mining the genome for susceptibility to diabetic nephropathy...
TRANSCRIPT
IdMtrt
*
†
‡
S
A
0©
2
Mining the Genome for Susceptibilityto Diabetic Nephropathy: The Role of
Large-Scale Studies and Consortia
Sudha K. Iyengar, PhD,* Barry I. Freedman, MD,† and John R. Sedor, MD‡
Summary: Approximately 30% of individuals with type 1 and type 2 diabetes developpersistent albuminuria, lose renal function, and are at increased risk for cardiovascular andother microvascular complications. Diabetes and kidney diseases rank within the top 10causes of death in Westernized countries and cause significant morbidity. Given theseobservations, genetic, genomic, and proteomic investigations have been initiated to betterdefine basic mechanisms for disease initiation and progression, to identify individuals at riskfor diabetic complications, and to develop more efficacious therapies. In this review we havefocused on linkage analyses of candidate genes or chromosomal regions, or coarse genome-wide scans, which have mapped either categorical (chronic kidney disease or end-stage renaldisease) or quantitative kidney traits (albuminuria/proteinuria or glomerular filtration rate).Most loci identified to date have not been replicated, however, several linked chromosomalregions are concordant between independent samples, suggesting the presence of a diabeticnephropathy gene. Two genes, carnosinase (CNDP1) on 18q, and engulfment and cellmotility 1 (ELMO1) on 7p14, have been identified as diabetic nephropathy susceptibilitygenes, but these results require authentication. The availability of patient data sets with largesample sizes, improvements in informatics, genotyping technology, and statistical methodol-ogies should accelerate the discovery of valid diabetic nephropathy susceptibility genes.Semin Nephrol 27:208-222 © 2007 Elsevier Inc. All rights reserved.Keywords: Genetics, albuminuria, linkage, association, candidate genes
mdporawtundtgwasTsam
n the past 15 years, studies with differentanalytic strategies have established that dia-betic nephropathy (DN) has a genetic pre-
isposition. DN does not segregate according toendelian rules, but rather is a multifactorial
rait that results from interplay between envi-onmental factors and multiple genes.1-8 Givenhis, strategies that were used successfully to
Departments of Epidemiology and Biostatistics, Ophthalmology and Genetics,Case Western Reserve University, Cleveland, OH.
Department of Internal Medicine/Nephrology, Wake Forest University Schoolof Medicine, Winston-Salem, NC.
Departments of Medicine and Physiology and Biophysics, Case School ofMedicine and Kidney Disease Research Center, Rammelkamp Center forResearch and Education, MetroHealth System Campus, Cleveland, OH.
upported by research grants U01DK57292, U01DK57293, UO1DK57298,RO1DK070942, and R01DK064719 from the National Institute of Diabetesand Digestive and Kidney Diseases and the Kidney Foundation of Ohio.
ddress reprint requests to Dr. Sudha K. Iyengar, Case Western ReserveUniversity, Wolstein Research Building, 1315, 2103 Cornell Rd, Cleveland,OH 44106-7281. E-mail: [email protected]
270-9295/07/$ - see front matter
s2007 Elsevier Inc. All rights reserved. doi:10.1016/j.semnephrol.2007.01.004Seminars08
ap the causal mutations for monogenic disor-ers could not be used for DN and other com-lex traits and new analytic plans were devel-ped. Initially modest-sized study collectionsanging from 100 to 200 nuclear families withffected siblings were assembled and analyzedith model-free methods in the hopes of iden-
ifying genes for DN.9-11 This was a time ofncertainty when the genome sequencing wasot finished, and it was unclear which studyesigns and sample sizes would best facilitatehe identification of genes for DN. Sample sizesenerally were based on the number of subjectsho could be enrolled at a single center within2- to 5-year time line, extrapolating successful
trategies for mapping single gene disorders.hus, these sample sizes would comprise abouteveral hundred meiotic products (individuals)nd allow a map resolution of approximately 1egabase (1 megabase � 1 centiMorgan), as-
uming a single locus genome-wide. However,
in Nephrology, Vol 27, No 2, March 2007, pp 208-222
ntt
ivdttrwmngwbcmnwbghusp
gduseeaeoTtnt
cmmeibctH
truldTlsmecfitCphan
eDfptl(nsmutbrfcwiergmcospsrpt
Diabetic nephropathy genes—genome-wide tools 209
either the sample sizes nor the map resolu-ions were adequate for identifying DN suscep-ibility variants.
Subsequently, seminal studies in genetic ep-demiology literature have suggested that indi-iduals with long-standing diabetes who wereiscordant for nephropathy were needed to dis-inguish between genes responsible for diabe-es and those regulating nephropathy.12 Theationale for this strategy was that individualsithout diabetes did not have the hyperglyce-ic exposure necessary to develop chronic kid-
ey disease (CKD), even if they did possessenes for DN susceptibility. Clearly, individualsho did not develop nephropathy after a num-er of years of exposure to chronic hypergly-emia possessed inherent (potentially genetic)echanisms of protection; contrasting the ge-
etic profile of diabetic individuals with andithout DN would distil out the common dia-etes genes, allowing identification of DNenes. The mechanisms by which persistentyperglycemia sensitizes the kidney in individ-als predisposed to develop DN is not under-tood completely, although tight control of hy-erglycemia certainly offers renoprotection.13
Concomitant with evolving approaches toene mapping, outpatient screens for kidneyisease (phenotypes) were validated. Albumin-ria or proteinuria could be quantitated on apot sample and identified individuals witharly CKD initiation and progression.14 Simplestimating equations, based on the serum cre-tinine level and demographic data, allowed anstimation of the glomerular filtration rate with-ut the need to collect urine over 24 hours.15-17
hus, the renal community was well positionedo implement gene mapping studies with largeumbers of enrolled patients with easily quan-ifiable kidney disease traits.
The best prospect for the successful identifi-ation of CKD susceptibility alleles would beultiple families with diabetes with multipleembers with CKD and/or end-stage renal dis-
ase (ESRD). Considering the genetic load, fam-lies with more than 2 individuals on dialysisecause of DN were ideal because even for aomplex trait this would imply that nephropa-hy genes were segregating within the family.
owever, these types of families were difficult co locate because of early mortality from mac-ovascular disease. Albuminuria (microalbumin-ria) may be a better predictor for cardiovascu-
ar disease than for ESRD,18,19 explaining theifficulty in identifying multiplex DN families.o relieve some of the onus of collecting fami-
ies with 2 or more DN-ESRD sibling pairs, thetudy designs were expanded further to accom-odate family members with less advanced dis-
ase. Quantitative measures of albuminuria andalculated glomerular filtration rate (glomerularltration rate calculated based on the Modifica-ion of Diet in Renal Disease equation or theockcroft-Gault equation) were used as endo-henotypes. Both these traits were shown to beeritable in at least some populations,20,21 andre routine measures of cardiovascular and re-al function in clinical settings.
