Interpretation of Linkage Data for a Huntington-Like Disorder Mapping to 4p15.3

Abstract

To The Editor: Kambouris et al. (2000) report on the mapping of a neurodegenerative disorder on the basis of a sibship of 10 individuals whose parents are first cousins. Using a model of autosomal recessive inheritance, linkage analysis detects a maximum two-point LOD score (Zmax) of 3.03 at recombination fraction (θ) 0. The authors of the report postulate the genetic interval as a 7-cM region bounded by D4S2366 and D4S2983, because all affected individuals are homozygous for the two markers (D4S431 and D4S394) in between. Figure 1 in their article demonstrates a haplotype analysis in which the parents (III:2 and III:3), although first cousins, share very few alleles in the putative linked region. First, the marker order presented in the report's figure 2b contradicts that presented in its haplotype analysis (fig. 1) and in the multipoint analysis (fig. 2a). The Marshfield sex-averaged linkage map places D4S2366 between D4S431 and D4S394. The haplotype and multipoint analyses place D4S2366 centromeric to D4S431 and D4S394. Since the parents share no alleles for D4S2366, interposing D4S2366 between D4S431 and D4S394 would abolish this region of putative homozygosity by descent among the affected individuals. It appears more likely that it is by chance alone that the two parents share a “2” allele for D4S431 and a “1” allele for D4S394. For example, the Foundation Jean Dausset CEPH genotype database reveals that the most common allele (205 bp) for D4S394 has a frequency of 41%. Thus, if allele 1 for D4S394 in the report's figure 1 is the 205-bp allele, the chances are 41% that parent III:2 inherited the 1 allele from the unrelated parent (II:1). Without genotype data for the parents and/or siblings of III:2 and III:3, identity by descent cannot be assumed. Kambouris et al. make the assumption that the disorder is recessive, apparently because of the consanguinity in the family. Although they report Zmax = 3.03 at θ = 0, under the assumption of 50% penetrance, the two-point LOD scores were likely calculated under a model of 100% penetrance. The two-point LOD scores would be expected to be lower under a model of 50% penetrance (two-point LOD score 2.7 at θ = 0 for the four fully linked markers). The data could also support a model of autosomal dominance with reduced penetrance with the disorder segregating with the red haplotype, if the disease is not penetrant in parent III:2 and individual IV:8. The same argument could be made for parent III:3 and individual IV:10 and the purple haplotype. Testing a dominant model assuming 90% penetrance demonstrated a Zmax of 1.94 at θ = 0, with marker D4S412 (data not shown). Even if it were assumed that the mode of inheritance is truly autosomal recessive, homozygous genotypes among the affected individuals are not absolutely required. If the linkage to this region is true, and if the red and purple haplotypes contain noncomplementing mutated alleles, the genetic interval would actually be defined by the telomeric recombination event in IV:2 and the centromeric recombination events in IV:4—that is, by D4S3023 and D4S1599, defining a nonrecombinant region of 15 cM. Finally, Kambouris et al. note that only chromosome 4 markers were genotyped. Testing markers at the already mapped locus on chromosome 20, for a similar Huntington-like disorder, would certainly seem pertinent. A two-point LOD score of 3.3 (not 3.0) is the generally accepted criterion for a 5% significance level (Lander and Schork 1994). A complete genome screen may well reveal another locus in which the parents are heterozygous for a common haplotype with a more convincing region of homozygosity.