Host Immunity Biomarkers of Neonatal Sepsis: a Prospective, Multicentre Study in Sub Saharan Africa

Background: Few biomarkers for sepsis diagnosis are commonly used in neonatal sepsis. Whilst host response is increasingly recognized in sepsis pathogenesis and prognosis, there is a need for evaluating the accuracy of new biomarkers targeting host response in regions where sepsis burden is high and medico-economic resources are scarce. The objective of the study is to evaluate diagnostic and prognostic accuracy of biomarkers of neonatal sepsis in Sub Saharan Africa. Methods: Prospective multicentre study conducted between April 2016 and March 2018 in three University hospitals and two sub-urbans healthcare centres in the south Benin region. Patients were followed from birth to 3 months of age. Accuracy of transcriptional (CD74, CX3CR1), proteic (PCT, IL-6, IL10, IP-10) biomarkers and clinical characteristics to diagnose and prognose neonatal sepsis were measured. At delivery, cord blood from all consecutive newborns were sampled and analysed, and infants were followed for a 12 weeks’ period. Results: From April 17, 2016 through March 12, 2018, a total of 581 newborns were enrolled. One hundred and seventy-two newborns developed neonatal sepsis (29.6%) and death occurred in forty-nine infants (8.4%). Among all tested biomarkers and clinical criteria, CD74/IP-10 ratio showed the best accuracy for neonatal sepsis diagnosis, while PCT accuracy was low. Among biomarkers and clinical criteria studied, only CD74 and PCT were independently associated with mortality, with CD74 showing an elevated predictive accuracy. Conclusion: Cord blood PCT had a low accuracy for diagnosing early onset neonatal sepsis in Sub Saharan African neonates, while CD74/IP-10 ratio had the best diagnostic accuracy. CD74 expression at birth had the best accuracy in prognosing sepsis mortality. biomarkers CD74, Cluster of differentiation 74; CX3CR1, C-X3-C motif chemokine receptor 1; PCT, Procalcitonin; IL-6/-10, Interleukin-6/-10; IP-10, Interferon inducible protein 10; CRP, C reactive protein; ; LMIC, Low- and middle-income country; SSA, sub-saharan Africa; GM, gestational malaria; E-/L-ONS, Early-/late-onset neonatal sepsis; W, week; AUC, Area under the cruve; GA, gestational age; mHLA-DR, monocyte Human leukocyte antigen-DR.

