Abstract
Background
Preoperative inflammation might cause and also be a marker for anastomotic leakage after anterior resection for rectal cancer. Available biomarker indices such as the modified Glasgow Prognostic Score (mGPS) or the C-reactive protein-to-albumin ratio (CAR) may be clinically useful for leakage assessment.
Methods
Patients who underwent anterior resection for rectal cancer during 2014–2018 from a multicentre retrospective cohort were included. Data from the Swedish Colorectal Cancer registry and chart review at each hospital were collected. In a subset of patients, preoperative laboratory assessments were available, constituting the exposures mGPS and CAR. Anastomotic leakage within 12 months was the outcome. Causally oriented analyses were conducted with adjustment for confounding, as well as predictive models.
Results
A total of 418 patients were eligible for analysis. Most patients had mGPS = 0 (84.7%), while mGPS = 1 (10.8%) and mGPS = 2 (4.5%) were less common. mGPS = 2 (OR: 4.11; 95% CI: 1.69–10.03) seemed to confer anastomotic leakage, while this was not seen for mGPS = 1 (OR 1.09; 95% CI: 0.53–2.25). A cut off point of CAR > 0.36 might be indicative of leakage (OR 2.25; 95% CI: 1.21–4.19). Predictive modelling using mGPS rendered an area-under-the-curve of 0.73 (95% CI: 0.67–0.79) at most.
Discussion
Preoperative inflammation seems to be involved in the development of anastomotic leakage after anterior resection for cancer. Inclusion into prediction models did not result in accurate leakage prediction, but high degrees of systemic inflammation might still be important in clinical decision-making.
Supplementary Information
The online version contains supplementary material available at 10.1007/s00384-023-04496-5.
Keywords: Anastomotic leak, Colorectal, Inflammation
Introduction
Anastomotic leakage after anterior resection for rectal cancer affects 10–20% of patients [1, 2], resulting in bowel dysfunction [3], permanent stoma and increased mortality [4]. While the pathophysiology of anastomotic leakage remains to be fully elucidated, there are established risk factors including male sex, low anastomosis, radiotherapy, advanced comorbidity [5] and smoking [6].
Previous research suggests associations with preoperative systemic inflammation, as measured by e.g. C-reactive protein (CRP) and albumin, and anastomotic leakage after gastrointestinal surgery [7–9]. However, few studies have been conducted on preoperative inflammation and the risk of anastomotic leakage in anterior resection for rectal cancer [10]. A well-known index of inflammation, with prognostic value for survival, is the modified Glasgow Prognostic Score (mGPS) [11]; another measure, increasingly used in research, is the CRP-to-albumin ratio (CAR) [8, 9, 12, 13]. As both measures are based on routine blood samples, they are easy to use and readily available markers of preoperative inflammation. If any such a measure can be validated as being predictive of anastomotic leakage, this could aid patients and surgeons in clinical decision-making, potentially reducing the use of temporary stomas in low-risk patients and avoiding an anastomosis altogether in high-risk patients.
In this multicentre retrospective study, our primary aim was to investigate whether preoperative mGPS or CAR are useful predictors of anastomotic leakage after anterior resection for rectal cancer. Our secondary purpose was to incorporate mGPS and CAR into predictive models for leakage and evaluate their accuracy.
Method
Study design
This is a retrospective multicentre cohort with 11 participating Swedish centres. Patients who underwent anterior resection from 2014 to 2018 were identified at each hospital. Exclusion criteria comprised other indications than rectal cancer and missing preoperative CRP and albumin. Data such as previous medical and surgical history, standing medication and surgical techniques were collected at each centre through chart review. Postoperative data comprising short- and long-term outcomes, including anastomotic leakage up to 12 months after index surgery, were also registered. The resulting information was entered in the RedCap system (Vanderbilt University, Tennessee, USA), which is a web-based application to capture data for clinical research.
Further data were retrieved through the Swedish Colorectal Cancer Registry (SCRCR). The SCRCR is a national registry with data from all hospitals operating patients with colorectal cancer in Sweden. It has been validated with excellent results, reaching a completeness of inclusion in 99% [14]. The RedCap database and SCRCR data were linked to ensure all patients registered in RedCap had surgery due to rectal cancer. Furthermore, data such as demographics, preoperative variables, tumour stage and height, neoadjuvant treatment and some perioperative data were added from the SCRCR.
