Young Onset Type 2 Diabetes Families Are the Major Contributors to Genetic Loci in the Diabetes UK Warren.pdf

8/13/2019 Young Onset Type 2 Diabetes Families Are the Major Contributors to Genetic Loci in the Diabetes UK Warren.pdf

http://slidepdf.com/reader/full/young-onset-type-2-diabetes-families-are-the-major-contributors-to-genetic 1/7

Young-Onset Type 2 Diabetes Families Are the MajorContributors to Genetic Loci in the Diabetes UK Warren2 Genome Scan and Identify Putative Novel Loci onChromosomes 8q21, 21q22, and 22q11Timothy M. Frayling,1 Steven Wiltshire,2 Graham A. Hitman,3 Mark Walker,4 Jonathan C. Levy,5

Mike Sampson,6 Christopher J. Groves,2 Stephan Menzel,2 Mark I. McCarthy,2,5

and Andrew T. Hattersley 1

A young onset of type 2 diabetes is likely to result, inpart, from greater genetic susceptibility. Young-onsetfamilies may therefore represent a group in which genes

are more easily detectable by linkage. To test thishypothesis, we conducted age at diagnosis (AAD) strat-ified linkage analyses in the Diabetes UK Warren 2sibpairs. In the previously published unstratifiedanalysis, evidence for linkage (logarithm of odds [LOD]>1.18) was found at seven loci. The LOD scores at theseseven loci were higher in the 245 families with AAD <55years (L55) compared with the 328 families with AAD>55 years (G55). Five of these seven loci (1q24–25,5q13, 8p21–22, 8q24.2, and 10q23.2) had LOD scores>1.18 in the L55 subset but only one (8p21–22) did inthe G55 subset. Two additional loci (8q21.13 and21q22.2) showed evidence for linkage in the L55 subsetalone. Another locus (22q11) showed evidence for link-

age in a subset of families with AAD <45 years. Using alocus-counting approach, the L55 subset had signifi-cantly more loci ( P0.01) than expected under the nullhypothesis of no linkage across the LOD score range0.59–3.0. In contrast, the G55 subset contained no moresusceptibility loci than that expected by chance. Inconclusion, young-onset families provide both dispro-portionate evidence for linkage to known loci and evi-dence for additional novel loci. Our data confirmour hypothesis that families segregating young-onsettype 2 diabetes represent a more powerful resource fordefining susceptibility genes by linkage. Diabetes 52:1857–1863, 2003

For many complex diseases, genetic predisposi-

tion is stronger in subjects diagnosed at younger ages. There is evidence that this is true in type 2diabetes. Younger type 2 diabetic subjects show

increased familial clustering of diabetes even when mono-genic forms are excluded. Several studies have shown high

prevalences of parental diabetes in subjects diagnosed at young ages (1–3). Weijnen et al. (4) demonstrated that younger age at diagnosis (AAD) results in an increasedrelative risk to siblings. This is likely to reflect increasedgenetic predisposition as well as shared environmentalfactors.

If type 2 diabetic patients diagnosed at a young age havea stronger genetic predisposition, two questions need to

be answered: 1) do these patients have a different genetic predisposition? and 2) are patients diagnosed at youngages more heavily loaded for the predisposing genes thatcontribute to disease across all ages? The answers to thesequestions will be crucial in refining strategies for type 2diabetes gene discovery.

If subjects diagnosed at younger ages have a stronger genetic predisposition, it is then likely that they provide a more powerful resource for mapping susceptibility genes.The hypothesis that families segregating young-onset dia-betes are a more powerful resource for mapping genescompared with families with patients diagnosed fromacross all age ranges has not been tested using genome-

wide genotype data. In some studies, such as FUSION (5),subjects diagnosed young contribute disproportionately tothe evidence for linkage at some loci. However, these

previous studies did not present the evidence for linkagefrom young-onset families from across all the genome.

To examine the hypothesis that young-onset families area more powerful resource for mapping genes, and todefine whether the localization of predisposing genes inthese families was similar or different to that found in allsubjects, we used the Diabetes UK Warren 2 sibling pair collection. A recent genome-wide scan using 573 of thesefamilies revealed seven putative loci on chromosomes 1, 5,7, 8, and 10 (6). None of these loci reached genome-widelevels of significance when using the Lander and Kruglyak

guidelines (7), which, though widely used, assume that

From the 1Department of Diabetes and Vascular Medicine, Peninsula MedicalSchool, Exeter, U.K.; 2Wellcome Trust Centre for Human Genetics, Oxford,U.K.; the 3Department of Diabetes and Metabolic Medicine, St. Bartholomew’sand the Royal London, Queen Mary’s School of Medicine & Dentistry, London,U.K.; the 4Department of Medicine, Medical School, University of Newcastle,Newcastle, U.K.; 5Oxford Centre for Diabetes, Endocrinology and Metabolism,University of Oxford, Oxford, U.K.; and 6Elsie Bertram Diabetes Centre,Norfolk and Norwich, University Hospital, Norwich, U.K.