The impetus to map genes for nephropathy,specially DN, was driven by the modeling ofN inheritance in several multigenerational,
amily data sets or by looking at concordance inhenotypes between family members. Segrega-ion and comingling analyses are the only estab-ished methods for determining if a phenotypeclinical measure or trait) fits a particular ge-etic model. Two segregation analyses haveuggested that a major locus controls the albu-in excretion rate.20,22 In one study, protein-
ria was analyzed as a continuous variable, withhe conclusion that proteinuria was influencedy multiple genes with variable effects. Theeport by Imperatore et al22 in diabetic Pimaamilies considered overt proteinuria as a dis-rete variable, and determined that this traitas regulated by a major gene effect. Two stud-
es of type 2 DN partitioned the genetic andnvironmental influences in albumin excretionate and estimated heritability, a measure ofenetic predisposition.23,24 Both studies esti-ated the heritability for urinary albumin ex-
retion to be approximately 30%. The estimatesf heritability for urine albumin excretion weretatistically significant, even after adjusting forotential confounding covariables such as age,ex, body weight, diabetes duration, and envi-onment, suggesting a major genetic effect forroteinuria.23 Finally, a renal biopsy study inype 1 diabetic siblings showed high degrees of
orrelation between severity and patterns ofgcl
nbittqsllkptinyecost
tpcasmidsbep2tmctaascpsfie
tlafiodestatc
GM
Gli1ttcbewqtsrdstmeltgortcndttsdts
210 S.K. Iyengar, B.I. Freedman, and J.R. Sedor
lomerular injury, despite the frequent lack ofoncordance of glycemic control between sib-ings.25
In constructing the studies for detecting ge-etic susceptibility to DN, consideration had toe given to several important issues. First, is DN
n type 1 diabetes a different entity from that inype 2 diabetes, such that genes for type 1 andype 2 DN differ? This remains an unanswereduestion, but one that hopefully will be re-olved by comparing the results from severalarge-scale gene mapping studies discussedater. However, current evidence suggests thatidney disease in type 1 and type 2 diabeticatients may result from common genes. Bothype 1 and type 2 DN cluster in the same fam-lies26 and show a similar disease tempo withephropathy onset approximately 10 to 15ears after diabetes onset.27 If hyperglycemicxposure can be considered an environmentalondition that primes the glomerulus regardlessf the cause, then the similar patterns of re-ponse under the influence of either type 1 orype 2 diabetes may have an explanation.
Second, how long should the diabetic rela-ives of probands with DN remain free of ne-hropathy after the initiation of diabetes to beonsidered truly unaffected (and contribute todiscordant relative pair)? Absent a more sen-
itive marker of DN progression, a lack of albu-inuria (�30 mg/g albumin-to-creatinine ratio
n repeated measurements) after 10 years ofiabetes was considered strong evidence of re-istance to DN. The figure of 10 years is notased on an arbitrary threshold but is based onpidemiologic studies that established that theeak of nephropathy initiation is within 10 to0 years after the initiation of diabetes,28,29 andhe risk for nephropathy plateaus approxi-ately 25 to 30 years after disease onset.8 In
ontrast, retinopathy risk increases with diabe-es duration. Intrinsic within this simple idea iscomplex model—individuals can be nephrop-thy-free because they do not carry a specificusceptibility allele at a particular locus, or theyan be nephropathy-free because they haverotective alleles elsewhere in the genome thatuppress the outcome of a deleterious allele, or,nally, their glomerular cells are resistant to the
ffects of the environmental insult, such that hhey can withstand hyperglycemic exposure foronger periods without consequence. The latterlso would be the result of the action of modi-er genes, just not genes directly controlling anbvious avenue to DN. It may not be possible toisentangle all these possibilities in a singlexperiment, and the genetic epidemiologytudy designs chosen may influence which ofhese outcomes are detectable. The strengthsnd weaknesses of the study designs pertinento gene identification are discussed later afteronsideration of some of these problems.
ENETIC COMPLEXITY: A DILEMMA OFODEST SAMPLE SIZES?
ene mapping studies run the gamut from col-ection of unrelated cases and controls, to fam-ly-based studies, to cohort studies (Fig. 1, Table). All of these study designs have been appliedo identifying genes for DN, and nested withinhese basic designs specific molecular (genetic)ontrasts have been devised with the goal ofest using the study population at hand. Severalarly studies attempted to identify genes for DNith small to modest sample sizes, but subse-uent investigations (both linkage and associa-ion) were unable to replicate findings in otheramples or attempts at replication were noteported. Few of these investigations have pro-uced results that meet modern guidelines fortatistical significance in the context of a mul-ifactorial trait.30 Primed by the successes inonogenic disorders (eg, polycystic kidney dis-
ase) in which the gene effect size is large, theack of speedy identification of DN genes ledhe general scientific audience to conclude thatenetic complexity precluded the identificationf genes for DN susceptibility. However, aseviewed later the sample sizes were modest inhe majority of investigations and ascertainmentriteria were not standardized between studies,or were the chromosome locations or candi-ate genes that were investigated similar be-ween the studies. Thus, direct comparison be-ween the studies was difficult and noturprisingly the results seldom were concor-ant. Some of these early reports reflected ei-her false-positives findings (type 1 error) orample-specific signals whose effect was en-
anced by the modest size of the sample. Fail-uppmhsermuissac
fsdsttpnsef
ac
Fdapt(cp((
Diabetic nephropathy genes—genome-wide tools 211
re to replicate gene mapping results in inde-endent samples, a key requirement forroving that a gene variant causes a disease,ay be the result of small sample sizes, geneticeterogeneity between samples from varioustudies, ascertainment bias (also leading to het-rogeneity), phenocopies, unaccounted envi-onmental correlates, and other epigeneticechanisms. Replication studies face some
nique design issues. The exact hypothesis be-ng examined in replication studies is not theame as the null hypothesis in the originaltudy. As an example, if in an initial scan therere 10 genes in total genome-wide that can
DNCase
DN Case
Diabetes Control
Affected relative pairs and family-based designs
DNCase
DNCase
DN Case
Diabetes Control
DNCase
DN Case
Diabetes Control
Affected relative pairs and family-based designs
DNCaseDN
Case
Family-based trios for Diabetic Nephropathy
DN Case
DNCase
DC
Family-based trios for Diabetic Nephropathy
DN Case
DNCase
DC
DN Case Diabetes ConDN Case Diabetes Con
DNCase
DN Case
Diabetes Control
Affected relative pairs and family-based designs
DNCase
DNCase
DN Case
Diabetes Control
DNCase
DN Case
Diabetes Control
Affected relative pairs and family-based designs
DNCaseDN
Case
Family-based trios for Diabetic Nephropathy
DN Case
DNCase
DC
Family-based trios for Diabetic Nephropathy
DN Case
DNCase
DC
DN Case Diabetes ConDN Case Diabetes ConA
B
C
igure 1. Study designs used in gene mapping experimesigns test for association whereas family based designsre more powerful for association testing in the contextower for rare variants. Some rare variants are difficult toypes of mutations. (A) Potential uses for the illustrated stc) gene � environment interaction; (d) genome-wideandidate genes; (b) gene � gene interaction; (c) gearent-of-origin effects; (f) sex-influenced traits. (C) Poteb) gene � gene interaction; (c) gene � environment inf) sex-influenced traits.
ause nephropathy, finding sufficient evidence o
or linkage to any 1 of these 10 is considered auccess. In a replication study, genetic evi-ence at a specific locus identified in the initialcan must be reproduced. If patient pheno-ypes in the second data set are not identical tohose characteristics in the first set, gene map-ing results may be at variance. This does notecessarily imply that the results of the secondcan or the first scan are false, merely that morevidence is needed to advance the experimentorward.