Exposure
The primary exposure was the mGPS, an index based on combining albumin plasma levels and elevated CRP levels. The prognostic score has three categories: 0, 1 or 2. Patients with CRP ≤ 10 mg/l and any albumin level are allocated a score of 0, while patients with CRP > 10 mg/l and albumin levels ≥ 35 g/l are scored as 1; patients with CRP > 10 mg/l and albumin < 35 g/l are scored as 2 [11]. The secondary exposure was the CAR. Both measures were calculated using preoperative blood samples ascertained by chart review, where these samples were taken within 30 days prior to surgery.
Outcome
The primary outcome was anastomotic leakage within 12 months after anterior resection. The definition of anastomotic leakage was in accordance with the international consensus definition from the International Study Group of Rectal Cancer [15]. Anastomotic leakage was thus defined as a communication between the intra- and extra-luminal compartments owing to a defect of the integrity of the intestinal wall at the level of anastomosis or suture/staple line, as well as an abscess in the proximity of the anastomosis. Anastomotic leakage was further divided into grades A, B and C. Grade A was defined as leakage requiring no active therapeutic intervention, corresponding to a radiological leakage with minimal clinical signs. Grade B was defined as leakage requiring an active therapeutic intervention, barring a laparotomy. Grade C was defined as leakage requiring laparotomy; in this study, relaparoscopy also rendered a grade C. Anastomotic leakage may also be classified into early or late, consistent with a diagnosis within or after 30 days, respectively [2]. In the SCRCR there is no formal definition of anastomotic leakage and it is only mandatory to register leakage within the first 30 days after primary surgery, leading to a potential underreporting of anastomotic leakage [16]. Using chart review instead in the current study, detailed data were collected on anastomotic leakage and subsequent management within a year from the primary surgery.
Statistical analysis
Continuous variables were reported using the median along with interquartile range (IQR), while categorical variables were reported using numbers and percentages. Fisher´s exact test was applied to test for associations in selected comparisons of categorical variables.
A directed acyclic graph [17] is presented in Fig. 1. The graph shows the proposed causes and effects that are involved in the development of anastomotic leakage. The following covariates were selected in order to have minimally sufficient adjustment set: age (years; continuous), body mass index (BMI; kg/m2; continuous), sex (male or female), clinical T stage (T1-2, T3, T4 or TX), clinical N stage (N0, N1-2 or NX), clinical M stage (M0 or M1), Charlson Comorbidity Index (CCI; 0, 1, 2 or ≥ 3), current smoking (no or yes), neoadjuvant therapy (none, radiotherapy or chemoradiotherapy).
Fig. 1.
Directed acyclic graph picturing the assumed relationship between different variables potentially involved in the development of anastomotic leakage. Modified Glasgow Prognostic Score (mGPS) indicates exposure and anastomotic leakage is outcome; exposure is replaced with C-reactive protein-to-albumin ratio in the secondary analysis. A minimal adjustment set to derive a total effect on the outcome from exposure consisted of variables such as age, body mass index (BMI), sex, clinical tumour stage (cTNM), Charlson Comorbidity Index (CCI), smoking, and neoadjuvant therapy. TME/PME = total/partial mesorectal excision; ASA = American Society of Anesthesiologists’ fitness grade
To estimate the effects of preoperative mGPS > 0 on anastomotic leakage within 12 months we used targeted maximum likelihood estimation [18, 19]. This is a two-step procedure designed to obtain an optimal bias-variance trade-off when estimating the effect of an exposure on outcome. The initial outcome and propensity score models were fitted using lasso regression [20], and the set of covariates described above was included in the models as potential confounders. The results are expressed as odds ratios (ORs), together with their 95% confidence intervals (CIs), calculated with influence-curve based standard errors [18].
Additionally, logistic regression analysis was performed to assess independent associations of mGPS with anastomotic leakage when including other relevant predictors in a multivariable prediction model. In addition to the covariates outlined above, the prediction model included the American Society of Anesthesiologists’ (ASA) fitness grade, extent of mesorectal excision (total [TME] or partial [PME]), defunctioning stoma (no or yes), previous surgery (no or yes), hospital volume and surgical technique (open or minimally invasive [conventional laparoscopy or robotically assisted laparoscopy]). A reduced model including only the subset of predictors selected by lasso regression was also fitted.
Analogous analyses as described for mGPS were also conducted with a categorized CAR as exposure. CAR was categorized by estimating the optimal location and number of cut-off points [18] for a continuous variable included in a pre-specified logistic regression model [21, 22]. The cut-off points were selected as those maximizing the area under the receiver operation characteristic curve. The pre-specified model used here included all predictors.