Address correspondence and reprint requests to Prof. Mark McCarthy,Oxford Centre for Diabetes, Endocrinology and Metabolism, Churchill Site,Old Road, Headington, Oxford OX3 7LJ, U.K. E-mail: [email protected].

Received for publication 6 January 2003 and accepted in revised form 11 April 2003.

AAD, age at diagnosis; G55, AAD 55 years; IRL, independent regions of linkage; LOD, logarithm of odds; L45, AAD 45 years; L55, AAD 55 years.

© 2003 by the American Diabetes Association.

DIABETES, VOL. 52, JULY 2003 1857



complete genetic information has been extracted from thefamilies. However, as the majority of type 2 diabetessibling pairs do not have parents available, and becauseour genome scan had a mean marker spacing of 9 cM, suchassumptions are not appropriate. Instead, genome-widesignificance levels were determined empirically from thedata. In addition, a recent study applied a complementaryapproach to assess results from the Diabetes UK Warren 2

genome scan (8). Simulations of the genome scan (usingthe same family structures, maps, and allelic frequencies)were performed under the null hypothesis of no linkageand were then compared with the observed number of locishowing evidence for linkage with the number expectedunder the null hypothesis of no linkage. The resultsshowed there were more loci with evidence for linkagethan expected by chance, even though it was not possibleto determine which loci were real and which signals hadappeared through stochastic variation.

In this study, we have examined subsets of young AADsibpair families from the Warren 2 genome scan. We havetested whether younger families contribute disproportion-

ately to the evidence for linkage at loci identified in our overall scan and whether they provide evidence for addi-tional loci.

RESEARCH DESIGN AND METHODS

Subjects. All subjects in this study are from the 573 families analyzed in the

previously reported genome-wide scan of type 2 diabetes Diabetes UK Warren

2 sibpairs. Details of the ascertainment and the inclusion and exclusion

criteria of these families are given in our previous publication (6). Briefl y,

index sibpairs were diagnosed between 35 and 70 years. Extensive efforts

were made to exclude other forms of diabetes by standard clinical criteria,

including personal and family history and an absence of anti-GAD autoanti-

bodies. All sibships were of British/Irish descent. Sibships with evidence of

bilineal inheritance and large sibships with a high proportion of affected

subjects were excluded.

Subsets. We reanalyzed our genome-wide scan on three subsets of familiesaccording to defined AAD cutoffs. These subsets were selected to have an

average AAD in affected family members of 45.0 (L45) and 55.0 (L55)

years. These represented the youngest 39 (7%) and 245 (43%) families,

respectively. The subsets were not exclusive, so all members of the L45 subset

were also in the L55 subset. In addition to the sibships, DNA was available

from two affected and one unaffected parents in the L45 subset and from six

affected and six unaffected parents in the L55 subsets. In addition, to assess

the contribution of young AAD families to the evidence for linkage, we

analyzed the complementary 328 families with an average AAD 55 years

(G55). Clinical characteristics of affected subjects from these subsets are

given in table one along with details of the total dataset for comparison. In

addition, because of its potential relevance to increased genetic susceptibility,

we calculated the proportion of subjects with a recalled maternal or paternal

history of diabetes. No subjects had a bilineal parental history of diabetes, as

this was an exclusion criterion. To minimize bias, we only included data from

probands who recalled the diabetes status of both parents. There was nodifference in the proportion of subjects recalling the diabetes status of both

parents across subsets (90% in all groups). Fisher ’s exact test and 2 test for

trend were used to compare frequencies of parental diabetes between groups.

Linkage analysis. Linkage analyses were performed using the genotype data

generated from the 573 families as previously reported (6). Briefl y, this

included 418 autosomal microsatellite markers genotyped on ABI-377 plat-

forms. These markers had an average spacing of 9 Haldane centiMorgans [cM

(H)]. To eliminate bias toward loci identified in the overall scan, we did not

include the extra markers used to fine map the region identified on chromo-

some 1q. Analysis was performed exactly as described in the analysis of the

573 sibships: a Haldane map function was used and multipoint model free

linkage statistics (the allele-sharing logarithm of odds [LOD] score of Kong

and Cox [9]) computed using ALLEGRO version 1.1b [10]). All LOD scores are

multipoint.

Significance of individual young AAD loci. To compare the evidence for

linkage obtained in each subset with that obtained from the total analysis, we

used permutation methods to derive empirical significance values for any

differences in LOD score observed. For each replicate, we selected either 39

or 245 families (equivalent to the numbers of pedigrees in the L45 and L55

subsets, respectively) at random from the 573 family dataset and determined

the LOD score. With 10,000 replicates, we were able to establish the frequency

with which this selection led to an increase in LOD score equivalent to or

greater than that detected in the observed data. The number of times the

actual LOD was reached or exceeded (within 20 cM of either side of the

observed peak) was then used to calculate an empirical P value for the effects

of subsetting. Any increase in the evidence for linkage in the young families

was not due to the extraction of extra information from these families.