Each method for gene mapping has its owndvantages and disadvantages, and these mayhange with advances in technology. All meth-
etes ntroletes
ntroletes
ntrol
Study Design Uses
a) candidate genes
b) gene x gene interaction
c) gene x environment interaction
d) genome-wide scans
e) parent-of-origin effects
f) sex-influenced traits
Study Design Uses
a) candidate genes
b) gene x gene interaction
c) gene x environment interaction
d) genome-wide scans
e) parent-of-origin effects
f) sex-influenced traits
Study Design Uses
a) candidate genes
b) gene x gene interaction
c) gene x environment interaction
d) genome-wide scans
etes ntroletes
ntroletes
ntrol
Study Design
a) candidate genes
b) gene x gene interaction
c) gene x environment interaction
d) genome-wide scans
e) parent-of-origin effects
f) sex-influenced traits
Study Design
a) candidate genes
b) gene x gene interaction
c) gene x environment interaction
d) genome-wide scans
e) parent-of-origin effects
f) sex-influenced traits
Study Design
a) candidate genes
b) gene x gene interaction
c) gene x environment interaction
d) genome-wide scans
or complex traits. Case control and trio (or TDT)-basedst for linkage or association. The first 2 designs generallymmon variants that cause disease. The latter has morect and only direct sequencing may be able to find thesesigns: (a) candidate genes; (b) gene � gene interaction;(B) Potential uses for the illustrated study designs: (a)environment interaction; (d) genome-wide scans; (e)ses for the illustrated study designs: (a) candidate genes;on; (d) genome-wide scans; (e) parent-of-origin effects;
DiabCo
DiabCo
DiabCo
NaseN
ase
trolstrols
DiabCo
DiabCo
DiabCo
NaseN
ase
trolstrols
ents fcan teof codete
udy descans.ne �ntial uteracti
ds are subject to ascertainment bias, and the
eCgbsffectrnw
astbtaDapp
iG1EpTwctdeidNopg(tppwl
212 S.K. Iyengar, B.I. Freedman, and J.R. Sedor
xtent of the bias depends on the study design.ohort studies that adequately represent theeneral population have the least bias, whereasoth family-based designs and case-control de-igns may have significant bias. The advantageor the latter is of course the increase in poweror gene finding, by recruiting subjects withxtreme phenotypes. Although identifying sus-eptibility genes is often within the means ofhe study, determining population-attributableisk (the fraction of risk that would be elimi-ated if the risk factor was removed) is difficultith common gene mapping designs.31-35
We have summarized large-scale studies thatre in the process of gene identification for DNusceptibility and perhaps also for nephroprotec-ion (Table 2). Most studies have a population-ased collection strategy, the exceptions beinghe Family Investigation of Nephropathy and Di-betes (FIND)36 and the Genetics of Kidneys iniabetes (GoKinD) studies.37 The FIND study hadmixed design with approximately half the sam-le comprising families and the other half com-
Table 1. Study Designs for Gene Mapping Stu
Characteristic Case-Control Design
Phenotypecontrast
Predefined choice of diseasecontrol status
Sample sizerequired
Small to modest
Collection process Can be clinic or center base
Type of genetichypothesisexamined
Only association analysis witgenes, environment or ge� environment
Ascertainment bias Yes; can lead to biased oddratio to estimate disease ri
Study oflongitudinaloutcomes
No
rising an admixture mapping approach in Mex- i
can Americans and African Americans. TheoKinD study focused on trios of an affected typeparticipant and their 2 parents. Although the
pidemiology of Diabetes Interventions and Com-lications/Diabetes Control and Complicationsrial (EDIC/DCCT) has a clinical trial base, effortsere made to assess the family history of compli-
ations among participants.26 Table 2 shows thathe outcomes under investigation and the studyesigns are fairly distinct, although each generallyxamines some DN trait. Further, the major focuss on type 1 and not the more common type 2iabetes; all of the studies but FIND and Theetherlands Cooperative Study on the Adequacyf Dialysis examined type 1 diabetes. Other inde-endent type 2 diabetes studies in specific ethnicroups also have been published in the literatureeg, the DN genome scan in African Americans,38
he DN case control single-nucleotide polymor-hism [SNP] scan in the Japanese39), but the sam-le sizes for the majority of these investigationsere modest. These cohorts can be used for rep-
ication when genes are found, but the contrast-
Retrospective/ProspectiveCohorts
LinkageStudies
an examine multiple diseasestages
Predefinedchoice ofdisease
arge Small tomodest
opulation-based and requiresepidemiologic principles offollow-up evaluation
Can be clinic orcentercontrolled
f data on familial relationshipscollected, either associationor linkage can be examined
Eitherassociation orlinkage testscan beperformed
o Yes; maymagnifyeffects of rarealleles
es No
dies
/ C
L
d P
hne
I
ssk
N
Y
ng study designs will make comparing the effect
ssmtsacduwr
rlstoi
pagtnstl
L
WaTci3
Diabetic nephropathy genes—genome-wide tools 213
ize of the gene quite difficult and the compari-ons should be accepted with caution. Thus, theajority of the studies have some bias, even ones
hat represent larger populations because of theampling scheme. Several analyses were initiateds cross-sectional studies and later expanded toohorts. The benefit of these studies is that theata are being collected prospectively and theyniformly are using strict phenotyping protocols,hich will allow for computation of attributable
isk.Recently the use of standardized criteria for
ecruitment and large sample sizes have al-owed the emergence of unifying results acrosstudies. This should increase confidence in fu-ure trials to identify DN genes. The prospectsf success already have been realized with the
Table 2. Large-Scale Investigations in Diabetic
Consortium Name Study Design Diab
EDIC/DCCT26,69 Prospective cohortlinked with tightglucose control trial
Type
EURODIAB70,71 Prospective cohort Type
FIND36 Cross-sectional familybased; mapping byadmixturedisequilibrium
Predotyp
FINNDIANE72 Prospective cohort Type
GENDIAN73 Case control;prospective
Type
GoKinD37,74,75 Cross-sectional trios(family based)
Type
NECOSAD76,77 Prospective cohort Both
EURODIAB, European Diabetes IDDM Complications Study GGenetic and Clinical Predictors of Morbidity, Mortality, anMellitus Type 2; NECOSAD, The Netherlands Cooperative
dentification of specific DN genes, for exam- a
le, carnosinase (CNDP1)40 and engulfmentnd cell motility 1 (ELMO1),39 although theseenes need to be validated in other popula-ions. Similarly, linkage and/or association sig-als have been replicated at several chromo-omal regions (eg, 10p, 18q), suggesting thathese regions harbor important susceptibilityoci (discussed later).
INKAGE STUDIES
e have compiled a list of linkage studies for DNnd all-cause nephropathy (Table 3).9-11,38,41-51
he early linkage studies considered DN as a di-hotomous outcome, but the later studies havencluded albuminuria as a quantitative trait (Table). Other studies with glomerular filtration rate as
hropathy
ype Outcome Locale
Multiplecomplications ofdiabetes
Multicenter UnitedStates/Canada
Multiplecomplications ofdiabetes
Multiple Europeancenters; UnitedStates
ntlyM
Severely affected anddiscordant relativepairs with DN;cases and controlswith and withoutDN
Multicenter UnitedStates
Follow-up evaluationof DN in 25% ofadult type 1diabetics
Finland
Follow-up evaluationof DN and othercomplications oftype 2 diabetes
Germany
Probands with CKDand parents
Multicenter UnitedStates
New ESRD casefollow-upevaluation
The Netherlands
INNDIANE, Finnish Diabetic Nephropathy Study; GENDIAN,etic Nephropathy with End Stage Renal Disease in Diabeteson the Adequacy of Dialysis.
