No Significant Differences in Presepsin Levels According to the Causative Microorganism of Bloodstream Infection

Abstract

Background

CD14 recognizes lipopolysaccharide (LPS), and presepsin is a fragment of soluble CD14. Still, it remains uncertain whether Gram-negative bacteria induce higher presepsin levels than other microorganisms. To address this question, this study aimed to analyze presepsin levels based on microorganisms isolated in blood cultures.

Materials and Methods

This study was a single-center study comprising suspected sepsis patients enrolled from July 2020 to September 2020. A total of 95 patients with a single isolate confirmed in blood culture were analyzed to evaluate if there are any differences in presepsin levels according to microbial isolates. Plasma presepsin level was measured using PATHFAST assay kit and analyzer (LSI Medience Corporation, Tokyo, Japan).

Results

There were 26 Gram-positive bacteremia, 65 Gram-negative bacteremia, and 3 fungemia patients with median presepsin levels of 869, 1,439, and 11,951 pg/mL, respectively. Besides, one case of algaemia demonstrated a presepsin level of 1,231 pg/mL. Our results showed no statistically significant difference in presepsin levels among patients with Gram-positive bacteremia, Gram-negative bacteremia, and fungemia. Furthermore, presepsin levels did not differ significantly among bloodstream infections caused by bacteria that were isolated from at least three different patients. In particular, Gram-positive bacteria such as Staphylococcus aureus and Enterococcus faecalis were able to induce presepsin levels comparable to those induced by Gram-negative bacteria.

Conclusion

We demonstrated that there were no significant differences in plasma presepsin levels according to microbial isolates in blood culture. The major cause of the variability in presepsin levels during bloodstream infection might be the immunogenicity of each microorganism rather than the presence of LPS in the microorganism.

Graphical Abstract

Introduction

Since the discovery of presepsin, most studies published to date have focused on evaluating the clinical significance of presepsin in diagnosing sepsis and predicting the mortality risk of sepsis compared to other markers such as C-reactive protein and procalcitonin [1,2]. Likewise, we recently suggested the value of presepsin in risk stratification and the potential benefit of using presepsin along with the Sepsis-3 criteria [3].

Presepsin, a soluble CD14 subtype (sCD14-ST), is a fragment of CD14 [4] and is secreted by human monocytes following phagocytosis [4,5]. Presepsin level rises after 1 hour from LPS exposure, then reaches its peak at 3 hours [6,7]. Since the primary function of CD14 is to mediate lipopolysaccharide (LPS)-induced signaling [8], and the outer membrane of Gram-negative bacteria are primarily composed of LPS, it has been suspected that Gram-negative bacterial infection could induce higher presepsin levels [14]. Nevertheless, conflicting results have been suggested regarding the difference in presepsin levels according to different causative microorganisms [9,10,11,12,13,14]. To date, there have been only a few studies regarding this issue [9,11,12,13,15,16]. However, most of the studies had a small sample size, and none of the studies reported the clinical characteristics of each group which could pose a potential bias. While the diagnostic and prognostic values of presepsin have been established in multiple studies [17,18], the question of whether presepsin response varies according to the type of causative microorganisms has been underappreciated. Thus, we aimed to elucidate whether there are significant differences in presepsin levels according to different microbial isolates in blood culture.

Materials and Methods

1. Study design

This study was conducted as a subgroup analysis of a population previously gathered to investigate the prognostic value of presepsin in suspected sepsis patients enrolled from July 2020 to September 2020 at Samsung Medical Center (a 1,989-bed, university-affiliated, tertiary care referral hospital located in Seoul, Korea). For the evaluation of presepsin in risk stratification, a total of 755 patients with suspected sepsis were analyzed in our previous study [3]. Among these patients, 109 showed positive blood cultures, 101 of whom had a single isolate and without replicates from a single patient with persistent bloodstream infection. For common contaminants such as Bacillus species, coagulase-negative staphylococci (CoNS), viridans group streptococci, and Corynebacterium species, the isolates were regarded as pathogens if any of the following conditions were fulfilled: (1) identical isolates from different culture bottles obtained during the same episode, and (2) identical isolates identified on different dates. Although a single occurrence of an isolate still had the possibility of true bacteremia, we decided not to include these cases to reduce analytical bias caused by potential contaminants. As a result, six cases were excluded, and 95 cases were included in the analysis (Fig. 1). In addition to presepsin levels and blood culture results, clinical characteristics such as age, sex, comorbidities, sequential organ failure assessment (SOFA) score, laboratory test results, and 28-day mortality were collected for all subjects. The origin of the infection was determined with the culture results of specimens from other sites and/or with clinical presentations. Patients with malignancy and/or solid organ transplantation were considered as immunocompromised status.

