resection were irresectable polyps or diverticulitis. The relevant categories from the Risk of Bias in Non-randomized Studies of Interventions tool were used to assess the risk of bias (Fig. 1 B). We reported a serious risk of bias in 9 studies, primarily attributed to the nonrandomized design of these studies 21 , 22 , 24 – 26 , 28 – 31 and a moderate risk of bias in the other 4 studies. 19 , 20 , 23 , 27 All details on AL and specific characteristics reported by each study are summarized in Supplemental S3 at http://links.lww.com/DCR/C410 . The reported definitions and diagnostic modalities for AL varied widely among the studies reviewed. Four studies (33%) did not report any specific definition for AL. 21 , 28 , 30 , 32 Furthermore, none of the studies applied the same definition for AL. The severity of AL was assessed using various classifications across the included studies. Four studies applied the International Study Group of Rectal Cancer classification, 19 , 26 , 27 , 30 whereas 2 studies used the Clavien-Dindo classification. 21 , 25 Four studies divided AL cases into symptomatic and asymptomatic or clinical and subclinical manifestations. 22 , 24 , 28 , 29 The other studies did not provide a specific classification or grading of severity of AL. The time frame in which AL was suspected or diagnosed was reported in 4 articles, with the latest reporting time being 6 months after surgery. 19 , 22 , 24 , 27 One study reported biochemical characteristics that might indicate surgical complications. 24 Eight studies (67%) used CT with or without contrast to confirm the diagnosis of AL. 20 , 22 – 25 , 27 – 29 Four studies reported performing radiological assessment and subsequent AL assessment only when clinical symptoms occurred. 19 , 20 , 25 , 27 Three other studies additionally performed routine scanning for AL before ileostomy closure (range, 6 weeks–3 months after surgery). 22 , 24 , 28 The type of reinterventions was specified in 10 studies 19 , 21 , 22 , 24 – 30 and ranged from antibiotic treatment to reoperation (laparotomy) with take down of anastomosis and end-colostomy construction. A total of 10 validated QoL questionnaires were administered at different time points within the studies. Four validated instruments were administered across the majority of studies (see Supplemental S4 at http://links.lww.com/DCR/C410 : The European Organization for Research and Treatment of Cancer QoL Questionnaire [EORTC QLQ]-C30 [Core] and -CR29 [CRC-specific], the 36-item Short-Form survey [SF-36] encompassing both a physical component summary [PCS] and a mental component summary [MCS], and the Fecal Incontinence QoL [FIQL] questionnaire). Six additional questionnaires were used in only 1 study (Supplemental S5 at http://links.lww.com/DCR/C410 ). These included the Cleveland Global QoL (CGQL), the EORTC IN-PATSAT32 questionnaire for assessing cancer care satisfaction, the GI QoL Index (GIQLI) addressing digestive disorders with both physical and emotional components, the EuroQoL Visual Analog Scale for patient self-rated health, the 12-item Short-Form survey evaluating health impact on daily life, and the Rotterdam Symptoms Check List (RSCL) questionnaire, which generally evaluates HRQoL. QoL was evaluated at different time points. Almost all studies compared QoL scores between patients with AL and patients without AL at specific time points but not always relative to baseline assessment (Fig. 2 ; Table 3 ). Detailed overview of questionnaires and impaired domains if applicable AL = anastomotic leakage; CGQL = Cleveland Global QoL; EORTC = European Organization for Research and Treatment of Cancer; EQ-VAS = the EuroQoL Visual Analog Scale; FIQL = Fecal Incontinence QoL; FISI = Fecal Incontinence Severity Index; GIQLI = Gastrointestinal QoL Index; IN-PATSAT32 = inpatient satisfaction; IQR = interquartile range; MCS = mental component score; NA = not applicable; PCS = physical component score; QLQ-CR29 = QLQ colorectal cancer–specific; QLQ-CR30 = QoL Questionnaire cancer-specific; QoL = quality of life; RSCL = Rotterdam Symptoms Check List; SF-12 = 12-item Short-Form survey; SF-36 = 36-item Short-Form survey. Multivariable linear and logistic regression. Not clinically relevant. Wilcoxon rank sum tests. Mann-Whitney U test. Kruskal–Wallis ANOVA. Multivariate analysis. Univariate and multivariate logistic regression. Wald’s tests using linear mixed-effects model and univariable Poisson regression analysis; Paired samples t test. Statistical analysis unclear. Schematic overview of results of QOL questionnaires based on different time points. *Not statistically significant but on an individual level AL was a determinant of a clinically relevant decrease; **at least 6 mo after the restoration of bowel continuity; ***subgroup: in patients with no stoma closure, there was a difference; ****modified version of the questionnaire. AL = anastomotic leakage; CGQL = Cleveland Global QoL questionnaire; EORTC = European Organization for Research and Treatment of Cancer; EQ-VAS = the EuroQoL Visual Analog Scale; FIQL = Fecal Incontinence; GIQLI = GI QoL Index; IQR = interquartile range; IN-PATSAT32 = inpatient satisfaction; MCS = mental component summary; PCS = physical component summary; QLQ-CR29 = QoL Questionnaire colorectal cancer–specific; QLQ-CR30 = QoL Questionnaire cancer–specific; QoL= quality of life; SF-12 = 12-item Short-Form survey; SF-36 = 36-item Short-Form survey; RSCL = Rotterdam Symptoms Check List. Based on EORTC QLQ-CR29 and -CR30 scores at 1 and 6 months postoperatively, Di Cristofaro et al identified AL as an independent predictor of lower QoL in multivariate analysis ( p < 0.001 and p = 0.004, respectively). 21 van Kooten et al 23 found that patients who developed AL reported a decrease in RSCL global health status and activity level within the first 3 months compared to preoperative scores, with some improvement at 6 months. In contrast, Marinatou et al 25 did not document any improvement based on GIQLI and EORTC QLQ-C30 questionnaires administered at 3 and 6 months. Instead, a significant decline in physical functioning and global and overall QoL scores was documented among patients with AL relative to patients without AL at 6 months. 25 Additional results from EORTC QLQ-C29 demonstrated significantly worse scores with respect to pain, stoma, and perianal skin-related complaints at 3 and 6 months in patients with AL. Also, SF-36 scores demonstrated significantly worse function among patients with AL versus patients without AL at 6 months, especially along emotional and social domains, which was not seen at 3 months. Impairments in functional outcomes based on SF-36 scores were also reported by Ashburn et al 20 among patients with AL compared to patients without AL after proctectomy. Arron et al 19 demonstrated that the decrease in EORTC QLQ-C30 scores observed among patients with AL at 6 months relative to patients without AL did not meet the threshold for clinical relevance, and AL status was not associated with the observed decrease. Among patients with a clinically relevant decrease in their 6-month scores relative to baseline, AL was an independent predictor of this decrease based on multivariate regression analysis. 19 Three studies reported QoL at 1 year after CRC resection. 20 , 23 , 25 van Kooten et al 23 demonstrated that HRQoL scores returned to baseline preoperative levels among rectal cancer patients with and without complications, whereas Marinatou et al 25 demonstrated persistently significant differences between AL and non-AL groups for perianal skin soreness and worse overall EORTC QLQ-C30, global GIQLI, and SF-36 scores. Ashburn et al 20 also documented significantly worse SF-36 scores in both the PCS and MCS domains at 1 year postoperatively in patients with AL compared to those without AL after restorative proctectomy. Mongin et al 28 evaluated QoL in patients undergoing restoration of bowel continuity at least 6 months before the assessment. Given that the median time of QoL assessment was 33 versus 30 months in patients with versus without AL, results were interpreted as representing longer-term QoL. No difference in SF-36 scores were found between the 2 groups. However, “blood and mucus in stool” scores of the EORTC QLQ-CR29 indicated significantly worse function in patients with AL, as did depression/self-perception FIQL scores. Ashburn et al noted that although the SF-36 PCS scores did not show significant differences beyond 12 months postoperatively (median, 3.2 years), MCS scores were still significantly worse in patients with AL after proctectomy. 20 Westerduin et al 31 identified 5 domains of the EORTC QLQ-30 and 2 functional and 5 emotional domains of the -CR29, which were significantly better beyond 1 year postoperatively in patients with AL compared to patients without AL. Hain et al 22 reported additional impaired -CR29 outcomes (more blood and mucus in stool, frequent bowel movements, and frequent urination per day) in patients with symptomatic AL compared to the combined groups of patients with no or asymptomatic AL. Di Re et al 29 also demonstrated lower mean EuroQoL visual analog scale scores among patients with AL versus patients without AL in a matched cohort at 1 year after surgery (range up to 5 years), although the difference did not reach statistical significance. At 18 and 24 months postoperatively, van Kooten et al found no differences in RSCL scores between patients with AL and patients without AL. 23 Arron et al 19 found no difference in overall HRQoL scores between patients with AL and patients without AL at 2 years relative to baseline EORTC QLQ-C30 scores. 19 Similar results were described when SF-36 scores were compared >2 years after surgery between patients with AL (median of 4 years postoperatively) and without AL (median of 6.4 years postoperatively). 