Average family size, as assessed by numbers of affected subjects, and

information extraction were identical across the L45, L55, and G55 subsets

(data not shown).

Ordered subset analysis. As a complementary analysis to those based on

predetermined age cutoffs (L45 and L55), we performed ordered subset

analyses at each locus. This method can be used for any quantitative trait. It

attempts to reduce the likely genetic heterogeneity underlying type 2 diabetes

by identifying more homogeneous groups of families on the basis of diabetes-

associated intermediate traits. This method was used in analyses by the

FUSION study (5). We used mean AAD to rank all 573 families from youngest

to oldest and, in a separate analysis, oldest to youngest. The LOD score for

linkage to type 2 diabetes was then repeatedly recalculated as families were

added in rank order. The overall maximum LOD score for any subset of

families is reported. The significance of this maximized LOD score derived

from these analyses was assessed by 10,000 permutations adding in families in

random order.

Estimation of expected numbers of independent regions of linkage. To

gain a better understanding of the strength of evidence for linkage obtained

from each subset, we compared the results from the subset genome scans

with results expected under the null hypothesis of no linkage using previously

described methods (8). Briefl y, using SIMULATE (11), 10,000 replicates of the

genome were simulated under the null hypothesis of no predisposing loci,

assuming no latent genotyping error, using the same map and allele frequen-

cies applied for the 573 families and the same pattern of missing genotype

information observed in the subsets of families. Each replicate was then

analyzed using ALLEGRO and IRLs, defined as maximum LOD scores 55

cM(H) apart (40 cM[K]), counted. To assess the significance of the observed

number of loci obtained from across the range of LOD score thresholds

compared with the number of loci expected under the null hypothesis of

no linkage, we also determined from our simulations the number of indepen-

dent regions of linkage (IRLs) with a cumulative probability of 0.95 and 0.99.

These probabilities are equivalent to a one-tailed P value of 0.05 and 0.01,

respectively.The locus-counting plots in Fig. 1 graphically represent the genome-wide

number of independent regions showing evidence for linkage (i.e., indepen-

dent loci). These plots show the number of IRLs observed in an actual

genome-wide scan experiment, in relation to the number expected by chance

under the null hypothesis of no linkage at all, across the whole range of

feasible LOD score thresholds. Each point for the null IRL distribution

represents the mean count of IRLs per genome scan, estimated from 10,000

replicates of the genome under the null, that meet or exceed the given LOD

score threshold. The actual IRL distribution trace shows the analogous counts

observed in the actual genome scan experiment itself: as it is based on one

single experiment, it has a step-like appearance. Any excess of loci observed

in the actual genome scan, over that expected by chance under the null, can

be clearly visualized from these traces. Each point on the P 0.05 limit trace

represents the number of IRLs, at the given LOD threshold, that must be met

or exceeded in the actual genome scan for the excess to be signi ficant at the

P 0.05 significance limit. As only a whole number of IRLs can be observedat any LOD threshold in an actual genome-scan experiment, the P 0.05 limit

trace shows individual points instead of a continuous line.

Power. To calculate the power of the young subsets, we performed simula-

tions (1,000 replicates of an average Warren 2 chromosome; that is, a

chromosome with the mean number of markers, mean marker spacing [9

cM(H)], mean number and distribution of alleles, and mean missing genotype

proportions) at various values of s

(relative risk to sibling) as previously

described (6). As expected, the young subsets had reduced power to detect

loci of specified s values compared with the total 573 pedigrees. However, the

245 L55 families still had very good power (95%) to detect loci at s

values

of 2.5 and good power (84%) to detect loci at s

values of 2.0 (at LOD scores

of 3.0). The 39 L45 families only had good power to detect loci of larger

effect at smaller thresholds (90% at s values of 7.5 with LOD 1). Despite this

low power, we have described the results from these 39 families because it

was decided a priori to analyze them. In addition these subjects have a higher

frequency of parental diabetes compared with other subsets, which is

consistent with them having a stronger genetic predisposition.

YOUNG-ONSET DIABETES AND GENETIC LOCI

1858 DIABETES, VOL. 52, JULY 2003



RESULTS

Characteristics of young AAD subsets. Table 1 shows

characteristics of affected subjects from the L45, L55, and

G55 subsets in comparison to the overall 573 families.

Families with younger AADs had increased parental his-tory of diabetes ( P 0.0009 for trend across L45, L55, and

G55 groups).

Linkage results. Table 2 gives details of the LOD scores

in the L45, L55, and G55 subsets for the seven loci detected

at LOD 1.18 in the overall genome scan of 573 families.

This LOD score threshold is used for convenience and

consistency with the previous study (6). LOD scores are

the maximum observed within 20 cM of either side of the

peak identified in the genome scan of 573 families; hence,

the LOD scores for the total dataset are not the exact sum

of those from the L55 and G55 groups.