Figure 1. Study flowchart.

ED, emergency department; HO, haemato-oncology; ICU, intensive care unit; BSI, bloodstream infection.

2. Ethics statement

This study was approved by the Institutional Review Board of the Samsung Medical Center (IRB No. 2019-05-170), and the need for informed consent was waived because this study was anonymous and utilized residual samples.

3. Presepsin measurement

The residual plasma samples after procalcitonin testing were utilized for presepsin measurement. These samples were collected within 24 hours of the blood culture being drawn. The PATHFAST assay kit and analyzer (LSI Medience Corporation, Tokyo, Japan) were used to measure presepsin levels. The PATHFAST system (LSI Medience Corporation) is a fully automated enzyme immunoassay system that utilizes the chemiluminescence principle to measure presepsin levels in approximately 17 min [19]. The upper limit of the analytical measurement range was 20,000 pg/mL. The imprecision assessment of the PATHFAST assay kit and analyzer (LSI Medience Corporation) demonstrated a within-run coefficient of variation (CV) and within-laboratory CV of less than 5%. For further details regarding the analytical performance, please refer to our previous publication [3].

4. Blood culture

Since the enrolled patients were suspected of having sepsis, blood cultures were routinely performed. The blood samples were collected in blood culture bottles (bioMérieux, Marcy l’Etoile, France) and incubated in BacT/ALERT 3D or BacT/ALERT Virtuo blood culture instruments (bioMérieux). The isolated microorganisms were identified using VITEK MS (bioMérieux), a matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) system for accurate and rapid identification of bacterial and fungal species [20,21,22].

5. Data analysis

The subjects were divided into different groups according to the category of the microorganism isolated from blood culture: (1) Gram-positive bacteria, (2) Gram-negative bacteria, and (3) fungi. Descriptive statistics of presepsin levels as well as other clinical characteristics were calculated for each group. To compare the clinical characteristics, Fisher’s exact test, Kruskal-Wallis test, and Wilcoxon’s test were performed as appropriate. For characteristics demonstrating statistically significant differences among the three compared groups, a post-hoc analysis was carried out and presepsin levels were further compared based on the subgroups classified with each categorizable characteristic. In addition, as presepsin secretion is an immune response against microorganisms, the presepsin levels were compared between immunocompromised and immunocompetent patients. As our previous study focused on the performance of presepsin in predicting 28-day mortality [3], the difference in presepsin levels according to 28-day mortality was also compared in each group as well. In order to compare the presepsin levels of different species, pathogens that were isolated in more than two different patients were analyzed. Throughout the analyses, box-and-whisker plots were prepared. To compare the presepsin levels, the Kruskal-Wallis test was conducted for the comparison among more than two groups, and Wilcoxon’s test was performed for the comparison between two groups. Statistical significance was set at P <0.05. The P values were adjusted with Bonferroni correction to account for multiple comparisons. All statistical analyses were performed with R 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria), and the plots were depicted with the ggplot2 3.3.3 package on R 4.0.2 (R Foundation for Statistical Computing).

