Skip Navigation


Human Molecular Genetics Advance Access originally published online on January 13, 2005
Human Molecular Genetics 2005 14(4):543-553; doi:10.1093/hmg/ddi051
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
14/4/543    most recent
ddi051v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (15)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Soares, M. L.
Right arrow Articles by Buxbaum, J. N.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Soares, M. L.
Right arrow Articles by Buxbaum, J. N.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Human Molecular Genetics, Vol. 14, No. 4 © Oxford University Press 2005; all rights reserved

Susceptibility and modifier genes in Portuguese transthyretin V30M amyloid polyneuropathy: complexity in a single-gene disease

Miguel L. Soares1,2, Teresa Coelho3,6, Alda Sousa4,5, Serge Batalov7, Isabel Conceição8, Maria L. Sales-Luís8, Marylyn D. Ritchie9,10, Scott M. Williams9,10,11, Caroline M. Nievergelt12, Nicholas J. Schork1,12, Maria João Saraiva2,5 and Joel N. Buxbaum1,*

1Department of Molecular and Experimental Medicine, Division of Rheumatology Research and the W.M. Keck Autoimmune Disease Center, The Scripps Research Institute, La Jolla, CA, USA, 2Amyloid Unit, Instituto de Biologia Molecular e Celular (IBMC), Porto, Portugal, 3Neuropsychophysiology Unit and 4UnIGENe, IBMC, 5Instituto de Ciências Biomédicas Abel Salazar, Porto, Portugal, 6Centro de Estudos de Paramiloidose, Porto, Portugal, 7Computational Biology Department, The Genomics Institute of the Novartis Research Foundation, San Diego, CA, USA, 8Department of Neurology, Faculty of Medicine, Centro de Estudos Egas Moniz, Hospital de Santa Maria, Lisbon, Portugal, 9Center for Human Genetics Research, 10Department of Molecular Physiology and Biophysics, 11Division of Cardiovascular Medicine, Department of Medicine, Vanderbilt University Medical Center, Nashville, TN, USA and 12Department of Psychiatry, University of California at San Diego, La Jolla, CA, USA

Received October 12, 2004; Accepted December 23, 2004

GenBank accession no. M11518


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
Familial amyloid polyneuropathy type I is an autosomal dominant disorder caused by mutations in the transthyretin (TTR) gene; however, carriers of the same mutation exhibit variability in penetrance and clinical expression. We analyzed alleles of candidate genes encoding non-fibrillar components of TTR amyloid deposits and a molecule metabolically interacting with TTR [retinol-binding protein (RBP)], for possible associations with age of disease onset and/or susceptibility in a Portuguese population sample with the TTR V30M mutation and unrelated controls. We show that the V30M carriers represent a distinct subset of the Portuguese population. Estimates of genetic distance indicated that the controls and the classical-onset group were furthest apart, whereas the late-onset group appeared to differ from both. Importantly, the data also indicate that genetic interactions among the multiple loci evaluated, rather than single-locus effects, are more likely to determine differences in the age of disease onset. Multifactor dimensionality reduction indicated that the best genetic model for classical onset group versus controls involved the APCS gene, whereas for late-onset cases, one APCS variant (APCSv1) and two RBP variants (RBPv1 and RBPv2) are involved. Thus, although the TTR V30M mutation is required for the disease in Portuguese patients, different genetic factors may govern the age of onset, as well as the occurrence of anticipation.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
Autosomal dominant disorders may vary in expression even within a given kindred. The basis of this variability is uncertain and can be attributed to epigenetic factors, environment or epistasis. We have studied familial amyloid polyneuropathy (FAP), an autosomal dominant disorder characterized by peripheral sensorimotor and autonomic neuropathy. It exhibits variation in cardiac, renal, gastrointestinal and ocular involvement, as well as age of onset. Over 80 missense mutations in the transthyretin gene (TTR) result in autosomal dominant disease (http://www.ibmc.up.pt/~mjsaraiv/ttrmut.html). The presence of deposits consisting entirely of wild-type TTR molecules in the hearts of 10–25% of individuals over age 80 reveals its inherent in vivo amyloidogenic potential (1).

FAP was initially described in Portuguese (2) where, until recently, the TTR V30M has been the only pathogenic mutation associated with the disease (3,4). Later reports identified the same mutation in Swedish and Japanese families (5,6). The disorder has since been recognized in other European countries and in North American kindreds in association with V30M, as well as other mutations (7).

TTR V30M produces disease in only 5–10% of Swedish carriers of the allele (8), a much lower degree of penetrance than that seen in Portuguese (80%) (9) or in Japanese with the same mutation. The actual penetrance in Japanese carriers has not been formally established, but appears to resemble that seen in Portuguese. Portuguese and Japanese carriers show considerable variation in the age of clinical onset (10,11). In both populations, the first symptoms had originally been described as typically occurring before age 40 (so-called ‘classical’ or early-onset); however, in recent years, more individuals developing symptoms late in life have been identified (11,12). Hence, present data indicate that the distribution of the age of onset in Portuguese is continuous, but asymmetric with a mean around age 35 and a long tail into the older age group (Fig. 1) (9,13). Further, DNA testing in Portugal has identified asymptomatic carriers over age 70 belonging to a subset of very late-onset kindreds in whose descendants genetic anticipation is frequent. The molecular basis of anticipation in FAP, which is not mediated by trinucleotide repeat expansions in the TTR or any other gene (14), remains elusive.



View larger version (22K):
[in this window]
[in a new window]
  		Figure 1. Distribution of FAP patients by age of disease onset. The curve shows the empirical distribution of the age of onset in 1208 Portuguese patients (13). The bars show distribution of the FAP patient sample used in this study by age of onset. Individuals were chosen to be representative of distinct classes of age of onset.

 
Variation in penetrance, age of onset and clinical features are hallmarks of many autosomal dominant disorders including the human TTR amyloidoses (7). Some of these clearly reflect specific biological effects of a particular mutation or a class of mutants. However, when such phenotypic variability is seen with a single mutation in the gene encoding the same protein, it suggests an effect of modifying genetic loci and/or environmental factors contributing differentially to the course of disease. We have chosen to examine age of onset as an example of a discrete phenotypic variation in the presence of the particular autosomal dominant disease-associated mutation TTR V30M. Although the role of environmental factors cannot be excluded, the existence of modifier genes involved in TTR amyloidogenesis is an attractive hypothesis to explain the phenotypic variability in FAP. The strongest evidence that the variable pathology of FAP V30M is influenced by environmental factors is the great disparity in penetrance between the Swedish and Portuguese gene carriers, given the substantial differences in climate and terrain between the North of Portugal and the North of Sweden, although direct experimental support for such an effect is lacking. On the other hand, there are also substantial genetic differences between the Swedish and Portuguese populations.

ATTR (TTR amyloid), like all amyloid deposits, contains several molecular components, in addition to the quantitatively dominant fibril-forming amyloid protein, including heparan sulfate proteoglycan 2 (HSPG2 or perlecan), SAP, a plasma glycoprotein of the pentraxin family (encoded by the APCS gene) that undergoes specific calcium-dependent binding to all types of amyloid fibrils, and apolipoprotein E (ApoE), also found in all amyloid deposits (15). The ApoE4 isoform is associated with an increased frequency and earlier onset of Alzheimer's disease (Aß), the most common form of brain amyloid, whereas the ApoE2 isoform appears to be protective (16). ApoE variants could exert a similar modulatory effect in the onset of FAP, although early studies on a limited number of patients suggested this was not the case (17).

