Laboratory

Genomic DNA was extracted from peripheral blood leucocytes. Flanking intron primer sequences for amplification of theNotch3 coding regions were:

exon 3 sense: 5′-TGTGCTGCCCAACCA AGCCA-3′

exon 3 antisense: 5′-ACTGACCACACCC CCGACTA-3′

exon 4 sense: 5′-TAGTCGGGGGTGTG GTCAGT-3′

exon 4 antisense: 5′-CCTCTGACTCTC CTGAGTAG-3′

Polymerase chain reaction conditions used in this study were 95°C for 3 minutes, and 35 cycles of 94°C for 1 minute, 62–65°C for 1 minute, 72°C for 1.5 minutes, and 72°C for 10 minutes. Polymerase chain reaction products of exon 3 (224 bp) and exon 4 (420 bp) were initially confirmed by sequencing on an ABI377. Restriction enzymes were identified for all but three of the molecular variants in exons 3 and 4 of the Notch3 gene. The remaining three variants (G291T, C499T, and G749A) were amplified with newly created mismatch primers (table 1). All DNA samples were then screened using the restriction enzyme diagnostic analysis array in table 2. Samples were digested alongside positive controls where appropriate. Restriction enzymes were deliberately chosen to recognise and cut the common wild-type sequence. This reduced the need to obtain positive controls for many of the rare mutations. Products were resolved using a 3% Metaphor gel.

STATISTICS AND POWER CALCULATIONS

Our study was designed to detect a clinically significant 2% occurrence of CADASIL mutations in stroke, dementia, and control groups with over 94%, 89% and 99% power, respectively. Allele frequencies were calculated using χ² tests. In all cases p<0.05 was considered to be statistically significant.

Results

The diagnostic array was found to work well (table2).

None of the 14 known mutations in exons 3 or 4 was present in 140 stroke, 110 dementia, or 234 control chromosomes.

Known polymorphisms C474A, A587G, and C594A did not show any frequency variations. The remaining polymorphism frequencies are shown in table3. Exon 3 polymorphism C381T occurred in greater frequencies in all cohorts than previously reported.4 Exon 4 polymorphism G684A occurred with lower frequency than previously reported.4 The two polymorphisms C381T and G684A were not in linkage disequilibrium with each other (D=0, p>0.05; χ²=0.68, 4 df).

There were no statistical differences in the allele frequencies between the control and stroke cohorts. In addition, there were no allele frequency differences between the control groups (control and FTD) and disease groups (stroke and dementia). Finally, no differences were seen between the dementia cohort and FTD group (table 3).

Discussion

We have developed a rapid and reliable test for screening the most common genetic mutations in the Notch3 gene using an array of restriction enzymes and mismatch primers that can be utilised by most laboratories. We have used this test to answer the question of whether CADASIL mutations occur more often than would be predicted in patients with sporadic stroke and dementia.5

None of the 14 CADASIL mutations screened were present in a well characterised cohort of patients with sporadic stroke and dementia. The allele frequencies determined here do not show significant variation between selected cohorts, although one polymorphism (C381T) was present in greater frequency and another (G684A) with lower frequency than previously reported.4 Other known polymorphisms did not show variation (A587T is a known rare polymorphism4). The differences in frequencies of molecular variants compared with previous reports4 may be accounted for by ancestral population differences. Interestingly, C381T and G684A were not in linkage disequilibrium with each other and will therefore need to be investigated independently in future studies, unless an appropriate haplotype can be identified.

The search for genes underlying complex disorders such as ischaemic stroke7 and dementia8 is currently topical with various strategies available for their identification.9 Although the rare monogenic disorders do not pose a major health problem, their importance rests with being able to extrapolate data to understand mechanisms underlying the more common phenotypes. At first sight the Notch3 gene seems a prime candidate gene for these diseases. However, our data suggest that the Notch3 gene is not a likely candidate gene for either sporadic stroke or dementia. However, it should be noted that this was not the primary purpose of this study and a larger affected population may need to be studied to detect a small attributable risk, as demonstrated by the angiotensin-1 converting enzyme gene.10

There may be shortcomings with our work that should be documented. Firstly, as with any study the question of power arises. Our study was well powered to detect a clinically significant (taken as⩾2%) occurrence of CADASIL in the common phenotypes. Even if a more conservative clinically significant occurrence of, say, 1% was chosen, the power of our study was 78%. Secondly, patients with stroke and dementia were deliberately chosen with a relatively low age of disease onset in an attempt to match the average disease onset of CADASIL. However, it is possible that in an even younger diseased cohort with stroke and dementia, Notch3 allele frequencies may be different. Finally, as our test for the most common of the CADASIL mutations used previously documented allele variants, the possibility exists that there are other unidentifiedNotch3 molecular variants not tested by us which may occur more often in sporadic stroke and dementia. However, we think that this is unlikely because had they occurred often they are likely to have been detected earlier.4 In addition, during the early testing of our diagnostic protocol direct fluorescent based sequencing failed to identify any new variants.

Clearly, the extent of the clinical phenotype for CADASIL can only be determined by assessing the frequency of mutations in patients with sporadic ischaemic stroke and dementia. Restriction enzyme array with mismatch primers allows both the majority and the most frequent of the mutations of the Notch3 gene to be rapidly screened and regional disease allele frequencies to be easily established, which is critical before any local diagnostic test can be set up.5

CADASIL is probably an underdiagnosed condition,1 and our test will ensure that the diagnosis can be quickly confirmed or excluded whenever there is a clinical suspicion. On the other hand, our data suggest that, despite the clinical resemblance, CADASIL is neither a frequent underlying masquerading cause of sporadic stroke or dementia nor a likely candidate gene for these conditions.