ORIGINAL ARTICLE
Procalcitonin as a Marker of Bacteremia in Patients With Fever
and Acute Lymphoblastic Leukemia
David Vyles, DO, Forrest Gnagi, MD, Blake Bulloch, MD, Jared Muenzer, MD, and Chengcheng Hu, PhD
Background: Children undergoing treatment for acute lymphoblastic
leukemia (ALL) often present to the emergency department (ED) with
a fever. They are at high risk of bacteremia secondary to being immu-
nocompromised. Recent reports indicate that procalcitonin (PCT) is a
useful marker of bacteremia in children.
Objective: Our objective was to evaluate the clinical utility of PCT as a
rapid marker of bacteremia in children with ALL presenting to the
ED with a fever. In addition, we compared the results of PCTwith white
blood cell (WBC) count, erythrocyte sedimentation rate (ESR) and
C-reactive protein (CRP).
Methods: Retrospective chart reviews were conducted of 492 patients
with a total of 735 visits presenting to the ED from January 2009 to June
2012 with fever and a history of ALL where a PCT and a blood culture
(BC) were obtained,. Positive BCs determined to be contaminants were
excluded. The predictive powers of PCT, WBC, ESR, and CRP for bac-
teremia were evaluated using the area under the receiver operating char-
acteristic curve with 95% confidence intervals (CI). In addition, each of
the 4 markers were also examined in a logistic regression model as a poten-
tial predictor of the BC result.
Results: A total of 735 PCT values were correlated with BC results.
There were 76 (10.3%) true-positive BCs. The area under the receiver
operating characteristic curve was 0.729 (95% CI, 0.661–0.792) for
PCT, 0.685 (95% CI, 0.531–0.823) for ESR, 0.622 (95% CI, 0.460–
0.796) for CRP, and 0.567 (95% CI, 0.483–0.649) for WBC. When
logistic regression was used, the transformation log PCT was signifi-
cantly associated with BC result whereas each of the other 3 markers,
after appropriate transformation to remove heavy skewness, was not
significant (all P > 0.1). A doubling of PCT was associated with an
odds ratio of 1.32 for positive BCs (95% CI, 1.15–1.53).
Conclusions: Procalcitonin value was significantly associated with
positive BC (P < 0.0001). The diagnostic performance of PCT was bet-
ter than the other markers of inflammation. Its use in the ED in a select
population of patients may be of significant value in identifying bacter-
emia. This has the potential to lead to a decrease in unwarranted use of
antibiotics, hospital length of stay, and health care expenditures.
Key Words: fever, leukemia, procalcitonin
(Pediatr Emer Care 2016;00: 00–00)
C hildren undergoing treatment for acute lymphoblastic leuke-
mia (ALL) often present to the emergency department (ED)
with a fever. They are at high risk of bacteremia secondary to be-
ing immunocompromised. Bacterial culture is the gold standard
for the diagnosis of bacteremia, but the results may take several
hours to detect and always carry the risk of contamination. Clini-
cians must decide on an appropriate therapy, which may involve
empiric antibiotics, based upon the risk of infection. In an initial
study, data showed that procalcitonin (PCT) seemed to have a high
From the Phoenix Children's Hospital, Phoenix, AZ.
Disclosure: The authors declare no conflict of interest.