At birth, the immune defences go through rapid development and strengthen during the rst years of life [4]. Studies have shown that the increased risk of early and late neonatal infections observed in both term and preterm infants are related to impaired innate immune function [5][6][7]. This innate immaturity is characterized by altered cytokines production, decreased function of antigen-presenting cells and has many similarities to the immunosuppression phenotypes observed in critically ill adults and septic patients [8,9]. In addition to factors known to be associated with neonatal infection [10,11], frequent parasitic infections in SSA, such as malaria, toxoplasmosis and schistosomiasis, affect foetal and perinatal growth and development [12,13]. In low-income countries, the speci c example of gestational malaria (GM) shows that sequestration and cytoadhesion of infected red blood cells in the placental intercellular space lead to local in ammatory responses, altered placental function and subsequently promote intrauterine growth retardation and prematurity [14,15]. The in utero impact of GM on immunity in early childhood is recognized and may explain the altered immune response to vaccines and the increased risk of infection [16,17].
Blood culture remains the gold standard for the diagnosis of neonatal sepsis, but results are delayed and have low sensitivity due to maternal antibiotic treatment, small sample volumes, and low bacteremia [18,19]. Other tests routinely used for the diagnosis of neonatal sepsis, such as C-reactive protein (CRP) and blood count (total and differential white blood cell counts, absolute and immature neutrophil counts, and the ratio of immature to total neutrophils), are respectively non-speci c and non-sensitive [20][21][22][23]. Procalcitonin (PCT) has been shown to be an excellent marker for severe bacterial infection in newborn [24][25][26]. However during the rst 48 hours following birth, high PCT values may re ect a state of stress rather than the presence of a neonatal infection [25,27]. PCT therefore suffers, in some cases, of a lack of sensitivity for diagnosis of neonatal sepsis. Several studies have shown that some pro-in ammatory (IL-6, IP-10) and anti-in ammatory (IL-10) cytokines are acute phase biomarkers that can be used for the diagnosis of bacterial infection in neonatal. Levels of IL-6, IL-10 and IP-10 are signi cantly higher in neonatal sepsis than in healthy controls [28][29][30]. IL-6 and IP-10 are associated with in ammation in newborn [31,32], while IL-10 is associated with immunosuppression and mortality in adult septic patients [33]. However, those cytokine pro les may be in uenced by the gestational age and time of onset of infection in newborn [29,34]. Based on a microarray study, a panel of genes has been identi ed in critically ill septic patients whose expression on peripheral blood could effectively help stratifying patients at increased risk of secondary infection and/or death [35,36]. Among these genes, CD74 and CX3CR1 were shown to be associated with mortality and/or the occurrence of secondary infections [37][38][39]. Low expression of CX3CR1 is associated with mortality and immunosuppression in adult patients with septic shock [37,39]. Similarly, low expression of CD74 is associated with mortality in adult patients with septic shock while increasing expression of CD74 in the rst days of intensive care is associated with the occurrence of secondary infections [38,40]. There are no data on the accuracy of CD74 and CX3CR1 for the prognosis or diagnosis of neonatal sepsis, which hamper current use of these biomarkers in neonates and children [22,23,27,41].
Few studies have accessed neonatal sepsis epidemiology in malaria endemic SSA and characterized it with extensive diagnostic workup including systematic blood cultures, pathogen molecular diagnosis and patients biomarkers of sepsis and immune status [42]. In this study, we evaluated the accuracy of transcriptional (CX3CR1, CD74) and proteic (IL6, IL10, IP-10, PCT) biomarkers and their association with clinics and maternal risk factors to diagnose and prognose neonatal sepsis and obtained the reference range of those biomarkers during the rst three months of life.

Study design and participants
Participants were delivered in two sub-urban health centres (sub-urban arm) and three urban University hospitals (hospital arm) in the Abomey-Calavi, Sô-Ava and Cotonou districts in the South Benin region where malaria is hyper-endemic [43]. In both arms, only infants born from mother living in the Abomey-Calavi district were recruited to facilitate the follow-up and minimize effect of geographical origins. In the sub-urban arm, that includes only normal geatation with low-risk delivery (no maternal risk factors for infection), all consecutive births were included, whereas in the hospital arm only newborns born from mothers with maternal-foetal risk factors for infection (prematurity, prolonged rupture of the membrane, maternal fever) were included. In both arms, the exclusion criteria included maternal HIV positive status, major congenital malformation and refusal of consent. All children from both arms were followed clinically on a bi-monthly base during the rst 3 months of life. The follow-up consisted of scheduled home visits and unscheduled emergency visits if the infant was ill. The study protocol was approved by the local institutional review board (CER-ISBA 85 − 5). Written informed consent was obtained from parents.

Exposures and neonatal sepsis de nition
The exposure were occurrence of gestational malaria (GM). GM was de ned as a malaria infection during pregnancy or at delivery. For women in the suburban arm, malaria screening was performed at each scheduled prenatal visit using a thick blood smear. Mothers from the Hospital arm were screened only at the time of delivery. In this group, antenatal malaria was established on the basis of mother's anamnesis. For both study arms, placental blood smear and mother's peripheral blood smear were performed. The Lambaréné technique was used to quantify parasitaemia with a detection threshold of ve parasites per microliter [44].
Neonatal sepsis was suspected in neonates with more than two of the following criteria being present: neutrophil count < 7500/mm3 or > 14 500/mm3, band form > 1500/mm3, immature/total neutrophils ratio > 0.16, platelets count < 150 000/mm3 and CRP > 10 mg/L. Suspected neonatal sepsis was considered as clinical sepsis when the following clinical signs were associated: temperature irregularity; respiratory distress or apnoea; seizures, altered tonus, irritability or lethargy; vomiting, altered feeding pattern, ileus; skin perfusion alteration, haemodynamic signs (tachycardia, hypotension); hypoglycaemic/hyperglycaemic, hyperlactatemia or identi cation of focal infection such as soft tissue infection or conjunctivitis. All newborns with a clinical sepsis were subsequently adjudicated by one independent pediatrician (PT) and sorted into 'presumed sepsis' and 'de nite clinical sepsis' grouped as "adjudicated sepsis". In discordant cases, a second independent pediatrician (ULT) performed the nal adjudication with access, in addition to the full medical le review, to microbiological cultures results.