For all seven loci highlighted in the genome scan of 573

families, the LOD scores in the L55 group were higher than

the LOD scores from the G55 families. Five of the seven

loci (1q24 –25, 5q13, 8p21–22, 8q24.2, and 10q23.2) occurred

at LOD scores 1.18 in the L55 families, but only one

(8p21–22) occurred over this nominal threshold in the G55

families. Four of the seven loci exceeded the empiricallydetermined LOD score of 1.56 for declaring suggestivelinkage— by chance, the LOD score threshold at whichonly one locus is expected to occur in a genome-wide scanof the Warren 2 data. This threshold is lower than thatindicated by Lander and Kruglyak’s guidelines (LOD 2.2)because of the incomplete inheritance information that canbe extracted from the Warren 2 data (8). The increase inLOD score in the L55 subset compared with the G55 subsetwas attenuated slightly for the 1q locus after the addition of extra markers (LOD 2.0 in L55, 0.78 in G55, and 1.98 in totalfamilies [6]). Empirically determined LOD score thresholdsfor suggestive linkage in the L45, L55, and G55 subsets are1.60, 1.56, and 1.56, respectively.

Table 3 gives details of three additional loci with someevidence for linkage detected in the young AAD families.These occurred on chromosomes 8q21.13 and 21q22 in theL55 subset and chromosome 22q11 in the L45 subset.Significance of individual loci: permutation to assess

significance of increases in LOD score. Table 4 showsthat families segregating young-onset diabetes contributedisproportionately to the evidence for linkage at some loci.Younger AAD families give significantly increased evidencefor linkage to the loci on chromosomes 8q24.2 ( P 0.01 for increase in LOD score), 10q23.2 ( P 0.02), 21q22 ( P

0.03), 5q32 ( P 0.006), and 22q11 ( P 0.003) than wouldbe expected if the same number of families had beenselected at random from the 573 dataset (Table 4).Significance of individual loci: ordered subset analy-ses. Ordered subset analyses were consistent with a subsetof the youngest AAD families contributing disproportion-ately to the evidence for linkage at loci on chromosomes

5q32 (empirical P 0.001) and 22q11 (empirical P 0.04)

FIG. 1. Actual and expected IRLs in 39 families with average AAD <45

years ( A), 245 families with average AAD <55 years ( B), and 328families with average AAD >55 years (C).

T.M. FRAYLING AND ASSOCIATES




than expected if families had been added into the analyses

in random order. No significant (at P 0.05) results were

obtained for other loci.

Performing ordered subset analyses on the loci identi-

fied by adding in families from old to young did not reveal

any significant results (at P 0.05; data not shown).

Numbers of observed and expected loci. We next used

the locus-counting approach to compare the number of

observed loci in the age-stratified subsets with the number

of loci that would be expected under the null hypothesis of

no linkage, across all LOD score thresholds 0.59. This

complements the analyses in which a nominal LOD score

threshold is used to highlight regions of linkage. Results

are shown in Fig. 1 A–C . The 245 L55 families produced

significantly more loci than expected under the null hypothesis of no linkage across most of the LOD score

thresholds from 0.59 –3.0 ( P 0.01 over much of the range

of LOD scores) (Fig. 1 B). In the 39 L45 families, the

number of observed loci exceeded only the number ex-

pected for LOD scores 2.0 (Fig. 1 A). In contrast to the

results in the younger-onset families, the G55 subset did

not show any more regions of linkage than expected by

chance (Fig. 1C ).

DISCUSSION

Our results show that in the Diabetes UK Warren 2genome-wide scan for type 2 diabetes, the evidence for linkage comes disproportionately from families diagnosedbefore 55 years of age. This indicates that the power of linkage analyses is likely to be higher in such families. It isconsistent with the hypothesis that genetic susceptibilityis enhanced in young-onset families.

For all seven loci with evidence for linkage in the overallgenome-wide scan of 573 families, the LOD score washigher in the L55 subset than the G55 subset. Of theseseven loci, fi ve were detectable (at LODs 1.18, nominal P 0.01) in the L55 subset, namely 1q24–25, 5q13,8p21–22, 8q24.2, and 10q23.2. In contrast, only one locushighlighted in the overall scan (8p21–22) was detected in

the G55 subset. In addition, our results provide evidencefor novel loci on chromosomes 8q21, 21q22, and 22q11.

For the loci identified in the L55 subset, on chromo-somes 8q24.2, 10q23.2, and 21q22, young families providesignificantly more evidence for linkage than would beexpected if the contribution to these loci were proportion-ate across all 573 families.

In the L45 subset, two loci (5q32 and 22q11) wereidentified above the nominal cutoff of LOD 1.18. One of these loci (5q32) occurs in the overall genome-wide scanof 573 families. For both these loci, both permutationanalyses and ordered subset analyses supported a role for a small number of the youngest AAD families contributingdisproportionately to the evidence for linkage.

