Linkage studies of schizophrenia: A simulation study of statistical power

Abstract

In planning for a linkage study, it is important to determine the number of pedigrees needed to show linkage. Our study overcomes some of the limitations of previous power studies by simulating multigeneration pedigrees to be compatible with the demographic and genetic epidemiological features of schizophrenia; these are variable age at onset, reduced fertility, and increased mortality after onset. We evaluate the power of these pedigrees by first simulating an ascertainment rule requiring at least three ill family members per pedigree and then simulating the trait and marker genotypes according to a single gene model known to fit epidemiological family study data. Our analysis allows for incomplete and age-dependent penetrance, phenocopies, and interpedigree heterogeneity. We present the power to detect linkage at several lod score thresholds since the multiple tests and phenotypic models required for complex diseases may necessitate using a lod score significance level greater than three. The sample size needed to achieve sufficient power is feasible if 50% of the pedigrees are linked to the marker under test. It may not be feasible to detect linkage if only 25% of the pedigrees are linked, even if a very closely linked marker is used. Our results indicate that to be certain of adequate statistical power, linkage analyses of schizophrenia will require very large samples that do not have a marked degree of genetic heterogeneity.

Power to detect linkage with heterogeneity in samples of small nuclear families

“…At a = 0.25, the power of samples of 200 families was .45 at r = 0 and .10 at r = 0.1, at a threshold of significance of lodh = 3.0; results here were similar in that, for a = 0.25, the smallest sample of S2 families (n = 220) studied had a power of 0.56. For ci = 0.5, Figure 3 in Chen et al [1992] can be used to estimate a power of 0.90 for approximately 90-95 families at r = 0, the same as was seen in the present data midway between two markers .05 M apart.…”

“…Figure 4b (above) suggests by extrapolation that results are similar here: about 80 S3 families would have been required for power = .50 at ci = 0.3, and about 30 such families at a = 0.5. Chen et al [1992] simulated samples of 200 families based on a population and ascertainment model consistent with epidemiological data for schizophrenia, assuming a rare dominant disease allele with very low (0.19) penetrance of the disease genotypes. Pedigrees with three available ill members (following specified extension rules and probabilities of cooperativeness) were selected; most families had three affecteds (but some had more), and a majority spanned three generations.…”

Genome-wide scans of three independent sets of 90 Irish multiplex schizophrenia families and follow-up of selected regions in all families provides evidence for multiple susceptibility genes

“…Without a good estimate of the number of important genes, plausible hypotheses about their mechanism of action, or some idea of the relative roles of epistasis and heterogeneity, debates over 'the' optimal set of ascertainment and analytic strategies will inevitably continue. 6,7 Furthermore, many power analyses [8][9][10][11] have shown that the sample sizes needed to detect susceptibility loci for complex diseases by linkage methods alone are far greater than for Mendelian disorders. Since collection of large samples is difficult and expensive, false-negative linkage results may remain a major problem.…”

Resemblance of psychotic symptoms and syndromes in affected sibling pairs from the Irish Study of High-Density Schizophrenia Families: evidence for possible etiologic heterogeneity

“…The most widely used linkage tests for heterogeneity (14,15) are a posteriori-they divide families into linked and unlinked subgroups solely on the basis of the family-by-family evidence for linkage. This test has only modest power, particularly when applied to the relatively small families most commonly ascertained in schizophrenia linkage studies (16)(17)(18)(19). If it is possible, on clinical grounds, to divide the sample before linkage into etiologically distinct subgroups, a great gain in power is possible.…”