Opportunities and Challenges for Health Disparities Research in the

Personal Genome Era

Carlos D. Bustamante Department Biomedical Data Sciences

and Genetics Stanford University School of Medicine

cdbustam@stanford.edu twitter: @cdbustamante

Key Driving Questions

* Why study diverse populations in clinicalgenomics research? What is the potentialscientific gain?

* What have we learned from broadeningrepresentation thus far in GWAS/Mendelian?

* How do we properly design multi-ethnic studies so we maximize the power of discovery andinterpretation?

* How do we modify protocols for recruitment,consent/enrollment process, and RoR in amulti- /trans- ethnic clinical genomic researchsetting? 2

worldwide. It aims to catalogue genetic vari-ants occurring in more than 1% of various populations throughout the world — includ-ing mixed-race North and South Americans as well as diverse populations from Africa, Europe, the Far East and South Asia.

Boosting genomic studies globally will also require initiatives that foster collabo-ration between countries, and enable the transfer of funding and technology beyond China, the United States and the European Union. The Human Heredity and Health in Africa Initiative, for example, is empower-ing local researchers. Supported by the US National Institutes of Health (NIH) and the UK Wellcome Trust in London, the initia-tive is the first major attempt to help Afri-can investigators study the genomic and environmental determinants of common diseases in African populations.

The private and philanthropic sector can also play an important part. The Slim Ini-tiative for Genomic Medicine, a collabora-tion between the Mexican National Human Genome Research Institute and the Broad Institute of the United States, was launched last year and is enabling researchers to study type 2 diabetes and cancer in Latin Ameri-can populations. This project is funded by the charitable foundation of the Mexi-can business magnate and philanthropist, Carlos Slim Helú.

For these efforts to be successful, research-ers and physicians in the participating developing countries cannot simply pro-vide samples. In addition, local expertise, resources and technology centres must be developed so that local populations benefit

directly from home-grown research — as at the BGI (formerly the Beijing Genomics Institute) in Shenzhen,

China. Local researchers will often better understand the history of local popula-tions, such as whether they have recently switched from a rural to an urban lifestyle, and so will have deeper insights into likely environmental effects.

At the same time, medical geneticists working in wealthy nations must include

their own minority and immigrant popu-lations in their studies and develop the tools needed to compare results between popu-lations. Many ‘exome’ sequencing projects (which sequence only

the coding regions of the genome) are moving in this direction. For example, the US National Heart, Lung and Blood Insti-tute aims to sequence the exomes of 7,000 people, roughly half of whom are African-Americans, to identify variants associated with cardiovascular and lung diseases.

PIECING TOGETHER THE MOSAICTo make medical genomics truly global, geneticists need new statistical methods to dissect the contribution of genetic, socio-cultural and environmental factors to both chronic and infectious disease.

Currently, ‘ancestry metrics’ are used to correct for the effect of shared ancestry on the results of association studies. But these methods do not work very well for groups whose genomes are a mosaic of fragments drawn from many different populations. Reference data from the relevant ancestral populations, including historically margin-alized populations such as native Americans and Australian Aborigines, will help geneti-cists to separate spurious from real associa-tions. And such understudied populations

must be properly on board for this to hap-pen. Researchers should gauge local values and concerns, and invest time and money into education and outreach to explain to the people they intend to study, as well as to the general public, why studying global (and local) health is so important.

The Center for Research on Genom-ics and Global Health headed by Charles Rotimi at the US National Human Genome Research Institute in Bethesda, Maryland, is beginning to gather the data and formulate the methods needed to understand the com-plex interplay that creates health disparities among ethnic groups for diseases such as type 2 diabetes, hypertension and obesity in Africa and elsewhere. Meanwhile, the Slim, the Wellcome Trust and the Bill & Melinda Gates Foundation based in Seattle, Wash-ington, have begun to support research in understudied populations. Ultimately, how-ever, global genomics needs the financial support of governments.

One way to encourage researchers to branch out may be for peer reviewers and granting bodies to stress the importance of racial and ethnic diversity in medical genetic studies. The NIH mandated the inclusion of diverse subjects in 1985. In the 26 years since, just 7% of GWAS have included minorities — perhaps because being more inclusive doesn’t win points for grant applicants.