In at least one instance of senile systemic amyloidosis, small amounts of AA-related material were found in TTR deposits (18). These could reflect either a passive co-aggregation or a contributory involvement of protein AA, encoded by the serum amyloid A (SAA) genes and the main component of secondary (reactive) amyloid fibrils, in the formation of ATTR.

Retinol-binding protein (RBP), the serum carrier of vitamin A, circulates in plasma bound to TTR. Vitamin A-loaded RBP and L-thyroxine, the two natural ligands of TTR, can act alone or synergistically to inhibit the rate and extent of TTR fibrillogenesis in vitro, suggesting that RBP may influence the course of FAP pathology in vivo (19).

We have analyzed coding and non-coding sequence polymorphisms in the RBP4 (serum RBP, 10q24), HSPG2 (1p36.1), APCS (1q22), APOE (19q13.2), SAA1 and SAA2 (11p15.1) genes with the goal of identifying chromosomes carrying common and functionally significant variants. At the time these studies were performed, the full human genome sequence was not completed and systematic single-nucleotide polymorphism (SNP) analyses were not available for any of the suspected candidate genes. We identified new SNPs in APCS and RBP4 and utilized polymorphisms in SAA, HSPG2 and APOE that had already been characterized and shown to have potential pathophysiologic significance in other disorders (16,20–22). The genotyping data were analyzed for association with the presence of the V30M amyloidogenic allele (FAP patients versus controls) and with the age of onset (classical- versus late-onset patients). Multi-locus analyses were also performed to examine the effects of simultaneous contributions of the six loci for determining the onset of the first symptoms.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
APCS and RBP4 variation survey
APCS is located on chromosome 1q22 and spans ~1.4 kb with two exons of 160 and 464 bp (23). To analyze APCS for sequence variation, a region of 1389 bp was sequenced from genomic DNA of 32 FAP subjects. The contigs of each individual were matched to a baseline reference sequence obtained from Genbank (accession no. D00097) to identify DNA variants. Data for the allelic variants of the polymorphisms identified in APCS are shown in Table 1. In total, five biallelic variants were identified across the scanned region. A5u.1 and A2.2 were observed together in a single individual among the 32 subjects surveyed. These rare variants were confirmed by sequencing newly amplified fragments, in addition to the opposite strand sequence agreement. The compound heterozygote developed the first symptoms at age 64 (late onset) and was the progenitor of a classical-onset individual in a kindred exhibiting genetic anticipation. It was not determined whether the substitutions were carried in the same or different alleles. The frequencies of the common SNPs were similar in classical- and late-onset patients.


View this table:
[in this window]
[in a new window]
  		Table 1. Sequence variants identified in APCS and RBP4
 
To assess variation in RBP4, we used a localized restriction fragment length polymorphism (RFLP) approach, focusing on intronic and non-coding sequences. Two overlapping amplicons spanning part of intron 4 and a 1 kb fragment upstream of exon 1 were digested with eight endonucleases, revealing three different electrophoretic patterns. The nature and position of the variable sites identified are shown in Table 1. Flanking polymorphism R5u.1 was typed for a subset of patients, but was not used in the association analyses. Genotype counts among 68 samples were 32 C/C, 35 C/T and 1 T/T, with allele frequencies of 0.73 and 0.27.

Single-locus allele and genotype frequencies, age of onset and disease susceptibility
Single-locus effects were compared by allele and genotype frequencies of each of the 10 variants at the six candidate loci. Variation at the HSPG2, APOE –491A/T, APOE, SAA1, SAA2 and RBP4 loci indicated no significant associations between genotype or allele frequency and phenotypes (Supplementary Material). The genotypic distributions at these loci in all groups did not deviate significantly from Hardy–Weinberg (H–W) equilibrium (data not shown).

FAP patients and controls differed significantly in allele and genotype frequencies for APCSv1 and APCSv2 (Table 2). The frequency of the APCSv1 G allele was greater in the V30M carriers than in the controls (P < 0.0001). Accordingly, genotype frequencies for heterozygotes and G/G homozygotes were also higher among the FAP subjects (P<0.0001). When the A/A genotype was used as the reference, the A/G genotype was eight times more common in the TTR V30M carriers (OR=8.76, 95% CI=2.52–30.44). For APCSv2, genotype G/G was less frequent among the mutation carriers (P=0.0143), in agreement with an increased frequency of the A allele relative to the controls (P=0.0178), although the differences were not statistically significant when corrected for multiple comparisons. Genotype G/A showed a 2-fold increased frequency among the FAP individuals, using genotype A/A as a reference (OR=2.58, 95% CI=1.31–5.11). The genotypic distribution at the two loci in both groups did not deviate significantly from H–W equilibrium.


View this table:
[in this window]
[in a new window]
  		Table 2. Genotype and allele frequencies at the APCS locus among FAP subjects and controls
 
Polymorphisms at APCS and RBP4 were then analyzed as haplotypes to compare the carrier and non-carrier populations, as well as to determine whether there were any associations with the age of disease onset. Table 3 shows the allelic and genotypic frequencies of haplotypes at the two loci among the FAP classes and the control group. The distribution of APCS haplotypes was significantly different between patients and controls (P<0.0001). Haplotype A-G-G was significantly less frequent among the FAP subjects than in the controls (P<0.0001). Conversely, the frequency of haplotype G-G-G was significantly increased in the V30M carriers when compared with the controls (P<0.0001), with an odds ratio of 8.88 (95% CI=2.63–30.01). Accordingly, the proportion of V30M carriers with the G-G-G/G-G-G and G-G-G/A-G-G genotypes was higher than in the control population (P=0.0002) (OR=7.73, 95% CI=2.20–27.14). These differences are significant even when corrected for multiple comparisons. Furthermore, the haplotypes at the APCS locus were out of H–W equilibrium in the FAP group (P=0.026), indicating that the alleles were not randomly combined to form genotypes within the FAP population.


View this table:
[in this window]
[in a new window]
  		Table 3. Genotypic and allelic frequencies of APCS and RBP4 haplotypes among FAP subjects and controls
 
No significant differences were detected between classical-onset cases and late-onset cases for any single variant. However, haplotype analyses revealed a higher prevalence of haplotype A-G-G among the late-onset cases (71.0%) than that observed in classical-onset cases (53.0%, P=0.019). Accordingly, A-G-G homozygotes were more frequent, although not significantly so, in the late-onset class (47.0 versus 24.0%, P=0.039). In addition, the APCS locus deviated from the H–W expectations in the classical-onset (P=0.008), but not in the late-onset class. Independent comparisons of each of the FAP onset subsets with the controls revealed that the genotypic distribution of haplotypes in the classical-onset group differed significantly from that in the controls (P<0.0001), whereas the late-onset class did not (P=0.109).

