Diagnostic and prognostic utility of D-dimer and heparin-binding protein in neonatal sepsis: A prospective case-control study

Abstract

BACKGROUND

Neonatal sepsis is a serious health problem, with high morbidity and mortality during the first 28 days of life. Clinical diagnosis at presentation is challenging due to the nonspecific signs and symptoms. Although blood culture is the gold standard for diagnosis, it is not always positive.

AIM

To evaluate the diagnostic and prognostic utility of D-dimer and heparin-binding protein (HBP) in neonatal sepsis.

METHODS

This prospective case-control study included 90 neonates in two groups: A sepsis group (n = 45) and a control group (n = 45) without sepsis. Sepsis group was further subdivided based on blood culture results into proven sepsis (n = 28 culture-positive sepsis) and suspected sepsis (n = 17 culture-negative sepsis). All neonates underwent complete history taking, thorough clinical examination and investigations [complete blood count, C-reactive protein (CRP), liver and kidney function tests, plasma D-dimer and HBP].

RESULTS

Levels of CRP, D-dimer and HBP were significantly higher in the sepsis group compared to the controls. At a cutoff value above 517.9 ng/mL, D-dimer outperformed CRP and HBP in distinguishing sepsis group from controls with 95.6% sensitivity and 97.8% specificity. D-dimer was also a better prognostic marker than the neonatal sequential organ failure assessment (nSOFA) for predicting mortality, with 100% sensitivity and 92.5% specificity vs 80% sensitivity and 82.5% specificity. There was a significant positive correlation between CRP, D-dimer and HBP.

CONCLUSION

D-dimer demonstrated superior diagnostic accuracy compared to CRP and HBP in predicting sepsis, and demonstrated superior prognostic accuracy compared to nSOFA in predicting the outcome of neonatal sepsis.

Key Words: C-reactive protein; D-dimer; Heparin-binding protein; Mortality; Neonatal sepsis

Core Tip: Neonatal sepsis is a serious health problem. Although early diagnosis is critical, clinical diagnosis at presentation is difficult due to its nonspecific signs and symptoms. Although blood culture is considered the gold standard for diagnosis, unfortunately, it is not always positive. Efforts have been made to help early suspicion of sepsis, for example, using the hematological scoring system. Nonetheless, there remains a pressing need for novel laboratory markers that can help differentiate bacterial infection from other pathological conditions. The current study investigates the diagnostic and prognostic value of D-dimer and heparin-binding protein in neonatal sepsis.

INTRODUCTION

Neonatal sepsis is a systemic inflammatory response syndrome in the presence of suspected or proven infection that occurs during the first 28 days of life[1]. Neonatal sepsis is a key contributor to neonatal mortality worldwide, and low- and middle-income countries (LMICs) are variably affected. In the last decade (2009–2018), the incidence was 3930 per 100000 Live births according to reports from LMICs[2]. Globally, approximately four million newborn deaths occur each year. The majority of these deaths usually occur in low-income countries, and almost one million of these deaths are attributed to infectious causes, including neonatal sepsis, meningitis, and pneumonia[3].

Despite advances in neonatal care, sepsis remains a significant cause of morbidity and mortality, particularly in preterm and very low birth weight babies[4]. Newborns are evaluated for the onset of sepsis based on symptoms and risk factors for sepsis in the neonatal intensive care unit (NICU) and newborn clinics. Focal infections mostly associated with sepsis are pneumonia, meningitis, necrotizing enterocolitis, and urinary tract infection[5].

Due to the nonspecific neonatal presentation of sepsis and the high risk of morbidity and mortality without treatment, many asymptomatic neonates undergo a sepsis workup if concerning risk factors are present. Although approximately 7% to 13% of all neonates are worked up for sepsis, only 50% develop positive cultures[4]. Efforts have been made to help the early suspicion of sepsis. The hematological scoring system (HSS) of Rodwell was established to classify if sepsis is unlikely, suspected, or very likely[6]. In addition, a neonatal sequential organ failure assessment (nSOFA) was established to predict mortality in neonatal sepsis[7].

Early detection and optimal treatment of infections that allow a decrease in the probability of progression to sepsis remain a common clinical challenge. Novel laboratory tests for the differentiation of bacterial infection from other types of disease are still in high demand. Although most of the biomarkers have been evaluated in sepsis, none are considered adequate for routine clinical use[8]. These biomarkers, such as C-reactive protein (CRP), procalcitonin (PCT), or peripheral white blood cell count (WBCs), are often used[9].