Multivariate imputation by chained equations was used to handle missing values [23] and the estimates from 10 imputed data sets were pooled according to Rubin’s rules [24]. All analyses were performed using R 4.2.1 statistical software [25]. The R-packages CatPredi [26], tmle [27], glmnet [28] and mice [29] were used for categorization, analysis and imputation.
Results
Patients
The original cohort included 1,126 eligible patients, while 708 patients were excluded due to unavailable data on preoperative CRP and/or albumin. After exclusion, 418 patients remained for analysis and a study flowchart is presented in Fig. 2.
Fig. 2.
Study flowchart. SCRCR = Swedish Colorectal Cancer Registry
Clinical and demographic variables are presented in Table 1. Most patients had mGPS = 0 (N: 354; 84.7%), while mGPS = 1 (N: 45; 10.8%) and mGPS = 2 (N: 19; 4.5%) were progressively rarer. While higher age, male sex, advanced comorbidity and current smoking were more common in the latter groups, BMI was similar across groups. Notably, patients with higher mGPS scores had more advanced tumours and subsequently received neoadjuvant therapy to a higher degree. Open surgery was also more common in these groups, while there were no apparent differences in type of mesorectal excision employed nor the use of defunctioning stoma. Duration of operation was longer and intraoperative bleeding was higher in the group with the highest mGPS score.
Table 1.
Clinical and demographic data for n = 418 patients (patients with missing on preoperative C-reactive protein and preoperative albumin are excluded; patients with missing covariate values are included), stratified by the modified Glasgow Prognostic Score (mGPS)
| mGPS = 0 (N = 354) |
mGPS = 1 (N = 45) |
mGPS = 2 (N = 19) |
Overall (N = 418) |
|
|---|---|---|---|---|
| Age (years) | ||||
| Median (IQR) | 65.8 (58–72) | 70.0 (62–75) | 70.0 (63–72.5) | 66.0 (59–72) |
| Sex | ||||
| Female | 141 (39.8%) | 17 (37.8%) | 5 (26.3%) | 163 (39.0%) |
| Male | 213 (60.2%) | 28 (62.2%) | 14 (73.7%) | 255 (61.0%) |
| Charlson Comorbidity Index | ||||
| 0 | 253 (71.5%) | 28 (62.2%) | 11 (57.9%) | 292 (69.9%) |
| 1 | 57 (16.1%) | 9 (20.0%) | 5 (26.3%) | 71 (17.0%) |
| 2 | 28 (7.9%) | 5 (11.1%) | 1 (5.3%) | 34 (8.1%) |
| ≥ 3 | 16 (4.5%) | 3 (6.7%) | 2 (10.5%) | 21 (5.0%) |
| ASA fitness grade | ||||
| I | 83 (23.4%) | 6 (13.3%) | 1 (5.3%) | 90 (21.5%) |
| II | 220 (62.1%) | 33 (73.3%) | 10 (52.6%) | 263 (62.9%) |
| III–V | 48 (13.6%) | 5 (11.1%) | 8 (42.1%) | 61 (14.6%) |
| Missing | 3 (0.8%) | 1 (2.2%) | 0 (0%) | 4 (1.0%) |
| Smoker | ||||
| No | 314 (88.7%) | 37 (82.2%) | 14 (73.7%) | 365 (87.3%) |
| Yes | 22 (6.2%) | 4 (8.9%) | 2 (10.5%) | 28 (6.7%) |
| Missing | 18 (5.1%) | 4 (8.9%) | 3 (15.8%) | 25 (6.0%) |
| Body mass index (kg/m2) | ||||
| Median (IQR) | 25 (23–28) | 25 (24–27) | 26 (24–28) | 25 (23–28) |
| Missing | 6 (2%) | 0 (0%) | 0 (0%) | 6 (1%) |
| Tumour height (cm) | ||||
| Median (IQR) | 10 (9–12) | 10 (9–12) | 12 (10–13) | 10 (9–12) |
| Missing | 0 (0%) | 1 (2%) | 0 (0%) | 1 (0%) |
| Clinical T stage | ||||
| cT1-2 | 141 (39.8%) | 15 (33.3%) | 2 (10.5%) | 158 (37.8%) |
| cT3 | 175 (49.4%) | 22 (48.9%) | 13 (68.4%) | 210 (50.2%) |
| cT4 | 27 (7.6%) | 7 (15.6%) | 4 (21.1%) | 38 (9.1%) |
| cTX | 11 (3.1%) | 1 (2.2%) | 0 (0%) | 12 (2.9%) |