RBP4 haplotype frequencies (Table 3) differed significantly between V30M carriers and controls (P=0.0002). Haplotype G-C was less prevalent in the FAP group than in the controls (74.0 versus 90.0%, P=0.0003). Haplotype G-T was absent in the control group, whereas accounting for 7.0% of the chromosomes among the patients (P=0.0007). As observed for the APCS locus, RBP4 genotype distribution in V30M individuals deviated from the H–W expectations (P=0.001). The haplotype frequencies also differed (but not significantly, P=0.041) between the late- and classical-onset FAP cases (Table 3), suggesting that RBP4 alleles might have an effect on disease onset. While the genotype distributions did not associate with any particular class, they were out of H–W equilibrium for the late-onset (P=0.008), but not for the classical-onset subset (P=0.380), although both classical- and late-onset subsets differed significantly from the control group to a similar extent (P=0.0009 and 0.0001, respectively).

In addition, the genetic distance based on all 10 polymorphisms typed in the six candidate genes showed a significant differentiation between the FAP and control groups (FST=0.00691, P<0.0069) and between the classical-onset and control groups (FST=0.01258, P<0.0001). Little differentiation was found between the late-onset and control groups (FST=0.00042, P>0.34) and between the late-onset and classical-onset groups (FST=0.00048, P>0.38). It is possible that a larger number of individuals in the late-onset group would have provided sufficient power to establish the significance of the difference. The 185 subjects from the combined FAP and control group were also clustered into three groups using a minimum variance cluster analysis based on a weighted allele sharing measure. A comparison with the three phenotypes showed significant association ({chi}2=9.766, P=0.045), further confirming genetic differentiation among the three phenotypic groups.

Multiple locus interactions, age of onset and disease susceptibility
The data were further analyzed to detect possible multiple locus interactions. Analyses were performed independently in each group by assessing single- and multi-locus deviations from H–W or linkage equilibrium (24–26). The rationale for this approach is that deviations from equilibrium, especially across loci (and more importantly, when assessing unlinked loci from different chromosomes, also referred to as gametic equilibrium), may indicate that the allelic combinations at different loci interact to modulate disease expression and increase or decrease susceptibility to disease.

HSPG2, APOE, APOE–491A/T, RBPv1, APCS, SAA1 and SAA2 alleles were compared in 120 possible combinations for each phenotypic class. Interestingly, 10 multi-locus combinations deviated (P<0.05) from equilibrium in the FAP group (Table 4). In contrast, no deviations from H–W were observed at any single locus or among loci in the control group. One would expect 5% of the combinations to deviate from equilibrium, on the basis of chance alone. The results of multi-locus interaction analyses were even more compelling among the FAP onset classes. Whereas the classical-onset group deviated from H–W equilibrium in four combinations, the late-onset class deviated from the H–W expectations in 25 combinations (Table 4). Because the APCS locus was individually out of H–W in the FAP group and in the classical-onset subset, H–W analyses with APCS were conditioned on genotype frequency, so as not to bias the comparisons by the intra-locus disequilibrium.


View this table:
[in this window]
[in a new window]
  		Table 4. Significant multilocus combinations for FAP, FAP classical-onset and FAP late-onset groups
 
To exclude the possibility that sample size differences accounted for the differences in significant comparisons observed among each phenotypic group, additional analyses were performed. Subsets of 85 FAP subjects sampled without replacement from the total FAP group (to match the number of individuals in the control group) and 40 patients sampled from the classical-onset subset (to match the number of individuals in the late-onset group) were re-analyzed for linkage disequilibrium. The sampling was repeated 10 times, and the number of times the original combinations resulted in a 0.05 level of significance was scored (26).

In the FAP group, all the original significant combinations were replicated in at least 8 of 10 re-analyses (80–100%). Similarly, in the classical-onset group, the original significant results were highly replicated in the 10 re-samplings (data not shown). These results clearly indicated that the FAP and control groups and the classical- and late-onset subsets differ from each other in the extent of possible gene interactions.

From the analysis of the multi-locus combinations that were out of H–W equilibria among the FAP classes, the two-, three- and four-site comparisons accounted for most of the significant results (Table 5). All loci, except for SAA1 and SAA2, were involved in all instances. To determine the specific genotypes that may affect the age of disease onset, multifactor dimensionality reduction (MDR) analysis (27–29) was performed using four different comparisons: (1) classical- versus late-onset (2) case versus control, (3) late-onset versus controls and (4) classical-onset versus controls. In our analysis, we did not include the TTR V30M mutation to determine only the modifiers that affect age of onset. In the first comparison, there were no genotypes that showed a significant association. The sample size in each of the disease categories is probably insufficient to detect small-to-moderate effects (28). In the case versus control comparison, there were also no significant genetic predictors. On the basis of the analyses described subsequently, we hypothesize that this is due to genetic heterogeneity, such that different genotypes predispose to classical onset and others to late onset, possibly consistent with the genetic background/cluster analysis results.


View this table:
[in this window]
[in a new window]
  		Table 5. Relative contribution of multi-locus comparisons
 
In the comparisons of classical-onset with controls and late-onset with controls, despite the potential confounding effects of different sample sizes, we found genotypes that were significant predictors of FAP phenotype (Fig. 2). The classical-onset versus the controls was best predicted by APCSv1 and APCSv2 genotypes with a prediction error of 29.45% (P < 0.024). The late-onset phenotype was best predicted by RBPv1, RBPv2 and APSCv1 genotypes (prediction error 21.7%, P < 0.006). The fact that the combinations of genotypes differed by age of onset when compared with controls, suggests that the underlying genetic risk factors differ for the two classes of disease. This would also explain why there were no significant effects when the control group was compared to all cases, since genetic heterogeneity has been shown to substantially reduce the power of these analyses.



View larger version (26K):
[in this window]
[in a new window]
  		Figure 2. Genetic models for classical-onset FAP versus control groups (A) and late-onset FAP versus control groups (B). (A) The best classical-onset genetic model incorporated only APCSv1 and APCSv2. Dark shaded boxes represent high-risk genotypes. Light shaded boxes represent low-risk genotypes. Unshaded boxes are empty cells or those that do not have enough information to indicate relative level of risk. Within each cell the left-hand bar represents the number of cases of that genotype and the right-hand bar represents the number of control individuals of that genotype. Prediction error P-value (0.024). (B) The best late-onset genetic model incorporated the RBP4v1, RBP4v2 and APCSv1. Prediction error P-value (0.006).

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 
It is likely that common diseases are genetically complex involving interactions at multiple loci to produce variations in clinical presentation. Most autosomal dominant disorders also display some clinical heterogeneity even when they are the result of the same mutation. FAP type I is an autosomal dominant single gene disorder related to mutations in TTR; however, even in the context of the V30M mutation, the clinical disorder shows differences in the age of onset, nature and degree of organ involvement and pace of progression. We analyzed phenotypic heterogeneity in FAP in terms of age of onset as a discrete (classical versus late), rather than continuous variable, to eliminate potential differences generated by any imprecise determination of the time an individual became symptomatic and to make the comparison more sensitive to possible genomic differences.