Reprints: David Vyles, MD, Department of Emergency Medicine, Phoenix
Children's Hospital, 1919 E Thomas Rd, Phoenix, AZ 85006
(e‐mail: dvyles@mcw.edu).
Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved.
ISSN: 0749-5161
Pediatric Emergency Care • Volume 00, Number 00, Month 2016
Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
negative predictive value (NPV) for sepsis when correlated with
blood culture (BC) results.1 Other inflammatory diagnostic labo-
ratory tests have been examined as potential markers of sepsis,
including total white blood cell (WBC) count, C-reactive pro-
tein (CRP), erythrocyte sedimentation rate (ESR), and cytokines
(tumor necrosis factor alpha [TNF-α], interleukin [IL]-1β, or
IL-6).2–5 However, these markers have lacked sufficient speci-
ficity for the diagnosis of bacteremia.
The development of a tool to predict bacterial infection using
PCT and other easily available laboratory tests would be useful to
avoid unnecessary exposure to antibiotics, invasive procedures,
and hospitalization. Procalcitonin is a calcitonin precursor5 that
has been found to be elevated in the serum or plasma of many con-
ditions, including bacteremia.1,2,6 It has been demonstrated in sev-
eral studies that its elevation is specific to bacterial infections and
does not tend to rise in the face of a viral etiology.2,7,8 Previous
studies have examined the use of PCT in a variety of different pop-
ulations and settings which range from adult intensive care units to
pediatric EDs.1,9 However, its use in pediatric oncology patients
presenting with a diagnosis of fever and ALL has yet to be thor-
oughly evaluated. Several studies have shown promising results
in the use of PCT compared with other biomarkers of infection
in oncology patients.8,10–16
The objective of this study was to evaluate the clinical utility
of PCT as a rapid marker for bacteremia in the ED population of
children with a fever and a history of ALL. The secondary objec-
tive was to compare PCT values to other commonly used diagnos-
tic markers of infection in this setting and evaluate if it is more
sensitive and specific to predict positive BC.
METHODS
Study Design
After approval by the institutional review board, a retrospec-
tive chart review was conducted on patients who presented to the
ED with fever and a diagnosis of ALL.
Study Setting and Population
The study took place at an urban, tertiary care, level-1, free-
standing children's hospital with an annual volume of 80,000 pa-
tients. The review period identified was from January 2009 to
June 2012.
Study Protocol
Patients were identified, and data were abstracted in children
with fever (≥38.3°C once, or ≥38.0°C taken on 2 occasions, at
least 1 hour apart) and a history of ALL, with PCT test results
and a BC drawn within 6 hours of each other. Results of the
obtained BCs were correlated with PCT, ESR, CRP, and WBC.
Exclusion criteria included the following: a patient receiving
antibiotics within the previous 24 hours of presenting to the ED,
patients with false-positive BCs due to suspected contaminants,
and patients with BCs and biomarkers of infection drawn more
than 6 hours beyond the defined window of PCT collection.
www.pec-online.com 1
Vyles et al Pediatric Emergency Care • Volume 00, Number 00, Month 2016

