The impact of procalcitonin to C-reactive protein ratio (PCR) kinetics as a predictor of mortality in nosocomial bloodstream infections

Background: This study was performed to investigate the combination of serum C-reactive protein (CRP) and procalcitonin (PCT) kinetics as a best marker in predicting mortality in patients with nosocomial blood stream infection (BSI). Methods: We retrospectively reviewed the medical records of patients ≥ 18 years of age with nosocomial BSIs hospitalized in intensive care units during the period from January 2016 to June 2018. Eighty-four patients who met the inclusion criteria were included in the study. Clinical, microbiological and biochemical data were compared in patients who survivors and deaths. Binary logistic regression analyses (backward LR) were used to identify independent risk factors. A receiver operating characteristic (ROC) curve analysis was performed to compare the predictive accuracy. The kinetic changes were expressed as Δ (delta) and dened the as difference between level on day 5 and level at day 1 of BSI. Results: Of the 84 included patients, 46 (58.4%) had survivors and 35 (41.6%) had deaths. In univariate analysis, renal disease (p = 0.007), cardiac disease (p = 0.042), septic shock (p = <0.001), maximum SOFA (p = <0.001) and APACHE-II (p <0.001), ΔCRP (p = 0.004), ΔPCT (p = <0.001), and ΔPCR (p = 0.025) were signicantly higher in non-survivors than in survivors. In the logistic regression analysis, APACHE-II score (odds ratio (OR) = 1.46, 95% condence interval (CI) = 1.20-1.78, p <0.001), ΔCRP (OR = 1.18, %95 CI =1.04-1.34, p = 0.009), ΔPCT (OR = 0.87, 95% CI = 0.79-0.95, p = 0.001), and ∆ PCR (OR = 36.78, 95% CI = 4.52-299.01, p = 0.001) were independent predictors of 28-day mortality. After a ROC analysis, the AUC of ∆ PCR


Background
The structure of nosocomial infections is dynamic and complex and often involves a large number of variables that adversely affect clinical outcomes [1]. Bloodstream infections (BSIs) in hospitalized patients are the leading cause of morbidity and mortality, and their early diagnosis and appropriate treatment is critical for positive outcome [2][3][4].
The in ammatory response is the main determinant of the unfavorable long-term consequences of sepsis. Biomarkers that can be used as an independent prognostic factor in assessing mortality should be objectively measured to re ect in ammatory processes and therapeutic responses [5]. These, together with their positive clinical results, also contribute signi cantly to health care costs. However, improper test requests and interpretations can also lead to unnecessary administration and procedures to patients [6].
Biomarkers need to provide a fast, common, and reliable method of identi cation. Existing biomarkers cannot fully re ect these features [7]. They do not replace a comprehensive clinical evaluation, but they increase the support data, the clinician must make rational decisions [6]. Page 3/16 The level of C-reactive protein (CRP) increases signi cantly in response to infection. However, it is also increasing in all non-speci c, in ammatory disorders. CRP provides only limited information on infection status and is far from a perfect test for diagnosing critical diseases [7][8][9]. CRP can be detected in the blood within 4 to 6 h after bacterial challenge and peaks at 36 to 50 h [9]. Procalcitonin (PCT) is an indicator of systemic in ammation and can help predict bacteremia, sepsis and mortality [10][11][12][13][14]. PCT concentrations rise rapidly, 2 to 3 h after bacterial infections, and PCT has a shorter half-life than that of CRP [9]. It differs better between CRP between infectious and non-infectious causes of critical disease [7].
PCT is useful as a parameter for predicting early systemic bacterial infections. PCT measurements in series can help predict mortality in sepsis and reduce antibiotic exposure [6,9]. Albumin (ALB) tends to decrease in levels during acute infections and is a strong indicator of outcome in infectious diseases. The relationship between low serum albumin level and high mortality has been reported in several studies [15,16].
Recently, the effectiveness of various biomarkers such as CRP, PCT, albumin, which can be used in clinical practice is investigated on sepsis and mortality. It has been stated that PCT to ALB ratio (PAR) is an early diagnostic factor in predicting sepsis, patients with high rates are more prone to septic shock [17]. CRP to ALB ratio (CAR) has been reported to correlate with hospital mortality, sepsis, and poor prognosis in patients with cancer [5,18,19]. There were two reports on the usefulness of PCT to CRP ratio (PCR) in various infectious conditions of adult patients [20,21]. It has been shown that the PCR can help differentiate proven sepsis from suspected sepsis when used with PCT in late onset neonatal sepsis [22].
In patients diagnosed with nosocomial BSI, a simple, fast and accessible parameter is needed to predict treatment response and mortality. In this study, we aimed to determine a more valuable predictor by comparing various in ammatory variables in predicting treatment response and mortality in BSIs.

