Hematological parameters in patients with bloodstream infection: A retrospective observational study

Introduction: To date, the relationship between the causative pathogens and the changes of hematological parameters was rarely referred and deserves further investigation. Methodology: A total of 825 adult patients, including 134 negative blood cultures patients and 691 bloodstream infection (BSI) patients, were screened for eligibility in this study. Receiver operating characteristic curves and binary logistic regression models were used to assess the power of hematological parameters to distinguish patients with BSI caused by different pathogens. Results: Except for platelet-to-lymphocyte ratio (PLR) and platelet larger cell count (P-LCC), the other hematological parameters investigated in the study were significantly different in patients with BSI caused by different pathogens, including Candida. The specific combinations of lymphocyte count (LYM), platelet count (PLT), neutrophil-to-lymphocyte ratio (NLR), mean platelet volume (MPV), MPV-to-PLT ratio (MPV/PLT), platelet larger cell ratio (P-LCR), and C-reactive protein (CRP) can improve the ability to distinguish various BSI from negative blood cultures. The highest area under the curve of was 0.753 (95% CI 0.709-0.797) for positive blood cultures, 0.715 (95% CI 0.658-0.771) for Gram-positive pathogens BSI, 0.777 (95% CI 0.730-0.824) for Gram-negative pathogens BSI, 0.797 (95% CI 0.747-0.846) for Escherichia coli BSI, 0.943 (95% CI 0.899-0.987) for Enterobacter aerogenes BSI, 0.830 (95% CI 0.740-0.921) for Pseudomonas aeruginosa BSI, and 0.767 (95% CI 0.695-0.839) for Staphylococcus aureus BSI. Conclusions: The specific combinations of hematological parameters can improve the power to distinguish patients with BSI caused by different pathogens. Attention to these parameters can be easily integrated into daily medical activities, without extra costs.


Introduction
Blood is usually sterile. However, once enter the blood, pathogens can interact with neutrophils, lymphocytes, monocytes, and platelets, leading to bacteremia and sepsis [1], known as bloodstream infection (BSI). Of course, this process also involves humoral immunity, cellular immunity, and endothelial activation mediated by antibodies, complement proteins, and interleukins [2]. Venous access devices and ports often increase the risk of BSI, especially in hematological and tumor patients in intensive care unit [3]. The mortality rate of BSI is about 25%, and up to 54% when developed to severe sepsis [4]. Currently, the most common pathogen is Staphylococcus, followed by Enterobacteriaceae, Pseudomonas, Enterococcus, Acinetobacter, and Candida [5,6].
Generally speaking, except for Salmonella, in a bacterial infection, the neutrophil count (NEU) will increase. By contrast, persistent lymphocytopenia was described in patients with Gram-positive and Gramnegative bacteremia [7]. Nevertheless, the changes of platelet count (PLT) and other platelet-related parameters in pathogens infection, particularly, BSI caused by different pathogens, are still ambiguity [8][9][10][11]. Platelets are classically known for their critical role in thrombosis, hemostasis, and wound repair. However, emerging evidence indicates that platelets also have a complex role in tumor growth, autoimmune disease, inflammation, and infection [12]. As the first cell at the site of injury, platelets act as primitive immune cells by interacting with invading pathogens, inducing platelet activation. Once activated, platelets release preformed granules that contain diverse bioactive molecules, including cytokines, chemokines [13], and antimicrobial molecules [14]. Besides, activated platelets can also promote the activation of dendritic cells and monocytes, enhancing adaptive immune responses [15]. Moreover, during BSI, platelets detect pathogens via their Toll-like receptors [16], glomming onto neutrophils. In response, the neutrophils release neutrophil extracellular traps (NETs), webs of DNA, ensnaring pathogens [17,18]. Platelets are needed for the recruitment of neutrophils to tissues of inflammation and infection [19]. Additionally, pathogens that enter the bloodstream will produce diverse extracellular proteins and toxins, resulting in platelet activation or inhibition of platelet activation [20]. Taken together, it is apparent that the interaction between platelets and pathogens not only has important consequences for the pathophysiological response to pathogenic infection but also affect PLT and other platelet-related parameters.
To date, the relationship between the causative pathogens and the changes of hematological parameters was rarely referred and deserves further investigation. Given the protective role of neutrophils, lymphocytes, and platelets in immunity against pathogens, in this retrospective observational study, we sought to systematically investigate the changes of hematological parameters and their differential ability in patients with BSI caused by different pathogens.

