Genetic Admixture and Asthma-Related Phenotypes inMexican American and Puerto Rican Asthmatics

Keyan Salari,1,2w Shweta Choudhry,1,2w Hua Tang,3 Mariam Naqvi,1,2 Denise Lind,4 Pedro C. Avila,1

Natasha E. Coyle,1,2 Ngim Ung,1,2 Sylvette Nazario,5 Jesus Casal,5 Alfonso Torres-Palacios,5

Suzanne Clark,1,2 Angie Phong,1 Ivan Gomez,5 Henry Matallana,1,2 Eliseo J. Perez-Stable,1 Mark D. Shriver,6

Pui-Yan Kwok,3 Dean Sheppard,1,2 William Rodriguez-Cintron,5 Neil J. Risch,7,8 Esteban GonzalezBurchard,1,2zn and Elad Ziv,1zn from the Genetics of Asthma in Latino Americans (GALA) Study

1Department of Medicine, University of California, San Francisco, California2Lung Biology Center, San Francisco General Hospital, San Francisco, California

3Fred Hutchinson Cancer Research Center, Seattle, Washington4Cardiovascular Research Institute, University of California, San Francisco, California

5San Juan VAMC, University of Puerto Rico School of Medicine, San Juan, Puerto Rico6Department of Anthropology, Pennsylvania State University; University Perk, Pennsylvania

7Department of Genetics, Stanford University School of Medicine, Palo Alto, California8Division of Research, Kaiser Permanente, Oakland, California

Genetic association studies in admixed populations may be biased if individual ancestry varies within the population andthe phenotype of interest is associated with ancestry. However, recently admixed populations also offer potential benefits inassociation studies since markers informative for ancestry may be in linkage disequilibrium across large distances. Inparticular, the enhanced LD in admixed populations may be used to identify alleles that underlie a genetically determineddifference in a phenotype between two ancestral populations. Asthma is known to have different prevalence and severityamong ancestrally distinct populations. We investigated several asthma-related phenotypes in two ancestrally admixedpopulations: Mexican Americans and Puerto Ricans. We used ancestry informative markers to estimate the individualancestry of 181 Mexican American asthmatics and 181 Puerto Rican asthmatics and tested whether individual ancestry isassociated with any of these phenotypes independently of known environmental factors. We found an association betweenhigher European ancestry and more severe asthma as measured by both forced expiratory volume at 1 second (r¼�0.21,p¼0.005) and by a clinical assessment of severity among Mexican Americans (OR: 1.55; 95% CI 1.25 to 1.93). We found nosignificant associations between ancestry and severity or drug responsiveness among Puerto Ricans. These results suggestthat asthma severity may be influenced by genetic factors differentiating Europeans and Native Americans in MexicanAmericans, although differing results for Puerto Ricans require further investigation. Genet. Epidemiol. 29:76–86, 2005.& 2005 Wiley-Liss, Inc.

Key words: admixture; asthma; Hispanics/Latinos

w,zThese authors contributed equally to the manuscript.Contract grant sponsor: National Institutes of Health; Contract grant numbers: K23 HL04464, HL07185, GM61390, NCMHD HealthDisparities Scholar, Extramural Clinical Research Loan Repayment Program for Individuals from Disadvantaged Backgrounds, 2001–2003; Contract grant sponsor: The Resource Centers for Minority Aging Research from the National Institute on Aging; Contract grantnumber: (RCMAR P30-AG15272); Contract grant sponsor: The National Institute of Nursing Research and The National Center onMinority Health and Health Disparities to EGB; Contract grant sponsor: the Asthma Clinical Research Network; Contract grant numbers:HL51823 and 3M01RR000083-38S30488; Contract grant sponsor: SFGH General Clinical Research Center; Contract grant number:M01RR00083-41 and U01-HL 65899; Contract grant sponsor: UCSF-Children’s Hospital of Oakland Pediatric Clinical Research Center,Oakland, CA; Contract grant number: M01 RR01271; Contract grant sponsor: University of California, San Francisco; Contract grantnumber: K12 AR47659; Contract grant sponsor: the American Lung Association of California to EGB; Contract grant sponsor: SandlerCenter for Basic Research in Asthma and the Sandler Family Supporting Foundation.nCorrespondence to: Esteban Gonzalez Burchard, Department of Medicine, University of California, San Francisco, CA 94143-0833.E-mail: eburch@itsa.ucsf.edu and Elad Ziv, Division of General Internal Medicine, Dept. of Medicine University of California,San Francisco, CA 94143-1732. Email: eziv@itsa.ucsf.eduReceived 31 August 2004; Accepted 24 November 2004Published online 26 May 2005 in Wiley InterScience (www.interscience.wiley.com)DOI: 10.1002/gepi.20079

Genetic Epidemiology 29: 76–86 (2005)

& 2005 Wiley-Liss, Inc.

INTRODUCTION

Genetic association studies in unrelated indivi-duals are currently being pursued for manydiseases due to their power and relative efficiency.Genetic association studies directly test a variantbelieved to be either the true disease-variant or inlinkage disequilibrium (LD) with the disease-variant. Recently, admixed populations such asAfrican Americans and Latino ethnic groups areknown to have areas of LD that can extend overlarge chromosomal regions due to allele frequencydifferences between the ancestral populations[Chakraborty and Weiss, 1988]. This increasedLD among admixed populations can facilitatemapping complex trait genes in an approachgenerally referred to as admixture mapping[Risch, 1992; McKeigue, 1998; Zheng and Elston,1999; McKeigue et al., 2000; Patterson et al., 2004;Smith et al., 2004].

A potential complication of studying admixedpopulations is the possibility of spurious associa-tions due to population stratification [Pritchardand Rosenberg, 1999; Cardon and Palmer, 2003;Ziv and Burchard, 2003]. If the risk of diseasevaries with ancestry proportions, this will createassociations of disease with genotypes at anylocus where allele frequencies differ betweenancestral populations. Thus, admixture mappingfirst requires appropriate adjustment for popula-tion stratification.