In our study, families segregating young-onset type 2

TABLE 1Details of affected siblings from all 573 families and subsets

L45 L55 G55 Total

Families 39 245 328 573Subjects (F/M) 41/42 235/280 322/369 557/649 Age at examination (years) 53.3 (9.3) 59 (8.2) 68.1 (6.6) 64.2 (8.6) Age at diagnosis (years) 42.1 (4.7) 49 (6.9) 60.6 (6.1) 55.6 (8.6)

BMI (kg/m

2

) 29.9 (5.4) 29.8 (5.2) 27.9 (4.7) 28.7 (5)Frequency of parental history of diabetes* 0.64 0.59 0.42 0.49Frequency parental diabetes (mother/father)† 0.34/0.29 0.44/0.15 0.28/0.14 0.35/0.14 Affected sibpairs

All possible 52 317 426 743Independent 45 278 372 650

Data are n and means (SD) unless otherwise indicated. Age at examination, age at diagnosis, and BMI all differ ( P 0.001) between subsetsand total group and two young subsets and 55 subset *Test for trend across three subsets, P 0.0009. L45 have increased paternal diabetesand L55 have increased maternal diabetes compared with both G55 and total groups (all P 0.05). †Sibships with both parents affected wereexcluded from the initial collection

TABLE 2Multipoint LODs for young AAD subsets for loci where LOD

1.18 in complete set of 573 sibships

Locus (marker) Position*

Subset (number of families)†

L45(39)

L55(245)

G55(328)

Total(573)

1q24-25 (D1S218) 209.8 0.06 2.38 0.50 1.505q13 (D5S647) 82.1 0.18 1.28 0.57 1.225q32 (D5S436) 160.8 2.13 0.70 0.54 1.227p15.3 (D7S493) 38.0 1.17 0.77 0.56 1.318p21-22 (D8S258) 42.2 0.13 1.68 1.43 2.558q24.2 (D8S272) 162.6 0.03 2.48 0.08 1.4110q23.2 (D10S1765) 125.2 0.56 3.07 0.12 1.99

*Position of peak LOD in centiMorgans in 573 families. †Empiricallydetermined LOD score thresholds for suggestive linkage in L45, L55,

and G55 subsets are 1.60, 1.56, and 1.56, respectively.

TABLE 3Loci detected at multipoint LOD 1.18 specific to young-onsetfamilies

Locus (marker) Position*

Subset (number of families)

Total(573)

L45(39)

L55(245)

G55(328)

8q21.13 (D8S275) 106.6 0.36 1.44 0.00 0.1421q22 (D21S266) 54.2 0.13 1.67 0.03 0.4422q11 (D22S420) 0 2.51 0.47 0.08 0.34

*Position of peak LOD in centiMorgans.





diabetes provide more evidence for linkage than expectedunder the null hypothesis of no linkage, even though noloci individually reach genome-wide levels of significance.For example, in the L55 subset, we saw significantly ( P

0.01) more loci across the whole LOD score thresholdrange of 0.59 –3.0 than expected by chance. For the L45subset, only loci at the higher end of the LOD score range(LOD 2.0) were observed in significant excess. In con-trast, the G55 subset yielded no more evidence for linkagethan that expected by chance under the null hypothesis of no genetic predisposition. This suggests families withsubjects diagnosed at older ages have reduced power todetect linkage, which raises the question as to whether

these families have any genetic component to their diabe-tes or whether their etiology is entirely environmental. Wewould argue that there is at least some genetic componentin older families but that the genes involved may have toosmall an effect to be detected by linkage. It is well knownthat the detection of linkage is very sensitive to the relativerisk associated with the locus and that most genome scansoperate close to the limits of detection for loci of modesteffect. For example our genome scan of 573 families had76 and 55% power to detect loci of values of 1.37 and1.28, respectively (6). This is not suf ficiently powerful todetect the predisposing variants described to date, such asthe E23K variant in the KCJN11 gene (Kir6.2) that predis-

poses to diabetes with an odds ratio of 1.23 (1.12–1.36)(12). In our large studies, the frequency of the predispos-ing K allele was very similar (40%) across all type 2diabetes cases, regardless of AAD (10).

Results from previous genome scans of type 2 diabetesindicate that there is no one major predisposing locus intype 2 diabetes (6,13–20). Our results confirm that this isalso the case in subjects affected at young ages: it is not

possible to conclude which linkage signals are real sus-ceptibility loci and which occur by chance. To gain a better understanding of which loci are likely to be real, wecompared our results to those of other genome scans. The

presence of replication across populations adds weight tothe body of evidence for that locus. Of the loci in our

study, those on chromosomes 1q24 –25 and 8p21–22 have

been observed in several other studies and noted before

(6,17,19). In addition, a locus on chromosome 5q35 has

been identified with a heterogeneity-LOD score of 1.37 in a study of 14 maturity-onset diabetes of the young (MODY)families (21) and with a nonparametric LOD score of 2.1 in26 early-onset (2 members diagnosed 45 years) Scan-dinavian families. There were, however, three families inboth the MODY and Scandinavian studies. These regionsare close enough to our peak that they could represent thesame gene. The locus recently identified on 5q in a largeFrench family is likely to be too proximal to represent thesame gene (22).