Biomarkers sampling
At birth and at follow-up visits, the clinical examination data of the children were collected. Blood samples were obtained at birth, then at week (W)1, W4, W8 and W12. The study protocol has been described in detail elsewhere [44]. Sampling and analytical methods are presented in Supplementary Appendix 2

Outcomes
The primary outcome was the diagnosis of clinical neonatal sepsis, and secondary outcome was mortality within the rst three months of life. Neonatal sepsis diagnosis was established by the local paediatrician based on the clinical examination of the child and initial laboratory workup including haemogram, C-reactive protein (CRP) and microbiological cultures (blood, cerebral uids and urine). Neonatal sepsis that occurred within the rst 72 hours following birth was considered as an early onset neonatal sepsis (EONS), and late onset (LONS) thereafter (for detailed algorithm for sepsis diagnosis, see published study protocol [44]).

Statistical analysis
An independent statistician (FB) (Soladis Inc. Lyon, France; https://www.soladis.com/ ) supported the statistical methodology and performed all analysis. Statistical analyses were performed using R software version 3.6.1. The variables were assessed for normality using Kolmogorov Smirnov test. Numbers and frequency were used for qualitative data and medians and IQR (inter-quartile range: [Q1-Q3]) for quantitative data. Qualitative variables were compared using the Chi-squared test (or Fisher's exact test for small expected numbers). The distribution of quantitative data was compared using Student's t-test (or the Mann-Whitney t-test when distribution was not normal or Welch test when homoscedasticity was rejected) if 2 groups were compared. If more than 2 groups, the distribution of quantitative data was compared using Anova test (or the Kruskal-Wallis test when distribution was not normal or when homoscedasticity was rejected).
To evaluate their diagnostic accuracy, data-driven analysis was performed. Selection of cut-off values or discrimination values de ning the positive and negative test results were performed. Several methods for selecting optimal cut-off values in diagnostic tests are proposed in the literature depending on the underlying reason for this choice. Here, we selected a cut-off to have a sensitivity of 0.95 and maximize speci city. This choice of cut-off was the same in the rest of the publication [45]. CD74/IP-10 was the score corresponding to the division of CD74 gene expression level by IP-10 serum concentration. We used three datasets to test more complex models, either genes (noted CX3CR1 & CD74), proteins (noted Protein biomarkers) or sets of biomarkers (noted All biomarkers). To avoid over tting and to compare our models, we used random sampling which takes place within each class and must preserve the overall distribution of data by class. To do this, we created a distribution, repeated 200 times, of 75/25% of the data. This distribution was used to optimize hyperparameters with package caret. Then we compared the average AUC and select the best average AUC for each dataset between all models. AUC accuracy were compared using Bootstrap approach. A p value < 0.05 was considered as signi cant. Statistical algorithms descriptions are displayed in Supplementary Appendix 2.
For comparison between clinical variables (VC) and biomarkers, we selected the clinical variables of interest following an expert opinion (PT) and corresponding to neonatal risk factors (eg. Gestational age, weight, maternal risk factors, multiple gestation, APGAR score) (Supplementary Table 1). In order to compare clinical and biomarker data, we transformed the data into the same referential to be able to compare them. We transformed the categorical clinical variables into a complete set of dummy variables [46]. Different transformations were then applied to the data set. A Yeo-Johnson transform is a non-linear transformation that reduces skewness and approximates a normal law. We centred (subtracts the mean of the variable's data) and scaled data (divides the standard deviation).
For the establishment of heatmap, Partial Last Squares (PLS) Regression was used to compare the two datasets. This algorithm comes from the mixOmics package. Biomarkers were de ated with respect to the information extracted/modelled from the local regression on Clinical Variables. Consequently, the latent variables computed to predict Biomarkers from Clinical Variables are different from those computed to predict Clinical Variables from Biomarkers. One matrix Clustered Image Map (CIM) is a 2dimensional visualization with rows and/or columns reordered according to some hierarchical clustering method to identify interesting patterns. The CIM allows to visualize correlations between variables. Generated dendrograms from clustering were added to the left side and to the top of the image. The used clustering method for rows and columns is the complete linkage method and the used distance measure is the distance Euclidean. We showed only variables with co-variances greater than max(covariance)/2.

