Blood group AB increases risk for surgical necrotizing enterocolitis and focal intestinal perforation in preterm infants with very low birth weight

Necrotizing enterocolitis (NEC) and focal intestinal perforation (FIP) are two of the most common emergencies of the gastrointestinal tract in preterm infants with very low birth weight (VLBW, birth weight < 1500 g). Identification of risk factors among these children is crucial for earlier diagnosis and prompt intervention. In this study, we investigated a relationship between ABO blood groups and the risk for surgical NEC/FIP. We genotyped the ABO locus (rs8176746 and rs8176719) in VLBW infants enrolled in a prospective, population-based cohort study of the German Neonatal Network (GNN). Of the 10,257 VLBW infants, 441 (4.3%) had surgical NEC/FIP. In univariate analyses, the blood group AB was more prevalent in VLBW infants with surgical NEC/FIP compared to non-AB blood groups (OR 1.51, 95% CI 1.07–2.13, p = 0.017; absolute risk difference 2.01%, 95% CI 0.06–3.96%). The association between blood group AB and surgical NEC/FIP was observed in a multivariable logistic regression model (OR of 1.58, 95% CI 1.10–2.26, p = 0.013) as well. In summary, our study suggests that the risk of surgical NEC and FIP is higher in patients with blood group AB and lower in those having non-AB blood groups.

www.nature.com/scientificreports/ system 24 . In contrast, maternal chorioamnionitis 25 and exposure to postnatal steroids with or without indomethacin treatment 26 are associated with FIP. In 2012, Thomson et al. showed that neonatal blood group AB is associated with an increased mortality in preterm infants with NEC 27 . Blood group antigens are not only on the surface of red blood cells but also occur in other tissues, including the intestinal surface. Moreover, they can be secreted into the lumen of the gut (and breast milk)-for ABO, for example, if the person carries another specific genetic variant in the gene FUT2 and is a so-called "secretor"-where these sugars serve as nutrients for intestinal microbes. Blood group antigens also serve as receptors for toxins, parasites, and bacteria, facilitating their colonization or invasion. They can also serve as false receptors, preventing binding to a target tissue. ABO antibodies contribute to the innate immune system and attack some bacteria or viruses that carry ABO antigens 28 .
In this study, we utilized a large cohort study of the German Neonatal Network (GNN) consisting of 10,257 VLBW infants to investigate the potential relationships between neonatal blood group AB and risk for developing surgical FIP and NEC.

Results
Baseline characteristics of study population. Of the 10,257 VLBW infants enrolled in this study, 3538 (34.5%) neonates had blood group O, 4773 (46.5%) blood group A, 1332 (13.0%) blood group B, and 614 (6.0%) blood group AB. Among them, there was no difference in gestational age, birth weight, gender, mode of delivery, and antenatal corticosteroid exposure (Table 1). However, differences in blood group distribution were observed for multiple birth (blood group B [390 multiples of 1332, 29 Blood group AB and risk for NEC/FIP. In univariate analyses, preterm infants with blood group AB were at higher risk for surgical NEC/FIP compared to the non-AB-groups (OR 1.51, 95% CI 1.07-2.13, p = 0.02; absolute risk difference 2.01%, 95% CI 0.06-3.96%). When the risk for surgical NEC/FIP or death was analyzed, the difference between the AB-and non-AB-group persisted ( Table 2). In a multivariable logistic regression model which included covariates such as sex, gestational age, birth weight, multiple birth, medical and surgical PDA treatments, we could observe a significantly higher risk of blood group AB for NEC/FIP when compared to non-AB-blood groups (OR 1.54, 95% CI 1.10-2.26, p = 0.01) (Fig. 1). We also found that medical treatment of PDA using indomethacin or ibuprofen was associated with increased risk for developing NEC/FIP (OR 1.50, 95% CI 1.21-1.85, p < 0.0001). In contrast, female sex (OR 0.68, 95% CI 0.55-0.83, p < 0.0001), higher gestational age (OR 0.73, 95% CI 0.68-0.78, p < 0.0001) and higher birth weight (OR 0.89, 95% CI 0.83-0.94, p < 0.0001) emerged as protective factors against NEC/FIP. As shown in Table 3, we did not detect any associations between blood group distribution and secondary outcome variables.