Other genome-wide scans of type 2 diabetes familieshave included analyses by young AAD. Unlike our study,however, none has shown that most of the evidence for linkage across the genome comes from the younger-onsetsubjects. Ghosh et al. (5) performed ordered subsetanalyses on regions identified in a total genome scan of 719Finnish sibpair-based type 2 diabetic families (index casediagnosed 35– 60 years). This showed that the 74 youngest

AAD families provided increased evidence for linkage to a region on chromosome 6 (nominal P 0.041 at 104.5 cM).

A recent study of 472 Ashkenazi Jewish type 2 diabeticsibpairs also used ordered subset analysis to dissect iden-tified loci but did not find any evidence for heterogeneity(20).

Other studies have focused exclusively on younger subjects. These include genome-wide scans in French (19),Utah Caucasian (17), and Scandinavian (23) type 2 diabeticfamilies. Results from the well-replicated locus on 1q21–24are particularly noteworthy. Subjects from the French andUtah Caucasian families linked to 1q were diagnosed onaverage at the relatively young ages of 49.5 and 50.6 years,respectively. In addition, in the Pima Indian genome scan(15), the evidence for linkage at this locus is increased in a subset of 55 sibpairs diagnosed 25 years.

It is not known whether subsetting by AAD and reassess-

ing the evidence for linkage across the whole genome in

other populations will produce similar results to ours. Such

analyses will depend on the initial ascertainment criteria

TABLE 4Multipoint LODs for all peaks with significance of increases compared to total families and ordered subset analyses

LocusPosition

(cM from pter)‡

Peak LOD in L45 or L55 subsets OSA (young-old)

LOD in young families ( P )* No. families LOD ( P )†No. families

(AAD cut-off)

1q24–25 (D1S218) 214.2 2.38 (0.09) 245 2.68 (0.14) 376 (59)

5q13 (D5S418) 64.1 1.28 (0.43) 245 1.63 (0.52) 257 (56)5q32 (D5S410) 170.3 2.13 (0.006) 39 3.67 (0.001) 23 (43)7p15.3 (D7S493) 38.0 0.77 (NA) 245 2.55 (0.07) 5 (39)8p12–22 (D8S258) 44.3 1.68 (NA) 245 2.46 (0.78) 572 (72)8q21.13 (D8S275) 106.6 1.44 (0.38) 245 1.55 (0.41) 257 (56)8q24.2 (D8S272) 166.9 2.48 (0.01) 245 3.00 (0.08) 284 (56)10q23.2 (D10S1765) 125.2 3.07 (0.02) 245 3.19 (0.10) 315 (57.5)21q22 (D21S266) 54.2 1.67 (0.03) 245 1.76 (0.10)§ 245 (55)22q11 (D22S420) 0 2.51 (0.003) 39 1.89 (0.04)§ 39 (45)

* P for increase in LOD in young subset compared with 573 families; † P for maximized LOD score in analysis of young to old families; ‡ positionof peak LOD in young subset; §OSA uses linear function of KAC Maximum LOD score statistic and all other analyses are performed usingexponential function so differences in LODs occur at chrom 21 and 22 despite same number of families in analyses. NA, not applicable as

young subset LOD not higher than peak in 573 families.





for the collection of families, ethnicity, and data on whencollections and diagnoses were made. It is likely that themore affected family members ascertained, the lower the

AAD compared with a general type 2 diabetic population.Therefore, if the initial collection focused on younger subjects or larger families, there may not be much value infurther stratification by young AAD. Different age cutoffsmay be required in non-Caucasian populations as the AAD

is generally lower compared with Caucasians. In addition,the prevalence of type 2 diabetes is increasing in young

people due to changes in obesity and physical activity,which may mean even younger ages at diagnosis areneeded to pull out the families who are most powerful for mapping genetic risk factors by linkage. Finally, it is

possible that AAD may be a surrogate for other traits, suchas insulin resistance or BMI, that are more apt to producethe optimum subset for detecting genes by linkage. De-spite these caveats, we would argue that analyses by AADare needed. Our results and the consistent finding that

young-onset type 2 diabetic subjects have an increasedfamily history provide a good a priori hypothesis that

younger subjects will provide the strongest evidence for linkage. This could both increase the evidence for locialready identified and provide evidence for novel locispecific to younger-onset subjects

In conclusion, families segregating young-onset type 2diabetes provide disproportionate evidence for linkage inour U.K. Caucasian genome-wide scan. Younger subjects

provide both a disproportionate amount of evidence for linkage to previously identified loci and evidence for additional loci. This suggests that, at least in Caucasian

populations, efforts to define type 2 diabetes genes bylinkage may be more powerful if focused on young AADsubjects.

