Abstracts 121

Clinical follow-up of the patient showed normal CH50(198) , C4(36), and C3(114) levels, and negative results for rheumatoid factor and ANA. H e r EBV titer was 1:640(2.85). Peripheral blood mononuclear cells had normal morphology by Wright 's stain. They showed a decrease in T cel ls(27-37%), an increase in MY4 + (50%) and I a + ( 3 8 - 6 3 % ) cells, and 9 - 1 6 % B1 + cells, by fluorescence. Mixed cells or nylon wool enriched T cells could not be sustained in continuous culture with N H S , while nylon wool adherent cells assumed lymphoblastoid morphology within 48 hr and continued to grow. CTL against 8 - 1 0 % of au- tologous peripheral blood cells could be generated by culture of the patient 's T cells with D R incompatible stimulator cells. These CTL were not active against A1, B8, or DR3 antigens of a third-party target cell. This patient has made an HLA-B8 specific autoantibody in the setting of predisposition to auto immune disease and recent EBV infection. Autoreactive T cytotoxic cells may be re- sponsible for the decline in the number of EBV transformed cells producing specific autoantibody, as suggested by autologous killing in vitro.

POWER OF EXCLUSION OF THE HLA-A,B SYSTEM IN PATERNITY TESTING. Richard H. Walker, Michael A. Meyers, and Linda M. Phillips; Beaumont Hospital. Royal Oak. MI

Although the class I antigens of the M H C are widely used as genetic markers in paternity tests, their current usefulness in this application has not been well established. One of the best criteria for utility is the ability of a genetic system to exclude a man falsely accused of pa te rn i ty - - the power of exclusion /PE). Various formulas have been proposed to calculate this value but they either fail to consider linkage disequilibrium between the A and B genes, amorphs, or codominance, or they are so complex that a computer is needed to derive the PE value. This study created false trios by moving the alleged father one case forward in our chronological series of paternity cases from the past 5 yr. It was assumed then that all of the men in these artificial trios were innocent and the HLA-A,B markers were subsequently examined for evidence of an exclusion. Black and white races were not mixed so that each race could be evaluated independently. Since the artificial trios were created by shifts in the next con- tiguous chronological position for the alleged fathers, in nearly every instance the same HLA reagents (lot number of typing trays) were used to define the antigens of each m e m b e r of the new trios. All individual typings required a minimum of two trays for the definition of the HLA antigens present. The repor ted phenotypes were examined independently by each of the co-authors for evidence of a direct and/or indirect exclusion. Seventeen percent of the exclusions were based on A locus markers, 38% were B locus markers, and 4 5 % were based on both A and B markers. Among 479 false white trios, the man was excluded in 447 cases yielding a power of exclusion for whites of 93 -+ 1.1%. One hundred and twenty-four men were excluded in 134 black artificial trios giving a value of 92 -+ 2 .3% for the PE in blacks for the HLA-A,B system. The average exclusion chance of a tested man (1-RMNE) was also derived from our data on the same cases and was found to be 94% for whites and 93% for blacks. This close agreement enhances the validity of the PE values derived from our artificial trios. The comparable results in whites and blacks were surprising since the much higher frequency of blanks in the blacks was expected to produce a relatively lower PE value for blacks. Analysis of the phenotypes of the two groups revealed greater heterogenei ty in the blacks which compensated for their higher frequency of blanks. A formula has been developed for defining phenotypic heterogenei ty which may prove useful in evaluating the HLA-A,B PE for races other than black and white.