Trying to figure out if a febrile patient showing up to the ED is having a bacterial infection is a conundrum. The main concern is bacteremia, since it confers higher mortality depending on the type of bacteremia a patient has. Inflammatory markers such as CRP and the neutrophil-to-lymphocyte ratio are relatively useful surrogates that may help upfront while blood cultures cook in the lab (and they can take anywhere from 24-48hrs before you get any useful info). Ideally, if one is dealing with a viral infection it would be helpful to know to avoid unnecessary antibiotic therapy. Enter the (semi) procalcitonin aka PCT.
Its use in differentiating bacterial from viral infections has gained some popularity following some data on its utility to de-escalate antibiotics in respiratory infections (see here). To cut to the chase, those who have a respiratory infection but have bacteria as the etiology will have higher PCT, with the levels rising within 3 to 6hrs, mimicking the host inflammatory response (1). In a sense, the higher the PCT levels the more severe and widespread the infection is, as it can be seen in this table (2; note to convert to ng/mL, multiply the value by 10):
In other words, levels <0.5ng/mL means infection is highly unlikely and looking at the suggested algorithm within the above paper (and here 3) you can see it is used as a cut-off to start antibiotics
There is some data to suggest that it can be useful in febrile patients with undifferentiated sepsis to predict whether or not they will have bacteremia, which may allow us to, if not de-escalate antibiotics, at least rule out bacterial infections as the etiology of a fever. Low numbers do not rule out localized infection but higher number suggest the presence of systemic infection. While it sounds promising, the data here suffers from a few issues:
- Different populations of patients are typically tested, so its utility may differ from ED to the ICU
- Different cut-offs for what allows you to “rule out” a bacteremia. In other words, each author is trying to use their population to determine a cut-off which may or may not be helpful
- The relatively low sensitivity
The first point where PCT measurements may be helpful would be the ED and there is some data to suggest its utility. In a cohort of 295 patients (4), a procalcitonin of 0.1 ng/mL was found in more patients with true bacteremia (75%) when compared to patients with contaminants (28.7%). Using a cut-off of 0.1475 ng/mL, the AUC-ROC yielded a sensitivity of 75%, specificity of 78.9%, and PPV and NPV of 16.9% and 98.2%, respectively. A retrospective study of 295 patients who presented to the ER found that, while PCT levels were correlated with blood culture positivity, it did not correlate with SIRS criteria, suggesting it may have some utility in differentiating other causes of SIRS from bacteremia (5):
At a threshold of 0.1475ng/mL, sensitivity, specificity, PPV, and NPV were 75%, 79%, 17%, and 98.2%, respectively. At a threshold of 1 ng/ML, the values were 37%, 93%, 23%, and 96.3%. In other words, in this cohort, a lower PCT was able to rule out bacteremia. Of course, the threshold at which you would have the highest NPV is way lower than the manufacturer threshold, so even a “normal” PCT of 0.5 ng/mL may not completely rule out bacteremia, but it retains a high enough NPV it could be used for this purpose.
