The prognostic value of serum procalcitonin measurements in critically injured patients: a systematic review

A systematic literature search was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Group [28] using the following databases from their inception to September 2018: PubMed, Ovid MEDLINE, EMBASE, and the Cochrane Library (see the completed PRISMA checklist in Additional file 1). Medical Subject Heading terms and keywords for procalcitonin, trauma, injury, prognosis, and predictive value were used. The search was limited to original studies on human subjects, published in the English language. Our complete search strategies are shown in Additional file 2. To identify other potentially relevant articles, the PubMed “related articles” feature was utilized and the two authors independently hand-searched the reference list of included articles and relevant reviews for additional citations.

Two reviewers (A.N.AR. and F.A.AH.) independently screened the titles and abstracts of all identified citations for potential eligibility. The following inclusion criteria were applied: (1) study participants (adults ≥ 16 years old trauma patients), (2) intervention (single or serial measurements of serum procalcitonin level from day of admission to trauma center or intensive care unit (ICU) following trauma), (3) comparison of prognostic performance of PCT levels compared to other potential biomarkers, (4) outcome (documentation of at least one of the outcomes of sepsis, organ dysfunction, mortality, or correlation of PCT with severity of injury), and (5) cohort or case-control study design.

Reviews, case reports, letters, conference abstracts, and editorials were excluded. Articles involving isolated injury to the central nervous system and pediatric or burn trauma were also excluded. Full-text articles of identified abstracts that were relevant with the inclusion criteria were assessed for final eligibility. The agreement between the two reviewers was assessed using Kappa statistic for the inter-rater reliability [29].

Two reviewers (A.N.AR. and F.A.AH) independently extracted data using a standardized recording tool to record the study design and setting, year of publication, country of origin, number of study participants, participant clinical characteristics, PCT testing system, kinetics of PCT, markers other than PCT, and study outcomes.

We defined the terms “SIRS” and “Sepsis” according to the American College of Chest Physician/Society for Critical Care Medicine [30], which supported definition of SIRS, sepsis, and severe sepsis as most studies were published prior to the release of Sepsis-3 [31].

The risk of bias of the included studies was assessed by two independent investigators (A.N.AR. and F.A.AH) using the Quality in Prognosis Studies (QUIPS) tool developed by Hayden et al. [32]. This tool consists of 30 criteria divided into six domains: patient selection, study attrition, prognostic factor measurement, outcome measurement, confounding measurement and account, and statistical analysis and reporting. Each criterion was scored using “yes,” “no,” or “unclear.” This scoring led to the overall judgment of “low,” “moderate,” or “high” risk of bias per domain. We considered a study to be of high quality when the bias was rated as low or moderate with respect to almost all of the domains. Conversely, a study was considered to be of low quality when the bias was rated high in most of the bias domains. Disagreements were resolved by consensus.

Figure 1 presents a flow diagram of study identification and subsequent inclusion. Among 2015 citations identified by the search, 19 studies fit the inclusion criteria including 4146 patients with multiple trauma. There was an excellent inter-investigator agreement on the inclusion of full-text articles in the systematic review (κ statistic = 0.88; 95% CI = 0.65–1.00).

Due to heterogeneity between studies with respect to definition of clinical outcomes and statistical methods, it was not possible to statistically pool the results. Instead, findings were reported descriptively.

Table 1 details the characteristics of the included studies. The included studies were published from 1998 to 2016, while the majority were published after 2006 (13/19; 68.4%). Sixteen studies were conducted in Europe [34, 37‐51] (84.2%), two were conducted in Asia [33, 35] (10.5%), and one was conducted in the USA [36] (5.2%). All studies were observational and non-interventional. Eleven studies were prospective cohort [34, 36‐39, 41‐43, 45, 46, 51], six studies were prospective case-control [33, 35, 44, 47‐49], one study was retrospective cohort study [40], and one study was retrospective case-control [50].