Asthma prevalence and severity vary widelyamong different racial and ethnic groups in theUnited States. Puerto Ricans have the highestasthma prevalence, morbidity, and mortalitywhile Mexican Americans’ asthma prevalence,morbidity and mortality is lower than Caucasians’and African Americans’ [Carter-Pokras andGergen, 1993; Homa et al., 2000]. Within each ofthese Latino groups, individuals are descendantsof highly diverse ancestral groups includingNative-American, European, and African popula-tions. Although Puerto Ricans and Mexicans sharea common language and have arisen throughseveral centuries of admixture, the relative pro-portions of these ancestral populations betweenthe contemporary Puerto Rican and MexicanAmerican population are different [Hanis et al.,1991]. Thus, we examined the association ofindividual ancestry with asthma severity andother asthma-related phenotypes in these twogroups separately. Using 44 markers highlyinformative for ancestry, we estimated individualancestry in 181 Puerto Rican and 181 Mexican

American asthmatic subjects. In each population,we tested the association between individualancestry and measures of airway dynamics,bronchodilator drug responsiveness, clinical as-sessments of severity, and serum IgE levels.

MATERIALS AND METHODS

STUDY PARTICIPANTS

As part of the ongoing Genetics of Asthma inLatino Americans (GALA) Study [Burchard et al.,2004], patients with asthma were enrolled in theSan Francisco Bay Area, California (SF), PuertoRico (PR), and Mexico City, Mexico (MX). Asth-matic subjects from the San Francisco Bay Area(n¼181) and Puerto Rico (n¼181) were included inthis analysis, and their characteristics are shownin Table I. Subjects were recruited from commu-nity schools, clinics, and hospitals. In all healthcare centers, medical records were reviewed toidentify patients with physician-diagnosed asth-ma, who then were contacted to participate in thestudy. Asthmatic subjects between the ages of 8and 40 years, with physician-diagnosed asthmaand two or more asthma symptoms (amongwheezing, coughing, and shortness of breath) inthe last two years, were eligible to participate.Recruitment was standardized across all clinicalcenters. Mexican Americans were recruited fromfive sites in Northern and Central California: SanFrancisco, Oakland, Redwood City, Salinas, andFresno. Puerto Ricans were recruited from ninesites in Puerto Rico: Bareceloneta, Barranquitas,Bayamon, Carolina, Catano, Corozal, Humacao,Mayaguez, and San Juan. Bilingual and biculturalphysicians specializing in asthma were present atall interviews.

TABLE I. Characteristics and pulmonary test results ofstudy subjectsa

CharacteristicMexican Americans

(n¼181)Puerto Ricans

(n¼181)

Age, yr 11.9 (10:15) 12.2 (10:15)Sex, % Male 63.0 59.1Body mass index, kg/m2 24.3 (20:29) 21.2 (17:26)FEV1, % predicted 94.2 (86:103) 84.8 (76:95)DFEV1, % 6.6 (3:10) 5.2 (1:10)Serum IgE, IU/ml 251 (106:623) 280 (102:636)% Mild 43.1 39.2% US born 64.1 0

aValues are expressed as median (25th:75th percentile), and weremissing for some subjects.

Admixture and Asthma Severity Among Latinos 77

ASTHMA AND MEDICAL QUESTIONNAIRE

Asthmatic subjects were assessed using amodified version of the 1978 ATS-Divisionof Lung Disease Epidemiology Questionnaire[Ferris, 1978]. All subjects’ forms and question-naires were available in English and Spanish.Ethnicity was self-reported. Subjects were enrolledonly if both biological parents and all biologicalgrandparents were of Puerto Rican (for the PRsite) or Mexican ethnicity (for the SF sites andMexico City). Local Institutional Review Boardsapproved the study, and all subjects signed formsindicating their consent to participate.

Pulmonary function tests and IgE measurements.Asthma is characterized by recurrent episodes ofwheeze, cough, and airway obstruction. Airwayobstruction is an indicator of asthma severity andcan be measured using spirometry. A standardmeasure of the severity of airway obstruction isforced expiratory volume in one second (FEV1).The lower the FEV1 is, the more severe the airwayobstruction is. Airway obstruction is reversiblewith the inhalation of medications such asalbuterol, the most commonly prescribed asthmamedication in the world. The reversibility ofairway obstruction is a measure of drug respon-siveness. Reversibility can be measured by per-forming spirometry before and after theadministration of albuterol and measuring thedifference (DFEV1).

Asthmatic subjects were instructed to withholdtheir bronchodilator medications for at least 8hours before lung function tests. Spirometry wasperformed according to established standards(Standardization of Spirometry, 1995), and base-line FEV1 is described herein as Pre-FEV1. Bronch-odilator drug responsiveness to albuterol isreported as percent change in forced FEV expira-tory volume in 1 second from baseline (DFEV1)after albuterol administration. Pulmonary func-tion tests are expressed as a percentage of thepredicted normal value using age-adjusted pre-diction equations from Hankinson and colleagues[Hankinson et al., 1999].

For this analysis, asthmatic subjects wereclassified as having ‘‘mild’’ or ‘‘moderate-severe’’asthma based on three yes/no questions related tomedication use and asthma symptoms: (1) is thepatient’s only asthma medication a bronchodilatorused on an as-needed basis? (2) without respect tomedications used, has the patient had dailyasthma symptoms for three or more months of

the previous year? and (3) has the patientexperienced nocturnal asthma 41 night per weekfor 3 or more months of the previous year?Asthmatic subjects were classified as mild if theyanswered yes to question 1 and no to all otherquestions. Subjects were classified as moderate-severe if they answered no to question 1 or yes toeither question 2 or 3. Total plasma IgE wasmeasured in duplicate using Uni-Cap technology(Pharmacia, Kalamazoo, MI).