N-integration and feature selection with Projection to Latent Structures models (PLS) with sparse
Discriminant Analysis was used to compare the two datasets and outcome (Circosplot). This algorithm comes from the mixOmics package. The circos plot represents the correlations between variables of different types, represented on the side quadrants. We showed within and between connexions between blocks, expression levels of each variable according to each class. The circos plots were built based on a similarity matrix, extended to the case of multiple data sets. We showed only variables with co-variances greater than max(correlation)/2 [46].

Cohort description
Between April 17, 2016 and March 12, 2018, a total of 581 newborns were enrolled, 420 (72.3%) in the hospital arm and 161 (27.7 %) in the sub-urban arm. Patients' characteristics are displayed in Table 1.
Median gestational age (GA) was 38.4 weeks (95% con dence interval [CI], 35.5 to 40 weeks), with a median birth weight of 2816 grams (95%CI, 2400 to 3150 grams). One hundred and eighty-four (31.6%) infants were born prematurely with thirty-four (5.8%) being less than 32 weeks of GA. More preterm infants were delivered in the hospital arm than in the sub-urban arm (41.2% vs 6.2%, p < 0.0001) and had a lower birth weight (2700 grams vs 3006 grams, p < 0.001). Forty-ve multiple gestations (40 twins, 5 triplicate) happened. Gestational malaria occurred in forty-three mothers (8.1%), affecting 49 infants (8.4%). Intermittent preventive treatment of malaria was performed in 470/531 (88.5%) of all mothers. According to the study design, main differences between arms were related to the perinatal risks with more premature infants and maternal risk factors of infection in the hospital arm. In total, 432/581 (74.3 %) babies were born from mothers with maternal risk factors of infection (386/531, 72.7%). One hundred and seventy-two infants (29.6%) developed a clinical sepsis according to the de nition. All occurred in the hospital arm. Among these, adjudication con rmed 168 cases (97.7%) (Fig. 1). All but ve clinical sepsis were EONS. In the hospital arm, 307/420 infants were hospitalized following birth, the remaining with uneventful delivery and normal perinatal adaptation were discharged. Only one infant from the sub-urban arm was hospitalized for prenatally undiagnosed omphalocele. The median hospital stay was 4 days (95% CI, 2 to 7 days). The global mortality was 49/581 (8.4%) with a signi cant difference between hospital and sub-urban arms, respectively 11.4 % (48/420) versus 0.62% (1/161) (p < 0.001). Most death occurred in the rst week of life (median 2 days) and were related to neonatal sepsis (n = 44) or prematurity complications (n = 5).  Cord blood levels of studied biomarkers are displayed in Table 3. Procalcitonin level was signi cantly higher in infants developing neonatal sepsis compared to both sub-urban arm and hospital healthy infants ( Fig. 2A). A signi cant difference in cord blood from non-septic neonates between both arms was observed. Diagnosis accuracy of cord blood PCT was low with an AUC of 0.62 (95% CI, 0.56 to 0.67) (  (Fig. 3A). However, CD74 and PCT were strongly correlated (correlation cut-off <-0.5 or > 0.5) with premature rupture of membrane (PROM), maternal fever, Apgar score, heart and respiratory rate (Fig. 3A). More speci cally, CD74 and PCT were inversely correlated with Apgar score, maternal fever, and to a lower extend with gestational age (Fig. 3B). At delivery, PCT, CD74, IL-6 and CX3CR1 levels were signi cantly different between survivors and non survivors (Fig. 4A). Prognosis accuracy of CD74 was elevated (AUC 0.78; 95% CI, 0.68 to 0.87) and low for CX3CR1 (AUC 0.51; 95% CI, 0.40 to 0.61) (Fig. 4B). CX3CR1, PCT and IP-10 levels were signi cantly different in survivors from both arms, suggesting that clinical conditions (maternal risk factors, prematurity) may affect those biomarkers (Fig. 4A). Out of studied biomarkers, only CD74 and PCT were associated with mortality in a partial least squares regression model. CD74 was positively correlated with Apgar score, gestational age, height and weight (Fig. 5). Interestingly models combining clinical criteria and biomarkers had elevated accuracy for mortality, but were statistically similar. Model associating CD74 expression, Apgar score at 1 minutes and abnormal amniotic uid had an AUC of 0.86 (95%CI, 0.79 to 0.92).

