In the past decade there have been a number of attempts at identifying AD linkage regions using affected sib-pairs or extended families [8
]. Although results have varied considerably and have sometimes been difficult to replicate, the most convincing linkage peaks have been reported from chromosomes 9, 10, 12, and 19.
In the present study, linkage peaks on chromosomes 1, 9, 10, 12, 19 and 21, previously implicated in a whole genome scan by Myers et al
. have been further analyzed using a modified version of the original sample with an added collection of ARPs from Sweden and Washington University. We could detect significant linkage to chromosome 19q13 in the immediate vicinity of the APOE
locus. This linkage peak was noticeably increased from the scan by Myers et al
. and the Swedish sample contributed considerably to the improved linkage (Table ). The influence of the APOE
locus on AD has been correlated to a lower age at onset [4
], as further demonstrated in our recent analysis of the chromosome 19 linkage [25
]. Accordingly, linkage analysis of the NIMH cohort by Blacker et al
. demonstrated the highest linkage to chromosome 19q13 in their subsample with earlier disease onset, whereas no linkage to this region was detected in their late onset sample [8
]. The only other significant linkage found in the present study was to chromosome 1p36 in the UK subsample. However, this peak was neither detected in the other subgroups, nor in the total sample.
In the original whole genome scan by Myers et al
., the most significant linkage peak was demonstrated on chromosome 10q21 (82 cM) in the whole sample. Blacker et al
. also found linkage in the region, to chromosome 10q22 (92 cM) in their total collection of NIMH samples. In the present study, we could not detect linkage to chromosome 10q21, even though a suggestive linkage of MLS 1.8 was detected to chromosome 10q22 (105 cM) in the APOE ε4-
sample. Although the sample size in the present study is smaller than in the study by Myers et al
. (380 and 451 ASPs, respectively), the previous study by Kehoe et al
] using 292 ASPs and including overlapping samples with the Myers study also detected linkage to chromosome 10q21. This suggests that the absence of a linkage peak on chromosome 10q21 in the present study might be due to sample differences between the studies rather than sample size. However, we cannot completely exclude that our linkage to chromosome 10q22 in the APOE ε4-
subsample coincides with the previously detected linkage to chromosome 10q21-22, although the positions of these peaks differ by 13-23 cM. Certain caution is also called for as the APOE ε4-
subsample is rather limited in size (42 ARPs).
Inconsistent results between linkage studies might reflect heterogeneity in sample cohorts, including age at onset, ethnic background and diagnostic criteria. Our finding of significant linkage to chromosome 19q13, but to no other regions in the total sample in combination with the results presented by Blacker et al., suggests that finding significant linkage to both chromosome 19q13 and additional regions in the same sample is uncommon.
In the past few years, whole genome association studies have successfully identified susceptibility loci for a number of complex conditions. However, APOE
is so far the only locus demonstrating strong association to AD [26
]. Sample sizes have turned out to be crucial and sample collections including thousands of cases have been analyzed for association [29
]. Increasing the number of samples in analyses also of ARPs would most likely be beneficial for the outcome and therefore further efforts to combine different sample collections should be made. It has been suggested that data from linkage analysis of affected sib-pairs could also be used to verify candidate susceptibility genes from association studies, since the frequency of a risk allele is expected to be higher in siblings sharing the locus than in population based cases [30