| Clinical N stage | ||||
| cN0 | 180 (50.8%) | 16 (35.6%) | 6 (31.6%) | 202 (48.3%) |
| cN1-2 | 162 (45.8%) | 29 (64.4%) | 12 (63.2%) | 203 (48.6%) |
| cNX | 12 (3.4%) | 0 (0%) | 1 (5.3%) | 13 (3.1%) |
| Clinical M stage | ||||
| cM0 | 335 (94.6%) | 43 (95.6%) | 18 (94.7%) | 396 (94.7%) |
| cM1 | 19 (5.4%) | 2 (4.4%) | 1 (5.3%) | 22 (5.3%) |
| Neoadjuvant therapy | ||||
| No | 176 (49.7%) | 18 (40.0%) | 6 (31.6%) | 200 (47.8%) |
| Radiotherapy | 134 (37.9%) | 18 (40.0%) | 10 (52.6%) | 162 (38.8%) |
| Radiochemotherapy | 44 (12.4%) | 9 (20.0%) | 3 (15.8%) | 56 (13.4%) |
| Surgical technique | ||||
| Open | 125 (35.3%) | 17 (37.8%) | 10 (52.6%) | 152 (36.4%) |
| Minimally invasive | 229 (64.7%) | 28 (62.2%) | 9 (47.4%) | 266 (63.6%) |
| Type of mesorectal excision | ||||
| Total | 271 (76.6%) | 31 (68.9%) | 15 (78.9%) | 317 (75.8%) |
| Partial | 83 (23.4%) | 14 (31.1%) | 4 (21.1%) | 101 (24.2%) |
| Defunctioning stoma | ||||
| No | 68 (19.2%) | 9 (20.0%) | 4 (21.1%) | 81 (19.4%) |
| Yes | 286 (80.8%) | 36 (80.0%) | 15 (78.9%) | 337 (80.6%) |
| Operation time | ||||
| Median (IQR) | 290 (213–377) | 260 (195–359) | 326 (219–397) | 289 (210–377) |
| Missing | 2 (1%) | 0 (0%) | 0 (0%) | 2 (1%) |
| Perioperative bleeding (ml) | ||||
| Median (IQR) | 100 (50–313) | 150 (50–350) | 300 (100–950) | 100 (50–350) |
| Missing | 10 (3%) | 0 (0%) | 0 (0%) | 10 (2%) |
IQR interquartile range, ASA American Society of Anesthesiologists
Modified Glasgow Prognostic Score and anastomotic leakage
In the analysed cohort, 74 patients (17.7%) had anastomotic leakage within 12 months (55 leaks occurred within 30 days, while 19 patients had late leakage). Of these, 10 were grade A, 42 were grade B, and 22 were grade C leaks. Patients with mGPS = 0 had a leak rate of 15.8% (n: 56), mGPS = 1 22.2% (n: 10) and mGPS = 2 42.1% (n: 8); Fisher’s exact test, p = 0.012. Table 2 demonstrates the average causal effect of preoperative mGPS > 0 on anastomotic leakage within 12 months with an OR of 1.09 (95% CI: 0.53–2.25) for mGPS = 1 and OR 4.11 (95% CI: 1.69–10.03) for mGPS = 2.
Table 2.
Average causal effect of preoperative modified Glasgow Prognostic Score (mGPS) and anastomotic leakage within 12 months, estimated by targeted maximum likelihood estimation, yielding odds ratios (ORs) with 95% confidence intervals (CIs)
| Exposure | Leak rate (%) | OR (95% CI)* |
|---|---|---|
| mGPS = 0 | 56 (15.8) | 1.00 (reference) |
| mGPS = 1 | 10 (22.2) | 1.09 (0.53–2.25) |
| mGPS = 2 | 8 (42.1) | 4.11 (1.69–10.03) |
*Covariates included in the lasso logistic regression models: Age, Body Mass Index, sex, clinical T stage, clinical N stage, clinical M stage, Charlson comorbidity index group, current smoking status, neoadjuvant therapy
CRP-to-albumin ratio and anastomotic leakage
Using the genetic algorithm [18], cut-off point optimization was performed on each imputed data set and the average of the estimated optimal cut-off points was used for categorizing CAR into ≤ 0.36 and > 0.36. Patients with CAR ≤ 0.36 had a leak rate of 15.6% (n: 57), while patients with CAR > 0.36 experienced a leak frequency of 32.1% (n: 17); Fishers’s exact test, p = 0.006 (Table S1). The average causal effect of preoperative CAR > 0.36 on anastomotic leakage, in comparison to CAR ≤ 0.36, was estimated at an OR of 2.25 (95% CI: 1.21–4.19).