D-dimer is a fibrin degradation product that is formed on cleavage of cross-linked fibrin by fibrinolysis[10]. In sepsis, there is an enhanced coagulopathy and fibrin formation with intravascular fibrin deposition. This leads to marked vascular inflammation and endothelial injury with pronounced fibrin breakdown, and elevations in plasma D-dimer[11,12].

In addition, the neutrophil-derived heparin-binding protein (HBP), which resides in the secretory and azurophilic granules, is released in the presence of bacteria[13]. HBP plays many critical roles in sepsis. It acts as a chemoattractant and immunological booster for a variety of immune cells, such as neutrophils, macrophages, and T-lymphocytes. In sepsis, HBP exacerbates the inflammatory responses, causes capillary leakage, and increases endothelial permeability[14].

Although there is a growing interest in the utility of such markers in diagnosing sepsis, there remains a lack of standardized cutoff values, especially in neonates[15], and the reports about their performance are contradictory[16-21]. The current study aims to evaluate the diagnostic and prognostic utility of D-dimer and HBP in neonatal sepsis.

MATERIALS AND METHODS

Study population

This prospective case-control study initially recruited 102 neonates. Based on inclusion and exclusion criteria, 90 neonates who met the inclusion criteria were ultimately included. They were recruited from the neonatal intensive care unit (NICU) of Pediatric Hepatology, Gastroenterology and Nutrition Department in National Liver Institute, Menoufia University, and the NICU of Damietta General Hospital from January 2020 to February 2022. They were divided into two main groups, the sepsis group (n = 45) and the non-sepsis control group (n = 45). The inclusion criteria for the sepsis group were based on the clinical symptoms and signs consistent with sepsis, with or without the presence of risk factors, in addition to HSS ≥ 3[6]. HSS ≥ 3 was selected as the cutoff with the best sensitivity and specificity in discriminating sepsis as described in previous reports[22]. The sepsis group was further subdivided according to blood culture into proven sepsis (n = 28 with culture-positive sepsis) and suspected sepsis (n = 17 with culture-negative sepsis). Proven sepsis patients have a positive blood culture in addition to sepsis clinical symptoms while suspected sepsis patients have the clinical signs of sepsis without laboratory evidence by culture[23]. All sepsis patients received antibiotics after blood sampling for laboratory investigations and sepsis workup. Control group were neonates admitted to NICU for causes other than sepsis (transient tachypnea of the newborn, indirect hyperbilirubinemia, infant of a diabetic mother, and/or intrauterine growth retardation) and had no clinical or laboratory evidence of infection at presentation and for at least 72 hours follow-up after admission, in addition to HSS ≤ 2[6].

Neonates with organ failure unrelated to sepsis, those with non-infectious critical illness, or who had received heparin therapy within three days of the study enrolment, or neonates who take drugs that affect neutrophil count, or those on antibiotic therapy or steroids, were excluded from the study. Based on exclusion criteria, twelve neonates were excluded (eight patients received antibiotics, and four received a heparinized umbilical catheter). Written informed consent was obtained from the legal guardians of all the participants in the study and the study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki and was approved for both centers by the institutional review board of the National Liver Institute Menoufia University (NLI-IRB 00003413 FWA0000227). The approval number is 00441.

Biochemical investigations

All included neonates underwent complete history taking and thorough clinical examination, complete blood cell count (CBC) with differential count, liver function tests, kidney function tests, CRP, urine, and stool analysis. Blood culture was performed only for the sepsis group. Imaging studies such as chest, abdomen, or joint X-ray, echocardiography, and abdominal ultrasonography were requested according to the clinical indications.

Measurement of D-dimer and HBP

Plasma D-dimer and HBP were measured at admission (before the use of antibiotics) using enzyme-linked immunosorbent assay kits (Cat. CEA506Hu and SEB461Hu, respectively, Cloud-Clone Corporation, United States). Briefly, 50 mL of standards, test samples, and reagents were added to designated wells and incubated at 37 °C for one hour. After washing three times, 100 mL of detection reagent was added and incubated at 37 °C for 30 minutes. After washing five times, 90 µL of substrate solution was added and incubated at 37 °C for 20 minutes, then the reaction was stopped with 50 µL of stop solution. Optical density was measured at 450 nm immediately, and sample concentrations were calculated using a standard curve. Intra-assay and inter-assay variations expressed as coefficient of variation were < 10%, and < 12% respectively. They were similar for both kits.