Examination of a set of genes encoding molecules involved in either TTR function or amyloid deposition revealed that the TTR V30M carriers differed from the control Portuguese population. Most, if not all, Portuguese V30M mutations are found on TTR haplotype I (defined by a set of six intronic SNPs), whereas haplotype III accounts for 47.5% of the Portuguese non-carrier chromosomes (30). The candidate gene studies show significant FST differences between the classical-onset FAP and control populations. The late-onset population shows a trend toward being different from the controls and the classical-onset group, but our study did not have sufficient power to demonstrate the significance of the difference. Similarly, the cluster analysis demonstrates that the control and classical-onset cases have a considerable genetic distance from each other, whereas the late-onset group appears to differ from both. It is possible that an analysis including a larger number of late-onset individuals and/or additional SNPs that were not known at the time these studies were performed might have strengthened the argument.

At the APCS locus, two of the three polymorphisms genotyped, a non-coding substitution in the 5' regulatory region and a silent mutation in exon 2, were more common in the V30M individuals. This finding is consistent with different ages of the individual SNPs and TTR V30M in the population. Haplotypes assembled with all three APCS SNPs were distributed differently among patients and controls (P≤0.0001). The haplotypes assembled for the two genotyped RBP4 SNPs were also distributed differently among the TTR V30M carriers and controls (P≤0.0002). The most common alleles (at APCS and RBP4) in the V30M carriers were the least common in the control population, supporting the idea that the mutation-bearing individuals differ from controls at loci other than TTR.

Notwithstanding the lack of independent associations of the RBP4 intronic polymorphisms with disease onset and susceptibility, the resulting haplotypes were significantly different (when controlled for multiple comparisons) between patients and controls (P=0.0002), but not significantly so among the FAP onset classes (P=0.041). Again the differences among the V30M carriers and controls are consistent with the notion that the V30M positives are a genetic subset of the general population. The group samples lacked the power to validate the haplotype differences between classical- and late-onset cases as true associations; nonetheless, the single-locus effect of RBP in disease onset should be investigated further.

The effects of combinations of variation in the six loci under study were examined by independently analyzing each phenotypic group for multiple locus deviations from H–W equilibrium. They have distinct chromosomal locations, except for the SAA loci, which are separated from each other by 17.5 kb on 11p15.1. Therefore, any deviations from H–W across unlinked loci would support a role of multi-locus interactions in FAP. For purposes of this analysis, we accepted a P<0.05 as significant, because we were more interested in the statistical behavior of each of the groups independently than in the individual comparisons. Although the FAP group deviated significantly from H–W expectations in 10 multi-locus combinations, we did not observe any disequilibrium in the control group. The number of deviations from equilibrium as observed in the FAP class is uncommonly high relative to the amount generally found for unlinked loci in human populations (31,32). The results obtained among the FAP onset subsets strongly support this notion and demonstrate a modulatory effect of gene–gene interactions on the onset of clinical symptoms in the presence of the TTR V30M mutation.

Although some of the departures from H–W among the multi-locus interactions maintain significance when corrected for multiple comparisons, it is interesting that no departures (P < 0.05) were found in the 120 comparisons in the controls and only four were found in the classical onset group, fewer than six, the number expected by chance alone. The difference between the controls and the classical-onset group are not significant. The number of differences between the total FAP group and the controls is significant (0/120 versus 10/120, P=0.0016) with the entire difference coming from the late-onset cohort (P < 0.0001).

The classical-onset group does not differ from the controls in terms of multi-locus genotype equilibria, but the late-onset group varies considerably, perhaps by carrying somewhat protective alleles at one or more interacting loci. If V30M is a relatively ancient mutation in the Portuguese population that resembled the controls, it resulted in classical onset. It is possible that the acquisition of one or more modifying alleles at other loci resulted in later onset and survival. There is no formal evidence for greater reproductive fitness among the individuals with late onset, because classical onset is predominant in Portugal despite the fact that individuals with very early onset might be lost from the reproducing population. If this were the case, genetic anticipation might result from the loss of a protective allele or the gain of a susceptibility allele, the child becoming homozygous at the relevant locus (loci), when a late-onset V30M carrier mates with an individual bearing the wild-type TTR gene in the context of a genome in which the interactive loci carry the ‘control’, rather than the ‘late-onset’, alleles.

Although the haplotype distributions indicate that the late-onset group differs from controls and the classical-onset group does not, the discrepancy is consistent with the notion that multi-locus, rather than single locus, effects are the major factor in differentiating the age of onset phenotypes (33). The decrease in the number of significant effects when the classical- and late-onset subjects are pooled (from 25 to 10) is also in agreement with this observation, further suggesting that the pooled samples have different patterns and that there is a qualitative difference in risk for classical- and late-onset subjects. These differences are particularly striking in view of the fact that none of the examined loci was independently and unequivocally associated with the age of onset in our study. However, it should be noted that the single locus association analyses were inherently more conservative as a Bonferroni correction was used.

The MDR analysis assessed single locus models and did not find them as predictive as the multi-locus models. Consequently, the onset of the first symptoms seems to be modulated by the interaction of multiple loci that modify the main effect of the TTR locus, rather than the isolated involvement of a single gene through a single pathway. Williams et al. (26) analyzed four unlinked candidate loci, none of which was independently associated with hypertension, and showed a similar combined effect that was not seen when the distribution of microsatellite markers, unlinked to the candidate genes, was independently examined.

The potential for different underlying models for classical and late onset is supported by the MDR analysis, which produces two distinct models when comparing each class with the controls. One could view the two onset classes as unique diseases. If this is the case, then the failure to detect a single predictive genetic model is consistent with two related, but different, diseases. This is exactly what would be expected in such a case of genetic heterogeneity (28). Using this approach, a major gene effect can be viewed as a necessary, but not sufficient, condition to explain the course of the disease. Analyzing the cases but omitting from the analysis of phenotype the necessary allele, in this case TTR V30M, can then reveal a variety of important modifiers that are distinct between the phenotypes.

The significant comparisons obtained in our study cohort indicate that the combined effects mainly result from two- and three-locus interactions involving all loci except SAA1 and SAA2 for susceptibility to disease. A considerable number of four-site combinations modulate the age of onset with SAA1 appearing in a majority of significant combinations in late-onset disease, perhaps indicating a greater role of the SAA variants in the age of onset of FAP.

The correlation between genotype and phenotype in so-called simple Mendelian disorders is often incomplete, as only a subset of all mutations can reliably predict specific phenotypes (34). This is because non-allelic genetic variations and/or environmental influences underlie these disorders whose phenotypes behave as complex traits. A few examples include the identification of the role of homozygozity for the SAA1.1 allele in conferring the genetic susceptibility to renal amyloidosis in FMF (20) and the association of an insertion/deletion polymorphism in the ACE gene with disease severity in familial hypertrophic cardiomyopathy (35). In these disorders, the phenotypes arise from mutations in MEFV and ß-MHC, but are modulated by independently inherited genetic variation. In this report, we show that interactions among multiple genes, whose products are confirmed or putative constituents of ATTR deposits, or metabolically interact with TTR, modulate the onset of the first symptoms and predispose individuals to disease in the presence of the V30M mutation in TTR. The exact nature of the effects identified here requires further study with potential application in the development of genetic screening with prognostic value pertaining to the onset of disease in the TTR V30M carriers.