Prediction modelling
mGPS
From the logistic regression model including all potential predictors, we found that mGPS = 2 (OR: 2.98; 95% CI: 1.01–8.82) had a stronger association to anastomotic leakage as compared to mGPS = 1 (OR: 1.42; 95% CI: 0.62–3.21). The reduced prediction model, including only predictors selected by lasso regression, showed similar results (Table S2). The full model resulted in an AUC of 0.73 (95% CI: 0.67–0.79), while the reduced model showed an AUC of 0.72 (95% CI: 0.66–0.79).
CAR
In both the full and reduced prediction models, CAR > 0.36 was found to be significantly associated with anastomotic leakage. As seen in Table S3, the full model resulted in an OR of 2.41 (95% CI: 1.16–5.00), and the reduced model displayed an OR of 2.32 (95% CI: 1.13–4.78). These models resulted in AUC 0.73 (95% CI: 0.66–0.79) and AUC 0.72 (95% CI: 0.66–0.79), respectively.
Discussion
Main results
In this multicentre study, the highest degree of preoperative inflammation, mGPS = 2, seemed to substantially increase the risk of anastomotic leakage within 12 months after anterior resection for rectal cancer, as compared to patients with mGPS = 0 or mGPS = 1. Similarly, using an internally derived cut-off, patients with a CAR > 0.36 were affected with a higher risk of leakage.
Incorporating the mGPS score and the CAR into prediction models for anastomotic leakage resulted in an only fairly performing model, particularly as the sample size did not permit internal validation, suggesting that more data are needed to accurately predict leakage after anterior resection for rectal cancer.
Strengths and weaknesses
The cohort is not truly population-based, although both university hospitals and county hospitals from different geographical regions throughout Sweden provided data. At each centre, chart review was compared to SCRCR data, minimising misclassifications. An advantage of the current report is the relatively short study period of five years, which ensures consistent management and surgical technique within centres. Moreover, the 12-month follow-up through chart review, instead of a conventional time frame of 30 days as captured by most registry studies [14, 16], means that more anastomotic leaks were discovered, as up to one third of the leaks occur 30 days after surgery [2]. Weaknesses of this study include the retrospective design, where in particular a high degree of exclusion due to lack of laboratory assessment is a threat to the validity of the study; this might introduce selection bias, especially as some centres had a more systematic approach of assessing preoperative bloods than others. Anecdotally, some of the missing samples were due to long waiting times to surgery, where the sampling was done outside of the allowed 30-day window. This also translated into a relatively small analytical sample, as evidenced by the wide confidence intervals. Although the use of directed acyclic graphs enabled consideration of confounding, residual confounding from unobserved covariates and misclassification are limitations common to any retrospective study. For example, there was no data concerning weight change and nutritional therapy, which might have influenced the results.
Literature review
A recently published large multicentre cohort identified low preoperative protein levels as an independent risk factor for anastomotic leakage after subtotal and total colectomy with ileorectal and ileosigmoid anastomosis in patients with colitis, polyposis disease and obstructed or synchronous colorectal cancer [30]. Although this patient cohort was heterogenous in nature and the study did not evaluate rectal cancer only, results were similar to the present study findings, despite a slightly different type of exposure. To our knowledge, only one other report has studied mGPS and the risk for anastomotic leakage after anterior resection due to rectal cancer. This small study identified mGPS = 2, as compared to mGPS = 0 or mGPS = 1, as an independent predictor for anastomotic leakage [10] with a very high point estimate and wide confidence intervals; the present study confirms this result, although with a lower estimate. In all likelihood, this is explained by the larger sample size in our study and another analytical approach, emphasising a causal interpretation. A few other studies have reported that the mGPS score and equivalent indices may be associated with postoperative complications in general [31, 32] after colorectal cancer surgery, further corroborating the findings in the present study. Similarly, there is a paucity of data concerning preoperative CAR and the subsequent risk of leakage. In a colorectal cancer cohort, comprising a similar number of anterior resections as in the present study, a CAR higher than 2.44 was found to be an independent risk factor for leakage [8]. Although CAR was shown to be an important predictor in the present study as well, differences such as no defunctioning stoma use and a markedly higher CAR cut-off make comparisons difficult, not the least as information on neoadjuvant treatment was missing from the above report.