ACKNOWLEDGMENTS

We thank William Duren (Department of Biostatistics,School of Public Health, University of Michigan, Ann

Arbor, MI) for the ordered subset software. T.M.F. is a career scientist of the South and West National HealthService Research Directorate. A.T.H. is a Wellcome TrustResearch Leave Fellow. Diabetes U.K. and the WellcomeTrust kindly provided financial support for the genomescan.

We thank all of the patients, nurses, and doctors whocontributed to the Diabetes UK Warren 2 collection.

REFERENCES

1. O’Rahilly S, Spivey RS, Holman RR, Nugent Z, Clark A, Turner RC: Type II

diabetes of early onset: a distinct clinical and genetic syndrome? BMJ

294:923–928, 1987

2. Owen K, Ayres S, Corbett S, Hattersley A: Increased risk of diabetes in

first-degree relatives of young-onset type 2 diabetic patients compared

with relatives of those diagnosed later. Diabetes Care 25:636 – 637, 2002

3. Frayling T, Walker M, McCarthy M, Evans J, Allen L, Lynn S, Ayres S,

Millauer B, Turner C, Turner R, Sampson M, Hitman G, Ellard S, Hattersley

A: Parent-offspring trios: a resource to facilitate the identification of type

2 diabetes genes. Diabetes 48:2475–2479, 1999

4. Weijnen CF, Rich SS, Meigs JB, Krolewski AS, Warram JH: Risk of diabetes

in siblings of index cases with type 2 diabetes: implications for genetic

studies. Diabet Med 19:41–50, 2002

5. Ghosh S, Watanabe RM, Valle TT, Hauser ER, Magnuson VL, Langefeld CD,

Ally DS, Mohlke KL, Silander K, Kohtamaki K, Chines P, Balow J J,

Birznieks G, Chang J, Eldridge W, Erdos MR, Karanjawala ZE, Knapp JI,

Kudelko K, Martin C, Morales-Mena A, Musick A, Musick T, Pfahl C, Porter

R, Rayman JB: The Finland-United States investigation of non-insulin-

dependent diabetes mellitus genetics (FUSION) study. I. An autosomal

genome scan for genes that predispose to type 2 diabetes. Am J Hum

Genet 67:1174 –1185, 2000

6. Wiltshire S, Hattersley AT, Hitman GA, Walker M, Levy JC, Sampson M,

O’Rahilly S, Frayling TM, Bell JI, Lathrop GM, Bennett A, Dhillon R,

Fletcher C, Groves CJ, Jones E, Prestwich P, Simecek N, Rao PV, Wishart

M, Foxon R, Bottazzo GF, Howell S, Smedley D, Cardon LR, Menzel S,

McCarthy MI: A genomewide scan for loci predisposing to type 2 diabetes

in a U.K. population (the Diabetes UK Warren 2 Repository): analysis of

573 pedigrees provides independent replication of a susceptibility locus on

chromosome 1q. Am J Hum Genet 69:553–569, 2001

7. Lander E, Kruglyak L: Genetic dissection of complex traits: guide-lines for

interpreting and reporting linkage results. Nat Genet 11:241–247, 1995

8. Wiltshire S, Cardon LR, McCarthy MI: Evaluating the results of genome-

wide linkage scans of complex traits by locus counting. Am J Hum Genet

71:1175–1182, 2002

9. Kong A, Cox NJ: Allele-sharing models: LOD scores and accurate linkage

tests. Am J Hum Genet 61:1179 –1188, 1997

10. Gudbjartsson DF, Jonasson K, Frigge ML, Kong A: Allegro, a new computer

program for multipoint linkage analysis. Nat Genet 25:12–13, 2000

11. Terwilliger JD, Speer M, Ott J: Chromosome-based method for rapid

computer simulation in human genetic linkage analysis. Genet Epidemiol

10:217–224, 1993

12. Gloyn AL, Weedon MN, Owen KR, Turner MJ, Knight BA, Hitman G,

Walker M, Levy JC, Sampson M, Halford S, McCarthy MI, Hattersley AT,

Frayling TM: Large-scale association studies of variants in genes encoding

the pancreatic -cell K ATP

channel subunits Kir6.2 ( KCNJ11) and SUR1

( ABCC8) confirm that the KCNJ11 E23K variant is associated with type 2

diabetes (Brief Genetics Report). Diabetes 52:568 –572, 2003

13. Mahtani M, Widen E, Lehto M, Thomas J, McCarthy M, Brayer J, Bryant B,

Chan G, Daly M, Forsblom C, Kanninen T, Kirby A, Kruglyak L, Munnelly

