***ECD*** maternal C-reactive protein as a predictor of neonatal sepsis

ABSTRACT

Systemic bacterial infection in the newborn has a significant impact on neonatal mortality and morbidity. Non-invasive prenatal markers of risk could be useful in the prediction and prevention of neonatal sepsis. We evaluated the association of maternal third-trimester serum level of C-Reactive Protein (CRP) with neonatal sepsis in a sample of infants in the JAKids pregnancy and birth cohort study. A population-based nested case–control design was used to identify cases and controls of neonatal sepsis from the subset of infants in the JAKids study whose mothers had serum archived in the early third trimester and who were admitted to newborn intensive care. Cases were 25 neonates with neonatal sepsis identified from hospital records. Controls were a random sample of 62 sepsis-free neonates matched to cases within three gestational age strata – ≤32 weeks, 33–36 weeks, and ≥37 weeks.

Mothers of neonatal sepsis cases ≥37 weeks had significantly higher mean levels of maternal CRP protein than mothers of controls (11.0 mg/dL ± 3.0 vs. 8.7 mg/dL ± 5.9; p < .05). Differences in maternal CRP were not found in sepsis cases born ≤32 weeks (9.5 mg/dL ± 4.2 vs 5.8 mg/dL ± 4.0, p = .23) nor in sepsis cases born at 33–36 weeks (9.0 mg/dL ± 3.6 vs 11.9 mg/dL ± 7.8, p = .34). Maternal third-trimester C-reactive protein levels were elevated in mothers of term-born neonates with sepsis, but not in the mothers of preterm neonates with sepsis.

KEYWORDS:

• Neonatal sepsis
• maternal serum
• c-reactive protein

Introduction

Sepsis is a clinical syndrome characterized by a set of hemodynamic, respiratory and metabolic alterations secondary to an infectious process that can trigger an abnormal systemic inflammatory response syndrome (SIRS). Neonatal sepsis refers to this process occurring during the first four weeks of life to infants of any gestational age at birth (Dessi et al., 2014). Neonates are predisposed to infection during the perinatal period both because of their high burden of exposure in the perinatal period and their relatively undeveloped immune system. The burden of disease attributed to neonatal infections varies by geographic region and by maternal and neonatal risk factors, but neonatal infections are estimated to cause approximately 36% of the estimated 4 million neonatal deaths annually in the world (Shane & Stoll, 2014). Neonatal sepsis is the third leading cause of neonatal mortality, behind only prematurity and intrapartum-related complications (Wang et al., 2014; Zea-Vera & Ochoa, 2015).

In recent years, a significant decrease in childhood mortality has been achieved worldwide. However, neonatal mortality has decreased at much lower rates, and currently represents 40% of all childhood mortality (Liu et al., 2012; Wang et al., 2014). Infants with neonatal infections are more likely to have adverse neurodevelopmental outcomes at follow-up, including cerebral palsy, lower mental and psychomotor development index scores, visual impairment and impaired growth, increasing the social and economic burden of this condition (Zea-Vera & Ochoa, 2015).

To decrease morbidity and mortality, and to improve the outcome of infants with sepsis, reliable identification of sepsis at an earlier stage is paramount (Ng et al., 2018; Shah & Padbury, 2014). Yet one of the major difficulties in the management of neonatal sepsis is getting an early and accurate diagnosis. Unlike older patients, newborns with sepsis have subtle presentations, and multiple conditions can resemble neonatal sepsis (Zea-Vera & Ochoa, 2015). Routine laboratory tests such as white blood cell (WBC) count, absolute neutrophil count, immature/total neutrophil ratio have low sensitivity. The ‘Gold Standard’, blood culture, is often negative in newborns because of the difficulty of obtaining sufficient blood and frequent prenatal administration of antibiotics, which can obscure blood culture findings (Su et al., 2014).

In recent years, much research on diagnostic and predictive tests for neonatal sepsis has focused on cytokines associated with inflammatory response, but nearly all such studies have been in neonates, not in pregnant women. C-reactive protein, an acute phase reactant synthesized by the liver, is one of the most extensively studied and most frequently used laboratory tests for the diagnosis of neonatal sepsis (Jeon et al., 2014; Shah & Padbury, 2014). The study we conducted is an attempt to see whether the risk of neonatal sepsis is raised by elevation of maternal C-reactive protein level in the third trimester of pregnancy. If information strongly predicting risk were available prior to birth, appropriate preventive activities, such as prophylactic use of antibiotics, might be employed.

