Blood Group Foundations and Carbohydrate Systems

Blood group results begin with inherited variants, but the route from gene to red-cell phenotype depends on the kind of antigen. Some genes encode membrane proteins. Others encode enzymes that add sugars to a precursor chain, and the completed carbohydrate becomes the antigen. This distinction explains why several independently inherited genes can shape one ABO or Lewis result and why serology may support more than one possible genotype.

Inheritance and expression

Genetic vocabulary

A gene occupies a defined chromosomal locus. Alternative sequences at that locus are alleles. A person’s genotype is the inherited allele combination, and the phenotype is the antigen expression observed or predicted from that combination. Family studies or molecular testing can establish genotype. Routine serology establishes phenotype and may support a probable genotype.¹

During gamete formation, the two alleles at a locus segregate so that each gamete receives one. Genes at separate, unlinked loci assort independently. Closely linked loci can travel together as a haplotype, with occasional recombination producing a new combination. Most blood group alleles are codominant, so a heterozygous cell can display both inherited antigens. Incomplete dominance has an intermediate heterozygous phenotype and occurs less often in routine blood group interpretation. Most blood group genes are autosomal. XG, XK, and ATP11C are X-linked examples, so a typical XY individual is hemizygous at those loci.

Sequence changes alter expression through several mechanisms. A missense variant substitutes an amino acid and may change an epitope or enzyme activity. A nonsense variant introduces a stop codon. An insertion or deletion can shift the reading frame. Larger deletions can remove a gene, and recombination can form a hybrid gene. The common ABO allele designated O.01.01 carries a single-nucleotide deletion, c.261delG, that shifts the reading frame and produces an inactive glycosyltransferase.^2,3

Family and population predictions

Pedigrees use squares for males, circles for females, a horizontal partner line, and descending lines for offspring. A filled symbol identifies the phenotype under study, and an arrow identifies the propositus whose finding began the family investigation. A trait appearing among siblings born to unaffected parents can support recessive inheritance. A trait found in successive generations can support dominant inheritance. Penetrance, a new variant, incomplete family information, and serologic error can change either pattern, so a pedigree supplies evidence together with laboratory testing.

The common ABO model has three allele classes. A and B are codominant, and O is functionally silent. Phenotypes A and B each allow two common genotypes; O and AB have one common genotype each.

For an AO × AO mating, each parent can transmit A or O. The four equally likely combinations are AA, AO, AO, and OO. The genotype ratio is 1:2:1. The expected phenotype distribution is 75% group A and 25% group O. The expectation describes many conceptions under the model; it cannot predict the type of one child.

Common mating combinations produce these phenotype sets:

The Hardy-Weinberg relationship estimates genotype frequencies from allele frequencies in a large, randomly mating population when mutation, migration, and differential survival have no meaningful effect on the locus. For two alleles with frequencies p and q:

p + q = 1

p² + 2pq + q² = 1

In a simplified two-allele RHD example, assume 9% of donors lack functional RHD on both haplotypes. Then q² = 0.09, q = 0.30, and p = 0.70. Among 2,000 donors, the model predicts 980 with two functional haplotypes, 840 with one functional and one null haplotype, and 180 with two null haplotypes. Serology groups the first 1,820 donors as D positive, although two genotype classes contribute to that phenotype. A three-allele system uses p + q + r = 1 and includes the three homozygous terms plus the three doubled heterozygous terms.

Blood group terminology

A blood group system contains one or more antigens controlled by a single gene or by a complex of closely linked homologous genes, and each system is genetically distinct. The International Society of Blood Transfusion (ISBT) maintains the official registry. Its August 2026 release recognizes 49 systems after the addition of JAMA as system 049.⁴

Each antigen in a system has a six-digit number. The first three digits identify the system and the last three identify the antigen. ABO antigen A is 001001. Related antigens whose system assignment is pending can enter a 200-series collection. The 700 series holds unassigned antigens whose prevalence is below 1% across human populations; the 901 series holds unassigned antigens whose prevalence exceeds 90%.⁵