Nep
etes T
1 DM
1 DM
minae 2 D
1 DM
2 DM
1 DM
roup; Fd DiabStudy
n endophenotype also are forthcoming. The first
gahiAsah
diac2ht
214 S.K. Iyengar, B.I. Freedman, and J.R. Sedor
enome scan for DN was performed in the Pimand included 98 affected pairs,9 but sample sizesave increased significantly since these types of
nvestigations were begun in the late 1990s.mong family based studies, the largest samplesizes will come from FIND and GoKinD for type 2nd type 1 DN, respectively. For example, FIND
Table 3. Summary of Linkage Studies for DN
PhenotypeDiabetes
Type Population
All-cause ESRD41 AfricanAmerican
All-cause ESRD42 AfricanAmerican
DN9 Type 2 Pima IndianAll-cause ESRD11 African
AmericanDN10 Type 1 CaucasianAll-cause ESRD43,44 African
AmericanDN45 Type 2 Turkish
Albumin-to-creatinine ratio46
Multipleethnicgroups
DN47 Type 2 CaucasianandAfricanAmerican
DN38 Type 2 AfricanAmerican
All-cause ESRD48 AfricanAmerican
Albumin-to-creatinine ratio49
Type 2 Multipleethnicgroups
DN50 Type 1 Finnish
DN and albumin-to-creatinine ratio51
Both, mostlytype 2
Multipleethnicgroups
ASP, concordantly affected sibling pair; DSP, discordant sibli*Genome scan not performed.†Does not comprise the full FIND sample.
as collected approximately 1,200 families from 4 e
ifferent ethnic groups, Caucasian, African Amer-can, Native American, and Mexican American,cross the United States. Similarly, the GoKinDollection encompasses 71 cases, 623 controls,72 case trios with a type 1 diabetic probandaving nephropathy, and 323 control trios ascer-ained by a diabetic proband without renal dis-
KD in the Literature
udy Design Sample Size Linkage
65 families Renin-angiotensin-aldosterone*
65 families? Cytokinegenes*
98 ASPs 3q, 7q, 9, 20142 ASPs Kallikrein
genes*66 DSPs 3q*129 ASPs and
356 ASPs10p and 10q
18 extendedfamilies
18q22
antitativeait
805 families 19, 12q
, DSP, USP 27 Caucasianand 38 AfricanAmericanfamilies
10p*
166 families 3q, 7p, 18q
483 extendedfamilies
13q33.3,9q34.3,4p15.32,1q25.1
antitativeait analysis
63 extendedfamilies
22q, 7q, 5q
P 83 DSPs 3q, 4p, 9q,22p
, DSP,SP,uantitativeaits
378 families† 7q, 10p, 14q,18q
r; USP, concordantly unaffected sibling pair.
and C
St
ASP
ASP
ASPASP
ASPASP
ASP
Qutr
ASP
ASP
ASP
Qutr
DS
ASPUqtr
ng pai
ase. This data set is better suited to genome-wide
ahbpd
trhpsdla2dhrt
lufo
t1litOhTtedbs7li(lswssp
Fmcti
Diabetic nephropathy genes—genome-wide tools 215
ssociation mapping than for linkage. Other co-orts described in Table 2, although not familyased, also have the potential to discover ne-hropathy genes through association analyses asescribed later.
In total, only 5 genome scans performed forype 2 DN and 1 for type 1 DN have beeneported in the literature. The definitions of DNave evolved over time, with microalbuminuricatients being considered affected in sometudies but not in others. The establishment ofiabetes before nephropathy initiation and the
ength of diabetes duration may have been vari-ble between studies as well, especially in typeDN. These subtle features have not been wellocumented and it is possible that undetectedeterogeneity has crept into the analyses as aesult of differences in the clinical characteris-ics of recruited patients.
In comparing the genome scans, most pub-ished studies have discovered at least onenique locus (Table 3). The best evidencerom the genome scans abstracted from the
* * * * * * * * *
D7S
2477
D7S
3056
D7S
513
D7S
3051D
7S1802
D7S
1808
D7S
817
D7S
2846
D7S
1818
Bowden, 2004:max LOD=3.59
ELMO1
01
23
45
0 50
LOD
Sco
re
igure 2. Comparison of results from four five studies oap (X-axis). The approximate areas of the best regio
hromosome. Either the log of the likelihood ratio (LODhe height of the bars is proportional to the significancen Table 3 and references 9, 38, 39, 49, and 51.
riginal articles shows little replication be- m
ween studies except at 10p and 18q. Within0p the marker D10S1435 shows evidence of
inkage in at least 3 studies for type 2 DN,ncluding FIND,40,42,43,52 and is validated fur-her in an association study of type 1 DN.53
n chromosome 18q, the carnosinase geneas been identified after a linkage scan of 18urkish families multiplex for DN45 and 2 of
he larger published linkage scans showedvidence in this region.38,40 Fine mapping toetermine if genetic variation in carnosinaseest explains the linkage signal in these 2tudies still is needed. Although chromosome
consistently has been reported to showinkage and association for DN, careful exam-nation of these loci does not show overlapFig. 2). Evidence against linkage at specificoci has not been compared rigorously acrosstudies, and loci that fail to meet genome-ide significance criteria in replication may
how signal corroboration at more modestignificance levels. One mechanism to com-are genetic evidence across studies is a
* * * * * * * * * * *D
7S2204
D7S
2212
D7S
821D
7S1799
D7S
3061
D7S
1804
D7S
1824D
7S2195
D7S
3070
D7S
3058
D7S
559
max LOD=2.04Imperatore, 1998:
Krolewski, 2006:max LOD=3.1
FIND Study:lue= 0.0000171
100 150 180
M
2 DN on chromosome 7, placed on the microsatelliteeach study are circumscribed along the length of theor the P value are marked along the Y-axis. Therefore,
linkage or association evidence. Each study is described
*
D7S
3046
p-va
c
f typens for) scoreof the
eta-analysis with the primary data from each
subbcwd
lvtgssheotp(tadtcgsaddblsodcwtbrtcleqpliuT
taf
utedwlp(ot2c(wmiaeitnbtdbb
CSA
CDmbcsbWafoeftc
216 S.K. Iyengar, B.I. Freedman, and J.R. Sedor
tudy, an approach that has not yet beenndertaken by the kidney community but haseen used with success to evaluate suscepti-ility alleles for type 2 diabetes and otheromplex genetic traits.54-56 Meta-analyses alsoould enable comparisons in the phenotypeefinitions between studies.Successful, systematic fine mapping of DN
inkage signals to find the causative genes orariants has been reported. Janssen et al40 iden-ified carnosinase (CNDP1) as the causativeene under the 18q linkage peak in the Turkishample and confirmed this finding in a sampleet obtained from Pima Indians. Other groupsave confirmed that the trinucleotide repeat inxon 2 of the CNDP1 gene, with homozygosityf a 5-leucine repeat in the leader peptide, pro-ects diabetes mellitus patients against ne-hropathy, particularly in Caucasian samplesB. I. Freedman, personal communication). In-erestingly, Zschocke et al57 were unable to seeny association between CNDP1 and coronaryisease or survival/longevity, suggesting thathis gene may be specific for renal failure sus-eptibility. It is unclear if a common set ofenes regulates both DN initiation and progres-ion. The distinction between initiation genesnd progression genes has been difficult to ad-ress. Individuals with renal insufficiency un-ergo a survival bottleneck as a result of comor-id cardiovascular disease, which frequently
eads to early mortality. Some investigators havepeculated that death of DN patients as a resultf cardiovascular disease is not random but isriven by a common gene set for kidney andardiovascular disease. Therefore, survivorsho progress to proteinuria and ESRD may re-
ain only a fraction of DN susceptibility alleles,iasing findings toward survival alleles or less-obust disease susceptibility variants. In prac-ice, using very strict trait definitions in case-ontrol studies may bias which genes areocated as a result of this phenomenon. How-ver, in family based studies the use of theuantitative data in all available siblings shouldrovide some safeguards against such a trend; a
inkage peak that retains a good proportion ofts signal after the exclusion of microalbumin-ric individuals should argue for a lack of bias.
he investigation by Zschocke et al57 addressed she issue of common genes for coronary diseasend survival bias in ESRD. Many more studiesocusing on this gene are underway.