It is tempting to focus on populations that are motivated, organized, medically compliant and otherwise easy to study. But by failing to develop resources, methodolo-gies and incentives for underserved people, we risk perpetuating the health disparities that plague the medical system. Those most in need must not be the last to receive the benefits of genetic research. ■

Carlos D. Bustamante and Francisco M. De La Vega are in the Department of Genetics, Stanford University School of Medicine, Stanford, California 94305, USA. Esteban G. Burchard is at the University of California, San Francisco, San Francisco, California 94143, USA.e-mail: cdbustam@stanford.edu

1. Need, A. C. & Goldstein, D. B. Trends Genet. 25,489–494 (2009).

2. Nature 467, 1026–1027 (2010).3. Manolio, T. A. et al. Nature 461, 747–753 (2009).4. McClellan, J. & King, M. C. Cell 141, 210–217

(2010).5. Gravel, S. et al. Proc. Natl Acad. USA doi:

10.1073/pnas.1019276108 (2011).6. Acuna-Alonzo, V. et al. Hum. Mol. Genet. 19,

2877–2885 (2010). 7. Waters, K. M. et al. PLoS Genet. 6, e1001078

(2010). 8. Adeyemo, A. & Rotimi, C. Public Health Genomics

13, 72–79 (2010).9. Kumar, R. et al. N. Engl. J. Med. 363, 321–330

(2010).10.Yang, J. J. et al. Nature Genet. 43, 237–241 (2011).

The authors declare competing financial interests:for details see go.nature.com/gvtxn4

“Being more inclusive doesn’t win points for grant applicants.”

SOU

RC

E: R

EF. 5

NATURE.COMThe human genome at ten:go.nature.com/ugle41

1 4 J U L Y 2 0 1 1 | V O L 4 7 5 | N A T U R E | 1 6 5

COMMENT

© 2011 Macmillan Publishers Limited. All rights reserved

worldwide. It aims to catalogue genetic vari-ants occurring in more than 1% of various populations throughout the world — includ-ing mixed-race North and South Americans as well as diverse populations from Africa, Europe, the Far East and South Asia.

Boosting genomic studies globally will also require initiatives that foster collabo-ration between countries, and enable the transfer of funding and technology beyond China, the United States and the European Union. The Human Heredity and Health in Africa Initiative, for example, is empower-ing local researchers. Supported by the US National Institutes of Health (NIH) and the UK Wellcome Trust in London, the initia-tive is the first major attempt to help Afri-can investigators study the genomic and environmental determinants of common diseases in African populations.

The private and philanthropic sector can also play an important part. The Slim Ini-tiative for Genomic Medicine, a collabora-tion between the Mexican National Human Genome Research Institute and the Broad Institute of the United States, was launched last year and is enabling researchers to study type 2 diabetes and cancer in Latin Ameri-can populations. This project is funded by the charitable foundation of the Mexi-can business magnate and philanthropist, Carlos Slim Helú.

For these efforts to be successful, research-ers and physicians in the participating developing countries cannot simply pro-vide samples. In addition, local expertise, resources and technology centres must be developed so that local populations benefit

directly from home-grown research — as at the BGI (formerly the Beijing Genomics Institute) in Shenzhen,

China. Local researchers will often better understand the history of local popula-tions, such as whether they have recently switched from a rural to an urban lifestyle, and so will have deeper insights into likely environmental effects.

At the same time, medical geneticists working in wealthy nations must include

their own minority and immigrant popu-lations in their studies and develop the tools needed to compare results between popu-lations. Many ‘exome’ sequencing projects (which sequence only

the coding regions of the genome) are moving in this direction. For example, the US National Heart, Lung and Blood Insti-tute aims to sequence the exomes of 7,000 people, roughly half of whom are African-Americans, to identify variants associated with cardiovascular and lung diseases.

PIECING TOGETHER THE MOSAICTo make medical genomics truly global, geneticists need new statistical methods to dissect the contribution of genetic, socio-cultural and environmental factors to both chronic and infectious disease.

Currently, ‘ancestry metrics’ are used to correct for the effect of shared ancestry on the results of association studies. But these methods do not work very well for groups whose genomes are a mosaic of fragments drawn from many different populations. Reference data from the relevant ancestral populations, including historically margin-alized populations such as native Americans and Australian Aborigines, will help geneti-cists to separate spurious from real associa-tions. And such understudied populations

must be properly on board for this to hap-pen. Researchers should gauge local values and concerns, and invest time and money into education and outreach to explain to the people they intend to study, as well as to the general public, why studying global (and local) health is so important.