MARKER SELECTION

Forty-four autosomal ancestry informative mar-kers (AIMs) were identified from the NCBI SNPdatabase (dbSNP) or previously reported litera-ture as being highly informative for ancestry andwere genotyped in Mexican and Puerto Ricansubjects. In addition, we genotyped our panel of44 AIMs in African, European, and Native-American populations to estimate their allelefrequencies in the ancestral populations. Theancestral allele frequencies were determined bygenotyping the markers in populations collectedfrom Sub-Saharan Africa (n¼481): Nigeria, CentralAfrican Republic, Sierra Leone; Europe (n¼243):Ireland, England, Germany and Spain; and Na-tive-American populations indigenous to theUnited States and Mexico (n¼148): Maya, Pima,Cheyenne and Pueblo. Delta (d) determinesancestral informativeness and is defined as theabsolute difference in allele frequency betweentwo ancestral populations. A d value of 1 impliescomplete ancestry informativeness; a d value of 0implies no informativeness for ancestry. Table IIshows the AIMs used in this study, along withtheir chromosomal locations and delta values.These markers were selected because they exhibithigh levels of allele frequency differences (d 40.3) between any two of the three parentalpopulations. There were 31, 33, and 23 ?informa-tive markers with d 4 0.3 between African andEuropean ancestry, African and Native-Americanancestry, and European and Native-Americanancestry, respectively.

Information regarding primer sequences, poly-morphic sites, and other relevant information hasbeen submitted to the dbSNP NCBI database,under the submitter handle PSU-ANTH and/orKWOK; the dbSNP reference SNP (rs) numbersare indicated in Table II and further informationcan be found at (http://www.ncbi.nlm.nih.gov/SNP/).

Salari et al.78

GENOTYPING

The 44 AIMs were genotyped using the Acy-cloPrime-FPt (PerkinElmer) method [Chen et al.,1999]. PCR conditions were: 6 ml volume withPlatinum Taq PCR buffer, 2.5 mM MgCl2, 2.4–4.0ng genomic DNA, 50 mM dNTPs, 0.1–0.2 mM

primers, 0.1–0.2 U Platinum Taq (Invitrogen) plus1 ml extra water to counteract evaporation. Cyclingconditions were: 951C for 2 min, 35 cycles of 921Cfor 10 sec, 581C for 20 sec, 681C for 30 sec, andfinal extension at 681C for 10 min. Enzymaticcleanup and single base extension genotypingreactions were performed with AcycloPrime-FP

TABLE II. Ancestry informative markersa

d

Marker name dbSNP accession Location African European Native American Afr/Eur Afr/NA Eur/NA

MID-575 rs140864 1p34.3 0.11 0.01 0.55 0.10 0.45 0.55MID-187 rs3188519 1p34.1 0.76 0.37 0.32 0.39 0.44 0.05FY-null rs2814778 1q23.2 0.00 0.99 0.99 0.99 0.99 0.00F13B rs6003 1q31.3 0.70 0.08 0.03 0.62 0.67 0.05TSC1102055 rs2065160 1q32.1 0.50 0.92 0.17 0.42 0.32 0.75

TSC0053441 rs723632 1q32.3 0.10 0.92 0.67 0.82 0.57 0.25WI-16857 rs3287 2p16.1 0.73 0.20 0.21 0.53 0.52 0.01WI-11153 rs17203 3p12.3 0.81 0.15 0.76 0.66 0.05 0.61TSC1157118 rs584059 3q22.3 0.49 0.14 0.47 0.35 0.03 0.33

GC1 rs7041 4q13.3 0.93 0.41 0.45 0.52 0.48 0.04TSC0255473 rs930072 5p13.2 0.96 0.10 0.45 0.86 0.51 0.35

SGC30610 rs3309 5q11.2 0.40 0.28 0.69 0.12 0.29 0.41SGC30055 rs3317 5q23.1 0.05 0.59 0.73 0.53 0.68 0.15WI-17163 rs3340 5q33.2 0.06 0.19 0.65 0.13 0.59 0.47

WI-9231 rs2763 7p22.3 0.14 0.16 0.52 0.03 0.38 0.36TSC0003523 rs1985080 7p14.3 0.10 0.64 0.97 0.54 0.87 0.32

WI-4019 rs2161 7q22.1 0.44 0.30 0.62 0.14 0.18 0.32

LPL rs285 8p21.3 0.97 0.52 0.45 0.45 0.52 0.07WI-11909 rs2695 9q21.31 0.81 0.86 0.22 0.05 0.59 0.64TSC0879894 rs518116 9q33.3 0.13 0.67 0.58 0.54 0.45 0.09TSC0000409 rs7349 10p11.22 0.04 0.87 0.96 0.83 0.92 0.08TSC0316844 rs994174 10q23.1 0.76 0.25 0.26 0.51 0.49 0.02D11S429 rs594689 11q11 0.09 0.47 0.13 0.37 0.04 0.34TYR-192 rs1042602 11q21 0.00 0.47 0.05 0.46 0.05 0.41

DRD2-BclI rs1079598 11q23.1 0.06 0.14 0.63 0.08 0.57 0.49

DRD2-Taq rs1800498 11q23.1 0.14 0.65 0.09 0.51 0.05 0.56

TSC0322307 rs326946 11q23.1 0.61 0.17 0.07 0.44 0.54 0.10TSC0072803 rs146026 13q13.1 0.26 0.92 0.83 0.66 0.57 0.09TSC0030203 rs736394 14q32.12 0.52 0.74 0.99 0.22 0.48 0.26OCA2 rs1800404 15q13.1 0.14 0.72 0.48 0.58 0.34 0.23WI-14319 rs2862 15q14 0.38 0.17 0.69 0.21 0.30 0.51