Microbiological investigations

Blood culture bottles (aerobic and anaerobic) were inoculated (1 mL of blood each) and incubated in a blood culture system (BacT/ALERT 3D system, bioMérieux, Durham, NC, United States) until positive results were obtained with a maximum of seven days. Bottles with positive signals were sub-cultured on MacConkey agar, blood, and chocolate plates. Growing colonies were identified by standard microbiological methods[24]. Antibiotic sensitivity was performed using the disk diffusion technique according to Clinical and Laboratory Standards Institute[25]. Vitek2 compact system (bioMérieux, France) was used for further confirmation of the identified bacterial and fungal isolates and for antibiotic susceptibility testing of the isolates.

Statistical analysis

Quantitative data were expressed as mean ± SD, while qualitative data were presented as numbers and percentages. For quantitative data, significance between the two groups was tested by either the Student t-test or Mann Mann-Whitney test according to data homogeneity as determined by the Shapiro–Wilk test[26]. For comparisons between more than two groups, one-way analysis of variance or Kruskal-Wallis tests were used as appropriate. χ² or Monte Carlo tests were applied to compare qualitative data as appropriate. Correlations were tested by Pearson’s or Spearman’s tests for parametric and non-parametric data respectively. The cutoff value for optimal clinical performance was determined from receiver operating characteristic curve analysis. Diagnostic performance was evaluated as sensitivity, specificity, positive predictive value, and negative predictive value. Univariate and multivariate logistic regression analyses were performed to detect independent variables associated with sepsis or mortality. The sample size was calculated to be 41 with a 90%CI, using OpenEpi (https://www.openepi.com/SampleSize/SSPropor.htm) based on the reported prevalence of neonatal sepsis in LMICs[2]. The effect size was evaluated by Cohen’s test. Results were considered statistically significant when the P value was ≤ 0.05. Data analysis was conducted using SPSS version 21 (SPSS Inc., Chicago, IL, United States).

RESULTS

Study population characteristics and sepsis risk factors

The studied groups were age- and sex-matched (P > 0.05 for both). There was no significant difference between the studied groups regarding gestational age, birth weight, or mode of delivery. The majority (nearly 90%) were delivered by cesarean section. In the sepsis group, the onset of sepsis was comparable between proven and suspected sepsis (P > 0.05). Neonatal resuscitation and delayed first cry were common in the sepsis group but absent among the controls. Regarding maternal risk factors, premature rupture of the membrane was significantly more frequent in the sepsis group compared to the controls. On the other hand, preeclampsia, antepartum hemorrhage, and maternal infections were reported only in the sepsis group but absent in the controls. Most neonates with sepsis survived, while mortality was 11.1% (5/45), all of whom belonged to the proven sepsis group (Table 1). Mortality rates were comparable between proven and suspected sepsis groups (P = 0.14). Analysis of the parameters between proven and suspected sepsis groups showed no significant differences (P > 0.05) except for length of hospital stay, urinary tract infection, and chorioamnionitis, which were significantly more frequent in the proven sepsis group.

Table 1 Demographic, basal characteristics and risk factors for sepsis of the studied groups, n (%)/mean ± SD.

Parameter	Sepsis group (n = 45)		P value1	Controls (n = 45)	P value2
	Proven sepsis (n = 28)	Suspected sepsis (n = 17)
Sex (Males)	15 (53.6)	10 (58.8)	0.73	22 (48.9)	0.77
Age (days)	11.88 ± 11.74	11.74 ± 10.29	0.81	11.74 ± 10.29	0.11
Gestational age
Term	14 (50.0)	10 (58.8)	0.56	27 (60.0)	0.69
Preterm	14 (50.0%)	7 (41.2)		18 (40.0)
Birth weight (kg)
Normal birth weight	18 (64.3)	14 (82.4)	0.33	36 (80.0)	0.18
Low birth weight	8 (28.6)	3 (17.6)		9 (20.0)
Very low birth weight	2 (7.1)	0 (0.0)		0.0
Apgar score	8.07 ± 2.16	8.71 ± 2.14	0.29	9.96 ± 0.21	< 0.0001
Resuscitation	13 (46.4)	5 (29.4)	0.25	0.0	< 0.0001
Delayed first cry	13 (46.4)	5 (29.4)	0.25	0.0	< 0.0001
Infant of diabetic mother	0.0	0.0	N/A	5 (11.1)	0.56
Sepsis onset
Early onset	14 (50.0)	7 (41.2)	0.56	0.0	N/A
Late onset	14 (50.0)	10 (58.8)		0.0
Outcome
Survivors	23 (82.1)	17 (100)	0.14	45 (100)	0.004
Nonsurvivors	5 (17.9)	0.0		0.0
Days of admission	17.68 ± 5.09	10.65 ± 1.93	0.03	4.78 ± 2.36	< 0.0001
Maternal risk factors
PROM	16 (57.1)	8 (47.1)	0.51	5 (11.1)	< 0.0001
Preeclampsia	6 (21.4)	1 (5.9)	0.23	0.0	0.02
Antepartum hemorrhage	3 (10.7)	2 (11.8)	1.0	0.0	0.03
Urinary tract infection	3 (10.7)	0.0	0.03	0.0	0.02
Chorioamnionitis	1 (3.6)	0.0	0.01	0.0	0.01
Mode of delivery
Vaginal	2 (7.1)	1 (5.9)	1.0	5 (11.1)	0.89
Caesarian	26 (92.9)	16 (94.1)		40 (88.9)