Patients and Study Design
This retrospective study was conducted in Bezmialem Vakif University Hospital which has 28-bed adult intensive care units (ICU), totally 550 beds. Adult patients aged ≥ 18 years in ICU diagnosed with BSIs during the period from January 2016 to June 2018 were included in the study. Patients with recurrent BSIs with the same microorganism, insu cient data, and transferred to another hospital or ICU were excluded.

Data Collection
Daily BSI surveillance have been done at our hospital by visiting patients and checking their laboratory results by the infection control team. Basic characteristics, medical history, comorbidities, sepsis/septic shock, microorganisms de ned in hemoculture, laboratory parameters (including PCT and CRP), and antimicrobial treatment have been recorded to surveillance forms. Maximum SOFA (sequential organ failure assessment) and APACHE-II (Acute Physiology and Chronic Health Evaluation-II) scores were calculated according to patients' digital records. In patients with suspected BSI, CRP and PCT levels are routinely checked daily or every 2 days. PCR was calculated based on routine laboratory parameters involving CRP (normal range values 0-0.5 mg/dL) and PCT (normal range values 0-0.05 ng/ml).

De nition
BSIs was diagnosed and recorded during daily active surveillance according to Centers for Disease Control and Prevention criteria [23]. Sepsis and septic shock diagnosis were made according to recent guidelines [24]. Primer outcome was determined as 28-day mortality rates. We evaluated the predictive ability of CRP, PCT, and their combination PCR. PCT, CRP, and PCR kinetic changes were expressed as ΔPCT, ΔCRP, and ΔPCR. For example, ΔPCT de ned the difference between PCT level on day 5 and PCT level at day 1 (basal or rst day of BSI). Values < 0 indicated decreasing PCT concentrations. CRP or PCR kinetics were also expressed in the same way. [25].

Statistical analysis
Categorical variables were presented as absolute frequency and percentage, and continuous variables were presented as mean, standard deviation (SD), Standard Error of Mean (SEM). The clinical characteristics of the two groups, survival and death, were compared using the Pearson's chi-squared test or Fisher's exact test for categorical variables, and the Student's-t independent test for continuous variables. Paired Student's t-test was used to compare the measurements at two time points (between the 5th day and 1st day) for CRP, PCT, and PCR. For the logistic regression analysis, univariate analysis was performed rst, and p values of less than 0.05 were considered statistically signi cant. Additionally, binary logistic regression ("backwards: LR" method) was used. Analysis of the receiver operating characteristic (ROC) curves and the area under the curves (AUCs) were performed to evaluate the PCR as predictive values for mortality in critically ill patients. In addition, we compared the ROC curves between the ΔCRP, ΔPCT and ΔPCR. All statistical analyses were performed with IBM SPSS software version 22.0 (IBM, Armonk, NY, USA).

Ethics
The study protocol was approved by the Ethics Committee of the Hospital of Bezmialem Vakif University.
The informed consent form was not needed because of the retrospective nature of the study.

Results
A total of 104 patients with nosocomial BSIs were enrolled to study in the ICU. We excluded 20 patients with transferred to another hospital or insu cient medical data (Fig. 1).