In addition to fine mapping efforts followingp linkage signals, another study58 used theransmission disequilibrium test (TDT)59,60 tovaluate candidate type 1 DN genes in trios, aesign that requires inclusion of a child affectedith DN and both parents (ie, a trio). Neuropi-
in 1 (NRP1) is located under the 10p linkageeak identified in several genome-wide scansdiscussed previously) and marginal evidencef association was reported for 2 NRP1 SNPs. Inhe same report, the B-cell leukemia/lymphoma
proto-oncogene (BCL2), which is located onhromosome 18q21 near the carnosinaseCNDP1) locus, was reported to be associatedith type 1 DN. The BCL2 association signalay indicate the presence of a second DN gene
n the 18q region or may reflect extended link-ge disequilibrium (correlation between mark-rs near each other on the same chromosome)n this region. The linkage scans that reportedhe 10p and 18q signals are based predomi-antly on type 2 diabetes, and the investigationy Ewens et al58 and by McKnight et al52 may behe first evidence for commonality of geneticeterminants between type 1 and type 2 DN,ut the results, although intriguing, remain toe confirmed in larger-scale studies.
ANDIDATE GENE STUDIES ANDYSTEMATIC GENOME-WIDESSOCIATION SCREENS
andidate gene analysis has been conducted forN and ESRD in hundreds of genes because it isore feasible for a single investigator to assem-
le a modest number of cases and controls. Theoncern with this approach is that the geneselected are driven solely by previously knowniology and therefore are limited in spectrum.e have not attempted to review the many
ssociation studies published in the literatureor single genes, but have focused this reviewn the newer studies that systematically havexamined more than 100 genes or have per-ormed genome-wide association scans. Our ra-ionale for this approach is that many singleandidate gene reports were based on a small
ample size and minimal coverage (one or onlyafscagS
ghgs2pbteatnEmSbcdswtcnmcsnf�rfttoasoaw4ha
a(JmTnliEdndatTailirectHnpar
DtwpqbitotaDrtcln(d(
Diabetic nephropathy genes—genome-wide tools 217
few markers) within or near the gene. There-ore, results are hard to interpret because mosttudies lack adequate power to detect an asso-iation. In recent years, standards for reportingssociations have changed,61,62 and even singleene investigations need to genotype multipleNPs in a sufficiently large sample.
Three published studies have canvassed theenome for DN variants using association orave genotyped a large number of genes at areater depth. These articles include a genomecan using 81,315 SNPs of 87 patients with typeDN and 92 controls without DN in a Japaneseopulation,39 scanning of 200 Irish type 1 dia-etic patients with overt nephropathy and 200ype 1 diabetic patients with normal albuminxcretion using 6,000 microsatellite markers,52
nd the evaluation of 115 candidate genes usinghe TDT approach for type 1 DN.58 The Japa-ese association scan lead to the discovery ofLMO1, a novel gene on chromosome 7, thatay be associated with DN. In the scan, 1,615
NP loci had significant P values of less than .01etween DN and control patients. Patients andontrols were ascertained at clinics and wereesignated according to their renal functionaltatus. Thus, patients with DN, that is, patientsith diabetic retinopathy and overt nephropa-
hy, were individuals with urinary albumin ex-retion rates of 200 �g/min or greater or uri-ary albumin-to-creatinine ratios of 300 mg/g orore of creatinine, or patients receiving
hronic renal-replacement therapy. Controlubjects included patients with diabetic reti-opathy but showing no evidence of renal dys-unction (ie, urinary albumin excretion rates20 �g/min or urinary albumin-to-creatinine
atios �30 mg/g creatinine). The SNPs chosenor genotyping were selected randomly fromhe gene-based Japanese SNP database. Ofhese, the best evidence was at SNP rs741301n 7p14. Two other nearby SNPs, rs7799004nd rs1558688, also showed good associationignals in a haplotype analysis. The evidenceutside of those 3 SNPs was weaker but stilldequate. The association of the rs741301 SNPith DN was confirmed in a second sample of
59 patients and 242 controls. However, theaplotype around this SNP shows very weak
ssociation, probably owing to very weak link- mge disequilibrium in the sample. The risk alleleG; reference allele is A) is rare in the Chinese,apanese, and Caucasian populations, but is theore common allele in the Yoruba population.he SNP itself is intronic—in intron 18—and isot functional, and falls within a conserved hap-
otype block between exons 16 and 22, suggest-ng that the sentinel SNP may not be the causativeLMO1 variant. Because of the reduced linkageisequilibrium (lack of correlation betweeneighboring markers) within ELMO1, it may beifficult to replicate this report until a causal vari-nt is identified. This hypothesis is supported byhe haplotype analyses reported in this article.he G risk allele is found in haplotypes bothssociated and not associated with DN, suggest-ng that the causative variant has not yet beenocated. Therefore, typing only the 3 most signif-cant SNPs from the Japanese sample, rs741301,s7799004, and rs1558688, may or may not showvidence of association in a different sample. Byomparing the haplotype blocks and examiningagging SNPs from several different populations inAPMAP63,64 we have estimated that in most eth-ic groups it will be necessary to genotype ap-roximately 90 to 100 SNPs to obtain full cover-ge of this gene, showing the need to planeplication studies carefully.
The genome scan by McKnight et al52 used aNA pooling paradigm. In this study, DNA from
he affected (patients) and control enrolleesas pooled separately and genotyped for ap-roximately 6,000 markers. The allele fre-uency profiles at each marker were comparedetween case and control pools, and the result-
ng profiles were ranked in order of the greatesto the least difference between pools. Markersn 10p (described previously) showed associa-ion with DN. Although the best evidence forssociation was on 10p, other loci including6S281, D4S2937, D2S291, and D17S515,
anked next in order of priority. Participants inhis study were type 1 diabetic patients andontrols from Northern Ireland and the Repub-ic of Ireland. Phenotype criteria for a case defi-ition required an onset of persistent proteinuria0.5 g protein/24 h) at least 10 years after theiabetes diagnosis, the presence of hypertensionblood pressure � 135/85 mm Hg and/or treat-
ent with antihypertensive agents), and the pres-edmomrianWsr
iTsgIIgCniaiart
sgetrgwccatcewppdnDgi
tsy
C
OataipnrWwrpuctnscidcae
vmttmc(uqtawTFbtootD
218 S.K. Iyengar, B.I. Freedman, and J.R. Sedor
nce of diabetic retinopathy. Control subjects hadiabetes for at least 15 years, normal urinary albu-in excretion, and normal blood pressures with-
ut treatment. DNA pooling, although not a com-only used disease gene mapping design, can
educe genotyping costs significantly, and whenntegrated into a 2-stage analytic plan it has beenn efficient approach for the association of ge-etic variants with other disease phenotypes.65
ith only 6,000 microsatellite markers, thecreen reported by McKnight et al52 has a lowesolution.
