Clinical value of combination detection of direct antiglobulin test and serum albumin globulin ratio in severe hyperbilirubinemia caused by ABO hemolytic disease of the newborn: A single-center retrospective analysis

Background To explore of a combination of antiglobulin test(DAT) and albumin globulin ratio(AGR) could predict the severity of ABO hemolytic disease of the newborn(ABO-HDN).Methods The measurement of DAT, AGR and combination detection of DAT and AGR was done to predict severe ABO-HDN hyperbilirubinemia in 270 full-term infants based on whether the infants received transfusions of blood components. The infants were divided into three groups according to the results of DAT and ARG and compared the differences of phototherapy day and hospitalization day of the three groups.Results Of the 270 cases enrolled in this study, 69 infants were DAT positive. Peak total bilirubin, AGR, and positive DAT were independently associated with the need for blood components transfusion. ROC curve analysis for blood components transfusion showed that DAT cutoff value >± with a sensitivity of 39.4% and a specificity of 83.9%, AGR cutoff value <2.05 with a sensitivity of 54.1% and a specificity of 85.7%, and combination detection of DAT and ARG with a sensitivity of 62.1% and a specificity of 91.2%. The AUCs for DAT, AGR, and combination detection of DAT and AGR were .621, .740, and .750 respectively. The phototherapy day and hospitalization day were significantly longer in group of AGR <2.05 and DAT >± than that of a group of AGR <2.05 and group of DAT >±.Conclusions DAT and ARG could be early predictors for the severity ABO-HDN hyperbilirubinemia and combination detection of DAT and AGR could further increase its predictive value.


Introduction
The pathogeny of hemolytic disease of the newborn (HDN) is maternal alloimmunization, the active IgG maternal red cell alloantibodies stimulated by fetal erythrocyte antigen could transport across the placenta and destroy involved fetal erythrocyte carrying the antigen 1 . HND is an important cause of neonatal morbidity and mortality with clinical presentation including mild clinical symptoms of neonatal jaundice, neonatal anemia and extramedullary hematopoiesis and severe symptoms of critical hyperbilirubinemia, anemia, hydrops fetalis even neonatal death 2 . ABO-HDN refers to occur mostly in infants with blood group A or B born to mothers with blood group O as the result of higher frequent occurrence of IgG anti-A and anti-B more in group O individuals, previous meta-analysis concerning the incidence of ABO incompatibility in pregnancies showed that about 15-25% of pregnancies can have ABO incompatibility and 10% develop HDN 3 . ABO-HDN has emerged as the leading cause of HDN and tends to be a milder disease usually treated successfully with phototherapy alone and rarely requires exchange transfusion therapy, intravenous immunoglobulin (IVIG) administration, or red blood cell (RBC) transfusion, compared with Rh-HDN, however, in severe cases, ABO-HDN may lead to severe fetal anemia with a risk for fetal death and to severe forms of neonatal hyperbilirubinemia with a risk for kernicterus or chronic bilirubin encephalopathy, which can lead to permanent brain damage 4 , previous reports 5, 6 have described the cases with severe anemia treated with RBC transfusion and with signi cant hyperbilirubinemia where exchange transfusion therapy is required. In our latest research 7 on exchange transfusion therapy enrolled 123 infants with severe hyperbilirubinemia in Wuhan demonstrated that ABO-HDN is the leading etiology of severe neonatal hyperbilirubinemia which should be treated with exchange transfusion therapy. Delay in recognizing severe ABO-HDN may potentially leads to signi cant hyperbilirubinemia that may require intervention, therefore, early and rapid identi cation of severe ABO-HDN is critical for pediatricians to discriminate high risk neonates, enhance early recognition of potentially severe neonatal jaundice, decrease bilirubin levels and instate treatments to minimize further hemolysis, ameliorate anemia 8,9 .
