Serum C-reactive protein is a useful marker to exclude anastomotic leakage after colorectal surgery

Anastomotic leakage is a complication of colorectal surgery. C-reactive protein (CRP) is an acute-phase marker that can indicate surgical complications. We determined whether serum CRP levels in patients who had undergone colorectal surgery can be used to exclude the presence of anastomotic leakage and allow safe early discharge. We included 90 patients who underwent colorectal surgery with primary anastomosis. Serum CRP levels were measured retrospectively on postoperative days (PODs) 1 – 7. Patients with anastomotic leakage (n = 11) were compared to those without leakage (n = 79). We statistically analysed data and plotted receiver operating characteristic curves. The incidence of anastomotic leakage was 12.2%. Diagnoses were made on PODs 3 – 24. The overall mortality rate was 3.3% (18.2% in the leakage group, 1.3% in the non-leakage group; P < 0.045). CRP levels were most accurate on POD 4, with a cutoff level of 180 mg/L, showing an area under the curve of 0.821 and a negative predictive value of 97.2%. Lower CRP levels after POD 2 and levels <180 mg/L on POD 4 may indicate the absence of anastomotic leakage and may allow safe discharge of patients who had undergone colorectal surgery with primary anastomosis.

of this study was to determine whether serum CRP levels of patients who had undergone emergency or elective colorectal surgery with primary anastomosis can be used to exclude the presence of anastomotic leakage and allow for safe and early discharge.

Methods
Ninety colorectal surgeries with primary anastomoses (ileocolic, colocolic, or colorectal) were performed in the General Surgery Department of Carapicuíba General Hospital between June 2014 and July 2018. The Research Ethics Committee of our institution approved this retrospective study and waived the need for informed consent. This study included patients of both gender who underwent elective or emergency colorectal surgery with primary anastomosis. The diseases that determined surgical treatment are presented in Table 1. Indications for emergency surgery included acute appendicitis, diverticulitis, and obstructive or perforated neoplasia. Patients who did not present at least 3 serum CRP levels within the first 7 postoperative days (PODs) were excluded. The patients were divided into 2 groups: leakage (n = 11) and non-leakage (n = 79). Clinical and demographic characteristics of the groups are shown in Table 1.
Serum CRP levels were evaluated on PODs 1 through 7 by immunoassays using the turbidimetric method with an Architect Plus C4000 analyser (Abbot, Lake Bluff, IL, USA). CRP levels >5 mg/L were considered altered. Patients were evaluated daily for the presence of abdominal pain, fever, volume, return of bowel habits, and/or appearance of abdominal drainage. Patients with altered parameters underwent laboratory and imaging examinations (CT or radiography). All patients received antibiotic prophylaxis, and mechanical preparation of the colon was conducted only for elective surgeries (71.1%). www.nature.com/scientificreports www.nature.com/scientificreports/ Anastomotic leakage was defined using the following clinical and radiologic criteria: 1) presence of air or abscess near the site of anastomosis identified on CT, 2) purulent discharge or enteric secretion through the drain, and 3) clinical signs of peritonitis and/or presence of faecal or purulent discharge during surgical re-approach. Antibiotics were restarted in patients with leakage.
Categorical variables were presented as frequency and percentage, and quantitative variables, as median and interquartile range. The Shapiro-Wilk test was used to define normality, whereas the Mann-Whitney U test was used for bivariate comparisons. Receiver operating characteristic (ROC) curves were plotted using the values generated by logistic regression analysis. Sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), accuracy, and area under the curve (AUC) were calculated. R language software (RStudio, Inc, Boston, MA, USA; www.rstudio.com) was used for statistical analysis with the level of significance set at 5% (P < 0.05).

Results
During the study period, 90 patients underwent colorectal surgery with primary anastomosis. The median age was 56 years and 55.6% of patients were male (Table 1). Ostomy closure (36.7%) and colon adenocarcinoma resection (28.9%) were the most common surgical indications. The use of abdominal drainage did not affect the onset of leakage (P = 0.694). Colic anastomoses were created using a mechanical stapler in 70 patients (77.8%) and sutures in 20 patients (22.2%).
There were no statistically significant differences in serum CRP levels in the first 3 PODs. After POD 4, however, there was a significant increase in serum CRP levels in patients with anastomotic leakage (median, 246.4 mg/L) compared with those without leakage (median, 113.5 mg/L; P = 0.002) (Online Resource 2). Serum CRP levels increased from POD 2 in patients with leakage and decreased in those without leakage (Fig. 1). Peak levels were seen on POD 5 in patients with leakage and on POD 2 in those without leakage.
Analyses of the ROC curves from PODs 3 through 5 are presented in Tables 2-4. Sensitivity, specificity, NPV, PPV, accuracy, and AUC are presented in Figs. 2-4. A cutoff value of 220 mg/L was established on POD 3 with an AUC of 0.643, NPV of 89.3%, PPV of 20%, sensitivity of 71%, and specificity of 45%. On POD 4, with a cutoff value of 180 mg/L, the AUC was 0.821, NPV was 97.2%, sensitivity was 72.3%, and specificity was 88.9%. Patients with CRP levels <180 mg/L on POD 4 had a 12.2% probability of developing anastomotic leakage.