Results

1. Characteristics of subjects and distribution of the identified microorganisms in blood cultures

Among the 95 patients, 26 were infected by Gram-positive bacteria, 65 by Gram-negative bacteria, three by fungi, and one by algae. Although the characteristics of each group did not significantly differ in most aspects, there were a few distinguishing features. Based on the origin of the infection, 8 of 26 Gram-positive bacteremia cases were of respiratory origin (30.8%), 20 of 65 Gram-negative bacteremia cases were of urinary origin (30.8%), and two of three fungemia cases were of bone and soft tissue origin (66.7%). The proportion of bone and soft tissue infection was significantly greater in the fungi group compared to the Gram-negative bacteria group (P = 0.023). The SOFA score at enrollment was significantly greater in the fungi group compared to the Gram-positive bacteria group (P = 0.035) and Gram-negative bacteria group (P = 0.033). The Gram-positive bacteria group had significantly lower total bilirubin levels compared to the Gram-negative bacteria group (P = 0.021) and fungi group (P = 0.048). Detailed information regarding the characteristics of each group is provided in Table 1.

Table 1. Characteristics of subjects based on the microorganism isolated from blood culture.

Characteristics		Gram-positive bacteria (n = 26)	Gram-negative bacteria (n = 65)	Fungi (n = 3)	P-value
Age, years		61.6 (50.9 - 70.6)	64.6 (54.4 - 71.7)	65.2 (56.9 - 65.2)	0.799
Sex, male		14 (53.8)	32 (49.2)	3 (100.0)	0.342
Clinical status
	Diabetes	3 (11.5)	10 (15.4)	0 (0.0)	0.841
	Cerebrovascular disease	4 (15.4)	4 (6.2)	1 (33.3)	0.087
	Chronic cardiac disease	4 (15.4)	8 (12.3)	1 (33.3)	0.397
	Chronic lung disease	0 (0.0)	3 (4.6)	0 (0.0)	0.597
	Chronic liver disease	2 (7.7)	14 (21.5)	0 (0.0)	0.231
	Chronic renal disease	1 (3.8)	8 (12.3)	0 (0.0)	0.585
	Malignancy	15 (57.7)	48 (73.8)	2 (66.7)	0.250
	Immunocompromised	15 (57.7)	50 (76.9)	2 (66.7)	0.118
Infection focus
	Respiratory	8 (30.8)	6 (9.2)	0 (0.0)	0.179
	Urinary	3 (11.5)	20 (30.8)	0 (0.0)	0.532
	Gastrointestinal	2 (7.7)	7 (10.8)	1 (33.3)	1.000
	Hepatobiliary and pancreatic	2 (7.7)	19 (29.2)	0 (0.0)	0.260
	Bone and soft tissues	5 (19.2)	2 (3.1)	2 (66.7)	0.008
	Others	6 (23.1)	11 (16.9)	0 (0.0)	1.000
SOFA score at enrollment		4 (2 - 6)	5 (2 - 7)	16 (9 - 16)	0.030
Sepsis at enrollmenta		19 (73.1)	41 (63.1)	3 (100.0)	0.370
Septic shock at enrollmentb		5 (19.2)	19 (29.2)	1 (33.3)	0.531
Laboratory tests
	WBC (103/mm3)	9.4 (1.9 - 17.3)	6.2 (2.1 - 11.0)	0.1 (0.0 - 16.2)	0.324
	Hemoglobin (g/dL)	9.9 (8.9 - 12.0)	9.8 (8.6 - 11.2)	9.8 (9.1 - 10.7)	0.980
	Platelet (103/mm3)	113.5 (50 - 154.2)	99 (31 - 180)	12 (6.5 - 40)	0.135
	Total bilirubin (mg/dL)	0.7 (0.4 - 1.5)	1.0 (0.7 - 3.4)	7.7 (4.9 - 19.5)	0.003
	Creatinine (mg/dl)	1.0 (0.7 - 1.3)	1.0 (0.7 - 1.5)	1.0 (0.8 - 1.4)	0.742
	Lactate (mmol/L)	1.9 (1.3 - 3.1)	2.1 (1.5 - 3.8)	2.4 (2.2 - 3.0)	0.502
	CRP (mg/dL)	13.8 (4.5 - 20.6)	9.1 (4.5 - 19.3)	11.4 (8.1 - 13.6)	0.809
	Procalcitonin (ng/mL)	2.7 (0.6 - 7.7)	2.1 (0.9 - 13.3)	4.2 (2.4 - 20.1)	0.832
28-day mortality		5 (19.2)	13 (20.0)	2 (66.7)	0.192

The data are presented as medians (IQRs) for continuous variables or as numbers (%) for categorical variables.

