eP405 - Trending C-Reactive Protein Levels as a Predictor of Anastomotic Leak

Purpose/Background: Anastomotic leak (AL) following colorectal surgery is a devastating complication. Early detection of ALs remains difficult as the development of clinical signs of deterioration is a late finding. C-reactive protein (CRP) is a systemic marker of inflammation which has been proposed as an early screening tool for AL. However, absolute cut-off values have been shown to have poor positive predictive values (PPV). We propose the use of the trend of daily postoperative CRP levels for early AL detection.Methods/Interventions: A retrospective chart review of patients undergoing surgery requiring at least one small bowel or colorectal anastomosis at a single tertiary care center was performed. Daily postoperative CRPs were ordered based on surgeon preference. Patients with ≤ 2 postoperative CRP values were excluded. Patients were grouped based on the development of an AL (AL group) or the lack thereof (NAL group) within 1 week. CRP trends were compared by fitting a mixed model that used a compound symmetry covariance matrix and fit using Restricted Maximum Likelihood.Results/Outcomes: Of the 255 patients evaluated, 29 were excluded due to ≤ 2 CRP values. Of the 226 remaining patients, 6 (2.7%) developed an anastomotic leak with the median day of diagnosis being postoperative day (POD) 3. Patient demographics including age, sex, comorbidities, and surgical indication were similar between groups. Ileocolic and colorectal anastomoses were the most common types of anastomoses in both groups. None of the 6 AL patients underwent pre-emptive diversion compared to 9.5% of NAL patients.A significant difference in postoperative CRP trends was found between groups. NAL patients showed an increase in CRP on postoperative day 2 followed by a decline as opposed to AL patients who demonstrated a consistent CRP increase until AL was treated (Figure 1). Surgical approach (minimally invasive or open) had no impact on CRP trends. Similarly, additional procedures performed concurrent with the colorectal operation had no impact on CRP trends. Trend in daily white blood cell count showed no difference between groups. Furthermore, AL patients did not show signs of physiologic alterations such as tachycardia, tachypnea, or hypotension at the time of the CRP that led to diagnosis. Of the NAL patients, a wide range of maximum CRPs between 5.3 to 414.6 was noted, demonstrating this marker’s large interpatient variability. Peak CRP was reached between POD 1-6 with a median of POD 2. A decline in CRP was found at a median of 3 days (range 2-7 days). Patients ranged a 6.3 to 63.4% (median 27.2%) CRP reduction on the first day of CRP decline. Lastly, the PPV of an upward trending CRP was 0.19 compared to 0.08 if an isolated CRP cut-off of 170 mg/L on POD 3 was used.Conclusion/

Discussion: Despite the small sample size, we found a significant difference in the postoperative CRP trend in patients with anastomotic leaks when other markers of underlying infection, such as WBC and vital signs, had yet to show alterations. This indicates trending CRPs daily in the postoperative period may allow for earlier detection of ALs. The CRP trends, rather than spot values, were also less subject to the variability associated with surgical modality or inclusion of additional operations. Additionally, an upward trending CRP value had a PPV over double that of a particular cut-off CRP. This screening modality may decrease the number of unnecessary CT scans by better aiding physicians in identifying high risk patients. Further studies are needed to confirm these findings.Figure 1Postoperative CRP trends in patients with or without ALs. Median and interquartile range are displayed.