Study design and data collection
This retrospective observational study was conducted using laboratory and clinical data collected from the microbial information system, laboratory information system, and laboratory electronic medical record system of the Second Hospital of Anhui Medical University, a 2,200-bed tertiary teaching hospital in Hefei (8.0 million inhabitants), Anhui province, China. Since the vast majority of our daily blood cultures were negative, to balance the sample size between groups, we collected data between February 2011 and June 2019 for positive blood cultures and May 2019 to June 2019 (randomly selected) for negative blood cultures.

Instrument and reagent
Serum CRP concentration was detected by Dimension EXL with LM automatic biochemistry analyzer (Siemens Healthcare Diagnostics, Newark, DE). Hematological parameters, such as NEU, lymphocyte count (LYM), PLT, mean platelet volume (MPV), platelet distribution width (PDW), and platelet larger cell ratio (P-LCR), were determined on Sysmex XE2100 hematology analyzer (Sysmex Corporation, Kobe). Subsequently, plateletcrit (PCT), neutrophil-tolymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), mean platelet volume-to-platelet count ratio (MPV/PLT), and platelet larger cell count (P-LCC) were calculated. Blood cultures were incubated in BACTEC FX automatic blood culture system (Becton Dickinson, Sparks, MD). The suspected positive bottles were removed and subjected to culturing tests by using Colombian blood agar plate, chocolate agar plate, and anaerobe 5% sheep blood agar plate [10]. Identification of microorganisms was performed with VITEKII Compact system (BioMérieux, Marcy L'Etoile) and matrix-assisted laser desorption/ionization time-offlight Microflex LT mass spectrometer (Bruker Daltonics, Hamburg). When no bacterial growth was detected for five days, the result was considered negative.

Statistical analysis
Statistical analyses were performed using SPSS version 19.0 (SPSS Inc., Chicago, IL). Firstly, all variables were tested for normal distribution by the Kolmogorov-Smirnov test. In accordance with the result of this test, the statistical significance of differences was tested using the Student's t-test or Mann-Whitney U test (two groups' comparison). In the case of multigroup comparison, One-Way ANOVA (LSD) or nonparametric (Kruskal-Wallis) test was applied. Through the above steps, we examined the relationship between all independent variables and dependent variable, consequently, some independent variables that may be meaningless were filtered out. Then, binary logistic regression analysis was conducted. Odds ratios (OR) and 95% CI were calculated to determine the strength of the association between hematological parameters and BSI pathogens. Besides, receiver operating characteristic (ROC) curves were constructed to investigate area under the curve (AUC), 95% confidence intervals (CI), sensitivity, specificity, and cut-off value of hematological parameters [21]. Continuous variables were reported as mean values ± standard deviation (SD) or median with interquartile range (IQR), while categorical variables were expressed as count and percentage. Two-sided p < 0.05 was considered to represent a statistically significant difference.

Ethics statement
Ethical approval was not required as all data used in this study were acquired retrospectively from the microbial information system, laboratory information system, and laboratory electronic medical record system.

Demographic and laboratory characteristics of study subjects
A total of 825 patients were included in the study. Blood cultures were positive in 691 patients. The demographic characteristics of all patients are summarized in Table 1. As for gender distribution, there was no significant difference between positive and negative blood cultures. Nevertheless, patients with positive blood cultures were older than those with negative blood cultures (p = 0.001). Furthermore,  Data are shown as number (%), mean (standard deviation, SD), or median (interquartile range, IQR) as appropriate. BSI: bloodstream infection; CRP: C-reactive protein; LYM: lymphocyte count; MPV: mean platelet volume; MPV/PLT: mean platelet volume-to-platelet count ratio; NEU: neutrophil count; NLR: neutrophil-to-lymphocyte ratio; PCT: plateletcrit; PDW: platelet distribution width; PLT: platelet count; PLR: platelet-to-lymphocyte ratio; P-LCC: platelet larger cell count; P-LCR: platelet larger cell ratio.
patients with isolated Gram-negative pathogens were older than that with Gram-positive pathogens (p < 0.001) and negative blood culture (p < 0.001) ( Table 2). Moreover, patients with E. coli BSI were older than those with S. aureus BSI (p < 0.001) and negative blood culture (p < 0.001) ( Table 3).

