SIR - Panic disorder is a common psychiatric condition defined by recurrent panic attacks and anticipatory anxiety, with a lifetime prevalence of 1-3%.1 Accumulating evidence predicts a familial component to this disorder, with a 2.6- to 20-fold relative risk in the first-degree relatives of probands,2 while twin studies show that about 40% of the liability towards panic disorder consists of heritable factors.3 The two tandemly linked monoamine oxidases (MAOA and MAOB) on the X chromosome are attractive candidate genes for panic disorder because inhibitors of these enzymes (MAOIs) are powerful antipanic agents. Biological studies of MAO activity in patients with panic disorder are equivocal, showing higher,4 lower,5 or no difference.6 In one recent case-control study, variation in a functional promoter polymorphism in the gene for MAOA7 was observed to be associated with panic disorder in females in two patient populations.8 Given this observation and the possible importance of MAO function in panic disorder, we have utilized a family-based design to look for linkage or association between panic disorder and this MAOA polymorphism.
In the current study, 620 individuals in 70 multiplex families, and 81 triads consisting of proband (62 female and 19 male), mother, and father were genotyped. Ascertainment and diagnostic protocols were as previously described.9,10 PCR amplification of the MAOA variable number tandem repeat was performed as described,7 with the addition of 1 M anhydrous betaine. PCR products were separated by electrophoresis on a 4% polyacrylamide gel on an ABI Prism 377 DNA Sequencer, and were analyzed using GENESCAN software. Genotypes were scored by PCR product size and numbered as in Deckert et al8 (3, 330 bp; 3a, 345 bp; 4, 360 bp; 5, 390 bp). These alleles, reflecting repeat number, correspond to numbers 1 to 4, respectively, in Sabol et al.7 Alleles in the 'triad' families were analyzed with the haplotype-based haplotype relative risk (HHRR) method,11,12 the transmission disequilibrium test (TDT) statistic,12 and the contingency table  2 tests. Monte-Carlo simulations were performed with the CLUMP program to assess the significance of the derived 2 test statistics for the large contingency tables with low numbers of rare alleles.13 Pedigrees were analyzed for linkage using both dominant and recessive genetic models of inheritance and a penetrance of 50%, assuming both genetic heterogeneity and homogeneity, under three diagnostic models (narrow, intermediate, and broad) using the FASTLINK package with the X-linked option for X-linked markers.14 There are 339, 385, and 431 affecteds in our pedigrees under the narrow, intermediate, and broad models, repectively. The triad probands all met criteria for definitely or probably affected (broad model).
We observed allele frequencies of 3¾36.2%; 3a¾2.9%; 4¾60.5% and 5¾0.4% for the 162 parents of the 81 triads. We did not observe the rare 2-repeat allele seen by Deckert et al. No significant difference in allele distribution of the maternally inherited alleles was seen in triads (Table 1, 2 = 1.75, 3 df, NS). The HHRR (P = 0.88 for HHRR-2xn, P = 0.50 for HHRR-LRT) and the TDT (P = 0.47) also showed no significant association. Sabol et al7 report a 2- to 10-fold difference in MAOA transcription between cell lines containing the 3a or 4 allele when compared with those with the 3 or 5 repeat alleles. When we compared allelic frequencies in the probands of the triads by grouping maternally inherited alleles (ie, 3a + 4 vs 3 + 5), there were no significant differences overall or by proband gender (data not shown). Since Deckert et al reported that MAOA transcription was higher with alleles 3a, 4, and 5, we also stratified allelic frequencies in the probands of the triads in this manner. Since only one maternal allele 5 was seen, again no significant association was seen. Sabol et al and Deckert et al report different transcriptional activities of their gene constructs, despite using the same human neuroblastoma cell line. This discrepancy may be due to a slight difference in their expression constructs. Although both constructs have nearly identical 3' ends, the fusion gene used by Deckert et al is 64 base pairs longer at the 5' end. A search of this region for DNA binding sites revealed potential Sp1, GC box, and AP-1 sites,15 potentially explaining the difference in expression patterns seen between the two groups. Panic disorder is about twice as frequent in females as males in the general population and Deckert et al only observed an association in females, thus we examined the triads when stratified by the sex of the proband and observed no relationship between panic disorder and MAOA promoter alleles (data not shown). We are unable to analyze our triad data for genotypic differences given the marker is on the X chromosome, resulting in a hemizygous father who does not pass on a non-transmitted control allele to the 'virtual control', but we note that Deckert et al only found significant differences in genotype frequencies by defining genotype in a manner at odds with their a priori hypothesis. Alleles were divided into short (low transcriptional activity) and long (high transcriptional activity) alleles, while genotypes are grouped into short/short and short/long vs long/long genotypes, without an explanation as to why short/long should be grouped with short/short and not separately or with long/long genotypes.
The allele frequencies of 220 independent alleles in the 70 multiplex families were 3¾44.1%; 3a¾3.2%; 4¾51.8% and 5¾0.9%. A lod score linkage analysis of these families was performed using several diagnostic and genetic models, as well as under models of genetic homogeneity or heterogeneity. No significant lod score was seen with any of the diagnostic and genetic models, or with analyses using homogeneity or heterogeneity. Z-max and Z-max under heterogeneity were zero for all models tested, with 
= 0.5. 'Non-parametric' tests, including SIBPAIR, TDT and the XRC-TDT,16 also showed no significant results, with P-values = 0.5 in both tests.
One major difference between this study and that of Deckert et al, is that this study has a family-based design, which reduces false positive results due to population stratification. Another possible issue relating to our non-replication concerns the low power to discern association with the relative risk cited by Deckert et al. Given X chromosomal transmission, this effectively reduces our 62 female triad probands to 31, which may be insufficient to detect association with a polymorphism that presumably confers a low relative risk. In conclusion, our results provide no additional evidence for the role of this MAOA promoter polymorphism in panic disorder. However, given the strong a priori hypothesis for the involvement of this gene in the pathogenesis of the disorder and the observed genetic association,8 further investigation with other samples and/or polymorphisms is probably warranted.
|