To investigate the association of a key maternal inflammatory marker, c-reactive protein (CRP) we designed a nested case–control study design of neonatal sepsis nested in a cohort of women with prenatal serum archived.

Materials and methods

Population of JAKids study

Briefly, the JA Kids Birth Cohort Study is the national birth cohort study of Jamaican children with the goal of improving the health and well-being of Jamaica’s children by providing data on the relationships among a wide range of family, school, community, environmental and individual variables in relation to child health and development. It provides national data on maternal health and well-being, pregnancy, paternal well-being and involvement, children’s status at birth and at various points thereafter. This pregnancy and birth cohort consists of all the children born in all fourteen parishes across Jamaica from July 1 to 30 September 2011. Eighty-seven percent (87%) of all mothers delivering in Jamaica during the study period were enrolled. In total, 9,766 mothers and 3,413 fathers were enrolled in the study, 5,204 of them antenatally, at 20–28 weeks of pregnancy with expected delivery dates between July and September 2011. Subsequently, contact was made with parents and their children at birth, 9 months, 12 months, 18 months and 24 months. (JA Kids; https://www.mona.uwi.edu/fms/jakids/)

Sub-population with maternal blood samples

When the study was underway, the opportunity arose to obtain blood samples from a subset of the pregnant women from 28 weeks of gestation onwards. Samples of serum and plasma were successfully obtained from 1,186 study participants and stored in −20 C freezers locally and then in −80 C freezers in Kingston, Jamaica. This archive of maternal serum and plasma allows for a variety of approved research uses under the highest level of security and confidentiality standards.

Case identification

Among the approximately 2,000 neonatal (first 28 days of life) admissions to neonatal care unit across the Island (JAKIDS, 2018), 505 cases of neonatal sepsis were identified utilizing ICD code P369 (10th edition) from medical records across Jamaica. Of 505 records with neonatal sepsis, 37 records were identified as being from the JAKids cohort and were linked to mothers who were recorded as having blood specimens. We excluded 12 infants for whom either the serum sample could not be located or in whom gestational age information was missing, leaving 25 cases for study. All cases and controls were singleton births.

Control Identification

Controls were children without sepsis selected from within the JA Kids Birth Cohort mothers who had banked serum samples, and who were also among the 1,495 infants who had been admitted to newborn care but who did not carry the diagnosis of sepsis. We identified 104 records of infants without sepsis whose medical records could be linked to their mothers, and in whom JAKids records indicated that a blood specimen was banked. Of the 104 potential controls, 62 remained for inclusion using the same inclusion criteria (sample located, gestational age available) as for cases.

Matching

Because preterm birth is a risk factor for neonatal sepsis, we separated cases into three categories of gestational age at birth (37 weeks and above, 33–36 weeks and up to 32 weeks) and matched controls within each of the three gestational age strata. The infants were also de-facto matched for NICU admission, because we used NICU admissions as the source of both cases and controls. This allowed us to match for severity of illness to some extent, but may have increased the possibility of a type 2 error, if other conditions in the NICU were associated with maternal inflammation.

Laboratory methods

Maternal serum levels of C-reactive protein were determined by immunoassays (RayBiotech Inc., Norcross, GA) utilizing an in vitro enzyme-linked immunosorbent assay (ELISA) for the quantitative measurement of human C-reactive protein in serum. Standard serum specimens were used in the laboratory run and analyte value was determined by interpolation from standard curves. The inter-and intra-assay coefficients of variation obtained in the laboratory were less than 12% and 10%, respectively. The sensitivity of the assays was 34 pg/mL for Human C-reactive protein (RayBio, 2012). Yoon et al. and Jeon et al. used antibody adsorption-particle agglutination assay and nephlometry using the Beckman Array System protein analyzer respectively for CRP analysis.

Data analysis

The laboratory levels of C-reactive protein in mothers were linked to records of neonates admitted to the neonatal care unit who were either cases or controls. Neonatal characteristics were compared between cases and controls. Birth weights were compared among the three gestational groups. The Chi-square test and Student t test were used to estimate the statistical significance of the comparisons for categorical and continuous variables, respectively. Means were calculated for cases and controls of each gestational group (Table 1). Gestational age is a significant effect modifier for neonatal sepsis; therefore, results are presented separately for each group. Birth weight, APGAR score at 1 minute and 5 minute were also examined.

