Sources of selection bias in evaluating social programs: An interpretation of conventional measures and evidence on the effectiveness of matching as a program evaluation method

This paper decomposes the conventional measure of selection bias in observational studies into three components. The first two components are due to differences in the distributions of characteristics between participant and nonparticipant (comparison) group members: the first arises from differences in the supports, and the second from differences in densities over the region of common support. The third component arises from selection bias precisely defined. Using data from a recent social experiment, we find that the component due to selection bias, precisely defined, is smaller than the first two components. However, selection bias still represents a substantial fraction of the experimental impact estimate. The empirical performance of matching methods of program evaluation is also examined. We find that matching based on the propensity score eliminates some but not all of the measured selection bias, with the remaining bias still a substantial fraction of the estimated impact. We find that the support of the distribution of propensity scores for the comparison group is typically only a small portion of the support for the participant group. For values outside the common support, it is impossible to reliably estimate the effect of program participation using matching methods. If the impact of participation depends on the propensity score, as we find in our data, the failure of the common support condition severely limits matching compared with random assignment as an evaluation estimator.

The major murine systemic lupus erythematosus susceptibility locus, Sle1, is a cluster of functionally related genes

The major murine systemic lupus erythematosus (SLE) susceptibility locus Sle1 is syntenic to a chromosomal region linked with SLE susceptibility in multiple human studies. Congenic analyses have shown that Sle1 breaks tolerance to chromatin, a necessary step for full disease induction that can be suppressed by specific modifier loci. In the present study, our fine mapping analysis of the location of Sle1 has determined that three loci within this congenic interval, termed Sle1a, Sle1b, and Sle1c, can independently cause a loss of tolerance to chromatin. Each displays a distinctive profile of serological and cellular characteristics, with T and B cell functions being more affected by Sle1a and Sle1b, respectively. The epistatic interactions of Sle1 with other susceptibility loci to cause severe nephritis cannot be accounted, however, by these three loci alone, suggesting the existence of an additional locus, termed Sle1d. These findings indicate that the potent autoimmune phenotype caused by the Sle1 genomic interval reflects the combined impact of four, separate, susceptibility genes. This level of genetic complexity, combined with similar findings in other systems, supports the possibility that many complex trait loci reflect the impact of polymorphisms in linked clusters of genes with related functions.

Genome scan of human systemic lupus erythematosus: Evidence for linkage on chromosome 1q in African-American pedigrees

Systemic lupus erythematosus (SLE) is an autoimmune disorder characterized by production of autoantibodies against intracellular antigens including DNA, ribosomal P, Ro (SS-A), La (SS-B), and the spliceosome. Etiology is suspected to involve genetic and environmental factors. Evidence of genetic involvement includes: associations with HLA-DR3, HLA-DR2, Fcγ receptors (FcγR) IIA and IIIA, and hereditary complement component deficiencies, as well as familial aggregation, monozygotic twin concordance >20%, λs > 10, purported linkage at 1q41–42, and inbred mouse strains that consistently develop lupus. We have completed a genome scan in 94 extended multiplex pedigrees by using model-based linkage analysis. Potential [log10 of the odds for linkage (lod) > 2.0] SLE loci have been identified at chromosomes 1q41, 1q23, and 11q14–23 in African-Americans; 14q11, 4p15, 11q25, 2q32, 19q13, 6q26–27, and 12p12–11 in European-Americans; and 1q23, 13q32, 20q13, and 1q31 in all pedigrees combined. An effect for the FcγRIIA candidate polymorphism) at 1q23 (lod = 3.37 in African-Americans) is syntenic with linkage in a murine model of lupus. Sib-pair and multipoint nonparametric analyses also support linkage (P < 0.05) at nine loci detected by using two-point lod score analysis (lod > 2.0). Our results are consistent with the presumed complexity of genetic susceptibility to SLE and illustrate racial origin is likely to influence the specific nature of these genetic effects.