in the lab. In previous studies, levels reported as less than

TABLE 1. Pathogen Etiology, all Cultures Obtained From 0.5 ng/mL were considered not indicative of sepsis whereas levels
Central Access
greater than 2 ng/mL were indicative of high sepsis probability.
Levels in between 0.5 and 2 suggested a repeat level be drawn af-
Pathogen No. Patients
ter 24 hours. We evaluated the PCT level as a continuous variable.
Coagulase-negative Staphylococcus 17 Blood cultures were performed using the bioMerieux BacTAlert
Klebsiella pneumonia 11 3D (bioMérieux, Inc., Durham, NC).
Pseudomonas aeruginosa 7
Staphylococcus aureus 7 Data Analysis
Enterococcus faecium 5 To study the capability of each marker to predict the result of
Enterobacter species 5 a certain culture, the area under the receiver operating characteris-
Alpha-hemolytic streptococci 4 tic curve (AUC) was used. Logistic regression models were used
Haemophilus influenzae 3 to associate all 4 markers with the result of each BC, with gener-
alized estimating equations used to account for multiple mea-
Escherichia coli 3
surements of the same patient. For each of the 4 markers and
Bacillus species 2 each BC, the sensitivity, specificity, positive predictive value
Streptococcus mitis 2 (PPV), and NPV were calculated for a selected biomarker.
Micrococcus luteus 2
Acinetobacter species 2 RESULTS
Salmonella arizonae 1
During a 3-year period of data collection, there were
Citrobacter freundii 1 492 patients with a total of 735 visits available for retrospective
Ochrobactrum anthropi 1 review. In evaluating the different biomarkers of infection, PCT
Mycobacterium chelonae 1 was measured at all visits, WBC was measured at all but 25 visits
Leuconostoc pseudomesenteroides 1 (96.6% of the time), CRP at 168 visits (22.9% of the time), and ESR
Staphylococcus epidermidis 1 at only 161 visits (21.9% of the time). Blood culture results were
available for all 735 visits, with 76 positives (10.3%) and 659 nega-
Blood cultures with diptheroids and coagulase-negative tives (89.7%). The bacterial etiology of each BC is listed Table 1.
Staphylococcus were considered negative or contaminants. If a There was a statistically significant difference in PCT (SD) levels
patient had a central line, defined as a portacath, peripher- between patients with a positive BC result (6 [16.4] ng/mL) versus
ally inserted central catheter, or broviac, then the growth of a negative BC result (0.8 [2.6] ng/mL) (P < 0.0001). The sum-
coagulase-negative Staphylococcus was considered a positive mary of culture results and the association of marker values with
BC. Blood culture results that had no growth at 5 days were culture results are summarized in Tables 1, 2, and 3a.
also considered negative. The AUC was used to check the predictive power of each
Patients were identified using the International Classification marker for BC result. In regard to BC results, the AUC levels of
of Diseases, Ninth Revision (ICD-9) codes 78060, 28800, 78061 PCT (0.729 [95% CI], 0.661–0.792) and ESR (0.685 [95% CI],
(fever unspecified, neutropenia not otherwise specified, fever pre- 0.531–0.823) were both significantly higher than 0.5, whereas
senting with conditions classified elsewhere, and fever in other AUC levels of WBC (0.501 [95% CI], 0.364–0.640) and CRP
diseases, respectively), and all with underlying ICD code/ (0.602 [95% CI], 0.355–0.829) were not significantly different
diagnosis of 20400 (acute lymphoid leukemia). In some instances, from 0.5, as shown in Table 4 and Figure 1.
patients had many ICD-9 codes for 1 chart created on a single ad- In addition, each of the 4 markers was examined in a logistic
mission. Subsequently, all but one of the ICD-9 codes were de- regression model as a potential predictor of the BC result, with
leted to prevent the appearance of multiple patient visits for a potential correlation of multiple visits of the same patient ad-
single ED visit and/or subsequent hospitalization. justed for the generalized estimating equations. The transfor-
It is standard in our ED that all children with a fever and an mation log PCT was significantly associated with BC result
oncologic diagnosis have a complete blood count with differential, whereas each of the other 3 markers, after appropriate trans-
BC either from the central line or peripherally drawn, and PCT formation to remove heavy skewness, was not significant (all
levels drawn. At our institution, all children with a fever and a cen- P > 0.1). A doubling of PCT was associated with an odds ratio
tral venous catheter have central line-associated bloodstream in- of 1.32 for positive BC (95% CI, 1.15–1.53). These results are
fections until proven otherwise and have a complete blood count summarized in Table 3b.
and BCs obtained from a central venous catheter on presentation The sensitivity, specificity, PPV, and NPV of PCT as a bio-
if present. All had empiric antibiotics initiated with ceftriaxone marker of infection are listed in Table 5.
if the patient was clinically stable and nonneutropenic; with cefe-
pime and gentamicin in the stable neutropenic patient; or with DISCUSSION
cefepime, gentamicin, and vancomycin in a ill-appearing or toxic- This study looked to evaluate the clinical utility of PCT as a
appearing neutropenic patient. Patients were either discharged rapid marker for bacteremia in the ED population of children with
home or admitted to the hospital based on laboratory findings
and clinical appearance.
Levels of PCT were determined using the Brahms Kryptor TABLE 2. Summary of Culture Results
compact immunoassay analyzer (Thermo Scientific, Berlin,
Germany) which requires a minimum of 200 μL of plasma, serum, Culture Type Blood
or whole blood and can detect a level as low as less than No. measurements 735
0.02 ng/mL in 20 to 25 minutes. The average time to obtain the No. positive results, (%) 76 (10.3%)
results of PCT on a blood sample is 60 minutes. This is subject No. patients 492
to some variation, with longer times pending degree of volume

2 www.pec-online.com © 2016 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Pediatric Emergency Care • Volume 00, Number 00, Month 2016 Procalcitonin in Pediatric Patients With ALL