The Center for Research on Genom-ics and Global Health headed by Charles Rotimi at the US National Human Genome Research Institute in Bethesda, Maryland, is beginning to gather the data and formulate the methods needed to understand the com-plex interplay that creates health disparities among ethnic groups for diseases such as type 2 diabetes, hypertension and obesity in Africa and elsewhere. Meanwhile, the Slim, the Wellcome Trust and the Bill & Melinda Gates Foundation based in Seattle, Wash-ington, have begun to support research in understudied populations. Ultimately, how-ever, global genomics needs the financial support of governments.

One way to encourage researchers to branch out may be for peer reviewers and granting bodies to stress the importance of racial and ethnic diversity in medical genetic studies. The NIH mandated the inclusion of diverse subjects in 1985. In the 26 years since, just 7% of GWAS have included minorities — perhaps because being more inclusive doesn’t win points for grant applicants.

It is tempting to focus on populations that are motivated, organized, medically compliant and otherwise easy to study. But by failing to develop resources, methodolo-gies and incentives for underserved people, we risk perpetuating the health disparities that plague the medical system. Those most in need must not be the last to receive the benefits of genetic research. ■

Carlos D. Bustamante and Francisco M. De La Vega are in the Department of Genetics, Stanford University School of Medicine, Stanford, California 94305, USA. Esteban G. Burchard is at the University of California, San Francisco, San Francisco, California 94143, USA.e-mail: cdbustam@stanford.edu

1. Need, A. C. & Goldstein, D. B. Trends Genet. 25,489–494 (2009).

2. Nature 467, 1026–1027 (2010).3. Manolio, T. A. et al. Nature 461, 747–753 (2009).4. McClellan, J. & King, M. C. Cell 141, 210–217

(2010).5.

6. Acuna-Alonzo, V. et al. Hum. Mol. Genet. 19, 2877–2885 (2010).

7. Waters, K. M. et al. PLoS Genet. 6, e1001078 (2010).

8. Adeyemo, A. & Rotimi, C. Public Health Genomics13, 72–79 (2010).

9. Kumar, R. et al. N. Engl. J. Med. 363, 321–330 (2010).

10.Yang, J. J. et al. Nature Genet. 43, 237–241 (2011).

The authors declare competing financial interests:for details see go.nature.com/gvtxn4

“Being more inclusive doesn’t win points for grant applicants.”

COMPARING THE UNCOMPARABLEThe rarer a genetic variant is within a population, the less likely it is to be found in all ethnic groups. One hundred people were sampled from each population.

Frequency of variants

Deg

ree

to w

hich

var

iant

s ar

e sh

ared

betw

een

popu

latio

ns (

%)

0

20

40

60

80

100

Rare variants (1%) Common variants (15%)

Europeans* Europeans & Chinese† European & African‡

*Comparison of individuals of European descent in Utah and in Tuscany, Italy. † Han Chinese individuals from Beijing compared with Utah sample ‡ Yoruba individuals from Ibadan, Nigeria, compared with Utah sample.

SOU

RC

E: R

EF. 5

NATURE.COMThe human genome at ten:go.nature.com/ugle41

1 4 J U L Y 2 0 1 1 | V O L 4 7 5 | N A T U R E | 1 6 5

COMMENT

© 2011 Macmillan Publishers Limited. All rights reserved

worldwide. It aims to catalogue genetic vari-ants occurring in more than 1% of various populations throughout the world — includ-ing mixed-race North and South Americans as well as diverse populations from Africa, Europe, the Far East and South Asia.

Boosting genomic studies globally will also require initiatives that foster collabo-ration between countries, and enable the transfer of funding and technology beyond China, the United States and the European Union. The Human Heredity and Health in Africa Initiative, for example, is empower-ing local researchers. Supported by the US National Institutes of Health (NIH) and the UK Wellcome Trust in London, the initia-tive is the first major attempt to help Afri-can investigators study the genomic and environmental determinants of common diseases in African populations.

The private and philanthropic sector can also play an important part. The Slim Ini-tiative for Genomic Medicine, a collabora-tion between the Mexican National Human Genome Research Institute and the Broad Institute of the United States, was launched last year and is enabling researchers to study type 2 diabetes and cancer in Latin Ameri-can populations. This project is funded by the charitable foundation of the Mexi-can business magnate and philanthropist, Carlos Slim Helú.