Changes of hematological parameters in negative, Gram-positive, Gram-negative, and Candida blood cultures
Similarly, other than PLR (p = 0.158) and P-LCC (p = 0.075), the other hematological parameters were significantly different among the four groups ( Table 2) Table 4 for more details.

Changes of hematological parameters in negative blood cultures and different pathogens BSI
Likewise, other than PLR (p = 0.178) and P-LCC (p = 0.254), the other hematological parameters were significantly different among negative blood cultures and different pathogens BSI (Table 3). Compare to negative blood cultures, E. coli BSI has higher values of CRP, NEU, MPV, MPV/PLT, NLR, and P-LCR, but lower values of PLT, LYM, and PCT (Supplementary Table 4. Changes of hematological parameters in patients with negative, Gram-positive, Gram-negative, and Candida blood cultures.
Besides, compared to S. haemolyticus BSI, E. coli BSI has a higher value of NLR (Supplementary Figure  6); E. aerogenes BSI has a lower value of LYM (Supplementary Figure 3). Moreover, compared to S. aureus BSI, both E. coli BSI and K. pneumoniae BSI have a higher value of P-LCR (Supplementary Figure  6). Similarly, compare to Candida BSI, both E. coli BSI and K. pneumoniae BSI have a higher value of MPV (Supplementary Figure 4); besides, K. pneumoniae BSI has a higher value of P-LCR as well (Supplementary Figure 6). See Table 5 for more details.

Differential ability of hematological parameters
ROC curves were constructed to evaluate the power of hematological parameters to distinguish patients with BSI caused by different pathogens. The AUC, optimal cutoff value, sensitivity, and specificity of each hematological parameter are presented in Supplementary Table 1   Furthermore, we performed univariate logistic regression analysis to examine the associations of each hematological parameter with different pathogens BSI, and calculated the standardized regression coefficient (β) and the odds ratio (OR) for each blood cell parameter. See Supplementary Table 2 for more details. The combined ROC curves of relevant hematological parameters were shown in Figure 3. The combinations of relevant hematological parameters can increase the differential ability to different pathogens BSI.

Discussion
Blood culture is the most definitive way to confirm BSI. Nonetheless, this gold standard needs at least one day to get the result and may be influenced by many factors [22]. Besides, partly due to differences in both pathogen and host, individual clinical responses to BSI vary greatly [23]. To date, early recognition, rapid microbiological diagnosis, as well as prompt initiation of appropriate antibiotics are always the goals of clinicians who confront probable BSI [24]. Therefore, there is an urgent need for efficient and rapid detection of BSI. For this purpose, it is important to explore BSI comprehensively and deeply, especially the relationship between the causative pathogens and the changes of hematological parameters.
Additionally, it has been demonstrated that platelets with higher MPV values have a larger surface area and more granules, which is associated with their activation [32]. Furthermore, MPV, MPV/PLT, and PLR were considered as diagnostic adjunct tests for BSI [33][34][35][36]. Nevertheless, Johansson et al. have reported that there was no association between bacterial species and the occurrence of thrombocytopenia [37]. However, each species of bacterium, and even individual strains, have different mechanisms for interacting with platelets [18]. Actually, the immune responses of the host to Grampositive pathogens are fundamentally different from Gram-negative pathogens [38,39]. Therefore, the kinds of causative pathogens should be taken into account. Opposite to our results, Djordjevic and coauthors reported [9] that patients with Gram-positive BSI have significantly lower values of MPV/PLT and PLR than those with negative blood cultures. Comparing to Gram-positive BSI and negative blood cultures, patients with Gram-negative BSI have the highest values of PLR. It should be noted that the population included in that research were patients with critically ill BSI and severe trauma. Altogether, to some extent, platelet-related parameters may reflect the distinction of platelet activation in BSI caused by different pathogens.
Although in a study by Wu et al., no significant differences were found in WBC, PLT, and CRP between BSI and negative blood cultures [11], while most of the previous studies have shown that the immune responses to BSI have obvious characteristics, such as the increase of NEU, NLR, PDW, and CRP [40], as well as the decrease of LYM [41]. These are consistent with our results. Furthermore, NLR and MPV were found to reflect the severity of BSI, as well as an independent predictor of death [9,42]. Currently, we provided the first retrospective observational study to systematically investigate the changes of more kinds of hematological parameters and their differential ability in patients with BSI caused by different pathogens, including Candida spp. Notably, our results provided some fundamental data, most of which were reported for the first time.
It is undeniable that our present study has some limitations. Firstly, the study was conducted at a single center, and the findings may not be readily suitable to patients with different demographic characteristics. Secondly, we excluded patients suffering from the severe underlying disease. Thirdly, patients with positive blood cultures were significantly older than those with negative blood cultures. Although this may be because older patients are more prone to BSI than younger patients, age-matched groups can better reflect the effects of different pathogens on hematological parameters. Fourthly, except for retrospective observational design, implementation of strict inclusion and exclusion criteria led to a lower number of patients with BSI caused by S. haemolyticus, E. faecalis, E. faecium, E. aerogenes, E. cloacae, P. aeruginosa, A. baumannii, and Candida. To explore the changes of hematological parameters and their differential ability in patients with BSI caused by more kinds of pathogens, further larger prospective studies are warranted.