Mechanisms
The inflammatory response to a tumour is linked to malnutrition and a decline in immunological response [32]. Importantly, the inflammation process leads to reduced albumin levels, regardless of the nutritional state. CRP is another acute phase protein, with a more immediate response to inflammation; as such, the combination of these inflammation-induced proteins might convey a more complete picture of the inflammatory state prior to surgery. The pathogenesis of anastomotic leakage is not fully understood, although it is recognized that leaks in almost all cases seem to be a multifactorial failure of wound healing rather than due to technical error [33]. Preoperative inflammation is associated with a higher incidence of postoperative infection, such as anastomotic leakage [34]. Potentially, preoperatively established systemic inflammation and changes in the local intestinal environment may persist postoperatively and influence the healing process [35] and could thus not only act as a marker for leakage, but also induce it.
Conclusion
A high degree of preoperative inflammation seems to be involved in the development of anastomotic leakage after anterior resection for cancer. However, inclusion of mGPS and CAR into prediction models with available clinical data did not result in accurate leakage prediction. Further research to delineate the role of preoperative inflammation is needed, as more precise modelling could facilitate clinical decision-making, potentially reducing the use of temporary stomas in low-risk patients and avoiding an anastomosis altogether in high-risk patients.
Supplementary Information
Below is the link to the electronic supplementary material.
Collaborators in RectoLeak study group
Anders Gerdin (Department of Surgical and Perioperative Sciences, Umeå University, Umeå, Sweden); Olle Sjöström (Department of Radiation Sciences, Umeå University, Umeå, Sweden); Maria Staffan (Department of Surgery, Falun Hospital, Falun, Sweden); Staffan Jangmalm (Department of Surgery, Växjö Hospital, Växjö, Sweden); Hanna Royson, Växjö (Department of Surgery, Växjö Hospital, Växjö, Sweden); Konstantinos Tsimogiannis (Department of Surgical Sciences, Uppsala University, Uppsala, Sweden); Kajsa Anderin (Department of Surgery, Ersta Hospital and Karolinska Institutet (KIDS), Stockholm, Sweden); Jonas Nygren (Department of Surgery, Ersta Hospital and Karolinska Institutet (KIDS), Stockholm, Sweden); Jennie Hurtig (Department of Surgery, Sunderby Hospital, Luleå, Sweden).
Authors’ contribution
Parisa Golshani drafted the manuscript, while Jennifer Park, Josefin Segelman, Peter Matthiessen, Marie-Louise Lydrup and Martin Rutegård reviewed and edited the manuscript text. Jenny Häggström performed the statistical analysis. Martin Rutegård and Jennifer Park conceived the study, while Martin Rutegård secured ethical approval. Funding was granted Martin Rutegård and Josefin Segelman. All of the above and the RectoLeak collaborators collected data for the study.
Funding
Open access funding provided by Umeå University. Knut and Alice Wallenberg Foundation, Swedish Society of Medicine, Cancer Research Foundation in Northern Sweden, Bengt Ihre’s Foundation, Swedish Cancer Society.
Data availability
Upon reasonable request, data and methodology can be shared. This also applies to the registry-based data used in the current study, while access to such data might be subject to external review by the Swedish Colorectal Cancer Registry steering committee.
Declarations
Ethics approval statement
The study was approved by the Regional Board of the Ethical Committee in Uppsala, Sweden (DNR: 2020-02285, 2020-03795, 2020-05126).
Permission to reproduce material from other sources
No material from other sources was included in this article.
Conflict of interest
None declared.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Contributor Information
Martin Rutegård, Email: martin.rutegard@umu.se.