For these efforts to be successful, research-ers and physicians in the participating developing countries cannot simply pro-vide samples. In addition, local expertise, resources and technology centres must be developed so that local populations benefit

directly from home-grown research — as at the BGI (formerly the Beijing Genomics Institute) in Shenzhen,

China. Local researchers will often better understand the history of local popula-tions, such as whether they have recently switched from a rural to an urban lifestyle, and so will have deeper insights into likely environmental effects.

At the same time, medical geneticists working in wealthy nations must include

their own minority and immigrant popu-lations in their studies and develop the tools needed to compare results between popu-lations. Many ‘exome’ sequencing projects (which sequence only

the coding regions of the genome) are moving in this direction. For example, the US National Heart, Lung and Blood Insti-tute aims to sequence the exomes of 7,000 people, roughly half of whom are African-Americans, to identify variants associated with cardiovascular and lung diseases.

PIECING TOGETHER THE MOSAICTo make medical genomics truly global, geneticists need new statistical methods to dissect the contribution of genetic, socio-cultural and environmental factors to both chronic and infectious disease.

Currently, ‘ancestry metrics’ are used to correct for the effect of shared ancestry on the results of association studies. But these methods do not work very well for groups whose genomes are a mosaic of fragments drawn from many different populations. Reference data from the relevant ancestral populations, including historically margin-alized populations such as native Americans and Australian Aborigines, will help geneti-cists to separate spurious from real associa-tions. And such understudied populations

must be properly on board for this to hap-pen. Researchers should gauge local values and concerns, and invest time and money into education and outreach to explain to the people they intend to study, as well as to the general public, why studying global (and local) health is so important.

The Center for Research on Genom-ics and Global Health headed by Charles Rotimi at the US National Human Genome Research Institute in Bethesda, Maryland, is beginning to gather the data and formulate the methods needed to understand the com-plex interplay that creates health disparities among ethnic groups for diseases such as type 2 diabetes, hypertension and obesity in Africa and elsewhere. Meanwhile, the Slim, the Wellcome Trust and the Bill & Melinda Gates Foundation based in Seattle, Wash-ington, have begun to support research in understudied populations. Ultimately, how-ever, global genomics needs the financial support of governments.

One way to encourage researchers to branch out may be for peer reviewers and granting bodies to stress the importance of racial and ethnic diversity in medical genetic studies. The NIH mandated the inclusion of diverse subjects in 1985. In the 26 years since, just 7% of GWAS have included minorities — perhaps because being more inclusive doesn’t win points for grant applicants.

It is tempting to focus on populations that are motivated, organized, medically compliant and otherwise easy to study. But by failing to develop resources, methodolo-gies and incentives for underserved people, we risk perpetuating the health disparities that plague the medical system. Those most in need must not be the last to receive the benefits of genetic research. ■

Carlos D. Bustamante and Francisco M. De La Vega are in the Department of Genetics, Stanford University School of Medicine, Stanford, California 94305, USA. Esteban G. Burchard is at the University of California, San Francisco, San Francisco, California 94143, USA.e-mail: cdbustam@stanford.edu

1. Need, A. C. & Goldstein, D. B. Trends Genet. 25,489–494 (2009).

2. Nature 467, 1026–1027 (2010).3. Manolio, T. A. et al. Nature 461, 747–753 (2009).4. McClellan, J. & King, M. C. Cell 141, 210–217

(2010).5. Gravel, S. et al. Proc. Natl Acad. USA doi:

10.1073/pnas.1019276108 (2011).6. Acuna-Alonzo, V. et al. Hum. Mol. Genet. 19,

2877–2885 (2010). 7. Waters, K. M. et al. PLoS Genet. 6, e1001078

(2010). 8. Adeyemo, A. & Rotimi, C. Public Health Genomics

13, 72–79 (2010).9. Kumar, R. et al. N. Engl. J. Med. 363, 321–330

(2010).10.Yang, J. J. et al. Nature Genet. 43, 237–241 (2011).

The authors declare competing financial interests:for details see go.nature.com/gvtxn4

“Being more inclusive doesn’t win points for grant applicants.”

COMPARING THE UNCOMPARABLEThe rarer a genetic variant is within a population, the less likely it is to be found in all ethnic groups. One hundred people were sampled from each population.

Frequency of variants

Deg

ree

to w

hich

var

iant

s ar

e sh

ared

betw

een

popu

latio

ns (

%)

0

20

40

60

80

100

Rare variants (1%) Common variants (15%)

Europeans* Europeans & Chinese† European & African‡

*Comparison of individuals of European descent in Utah and in Tuscany, Italy. † Han Chinese individuals from Beijing compared with Utah sample ‡ Yoruba individuals from Ibadan, Nigeria, compared with Utah sample.