Laboratory records keep genes, antigens, antibodies, and phenotypes distinct:

• Gene symbols are italicized, such as ABO, FUT1, and A4GALT.
• Antigen symbols are plain type, such as A, K, Fy^a, and P^k.
• Antibody names begin with anti-, such as anti-Fy^a and anti-P1.
• A single antigen can be recorded as present or absent, such as K+ or K−.
• An antithetical pair appears in one set of parentheses, such as Fy(a−b+).
• Numeric phenotypes use the system symbol, a colon, and antigen numbers, such as SC:−1,2.
• Phenotypes from different systems are separated with semicolons, such as S+s+; K−; Fy(a+b−).

A null red cell can be especially useful in an antibody investigation. If plasma reacts with common cells and loses reactivity with a system-null cell, the absent system becomes a strong candidate for the antibody’s target. Reagent rarity and other antigen differences still require a complete investigation.

ABO and H

Antigen construction

ABO and H antigens are the terminal sugars of oligosaccharide chains on glycoproteins and glycolipids. A glycosyltransferase attaches each sugar in a defined linkage. Type 2 precursor chains predominate on red cells, while type 1 chains predominate in secretions. FUT1 adds fucose to a red-cell precursor to create H. An A or B transferase can then modify H. FUT2 creates H on type 1 chains in secretory tissues.⁶

Group O cells retain abundant unmodified H. A and B cells retain smaller amounts because their transferases consume H as substrate. The relative red-cell H expression is commonly summarized as O > A₂ > B > A₂B > A₁ > A₁B. Neonatal ABH expression is weaker and less branched, then approaches adult expression during early childhood. Reagent strength, method, and the inherited allele affect the measured antigen density.^1,6

Secretor status

A person with at least one functional FUT2 allele forms soluble H on type 1 chains in secretory tissues. The inherited ABO transferase then creates soluble A or B when the corresponding allele is present. A person with two nonfunctional FUT2 alleles lacks soluble ABH in saliva and other affected secretions. Red-cell ABH expression remains under FUT1 and ABO control.

Hemagglutination inhibition can demonstrate soluble antigen. Patient saliva is incubated with a known antibody, then indicator red cells are added. Soluble antigen neutralizes the corresponding antibody, so the indicator cells remain unagglutinated. Agglutination shows that active antibody remained. The method requires validated reagents, specimen preparation, and controls.

A₁, A₂, and weaker expression

A₁ and A₂ account for most group A phenotypes. A₁ cells react with anti-A and the lectin from Dolichos biflorus. A₂ cells react with anti-A and remain nonreactive with a properly controlled anti-A₁ lectin. A₂ transferase converts a smaller set of H structures, leaving more H on the cell. Ulex europaeus anti-H lectin therefore usually reacts more strongly with A₂ cells than with A₁ cells.^2,6

In many studied donor populations, A₁ accounts for about 80% of group A and A₂ for about 20%, with substantial population variation. Anti-A₁ has been reported in approximately 1%–8% of A₂ individuals and 22%–35% of A₂B individuals. These are teaching estimates from defined populations, and local frequencies may differ.^1,3

Some A₂ and A₂B individuals produce anti-A₁. It is usually a cold, naturally occurring antibody that appears as an extra reverse-grouping reaction. Thermal range determines its clinical importance. A newborn’s weak and less-branched A expression can also produce anti-A₁-lectin nonreactivity. A single neonatal lectin result leaves the subgroup unresolved.

Rare A and B alleles produce still weaker expression. Their results vary with reagent clones and methods, so the reaction pattern leads to additional testing under the laboratory procedure.

Cis-AB is a rare allele that can express A and B specificity from one chromosome. It often produces stronger A expression with weaker or mixed-field B expression and may place an unexpected weak anti-B in the plasma. Family or molecular study can distinguish this inheritance from ordinary A and B alleles carried on separate chromosomes.