If the effects of additional single or interacting genes dictate the heterogeneity of phenotype, as reflected in variability of onset and clinical expression (with the same TTR mutation), the products encoded by alleles at such loci could contribute to the process of wild-type TTR deposition in elderly individuals without a mutation (senile systemic amyloidosis), a phenomenon not readily recognized as having a genetic basis because of the insensitivity of family history in the elderly.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
SUPPLEMENTARY MATERIAL
 REFERENCES
 
Subjects and DNA samples
The present study included 100 V30M carriers representative of unrelated FAP kindreds from Portugal. Only one member from any kindred was selected for analysis. The male-to-female ratio was 1 : 1 (49 men and 51 women). Ninety-two individuals were clinically affected and eight subjects (five men and three women) were asymptomatic at the time the study was initiated. All individuals referred to as carriers, whether affected or not, were determined to have the TTR V30M allele by a restriction assay described elsewhere (36). The 92 affected subjects had been diagnosed with FAP according to the established clinical criteria (2). They were chosen to represent two subsets of the affected population that varied in the age of onset, one with onset before age 40 and the other with onset after age 50. Eighty-five unrelated Portuguese individuals, who did not carry the V30M mutation, served as negative controls. Blood samples were obtained with informed consent, and DNA was extracted by standard techniques.

Age of onset
Among the 92 affected subjects, the age of onset ranged from 21 to 80 years, with an average of 41.3±15.0 (SD) years (Fig. 1). For the purposes of clarity of definition in the face of uncertainty in precisely ascertaining age at the beginning of clinical disease, they were divided into two groups on the basis of age of onset: classical (below age 40) and late (above age 50) (8,9,13). In the classical-onset subset (n=60), the mean age of onset was 31.3±4.9 years (range 21–40), whereas the late-onset subset (n=40) presented with symptoms at an average age of 60.8±7.2 years (range 51–80). The male-to-female ratio in the classical-onset subset was ~1 : 1 [29 men (mean age of onset 29.6±5.3 years, range 21–40) and 31 women (mean age of onset 33.0±3.9 years, range 27–39)]. In the late-onset cohort, the male-to-female ratio was 1 : 1.2 [18 men (mean age of onset 59.9±6.5 years, range 52–72) and 22 women (mean age of onset 61.5±7.8 years, range 51–80)]. The asymptomatic subjects were included in the latter group, because the youngest among them was already over 53 years of age at the time the study was initiated (average age: 62.9±8.7 years, ranging from 53 to 74 years). Three of the carriers were asymptomatic over 71 years of age.

Polymerase chain reaction
Polymerase chain reaction (PCR) assays were assembled in 25 µl reactions containing 10–50 ng of genomic DNA, 1.5–2.0 mM MgCl2, 0.25 mM of each dNTP, 0.2–0.5 µM of each primer and 1.25–2.5 U of AmpliTaq Gold DNA Polymerase (PE Applied Biosystems) or Biolase Taq DNA Polymerase (Doc Frugal Scientific Corp.). Amplification was performed in a MWG Biotech Primus 96 HPL thermal cycler with an initial denaturation at 96°C for 1–6 min followed by 35 cycles with denaturation at 94°C for 30 s, annealing at temperatures specified below for 30 s and extension at 72°C for 30 s, unless otherwise stated. The final extension was at 72°C for 10 min.

APCS genomic DNA sequencing
A region of 1389 bp of APCS complete coding sequence (GenBank accession no. D00097) was amplified from genomic DNA of 32 FAP individuals (14 early-onset, 16 late-onset and two asymptomatic carriers aged 72 and 74, classified as late-onset) in three overlapping segments and sequenced using the same primers employed for PCR amplification as follows: (i) 5'-GCCATCACTTGTCTCTAATAAATAACT-3' and 5'-AGGCTGTAGGCACGAGAGAG-3' (584 bp, 58°C); (ii) 5'-ACTTGATCACACCGCTGGAG-3' and 5'-CCAATCTCTCCCACAAAGGA-3' (403 bp, 58°C); and (iii) 5'-GATTCCTATGGGGGCAAGTT-3' and 5'-GTGCTCTGGCTTTGCGATA-3' (540 bp; 58°C).

PCR amplification was performed as described earlier. Following column purification with a PCR-purification kit (Qiagen), cycle sequencing of both strands was performed with dye terminator chemistry (PE Applied Biosystems), according to the manufacturer's instructions. The sequencing reactions were electrophoresed on an ABI Prism 377 DNA sequencer (PE Applied Biosystems), and the data were analyzed with sequencer 3.0 (Gene codes).

RBP4 RFLP analysis
Two overlapping segments of ~2000 bp each spanning most of RBP4 intron 4 (adjacent regions to exons 4 and 5) and 1000 bp immediately upstream exon 1 (5' UTR) were PCR amplified from genomic DNA of 10 FAP individuals with the following primers: (i) 5'-CACGGCATACGTGTCGTAGT-3' and 5'-TTCTGTTCCCAGCCTTTCTG-3' (1988 bp, 60°C); (ii) 5'-AGTGGTGGAGTTGGGATTTG-3' and 5'- ATGAAGTACTGGGGCGTAGC-3' (2024 bp, 60°C); and (iii) 5'-GCCGTATCCCACCTCACC-3' and 5'-ATTGTGCCATCACACAGGAA-3' (1055 bp, 60°C).

The amplicons were subsequently digested for ~14 h with 10 U of ApaLI, BamHI, BseRI, BsrI, DraI, HhaI and XbaI (intron 4) and MnlI (5'-UTR). All restriction enzymes used in this study were purchased from New England BioLabs, except for XbaI acquired from Gibco BRL. Following electrophoresis in 4–20% TBE polyacrylamide gels (BioRad), the digestion products were stained with ethidium bromide and visualized under ultraviolet (UV) light, rendering three different electrophoretic patterns. The emerging putative SNPs (two in intron 4 and one in the 5'-UTR, Table 1) were confirmed by PCR amplification from genomic DNA of 10 additional subjects, using the cycling conditions as described earlier and the following primers: (i) RBP4v1, 5'-GCCCACATGCTTCACCTCTA-3' and 5'-TTCTGTTCCCAGCCTTTCTG-3' (491 bp, 58°C); (ii) RBP4v2, 5'-AGGCACAGAGAAGTGAAGGAA-3' and 5'-GCAGCAAAACATCCCAAAGT-3' (532 bp, 56°C); and (iii) RBP4v3, 5'-CACTTGTGCAGGAATTTTGG-3' and 5'-TGTTTTGATTCAATTGGCTACTG-3' (421 bp, 62°C for 1 min).

SNP genotyping
HSPG2 gene.
A 241 bp fragment encompassing the G/T intronic variant was amplified using primers 5'-CATGTCCCATGCCCCACGTGTGCT-3' and 5'-ATTGTAGCTGTGGCAGGCAAACTC-3' (60°C, 40 cycles) (21). The amplicons were digested with 10 U of BamHI for 14 h in 15 µl reactions with 10 µl of template. The digested products were electrophoresed in 4–20% TBE polyacrylamide gels (Invitrogen), stained with ethidium bromide and visualized under UV light, yielding two fragments of ~150 and ~100 bp when the recognition site was present.