¹Proven sepsis vs suspected sepsis.

²Sepsis group vs controls.

N/A: Not applicable; PROM: Premature rupture of membrane.

Types of infection, causative organisms, and treatment

Respiratory infections were the most frequent type of infection (60%) in the sepsis group, while gastrointestinal, central nervous system and joint infections were less frequent. On the other hand, the majority of controls (82.2%) presented with neonatal jaundice (Table 2). Blood cultures were positive in 62.2% (28/45) of the sepsis group. The frequency of incriminated organisms was nearly comparable. Staphylococcus and Streptococcus were detected in nearly half of the cultures (Table 3). All patients in the sepsis group received antimicrobials, namely glycopeptide and carbapenem (60.7%), while only three patients in the control group with transient tachypnea of the newborn received prophylactic antibiotics. Significance analyses of the parameters between the proven and suspected sepsis groups were all comparable (P > 0.05), except for the type and duration of antimicrobial therapy, which were significantly higher in the proven sepsis group (Table 2).

Table 2 Initial diagnosis, involved organisms, and antimicrobial type and duration, n (%)/mean ± SD.

Initial diagnosis	Sepsis group (n = 45)		P value1	Controls (n = 45)	P value2
	Proven (n = 28)	Suspected (n = 17)
Respiratory disease
Respiratory distress syndrome	8 (28.5)	3 (17.6)	0.405	0.0	N/A
Pneumonia	8 (28.5)	8 (47.1)		0.0
Meconium aspiration	2 (7.1)	0.0		0.0
TTN	0.0	0.0		4 (8.9)
Gastrointestinal disease
Gastroenteritis	1 (3.6)	3 (17.6)	0.4	0.0	N/A
Necrotizing enterocolitis	2 (7.1)	0.0		0.0
Central nervous system
Infection	2 (7.1)	0.0	1.0	0.0	N/A
Ruptured meningocele	1 (3.6)	1 (5.9)		0.0
Others
Septic arthritis	2 (7.1)	0.0	0.467	0.0	N/A
Neonatal jaundice	5 (17.9)	4 (23.5)		37 (82.2)	< 0.0001
Antimicrobials
No treatment	0.0	0.0	0.002	42 (93.3)	< 0.0001
B-Lactam and aminoglycoside	4 (14.3)	7 (41.2)		3 (6.7)
B-Lactam and cephalosporins3	4 (14.3)	10 (58.8)		0.0
Aminoglycoside and carbapenem	3 (10.7)	0.0		0.0
Glycopeptide and carbapenem	17 (60.7)	0.0		0.0
Duration of antimicrobials (days)	15.8 ± 4.9	10.1 ± 2.2	0.04	1.5 ± 0.21	< 0.0001

¹Proven sepsis vs suspected sepsis.

²Sepsis group vs controls.

³Third generation.

N/A: Not applicable; TTN: Transient tachypnea of newborn.

Table 3 Sepsis serological markers according to the causative organism, sepsis onset, gestational age and sepsis outcome, mean ± SD.