26 Riss et al 30 described no significant difference in mental and physical QoL scores measured by the 12-item Short-Form survey at a median follow-up time of 106.8 months after rectal surgery (range of 32.5–170.4 months) comparing AL to no AL patients from a matched cohort. Two additional studies evaluated the longer-term impact of AL on QoL. 24 , 27 Lim et al 24 assessed the EORTC QLQ-C30 in patients without AL, with subclinical leaks, and with clinical leaks with and without ileostomy closure (overall median follow-up time of 26 months; interquartile range, 19–37 months). They found worse scores in patients with clinical leaks in whom ileostomy reversal was not possible. Miura et al 27 did not find significant differences in overall modified FIQL scores when comparing patients with AL and patients without AL at a median time of 63 months after low rectal cancer surgery. van Kooten et al 23 conducted a supplementary analysis on EORTC QLQ-C30 and -CR29 outcomes 14 years postsurgery but found statistically significant differences between patients with AL and patients without AL. 23 Some additional outcomes that might influence QoL are summarized in Supplemental S6 at http://links.lww.com/DCR/C410 . Neoadjuvant treatments were described by 9 studies, 19 , 20 , 22 , 24 , 25 , 28 – 31 1 of which found chemoradiation to be significantly different between patients with AL and patients without AL and 1 of which found radiotherapy to be significantly different. 25 , 31 Diverting stoma rates between patients with AL and patients without AL were compared in 6 studies, 19 , 20 , 25 , 29 – 31 of which 2 found significant differences (more diverting in the AL group). 20 , 30 Stoma status during follow-up was clearly described by 2 studies, 19 , 25 which showed significant differences between patients with AL and patients without AL within the first year after surgery. Two additional studies described permanent stoma rates related to AL. 24 , 27 Di Re et al 29 additionally analyzed oncological outcomes as disease-free survival at 1, 3, and 5 years after surgery, and they found that rates were not significantly different between patients with AL and patients without AL. Overall, there was a lack of comparing types of (re-)interventions.

This systematic review appraised and synthesized the evidence on the impact of AL on QoL after oncological colorectal resections. In total, the studies comprised 4618 individual patients, with an overall incidence of AL of 12.4% (N = 572). QoL was assessed using 10 validated questionnaires administered at postoperative time points ranging from 1 month to 14 years. Overall, AL was found to negatively impact QoL at 6 and even 12 months postoperatively, with variable degrees of subsequent improvement. The heterogeneity in questionnaires administered and variable times of assessment hindered our data analysis and may account for some conflicting results across studies. In a comprehensive systematic review of research studies on QoL and HRQoL, Haraldstad et al 33 concluded that the majority experienced conceptual and methodological challenges with no clear consensus on how QoL should be measured. The use of various assessment tools and questionnaires in different studies hinders meaningful comparisons between similar study populations. 34 Adoption of the standard set of outcomes for CRC proposed by the International Consortium for Health Outcomes Measurements (ICHOM) may avoid some of these issues. 35 In this consortium, it was recommended to use the EORTC QLQ-C30 tool to capture overall QoL and the -CR29 to capture CRC-specific outcomes. The optimal time for QoL assessment was also addressed, with recommendations to administer questionnaires at baseline (before surgery), 6 months after surgery, and then annually for up to 10 years. Our research team suggests following the ICHOM recommendations. Other patient and treatment variables such as the ASA score, BMI, anastomotic height, and adjuvant radiotherapy may impact QoL after CRC resections. 36 – 39 Only 2 of the 13 included studies performed multivariate logistic regression analyses to investigate whether differences in QoL scores observed between AL and non-AL groups were due to the leak or driven by other factors such as neoadjuvant treatment, surgical procedure, or reintervention. 19 , 22 Ideally, all studies should have performed such an analysis to verify whether AL is an independent factor that influences QoL. Besides, not all studies compare outcomes relative to baseline function, which weakens the interpretation of the functional scores at subsequent postoperative time points. As a result, it was difficult to draw valid conclusions comparing the included studies. The observed decline in QoL scores reported among patients with AL in the first 6 and even 12 months may be due to several reasons. AL delays recovery and results in additional postoperative complications, higher rates of reintervention, and increased mortality within the first 30 days after surgery. 