These 2 whole-genome association screensllustrate 2 approaches to gene mapping design.he phenotype definition in the Japanesecreen was much less stringent whereas theenotyping design was more comprehensive.n contrast, the ascertainment scheme in therish scan was much more restrictive, but theenotyping screen was less comprehensive.urrently, there is a move to very dense ge-ome-wide association screens with genotyp-
ng of 500,000 to 1,000,000 markers to ensuredequate coverage,66 and neither of these 2nvestigations meet that criterion. Their effortss landmark studies in the DN arena should beegarded as the initiation points for more de-ailed studies.
Finally, a study by Ewens et al,58 an extensiveurvey of 115 known and novel candidateenes, deserves mention. The study by Ewenst al58 examined 72 families with type 1 diabe-es and nephropathy, defined as ESRD or 2 of 3andom urine albumin-to-creatinine ratiosreater than 300 �g/mg collected at least 6eeks apart, and used the TDT to test for asso-
iation. This family based association studyombined both linkage and association testingnd controlled for population substructureshat resulted in false-positive results in case-ontrol designs from gene mapping. Modestvidence was detected for association of DNith multiple candidate genes, including aqua-orin 1, B-cell leukemia/lymphoma 2 (BCL2)roto-oncogene, catalase, glutathione peroxi-ase 1, insulin-like growth factor (IGF)1, lami-in alpha 4, laminin gamma 1, mothers againstPP homolog (SMAD), SMAD 3, transformingrowth factor beta receptors II and III, tissue
nhibitor of metalloproteinase 3, and upstream branscription factor 1. However, the sampleize is this study was small and the results haveet to be confirmed in other studies.
ONCLUSIONS AND FUTURE DIRECTIONS
ne third of all diabetic subjects remain undi-gnosed in the United States.67 The complica-ions of diabetes accrue in these individualsnd, as shown by Koopman et al,67 after adjust-ng for age and diagnosed or undiagnosed hy-ertension, the association between undiag-osed diabetes and nephropathy persists (oddsatio, 2.35; 95% confidence interval, 1.38-4.01).
ith an unprecedented increase in the world-ide incidence of diabetes looming in the ho-
izon, it is very important that multiple ap-roaches to identifying individuals at risk bendertaken to reduce the burden of diabeticomplications. In this review, we have shownhat an array of techniques has been used by theephrology community to map genes for DNusceptibility. Each of these techniques has spe-ific advantages and disadvantages, but the find-ngs require replication to be meaningful. Asescribed earlier, replication may not be per-eptible at first glance and careful secondarynalysis may be necessary to reconcile appar-ntly discrepant results.
Replication will be particularly hard if rareariants cause disease. The refocusing of geneapping efforts from large-scale linkage studies
o large-scale association studies has changedhe paradigm from finding both rare and com-on variants to finding common variants. Dis-
overy of the polycystic kidney disease genesPKD1 and PKD2) was successful through these of traditional linkage mapping and subse-uent fine mapping in families. The allelic spec-rum of mutations in PKD1 and PKD2 is size-ble,68 and it is unclear if association methodsould have been able to find these genes.herefore, the results of a single cohort (eg,IND, EDIC/DCCT, or GoKinD) should be note evaluated in isolation but rather put in con-ext of the type of study design and the powerf the design to find genes for specific elementsf type 1 or type 2 DN. For example, the kidneyraits emphasized in FIND are extreme type 2N phenotypes, with the majority of probands
eing on dialysis or having greater than 1 gabrrvacpfptmiwisEsntDagm
estasMFin(aSastttmGtaitap
tae
ATPt
R
1
1
1
Diabetic nephropathy genes—genome-wide tools 219
lbuminuria. However, siblings of FIND pro-ands span the range of albuminuria from noenal disease (�30 �g/mg albumin-to-creatinineatio) to proteinuric patients (�1 g/24 h), withariable duration of diabetes. In contrast, thescertainment strategies of EDIC/DCCT ex-luded subjects with overt albuminuria and ne-hropathy complications have developed over
ollow-up periods of up to 23 years. GoKinDrobands included subjects with type 1 diabe-es and both microalbuminuria and normoalbu-inuria. The FIND design has the potential to
dentify both rare and common disease genes,hereas the power in EDIC/DCCT and GoKinD
s geared toward common disease genes. Con-idering the strengths of the other cohorts,DIC/DCCT and GoKinD have a better repre-entation from the population with less pro-ounced disease, features that are conducive tohe identification of initiation genes. EDIC/CCT also has longitudinal data and may beble to distinguish initiation genes from pro-ression genes using survival analyses–typeethods.With the advent of new SNP technology,
nabling thousands of markers to be genotypedimultaneously, new DN gene mapping initia-ives are in progress. FIND is in the process of
6,000� marker genome-wide SNP linkagecan. Admixture scans in African Americans andexican Americans are also on the agenda for
IND. GoKinD has been accepted for genotyp-ng by a new public-private initiative called Ge-etic Association Information Network (GAIN)http://www.fnih.org/GAIN/GAIN_home.shtml)nd will undergo the genotyping for 1,000,000NP markers. Similarly, the EDIC/DCCT cohortlso is expected to initiate a very dense genomecan in that sample. Therefore, we anticipatehat a vast amount of data will be generatedhrough these larger-scale initiatives in the fu-ure. A comparison of the variety of genetic andolecular data between studies such as FIND,oKinD, and EDIC/DCCT is anticipated to de-
ect genes for DN. The identification of ELMO1nd CNDP1 already has spawned investigationsnto the biology of these genes in an effort toranslate the basic science to clinical medicinend therapeutic approaches. In conclusion, the
rospects for DN genetics are very positive, andhe interplay between these larger initiativesnd other genome-scale projects should help tolucidate the pathobiology of DN.
cknowledgmentshe authors would like to thank Shannon Quade,aula Wedig, and Sarah Ialacci for technical assis-ance with the manuscript.
EFERENCES1. Seaquist ER, Goetz FC, Povey S. Diabetic nephropa-
thy: an hypothesis regarding genetic susceptibility forthe disorder. Minn Med. 1986;69:457-9.
2. Scalvini T, Spandrio S, di-Stefano O, et al. Severemicrovascular disease in type II diabetes of early on-set. A family study. Acta Diabetol Lat. 1989;26:58-74.
3. Seaquist ER, Goetz FC, Rich S, et al. Familial clusteringof diabetic kidney disease. Evidence for genetic sus-ceptibility to diabetic nephropathy. N Engl J Med.1989;320:1161-5.
4. Pettitt DJ, Saad MF, Bennett PH, et al. Familial predis-position to renal disease in two generations of PimaIndians with type 2 (non-insulin-dependent) diabetesmellitus. Diabetologia. 1990;33:438-43.
5. Kalter LO, Van-Dyk DJ, Leibovici L, et al. Risk factorsfor development of diabetic nephropathy and reti-nopathy in Jewish IDDM patients. Diabetes. 1991;40:204-10.
6. Freedman BI, Spray BJ, Tuttle AB, et al. The familialrisk of end-stage renal disease in African Americans.Am J Kidney Dis. 1993;21:387-93.
7. Krolewski AS, Doria A, Magre J, et al. Molecular ge-netic approaches to the identification of genes in-volved in the development of nephropathy in insulin-dependent diabetes mellitus. J Am Soc Nephrol. 1992;3:S9-17.