Role of the major histocompatibility complex class II Ea gene in lupus susceptibility in mice

The gene(s) encoded within major histocompatibility complex (MHC) act as one of the major genetic elements contributing to the susceptibility of murine systemic lupus erythematosus (SLE). We have recently demonstrated that lupus susceptibility is more closely linked to the I-E− H-2b haplotype than to the I-E+ H-2d haplotype in lupus-prone BXSB and (NZB × BXSB)F1 hybrid mice. To investigate whether the reduced susceptibility to SLE in H-2d mice is related to the expression of the MHC class II Ea gene (absent in H-2b mice), we determined the possible role of the Ea gene as a lupus protective gene in mice. Our results showed that (i) the development of SLE was almost completely prevented in BXSB (H-2b) mice expressing two copies of the Ead transgene at the homozygous level as well as in BXSB H-2k (I-E+) congenic mice as for H-2d BXSB mice, and (ii) the expression of two functional Ea (transgenic and endogenous) genes in either H-2d/b (NZB × BXSB)F1 or H-2k/b (MRL × BXSB)F1 mice provided protection from SLE at levels comparable to those conferred by the H-2d/d or H-2k/k haplotype. In addition, the level of the Ea gene-mediated protection appeared to be dependent on the genetic susceptibility to SLE in individual lupus-prone mice. Our results indicate that the reduced susceptibility associated with the I-E+ H-2d and H-2k haplotypes (versus the I-E− H-2b haplotype) is largely, if not all, contributed by the apparent autoimmune suppressive effect of the Ea gene, independently of the expression of the I-A or other MHC-linked genes.

Surface blebs on apoptotic cells are sites of enhanced procoagulant activity: implications for coagulation events and antigenic spread in systemic lupus erythematosus.

The restriction of phosphatidylserine (PtdSer) to the inner surface of the plasma membrane bilayer is lost early during apoptosis. Since PtdSer is a potent surface procoagulant, and since there is an increased incidence of coagulation events in patients with systemic lupus erythematosus (SLE) who have anti-phospholipid antibodies, we addressed whether apoptotic cells are procoagulant and whether anti-phospholipid antibodies influence this. Apoptotic HeLa cells, human endothelial cells, and a murine pre-B-cell line were markedly procoagulant in a modified Russell viper venom assay. This procoagulant effect was entirely abolished by addition of the PtdSer-binding protein, annexin V, confirming that it was PtdSer-dependent. The procoagulant effect was also abolished by addition of IgG purified from the plasma of three patients with anti-phospholipid antibody syndrome, but not IgG from normal controls. Confocal microscopy of apoptotic cells stained with fluorescein-isothiocyanate-conjugated-annexin V demonstrated (Ca2+)-dependent binding to the surface of membrane blebs o apoptotic cells, but not to intracellular membranes. Recent data indicate that the surface blebs of apoptotic cells constitute an important immunogenic particle in SLE. We propose that the PtdSer exposed on the outside of these blebs can induce the production of anti-phospholipid antibodies, which might also enhance the immunogenicity of the bleb contents. When apoptosis occurs in a microenvironment in direct contact with circulating plasma, the unique procoagulant consequences of the apoptotic surface may additionally be expressed. This might explain the increased incidence of pathological intravascular coagulation events that occur in some lupus patients who have anti-phospholipid antibodies.

In vivo expression of a single viral DNA-binding protein generates systemic lupus erythematosus-related autoimmunity to double-stranded DNA and histones.

Although the origin of autoimmune antibodies to double-stranded DNA is not known, the variable-region structures of such antibodies indicate that they are produced in response to antigen-selective stimulation. In accordance with this, results from experiments using artificial complexes of DNA and DNA-binding polypeptides for immunizations have indicated that DNA may induce these antibodies. Hence, the immunogenicity of DNA in vivo may depend upon other structures or processes that may render DNA immunogenic. We report that in vivo expression of a single DNA-binding protein, the polyoma virus T antigen, is sufficient to initiate production of anti-double-stranded DNA and anti-histone antibodies but not a panel of other autoantigens. Expression of a mutant, non-DNA-binding T antigen did result in strong production of antibodies to the T antigen, but only borderline levels of antibodies to DNA and no detectable antibodies to histones. Nonexpressing plasmid DNA containing the complete cDNA sequence for T antigen did not evoke such immune responses, indicating that DNA by itself is not immunogenic in vivo. The results represent a conceptual advance in understanding a potential molecular basis for initiation of autoimmunity in systemic lupus erythematosus.