APOE gene.
Restriction isotyping of APOE was conducted with minor modifications (37). Briefly, following amplification with primers 5'-TCCAAGGAGCTGCAGGCGGCGCA-3' and 5'-ACAGAATTCGCCCCGGCCTGGTACACTGCCA-3' (65°C, 50 cycles, extension for 1.5 min, DMSO), the amplicons were digested with 10 U of HhaI for >2 h and analyzed by PAGE as described earlier.

APOE–491A/T gene.
A 1423 bp fragment of APOE transcriptional regulatory region was amplified using primers 5'-CAAGGTCACACAGCTGGCAAC-3' and 5'-TCCAATCGACGGCTAGCTACC-3' (68°C, extension for 40 s). A nested amplification was then performed with primers 5'-TGTTGGCCAGGCTGGTtTtAA-3' and 5'-CCTCCTTTCCTGACCCTGTCC (same conditions), in which the lower case letters correspond to mismatched nucleotides to create an artificial restriction site at position –491 when an adenosine is present. Ten microlitres of the 228 bp PCR product was digested with 10 U of DraI for ~14 h and separated by gel electrophoresis (as described earlier) in two fragments of 19 and 209 bp (22).

APCS gene.
Three of the five SNPs identified in APCS by genomic sequencing were genotyped by PCR amplification of the target sequences using primers with mismatched nucleotides to create artificial restriction sites for allele detection. The SNPs, designated APCSv1, APCSv2 and APCSv3 (Table 1), were amplified using the following primers: (i) 5'-CTTCACCGTAAGCGCgA and 5'-AGGCTGTAGGCACGAGAGAG-3' (527 bp, 55°C); (ii) 5'-GTCTGCGACAGGGTTACTTTcT-3' and 5'-CAAATCCCCAATCTCTCCCAC-3' (119 bp, 60°C); and (iii) 5'-GATTCCTATGGGGGCAAGTT-3' and 5'-GATAAAGATATGGGACTATGACgtG-3' (402 bp, 52°C), respectively, in which the lowercase letters are the mismatched nucleotides. PCR cycling was conducted as earlier, and the amplicons were subsequently digested for ~14 h with 10 U of BseRI (APCSv1, for which the A-to-G transition destroys the restriction site), XbaI (APCSv2, for which the G-to-A transition creates a restriction site). and ApaLI (APCSv3, for which the G-to-T transversion destroys the restriction site). The digested products were gel electrophoresed in 4–20% TBE polyacrylamide gels (BioRad), stained with ethidium bromide and visualized under UV light.

RBP4 gene.
Genotyping for the newly identified SNPs was performed with the primers specified in RBP4 RFLP Analysis. PCR products were digested with HhaI (RBP4v1) yielding two bands of 156 and 335 bp when the recognition site was present, with BsrI (RBP4v2) resulting in two fragments of 204 and 328 bp and MnlI (RBP4v3), in which digestion the 113 bp fragment is cut into two fragments of 52 and 62 bp, in addition to the resulting 102 and 206 bp bands.

SAA1 and SAA2 genes.
Restriction isotyping of SAA genes was performed as described elsewhere (38). In brief, primer pairs (i) 5'-GCCAATTACATCGGCTCAG-3' and 5'-TGGCCAAAGAATCTCTGGAT-3' (518 bp, 60°C for 1 min) and (ii) AGAGAATATCCAGAGACTCACAGGC-3' and 5'-CAGGCCAGCAGGTCGGAAGT-3' (115 bp, 60°C for 1 min) were used to amplify SAA1 and SAA2 polymorphic regions, respectively. Following digestion with BclI and BanI (SAA1), and NcoI (SAA2), the amplicons were separated by PAGE revealing the allelic variants as follows: 176 and 317 bp fragments (SAA1.1), 73, 176 and 225 bp (SAA1.2); 73, 176 and 244 bp (SAA1.3); 25 and 90 bp (SAA2.1); and 115 bp (SAA2.2).

Haplotype determination
APCS and RBP4 haplotypes were assigned unambiguously in the case of complete homozygozity or heterozygozity at a single site. For APCS, in complex cases with more than one heterozygous site, linkage phase was resolved by haplotype inference (39). For RBP4, the haplotypes were assembled with RBP4v1 and RBP4v2 SNPs only, in which case haplotypes for double heterozygotes were not inferred to avoid haplotyping errors. RBP4 haplotypes were used only for haplotype association and MDR analyses. For multi-locus disequilibrium comparisons, only RBPv1 was employed so as not to reduce the study sample size. RBP4v3 was neither used to integrate the RBP4 haplotypes nor it included in single-locus association analysis due to technical difficulties of genotyping.

Statistical analysis
Statistical analyses were performed with tools for population genetics analyses software (TFPGA 1.3) for single-locus allele frequency differences (40). Allelic comparisons were performed with exact test statistics for population differentiation conducted with the algorithm of Raymond and Rousset (41). Single-locus genotype frequencies were compared among groups by the use of CLUMP and Fisher exact statistics through an exact calculation of the hypergeometric distribution (Freeman–Halton) as implemented by StatXact-5 (Cytel Software, Cambridge, MA, USA) (42). Haplotype frequency differences were analyzed using the StatXact-5 and the GraphPad Instat 3.01 software package (GraphPad Software, San Diego, CA, USA). The analyses were performed using Fisher exact statistics. The respective odds ratios and confidence intervals were obtained using the approximation of Woolf. Levels of significance were corrected for multiple (14 SNPs + nine haplotypes) comparisons using the method of Bonferroni, i.e. P < 0.002 for SNP and haplotype comparisons. P-values presented in the text are always uncorrected for multiple comparisons.

Single- and multi-locus H–W equilibrium analyses were performed with genetic data analysis (GDA 1.0) software, on the basis of the work of Weir (http://hydrodictyon.eeb.uconn.edu/people/plewis/software.php/) (24,25). In the analysis of the multi-locus comparisons, we used an arbitrary significance level of P < 0.05 to reflect the statistical distribution of P-values within the control, total FAP, classical- and late-onset groups. The intent was to measure the behavior of the groups independently rather than the significance of any single finding, although one comparison did achieve significance in the late-onset group, even after correction for multiple testing (P=0.0003).

To test for genetic differentiation between the control, classical- and late-onset groups, we calculated pairwise FST values using Arlequin v. 2.0 (43). In addition, all subjects were clustered into three groups, on the basis of a weighted identical-by-state allele sharing measure using a minimum variance cluster analysis (Nievergelt, personal communication) (44). Association of the assessed clusters with the control, classical- and late-onset groups was tested with a Pearson chi-square test.

To determine the specific high- and low-risk genotypes, the MDR analysis was performed (27–29). This method produces the genetic model that most successfully predicts class or phenotype, on the basis of genotype information using cross-validation consistency and prediction error. We used the MDR analysis to determine which genotypes predict control versus late-onset and control versus classical-onset FAP. We used the Bonferroni correction for the single locus tests because significant findings (P<0.05) could occur by chance. However, for the MDR analysis, this is not an issue because using permutations to estimate statistical significance eliminates the problem of multiple comparisons by empirically deriving the P-value in each case (27).


   SUPPLEMENTARY MATERIAL
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
REFERENCES
 
Supplementary Material is available at HMG Online.