Conclusions
Taken together, the specific combinations of hematological parameters can improve the power to distinguish patients with BSI caused by different pathogens. Attention to these parameters can be easily integrated into daily medical activities, without extra costs.

Annex -Supplementary Items
\ Supplementary Figure 1. Hematological parameters among and between negative, Gram-positive, Gram-negative, and Candida blood cultures. (a) platelet count (PLT) and C-reactive protein (CRP); (b) neutrophil count (NEU) and lymphocyte count (LYM); (c) platelet distribution width (PDW) and mean platelet volume (MPV); (d) plateletcrit (PCT) and mean platelet volume-to-platelet count ratio (MPV/PLT); (e) neutrophil-to-lymphocyte ratio (NLR) and platelet larger cell ratio (P-LCR). Comparisons should be read from left to right. The estimate is located at the intersection of the column-defining blood cultures and the row-defining blood cultures. Significant results are bolded and underscored. Green means that, in comparison, the former is lower than the latter. Red means that, in comparison, the former is higher than the latter.

Supplementary Figure 2. Platelet count (PLT) and C-reactive protein (CRP) values among and between negative blood cultures and bloodstream infection (BSI) caused by different pathogens. Comparisons should be read from left to right.
The estimate is located at the intersection of the column-defining blood cultures and the row-defining blood cultures. Significant results are bolded and underscored. Green means that, in comparison, the former is lower than the latter. Red means that, in comparison, the former is higher than the latter. Figure 3. Neutrophil count (NEU) and lymphocyte count (LYM) values among and between negative blood cultures and bloodstream infection (BSI) caused by different pathogens. Comparisons should be read from left to right. The estimate is located at the intersection of the column-defining blood cultures and the row-defining blood cultures. Significant results are bolded and underscored. Green means that, in comparison, the former is lower than the latter. Red means that, in comparison, the former is higher than the latter. Figure 4. Platelet distribution width (PDW) and mean platelet volume (MPV) values among and between negative blood cultures and bloodstream infection (BSI) caused by different pathogens. Comparisons should be read from left to right. The estimate is located at the intersection of the column-defining blood cultures and the row-defining blood cultures. Significant results are bolded and underscored. Green means that, in comparison, the former is lower than the latter. Red means that, in comparison, the former is higher than the latter. Figure 5. Plateletcrit (PCT) and mean platelet volume-to-platelet count ratio (MPV/PLT) values among and between negative blood cultures and bloodstream infection (BSI) caused by different pathogens. Comparisons should be read from left to right. The estimate is located at the intersection of the column-defining blood cultures and the rowdefining blood cultures. Significant results are bolded and underscored. Green means that, in comparison, the former is lower than the latter. Red means that, in comparison, the former is higher than the latter.