Bombay and para-Bombay phenotypes

Classical Bombay, Oh, results when red cells lack functional FUT1 and secretions also lack functional FUT2. Their red cells lack H substrate, preventing inherited ABO transferases from forming A or B. Forward grouping resembles group O, yet the plasma contains anti-A, anti-B, and anti-H. The anti-H usually has a broad thermal range, activates complement, and reacts strongly with ordinary group O cells because those cells carry abundant H. Red-cell support requires compatible Bombay units.^1,3,6

Para-Bombay phenotypes retain a source of small amounts of H, commonly through functional FUT2 with absent or markedly reduced FUT1 activity. Soluble ABH can adsorb onto red cells, and weak A or B may be demonstrable by adsorption and elution. Anti-H varies in strength and thermal range. Compatibility testing and reference-laboratory characterization guide selection because some examples react at 37 °C.

Discrepant ABO results

Forward grouping detects antigen on patient red cells. Reverse grouping detects expected antibodies in patient plasma. A discrepancy exists when those results fail to support one ABO interpretation. The weakest or unexpected reaction often identifies the affected side of the test. Identity, labeling, reagent, centrifugation, cell suspension, and transcription checks come first. A fresh specimen may be required.²

For example, patient cells react 3+ with anti-A and 0 with anti-B. Plasma reacts 2+ with A₁ cells and 4+ with B cells, while O cells and the autocontrol are nonreactive. The forward result supports group A. The isolated extra reaction with A₁ cells supports an A₂ phenotype with anti-A₁. Anti-A₁ lectin and additional selected cells can test that explanation. The laboratory resolves the discrepancy before assigning a final ABO type; emergency component selection follows its validated policy.

Lewis

Lewis antigen expression reflects the interaction of FUT3 and FUT2 on type 1 chains. FUT3 can add fucose to the precursor to form Le^a. When FUT2 acts first, it creates type 1 H, and FUT3 can then form Le^b. Le^a and Le^b are separate products of these interacting pathways.⁶

Lewis-active glycolipids are made mainly in tissues, circulate in plasma, and adsorb onto the red-cell membrane. The red-cell phenotype can therefore change as the plasma environment changes.

The Le(a−b−) red-cell phenotype permits more than one secretor genotype, so Lewis typing alone cannot establish secretor status. Weak-secretor alleles can produce Le(a+b+), a phenotype seen more often in some Asian populations. Cord cells express little Lewis antigen, and expression develops as circulating glycolipids become available. Pregnancy can temporarily weaken the red-cell phenotype. Transfused Lewis-positive cells can shed adsorbed antigen in a Lewis-negative recipient.

Anti-Le^a and anti-Le^b occur most often in people with the Le(a−b−) phenotype. They are predominantly IgM, frequently bind complement, and usually react below 37 °C. Soluble Lewis substance can neutralize antibody, and transfused cells can lose Lewis antigen. These features make hemolytic transfusion reactions uncommon. An antibody reactive at 37 °C receives compatible red-cell units by the indirect antiglobulin test under the laboratory procedure.⁷

Anti-Le^bH reacts best when both Le^b and abundant H are present, often with group O or A₂ cells. Anti-Le^bL recognizes Le^b across ABO groups. Lewis antigens are weak on fetal cells and most Lewis antibodies are IgM. These features generally exclude Lewis antibodies as causes of hemolytic disease of the fetus and newborn.

Ii

i and I describe a developmental change in poly-N-acetyllactosamine chains. i is the linear chain. The branching enzyme encoded by GCNT2 converts it to branched I. Cord red cells are i rich and I poor. During approximately the first 18 months of life, I increases toward adult expression and i falls to trace levels. Rare GCNT2 variants can preserve the adult-i phenotype.^6,8

Low-level autoanti-I is common and usually reacts only at cold temperatures. A broad thermal range can allow reaction near body temperature, produce complement-mediated hemolysis, and interfere with ABO typing, antibody screening, and crossmatching. Anti-I commonly accompanies cold agglutinin disease and can appear transiently with Mycoplasma pneumoniae infection. Anti-i is associated with infectious mononucleosis and some lymphoproliferative disorders. Thermal amplitude and evidence of hemolysis carry more interpretive weight than a cold titer alone.