^aSepsis, delta SOFA score ≥2.

^bSeptic shock, patients who met septic shock criteria according to the new Sepsis-3 definition.

SOFA, sequential organ failure assessment; WBC, white blood cell count; CRP, C-reactive protein; IQR, interquartile range.

The distribution of the isolated species is shown in Figure 2. Among the 26 Gram-positive bacteremia cases, Staphylococcus aureus was the most common isolate, comprising nine cases (34.6%). Among a total of 65 Gram-negative bacteremia cases, Escherichia coli, Klebsiella pneumoniae, and Pseudomonas aeruginosa made up 28 (43.1%), 15 (23.1%), and six (9.2%) cases, respectively. There was one Gram-positive bacillus and one Gram-negative bacillus that were unidentifiable by MALDI-TOF MS. In particular, all three fungal isolates belonged to Candida species. There was one case of algaemia from Prototheca zopfii infection.

Figure 2. Number of patients for each isolated microorganism.

ssp., subspecies.

2. Presepsin levels in Gram-positive bacteremia, Gram-negative bacteremia, and fungemia

Median presepsin levels were 869 pg/mL (interquartile range [IQR]: 628 – 2,514), 1,439 pg/mL (IQR: 753 - 2,841), and 11,951 pg/mL (IQR: 6,632 - 15,976) for Gram-positive bacteremia, Gram-negative bacteremia, and candidemia, respectively. It was shown that there is a significant difference in presepsin levels among the three groups (P = 0.046). Wilcoxon’s test demonstrated adjusted P values of 0.260 for “Gram-positive bacteria vs. Gram-negative bacteria,” 0.067 for “Gram-positive bacteria vs. fungi,” and 0.372 for “Gram-negative bacteria vs. fungi,” none of which were statistically significant (Fig. 3A). The distribution of presepsin levels in the Gram-positive and Gram-negative bacteremia groups as well as the statistical significance of the difference are presented in Table 2 along with those reported in previous studies [9,11,12,13,15,16].

Table 2. Summary of previous reports comparing the difference in presepsin levels between Gram-positive and Gram-negative infections.

Studies	Number of cases		Presepsin levels (pg/mL)		P-value
	Gram-positive	Gram-negative	Gram-positive	Gram-negative
Endo et al. (2012) [9]	22	36	2,881 ± 4,374a	2,641 ± 3,709a	0.532
Romualdo et al. (2014) [11]	8	26	1,085 ± 2,062a	1,552 ± 2,356a	0.808
Rabensteiner et al. (2014) [15]	102	124	1,078 (670 - 2,422)b	1,295 (777 - 2,654)b	NA
Masson et al. (2015) [12]	112	98	963 (559 - 2,297)b	1,398 (763 - 2,774)b	0.020
Matera et al. (2017) [13]	20	15	NA	NA	<0.05
Lu et al. (2018) [16]	8	18	871 (556 - 1,077)c	1,448 (554 - 3,890)c	0.013
Current	26	65	869 (628 - 2,514)b	1,439 (753 - 2,841)b	0.087

The presepsin levels are presented in ^amean±SD or ^bmedian (IQRs) or ^cmedian (min, max).

NA, not available.

3. Presepsin levels in bloodstream infection caused by microorganisms other than bacteria

In the candidemia group, which consisted of three cases, the presepsin levels for each case were as follows: >20,000 pg/mL, 11,951 pg/mL, and 1,313 pg/mL. The patient with the highest presepsin level had experienced a polymicrobial bacteremia involving Enterococcus faecium, E. coli, and Enterobacter cloacae complex two days before Candida parapsilosis was identified exclusively.

Interestingly, a very rare case of human algaemia was included in our study. The patient, a 54-year-old woman undergoing chemotherapy for B-lymphoblastic leukemia, exhibited a presepsin level of 1,231 pg/mL and did not have any other origin of infection. She presented with a fever in the emergency department, and Prototheca zopfii, an alga, was isolated from blood culture.