Sepsis group (n = 45)	CRP (mg/L)	D-dimer (ng/mL)	HBP (ng/mL)
Causative organism
No growth (n = 17)	21.35 ± 16.16	1915.9 ± 1093.1	261.79 ± 147.93
Pneumococci (n = 5)	45.60 ± 31.06	2481.68 ± 682.1	439.56 ± 226.10
Staphylococcus (n = 7)	42.29 ± 34.40	3317.7 ± 2350.7	462.05 ± 290.22
Klebsiella (n = 5)	40.40 ± 37.80	2389.1 ± 1936.5	247.35 ± 210.26
Group beta Streptococci (n = 6)	42.83 ± 41.82	2028.4 ± 1443	405.01 ± 344.63
Enterobacter (n = 5)	41.60 ± 34.48	2623.1 ± 3104.9	341.47 ± 299.08
P value	0.39	0.80	0.41
Sepsis onset
Early onset (n = 21)	30.29 ± 33.23	2283 ± 1666.07	345.2 ± 249.2
Late onset (n = 24)	38.25 ± 26.68	2395.3 ± 1785.1	333.8 ± 235.2
P value	0.158	0.829	0.666
Gestational age
Term (n = 24)	63.08 ± 3.52	2647.3 ± 1674.9	387.75 ± 251.19
Preterm (n = 21)	50.76 ± 2.88	1995.2 ± 1727.3	283.37 ± 216.95
P value	0.13	0.07	0.15
Birth weight
Normal birth weight (n = 32)	39.37 ± 32.73	2535.1 ± 1879.8	362.13 ± 262.7
LBW (n = 11)	21.27 ± 17.8	1528.4 ± 755.4	300.16 ± 170.1
VLBW (n = 2)	30.0 ± 8.48	3749.2 ± 1126.6	182.85 ± 52.11
P value	0.252	0.91	0.754
Sepsis outcome
Survivors (n = 40)	33.05 ± 29.53	1933.2 ± 1181.6	317.1 ± 221.6
Nonsurvivors (n = 5)	46.40 ± 33.06	5621.3 ± 1873.7	514.8 ± 326.2
P value	0.33	< 0.0001	0.23
Effect size (Cohen's d)	0.43	2.35	0.7

LBW: Low birth weight; VLBW: Very low birth weight; CRP: C-reactive protein; HBP: Heparin binding protein.

CRP, D-dimer, and HBP among the studied groups

CRP, D-dimer, and HBP were significantly higher in the proven sepsis compared to both the suspected sepsis and control groups. Moreover, these markers were significantly higher in the suspected sepsis group compared to controls (Figure 1A-C). Other serological markers of sepsis as WBCs and neutrophil percentages, were significantly higher in the sepsis group compared to controls, while platelets were significantly lower in the sepsis group than in controls, but they were comparable in the proven sepsis and the suspected sepsis groups (Figure 1D-F). Comparison of CRP, D-dimer, and HBP levels according to the causative organism, sepsis onset, gestational age, and birth weight showed no significance (Table 3).

Figure 1 Serological parameters of sepsis in the different studied groups. A: C-reactive protein; B: D-dimer; C: Heparin binding protein; D: White blood cells; E: Neutrophils; F: Platelets. Box and Whiskers plot. Minimum, 25^th percentile, median, 75^th percentile, and maximum are indicated. Horizontal line represents P value between the designated groups. The asterisk represents outliers. mean ± SD are plotted. CRP: C-reactive protein; HBP: Heparin binding protein; WBCs: White blood cells.

Diagnostic performance of sepsis biomarkers

D-dimer performed better than CRP and HBP in predicting sepsis vs controls [area under the curve (AUC) were 99.5%, 84.5%, and 91.03% respectively; Figure 2]. Univariate regression for markers of sepsis revealed that CRP, D-dimer, and HBP, but not CBC parameters, were significantly associated with sepsis compared to controls and associated with proven sepsis compared to suspected sepsis. In addition, multivariate regression revealed that D-dimer and HBP were independent predictors for sepsis compared to controls, while HBP was the only independent predictor of proven sepsis compared to suspected sepsis (Table 4).

Figure 2 Performance of serological markers in discriminating sepsis. A: C reactive protein; B: D-dimer; C: Heparin binding protein. Sensitivity is plotted on Y axis and specificity on X axis. Performance percentages are in the lower right box. Selection of cutoff values is based on the best performance (highest sensitivity and specificity simultaneously). AUC: Area under the curve; PPV: Positive predictive value; NPV: Negative predictive value.

Table 4 Univariate and multivariate logistic regression analysis for the independent parameters associated with sepsis.