Accelerated or abnormal erythropoiesis can increase i expression on adult cells because less-mature red cells enter circulation. The finding can occur with leukemia, megaloblastic or hypoplastic anemia, thalassemia, sickle cell disease, and chronic hemolysis. The antigen pattern supplies context for serologic testing across these disorders.

P1PK and Globoside

The historical P blood group was divided after its antigens were assigned to separate genes. The P1PK system contains P1, P^k, and NOR under A4GALT. The Globoside system contains P, PX2, and ExtB under B3GALNT1. These systems remain biochemically connected. A4GALT converts lactosylceramide to P^k, and B3GALNT1 converts P^k to P. A4GALT also creates P1 from paragloboside through a parallel branch.^5,6,9

ExtB is expressed when B3GALNT1 acts on the corresponding B-bearing precursor. Its expression therefore follows both the Globoside pathway and the person’s ABO type.

Anti-P1 is commonly a naturally occurring IgM antibody in P₂ individuals. It usually reacts best in the cold and lacks clinical significance at 37 °C. P1 expression varies among cells and can weaken during reagent-cell storage, which can hide an anti-P1. A 37 °C-reactive example requires appropriate compatible units.

One U.S.-licensed anti-P1 reagent insert cites P₁ phenotype estimates of 79% in White people and 94% in Black people, with corresponding P₂ estimates of 21% and 6%. These are reference-population estimates; prevalence varies among populations.¹⁰

People with the rare p phenotype can produce anti-PP1P^k, a mixture directed against the missing P1, P^k, and P antigens. People with a P^k phenotype can produce anti-P. Anti-PX2 can accompany anti-P because P^k cells also lack PX2. These antibodies can include IgM and IgG, activate complement, and cause severe hemolytic transfusion reactions. They are also associated with recurrent early pregnancy loss. Compatible blood usually requires rare-donor resources.^1,9

Autoanti-P is the Donath-Landsteiner antibody of paroxysmal cold hemoglobinuria. This biphasic IgG binds red cells during cold exposure and fixes complement; warming permits completion of complement lysis. The laboratory demonstrates that pattern with a properly controlled Donath-Landsteiner test. Its P specificity is distinct from alloanti-P made by a person whose cells lack P.

References

1. Bloch EM, Campbell-Lee S, McKenna DH Jr, Montemayor-Garcia C, Schwartz J, Shaz B, Storry J, eds. Technical Manual. 22nd ed. AABB; 2026. Accessed August 26, 2026.
2. International Society of Blood Transfusion. ISBT Blood Group Database. Accessed August 26, 2026.
3. Dean L. Blood Groups and Red Cell Antigens. National Center for Biotechnology Information; 2005.
4. International Society of Blood Transfusion. ISBT Blood Group Database, August 2026 release, version 18. Updated August 3, 2026. Accessed August 26, 2026.
5. International Society of Blood Transfusion Working Party on Red Cell Immunogenetics and Blood Group Terminology. Blood group terminology. Accessed August 26, 2026.
6. Stanley P, Wuhrer M, Lauc G, Stowell SR, Cummings RD. Structures common to different glycans. In: Varki A, Cummings RD, Esko JD, et al, eds. Essentials of Glycobiology. 4th ed. Cold Spring Harbor Laboratory Press; 2022. doi:10.1101/glycobiology.4e.14
7. Canadian Blood Services. Serological best practices. Updated November 7, 2025. Accessed August 26, 2026.
8. Cooling L. An update on the I blood group system. Immunohematology. 2019;35(3):85-90. PMID: 31621365.
9. Stenfelt L, Hellberg Å, Westman JS, Olsson ML. The P1PK blood group system: revisited and resolved. Immunohematology. 2020;36(3):99-103. doi:10.21307/immunohematology-2020-048.
10. Bio-Rad Medical Diagnostics GmbH. Seraclone Anti-P1 Package Insert. Accessed August 26, 2026.