SO

UR

CE:

REF

. 5

NATURE.COMThe human genome at ten:go.nature.com/ugle41

COMMENT

© 2011 Macmillan Publishers Limited. All rights reserved

COMPARING THE UNCOMPARABLE The rarer a genetic variant is within a population, the less likely it is to be found in all ethnic groups. One hundred people were sampled from each population.

Europeans* Europeans & Chinese† European & African‡

Deg

ree

to w

hich

var

iant

s ar

e sh

ared

betw

een

popu

latio

ns (

%)

0

20

40

60

80

100

Rare variants (1%) Common variants (15%) Frequency of variants

*Comparison of individuals of European descent in Utah and in Tuscany, Italy. † Han Chinese individuals from Beijing compared with Utah sample ‡ Yoruba individuals from Ibadan, Nigeria, compared with Utah sample.

Bustamante, Burchard, and De La Vega Gravel, S. et al. Proc. Natl Acad. USA doi:

1 4 J U L Y 2 0 1 1 | V O L 4 7 5 | N A T U R E | 1 6 5 10.1073/pnas.1019276108 (2011).

6789:%) =>?) D?F) =>@) ?BB)

/"8;4() @A) D@D) E>BA) ?BB)

6#9%) I) ?AA) ?D=) @BB)

-:(%'9%) A?) =A) I) I)

6&(89:%#) C?A) DBC) DDE) ?BB)

<;$%G) D=@C) DB@>) ==CB) EDBB)

Broadening representation in genomics...

!"#$%&%'$()*%+) 123,) ,-,./0) DBBB)2(';&(#) and collaborators

HLA B*5701 causes Abacavir hypersensitivity in HIV anti-retroviral therapy

Incidence: 3-5% Onset: 4-6 wks after initiation of abacavir therapy Symptoms: Fever, skin rash, fatigue, GI symptoms (nausea, vomiting, diarrhea, abdominal pain), and respiratory tract symptoms (pharyngitis, dyspnea, or cough) Management: Discontinue abacavir Do not re-start abacavir; severe symptoms will recur within hours, including life-threatening hypotension and death

Gujarati (India)Massai (Kenya)

Utah (U.S. whites)Tuscans (Italy)Luhya (Kenya)

Mexican AmericansChinese in Denver

Han ChineseAfrican Americans

Japanese from TokyoYorubans

0 0.05 0.1 0.15 0.2

Early results suggest broadening ethnicity for GWAS works

! Medical genetics continues to suffer from a European bias (Bustamante et al. Nature 2011), although that is slowly changing:

Multiethnic Genomewide Association (MEGA) Array

PAGE-II Custom Content, 42K Exonic Variants, 200K Functional Variants, 40K

(1.7M SNPs)GWAS Scaffold, 360K

African Power Diaspora Scaffold, 700K

Human Core Scaffold, 300K

Human Exome Array, 250K

Lessons Learned thus far… * 1000 Genomes and other population scale studies have demonstrated

“Common variants are rare and shared; Rare variants are common and largely population private”

* Properly powered GWAS studies in understudied populations yieldnovel variants at previously associated genes (e.g., LDLR/PCSK9) and new genes underlying previously studied phenotypes (e.g.,SLC16A11 in T2D for H/L).

* NHGRI efforts in diversifying medical and population genomics haveled to important reagents for multi-ethnic GWAS (e.g., MEGA) andnow sequencing (e.g., Centers for Common and Mendelian Diseases)

* Sequencing based population screening (e.g., CFTR sequencing inCA) yields a broad spectrum of alleles that are rare, many VUSs, andrequire additional clinical data for interpretation.

* Building diversity into CSER will likely yield new opportunities for biology and improve patient care/health outcomes for minority populations.

8

Key challenges ahead: * Should representation in sequencing studies be proportional

(65%, 15%, 10%, 5%, etc.) or stratified (25%, 25%, 25%,etc.)

* VUS rates are higher in non-white populations. Does this pose a challenge to genomic medicine and how do weaddress it?

* Inclusion doesn’t mean just ethnic diversity. How does SES, education, etc. impact enrollment, genomeinterpretation, and ROR?

* New technologies pose a risk to broadening healthdisparities. While overall improvement in health outcomes across populations is expected, rates of improvementcould vary by race, ethnicity, SES, etc. Can we study this directly and develop countermeasures?

9