CYP19 rs4646 15q21.2 0.32 0.29 0.72 0.03 0.40 0.43

Rs223830 rs223830 16q13 0.03 0.19 0.64 0.16 0.62 0.45

MC1R-314 rs2228478 16q24.3 0.51 0.14 0.04 0.37 0.47 0.09WI-14867 rs2891 17p13.2 0.02 0.51 0.43 0.49 0.40 0.08WI-7423 rs2816 17p12 0.00 0.49 0.08 0.49 0.07 0.42

TSC0745571 rs203096 17q21.33 0.65 0.72 0.28 0.07 0.37 0.44

TSC0042312 rs2077863 18p11 0.51 0.93 0.93 0.41 0.42 0.00CKM rs4884 19q13.32 0.15 0.29 0.86 0.13 0.71 0.57

TSC0000580 rs1989486 19q13.42 0.04 0.58 0.40 0.53 0.36 0.17MID-154 rs3188520 20q11.22 0.83 0.35 0.44 0.48 0.39 0.09TSC0050288 rs722098 21q21.1 0.90 0.18 0.72 0.73 0.19 0.54

TSC1023401 rs235936 21q21.3 0.18 0.49 0.37 0.31 0.20 0.11MID-93 rs16383 22q13.2 0.27 0.80 0.11 0.52 0.17 0.69

Mean 0.42 0.42 0.30

aThe marker names, accession numbers, chromosomal locations, allele frequencies and differences in frequencies (d) between African andEuropean (Afr/Eur), African and Native-American (Afr/NA), and European and Native-American (Eur/NA) populations are shown.d values 40.3 are in bold.

Admixture and Asthma Severity Among Latinos 79

kits. Plates were read on an EnVision fluorescencepolarization plate reader (PerkinElmer).

STATISTICAL ANALYSIS

Individual admixture estimates were derivedfrom a maximum likelihood algorithm, as im-plemented in the program IBGA [Parra et al.,2001]. Group admixture levels were calculatedwith the program ADMIX, which implements aweighted least squares method [Long, 1991]. Totest the association between quantitative asthma-related traits (Pre-FEV1 DFEV1, and serum IgE)and ancestry, individual ancestry was incorpo-rated as a continuous variable into linear regres-sion models. Serum IgE levels were log-transformed to make the distribution approxi-mately normal and all subsequent references toIgE levels indicate this transformation. We usedlogistic regression models to test the associationbetween individual ancestry and asthma severity,defined as a qualitative phenotype. In each model,we used the maximum likelihood estimate of theindividual ancestry as a predictor. We entered age,sex, asthma duration, regular use of asthmamedication, and body mass index (BMI) into themodel as covariates. Regular use of asthmamedication was considered a covariate since itmay be a marker of access to care and, thus, mayconfound the association between severity andancestry. To adjust for potential environmentalinteractions, second-hand exposure to environ-mental tobacco smoke (ETS) and birthplace werealso incorporated into models as covariates. ETSexposure was determined to be present if one orboth parents reported current smoking and absentif both parents were current non-smokers. We alsoincluded the clinic or hospital from which theparticipant was recruited as a categorical covariatein the model. For Mexican Americans, we alsoincluded socioeconomic status as a covariate. Weused the U.S. census tract definitions of socio-ecomic status to classify individuals into 1 of 4socioeconomic categories: low, moderate, middle,and upper. These categories are based on themedian household income for the census tract forthe year 2000. We entered the categories ascategorical variables. For Puerto Ricans, we werenot able to characterize SES from the datacollected.

We used a forward stepwise procedure to selectcovariates for each model. Tests for individualmarker associations were performed using alinear regression model with the assumption of

an additive genetic model (genotypes entered as0, 1, or 2 alleles). We adjusted the associationsfor individual markers using a multivariate modelin which individual ancestry was entered as acovariate. Regression models were estimatedusing the program STATA 8.0 S/E (CollegeStation, TX). We present nominal p values in thetext and tables. However, for associations thatmeet a nominal significance level, we alsoconsider a more conservative Bonferroni correc-tion as noted in the text.

RESULTS

ADMIXTURE ESTIMATES

Among the Mexican Americans as a group,ancestry was estimated to be 3.470.97% African,44.971.7% European, and 51.771.7% NativeAmerican. Among Puerto Ricans, ancestry wasestimated as 16.271.6% African, 65.572.2% Eur-opean, and 18.372.1% Native American. Esti-mates for individual admixture demonstrate avery wide range of genetic ancestry amongindividuals in these two highly admixed Latinopopulations (Fig. 1), although the standard errorson individual admixture estimates are large.

MEXICAN-AMERICAN POPULATION

We found a significant relationship betweenindividual ancestry and asthma severity (asdefined by Pre-FEV1) in our Mexican-Americanpopulation (Table III, Fig. 2). European ancestrywas associated with more severe asthma [definedby lower Pre-FEV1 (Pearson: r¼�0.211, p¼0.0051;Spearman: rho¼0.228, p¼0.0024)]. European an-cestry remained a significant predictor of Pre-FEV1 in a multivariate regression model afteradjustment for age, sex, asthma duration, regularuse of asthma medication, ETS exposure, birth-place, socioeconomic status, recruitment site, andBMI. The forward stepwise regression modelidentified age and ETS exposure as the only othersignificant effectors of Pre-FEV1. A decrease of1.7% (95% CI: 0.6% �2.8%) Pre-FEV1 was ob-served per 10% increase in European ancestry(Table III). There is a strong inverse correlationbetween Native-American and European ancestry.Therefore, any negative association with Eur-opean ancestry would be expected to have apositive association with Native-American ances-try. As expected, in models where Native-American ancestry was the main predictor,Native-American ancestry was associated with