All studies evaluated the role of PCT in predicting at least one clinical outcome in critically injured patients (Fig. 2). Twelve studies (63.3%) were conducted in the critical care units [35‐37, 39, 40, 42‐45, 48, 50, 51]. Overall, the aggregate study population included a total of 4146 patients with trauma in whom serum PCT levels were used to predict posttraumatic complications. The mean age ranged between 34 and 49 years. The mean injury severity score ranged from 21 to 32. The mechanism of injury was blunt and/or penetrating trauma. All studies were civilian, with none comprising military populations or combat injuries.

Figure 3 summarizes the risk of bias assessment for all included studies. Most studies were assessed to be of low to moderate risk of bias. Nine studies demonstrated high risk of bias in at least one domain [33‐35, 37, 39, 41, 42, 44, 46]. Study confounding domain was deemed to be at moderate to high risk of bias in 14 studies (73.7%) [33‐35, 37‐46, 51]. The majority did not account for potential confounding factors in the study designs and/or made adjustment for the effects of the confounders in the analysis, while 21.1% of these studies did not name any confounder [35, 43, 44, 46]. The risk of bias was moderate to high in the domain of statistical analysis in 10 studies (52.6%) [33‐35, 37, 39, 41, 42, 44, 46, 51]. Only 7 studies used statistical models to assess prognostic relationships [36, 37, 40‐42, 45, 47]. In one study, mortality data were presented in graph only [34].

PCT levels were measured from the serum sample in all studies, using immunoluminometric assay (LUMItest) in 11 studies (57.8%) [38, 40, 42‐45, 47‐51] and Kryptor Assay in 3 studies (15.7%) [36, 37, 41]. Other techniques used to determine PCT levels were Roche Cobas e 411 [33], VIDAS system [35], enzyme-linked fluorescent immunoassay (ELFA) [34], and chemiluminescence analyzer [39]. Most studies showed rapid kinetics of PCT levels with peak levels reached on day 1 post trauma [33, 36, 38‐43, 45, 47‐51], and to a lesser extent on day 2 [34, 37]. PCT levels declined rapidly thereafter towards the normal range. Sakran et al. [36] and Haasper et al. [37] demonstrated that a biphasic rise in PCT after day 7 was associated with development of sepsis.

Nine studies (47.4%) [33, 39, 41, 42, 45‐47, 50, 51] assessed the correlation between initial PCT level and the severity of injury using Injury Severity Score (ISS) [52]. All studies showed a correlation between initial PCT levels and ISS. When patients were categorized into those with severe trauma (ISS > 20) or moderate trauma (ISS < 20), the initial PCT was significantly higher in patients with severe trauma [41, 42, 45, 50].

Four studies assessed the association between PCT level and injury pattern, three of which showed that serum levels of PCT were higher among patients with abdominal injury [35, 41, 45], whereas one study showed no correlation between PCT level and injury pattern [50].

Sixteen studies (84.2%) assessed the utility of serum PCT level as a marker for sepsis [33, 35‐40, 42‐50]. In patients who developed systemic or septic complications, the kinetic of PCT was similar to those without complications. After reaching the peak level on day 1 after trauma, an immediate decline was observed towards normal range [36, 38, 40, 45, 50]. However, the initial peak PCT was significantly higher in patients who subsequently developed sepsis compared to those without sepsis and the difference remained significant between the two groups during the study period [33, 36, 38‐41, 45, 46, 49, 50]. Further, patients who developed sepsis demonstrated a significant increase of peak PCT levels compared with patients with non-infectious systemic inflammation [36, 42, 43, 50]. Because the severity of injury itself can significantly influence circulating PCT levels, Wanner et al. demonstrated a 3.9-fold increase of initial peak PCT levels in injured patients with sepsis compared with patients with SIRS [50]. In addition, in a subgroup analysis of injured patients with low ISS (ISS < 25) compared with patients with high ISS (ISS ≥ 25), PCT levels remained significantly elevated in patients with sepsis compared with patients without sepsis in those subgroups [50]. Castelli et al. observed a significant increase in PCT level in trauma patients at day of sepsis diagnosis as compared with levels measured on day 1 before the diagnosis [39]. Rajkumari et al. reported that all patients with serum PCT level > 2 ng/ml developed infections after trauma [35]. Three studies measured the sensitivity and specificity for the diagnosis of sepsis as compared with non-infectious systemic inflammation. With cutoff point of procalcitonin level of 1.5 ng/mL, the sensitivity to detect sepsis ranged from 42 to 76%. In those same studies, the specificity ranged from 73 to 77% [45, 47, 50].