Table 1. Demographic and delivery data in different groups of gestation weeks.

Characteristics	Group 1 (GA = 37 weeks and above)		Group 2 (GA = 33–36 weeks)		Group 3 (GA = upto 33 weeks)
	Case (n = 15)	Control (n = 47)	Case (n = 5)	Control (n = 11)	Case (n = 5)	Control (n = 4)
Gestational age (Week)	38.1 ± 1.2	39.1 ± 1.1	34.6 ± 1.1	35.1 ± 1.0	29.8 ± 1.5	28.8 ± 2.2
Male(%)	66.67	61.70	20.0	54.55	20.00	50.00
Birth Weight (g)	2947.7 ± 816	3328.7 ± 672	1980 ± 380	2476 ± 567	1396 ± 232	1625 ± 765
Apgar Score (1 min)	8.1 ± 1.6	7.5 ± 2.4	7.8 ± 2.7	6.5 ± 2.5	7.6 ± 2.2	6.3 ± 3.2
Apgar Score (5 min)	9.0 ± 0.4	8.5 ± 1.7	8.2 ± 1.8	8.0 ± 2.0	7.8 ± 2.2	7.3 ± 2.1

All statistical analyses were conducted using software (SAS, Version 9.3 for Windows; SAS Institute, Cary, NC). The study protocol was approved by the institutional review board of Michigan State University.

Results

Demographic and delivery data were compared between cases and controls of different groups of gestational weeks (Table 1). The gestational age did not differ significantly between cases and controls after matching. Cases had lower birthweights than controls in all the three gestational groups (2947 g ± 816 vs. 3328 g ± 672 in group 1 with gestation week 37 weeks and above; 1980 g ± 380 vs 2476 g ± 567 in group 2 with gestation week 33–36 weeks and 1396 g ± 232 vs. 1625 g ± 765 in group 3 with gestation week 32 weeks and less). No significant differences were observed between cases and controls in any gestational group with regard to APGAR score at 1 minute and at 5 minutes. Controls were somewhat more likely to be males.

Cases had significantly higher mean level of C-reactive protein than controls in the group of gestation week 37 weeks and above (11.0 mg/dL;95% CI, 9.4–12.7 vs. 8.7 mg/dL;95% CI,6.9–10.4; p < .05). Among infants of 32 weeks of gestation and less (9.5 mg/dL; 95% CI,4.2–14.7 vs 5.8 mg/dL; 95% CI, 0.0–12.1), mothers of cases also had higher levels of CRP than control mothers, but the finding did not achieve statistical significance. In infants born at gestation weeks 33–36, CRP in mothers was not elevated (9.0 mg/dL;95%CI, 4.5–13.5 vs. 11.9 mg/dL;95%CI, 6.6–17.1; Table 2).

Table 2. Standard error and 95% confidence intervals of maternal CRP in different gestational age groups of cases and controls.

Outcome variable of Interest	Group 1 (GA = 37 weeks and above)		Group 2 (GA = 37 weeks and above)		Group 3 (GA = upto 32 weeks)
	Case (n = 15)	Control (n = 47)	Case (n = 5)	Control (n = 11)	Case (n = 5)	Control (n = 4)
CRP (mg/dl)
Mean	11.0 ± 3.0	8.7 ± 5.9	9.0 ± 3.6	11.9 ± 7.8	9.5 ± 7.8	5.8 ± 4.0
Standard error	0.8	0.9	1.6	2.4	1.9	2
Confidemce Interval	9.4–12.7	6.9–10.4	4.5–13.5	6.6–17.1	4.2–14.7	0.0–12.1
P Value	<0.05		0.34		0.23

Values are given as the mean ± standard deviation. CRP, C-reactive protein.

Discussion

In this gestational-age matched nested case–control study of neonatal sepsis, we found that C-reactive protein in maternal serum collected from 28 weeks of gestation onwards was significantly higher in term-born infants, and trended towards being higher in very premature cases of sepsis than in matched controls. This finding was, however, not replicated in moderately premature cases of sepsis. To our knowledge, this is the first study to compare maternal C-reactive protein in the third trimester of pregnancy in infants with and without neonatal sepsis, that is not restricted to premature or low birthweight newborns.