TABLE 3A. Association of Marker Values and Culture Results

PCT WBC CRP ESR

Number with Number with Number with Number with
Culture Number marker, (%) Mean (SD) marker, (%) Mean (SD) marker, (%) Mean (SD) marker, (%) Mean (SD)
Blood + 76 76 (100%) 6 (16.4) 74 (97.4%) 7 (7.6) 14 (18.4%) 11.8 (13.3) 10 (13.2%) 19.4 (15.4)
− 659 659 (100%) 0.8 (2.6) 636 (96.5%) 8.1 (7) 154 (23.4%) 5.9 (7) 151 (22.9%) 37.4 (29.5)
P <0.0001 0.008 0.098 0.007

TABLE 3B. Partial AUC (False-Positive Ratio < 0.5) of Markers to Predict BC Result

PCT WBC CRP ESR

pAUC (95% CI) 0.268 (0.216–0.317) 0.167 (0.117–0.222) 0.224 (0.113–0.335) 0.200 (0.091–0.331)

TABLE 4. Area Under the Receiver Operating Characteristic Curve of Markers to Predict Blood Culture Result

PCT WBC CRP ESR

AUC (95% CI) 0.729 (0.661–0.792) 0.567 (0.483-0.649) 0.622 (0.460–0.796) 0.685 (0.531–0.823)

a fever and a history of ALL. We found a significant difference in readily generalized to other patient populations. In addition, in our
PCT values between children with a positive BC result compared ED during the study time, it was standard to obtain a PCT value
with those with a negative BC result. with BC; it was also standard to obtain a WBC at this time as well.
Procalcitonin levels were significantly higher in patients The obtainment of CRP and ESR was at the discretion of the
with fever and ALL in the population of those with a positive ordering provider. Therefore, although there is valuable informa-
BC result compared with those with a negative culture result tion comparing all 4 biomarkers of infection, the populations are
(6 vs 0.8 ng/mL, respectively). A study completed by Reitman not directly comparable.
et al16 showed PCT to be predictive of bacterial infection on sub- In conclusion, PCT value was significantly associated with
sequent measurements in febrile neutropenic oncology patients. positive BC in this pediatric population of febrile children with
In our study, PCT seemed to be an effective biomarker at ini- ALL, and the diagnostic performance of PCTwas better than other
tial presentation in predicting bacterial infection in pediatric markers of inflammation that were measured. Further studies
ALL patients.
The diagnostic performance of PCT was better than other
markers of inflammation. Our results showed PCT to be superior
to ESR, CRP, and WBC. These findings are consistent with that of
Hatzistilianou et al,11 who evaluated PCT, CRP, TNF-α, IL-1b,
IL-8, and soluble TNF receptor II. These biomarkers were eval-
uated to see which was best in the early and rapid diagnosis
of infection in neutropenic children with acute ALL. The result
of this retrospective chart review was that a highly statistical
difference was described between PCT levels in bacterial and
nonbacterial episodes.
This study showed promise in the use of PCT in evaluating
pediatric patients with ALL, but largely differed from our study
because this study assessed the value of many inflammatory
markers and cytokines as markers of early infection in children
with neutropenia and ALL. The focus of our study was to solely
assess the value of PCT as an early marker of bacteremia in chil-
dren with ALL, regardless of neutrophil count. Our study focused
solely on febrile children with ALL who presented to the ED and
elucidated whether there was statistical significance in PCT level
and eventual diagnosis of bacteremia.11 This has the potential
to lead to a decrease in unwarranted use of antibiotics, hospital
length of stay, and health care expenditures.
The limitations of this study include a relatively small subset FIGURE 1. Area Under the Receiver Operating Characteristic
of patients, because febrile pediatric patients with ALL cannot be Curve of PCT Versus Other Biomarkers of Infection.