8. Quinn M, Angelico MC, Warram JH, et al. Familialfactors determine the development of diabetic ne-phropathy in patients with IDDM. Diabetologia.1996;39:940-5.
9. Imperatore G, Hanson RL, Pettitt DJ, et al. Sib-pairlinkage analysis for susceptibility genes for microvas-cular complications among Pima Indians with type 2diabetes. Pima Diabetes Genes Group. Diabetes.1998;47:821-30.
0. Moczulski DK, Rogus JJ, Antonellis A, et al. Majorsusceptibility locus for nephropathy in type 1 diabe-tes on chromosome 3q: results of novel discordantsib-pair analysis. Diabetes. 1998;47:1164-9.
1. Yu H, Bowden DW, Spray BJ, et al. Identification ofhuman plasma kallikrein gene polymorphisms andevaluation of their role in end-stage renal disease.Hypertension. 1998;31:906-11.
2. Rogus JJ, Krolewski AS. Using discordant sib pairs tomap loci for qualitative traits with high sibling recur-
rence risk. Am J Hum Genet. 1996;59:1376-81.1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
220 S.K. Iyengar, B.I. Freedman, and J.R. Sedor
3. Hoogwerf BJ, Brouhard BH. Glycemic control andcomplications of diabetes mellitus: practical implica-tions of the Diabetes Control and Complications Trial(DCCT). Cleve Clin J Med. 1994;61:34-7.
4. Jones SL, Viberti GC. Hypertension and microalbu-minuria as predictors of diabetic nephropathy. Diabe-tes Metab. 1989;15:327-32.
5. Levey AS, Coresh J, Greene T, et al. Using standard-ized serum creatinine values in the modification ofdiet in renal disease study equation for estimatingglomerular filtration rate. Ann Intern Med. 2006;145:247-54.
6. Stevens LA, Coresh J, Greene T, et al. Assessing kid-ney function—measured and estimated glomerularfiltration rate. N Engl J Med. 2006;354:2473-83.
7. Peterson JC, Adler S, Burkart JM, et al. Blood pressurecontrol, proteinuria, and the progression of renaldisease. The Modification of Diet in Renal DiseaseStudy. Ann Intern Med. 1995;123:754-62.
8. de Jong PE, Curhan GC. Screening, monitoring, andtreatment of albuminuria: public health perspectives.J Am Soc Nephrol. 2006;17:2120-6.
9. Borch-Johnsen K, Feldt-Rasmussen B, Strandgaard S,et al. Urinary albumin excretion. An independentpredictor of ischemic heart disease. ArteriosclerThromb Vasc Biol. 1999;19:1992-7.
0. Fogarty DG, Hanna LS, Wantman M, et al. Segregationanalysis of urinary albumin excretion in families withtype 2 diabetes. Diabetes. 2000;49:1057-63.
1. Langefeld CD, Beck SR, Bowden DW, et al. Heritabil-ity of GFR and albuminuria in Caucasians with type 2diabetes mellitus. Am J Kidney Dis. 2004;43:796-800.
2. Imperatore G, Knowler WC, Pettitt DJ, et al. Segrega-tion analysis of diabetic nephropathy in Pima Indians.Diabetes. 2000;49:1049-56.
3. Fogarty DG, Rich SS, Hanna L, et al. Urinary albuminexcretion in families with type 2 diabetes is heritableand genetically correlated to blood pressure. KidneyInt. 2000;57:250-7.
4. Forsblom CM, Kanninen T, Lehtovirta M, et al. Heri-tability of albumin excretion rate in families of pa-tients with type II diabetes. Diabetologia. 1999;42:1359-66.
5. Fioretto P, Steffes MW, Barbosa J, et al. Is diabeticnephropathy inherited? Studies of glomerular struc-ture in type 1 diabetic sibling pairs. Diabetes. 1999;48:865-9.
6. The Diabetes Control and Complications Trial Re-search Group. Clustering of long-term complicationsin families with diabetes in the diabetes control andcomplications trial. The Diabetes Control and Com-plications Trial Research Group. Diabetes. 1997;46:1829-39.
7. Biesenbach G, Janko O, Zazgornik J. Similar rate ofprogression in the predialysis phase in type I and typeII diabetes mellitus. Nephrol Dial Transplant. 1994;9:1097-102.
8. Svensson M, Nystrom L, Schon S, et al. Age at onset
of childhood-onset type 1 diabetes and the devel- 4opment of end-stage renal disease: a nationwidepopulation-based study. Diabetes Care. 2006;29:538-42.
9. Kussman MJ, Goldstein H, Gleason RE. The clinicalcourse of diabetic nephropathy. JAMA. 1976;236:1861-3.
0. Lander E, Kruglyak L. Genetic dissection of complextraits: guidelines for interpreting and reporting link-age results. Nat Genet. 1995;11:241-7.
1. Burton PR, Palmer LJ, Jacobs K, et al. Ascertainmentadjustment: where does it take us? Am J Hum Genet.2000;67:1505-14.
2. Dawson DV, Elston RC. A bivariate problem in humangenetics: ascertainment of families through a corre-lated trait. Am J Med Genet. 1984;18:435-48.
3. Elston RC, Sobel E. Sampling considerations in thegathering and analysis of pedigree data. Am J HumGenet. 1979;31:62-9.
4. Ginsburg E, Malkin I, Elston RC. Sampling correctionin pedigree analysis. Stat Appl Genet Mol Biol. 2003;2:Article 2.
5. Ginsburg E, Malkin I, Elston RC. Sampling correctionin linkage analysis. Genet Epidemiol. 2004;27:87-96.
6. Knowler WC, Coresh J, Elston RC, et al. The FamilyInvestigation of Nephropathy and Diabetes (FIND):design and methods. J Diabetes Complications. 2005;19:1-9.
7. Mueller PW, Rogus JJ, Cleary PA, et al. Genetics ofKidneys in Diabetes (GoKinD) study: a genetics col-lection available for identifying genetic susceptibilityfactors for diabetic nephropathy in type 1 diabetes.J Am Soc Nephrol. 2006;17:1782-90.
8. Bowden DW, Colicigno CJ, Langefeld CD, et al. Agenome scan for diabetic nephropathy in AfricanAmericans. Kidney Int. 2004;66:1517-26.
9. Shimazaki A, Kawamura Y, Kanazawa A, et al. Geneticvariations in the gene encoding ELMO1 are associatedwith susceptibility to diabetic nephropathy. Diabetes.2005;54:1171-8.
0. Janssen B, Hohenadel D, Brinkkoetter P, et al. Car-nosine as a protective factor in diabetic nephropathy:association with a leucine repeat of the carnosinasegene CNDP1. Diabetes. 2005;54:2320-7.
1. Yu H, Bowden DW, Spray BJ, et al. Linkage analysisbetween loci in the renin-angiotensin axis and end-stage renal disease in African Americans. J Am SocNephrol. 1996;7:2559-64.
2. Freedman BI, Yu H, Spray BJ, et al. Genetic linkageanalysis of growth factor loci and end-stage renaldisease in African Americans. Kidney Int. 1997;51:819-25.
3. Freedman BI, Rich SS, Yu H, et al. Linkage heteroge-neity of end-stage renal disease on human chromo-some 10. Kidney Int. 2002;62:770-4.
4. Yu H, Sale M, Rich SS, et al. Evaluation of markers onhuman chromosome 10, including the homologue ofthe rodent Rf-1 gene, for linkage to ESRD in blackpatients. Am J Kidney Dis. 1999;33:294-300.