K, Parkkonen M, Reeve-Daly M, Weaver A, Brettin T, Duyk G, Lander E,

Groop L: Mapping of a gene for type 2 diabetes associated with an insulin

secretion defect by a genome scan in Finnish families. Nat Genet 14:90 –94,

1996

14. Hanis CL, Boerwinkle E, Chakraborty R, Ellsworth DL, Concannon P,

Stirling B, Morrison VA, Wapelhorst B, Speilman PS, Gogolin-Ewens KJ,

Shephard JM, Williams SR, Risch N, Hinds D, Iwasaki N, Ogata M, Omori

Y, Petzold C, Rietzsch H, Shcrofer HE, Schulze J, Cox NJ, Menzel S, Boriraj

VV, Chen X, Lim LR, Lindner T, Mereu LE, Wang YQ, Xiang K, Yamagata K,

Yang Y, Bell GI: A genome wide search for human non-insulin dependent

(type 2) diabetes genes reveals a major susceptibility locus on chromo-

some 2. Nat Genet 13:161–166, 1996

15. Hanson RL, Ehm MG, Pettit DJ, Prochazka M, Thompson DB, Timberlake

D, Foroud T, Kobes S, Baier L, Burns DK, Almasy L, Blangero J, Garvey

WT, Bennett PH, Knowler WC: An autosomal genomic scan for loci linked

to type II diabetes mellitus and body-mass index in Pima Indians. Am J

Hum Genet 63:1130 –1138, 1998

16. Ghosh S, Watanabe RM, Hauser ER, Valle T, Magnuson VL, Erdos MR,

Langefeld CD, Balow J, Jr., Ally DS, Kohtamaki K, Chines P, Birznieks G,

Kaleta HS, Musick A, Te C, Tannenbaum J, Eldridge W, Shapiro S, Martin

C, Witt A, So A, Chang J, Shurtleff B, Porter R, Boehnke M: Type 2 diabetes:

evidence for linkage on chromosome 20 in 716 Finnish affected sib pairs.

Proc Natl Acad Sci U S A 96:2198 –2203, 1999

17. Elbein SC, Hoffman MD, Teng K, Leppert MF, Hasstedt SJ: A genome-wide

search for type II diabetes susceptibility genes in Utah Caucasians.

Diabetes 48:1175–1182, 1999

18. Ehm M, Karnoub M, Sakul H, Gottschalk K, Holt D, Weber J, Vaske D,

Briley D, Briley L, Kopf J, McMillen P, Nguyen Q, Reisman M, Lai E, Joslyn

G, Shepherd N, Bell G, Wagner M, Burns D: Genomewide search for type

2 diabetes susceptibility genes. Am J Hum Genet 66:1871–1881, 2000

19. Vionnet N, Hani EH, Dupont S, Gallina S, Francke S, Dotte S, Matos FD,

Durand E, Lepretre F, Lecoeur C, Gallina P, Zekiri L, Dina C, Froguel P:

Genomewide search for type 2 diabetes-susceptibility genes in French

whites: evidence for a novel susceptibility locus for early-onset diabetes on

chromosome 3q27-qter and independent replication of a type 2-diabetes

locus on chromosome 1q21– q24. Am J Hum Genet 67:1470 –1480, 2000

20. Permutt MA, Wasson JC, Suarez BK, Lin J, Thomas J, Meyer J, Lewitzky S,

Rennich JS, Parker A, DuPrat L, Maruti S, Chayen S, Glaser B: A genome

scan for type 2 diabetes susceptibility loci in a genetically isolated

population. Diabetes 50:681– 685, 2001

21. Frayling TM, Lindgren CM, Chevre JC, Menzel S, Wishart M, Benmezroua





Y, Brown A, Evans JC, Subba Rao P, Dina C, Lecoeur C, Kanninen T,

Almgren P, Bulman MP, Wang Y, Mills J, Wright-Pascoe R, Mahtani MM,

Prisco F, Costa A, Cognet I, Hansen T, Pedersen O, Ellard S, Tuomi T,

Groop LC, Froguel P, Hattersley AT, Vaxillaire M: A genome-wide scan in

families with maturity-onset diabetes of the young: evidence for further

genetic heterogeneity. Diabetes 52:872– 881, 2003

22. Martin LJ, Comuzzie AG, Dupont S, Vionnet N, Dina C, Gallina S, Houari M,

Blangero J, Froguel P: A quantitative trait locus influencing type 2 diabetes

susceptibility maps to a region on 5q in an extended French family.

Diabetes 51:3568 –3572, 2002

23. Lindgren CM, Widen E, Tuomi T, Li H, Almgren P, Kanninen T, Melander

O, Weng J, Lehto M, Groop LC: Contribution of known and unknown

susceptibility genes to early-onset diabetes in Scandinavia: evidence for

heterogeneity. Diabetes 51:1609 –1617, 2002



Documents

Young Onset Type 2 Diabetes Families Are the Major Contributors to Genetic Loci in the Diabetes UK Warren.pdf