Currently, postpartum examinations of screening for ABO, Rh blood group type, irregular antibodies and direct antiglobulin test (DAT) are performed to identify neonates who are at higher risk for HDN in many hospitals 10 . DAT (previously termed direct Coombs' test), rstly described in 1945 by Cambridge immunologist Robin Coombs 11 , has been widely used to detect antibodies or complement bound directly to the patient's RBCs, indicating in vivo sensitization to diagnose autoimmune hemolytic anemias as well as HDN, hence, DAT is regarded as the cornerstone in the diagnosis of HDN 12 and in 2016, the guideline of evidence-based recommendations for the application of blood grouping and red cell antibody testing in pregnancy to predict and prevent HDN developed in accordance with the British Committee for Standards in Hematology methodology showed that DAT on a neonatal specimen may help to determine the etiology of neonatal hyperbilirubinemia when HDN is suspected 13 , however, the incidence of positive DAT in newborns with ABO-HDN is low, leading to the positive DAT or negative DAT may not rule in or out ABO-HDN. Additionally, the negative DAT may not exclude clinically signi cant hyperbilirubinemia caused by a non-immune hemolytic etiology including G6PD de ciency, sepsis, intracranial hemorrhage, hereditary spherocytosis and pyruvate kinase de ciency as well. Moreover, the low concentration of the anti-A or anti-B antibody occurs in ABO-HDN may not be detected by DAT because of the insu cient amount of anti-A or anti-B antibody, could still increase bilirubin level and generate mild hemolysis 14 , meanwhile, the DAT itself may be affected by many factors including anti-RBC IgA, low a nity antibody, antihuman globulin activity and centrifugation technique, in our previous research which was concerning the interference in DAT showed that rheumatoid factor in under-tested samples can lead to both false decreases and false increases in DAT 15 . Therefore, DAT could not support the diagnosis of severe hemolysis alone and combined additional identi ed factors and investigations are required to diagnose the severity of ABO-HDN. Albumin globulin ratio(AGR), calculated as albumin/globulin, is actually a re ection of all non-albumin proteins, has been reported to be a novel prognosticator of many diseases including different types of cancer and some other diseases such as polyangiitis and heart failure 16,17 . The potential decrease of albumin concentration caused by bilirubin metabolism may lead to the decline of AGR level and no study has investigated the clinical value of association of AGR with the severity of ABO-HDN so far, therefore, in the study, our aim is to explore whether the combination detection of AGR and DAT would predict the severity of ABO-HDN.

Data collection
The basic conditions of all infants were collected from electronic medical record system anonymously, including age of hospitalization, sex, gestational age, delivery mode, feeding mode, age of jaundice, age of peak total bilirubin, birth weight, body weight at hospitalization, days of phototherapy, days of hospitalization, as well as the results of laboratory tests during hospitalization, including ABO blood type, peak total bilirubin, HGB, RBC, AGR and DAT at the same time, the general data of the mothers were recorded, including age, gestational day, pregnancy and delivery history and so on.

Outcome measures
The primary outcome was the infants received transfusions of blood components, including IVIG, albumin, RBC and exchange transfusion therapy during hospitalization. The secondary outcomes included the days of phototherapy and the length of hospitalization.

Statistical analysis
The measurement data were tested by Kolmogorov-Smirnov test or Shapiro-Wilk test to see if they were of normal distribution. Normally distributed data were expressed by mean ± standard deviation (x ± s) and data that were not normally distributed were presented as median and 25th-75th centiles. A univariate binary logistic regression analysis was used to compare the differences of variables between the infants with and without blood components transfusion, after identi cation of predictors, a multivariate binary logistic regression analysis was performed to identify whether the AGR and DAT were ultimate predictors. A receiver operating characteristic (ROC) curve was made to compare the difference of prediction value of AGR, DAT and combination detection of AGR and DAT and the result is presented as the area under the curve (AUC). To nd the cutoff value for AGR and DAT as predictors for blood components transfusion in ABO-HDN infants, the Youden index was calculated. The Youden index is presented as the fetal AGR and DAT levels with corresponding sensitivity and speci city. A threeindependent samples nonparametric test was performed to compared differences of phototherapy day and hospitalization day among group of AGR < 2.05, group of DAT > ± and group of AGR < 2.05 and DAT > ±. The difference of P value less than 0.05 was considered to be signi cant. These analyses were performed using the SPSS 20.0 statistical software (IBM, Chicago, Illinois, USA). The gures were performed using the GraphPad Prism 8.0 software (San Diego, California, USA).