The most common identified sources of infection resulting in sepsis were the lungs [36, 39, 40, 42, 43, 45, 48, 50], bloodstream [36, 39, 40, 42, 43, 45], urinary tract infection [36, 39, 40, 45, 48], soft tissue infection [39, 42, 43, 45, 48, 50], peritonitis [39, 43, 45, 50], and bacterial meningitis [43, 50]. However, four studies reported no significant difference in the PCT level between patients who developed sepsis and those who did not [37, 44, 47, 48], although one study evaluated the predictive value of PCT level for the development of posttraumatic pneumonia and reported higher PCT levels among patients who developed pneumonia compared with those without, but the difference did not reach significance [44].

Nine studies (47.4%) assessed the utility of serum PCT level in predicting the future occurrence of MOD and/or MOF in injured patients [35, 37, 39, 43, 45, 47, 49‐51]. Studies used either the Sequential Organ Failure Assessment (SOFA) [53] or the Multiple Organ Failure Score developed by Goris to assess the severity of organ dysfunction [54, 55]. Seven studies demonstrated significantly higher initial PCT levels in patients who subsequently developed MODS compared to those patients without MODS [37, 39, 43, 47, 49‐51]. Haasper et al. observed an elevated PCT level preceding the development of MODS by 3 days [37]. Hensler et al. divided patients with MOF into early and late MOF using day 3 as a cutoff point to avoid biased results by patients who developed MOF already on day 1. Results remained significant for both groups when compared with patients without MOF [47]. However, two studies showed no correlation between PCT levels and severity of organ dysfunction [35, 45].

Among seven studies assessing the value of procalcitonin serum level determination in predicting death following trauma, three studies demonstrated significant elevation of PCT levels in non-survivors compared with that in survivors [34, 42, 45]. The difference in PCT level between the two groups remained significant during the first week after trauma in two of those studies [42, 45]. Meisner et al. showed that by the end of the first week, the difference between survivors and non-survivors increased up to 15-fold in patients with fatal outcome [45]. A study by Sakran et al. demonstrated significantly increased mortality among patients with a PCT level of ≥ 5 ng/ml compared with that of patients with less than 5 ng/ml, with odds ratio of 3.65 (95% CI 1.23–4.61, p = 0.01) [36]. Conversely, two studies reported no association between PCT level and fatal outcome [47, 50].

Fourteen studies (73.7%) assessed serum PCT levels and other biomarkers simultaneously to predict posttraumatic complications (Table 2). Five studies demonstrated slow induction of C-reactive protein (CRP) after trauma with peak levels reaching on day 3 after trauma [38‐40, 43, 45]. PCT was superior to CRP in predicting septic complications in these studies.

Interleukin 6 (IL-6) has similar kinetics to those of PCT. IL-6 increases during the early phase after trauma with peak levels reached on day 0 after trauma [37, 38, 40, 49]. While Billeter et al. observed a significant difference in IL-6 levels between patients who developed sepsis and those who did not on days 3 and 5 after trauma [40], Keel et al. observed this difference only after day 5 [38].

Haasper et al. demonstrated low sensitivity of IL-6 for predicting sepsis and low specificity for predicting both sepsis and MODS [37].