4 , 40 This often prolongs the length of hospital stay and adversely impacts mobility and the ability of patients to care for themselves. 41 – 43 Furthermore, some patients require stoma construction, which impairs role and social functioning scores. 44 In the current study, there was a lack of correlation between stoma status and QoL outcomes. One study excluded patients with a stoma, 28 whereas other studies did include them but did not draw strong conclusions on any association between stoma formation and QoL scores. AL has also been associated with higher rates of local recurrence and distant metastases in patients with CRC. 45 , 46 Although smaller cohort studies have not found the same association between AL and colon cancer outcomes, the fear of (local) recurrence, as well as additional treatments required to mitigate the higher risk of recurrence, may further negatively impact QoL. 47 , 48 Moreover, AL has been shown to be an independent risk factor for worse defecatory function (low anterior resection syndrome) and sexual function after CRC resections. 49 – 52 Although these functional outcomes were not specifically assessed in the current study, it is crucial to consider their impact on overall QoL. 53 To our knowledge, this is the first systematic review of the effects of AL on QoL in patients undergoing oncological colon and rectal resections. This study has limitations. First, a high heterogeneity in AL reporting was found in the included articles. It was often unclear what type of intervention and reoperation was performed to manage leaks. Because these elements are important when comparing outcomes, standardizing the reporting and management of leaks would be helpful. Subsequently, some studies only included patients with rectal cancer, whereas others included all types of colorectal surgeries. Second, a wide range of QoL questionnaires and time frames for assessment was used across the different studies. Although only studies using validated instruments were included, the heterogeneity of questionnaires created challenges when comparing outcomes across studies and interpreting results. The use of patient-centered methods, such as patient-reported outcome measures, may be even more informative to gain more insight into overall changes. 54 Due to the heterogeneity of the included studies, comparisons across studies are limited and a meta-analysis was not possible to perform. Finally, all included studies demonstrated a moderate to serious risk of bias, which results in a low level of evidence, and caution is warranted by the presented findings.

This systematic review demonstrated that QoL of patients with CRC may be compromised after AL up to 1 year, but assessment and reporting of QoL needs to be standardized to draw clear conclusions. In addition to exploring strategies for preventing and effectively managing AL, it is crucial to investigate long-term sequelae on patients’ QoL in future research. We recommend incorporating a standardized QoL assessment for patients with CRC who have experienced AL and integrating this outcome measure into a core outcome set for research focused on AL in the colorectal field. Continuous assessment and monitoring of QoL in patients undergoing CRC resection is essential to better support patients throughout their recovery. We emphasize the relevance of uniform reporting of AL outcomes to facilitate comparisons of results in future research. To reach this goal, we advise following the proposed questionnaires and time point as described by the CRC ICHOM working group, 35 and work on a standardized reporting framework for AL-related research within the Consensus on Reporting and Defining Colorectal Anastomotic Leaks project.

The authors express their gratitude to Gregor Franssen, who ensured the appropriate search strategy as a professional clinical librarian at Maastricht University. They also thank the CoReAL Collaborative that has worked to highlight the QoL knowledge gap. CoReAL Collaborative: Michel Adamina, Alberto Arezzo Mahdi Al-Taher, Tan Arulampalam, Saba Balvardi, Himani Bhatt, Marta Botti, Marylise Boutros, David A. Clark, Freek Daams, Jennifer S. Davids, Anse De Sadeleer, Abe Fingerhut, Nader Francis, Zoe Garoufalia, Roel Hompes, Neil H. Hyman, Mehraneh D. Jafari, John T. Jenkins, Audrey C.H.M. Jongen, Deborah S. Keller, Samuel H. Lai, Jérémie H. Lefevre, Bibi Martens, Justin A. Maykel, Jeongyoon Moon, Nariaki Okomoto, Ian Paquette, Gianluca Pellino, Sherief F. Shawki, Benjamin D. Shogan, Chelliah Selvasekar, Simon N.G. Siu Man, Jasper Stijns, Patricia Tejedor, William Tzu-Liang Chen, Christiaan van Der Leij, Steven D. Wexner, Elizabeth Wick, and Marina Yiasemidou.