Results
During the study period, 510 infants were con rmed as ABO incompatibility and admitted to Wuhan Children's Hospital (Wuhan Maternal and Child Healthcare Hospital), Tongji Medical College, Huazhong University of Science & Technology, of which 465 cases were diagnosed with hyperbilirubinemia during the study period and thus were eligible for this study. A total of 185 cases were excluded, as 46 cases were discharge required by guardians of neonates, 43 cases were excluded because of the gestational age was less than 37 weeks, 52 cases with other causes of hyperbilirubinemia, including 40 cases of sepsis and 12 cases of intracranial hemorrhage, respectively, 21cases received operations during hospitalization, 19 infants has the history for treatment in other hospitals and 4 infants with severe congenital malformation. The derivation of the study population is shown in Fig. 1. The baseline characteristics of the study group are presented in Table 1. Finally, 270 cases were enrolled in this study, of which 109 infants received blood components infusion. Among them, 80 infants received one kind of blood components, 29 infants received more than one kind of blood components, respectively and the number of infants who received one time, two times, three times or more than three times of blood components infusions were 65, 29, 13 and 2, respectively.
Univariate binary logistic regression analysis of potential predictors for blood components transfusion was performed ( were statistical signi cantly associated with need for blood components transfusion. The factors were entered in a multivariate logistic regression model to assess the independent association of these risk factors on the need for blood components transfusion ( Table 2) An ROC curve was plotted and the AUCs for DAT and AGR were 0.621 and 0.740, respectively. The Youden index was calculated at 0.398 with a cutoff AGR of < 2.05 with a sensitivity of 54.1% and a speci city of 85.7%, the Youden index was calculated at 0.233 with a cutoff DAT level of > ± with a sensitivity of 39.4% and a speci city of 83.9%. An ROC curve was plotted and the AUC for combination detection of DAT and ARG was 0.750 (Fig. 2).
A three-independent samples nonparametric test was performed and there are signi cant statistical difference of phototherapy day and hospitalization day among group of AGR < 2.05, group of DAT > ± and group of AGR < 2.05 and DAT > ± (Fig. 3).  22 which may lead to acute and chronic nervous system damage with no early intervention. In severe ABO-HDN cases, transfusions of blood components were required usually to reduce jaundice, prevent and treat anemia, however, blood components transfusion may lead to transfusion transmitted diseases, anaphylaxis, hypersensitivity, thrombosis, electrolyte disturbance and renal failure as well as high hospitalization cost.  24 The incidence of positive DAT is higher than that of previous reports, in addition to race differences, that DAT was used as screening test to perform in all cases with suspected ABO-HDN instead of infants with con rmed ABO-HDN in our study, could be the main reason cause of this difference, hence, even if the positive rate of DAT has risen with the increase of the possibility of ABO-HDN in the study population, the incidence of positive DAT in ABO-HDN is still low, illustrating DAT a poor positive predictor of ABO-HDN. Moreover, we investigated the clinical value of DAT in severe hyperbilirubinemia in the present study, although ABO-HDN refers to be the major reason of positive DAT, whether DAT could be a prediction factor for severe hyperbilirubinemia is still controversial, some studies reported that the positive DAT has only a poor predictive value for severe hyperbilirubinemia because of only approximate 23% cases of ABO-HDN with positive DAT will continue to have signi cant hyperbilirubinemia 25,26 , however, this study shows that DAT has a predictive value for the severe ABO-HDN hyperbilirubinemia with the DAT level cutoff value of ± has the highest sensitivity and speci city in terms of accurately predicting severe ABO-HDN hyperbilirubinemia, Mehta R's study 14 enrolled 901neonates with gestational age > 34 weeks and birth weight > 2000 g showed that the risk for hyperbilirubinemia requiring phototherapy in the DAT positive infants is signi cantly higher than that in the neonates with negative DAT (OR 6.78, 95% CI 2.38- 19.33) and previous studies demonstrated that ABO-HDN with a positive DAT is considered a major risk factor for the development of severe hyperbilirubinemia and neurotoxicity 27,28 as well.