Salari et al.80

Mex

ican

Am

eric

an IA

Est

imat

es

0%20%

40%

60%

80%

100%

17

Ind

ivid

ual

% IA

0%20%

40%

60%

80%

100%

% IA

Afr

ican

Eur

opea

nN

ativ

e A

mer

ican

Pu

erto

Ric

an IA

Est

imat

es

1319

2531

3743

4955

6167

7379

8591

9710

310

911

512

112

713

313

914

515

115

716

316

917

518

1

17

Ind

ivid

ual

1319

2531

3743

4955

6167

7379

8591

9710

310

911

512

112

713

313

914

515

115

716

316

917

518

1

Fig

.1.I

nd

ivid

ual

an

cest

ry(I

A)

est

imate

s.E

stim

ate

sfo

r181

Mexi

can

-Am

eri

can

ast

hm

ati

cs(t

op

)an

d181

Pu

erto

-Ric

an

ast

hm

ati

cs(b

ott

om

)are

sho

wn

,clu

ste

red

by

ad

mix

ture

lev

el.

Th

ed

istr

ibu

tio

no

fin

div

idu

al

an

cest

rye

stim

ate

sin

Me

xica

nA

me

rica

ns

cov

ers

the

ran

ge

of

Eu

rop

ea

na

nd

Na

tiv

e-A

me

rica

np

rop

ort

ion

s,w

hil

eA

fric

an

an

cest

ryco

ntr

ibu

tes

ve

ryli

ttle

toth

isp

op

ula

tio

n.

Ho

we

ve

r,in

Pu

erto

Ric

ans,

Afr

ica

na

nd

Eu

rop

ea

na

nce

stri

es

sho

wa

hig

hd

eg

ree

of

va

ria

bil

ity,

wh

ile

Na

tiv

e-A

me

rica

na

nce

stry

ex

hib

its

am

ore

rest

rict

ed

pa

tte

rno

fv

ari

ati

on

.

milder asthma [defined as higher measures ofPre-FEV1 (Pearson: r¼0.176, p¼0.0197; Spearman:rho¼0.186, p¼0.0139)]. After correction for theaforementioned potential confounders, Native-American ancestry remained significantly asso-ciated with higher Pre-FEV1 values. Additionally,ETS exposure was found to be associated withPre-FEV1. We found no association between Pre-FEV1 and African ancestry (r¼0.092, p¼0.224). Noassociations were found between individual an-cestry and either DFEV1 or IgE levels amongMexican Americans.

We also tested the association between ancestryand asthma severity, comparing the proportion ofancestry among mild versus severe asthmatics, as

defined by asthma medication use and clinicalsymptoms. We found an association betweenhigher European ancestry and asthma severity(Table IV). For each 10% increase in Europeanancestry there was an approximate 37% riskincrease for severe asthma. Adjustment for Pre-FEV1, age, sex, asthma duration, ETS exposure,birthplace, and BMI did not affect the association.Conversely, Native-American ancestry was asso-ciated with decreased risk of asthma severity, withan odds ratio of 0.66 (95% CI 0.54–0.81, p¼0.00006)for having severe asthma versus mild asthma per10% increase in Native-American admixture. Bothof the associations between European ancestryand lower Pre-FEV1 and increased clinical severityare significant after adjustment for multiplehypothesis testing, considering that 4 phenotypeswere tested against ancestry (Bonferroni adjustedcutoff for significance is p¼0.0125).

We also noted that there was a higher Europeanadmixture in U.S.-born Mexicans (n¼116) than inMexico-born Mexicans (n¼65; 44.7% vs. 38.6%,p¼0.036), although the adjustment for birthplacedid not change the strength of the associationbetween ancestry and asthma severity.

PUERTO RICAN POPULATION

Among Puerto Rican asthmatics, we found nosignificant association between African ancestryand asthma-related phenotypes after adjustmentfor multiple hypothesis testing. There was aborderline association between African ancestryand higher bronchodilator drug responsiveness[DFEV1 (Pearson: r¼0.163, p¼0.0321) at the nom-inal level of significance (Table V). However, thisassociation became non-significant when weadjusted for known potential confounders (bCoefficient: 0.63; 95% CI: �0.27 to 1.54, p¼0.16)and was not included in the final model (Table V).

TABLE III. Association of ancestry with asthma severity (Pre-FEV1)

Population Significant predictors Regression coefficient 95% Confidence interval p value

Mexican AmericansUnadjusted European IA �1.50a �2.54 to �0.45 0.005Adjustedb European IA �1.71a �2.83 to �0.59 0.003

ETS exposure 6.41 1.68 to 11.1 0.008

Puerto Ricans�16.7 to 13.2 0.82

Adjustedb Age �0.35 �0.66 to �0.048 0.023

aIndividual ancestry coefficients are reported as change in Pre-FEV1 per 10% increase in individual ancestry.bVariables included in each multivariable model were determined using a forward stepwise selection approach.

0

0

0

0

0 20 40 60 80

European Individual Ancestry (%)

Fig. 2. Association between European ancestry and asthmaseverity (defined as baseline Pre-FEV1) among Mexican Amer-

icans. The regression model shows a significant negative

correlation between European admixture and asthma severity

(r¼�0.211, p¼0.0051); that is, European ancestry is associatedwith more severe asthma as defined by Pre-FEV1.

Salari et al.82

We found no significant associations betweenbronchodilator drug responsiveness and Eur-opean ancestry (r¼�0.0589, p¼0.44) or Native-American ancestry (r¼�0.0814, p¼0.29). No asso-ciations between ancestry and quantitative orqualitative measures of asthma severity or IgElevels were observed in Puerto Ricans.