Parameter	Univariate		Multivariate
	OR (95%CI)	P value	OR (95%CI)	P value
Sepsis vs controls
CRP (mg/L)	1.25 (1.09-1.43)	0.001	1.20 (0.96-1.51)	0.11
HBP (ng/mL)	1.04 (1.02-1.06)	0.001	1.03 (1.00-1.05)	0.04
D-dimer (ng/mL)	1.09 (1.01-1.18)	0.03	1.15 (0.93-1.42)	0.05
WBCs (× 103/mm3)	1.38 (1.23-1.56)	< 0.0001	1.28 (0.99-1.65)	0.054
Neutrophils (%)	1.16 (1.11-1.24)	< 0.0001	1.00 (0.89-1.12)	0.99
Platelets (× 103/mm3)	0.43 (0.0-8.48 × 1052)	0.99	NI
Proven vs suspected sepsis
CRP (mg/L)	1.03 (1.00-1.06)	0.03	1.01 (0.97-1.06)	0.48
HBP (ng/mL)	1.00 (1.00-1.01)	0.007	1.007 (1.00-1.01)	0.04
D-dimer (ng/mL)	1.00 (1.00-1.00)	0.01	1.00 (1.00-1.00)	0.21
WBCs (× 103/mm3)	1.03 (0.94-1.13)	0.45	NI
Neutrophils (%)	1.02 (0.98-1.07)	0.26	NI
Platelets (× 103/mm3)	0.99 (0.95-1.03)	0.57	NI

CRP: C-reactive protein; HBP: Heparin binding protein; WBCs: White blood cells; NI: Not included.

Correlation of D-Dimer and HBP with other biomarkers of sepsis

D-dimer and HBP levels had a significant positive correlation with CRP, WBCs, and neutrophils percentage, and had a significant negative correlation with platelets (Table 5).

Table 5 Correlation between sepsis parameters among the studied patients.

Variable	D-dimer (ng/mL)		Heparin binding protein (ng/mL)
	r	P value	r	P value
White blood cells (× 10³/mm³)	0.52	< 0.0001	0.5	< 0.0001
Neutrophils (%)	0.50	< 0.0001	0.5	< 0.0001
Platelets (× 10³/mm³)	-0.53	< 0.0001	-0.4	< 0.0001
C-reactive protein (mg/L)	0.64	< 0.0001	0.76	0.01
Heparin binding protein (ng/mL)	0.7	0.01

Prognostic value of sepsis biomarkers

D-dimer, but not CRP or HBP, was significantly higher in nonsurvivors compared to survivors (Table 3). D-dimer also outperformed nSOFA in predicting mortality in the sepsis group (AUC: 97.8 % vs 79.2%; P value < 0.0001 vs P value = 0.02, respectively) (Figure 3).

Figure 3 Performance of mortality predictors. A: D-dimer; B: Neonatal sequential organ failure assessment. Sensitivity is plotted on Y axis and specificity on X axis. Performance percentages are in the lower right box. Selection of cutoff values is based on the best performance (highest sensitivity and specificity simultaneously). nSOFA: Neonatal sequential organ failure assessment; AUC: Area under the curve; PPV: Positive predictive value; NPV: Negative predictive value.

Platelet counts as a potential confounder

Given the negative correlation between platelet count and D-dimer, platelet count was evaluated as a potential confounder in the association between D-dimer and mortality. Platelet count was included as a covariate in a multivariable regression model to adjust for its potential confounding effect. The strength of the association between D-dimer and mortality remained similar, suggesting that platelet count did not confound this relationship (Table 6).

Table 6 Univariate and multivariate logistic regression analysis for the independent parameters associated with mortality.

Parameter	Univariate			Multivariate
	OR (95%CI)		P value	OR (95%CI)	P value
Survivors vs nonsurvisors
Platelets (× 103/mm3)	0.987 (0.943-1.032)	0.569		0.987 (0.922-1.057)	0.716
D-dimer (ng/mL)	1.002 (1.0-1.004)	0.021		1.002 (1.0-1.003)	0.023

OR: Odds ratio.

DISCUSSION

CRP is frequently performed during sepsis workup. The current study showed that CRP was significantly higher in both proven and suspected sepsis than in the control group. At a cutoff value of > 6, it had a sensitivity of 75.6% and a specificity of 91.1% for discriminating sepsis (AUC = 84.5%). This result is in agreement with that of Younis et al[27] and Hamam et al[28]. In addition, Kumar et al[29] reported a similar performance of CRP in discriminating sepsis (AUC = 80%). However, elevation in CRP is not necessarily diagnostic of sepsis, as elevations may also occur as a physiologic phenomenon after birth or non-infectious conditions such as premature rupture of membranes, meconium aspiration, and perinatal asphyxia[30]. Therefore, it cannot be used as a sole marker for the diagnosis of sepsis[31].

