The ABO system was the first human blood group system discovered by Karl Landsteiner in 1900 1 and it remains the most important and widely studied blood group system in transfusion medicine.2 Rare genetic variations resulting in weak ABO phenotypes, known as ‘weak ABO subgroups’, is a worldwide long-term research topic. In recent years, various subgroups with different molecular etiologies have been reported.3-7

The true frequencies of the ABO subgroup phenotypes have not been well studied due to the relatively low prevalence and the potential for missed detection. This low prevalence also obscures the relationship between the phenotype and genotype. The detection rate of the ABO subgroup in the Han Chinese population is reportedly approximately 0.015 %,8 yet this might be lower than the true frequency due to the lack of an estimate of the missed detection rate (MDR).

The frequencies of certain weak ABO subgroups have been reported in several populations.9 A3 is the most frequent weak A phenotype in Canadians, with a frequency of 0.0011 %.10 The frequency of Ax in France has been estimated to be 1 in 77,000 (0.003 % of group A) 11 and 1 in 40,000.12 Other A subgroups include Am and Ael, which are very rare subgroups with low frequencies and different ABO gene variations.13,14 Among B subgroups, B3 is relatively common, with a frequency of 1 in 10,000 group B French donors,15 1 in 900 group B Chinese donors and 1 in 1800 group A1B Chinese donors.16 Cis-AB and B(A) phenotypes have also been frequently detected in Chinese ABO subgroup individuals, with a total frequency of 0.0083 %.17 In Japan, Bm is common (a total frequency of 0.0244 %) among B subgroups.18

Our previous study focused on molecular genetic analysis of ABO subgroup individuals, identification of novel subgroup alleles and elucidation of the molecular mechanisms involved.4,19,20 However, the distribution characteristics of these alleles and their individual differences in the antigen expression remain largely unknown.

Therefore, we systematically and comprehensively analyzed many ABO group samples, estimated the MDR of the typing process, and investigated the relative percentages (RPs) and antigen expressions of different ABO subgroup alleles. This study is the first to propose an MDR around the world and to provide a practical estimation method. The value obtained in this study is closer to the true prevalence, which is highly important as a reference in blood typing and investigation of subgroup frequency.

This study was conducted on 2.9 million volunteer blood donors in Shanghai, which is the largest sample size to date of studies on ABO subgroups.

According to the statistical analysis of ABO subgroups of repeat blood donors, up to 27.81 % of ABO subgroups were not identified during ABO blood typing, which might indicate that the frequency reported in the literature is much lower than the true value. The MDR of the AB subgroup was 33.60 %, which was much greater than that of the A and B subgroups. Overall, the AB subgroup was more prone to missed detection than the A and B subgroups. One of the reasons might be that the H antigen of the A and B subgroups is enhanced, which is better to confirm the subgroup using the anti-H reagent, but that the H antigen of the AB subgroup is enhanced only when the allele is O.

Accounting for MDR, the ABO subgroup frequency of blood donors was found to be 0.044 % in Shanghai, which was much greater than previously reported.8 Additionally, we report the detection rates of the A, B and AB subgroups (0.003 %, 0.010 %, and 0.018 %, respectively) and obtained relatively accurate prevalence rates (0.004 %, 0.013 %, and 0.027 %, respectively) in Shanghai. Obviously, regardless of the detection rate or frequency, the AB subgroup is the most common ABO subgroup in the Chinese blood donor population. The A subgroup showed the lowest frequency in the voluntary blood donor population; the most common A subgroups, such as A2 and Aint, were excluded from in this study.

There are many reasons for the high MDR of ABO subgroups. For instance, the initial screening and serological identification of ABO subgroups are complex, and each may involve missed detection, such as interbatch differences in reagents, the limit of the automatic blood group analyzer, and individual (skill) differences among technicians. We utilized three automatic platforms. Samples (n = 1,874,808) were initially screened twice from January 2009 to October 2014: one by Galileo or the PK7300 system with monoclonal anti-A anti-B from Immucor; and one by the Tecan–Microplate Reader system with monoclonal anti-A and anti-B from SHPBC. The detection rate was 0.029 %. From November 2014 to December 2017, all samples (1,070,835) were screened once using one of the platforms with monoclonal anti-A and anti-B from Immucor. The detection rate was 0.035 %. Compared to that of the previous two initial screenings, the detection rate was slightly better (p-value >0.05). Therefore, the second screening did not increase the missed detection rate of ABO subgroups, but it indicated that there are differences in the ability of subgroup detection between different automatic platforms. In addition, the complex characteristics of ABO subgroups are one of the reasons for missed detection, such as Ael. Due to the presence of irregular anti-A, it is likely to be mistaken as the O type by the automatic blood group analyzer. Therefore, in practical work, typing cannot rely entirely on an automatic blood group analyzer.

In general, we should employ a series of measures to minimize MDR strains. First, all microplate results should be manually reviewed during the initial screening. Second, it is necessary to establish a complete and systematic operating procedure for blood typing and strictly control the centrifuge, incubation, and environmental temperature and humidity conditions during ABO blood typing, and technicians should be regularly trained to avoid weak agglutination and missed detection caused by nonstandard operations. Third, tube agglutination should be standardized. On the basis of the AABB observation agglutination method,22 a shaking step was added in this study to observe the firmness of the agglutination block to avoid missing weak agglutination.

To investigate the clinical hazard of missed ABO subgroups, we traced the transfusion effects of several ABO subgroup blood samples to the transfusion departments of multiple tertiary hospitals and found no data on transfusion reactions or invalid transfusions. A possible explanation is that subgroups with strong antigen expression, such as A3B and cisAB, are less likely to be missed during detection, and for subgroups with weak antigen expression or weak anti-A, anti-B is not sufficient to cause a severe transfusion reaction and thus has a little impact on patients.

The ABO blood group system is clearly a critical player in the modern era of genomic medicine.23BA.04 was the most common mutated allele in Shanghai blood donors, and B alleles were the most complicated. This study identified several new ABO alleles that are being submitted. Although our understanding of the structures of A and B transferases and their enzymology has dramatically improved,24-26 the effects of these enzymes on the serological characteristics of different ABO subgroups have yet to be elucidated. Moreover, the relationship between serological characteristics and molecular mechanisms has received little attention. This study preliminarily evaluated the correlation between these factors using statistical methods and provided theoretical evidence for future molecular diagnoses.

The weakest B antigen expression was found for BW.03, with Bvar-1 having the strongest expression. Two critical mutations in Bvar-1 are p.Val175Met and p.Arg249Gln, which are near the four critical residues 21. These findings suggest that mutations near the four critical residues have weak effects on antigen expression. The critical mutation sites in BW.03 are p.Asp302Gly and p.Arg241Trp 27. Both these sites are located in the C-terminal domain, which binds disaccharide acceptors. These findings suggest that amino acid changes in the receptor-binding site have a strong effect on antigen expression, decreasing expression of the B antigen and causing weak serological agglutination intensity, with no obvious individual differences.

This comparison revealed significant relationships among the expression level, individual differences in antigens, ABO gene mutation sites and amino acid changes, but these relationships need to be verified by additional experiments. This research involved a large sample size and long study period, and some blood donors may be related. However, due to the privacy protection of blood donors, data on the relationships between blood donors were unavailable - this was the limitation of this study.