4. Presepsin levels in bloodstream infections by different microbial species

Presepsin levels were compared between isolates that were identified in more than two patients. The Kruskal–Wallis test showed that presepsin levels did not significantly differ among different isolates (P = 0.670; Fig. 3B). Even when the isolates of interest were limited to those identified from more than five patients, there was still no statistically significant difference in presepsin levels among different species (P = 1.000). The median presepsin levels in bacteremia caused by S. aureus, E. coli, K. pneumoniae, P. aeruginosa were 2,688 pg/mL (IQR: 821 - 3,376), 1,490 pg/mL (IQR: 737 - 3,329), 1,448 pg/mL (IQR: 821 - 3,225), and 1,558 pg/mL (IQR: 1,141 - 1,985), respectively.

Figure 3. Presepsin levels among different microorganisms: (A) Differences of presepsin levels among Gram-positive bacteria, Gram-negative bacteria, and fungi; (B) Difference of presepsin levels among microorganisms isolated from at least three different patients.

ssp., subspecies.

5. Presepsin levels depending on various clinical characteristics

Although the median presepsin level was higher in the immunocompetent group compared to the immunocompromised group, the difference was not statistically significant (P = 0.060; Supplementary Fig. 1A). When comparing presepsin levels across different types of microorganisms while considering the host immune status, we observed that fungemia exhibited significantly higher presepsin levels than Gram-positive bacteremia in the immunocompromised subgroup (P = 0.044). Otherwise, no significant differences were observed in the presepsin levels (Supplementary Fig. 1B). While presepsin levels exhibited significant differences based on the presence of bone and soft tissue infections (P = 0.900; Supplementary Fig. 1C), hyperbilirubinemia (P < 0.001; Supplementary Fig. 1E) and 28-day mortality (P = 0.006; Supplementary Fig. 1G), the presepsin levels did not significantly differ across different types of microorganisms within each subgroup defined by the aforementioned characteristics (Supplementary Fig. 1D, 1E and 1H).

Discussion

Although the clinical significance of presepsin in sepsis has been established in multiple previous studies [17], whether there is a significant difference in presepsin response depending on the causative pathogen is still not well established. The secretion of presepsin from monocytes is triggered by the bacterial phagocytosis [4,5]. While presepsin is a fragment of CD14 which recognizes LPS, a major component of the outer membrane of Gram-negative bacteria, the possibility of disparity in presepsin levels depending on different pathogens has not attracted enough attention.

To our knowledge, there have been six previous studies that investigated this issue from sepsis patients (Table 2) [9,11,12,13,15,16]. While most previous studies reported a higher mean or median presepsin level in the Gram-negative group compared to the Gram-positive group [11,12,13,15,16], only three studies demonstrated a statistically significant difference (P <0.05) [12,13,16]. A recent study attempted to address this by utilizing an in vitro coculture method, which demonstrated a statistically significant difference in presepsin levels between Gram-positive bacteria and Gram-negative bacteria (P <0.001) [14]. In our study, while Gram-negative bacteremia exhibited higher presepsin levels compared to Gram-positive bacteremia, the difference was not statistically significant. Although there were only three candidemia cases included in this study, which may not be representative of the whole spectrum of fungemia, they elicited presepsin levels higher than both Gram-positive and Gram-negative bacteremia groups. While one of the candidemia cases had preceding bacteremia, we believe that the magnitude of the presepsin level was more likely to be affected by the Candida infection, considering the early elevation and short half-life of presepsin [7]. Previous studies have demonstrated increased presepsin levels in patients with fungal infections and also through in vitro experiments [23,24]. Considering the high SOFA scores of the fungemia group, high presepsin levels in the three candidemia cases would be associated with the severity of infection as previously suggested [23]. Furthermore, a case of algaemia caused by P. zopfii exhibited a presepsin level comparable to that of Gram-negative bacteremia. Of note, human protothecosis has been scarcely reported in the literature, and previously reported cases were generally fatal [25,26].