INDIVIDUAL MARKER ASSOCIATIONS

Associations between individual AIMs andasthma phenotypes were evaluated with andwithout correction for individual ancestry esti-mates. Table VI shows the independent associa-tions between single-locus genotypes and the

TABLE VI. Single-locus genotype associations with Pre-FEV1 and DFEV1

Pre-FEV1 DFEV1

Population Marker p valuea p value (IA)b Marker p valuea p value (IA)b

Mexican Americans DRD2-Taq 0.023 0.216 WI-16857 0.036 0.016OCA2 0.036 0.063 WI-9231 0.017 0.008

WI-14867 0.062 0.039 MID-93 0.019 0.031MID-93 0.024 0.142

Puerto Ricans TSC0072803 0.05 0.05 LPL 0.057 0.028MID-154 0.043 0.05 MC1R-314 0.017 0.02

TSC0050288 0.037 0.04

aNominal significance level for the unadjusted regression of marker genotype as a predictor of clinical outcome.bNominal significance level for the regression, after adjustment for individual ancestry.

TABLE IV. Association of ancestry with clinical severitya

Population Significant predictors Odds ratio 95% Confidence interval p value

Mexican AmericansUnadjusted European IAb 1.37 1.14 to 1.63 0.0006Adjustedc European IAb 1.55 1.25 to 1.93 0.00008

BMI (per kg/m2) 1.07 1.007 to 1.13 0.028Clinic 3 (Redwood City) 0.25 0.08 to 0.80 0.019

Puerto RicansAdjustedc Pre-FEV1 0.97 0.95 to 0.99 0.027

Clinic 9 (San Juan) 0.45 0.21 to 0.97 0.042

aAsthma severity (mild or severe) was assigned by qualitative, clinical measures of severity (asthma medication use, recurrent daily asthmasymptoms, and nocturnal asthma symptoms).bIndividual ancestry coefficients are reported as change in Pre-FEV1 per 10% increase in individual ancestry.cVariables included in each multivariable model were determined using a forward stepwise selection approach.

TABLE V. Association of ancestry with bronchodilator drug responsiveness (DFEV1)

Population Significant predictors Regression coefficient 95% Confidence interval p value

Mexican AmericansAdjusteda Pre-FEV1 �0.16 �0.24 to �0.084 0.00008

Clinic 5 (Fresno) 3.55 0.51 to 6.60 0.022Clinic 4 (Salinas) 4.99 0.19 to 9.78 0.042

Puerto RicansUnadjusted African IAb 0.94b 0.08 to 1.80 0.032Adjusteda Pre-FEV1 �0.21 �0.29 to �0.13 0.00007

Age (yrs) �0.26 �0.51 to �0.02 0.037Duration of asthma (yrs) 0.36 0.08 to 0.63 0.012

aVariables included in each model were determined using a forward stepwise selection approach.bIndividual ancestry coefficients are reported as change in DFEV1 per 10% increase in individual ancestry.

Admixture and Asthma Severity Among Latinos 83

phenotypes Pre-FEV1 or DFEV1 among MexicanAmericans and Puerto Ricans. In the unadjustedregression model, three out of 44 markers (7%)were nominally associated with Pre-FEV1 inMexican Americans at the 0.05 significance level.After correction for individual ancestry, only onemarker was significantly associated with Pre-FEV1, suggesting some degree of confoundingbetween ancestry and Pre-FEV1, consistent withthe association between European ancestry andPre-FEV1 among Mexican Americans. There wereseveral other modest associations between indivi-dual markers and other phenotypes at thenominal level of significance. However, none ofthese would be considered significant once thenumber of tests is taken into consideration.

DISCUSSION

Using 44 ancestry informative genetic markers(AIMs), we estimated individual ancestry among181 Puerto Rican and 181 Mexican Americanasthmatics. Among Mexican Americans, we foundan association between European ancestry andasthma severity, as defined by lower Pre-FEV1.Interestingly, an independent qualitative measureof asthma severity, based on asthma medicationuse and clinical symptoms, was also associatedwith European ancestry and, thus, corroboratedour initial findings. The observation of greaterasthma severity in individuals with higher Eur-opean ancestry suggests that one or more alleles athigher frequency among Europeans may increaseasthma severity in Mexican American popula-tions. However, this association may also be dueto unmeasured environmental factors. For exam-ple, a certain environmental factor may be morecommon in regions of Mexico that have higherEuropean ancestry. This may lead to an associa-tion between European ancestry and asthmaseverity, even though there may be no geneticfactors that underlie this association [Risch et al.,2002]. In addition, the association we observedmay be due to a bias if asthmatic patients withhigher European ancestry and less severe asthmaare less likely to use the community clinics fromwhich the patients in this study were recruited.Thus, additional studies of the Mexican popula-tion accounting for other potential environmentalrisk factors for asthma and studying this groupin diverse regions may help to clarify thisassociation.

We did not observe a similar trend towardsincreased asthma severity with European ancestryor decreased asthma severity with Native-Amer-ican ancestry in Puerto Ricans. This may be due tothe fact that the association between Europeanancestry and more severe asthma among MexicanAmericans is due to an unmeasured environmen-tal confounder as discussed above. Alternatively,there may be a gene–environment interaction thatunderlies this difference. For example, there maybe an environmental factor that increases theseverity of asthma only in the presence of acertain genotype, which is higher among people ofEuropean ancestry. Furthermore, the Puerto Ricanpopulation is somewhat more complex in geneticancestry. Puerto Ricans participating in the GALAstudy have substantial contributions from threedifferent ancestral populations, while MexicanAmericans are primarily European and NativeAmerican, with little African ancestry. Thus, morecomplex interactions between these ancestriesmay have altered the association. Moreover, theNative-American ancestral population of PuertoRico may be distinct from that of Mexico, asgenetic distance between Native-Americangroups has been shown to be relatively high[Rosenberg et al., 2002]. A reliable estimate for theNative-American ancestral population of PuertoRico is also difficult to ascertain, as very fewindividuals of the indigenous Taıno Indian tribesurvived after the European conquest of the island[Rouse, 1993].