   ACKNOWLEDGEMENTS
 
We are indebted to Drs Rosário Santos and Laura Vilarinho from Instituto de Genética Médica Jacinto Magalhães, Porto, for providing the control DNA samples. We acknowledge the excellent technical support of Paul Moreira in performing DNA diagnosis of the V30M mutation, and the invaluable assistance of Isabel Friães and Laurinda Teixeira in collecting blood samples from the Portuguese carriers. We are grateful to Jim Koziol for his thoughtful comments regarding the statistical analyses and Dr Ernest Beutler for his review of the manuscript. Miguel Soares was recipient of a fellowship (PRAXIS XXI, BD/11564/97) from Fundação para a Ciência e Tecnologia, Portugal. The work was supported by NIH R01 AG19259 and funds from the Department of Molecular and Experimental Medicine (to J.B.), by Fundação para a Ciência e Tecnologia (to M.J.S.) and Comissão de Fomento da Investigação em Cuidados de Saúde, Project no. 106 (to I.C. and M.L.S.L.).


   FOOTNOTES
 
* To whom correspondence should be addressed at: Department of Molecular and Experimental Medicine, The Scripps Research Institute, 10550 North Torrey Pines Road, MEM-230, La Jolla, CA 92037, USA. Tel: +1 8587848895; Fax: +1 8587848891; Email: jbux{at}scripps.edu


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 SUPPLEMENTARY MATERIAL
 REFERENCES
 

  1. Westermark, P., Sletten, K., Johansson, B. and Cornwell, G.G. III (1990) Fibril in senile systemic amyloidosis is derived from normal transthyretin. Proc. Natl Acad. Sci. USA, 87, 2843–2845.[Abstract/Free Full Text]

  2. Andrade, C. (1952) A peculiar form of peripheral neuropathy. Familial atypical generalized amyloidosis with special involvement of the peripheral nerves. Brain, 75, 408–427.[Free Full Text]

  3. Saraiva, M.J.M., Birken, S., Costa, P.P. and Goodman, D.S. (1984) Amyloid fibril protein in familial amyloidotic polyneuropathy, Portuguese type: definition of molecular abnormality in transthyretin (prealbumin). J. Clin. Invest., 74, 104–119.[Web of Science][Medline]

  4. de Carvalho, M., Moreira, P., Evangelista, T., Ducla-Soares, J.L., Bento, M., Fernandes, R. and Saraiva, M.J. (2000) New transthyretin mutation V28M in a Portuguese kindred with amyloid polyneuropathy. Muscle Nerve, 23, 1016–1021.[CrossRef][Web of Science][Medline]

  5. Dwulet, F.E. and Benson, M.D. (1984) Primary structure of an amyloid prealbumin and its plasma precursor in a heredofamilial polyneuropathy of Swedish origin. Proc. Natl Acad. Sci. USA, 81, 694–698.[Abstract/Free Full Text]

  6. Tawara, S., Nakazato, M., Kangawa, K., Matsuo, H. and Araki, S. (1983) Identification of amyloid prealbumin variant in familial amyloidotic polyneuropathy (Japanese type). Biochem. Biophys. Res. Commun., 116, 880–888.[CrossRef][Web of Science][Medline]

  7. Buxbaum, J.N. and Tagoe, C.E. (2000) The genetics of the amyloidoses. Annu. Rev. Med., 51, 543–569.[CrossRef][Web of Science][Medline]

  8. Sousa, A., Andersson, R., Drugge, U., Holmgren, G. and Sandgren, O. (1993) Familial amyloidotic polyneuropathy in Sweden: geographical distribution, age of onset, and prevalence. Hum. Hered., 43, 288–294.[CrossRef][Web of Science][Medline]

  9. Sousa, A., Coelho, T., Lobato, L. and Sequeiros, J. (1995) Genetic epidemiology of familial amyloidotic polyneuropathy (FAP)-type I in Póvoa do Varzim and Vila do Conde (North of Portugal). Am. J. Med. Genet., 60, 512–521.[CrossRef][Web of Science][Medline]

  10. Ikeda, S., Nakazato, M., Ando, Y. and Sobue, G. (2002) Familial transthyretin-type amyloid polyneuropathy in Japan: clinical and genetic heterogeneity. Neurology, 58, 1001–1007.[Abstract/Free Full Text]

  11. Sequeiros, J. and Saraiva, M.J.M. (1987) Onset in the seventh decade and lack of symptoms in heterozygotes for the TTR Met30 mutation in hereditary amyloid neuropathy-type I (Portuguese, Andrade). Am. J. Med. Genet., 27, 345–357.[CrossRef][Web of Science][Medline]

  12. Coelho, T., Sousa, A., Lourenco, E. and Ramalheira, J. (1994) A study of 159 Portuguese patients with familial amyloidotic polyneuropathy (FAP) whose parents were both unaffected. J. Med. Genet., 31, 293–299.[Abstract/Free Full Text]

  13. Sousa, A. (1995) Phenotypic Variability in Familial Amyloidotic Polyneuropathy: A Study of Quantitative Genetics in Portugal and Sweden. PhD Thesis, University of Porto, Portugal.

  14. Soares, M., Buxbaum, J., Sirugo, G., Coelho, T., Sousa, A., Kastner, D. and Saraiva, M.J. (1999) Genetic anticipation in Portuguese kindreds with familial amyloidotic polyneuropathy is unlikely to be caused by triplet repeat expansions. Hum. Genet., 104, 480–485.[CrossRef][Web of Science][Medline]

  15. Kisilevsky, R. (2000) The relation of proteoglycans, serum amyloid P and apo E to amyloidosis current status, 2000. Amyloid, 7, 23–25.[Web of Science][Medline]

  16. Corder, E.H., Saunders, A.M., Strittmatter, W.J., Schmechel, D.E., Gaskell, P.C., Small, G.W., Roses, A.D., Haines, J.L. and Pericak-Vance, M.A. (1993) Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer's disease in late onset families. Science, 261, 921–923.[Abstract/Free Full Text]

  17. Satoh, S., Tokuda, T., Ikeda, S., Sekijima, Y., Yanagisawa, N., Hidaka, H. and Kametani, F. (1996) No association between apolipoprotein E4 allele and the age of onset in type I familial amyloid polyneuropathy. Neurosci. Lett., 204, 209–211.[CrossRef][Web of Science][Medline]

  18. Westermark, P., Pitkänen, P., Benson, L., Vahlquist, A., Olofsson, B.O. and Cornwell, G.G. III (1985) Serum prealbumin and retinol-binding protein in the prealbumin-related senile and familial forms of systemic amyloidosis. Lab. Invest., 52, 314–318.[Web of Science][Medline]

  19. White, J.T., Kelly, J.W. (2001) Support for the multigenic hypothesis of amyloidosis: the binding stoichiometry of retinol-binding protein, vitamin A, and thyroid hormone influences transthyretin amyloidogenicity in vitro. Proc. Natl Acad. Sci. USA, 98, 13019–13024.[Abstract/Free Full Text]

  20. Cazeneuve, C., Ajrapetyan, H., Papin, S., Roudot-Thoraval, F., Genevieve, D., Mndjoyan, E., Papazian, M., Sarkisian, A., Babloyan, A., Boissier, B. et al. (2000) Identification of MEFV-independent modifying genetic factors for familial Mediterranean fever. Am. J. Hum. Genet., 67, 1136–1143.[Web of Science][Medline]