Discussion
Anastomotic leakage is an undesirable complication of colorectal surgery 1 , resulting in increased length of hospitalization, increased treatment costs, delayed return of intestinal homeostasis, and decreased survival 34,35 . Because vital signs and leukocyte numbers are slow in responding, it is important to identify tools to detect early leakage 36,37 . Early diagnosis is essential to reduce morbidity and mortality 33,38-40 because delayed diagnosis can increase mortality by 18% 37 .
CRP is an acute-phase protein produced by hepatocytes after inflammatory stimulation 18,25 . It is a useful marker to monitor and identify postoperative complications because it has a short half-life 38,40 . This protein has been shown to be as effective and sensitive as a predictor of anastomotic leakage 34,35,[39][40][41] and postoperative infection 15,19,38 . In fact, increased CRP levels are more sensitive to diagnose surgical complications than are increased erythrocyte sedimentation rate, leukocytes, body temperature, and heart rate 25 22 . Medications such as corticosteroids and statins may also alter this response, which could decrease serum CRP levels and alter the interpretation of cutoff levels 22 .
The surgical approach also influences serum CRP levels. Waterland et al. found higher CRP levels in patients who underwent open surgery than in those who underwent laparoscopic surgery 40    www.nature.com/scientificreports www.nature.com/scientificreports/ reported that a cutoff level of 135 mg/L on POD 5 was a good predictor of leakage 39 . In another study, Muñoz et al. evaluated only patients who underwent elective laparoscopic colorectal cancer resection using the enhanced recovery after surgery (ERAS) protocol. In their study, CRP had a high NPV on POD 3 with a cutoff level of 163 mg/L 33 .
Singh et al. conducted a systematic review of 6 studies including >2400 patients 41 and found that CRP levels were comparable in terms of accuracy on PODs 3, 4, and 5. On the other hand, Warschkow et al. conducted a meta-analysis and reported that CRP levels were more accurate on POD 4, demonstrating a high NPV for postoperative complications, with a cutoff value of 135 mg/L (38). Our study showed a high NPV, sensitivity, and specificity with a cutoff value of 180 mg/L on POD 4. This high cutoff value may be related to the inclusion of patients who underwent emergency colorectal surgery. This study also identified decreased CRP levels on POD 2 in patients without leakage, similar to the findings reported by Woeste et al. 34 . However, it is difficult to compare studies because of the non-standardization of anastomotic leakage definitions, day and time of CRP testing, patient selection, and surgical approach 22,34,41 .
Nonetheless, most studies support the notion that patients with anastomotic leakage present higher and sustained elevation of serum CRP levels in the postoperative period compared with patients without leakage 33,34,36,39,42 . According to several studies, increased serum CRP levels precede radiologic and clinical diagnosis of anastomotic leakage. They reported that the detection of sustained serum CRP elevation may decrease the time for indicating reoperation, which could lead to lower mortality rates and hospital costs 34,39 .
Sawyer et al. analysed the differences between short and extended use of antibiotics in >500 patients who underwent complicated intra-abdominal infection treatment and colorectal surgery. They found no significant differences in terms of surgical site infection, recurrent intra-abdominal infection, or death 43 . These findings suggest that with short-term use of effective and safe antibiotic therapy, patients undergoing emergency colorectal surgery could also benefit from the analysis of serial CRP levels, which could provide the possibility of early and safe hospital discharge 43 .
Patients tend to be discharged early, between PODs 4 and 5, with the advent of multimodal accelerated postoperative recovery protocols such as ERAS 33,35,41 . Because most surgical complications occur after patients are discharged, between PODs 5 and 8, a marker such as CRP, which has a high NPV on POD 4, could be used to exclude anastomotic leakage and other postoperative complications 38 . In addition to the use of scores to identify patients at high risk of anastomotic leakage, postoperative investigation protocols for patients with sustained elevation of CRP levels after POD 2 or with levels above the cutoff on POD 4 should be generated 13,38 . Because of the high NPV, serum CRP levels on POD 4 seem to play an important role in the exclusion of anastomotic leakage 33,35,39,40 .
In conclusion, serum CRP levels can be routinely analysed in patients who undergo elective or emergency colorectal surgery. Decreased CRP levels after POD 2 can exclude anastomotic leakage because they are not influenced by factors such as individual inflammatory response, type of approach, or surgical indication. A cutoff level of 180 mg/L on POD 4 can indicate high reliability for hospital discharge due to a low probability of anastomotic leakage.
Postoperative serum CRP levels in patients who undergo colorectal surgery with primary anastomosis could become a useful marker for the exclusion of anastomotic leakage. This was a single-centre study with a small sample size; therefore, prospective multicentre studies with a greater number of patients are necessary to confirm our findings and extend them to clinical practice.

Data availability
All data generated or analysed during this study are included in this published article.