Three studies have found associations between third trimester maternal C-reactive protein and neonatal sepsis, but in high-risk populations of mothers and infants and with measurements of CRP closer to the time of birth. Yoon et al. reported that maternal CRP during preterm labor in women with intact membranes was significantly increased (2.1 mg/dL vs 0.4 mg/dL) in mothers whose infants later had neonatal sepsis (Yoon et al., 1996).

Jeon et al. also studied a select population, infants below 37 weeks gestation admitted to intensive care (mean gestational age 32 weeks ± 3.2, mean birth weight 1,887 g ± 623) and they found maternal C-reactive protein, measured in the peripartum period, to be significantly higher in mothers of infants with neonatal sepsis than in controls (3.55 mg/dL ± 2.69 vs. 0.48 mg/dL± 0.31, p= 0.0001; Jeon et al., 2014).

Lee et al. conducted a retrospective cohort study of women with preterm labor or preterm premature rupture of membranes (23–35 week of gestation) and found that a CRP ≥ 8 mg/dl measured 72 hours or less prior to birth carried a significantly higher risk of early neonatal sepsis (RR = 2.81 (95% CI 1.21–6.51; Lee et al., 2012). Thus we find support for our hypothesis that maternal CRP may be a predictor of neonatal sepsis, but thus far only in premature infants. Our study extends the above findings to term infants.

Strengths of this present study include its use of a nested case–control design, making use of a serum data bank of maternal third trimester blood specimens, to concurrently compare the distribution of maternal C-reactive protein levels in three different gestational age groups and its association with neonatal sepsis. Gestational age groups were well defined and the outcome was restricted to neonatal sepsis, and cases and controls were selected from the same birth cohort. Unlike previous studies, we assessed CRP long before birth, during prenatal care, and did not restrict our attention to prematurely born babies. Our study is particularly innovative in that it examined, for the first time, the link of maternal CRP to neonatal sepsis in term-born infants.

Despite these strengths, some study limitations need to be mentioned. First, the sample size was small. Second, the study used maternal serum specimens that had been stored for up to a few months at −20°C before long-term storage at −80°C at the University of the West Indies in Kingston, Jamaica. Biomarker stability may possibly be stronger at storage temperature of −80°C than at −20°C (Hubel, Spindler & Skubitz, 2014), but epidemiologic studies with long-term storage of maternal serum at −20°C have yielded serum protein levels suitable for research (Trabert, Longnecker et al., 2011). Moreover, the levels of CRP found in this study were not lower than found in other studies, an observation which does not support concerns about analyte decay. The impact of storage temperature on analyte values in this study is unknown; however, both case and control specimens were handled in a similar way, therefore, differential effect (bias) across case–control and different gestational groups is unlikely.

Third, we selected controls from NICU admissions, and this could have reduced the possibility of finding differences compared to a study in which otherwise healthy newborns served as controls. Had resources been available, we would have created another matched sample derived from JAKids babies who had not been admitted to the NICU, a procedure that would have been feasible in all but the smallest gestational age grouping, where NICU admission is universal.

A fourth consideration is that analytes other than C-reactive protein might be of interest in infants with neonatal sepsis (Lee et al., 2012; Sharma et al., 2018; Su et al., 2014). Future studies should assess additional inflammatory and infectious biomarkers in pregnancy. While prenatal banked specimens provided a unique opportunity to investigate the study questions, the one-time-only blood collection limited our ability to fully understand the concentration change over the course of pregnancy and labor and delivery. Such a longitudinal investigation would represent an important next step.

In summary, this study suggested a relationship between neonatal sepsis and maternal third trimester C-reactive protein and indicated the possibility that this effect might be differential by gestational age. Our study results point to the need for further research in a larger population group stratified by gestational age to further examine maternal pregnancy biomarkers which might permit antenatal prediction of neonatal sepsis, and thereby improve outcomes in neonatal sepsis.

Acknowledgments

The authors would like to thank Angella-Turner-Dawkins and Erica Jones-Edwards from the Department of Basic Medical Sciences, UWI, Mona for their analysis of serum samples.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the National Institutes of Health (NIH) Research Training Program in Perinatal Epidemiology [T32-HD046377] at Michigan State University (MSU). Dr. Manandhar initiated this work while she was a post-doctoral fellow in this program.