on behalf of the RectoLeak study group:
Anders Gerdin, Olle Sjöström, Maria Staffan, Staffan Jangmalm, Hanna Royson, Konstantinos Tsimogiannis, Kajsa Anderin, Jonas Nygren, and Jennie Hurtig
References
- 1.Boström P, Haapamäki MM, Rutegård J, Matthiessen P, Rutegård M. Population-based cohort study of the impact on postoperative mortality of anastomotic leakage after anterior resection for rectal cancer. BJS Open februari. 2019;3(1):106–111. doi: 10.1002/bjs5.50106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Borstlap WAA, Westerduin E, Aukema TS, Bemelman WA, Tanis PJ, Dutch Snapshot Research Group (2017) Anastomotic Leakage and Chronic Presacral Sinus Formation After Low Anterior Resection: Results From a Large Cross-sectional Study. Ann Surg 266(5):870–7 [DOI] [PubMed]
- 3.Jutesten H, Buchwald PL, Angenete E, et al (2022) High risk of low anterior resection syndrome in long-term follow-up after anastomotic leakage in anterior resection for rectal cancer. Dis Colon Rectum 65:1264–1273 [DOI] [PubMed]
- 4.Holmgren K, Kverneng Hultberg D, Haapamäki MM, Matthiessen P, Rutegård J, Rutegård M. High stoma prevalence and stoma reversal complications following anterior resection for rectal cancer: a population-based multicentre study. Colorectal Dis. 2017;19(12):1067–1075. doi: 10.1111/codi.13771. [DOI] [PubMed] [Google Scholar]
- 5.Pommergaard HC, Gessler B, Burcharth J, Angenete E, Haglind E, Rosenberg J. Preoperative risk factors for anastomotic leakage after resection for colorectal cancer: a systematic review and meta-analysis. Colorectal Dis. 2014;16(9):662–671. doi: 10.1111/codi.12618. [DOI] [PubMed] [Google Scholar]
- 6.Sørensen LT, Jørgensen T, Kirkeby LT, Skovdal J, Vennits B, Wille-Jørgensen P. Smoking and alcohol abuse are major risk factors for anastomotic leakage in colorectal surgery. Br J Surg juli. 1999;86(7):927–931. doi: 10.1046/j.1365-2168.1999.01165.x. [DOI] [PubMed] [Google Scholar]
- 7.Oshi M, Kunisaki C, Miyamoto H, Kosaka T, Akiyama H, Endo I. Risk Factors for Anastomotic Leakage of Esophagojejunostomy after Laparoscopy-Assisted Total Gastrectomy for Gastric Cancer. Dig Surg. 2018;35(1):28–34. doi: 10.1159/000464357. [DOI] [PubMed] [Google Scholar]
- 8.Yu Y, Wu Z, Shen Z, Cao Y. Preoperative C-reactive protein-to-albumin ratio predicts anastomotic leak in elderly patients after curative colorectal surgery. Cancer Biomark Sect Dis Markers. 2020;27(3):295–302. doi: 10.3233/CBM-190470. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Paliogiannis P, Deidda S, Maslyankov S, Paycheva T, Farag A, Mashhour A (2021) C reactive protein to albumin ratio (CAR) as predictor of anastomotic leakage in colorectal surgery. Surg Oncol 38:101621 [DOI] [PubMed]
- 10.Sakamoto W, Ohki S, Kikuchi T, Okayama H, Fujita S, Endo H (2020) Higher modified Glasgow Prognostic Score and multiple stapler firings for rectal transection are risk factors for anastomotic leakage after low anterior resection in rectal cancer. Fukushima J Med Sci 66(1):10–6 [DOI] [PMC free article] [PubMed]
- 11.McMillan DC. The systemic inflammation-based Glasgow Prognostic Score: a decade of experience in patients with cancer. Cancer Treat Rev augusti. 2013;39(5):534–540. doi: 10.1016/j.ctrv.2012.08.003. [DOI] [PubMed] [Google Scholar]
- 12.Xu HJ, Ma Y, Deng F, Ju WB, Sun XY, Wang H (2017) The prognostic value of C-reactive protein/albumin ratio in human malignancies: an updated meta-analysis. OncoTargets Ther 10:3059–70 [DOI] [PMC free article] [PubMed]
- 13.Ge X, Cao Y, Wang H, Ding C, Tian H, Zhang X (2017) Diagnostic accuracy of the postoperative ratio of C-reactive protein to albumin for complications after colorectal surgery. World J Surg Oncol 15:15 [DOI] [PMC free article] [PubMed]
- 14.Moberger P, Sköldberg F, Birgisson H. Evaluation of the Swedish Colorectal Cancer Registry: an overview of completeness, timeliness, comparability and validity. Acta Oncol Stockh Swed december. 2018;57(12):1611–1621. doi: 10.1080/0284186X.2018.1529425. [DOI] [PubMed] [Google Scholar]
- 15.Rahbari NN, Weitz J, Hohenberger W, Heald RJ, Moran B, Ulrich A (2010) Definition and grading of anastomotic leakage following anterior resection of the rectum: a proposal by the International Study Group of Rectal Cancer. Surgery 147(3):339–51 [DOI] [PubMed]