We found that D-dimer levels were significantly elevated in both proven and suspected sepsis groups compared to the control group. El-Shahat et al[16] and others[32] reported similar results in neonatal sepsis. On the other hand, Brahmana et al[17] found that D-dimer levels were comparable between the neonatal sepsis and the control groups. This difference could be related to the gestational age of the recruited neonates, as 76% of Brahmana’s study population were preterm[17]. In preterm infants, the fibrinolytic system is still premature, which can result in coagulation disorder, and the fibrinolytic products can be detected normally. In contrast, in term infants after the first 24 hours of life, fibrinolysis breakdown products are typically at normal levels, and any increase may indicate pathological disorders such as thrombosis, bleeding, or sepsis[33].

In the current study, D-dimer was a significant independent predictor for sepsis. At a cutoff value of > 517.9 ng/mL, it had a sensitivity of 95.6% and a specificity of 97.8%. In agreement with our findings, Meini et al[34] reported that D-dimer was a significant predictor of sepsis. Furthermore, Kumar et al[29] reported a sensitivity of 90% and a specificity of 58%. Al-Biltagi et al[35] reported similar results but with lower sensitivity (72.7%) and specificity (86.7%) in diagnosing neonatal sepsis.

In our study, D-dimer levels showed a significant direct correlation with CRP, neutrophils, WBCs, and HBP and a negative correlation with platelet counts. This result is in agreement with that of Al-Biltagi et al[35] who reported that D-dimer levels had a significant direct correlation with CRP and negatively correlated with platelet count. This significant correlation may reflect early or developing disseminated intravascular coagulation associated with increased platelet consumption.

We further investigated the value of HBP in neonatal sepsis. HBP has extensive antimicrobial activity[31], and is rapidly released within 30 minutes following leukocyte membrane stimulation[36]. We found that HBP levels were significantly higher in proven and suspected sepsis groups compared to controls. This finding aligns with that of Deng et al[37]. Similarly, Huang et al[38] found a significant HBP elevation in patients with respiratory failure compared to those without. Additionally, Liu et al[18] found that pediatric ICU patients with severe sepsis had significantly elevated levels compared to those with sepsis.

In the current study, HBP levels showed a significant direct correlation with CRP, neutrophils, WBCs, and D-dimer and a negative correlation with platelet counts. In adults, Elsayed et al[39], reported a significant positive correlation between HBP and WBCs. Moreover, Linder et al[40] reported a positive correlation between HBP and each of WBC, neutrophils, SOFA score, and D-dimer in addition to a negative correlation with platelets in patients with sepsis.

We found that plasma HBP, at a cutoff level of > 137.55 pg/mL, had a sensitivity of 82.22%, and a specificity of 88.90% for the prediction of neonatal sepsis. This result agrees with that of Deng et al[37] who reported that HBP had a sensitivity of 80.4% and a specificity of 88.4% in predicting sepsis. Similarly, Zhou et al[41] found that HBP had a sensitivity of 84.9% and a specificity of 78.3% in diagnosing sepsis. Moreover, Yang et al[31] found that HBP could predict infection with 89% sensitivity and 73% specificity.

In the current study, 17 (37.8%) patients with sepsis were culture-negative (suspected sepsis). Despite the high sensitivity in detecting low bacterial loads, many studies have reported negative blood cultures when investigating sick infants[42]. De Prost et al[43] reported the occurrence of culture-negative sepsis in 41.5% of their cohort, which was associated with lower ICU mortality. However, identification of a pathogen was not independently associated with mortality in adjusted analyses. The reasons for culture-negative sepsis may include antibiotics intake before the onset of organ dysfunction, thus obscuring conventional cultures, or the possibility that sepsis is caused by unusual organisms that are difficult to identify in routine practice. Additionally, some patients might actually have a noninfectious cause for the clinical syndrome[43,44].