© 2016 Wolters Kluwer Health, Inc. All rights reserved. www.pec-online.com 3

Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
Vyles et al Pediatric Emergency Care • Volume 00, Number 00, Month 2016

TABLE 5. Properties of Diagnostic Tests for Blood Culture

Procalcitonin, ng/ml PCT

Sensitivity Specificity PPV NPV
≥0.5 0.592 0.748 0.213 0.941
(0.474, 0.711) (0.698, 0.793) (0.151, 0.282) (0.917, 0.963)
≥1.0 0.474 0.847 0.263 0.933
(0.346, 0.603) (0.803, 0.882) (0.173, 0.357) (0.909, 0.954)
≥2.0 0.263 0.935 0.317 0.917
(0.159, 0.378) (0.911, 0.957) (0.196, 0.46) (0.891, 0.942)

should investigate generalizing significant PCT values as predic-
tors of disease in other patient populations and disease processes.
REFERENCES
1. Kasem AJ, Bulloch B, Henry M, et al. Procalcitonin as a marker of
bacteremia in children with fever and a central venous catheter presenting
to the emergency department. Pediatr Emerg Care. 2012;28:1017–1021.
2. Gendrel D, Bohuon C. Procalcitonin as a marker of bacterial infection.
Pediatr Infect Dis J. 2000;19:679–687.
3. Van Rossum AM, Wulkan RW, Oudesluys-Murphy AM. Procalcitonin as
an early marker of infection in neonates and children. Lancet Infect Dis.
2004;4:620–630.
4. Fernández Lopez A, Luaces Cubells C, García García, JJ, et al.
Procalcitonin in pediatric emergency departments for the early diagnosis
of invasive bacterial infections in febrile infants: results of a multicenter
study and utility of a rapid qualitative test for this marker. Pediatr Infect Dis
J. 2003;22:895–903.
5. Ahn S, Lee YS, Lim KS, et al. Adding procalcitonin to the MASCC
risk-index score could improve risk stratification of patients with febrile
neutropenia. Support Care Cancer. 2013;21:2303–2308.
6. Assicot M, Gendrel D, Carsin H, et al. High serum procalcitonin
concentrations in patients with sepsis and infection. Lancet. 1993;341:
515–518.
7. Charles PE, Ladoire S, Aho S, et al. Serum procalcitonin elevation in
critically ill patients at the onset of bacteremia caused by either Gram
negative or Gram positive bacteria. BMC Infect Dis. 2008;8:38.
8. Koivula I, Hämäläinen S, Jantunen E, et al. Elevated procalcitonin
predicts Gram-negative sepsis in haematological patients with febrile
neutropenia. Scand J Infect Dis. 2011;43:471–478.
9. Müller B, Becker KL, Schächinger H. Calcitonin precursors are reliable
markers of sepsis in a medical intensive care unit. Crit Care Med.
2000;28:977–983.
10. Fleischhack G, Kambeck I, Cipic D, et al. Procalcitonin in paediatric
cancer patients: its diagnostic relevance is superior to that of C-reactive
protein, interleukin 6, interleukin 8, soluble interleukin 2 receptor and
soluble tumour necrosis factor receptor II. Br J Haematol. 2000;111:
1093–1102.
11. Hatzistilianou M, Rekliti A, Athanassiadou F, et al. Procalcitonin as an
early marker of bacterial infection in neutropenic febrile children with acute
lymphoblastic leukemia. Inflamm Res. 2010;59:339–347.
12. Hatzistilianou M, Rekleity A, Athanassiadou F, et al. Serial procalcitonin
responses in infection of children with secondary immunodeficiency.
Clin Invest Med. 2007;30:E75–E85.
13. Hitoglou-Hatzi S, Hatzistilianou M, Gougoustamou D, et al. Serum
adenosine deaminase and procalcitonin concentrations in neutropenic
febrile children with acute lymphoblastic leukaemia. Clil Ex Med. 200;5:
60–65.
14. Miedema KG, de Bont ES, Elferink RF, et al. The diagnostic
value of CRP, IL-8, PCT, and sTREM-1 in the detection of bacterial
infections in pediatric oncology patients with febrile neutropenia.
Support Care Cancer. 2011;19:1593–1600.
15. Phillips RS, Wade R, Lehrnbecher T, et al. Systematic review and
meta-analysis of the value of initial biomarkers in predicting adverse
outcome in febrile neutropenic episodes in children and young people with
cancer. BMC Med. 2012;10:6.
16. Reitman AJ, Pisk RM, Gates JV 3rd, et al. Serial procalcitonin levels to
detect bacteremia in febrile neutropenia. Clin Pediatr (Phila). 2012;51:
1175–1183.

4 www.pec-online.com © 2016 Wolters Kluwer Health, Inc. All rights reserved.

Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.