5. Vardarli I, Baier LJ, Hanson RL, et al. Gene for suscep-
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
6
6
6
6
6
6
6
6
6
7
7
7
7
Diabetic nephropathy genes—genome-wide tools 221
tibility to diabetic nephropathy in type 2 diabetesmaps to 18q22.3-23. Kidney Int. 2002;62:2176-83.
6. Freedman BI, Beck SR, Rich SS, et al. A genome-widescan for urinary albumin excretion in hypertensivefamilies. Hypertension. 2003;42:291-6.
7. Iyengar SK, Fox KA, Schachere M, et al. Linkageanalysis of candidate loci for end-stage renal diseasedue to diabetic nephropathy. J Am Soc Nephrol.2003;14:S195-201.
8. Freedman BI, Bowden DW, Rich SS, et al. A genomescan for all-cause end-stage renal disease in AfricanAmericans. Nephrol Dial Transplant. 2005;10:2719-27.
9. Krolewski AS, Poznik GD, Placha G, et al. A genome-wide linkage scan for genes controlling variation inurinary albumin excretion in type II diabetes. KidneyInt. 2006;69:129-36.
0. Osterholm AM, He B, Pitkaniemi J, et al. Genome-wide scan for type 1 diabetic nephropathy in theFinnish population reveals suggestive linkage to asingle locus on chromosome 3q. Kidney Int. 2006;71:140-5.
1. Iyengar SK, Abboud HE, Goddard KAB, et al, and onbehalf of the Family Investigation of Nephropathyand Diabetes. Genome-wide scans for diabetic ne-phropathy and albuminuria in multi-ethnic popula-tions. the Family Investigation of Nephropathy andDiabetes. Diabetes. In press.
2. McKnight AJ, Maxwell AP, Sawcer S, et al. A genome-wide DNA microsatellite association screen to iden-tify chromosomal regions harboring candidate genesin diabetic nephropathy. J Am Soc Nephrol. 2006;17:831-6.
3. An P, Freedman BI, Hanis CL, et al. Genome-widelinkage scans for fasting glucose, insulin, and insulinresistance in the national heart, lung, and blood insti-tute family blood pressure program: evidence of link-ages to chromosome 7q36 and 19q13 from meta-analysis. Diabetes. 2005;54:909-14.
4. Altshuler D, Hirschhorn JN, Klannemark M, et al. Thecommon PPARgamma Pro12Ala polymorphism is as-sociated with decreased risk of type 2 diabetes. NatGenet. 2000;26:76-80.
5. Grant SF, Thorleifsson G, Reynisdottir I, et al. Variantof transcription factor 7-like 2 (TCF7L2) gene confersrisk of type 2 diabetes. Nat Genet. 2006;38:320-3.
6. Tsuchiya T, Schwarz PE, Bosque-Plata LD, et al. Asso-ciation of the calpain-10 gene with type 2 diabetes inEuropeans: results of pooled and meta-analyses. MolGenet Metab. 2006;89:174-84.
7. Zschocke J, Nebel A, Wicks K, et al. Allelic variationin the CNDP1 gene and its lack of association withlongevity and coronary heart disease. Mech AgeingDev. 2006;127:817-20.
8. Ewens KG, George RA, Sharma K, et al. Assessment of115 candidate genes for diabetic nephropathy bytransmission/disequilibrium test. Diabetes. 2005;54:3305-18.
9. Spielman RS, McGinnis RE, Ewens WJ. Transmission 7
test for linkage disequilibrium: the insulin generegion and insulin-dependent diabetes mellitus(IDDM). Am J Hum Genet. 1993;52:506-16.
0. Spielman RS, Ewens WJ. A sibship test for linkagein the presence of association: the sib transmission/disequilibrium test. Am J Hum Genet. 1998;62:450-8.
1. Freimer NB, Sabatti C. Guidelines for association stud-ies in human molecular genetics. Hum Mol Genet.2005;14:2481-3.
2. Ehm MG, Nelson MR, Spurr NK. Guidelines for con-ducting and reporting whole genome/large-scale as-sociation studies. Hum Mol Genet. 2005;14:2485-8.
3. The International HapMap Consortium. The Interna-tional HapMap Project. Nature. 2003;426:789-96.
4. The International HapMap Consortium. Integratingethics and science in the International HapMapProject. Nat Rev Genet. 2004;5:467-75.
5. Hinds DA, Seymour AB, Durham LK, et al. Applicationof pooled genotyping to scan candidate regions forassociation with HDL cholesterol levels. Hum Genom-ics. 2004;1:421-34.
6. Barrett JC, Cardon LR. Evaluating coverage of ge-nome-wide association studies. Nat Genet. 2006;38:659-62.
7. Koopman RJ, Mainous AG III, Liszka HA, et al. Evi-dence of nephropathy and peripheral neuropathy inUS adults with undiagnosed diabetes. Ann Fam Med.2006;4:427-32.
8. Ong AC, Harris PC. Molecular pathogenesis of AD-PKD: the polycystin complex gets complex. KidneyInt. 2005;67:1234-47.
9. Epidemiology of Diabetes Interventions and Compli-cations (EDIC). Design, implementation, and prelim-inary results of a long-term follow-up of the DiabetesControl and Complications Trial cohort. DiabetesCare. 1999;22:99-111.
0. Lloyd CE, Stephenson J, Fuller JH, et al. A comparisonof renal disease across two continents; the epidemi-ology of diabetes complications study and the EURO-DIAB IDDM Complications Study. Diabetes Care.1996;19:219-25.
1. Roglic G, Colhoun HM, Stevens LK, et al. Parentalhistory of hypertension and parental history of diabe-tes and microvascular complications in insulin-depen-dent diabetes mellitus: the EURODIAB IDDM Compli-cations Study. Diabet Med. 1998;15:418-26.
2. Pettersson-Fernholm K, Forsblom C, Perola M, et al.Polymorphisms in the nephrin gene and diabetic ne-phropathy in type 1 diabetic patients. Kidney Int.2003;63:1205-10.
3. Boger CA, Haak T, Gotz AK, et al. Effect of ACE andAT-2 inhibitors on mortality and progression to mi-croalbuminuria in a nested case-control study of dia-betic nephropathy in diabetes mellitus type 2: resultsfrom the GENDIAN study. Int J Clin Pharmacol Ther.2006;44:364-74.
4. Cordovado SK, Hancock LN, Simone AE, et al. High-
7
7
7
222 S.K. Iyengar, B.I. Freedman, and J.R. Sedor
resolution genotyping of HLA-DQA1 in the GoKinDstudy and identification of novel alleles HLA-DQA1*040102, HLA-DQA1*0402 and HLA-DQA1*0404.Tissue Antigens. 2005;65:448-58.
5. Hancock LN, Cordovado SK, Hendrix M, et al. Identifi-cation of two novel DQA1 alleles, DQA1*0107 andDQA1*0602, by sequence-based typing in the GoKinDpopulation. Hum Immunol. 2005;66:1248-53.
6. van der Sman-de Beer, Verhagen C, Rombach SM, et
al. ACE I/D polymorphism is associated withmortality in a cohort study of patients starting withdialysis. Kidney Int. 2005;68:2237-43.
7. Jager KJ, Merkus MP, Boeschoten EW, et al. Dialysis inThe Netherlands: the clinical condition of new pa-tients put into a European perspective. NECOSADStudy Group. Netherlands Cooperative Study on theAdequacy of Dialysis phase 1. Nephrol Dial Trans-
plant. 1999;14:2438-44.