An ED study of over 3300 blood samples found similar results as prior studies (7):
Interestingly, the investigators combined other inflammatory markers including body temperature, CRP, and ANC and found that the addition of PCT to inflammatory markers yielded a better sensitivity:
What about the inpatient setting? A retrospective study of 628 inpatients with bacteremia compared CRP, WBC, and procalcitonin on admission (7). Of these, 162 had positive blood cultures, with 57 having coagulase-negative staphylococcus (CoNS). While there was a correlation with higher procalcitonin levels and positive blood culture results, the PPV was significantly lower than the NPV. Notably, a cut-off value of 0.5 ng/mL yielded a NPV of 93%, assuming a prevalence of roughly 20% for bacteremia:
When compared to WBC, CRP, and neutrophil-lymphocyte ratio, procalcitonin had a better performance:
PCT had a higher utility in CoNS to differentiate between contamination and true bacteremia, but again it was not quite perfect in its performance:
In another of 165 febrile patients with suspected community acquired infection, patients with bacteremia had higher levels of inflammatory markers, including PCT (8):
The sensitivity/specificity was calculated for different cutoffs, with the best one being 0.4ng/mL:
When looking at AUR-ROC for CRP and ESR when compared to PCT, the latter performed better than the previous 2:
A similar study of 156 patients admitted to an anesthesia ICU found that while procalcitonin was higher in the bacteremic group than in the non-bacteremic group (11.9 ng/dL or 0.119 ng/mL vs 5.9 ng/dL or 0.059 ng/mL), the difference was not statistically significant (p =0.168). Using a cut-off of 0.093ng/mL, this yielded a sensitivity/specificity of 35% and 83%, respectively. Moreover, the difference in PCT levels was also not statistically different between gram-negative or gram positive bacteremia (p = 0.145). There was a difference in PCT levels in SIRS (0.9 ng/dL), sepsis (4.2), and severe sepsis/septic shock (14.2), however this did not help predict bacteremia. Indeed, in this cohort the NPV was not incredibly helpful:
So there is a correlation, but the different cut-offs make it a bit difficult to use it as an adequate rule out test. In a cohort of over 1000 patients, procalcitonin levels were used to predict results for a SeptiFast multi-pathogen PCR test from blood (i.e. similar to a respiratory viral panel, 10). Here, the test seemed to perform better but specifically to predict the SeptiFast PCR rather than a true blood culture:
Using a cut-off of 0.37ng/mL, the sensitivity of PCT to predict blood culture results was 87%. In a cohort of over 300 patients with at least one positive blood cultures, those with a PCT level >0.5ng/mL were more likely to have leukocytosis, tachypnea, requirement for vasopressor support, and more likely to be admitted to the ICU when compared to those who had a PCT level <0.5ng/mL (11). Moreover, in this cohort 59% of those with 2 sets of positive blood cultures had low PCT compared to 62% with high PCT levels (p = 0.634). Indeed, there was no statistical difference in PCT levels based on organism isolated, suggesting at least in this cohort the NPV was not good enough as a rule-out test:
One retrospective study of 507 of bacteremia found that bacteremia due to E. coli tended to have the highest procalcitonin level, significantly higher than Staph aureus, but lower than CoNS (though still above the usual threshold, 12):
Moreover, higher procalcitonin levels were associated with higher 28-day mortality:
So Is it Useful?
Maybe? Given we have no clear cut-off, the only utility would be in a completely undetectable procalcitonin level, though some labs may not report a value under 0.5ng/mL. Indeed, the above studies may not have much in the way of external validity as they used values significantly lower than the manufacturer cut-off. Even using a cut-off of 0.5ng/mL may not yield the greatest performance. A meta-analysis of 58 studies with over 16,000 patients found that at a threshold of 0.5ng/mL, the overall sensitivity and specificity was 76 and 69%, respectively (13):
Notably, the test had higher sensitivity in ICU patients while its use in those with immunosuppression had the lowest sensitivity. Further, the NPV was highest in the ICU population as well:
Another meta-analysis of 17 studies and 2008 patients focused on the ED population (14). Here, the pooled sensitivity/specificity using studies with a 0.5 ng/mL threshold was 76% and 70%, respectively, though the performance of the test varied widely depending on the population:
- Procalcitonin is released 3-6 hours after infection with bacteria
- It allows you to differentiate bacterial infection from viral infection
- Its utility in predicting bacteremia is ok, at best given the different cut-offs used in different studies
- Higher PCT levels = more likely it is bacterial infection, but if your lab reports a value of <0.5ng/mL, the test performance may not be the best to rule out bacteremia
- It is better than nothing, but would not bank on this being the sole arbiter of giving empiric antibiotics for possible bacteremia
- Brunkhorst FM, Heinz U, Forycki ZF. Kinetics of procalcitonin in iatrogenic sepsis. Intensive Care Med. 1998 Aug;24(8):888-889.
- Samsudin I, Vasikaran SD. Clinical Utility and Measurement of Procalcitonin. Clin Biochem Rev. 2017 Apr;38(2):59-68. PMID: 29332972; PMCID: PMC5759088.
- Bouadma L, Luyt CE, Tubach F, Cracco C, Alvarez A, Schwebel C, Schortgen F, Lasocki S, Veber B, Dehoux M, Bernard M, Pasquet B, Régnier B, Brun-Buisson C, Chastre J, Wolff M; PRORATA trial group. Use of procalcitonin to reduce patients’ exposure to antibiotics in intensive care units (PRORATA trial): a multicentre randomised controlled trial. Lancet. 2010 Feb 6;375(9713):463-74. doi: 10.1016/S0140-6736(09)61879-1. Epub 2010 Jan 25. PMID: 20097417.
