phenotype (HLA antigens expressed) but doesn’t identify which alleles are linked together on the chromosome as a haplotype. The only way to know for sure is through family segregation analysis. When family data are not available, the EM algorithm has been used to predict haplotypes (45,52).

When two or more genes are on the same chromosome, they are said to be linked. When alleles of linked genes occur in haplotypes more frequently than would be expected on the basis of chance alone, those genes are said to be in linkage disequilibrium. The HLA gene complex is in linkage disequilibrium (7, 10). Apparently high disequilibrium across the DR-DQ subregion coupled with a lack of recombination have resulted in specific associations between DQA1 and DQB1 alleles and between DRB1 and DQ alleles.

These associations may differ among individuals of different racial/ethnic backgrounds. At the level of class I–class II associations, the best known example of linkage disequilibrium is the HLA-A1, -Cw7, -B8, -DR3, -DR52, -DQ2 haplotype, which occurs approximately four times more frequently than would be expected by chance. It is thought that combinations such as this one (often called extended haplotypes) account for at least 30 percent of HLA allele combinations in whites.

The frequencies of HLA alleles and haplotypes found in individuals differ among ethnic/racial groups (9,38,43,77). Some alleles and haplotypes are common to several populations; others may be predominantly confined to one particular population group. Most of the alleles are rarely observed. For example, in the United States, DRB1*03 (or DR3) is carried by 10–23 percent of four U.S. population groups (Table F-2) (77). The allele DRB1*0301 is common to most population groups, but DRB1*0302