  21. Hansen, P.M., Chowdhury, T., Deckert, T., Hellgren, A., Bain, S.C. and Pociot, F. (1997) Genetic variation of the heparan sulfate proteoglycan gene (perlecan gene). Association with urinary albumin excretion in IDDM patients. Diabetes, 46, 1658–1659.[Web of Science][Medline]

  22. Bullido, M.J., Artiga, M.J., Recuero, M., Sastre, I., Garcia, M.A., Aldudo, J., Lendon, C., Han, S.W., Morris, J.C., Frank, A. et al. (1998) A polymorphism in the regulatory region of APOE associated with risk for Alzheimer's dementia. Nat. Genet., 18, 69–71.[CrossRef][Web of Science][Medline]

  23. Mantzouranis, E.C., Dowton, S.B., Whitehead, A.S., Edge, M.D., Bruns, G.A. and Colten, H.R. (1985) Human serum amyloid P component. cDNA isolation, complete sequence of pre-serum amyloid P component, and localization of the gene to chromosome 1. J. Biol. Chem., 260, 7752–7756.[Abstract/Free Full Text]

  24. Lewis, P.O. and Zaykin, D. (2001) Genetic Data Analysis: Computer program for the analysis of allelic data. Version 1.0 (d16c).

  25. Weir, B.S. (1996) Genetic Data Analysis II. Sinauer Associates, Sunderland, MA.

  26. Williams, S.M., Addy, J.H., Phillips J.A., III, Dai, M., Kpodonu, J., Afful, J., Jackson, H., Joseph, K., Eason, F., Murray, M.M. et al. (2000). Combinations of variations in multiple genes are associated with hypertension. Hypertension, 36, 2–6.[Abstract/Free Full Text]

  27. Ritchie, M.D., Hahn, L.W., Roodi, N., Bailey, L.R., Dupont, W.D., Parl, F.F. and Moore, J.H. (2001) Multifactor-dimensionality reduction reveals high-order interactions among estrogen-metabolism genes in sporadic breast cancer. Am. J. Hum. Genet., 69, 138–147.[CrossRef][Web of Science][Medline]

  28. Ritchie, M.D., Hahn, L.W. and Moore, J.H. (2003) Power of multifactor dimensionality reduction for detecting gene–gene interactions in the presence of genotyping error, missing data, phenocopy, and genetic heterogeneity. Genet. Epidemiol., 24, 150–157.[CrossRef][Web of Science][Medline]

  29. Hahn, L.W., Ritchie, M.D. and Moore, J.H. (2003) Multifactor dimensionality reduction software for detecting gene–gene and gene–environment interactions. Bioinformatics, 19, 376–382.[Abstract/Free Full Text]

  30. Soares, M.L., Coelho, T., Sousa, A., Holmgren, G., Saraiva, M.J., Kastner, D.L. and Buxbaum, J.N. (2004) Haplotypes and DNA sequence variation within and surrounding the transthyretin gene: genotype—phenotype correlations in familial amyloid polyneuropathy (V30M) in Portugal and Sweden. Eur. J. Hum. Genet., 12, 225–237.[CrossRef][Web of Science][Medline]

  31. Sinnock, P. and Sing, C.F. (1972) Analysis of multilocus genetic systems in Tecumseh, Michigan. I. Definition of the data set and tests for goodness-of-fit to expectations based on gene, gamete, and single-locus phenotype frequencies. Am. J. Hum. Genet., 24, 381–392.[Web of Science][Medline]

  32. Hernandez, J.L., Elston, R.C. and Ward, L.J. (1990) Gametic equilibrium between 24 polymorphic markers. Hum. Genet., 85, 343–348.[Web of Science][Medline]

  33. Moore, J.H. and Williams, S.M. (2002) New strategies for identifying gene–gene interactions in hypertension. Ann. Med., 34, 88–95.[CrossRef][Web of Science][Medline]

  34. Dipple, K.M. and McCabe, E.R. (2000) Phenotypes of patients with ‘simple’ Mendelian disorders are complex traits: thresholds, modifiers, and systems dynamics. Am. J. Hum. Genet., 66, 1729–1735.[CrossRef][Web of Science][Medline]

  35. Tesson, F., Dufour, C., Moolman, J.C., Carrier, L., al Mahdawi, S., Chojnowska, L., Dubourg, O., Soubrier, E., Brink, P., Komajda, M. et al. (1997) The influence of the angiotensin I converting enzyme genotype in familial hypertrophic cardiomyopathy varies with the disease gene mutation. J. Mol. Cell Cardiol., 29, 831–838.[CrossRef][Web of Science][Medline]

  36. Almeida, M.R., Alves, I.L., Sakaki, Y., Costa, P.P. and Saraiva, M.J.M. (1990) Prenatal diagnosis of familial amyloidotic polyneuropathy: evidence for an early expression of the associated transthyretin methionine 30. Hum. Genet., 85, 623–626.[Web of Science][Medline]

  37. Hixson, J.E. and Vernier, D.T. (1990) Restriction isotyping of human apolipoprotein E by gene amplification and cleavage with HhaI. J. Lipid Res., 31, 545–548.[Abstract]

  38. Moriguchi, M., Terai, C., Koseki, Y., Uesato, M., Nakajima, A., Inada, S., Nishinarita, M., Uchida, S., Nakajima, A., Kim, S.Y. et al. (1999) Influence of genotypes at SAA1 and SAA2 loci on the development and the length of latent period of secondary AA-amyloidosis in patients with rheumatoid arthritis. Hum. Genet., 105, 360–366.[CrossRef][Web of Science][Medline]

  39. Clark, A.G. (1990) Inference of haplotypes from PCR-amplified samples of diploid populations. Mol. Biol. Evol., 7, 111–122.[Abstract]

  40. Miller, M.P. (1997) Tools for Population Genetics Analyses (TFPGA 1.3): A Windows Program for the Analysis of Allozyme and Molecular Genetic Data (Distributed by the author).

  41. Raymond, M. and Rousset, F. (1995) An exact test for population differentiation. Evolution, 49, 1280–1283.[CrossRef][Web of Science]

  42. Sham, P.C. and Curtis, D. (1995) Monte Carlo tests for associations between disease and alleles at highly polymorphic loci. Ann. Hum. Genet., 59, 97–105.[Web of Science][Medline]

  43. Schneider, S, Roessli, D and Excoffier, L. (2000) ARLEQUIN ver 2.000: A Software for Population Genetics Data Analysis.

  44. Lynch, M. and Ritland, K. (1999) Estimation of pairwise relatedness with molecular markers. Genetics, 152, 1753–1766.[Abstract/Free Full Text]


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 Clin Med ResHome page
T. Rudolph, M. W. Kurz, and E. Farbu
Late-Onset Familial Amyloid Polyneuropathy (FAP) Val30Met Without Family History
Clin. Med. Res., September 1, 2008; 6(2): 80 - 82.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
14/4/543    most recent
ddi051v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (15)
Right arrowRequest Permissions
Google Scholar
Right arrow Articles by Soares, M. L.
Right arrow Articles by Buxbaum, J. N.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Soares, M. L.
Right arrow Articles by Buxbaum, J. N.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?