- 16.Rutegård M, Kverneng Hultberg D, Angenete E, Lydrup ML (2017) Substantial underreporting of anastomotic leakage after anterior resection for rectal cancer in the Swedish Colorectal Cancer Registry. Acta Oncol 56(12):1741–5 [DOI] [PubMed]
- 17.Textor J, van der Zander B, Gilthorpe MS, Liskiewicz M, Ellison GT (2016) Robust causal inference using directed acyclic graphs: the R package ”dagitty”. Int J Epidemiol 45(6):1887–94 [DOI] [PubMed]
- 18.Van Der Laan MJ, Rubin D (2006) Targeted maximum likelihood learning. Int J Biostat 2(1):11 [DOI] [PMC free article] [PubMed]
- 19.Rose S, van der Laan MJ (2011) A Targeted Maximum Likelihood Estimator for Two-Stage Designs. Int J Biostat 7(1):17 [DOI] [PMC free article] [PubMed]
- 20.Tibshirani R. Regression Shrinkage and Selection Via the Lasso. J R Stat Soc Ser B Methodol. 1996;58(1):267–288. [Google Scholar]
- 21.Barrio I, Arostegui I, Rodríguez-Álvarez MX, Quintana JM. A new approach to categorising continuous variables in prediction models: Proposal and validation. Stat Methods Med Res december. 2017;26(6):2586–2602. doi: 10.1177/0962280215601873. [DOI] [PubMed] [Google Scholar]
- 22.Barrio I, Roca-Pardiñas J, Arostegui I. Selecting the number of categories of the lymph node ratio in cancer research: A bootstrap-based hypothesis test. Stat Methods Med Res mars. 2021;30(3):926–940. doi: 10.1177/0962280220965631. [DOI] [PubMed] [Google Scholar]
- 23.van Buuren S (2018). Flexible Imputation of Missing Data, Second Edition (2nd ed.). Chapman and Hall/CRC
- 24.Rubin DB, Schenker N. Multiple Imputation for Interval Estimation From Simple Random Samples With Ignorable Nonresponse. J Am Stat Assoc. 1986;81(394):366–374. doi: 10.1080/01621459.1986.10478280. [DOI] [Google Scholar]
- 25.R Core Team (2020). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/
- 26.CatPredi: Optimal Categorisation of Continuous Variables in Prediction Models version 1.3 from CRAN [Internet]. [citerad 11 december 2022]. Tillgänglig vid: https://rdrr.io/cran/CatPredi/
- 27.Gruber S, van der Laan M (2012) tmle: An R package for targeted maximum likelihood estimation. J Stat Softw 51(13):1–35
- 28.Friedman J, Hastie T, Tibshirani R. Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw. 2010;33(1):1–22. doi: 10.18637/jss.v033.i01. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.van Buuren S, Groothuis-Oudshoorn K (2011) mice: Multivariate Imputation by Chained Equations in R. J Stat Softw 45:1–67
- 30.Ocaña J, Pastor P, Timoteo A, et al (2023) Risk factors for anastomotic leakage and postoperative outcomes after total and subtotal colectomy: A nationwide retrospective cohort study (RIALTCOT Study Collaborative Group). Colorectal Dis 25:420–430
- 31.Richards CH, Leitch EF, Horgan PG, Anderson JH, McKee RF, McMillan DC (2010) The relationship between patient physiology, the systemic inflammatory response and survival in patients undergoing curative resection of colorectal cancer. Br J Cancer 103(9):1356–61 [DOI] [PMC free article] [PubMed]
- 32.Bailon-Cuadrado M, Perez-Saborido B, Sanchez-Gonzalez J, Rodriguez-Lopez M, Mayo-Iscar A, Pacheco-Sanchez D. A new dimensional-reducing variable obtained from original inflammatory scores is highly associated to morbidity after curative surgery for colorectal cancer. Int J Colorectal Dis september. 2018;33(9):1225–1234. doi: 10.1007/s00384-018-3100-0. [DOI] [PubMed] [Google Scholar]
- 33.Lam A, Fleischer B, Alverdy J (2020) The Biology of Anastomotic Healing-the Unknown Overwhelms the Known. J Gastrointest Surg Off J Soc Surg Aliment Tract 24(9):2160–6 [DOI] [PMC free article] [PubMed]
- 34.Moyes LH, Leitch EF, McKee RF, Anderson JH, Horgan PG, McMillan DC (2009) Preoperative systemic inflammation predicts postoperative infectious complications in patients undergoing curative resection for colorectal cancer. Br J Cancer 100(8):1236–9 [DOI] [PMC free article] [PubMed]
- 35.Holmgren K, Jonsson P, Lundin C, Matthiessen P, Rutegård J, Sund M (2022) Preoperative biomarkers related to inflammation may identify high-risk anastomoses in colorectal cancer surgery: explorative study. BJS Open 6(3):zrac072 [DOI] [PMC free article] [PubMed]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Upon reasonable request, data and methodology can be shared. This also applies to the registry-based data used in the current study, while access to such data might be subject to external review by the Swedish Colorectal Cancer Registry steering committee.