Regarding the species-level differences in presepsin levels, while one study compared the presepsin levels among Gram-negative species and reported a significantly higher presepsin level in infections caused by Pseudomonadales (Acinetobacter baumannii and P. aeruginosa) [13], this trend was not evident in our study. A previous in vitro coculture study demonstrated that Gram-positive bacteria were also capable of inducing presepsin secretion, and bacteria such as S. aureus were able to induce presepsin secretion to a level comparable to that of Gram-negative bacteria [14]. Particularly, S. aureus exhibited higher presepsin levels than E. coli and P. aeruginosa [14]. This might be due to the different immunogenicity of each pathogen, different phagocytosis sensitivity, and the association of CD14 with ligands other than LPS [14]. Our study also revealed that Gram-positive bacteria such as S. aureus and Enterococcus faecalis are able to induce presepsin levels comparable to those induced by Gram-negative bacteria.

Even when the analysis was segmented into various subgroups based on host immune status, bilirubin level, origin of infection, and 28-day mortality, the presepsin levels in different types of microorganisms did not exhibit significant differences. Moreover, presepsin level was significantly higher in the decedent group but did not differ by host immune status and origin of infection. The results of our study imply that presepsin would be a feasible sepsis marker regardless of the host immune status and the origin of the infection.

Taking all these findings into consideration, it is plausible to think that the immunogenicity of each microorganism contributes to the rise in presepsin levels instead of the presence of LPS in the outer membrane of the microorganism. Although the primary function of CD14, the precursor of presepsin, is to facilitate Toll-like receptor 4 (TLR4) in response to the recognition of LPS and induction of TLR4 endocytosis [8], monocytes incubated with LPS were unable to induce presepsin secretion [5]. Rather, as it has been reported that sterile phagocytic stimulus can induce presepsin secretion [5], it is implied that phagocytosis of any microorganism would result in CD14 consumption and cleavage, which leads to presepsin secretion, whereas LPS would not play a crucial role in presepsin response despite its relationship with CD14. Besides, CD14 is also capable of interacting with various pattern recognition receptors that contribute to the innate immunity [12,27,28], which could potentially account for the elevated presepsin levels in response to different types of pathogens. In brief, while it is uncertain whether CD14 plays a direct role in presepsin secretion in response to microorganisms other than Gram-negative bacteria, our results showed that bloodstream infections by Gram-positive bacteria, fungi, and even an alga were able to produce presepsin levels comparable to that of Gram-negative bacteremia.

To the best of our knowledge, our study is the first study to primarily focus on the difference in presepsin levels among various microorganisms. In turn, this study is also the first to describe the clinical characteristics of each group classified with the type of microorganism, which could potentially cause bias in presepsin levels. There are a few limitations to note in our study. First, the data did not address the load of the pathogen in the bloodstream. Thus, the difference in presepsin levels among the species could be a result of different bacterial and fungal burdens. Second, the species incorporated in our analysis showed a skewed distribution with certain predominant pathogens in each group. However, this may better represent the clinical spectrum of patients with suspected sepsis since the subjects were enrolled in a prospective manner before the culture results were reported. Third, the samples used in this study may not necessarily represent the peak level and would be those from various stages of presepsin kinetics. While this could be a potential source of bias, accurately determining the timing of peak presepsin level in clinical practice would be nearly impossible. Fourth, considering the age distribution of the study population, it is likely that immunosenescence would have affected the presepsin response. Still, since there were no significant differences in age across the three compared groups, the effect of immunosenescence would apply to all groups.

In conclusion, there were no significant differences in presepsin levels among different types of microorganisms. The presepsin levels in Gram-positive bacteremia, candidemia, and protothecosis were all comparable to those induced by Gram-negative bacteremia despite the lack of LPS in these microorganisms. It is implied that presepsin could serve as a biomarker of sepsis regardless of the type of pathogen. The results of our study should be validated on a larger scale, and further research to elucidate the underlying mechanisms of presepsin secretion in response to various microorganisms is warranted.

SUPPLEMENTARY MATERIAL

Supplementary Figure 1

Presepsin levels in various clinical subgroups.