A genetic interpretation of the association thatwe observed of more severe asthma amongMexican Americans with higher European ances-try suggests that Native Americans have lesssevere asthma compared to European-Americanpopulations. There are limited reports of asthmaprevalence among the Native Americans. Amer-ican Indian and Alaskan Native children whowere cared for by the Indian Health Servicesbetween 1979 and 1989 had rates of asthmadiagnosis at the time of hospital discharge similarto rates seen in white children [Hisnanick et al.,1994]. A more recent study found a non-signifi-cant lower rate of asthma among Native Amer-icans compared with European Americans in asurvey of school age children [Stout et al., 1999].However, the data comparing the phenotypes weanalyzed, clinical and spirometry-based measuresof asthma severity among diagnosed asthmatics,is even more limited than prevalence data andthus it is difficult to make any conclusionsabout asthma severity among European-American

Salari et al.84

asthmatics vs. Native-American asthmatics. Sinceour Puerto Rican data does not demonstrate anassociation between increased European ancestryand severity, our results in Mexican Americansshould be interpreted with caution. Eventually,only identification of a locus that underlies aseverity difference between Europeans and Na-tive Americans can confirm that this difference isdue to a genetic rather than an environmentaldifference.

The individual ancestry estimates that weobtained are statistical estimates, and for eachindividual these estimates may have considerablestatistical error. However, since the error shouldbe random with respect to the phenotypes ofinterest, the estimates of association should bebiased towards the null hypothesis. Therefore, theerror in estimating the associations in our analysislikely led to underestimating the strength of theassociations with individual ancestry and reducedpower.

In addition to associations between ancestryand asthma phenotypes, we found several indivi-dual AIMs to be associated with asthma severity(Pre-FEV1) and bronchodilator drug responsive-ness (DFEV1) in our Latino ethnic groups, evenafter correction for individual ancestry. Themarkers that were associated with severity amongMexican Americans in univariate analyses wereno longer significantly associated after correctionfor individual ancestry, suggesting that theseassociations were false positives due to stratifica-tion. Some of the other associations were noteliminated after correction for individual ancestry.These associations are all of marginal significanceand are likely the result of type I error. None ofthese markers would meet significance aftercorrection for multiple hypotheses testing. How-ever, since only a very small fraction of thegenome was covered with the markers used inthis study, we would not expect to find loci forasthma severity. Once adequate coverage of thegenome is achieved with AIMs, we should be ableto more definitively determine whether there areany specific loci that underlie the associationswith ancestry that we describe here. Globalstudies of allele frequency differences amongdifferent racial/ethnic populations suggest thatgenes that modify immune response are morelikely to have large allele frequency differences,presumably due to selection [Akey et al., 2002].Since such genes may affect asthma severity, thenadmixture mapping for asthma severity modifiersmay be a feasible approach.

In summary, our results demonstrate thatrecently admixed populations, such as Latinoethnic groups, provide both a challenge andopportunity for genetic association studies. Thepresence of population stratification in admixedpopulations leads to spurious associations be-tween unlinked markers and disease. However,such populations offer a potential advantage indisease gene mapping. Given that linkage dis-equilibrium in highly admixed populations spansextended regions of the genome, relatively fewmarkers that are informative for ancestry mayprovide coverage for detecting linkage to diseasealleles across the entire genome, compared to thenumber of random markers required for tradi-tional LD mapping [Shriver et al., 2003; McKeigue2005; Patterson et al., 2004; Smith et al., 2004].Therefore, with the use of ancestry informativemarkers, admixture mapping may be a feasiblemethod for discovering genes underlying asthmaseverity in admixed populations.

ACKNOWLEDGMENTS

The authors acknowledge the families and thepatients for their participation. The authors alsothank the numerous health care providers andcommunity clinics for their support and participa-tion in the GALA Study. In addition to theprimary clinical centers of the investigators,participating community clinics and hospitalsinclude: La Clinica de La Raza, Oakland, CA;UCSF-Children’s Hospital of Oakland PediatricClinical Research Center, Oakland, CA; GeneralClinical Research Center, SFGH, San Francisco,CA; Alliance Medical Center, Healdsburg, CA;Santa Clara Valley Medical Center, San Jose, CA;Fair Oaks Family Health Center, Redwood City,CA; Clinica de Salud del Valle de Salinas, Salinas,CA; Natividad Medical Center, Salinas, CA;Asthma Education and Management Program,Community Medical Centers, Fresno, CA; Diag-nostic Health Centers of: Corozal, Naranjito,Catano, Orocovis, Barranquitas, and San AntonioHospital of Mayaguez. The authors also acknowl-edge Monica Toscano, MariaElena Alioto, CarmenJimenez, Yannett Marcano, Pedro Yapor, MD,Alma Ortiz, MD, Lisandra Perez, MD, and SheilaGonzalez, MD, for their assistance with recruit-ment. The authors especially thank Jeffrey M.Drazen, MD, Ed Silverman, MD, PhD, Homer A.Boushey, MD Scott Weiss, MD, and Jean G. Ford,

Admixture and Asthma Severity Among Latinos 85

MD, for all of their effort towards the creation ofthe GALA Study.

ELECTRONIC DATABASEINFORMATION

http://www.ncbi.nlm.nih.gov/SNP/

REFERENCES

Akey JM, Zhang G, Zhang K, Jin L, Shriver MD. 2002.Interrogating a high-density SNP map for signatures ofnatural selection. Genome Res 12:1805–1814.