Biomarkers reference range in the rst 12 weeks of life
In healthy newborns, all biomarkers but PCT had a signi cant lower value on cord blood than in the rst week following birth (Fig. 6). Whereas most biomarkers range did not vary between weeks 1 and 12, CD74 progressively increased during the rst eight weeks of life. Following neonatal sepsis, at week 4, CD74 was signi cantly lower in septic than in healthy newborns. We did not observe any signi cant differences in biomarker pro les between newborns exposed or not to gestational malaria.

Discussion
This multicentre study performed in sub Saharan Africa showed that cord blood PCT has a low accuracy for diagnosing neonatal sepsis whilst CD74/IP-10 ratio had the best diagnostic accuracy among tested biomarkers and clinical criteria. In addition, the study showed that CD74 accuracy was elevated in prognosing mortality. The evaluation of clinical and biomarkers for neonatal sepsis diagnosis and prognosis in setting with high prevalence of sepsis and limited resources is bringing for the rst time detailed assessment of potential biomarkers utility and establishment of reference values in neglected pediatric populations. As such, this study is a radical breach in biomarkers development paradigms by refocusing their evaluation in setting with high prevalence of neonatal sepsis. Several large studies, occurring in setting with high resources and low sepsis prevalence, have suggested that cord blood PCT may be a robust biomarker for neonatal sepsis diagnosis, reporting elevated sensitivity and speci city for cut-off values ranging between 0.6-0.7 ng/mL [47][48][49]. These encouraging data were not con rmed by all. Few series in extreme premature infants and in setting where EONS prevalence was higher showed lower sensitivity and speci city ranging between 48.7 to 69% and 68.6 to 70%, respectively [50,51]. Our study con rms the inaccuracy of cord blood PCT in diagnosing neonatal sepsis. The lack of speci city that we observed may be related to the effect of maternal and neonatal medical or environmental conditions on PCT levels. In a large series involving 2'151 infants (26 with EONS), Joram et al. identi ed gestational age (28-32 weeks) and pH < 7.10 to be the only factors associated with increased PCT levels [48]. In our series, PCT was strongly correlated with PROM, maternal fever, Apgar score, gestational age, infant heart and respiratory rate, suggesting that prenatal condition and newborn adaptation impact PCT level as displayed by a signi cant PCT level difference in healthy infants born from either the hospital (with maternal risk factors) or the sub-urban arm (no maternal risk factors). Interestingly, CD74/IP-10 ratio was shown for the rst time to have a better accuracy for sepsis diagnosis than any studied biomarkers and clinical combinations. CD74, the invariant chain involved in MHC class II molecules transport, is identi ed as a prognosis marker in adult critically ill patients developing healthcare associated infections [40]. In our study, CD74 expression alone has a low accuracy for neonatal sepsis diagnosis, but its association with IP-10, the interferon-γ inducible protein-10, emerges as a strong combination of biomarker (AUC 0.77, 95% CI: 0.72 to 0.81). IP-10 is associated with various in ammatory conditions such as autoimmune diseases, hemophagocytic syndromes and viral infections where its combination with phospholipase A2, or IL-10 were suggested for sepsis diagnosis [31,[52][53][54]. Combination of CD74 and IP-10, illustrating both antigen presentation ability and response to type II interferon stimulation, may be representative of both faces of neonatal immunological competency to infection. Ontogeny of antigen presentation in early life is thought to be a key factor in determining age-speci c responses to microbes and other antigens. IP-10 production in infants re ects a consequent Th1 response and as such, it may represent in conjunction with a reduced CD74 expression, despite IFNγ stimulation, a signature for severely dysregulated response associated with neonatal sepsis.
Although well recognized in cancer, few publications relate CD74 utility in sepsis, none in children and neonates [55,56]. In contrast, in a study of critically ill septic adults, low expression of CD74 was frequently shown to be independently associated with mortality (OR 3.4, 95% CI: 1.2 to 9.8, p = 0.026) 38 whereas secondary infection occurrence was associated with increased CD74 expression in the rst day following admission for septic shock [40]. Our study further extend CD74 prognosis utility in neonatal sepsis. CD74 expression on cord blood of < 9.47 cutoff value (Youden index 0.09) was predictive of death in neonates with a sensitivity of 95% and a speci city of 13%. Indirectly those results con rm those related to histocompatibility leukocytes antigen-DR expression on monocytes (mHLA-DR) that is known to be highly correlated (r = 0.87) with CD74 expression [38]. Low mHLA-DR was shown to be associated with EONS in premature infants and mortality following LONS [57,58]. The underlying pathophysiologic rational behind the association of low CD74/low mHLA-DR in infants and sepsis mortality may be resumed by an altered antigen presentation, either acquired or secondary to a perinatal event. Primary defect or decreased expression of mHLA-DR was shown to be associated with prematurity and altered myeloid cells functionality in extremely immature infants [7,59]. Secondary decrease of mHLA-DR in response to diverse aggression is well recognized. Decrease in mHLA-DR expression was demonstrated following sepsis, and is recognized as one of the main immunological caracterization of post-infective immune failure [60-62]. As such, the prolonged alteration of mHLA-DR was shown to be associated with mortality in adults with sepsis [60, 63-65]. Serial mHLA-DR kinetics in septic newborns is largely unknown although a rise in HLA-DR negative monocytes following severe sepsis is reported [66]. Interestingly, in our study, infants surviving sepsis showed a prolonged alteration of CD74 expression up to one month following EONS. Nevertheless, such prolonged post-sepsis reduction in CD74 transcription, as well as reduced mHLA-DR has never been described in both infants and adults. Newborns are known to have a rapidly evolving immunity during the rst months of life

