A Different Approach to the Use of C-Reactive Protein and Procalcitonin in Postoperative Infectious Complications
We read with interest the study by Giaccaglia et al1 regarding the use of inflammatory markers in postoperative infectious complications in colorectal surgery. In that multicentric, prospective, and observational study the authors corroborate previous findings on the use of C-reactive protein (CRP) and procalcitonin (PCT): both markers may be useful in this sense, but PCT measurement may not offer an advantage over CRP for predicting infectious complications,2–4 unless combined with CRP at the 5th postoperative day.1 We would like to draw attention to the group “complications other than anastomotic leak”. In this group, Giaccaglia et al1 include both noninfectious complications and infectious complications such as wound infection, pneumonia, and urinary tract infection. Analyzing a mix of noninfectious and infectious complications may lead to biased results because noninfectious complications such as bleeding, acute kidney injury, and urinary retention are not expected to alter the postoperative course of CRP, whereas infectious complications will.5
To date, the vast majority of works assessing the utility of CRP or PCT in postoperative settings have attempted to find a cut-off value on a specific postoperative day, using receiver operating characteristic curves and the corresponding area under the curve, which will identify patients developing or at higher risk of infectious complications, especially anastomotic leak. Different postoperative days have been evaluated with quite variable CRP thresholds and diagnostic performance.1–4,6–8 Nevertheless, although good specificity and negative predictive value have been reported, disparity in CRP thresholds and low sensitivity and positive predictive value (PPV) remain major drawbacks of this method.1,6,7
It has been reported that postoperative values of CRP in patients with no complications are mostly influenced by the extent of surgical trauma (surgical procedure and approach), and personal variability.5,8,9 The cut-off point approach for the use of inflammatory markers may underestimate this manifest variability, which could explain the discrepancy in CRP cut-off values among studies and the low sensitivity and PPV found by Giaccaglia et al1 (59.3%/73.9% and 19.5%/22.1%, on 3rd/5th postoperative day, respectively) and other authors.6,7 Because postoperative values of CRP present such great variability depending on the surgical procedure and approach undertaken,8,9 to optimize the performance of the cut-off point method, we would need to estimate a CRP threshold for each postoperative day, surgical procedure, and approach, making this methodology awkward and impractical. Therefore, the assessment of CRP on an individual basis could simplify its use and improve its performance for detecting/predicting infectious complications.
After a surgical aggression, CRP peaks at 48 hours in the majority of cases and then progressively decreases, diminishing by half at approximately the 5th postoperative day, if no complication occurs.5–7,10 Although the magnitude of the systemic inflammatory response, as evidenced by CRP absolute values, may vary, postoperative kinetics appear to be constant.5 We postulate that the postoperative peak of CRP at 48 hours demonstrates all the inflammation attributable to surgical trauma, and hence, a value superior to that peak at any point during the subsequent postoperative course would be manifesting additional inflammation produced by an infectious complication. Based on this thesis, we propose a different approach for the use of CRP in which for each subject, his or her own postoperative peak is used as a self-reference value, comparing the value on the day on which an infectious complication is suspected with the value presented at 48 hours (peak). This practice has been reported useful in two retrospective studies,5,9 showing good diagnostic accuracy (90.7%) and performance (sensitivity 82.9%, specificity 93.4%, PPV 81.0%, and negative predictive value 94.2%) in the case of anastomotic leak/intra-abdominal abscess.