Discussion
In our study, we evaluated the kinetics of in ammatory variables including CRP and PCT in terms of their prognostic power in predicting mortality in patients with nosocomial BSI. The main nding of the study is that ∆PCR is signi cantly more effective in predicting 28-day mortality than ∆CRP and ∆PCT alone. There are several studies investigating the predictive value of in ammatory indicators such as CRP and PCT in mortality in critically ill patients [5,8,26,27]. According to our knowledge, this is the rst article reporting the association between ∆PCR and hospital mortality in patients with nosocomial BSI. Our data suggest that combined assessment of ΔPCT and ΔCRP could increase the predictive value of these parameters. For ∆PCR, it was found that the intersection point of -0.05 is predicting the 28 days mortality by ROC curve. The results of this study make contribution to the literature on the prognostic role of ∆PCR on mortality in critically ill patients.
Timely diagnosis and treatment are required to reduce bacteremia-related morbidity and mortality [9]. Serum PCT and CRP have good clinical diagnosis and prognostic value for patients with sepsis and septic shock. Kinetic studies of PCT and CRP may increase sensitivity and accuracy when evaluating the prognosis of patients with sepsis and septic shock [28]. They have been used for early detection of infection and directing antibiotic therapy [8]. PCT level was higher in patients with nosocomial BSI and was more useful for predicting nosocomial BSI than CRP or white blood cell count [4]. PCT change faster than CRP in response to bacterial infection, and proper antibiotic therapy is associated with a rapid decrease in PCT levels [25]. A systemic review and meta-analysis study showed that PCT levels were signi cantly different between surviving and non-surviving sepsis patients [29]. The prognostic value of CRP and PCT kinetics has been studied in several types of infection, with mortality as the main outcome variable. Initial CRP and PCT cannot be considered useful markers in patients with acute and chronic conditions [8]. In recent studies, different results have been observed on the relationship of PCT and CRP levels with mortality [30,31]. CRP and PCT kinetics in the rst days were associated with use of appropriate (active) empirical therapy [26,32]. A large multicenter prospective study showed that the decrease in PCT levels by more than %80 was a predictor of mortality and can be used in managing sepsis [26]. In another prospective study, initial levels of CRP and PCT and their combinations in patients with sepsis, were found to have limited value in predicting 28 days mortality [8]. In our study, we did not nd basal (day 1 of BSI) levels of CRP, PCT and PCR signi cant at 28-day mortality. However, we found the kinetics that re ect the differences between the 5th and 1st days to predict mortality. In particular, PCR kinetics was remarkable in terms of prediction (OR = 36.8). It is thought that PCR kinetics are thought to be superior in predicting mortality due to PCT levels rises and falls earlier than CRP and CRP are affected more by non-infectious diseases than PCT.
Recently, there were reports on the bene t of the PCR in various infectious conditions of adult patients. In a study on the adult pneumonia, log PCR was signi cantly higher in patients with Legionella pneumonia compared with pneumococcal pneumonia concluding that PCR resulted in excellent discrimination between infections caused by these two pathogens. It was because increase of CRP was more signi cant than PCT in Legionella pneumonia patients. [20]. Similarly, different increasing patterns of CRP and PCT were found in suspected sepsis group and the proven sepsis group in the present study, which was re ected in the decreased log PCR as well [22]. In addition, Hangai et al. reported that PCR showed best performance in discrimination between tumor fever and infection induced fever in patients with hematological diseases (CRP AUC 0.67, PCT AUC 0.70, and PCR AUC 0.75). These results implicate that PCT and CRP differently respond to various infectious conditions, which makes PCR as a potential tool in differentiation of various infectious diseases [21]. In other study, PCR showed highest AUC compared with CRP and PCT in discrimination of proven sepsis and suspected sepsis in neonates even though there was no statistical difference [22]. In our study, ROC analysis showed that PCT was more useful than CRP in terms of the AUC. Furthermore, ∆PCR was the best test to discriminate between surviving and non-surviving, showing the highest AUC.
Although, there are many scoring systems used to estimate the prognosis of critically ill patients, the ∆PCR is valuable because it is relatively simple and easy to use in all settings. However, there are several limitations of this study. Firstly, it is a retrospective single center study done on a selected population. There may have been some selection bias because of the study's retrospective design. The patient population might be heterogenous because of various hospitalization settings, including intensive care units. Secondly, PCT and CRP results were reported routinely at our hospital so our clinicians were not blind to the results. This can cause bias and effect the predictive value of PCR. Thirdly, we couldn't investigate the associations of general mortality because cases without BSIs and nosocomial infections other than BSIs were excluded. Nonetheless, PCR was not used in follow up or making up critical decisions for patients.

Conclusions
This study indicates that higher PCR kinetics was associated with higher mortality rate in ICU patients with nosocomial BSI. We suggest to use ∆PCR as an auxiliary test in the follow-up of these patients. But the sensitivity and speci city rate of ∆PCR for predicting mortality was not at the ideal height. So, larger multicentral studies are needed to verify this result.   ROC curves of ∆PCR, ∆PCT and ∆CRP as 28-day mortality predictors.