- Albrich WC, Mueller B. Predicting bacteremia by procalcitonin levels in patients evaluated for sepsis in the emergency department. Expert Rev Anti Infect Ther. 2011 Jun;9(6):653-6. doi: 10.1586/eri.11.42. PMID: 21692670.
- Riedel S, Melendez JH, An AT, Rosenbaum JE, Zenilman JM. Procalcitonin as a marker for the detection of bacteremia and sepsis in the emergency department. Am J Clin Pathol. 2011 Feb;135(2):182-9. doi: 10.1309/AJCP1MFYINQLECV2. PMID: 21228358.
- Kim SY, Jeong TD, Lee W, Chun S, Min WK. Procalcitonin in the assessment of bacteraemia in emergency department patients: results of a large retrospective study. Ann Clin Biochem. 2015 Nov;52(Pt 6):654-9. doi: 10.1177/0004563214568685. Epub 2015 Jan 9. PMID: 25575698.
- Wang H, Yin F, Shen DX, Zhang YJ, Luo YP, Liu CJ, Wang KF, Zhou G, Ye LY, Chen G, Wang XN. Predictive value of procalcitonin for excluding bloodstream infection: results of a retrospective study and utility of a rapid, quantitative test for procalcitonin. J Int Med Res. 2013 Oct;41(5):1671-81. doi: 10.1177/0300060513497558. Epub 2013 Sep 24. PMID: 24065454.
- Chirouze C, Schuhmacher H, Rabaud C, Gil H, Khayat N, Estavoyer JM, May T, Hoen B. Low serum procalcitonin level accurately predicts the absence of bacteremia in adult patients with acute fever. Clin Infect Dis. 2002 Jul 15;35(2):156-61. doi: 10.1086/341023. Epub 2002 Jun 17. PMID: 12087521.
- Demirdal T, Sen P, Nemli SA. Diagnostic Value of Procalcitonin in Predicting Bacteremia in Intensive Care Unit. Indian J Crit Care Med. 2018 Feb;22(2):78-84. doi: 10.4103/ijccm.IJCCM_437_17. PMID: 29531446; PMCID: PMC5842461
- Mencacci A, Leli C, Cardaccia A, Meucci M, Moretti A, D’Alò F, Farinelli S, Pagliochini R, Barcaccia M, Bistoni F. Procalcitonin predicts real-time PCR results in blood samples from patients with suspected sepsis. PLoS One. 2012;7(12):e53279. doi: 10.1371/journal.pone.0053279. Epub 2012 Dec 27. PMID: 23300907; PMCID: PMC3531374.
- Kellie J Goodlet, Emily A Cameron, Michael D Nailor, Low Sensitivity of Procalcitonin for Bacteremia at an Academic Medical Center: A Cautionary Tale for Antimicrobial Stewardship, Open Forum Infectious Diseases, Volume 7, Issue 4, April 2020, ofaa096, https://doi.org/10.1093/ofid/ofaa096
- Ogasawara S, Saito N, Hirano R, Minakawa S, Kimura M, Kayaba H. Clinical relevance of procalcitonin values in bacteremia. J Infect Chemother. 2020 Oct;26(10):1048-1053. doi: 10.1016/j.jiac.2020.05.023. Epub 2020 Jun 25. PMID: 32595106.
- Hoeboer SH, van der Geest PJ, Nieboer D, Groeneveld AB. The diagnostic accuracy of procalcitonin for bacteraemia: a systematic review and meta-analysis. Clin Microbiol Infect. 2015 May;21(5):474-81. doi: 10.1016/j.cmi.2014.12.026. Epub 2015 Jan 14. PMID: 25726038.
- Jones AE, Fiechtl JF, Brown MD, Ballew JJ, Kline JA. Procalcitonin test in the diagnosis of bacteremia: a meta-analysis. Ann Emerg Med. 2007 Jul;50(1):34-41. doi: 10.1016/j.annemergmed.2006.10.020. Epub 2006 Dec 11. PMID: 17161501.