Conclusion
In this study cord blood PCT was found to have a low accuracy for neonatal sepsis diagnosis and CD74 to be predictive of mortality in infants with neonatal sepsis. Kinetics of CD74 expression following neonatal sepsis showed a prolonged impairment in survivors. In addition, CD74/IP-10 ratio was found to have a good accuracy for neonatal sepsis diagnosis, depicting a novel signature for defective Th1 response and altered antigen presentation seen in sepsis and sepsis-associated immune depression.

Declarations
Ethics approval and consent to participate: The study protocol was approved by the local institutional review board (CER-ISBA 85-5). Written informed consent was obtained from parents.

Consent for publication:
Not applicable Availability of data and materials: Data are available upon request to the corresponding author.
Competing interest: JYM, LV, AP are employed by an in vitro diagnostic company, bioMérieux SA. The remaining authors declare that this research was conducted in the absence of any commercial or nancial relationship that could cause potential con ict of interest.     Cord blood biomarkers level according to neonatal sepsis prognostic A) Transcriptional and protein biomarkers expression in sepsis non survivors neonates and survivors neonates. CD74 and CX3CR1 mRNA level were evaluated by RT-qPCR with ABI7500 fast and plasma PCT; IL6; IL10 and IP-10 concentrations were measured by multiplexed assay with the Ella platform in cord blood sample obtained from non-surviving neonates and surviving neonates.   Biomarkers kinetics during 3 months of follow-up The gure shows the biomarkers pro le during the rst 3 months of life in healthy infants with median value (in blue), 95%con dence interval (in gray), and median value in septic infants (in red). Biomarkers were measured at delivery in cord blood (Cord) and at rst, four, eight and twelve weeks in peripheral blood. Data are expressed as medians. Anova test was used to compare biomarkers levels.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.