The combination detection 29 of DAT and detection factors including the blood cell indices, pre-discharge total bilirubin level, cord serum albumin and cord bilirubin/albumin ratio 30 would be more bene cial than that of DAT alone for predicting the severity of ABO-HDN. Among them, cord serum albumin and cord bilirubin/albumin ratio have been well researched as the result of the mechanism of bilirubin metabolism in infants, in plasma, bilirubin binds to albumin to form the bilirubin albumin complex in order to transport to the liver for further metabolism, on one hand, bilirubin binding to albumin increases the water solubility of bilirubin and improves the plasma capacity of transporting bilirubin, on the other hand, it limits the free permeability of bilirubin to various cell membranes and avoids toxic effect of bilirubin on tissues and cells 31,32 , therefore, low serum albumin level decreases bilirubin clearance and thus increases signi cant hyperbilirubinemia, previous studies 33,34 found that term infants cases with low cord albumin < 2.8 g/dl developed more signi cant hyperbilirubinemia requiring phototherapy and exchange transfusion and Khairy MA and colleagues 33 showed that neonates with cord bilirubin/albumin ratio < 0.61 were at risk of developing signi cant hyperbilirubinemia needing interventions at the same time. However, both cord serum albumin level and cord bilirubin/albumin ratio are considered to re ect ability of albumin to bind bilirubin, AGR we assessed in the current study re ect the infant level of albumin remaining after binding bilirubin resulting in that high level of bilirubin may lead to decline of albumin level. Moreover, since this is the rst analysis assessing the clinical value of AGR in severe ABO-HDN hyperbilirubinemia to the best of our knowledge, we plotted an ROC curve analysis and showed that AGR cutoff value < 2.05 had a good predictive value with a sensitivity of 54.1% and a speci city of 85.7%, meanwhile, we also demonstrated that combination detection of DAT and ARG had a better predictive value than that of respective detection of DAT and ARG in prediction of severe ABO-HDN hyperbilirubinemia (AUC for combination detection 0.750 VS AUC for DAT 0.621 and AUC for AGR 0.740). Finally, after dividing the infants into three groups according to the results of DAT and ARG and comparing phototherapy day and hospitalization day of the three groups, we found that group of AGR < 2.05 and DAT > ± had signi cant longer phototherapy day and hospitalization day than that of group of AGR < 2.05, group of DAT > ± respectively, prompting that the condition of infants with AGR < 2.05 and DAT > ± is more severe and meaning more hospital costs. With lack of studies done on ARG as a prediction factor of severe ABO-HDN hyperbilirubinemia, this work opens the window for further studies to be performed in this eld and we are aware that larger scale trials including Multi ethnic researches and preterm neonates are needed.
To summarize, in this study DAT and AGR proved to predict the severity of ABO-HDN hyperbilirubinemia in term neonates. Infants with either DAT > ± or AGR < 2.05 were at risk of developing signi cant ABO-HDN hyperbilirubinemia. These can be considered possible predictors for severe ABO-HDN hyperbilirubinemia and combination detection of DAT and ARG can improve the predictive value. We recommend that pediatricians pay close attention to the results of DAT and AGR when judging the severity of ABO-HDN hyperbilirubinemia, particularly the term infants with DAT > ± and AGR < 2.05.
. the rst draft of the manuscript text, Duan Ling and Chen Ping prepared table1-2 and gures1-3. Hu Hongbing guided the process of the whole research, including revised the manuscript text and all authors commented on previous versions of the manuscript. All authors read and approved the nal manuscript.