Discussion
We found that neonates with blood group AB are at increased risk of developing surgical NEC and FIP compared to individuals with non-AB-blood groups. Consequently, blood group AB may be considered as a novel risk factor for developing NEC/FIP in VLBW infants besides the well-known factors including gestational age, hemodynamically relevant PDA, and male gender. The awareness of this association should contribute to earlier diagnosis and treatment of both potentially devastating diseases. However, we are aware that blood group AB is infrequent, and only a small proportion of VLBW neonates can be potentially observed more closely.
The ABO blood group system comprises the carbohydrate molecules (A, B, and H antigens) that are expressed on the extracellular surface of red blood cells as well as a variety of human tissues, including platelets, leukocytes, plasma proteins, the vascular endothelium, and the mucus layer 29 . These ABO antigens locating on tissues are histo-blood group antigens (HBGAs). The proposed functions of HBGAs are mostly cell-cell recognition, facilitation of intracellular uptake, signal transduction, enzymatic activity, and glycocalyx formation 30 . Furthermore, blood group antigens serve as receptors and coreceptors for bacteria, parasites, and viruses, contributing to host  www.nature.com/scientificreports/ susceptibility or resistance to various infectious and non-infectious diseases. Finally, several blood groups can additionally modify the innate immune response to infection 28 .
To date, only one study investigated the effect of neonatal and maternal blood group distribution on NEC mortality. In 2012, Thomson and colleagues analyzed 276 neonates with Bell stage II-III NEC and found that neonates with blood group AB (n = 60, 21.7%) had a significantly higher risk of mortality from NEC compared to other blood groups (Hazard ratio 2.87, 95% CI 1.40-5.89, p = 0.003) 27 . In contrast, in our cohort, we could not confirm an association between blood group and risk of mortality related to NEC/FIP. These discordant findings can be partially explained by the fact that the mortality from NEC/FIP in our patient cohort was very low (n = 31, 7.0% of all infants with surgery for FIP or NEC) compared to those reported by others (up to 32% 31, 32 ), which may have affected the statistical power and validity.
In the intestine, blood group antigens are found on red blood cells, mucosal epithelia, in the mucus layer, and in biologic fluids (saliva, intestinal contents, milk) [33][34][35] . Individuals who are non-secretors of blood group antigens do not have ABH antigens expressed on the mucosal secretions and surfaces 36 . HBGAs are recognized as receptors by numerous pathogens, including, noroviruses [37][38][39] , rotavirus [40][41][42] , coronaviruses 43 , and Helicobacter pylori 44 . H. pylori strains, for example, binds to all three ABH blood group antigens via the Blood group Antigen Binding Adhesin 45 . The most well-known example of blood-group-associated diseases is cholera, which affects predominantly individuals with blood group O leading to more severe symptoms than those with other blood groups 46 . Although the primary receptor of the cholera toxin is the GM1 ganglioside, it can also bind to HBGAs, facilitating adhesion on the surface of the enterocytes 47 . More recently, a study conducted by the Severe Covid-19 GWAS Group has shown that individuals with blood group A are at higher risk for developing symptomatic Covid-19 with respiratory failure compared to other blood groups suggesting the potential involvement of the ABO blood group system in the pathogenesis of Covid-19 infection 48 . Furthermore, there is increasing evidence for a HBGA-mediated modulation of the intestinal microbiome. For example, certain lactobacilli such as L. gasseri OLL2755 or L. gasseri OLL2877 bind to HBGAs 49 . Healthy subjects with different HBGA were shown to possess distinct profiles of fecal microbiota with a blood group secretor status being associated with less diversity at higher-order taxa and an expansion of distinct bacterial families, e.g., Clostridiales, in the presence of blood group A antigens 50,51 .
While less is known about an association between FIP and the pattern of gut colonization, multiple evidence has been provided that NEC is preceded by dysbiotic microbiota profiles 52,53 . The impact of HBGAs on the shaping of the gut microbiota might represent the pathogenetic clue explaining the increased occurrence of NEC in neonates with blood group AB and should be further elucidated in future studies.
Finally, whether packed red blood cell transfusion increases the risk for NEC (so-called transfusion-associated necrotizing enterocolitis) is subject of ongoing discussions 54,55 . As blood products with blood group AB are scarce and individuals with blood group AB are universal recipients, blood group compatible but not identical transfusions might be more frequent in patients with blood group AB compared to individuals with non-AB. Furthermore, in some institutions, preterm infants are transfused with blood group O, irrespective of their own blood group 27 . As individuals with blood group AB express A and B antigens in various tissues, the transfusion of blood products containing isoagglutinins (i.e., anti-A or anti-B antibodies) might augment the risk for surgical NEC/FIP in preterm infants carrying AB antigens. However, to test this hypothesis, register studies in VLBW preterm infants should include information on the type (blood group identical or blood-group compatible) and timing (transfusion bevor NEC or after) of red blood cell transfusions.
We are aware of several limitations of our study, including the low prevalence of blood group AB among neonates with NEC/FIP (n = 39, 5.9%). Furthermore, unrecognized confounders such as the prenatal course, center-specific protocols for feeding advancement, and availability of donor milk in some centers might have influenced the results of our study. Overall the association of blood group AB and NEC/FIP needs to be tested in other populations, e.g., in cohorts with a higher prevalence of blood group AB. We are also aware that the intraoperative findings and histopathology of FIP are different from that of NEC. However, histopathological data is lacking in the GNN cohort, thus making precise differentiation between both diseases impossible. Moreover, as both entities have an overlapping clinical presentation and there is currently no well-defined set of objective criteria to intra-operatively distinguish NEC from FIP, it is possible that in the central GNN database, a code for "NEC" was chosen instead of "FIP" and vice versa. To overcome both limitations, we used "surgical FIP/NEC" as a robust outcome parameter.
On the other hand, this study has several strengths. Our cohort is quite homogeneous in ethnicity, the degree of prematurity, and the clinical treatment strategies. To the best of our knowledge, we present the first study investigating the association between ABO blood group and risk for NEC/FIP in a large, well-defined cohort of VLBW infants with stringent criteria for NEC/FIP (both requiring surgical intervention). Furthermore, we also excluded neonates with the diagnosis "medical NEC" (not requiring surgery), as this condition is difficult to distinguish from other disease entities. Another strength of our study is that the blood groups were determined genetically, which represents the most accurate method.