Our study demonstrated an overall mortality of 11.1% of the sepsis group, which represented 17.8% of the proven sepsis group. Although nonsurvivors in our study were restricted to the proven sepsis group, there was no significant difference between proven and suspected sepsis regarding mortality (P > 0.05). Similar to our findings, Sigakis et al[45] reported an overall mortality of 13%, with 12% for culture-negative and 16% for culture-positive patients. They noted that mortality is not impacted by culture positivity and that many clinical characteristics are similar between patients with or without microbiologically documented infection. The most important predictor of culture negativity was the receipt of antibiotics within the preceding 48 hours. A single-center study of ICU patients in Asia found that culture-negative patients had fewer comorbidities and a lower risk of death. After controlling for other factors, having a positive culture was not significantly associated with mortality[44]. Li et al[46] found that culture-negative and culture-positive patients with sepsis or septic shock demonstrated similar mortality rates. It may be hypothesized that the lower mortality in culture-negative sepsis is due to a milder insult and lower bacterial burden, and correspondingly, the inability to capture the microorganisms in cultures[44].

In contrast, Hazwani et al[47] demonstrated that the mortality rate in the culture-positive group was higher than that in the culture-negative group (43% vs 20% respectively). Another study found that culture-negative patients had an even higher burden of comorbidity and a higher risk of death[48].

Total mortality rates due to neonatal sepsis were estimated at 27.4%, reaching as high as 59%[49]. Neonatal sepsis is a primary cause of neonatal mortality within LMICs accounting for 99% of global neonatal mortality[50]. In the current study, we found that D-dimer, but not CRP or HBP, was significantly higher in nonsurvivors compared to survivors, and at a cutoff value of 3322 ng/mL, it had a sensitivity of 100% and a specificity of 92.5% in predicting mortality. Similar to our result, Rodelo et al[49] revealed that high D-dimer levels are associated with 28-day mortality. Although increased D-dimer values have been associated with worse clinical outcomes and death in some studies[16,19,51], others have failed to confirm such association, revealing that the prognostic value of D-dimer may be modest or poor in sepsis patients[20,52].

Identification and prediction of patients with a high risk of mortality may help in their proper management. Many clinicians recommend that the early introduction of adequate antibiotic therapy and resuscitation fluid decreases the mortality rate among patients with sepsis[53]. The mortality rate increases by up to 7.5% for every hour that proper treatment is delayed[54]. These patients, consequently, could require special attention upon admission to the emergency service[49]. The nSOFA scoring system is widely used to identify the presence of life-threatening organ dysfunction among preterm infants with sepsis[55]. When we compared D-dimer with nSOFA, we found that D-dimer performed better (100% sensitivity and 92.5% specificity) than nSOFA (80% sensitivity and 82.5% specificity) in predicting mortality in our study population.

In the current study, platelets were not associated with sepsis outcome (P = 0.716) but they had a significant negative correlation with D-dimer (P < 0.0001). Therefore, their inclusion in the regression model as a covariate was intended to control for their potential confounding effect. Fortunately, platelets did not change the strength of D-dimer in predicting mortality.

We found that HBP levels were comparable in survivors and nonsurvivors (P > 0.05). Therefore, it could not be used to predict mortality. Similar to our results, Katsaros et al[21] showed that HBP could not predict 28-day mortality in the emergency department and the plasma levels of HBP did not correlate with survival in non-septic and septic patients with shock. On the other hand, Liu et al[18] found that HBP can serve as a predictor for in-hospital mortality in the pediatric ICU. This apparent contradiction between our results and those of Liu et al[18] may stem from the larger number of patients in Liu et al’s study[18], as it is a multicenter one. In addition, the study population consisted of patients with sepsis and severe sepsis where the majority of mortality occurred in the severe sepsis group. Furthermore, the mean age was relatively higher in our study, which may affect HBP levels.

The limitations of the current study include the relatively small sample size, particularly for subgroup mortality analysis, and the fact that it is a two-center study. In addition, other markers of infection like PCT and lactate were not available for comparison, and there was no longitudinal follow-up of survivors. Future multicenter studies with larger sample sizes for validation are strongly recommended.

The strength of the study is that it is among the few prospective studies comparing HBP and D-dimer simultaneously in neonates.

CONCLUSION

D-dimer demonstrated superior diagnostic accuracy compared to CRP and HBP, and demonstrated superior prognostic accuracy compared to nSOFA in predicting sepsis outcome. The demonstrated significant performance of D-dimer in diagnosis and prognosis of neonatal sepsis, in addition to being simple and readily available test in most hospitals, suggest its feasibility to be included into sepsis diagnostic models.