Burchard EG, Avila PC, Nazario S, Casal J, Torres A, Rodriguez-Santana JR, Toscano M, Sylvia JS, Alioto M, Salazar M, Gomez I,Fagan JK, Salas J, Lilly C, Matallana H, Ziv E, Castro R, SelmanM, Chapela R, Sheppard D, Weiss ST, Ford JG, Boushey HA,Rodriguez-Cintron W, Drazen JM, Silverman EK, Genetics ofAsthma in Latino Americans (GALA) Study. 2004. Lowerbronchodilator responsiveness in Puerto Rican than inMexican subjects with asthma. Am J Respir Crit Care Med169:386–392.

Cardon LR, Palmer LJ. 2003. Population stratification andspurious allelic association. Lancet 361:598–604.

Carter-Pokras OD, Gergen PJ. 1993. Reported asthma amongPuerto Rican, Mexican-American, and Cuban children, 1982through 1984. Am J Public Health 83:580–582.

Chakraborty R, Weiss KM. 1988. Admixture as a tool for findinglinked genes and detecting that difference from allelicassociation between loci. Proc Natl Acad Sci USA 85:9119–9123.

Chen X, Levine L, Kwok PY. 1999. Fluorescence polarization inhomogeneous nucleic acid analysis. Genome Res 9:492–498.

Ferris BG. 1978. Epidemiology Standardization Project (AmericanThoracic Society). Am Rev Respir Dis 118:1–120.

Hanis CL, Hewett-Emmett D, Bertin TK, Schull WJ. 1991. Originsof U.S. Hispanics. Implications for diabetes. Diabetes Care14:618–627.

Hankinson JL, Odencrantz JR, Fedan KB. 1999. Spirometricreference values from a sample of the general U.S. population.Am J Respir Crit Care Med 159:179–187.

Hisnanick JJ, Coddington DA, Gergen PJ. 1994. Trends in asthma-related admissions among American Indian and Alaskan nativechildren from 1979 to 1989. Universal health care in the face ofpoverty. Arch Pediatr Adolesc Med 148:357–363.

Homa DM, Mannino DM, Lara M. 2000. Asthma mortality in U.S.Hispanics of Mexican, Puerto Rican, and Cuban heritage, 1990–1995. Am J Respir Crit Care Med 161:504–509.

Long JC. 1991. The genetic structure of admixed populations.Genetics 127:417–428.

McKeigue PM. 1998. Mapping genes that underlie ethnicdifferences in disease risk: methods for detecting linkage inadmixed populations, by conditioning on parental admixture.Am J Hum Genet 63:241–251.

McKeigue PM. 2005. Prospects for admixture mapping of complextraits. Am J Hum Genet 76:1–7.

McKeigue PM, Carpenter JR, Parra EJ, Shriver MD. 2000.Estimation of admixture and detection of linkage in admixedpopulations by a Bayesian approach: application to African-American populations. Ann Hum Genet 64:171–186.

Parra EJ, Kittles RA, Argyropoulos G, Pfaff CL, Hiester K, BonillaC, Sylvester N, Parrish-Gause D, Garvey WT, Jin L, McKeiguePM, Kamboh MI, Ferrell RE, Pollitzer WS, Shriver MD. 2001.Ancestral proportions and admixture dynamics ingeographically defined African Americans living in SouthCarolina. Am J Phys Anthropol 114:18–29.

Patterson N, Hattangadi N, Lane B, Lohmueller KE,Hafler DA, Oksenberg JR, Hauser SL, Smith MW, O’Brien SJ,Altshuler D, Daly MJ, Reich D. 2004. Methods for high-densityadmixture mapping of disease genes. Am J Hum Genet 74:979–1000.

Pritchard JK, Rosenberg NA. 1999. Use of unlinked geneticmarkers to detect population stratification in associationstudies. Am J Hum Genet 65:220–228.

Risch N. 1992. Mapping genes for complex disease usingassociation studies with recently admixed populations. Am JHum Genet Suppl 51:13.

Risch N, Burchard E, Ziv E, Tang H. 2002. Categorization ofhumans in biomedical research: genes, race and disease.Genome Biol 3:1–12.

Rosenberg NA, Pritchard JK, Weber JL, Cann HM, Kidd KK,Zhivotovsky LA, Feldman MW. 2002. Genetic structure ofhuman populations. Science 298:2381–2385.

Rouse I. 1993. The Tainos: the rise and decline of thepeople who greeted Columbus. New Haven: Yale UniversityPress.

Shriver MD, Parra EJ, Dios S, Bonilla C, Norton H, Jovel C, Pfaff C,Jones C, Massac A, Cameron N, Baron A, Jackson T,Argyropoulos G, Jin L, Hoggart CJ, McKeigue PM, Kittles RA.2003. Skin pigmentation, biogeographical ancestry andadmixture mapping. Hum Genet 112:387–399.

Smith MW, Patterson N, Lautenberger JA, Truelove AL,McDonald GJ, Waliszewska A, Kessing BD, Malasky MJ, ScafeC, Le E, De Jager PL, Mignault AA, Yi Z, De The G, Essex M,Sankale JL, Moore JH, Poku K, Phair JP, Goedert JJ, Vlahov D,Williams SM, Tishkoff SA, Winkler CA, De La Vega FM,Woodage T, Sninsky JJ, Hafler DA, Altshuler D, Gilbert DA,O’Brien SJ, Reich D. 2004. A high-density admixture map fordisease gene discovery in african americans. Am J Hum Genet74:1001–1013.

Standardization of Spirometry 1995. (1994 Update). Am ThoracSoc Am J Respir Crit Care Med 152:1107–1136.

Stout JW, Sullivan M, Liu LL, Grossman DC. 1999. Asthmaprevalence among American Indian and Alaska Nativechildren. Public Health Rep 114:257–261.

Zheng C, Elston RC. 1999. Multipoint linkage disequilibriummapping with particular reference to the African-Americanpopulation. Genet Epidemiol 17:79–101.

Ziv E, Burchard EG. 2003. Human population structure andgenetic association studies. Pharmacogenomics 4:431–441.

Salari et al.86