Conclusion
Here, we demonstrate that among VLBW infants, blood group AB is associated with an increased risk of NEC/ FIP. Therefore, blood group AB may be considered as an additional risk factor for NEC/FIP. The underlying mechanism of the blood group AB in NEC/FIP development, including potential interactions between intestinal bacteria and HBGAs needs to be further elucidated. Definitions of outcome parameters. Primary outcome was the frequency of NEC/FIP in infants with blood group AB (n = 614, 6.0%) compared to 9643 (94.0%) non-AB neonates. Secondary outcome parameters included NEC/FIP related and overall mortality, occurrence of complications such as IVH, PVL, BPD, retinopathy of prematurity (ROP) or sepsis, and treatment modalities such as surgical or medical treatment of PDA according to blood group distribution in infants with blood group AB versus non-AB.
Sepsis was either clinical sepsis or blood culture-proven sepsis. According to our previous publications 56,57 , clinical sepsis was defined as sepsis with at least two signs (temperature > 38.0 °C or < 36.5 °C, tachycardia > 200/min, new-onset or increased frequency of bradycardia or apnea, hyperglycemia > 7.8 mmol/l, base excess < − 10 mval/l, change in skin color, increased oxygen requirements) and one laboratory sign (C-reactive protein > 10 mg/l, immature/neutrophil ratio > 0.2, white blood cell count < 5/nl, platelet count < 100/nl) and antibiotic treatment for ≥ 5 days, but no detection of a causative agent in the blood culture. Blood-culture confirmed sepsis was defined as clinical sepsis with proof of a causative agent in the blood culture 58 .
Focal intestinal perforation (FIP): focal intestinal perforation with the need for laparotomy. Surgical Necrotizing enterocolitis (NEC): clinical NEC with the need for laparotomy with or without resection of the necrotic gut and the macroscopic diagnosis of NEC. The attending pediatric surgeon differentiated in NEC or FIP.
FIP/NEC-related mortality was the subgroup of children with diagnosed NEC in Bell stages II or III without surgery but death (n = 11). In monitoring visits, the probable cause of death was determined.
Small for gestational age (SGA): Birth weight below the 10 th percentile according to Voigt et al. 60 . Patent ductus arteriosus (PDA) surgery: surgical ligation of PDA. Patent ductus arteriosus (PDA) pharmacological treatment: therapy with indomethacin or ibuprofen and PDA.
Bronchopulmonary dysplasia (BPD): need for ventilation support or oxygen supplementation at 36 weeks corrected age.
Death was defined as mortality during the primary stay in the hospital.
Genotyping. Patients' data and samples were coded before genotyping and analysis. Umbilical cord tissue of participating infants was collected after birth and stored at − 20 °C until transfer to the University of Lübeck. Genotyping was performed using the previously described methodology 56 . DNA was extracted using standard protocols for commercial DNA purification kits (Gentra Puregene Tissue Kit, Qiagen, Hilden, Germany www.nature.com/scientificreports/ Statistical analysis. Study population was grouped by the blood groups, A, B, O, or AB and additionally non-AB (A, B, and O). Groups were tested with χ 2 test and Mann-Whitney-U-test to describe differences descriptively. The association of non-AB vs. AB with the predefined outcomes (surgical treatment of NEC/FIP, and surgical treatment of NEC/FIP or death) was determined by OR, absolute risk difference, corresponding confidence interval, and p value from χ 2 test. In addition, the association of the blood groups non-AB vs. AB on several secondary outcomes (overall mortality, mortality from NEC/FIP, IVH, PVL, surgical treatment of PDA, medical treatment of PDA, BPD, ROP, sepsis) was determined by OR, absolute risk difference and corresponding confidence interval. The association of non-AB vs. AB with surgical treatment of NEC/FIP was calculated in multivariable logistic regression analyses with adjustment for known influencing variables (gestational age, birth weight as continuous variables), and further covariates (antenatal steroids, sex, drug treatment of PDA, surgical treatment of PDA, and multiple births). The type I error level was set to 0.05. Data analysis was performed using SPSS 25.0 (IBM, Munich, Germany). Data visualization was done by GraphPad Prism version 6.21 for Windows (GraphPad Software, San Diego, CA).

Data availability
The datasets generated and analyzed during the current study are not publicly available but can be reviewed on reasonable request by Wolfgang Göpel (Head of Studies, GNN).