ETHNICITY AND HEALTH

Cancer incidence rates among South Asiansin four geographic regions India SingaporeUK and USTanuja Rastogi1 Susan Devesa1 Punam Mangtani2 Aleyamma Mathew3 Nicola Cooper4

Roy Kao5 and Rashmi Sinha1

Accepted 1 October 2007

Background Data are limited regarding cancer incidence among Indians residingin different geographic regions around the world Examining suchrates may provide us with insights into future aetiological researchpossibilities as well as screening and prevention

Methods Incidence rates for all cancers combined and 19 specific cancerswere obtained for India from Globocan 2002 for Indians inSingapore from Cancer Incidence in Five Continents (VIII) andfrom national data sources for South Asians (SA) in the UnitedKingdom (UK) and for Asian IndiansPakistanis (AIP) and whitesin the United States (US)

Results We observed the lowest total cancer incidence rates in India (111and 116 per 100 000 among males and females respectively age-standardized to the 1960 world population) and the highest amongUS whites (362 and 296) Cancer incidence rates among Indiansresiding outside of India were intermediate Singapore (102 and132) UK (173 and 179) and US ranges 152ndash176 and 142ndash164 Asimilar pattern was observed for cancers of the colorectumprostate thyroid pancreas lung breast and non-Hodgkin lym-phoma In contrast rates for cancers of the oral cavity oesophaguslarynx and cervix uteri were highest in India Although littlegeographic variability was apparent for stomach cancer incidenceIndians in Singapore had the highest rates compared with anyother region The UK SA and the US AIP appear with adopt thecancer patterns of their host country

Conclusion Variations in environmental exposures such as tobacco use diet andinfection as well as better health care access and knowledge mayexplain some of the observed incidence differences

Keywords Cancer incidence India Asian Indians South Asians SingaporeIndians

Corresponding author Division of Cancer Epidemiology andGenetics National Cancer Institute 6120 Executive Blvd Rm3046 EPS Rockville MD 20852 USAE-mail sinharnihgov

1 Division of Cancer Epidemiology amp Genetics (DCEG) NationalCancer Institute (NCI) NIH DHHS Rockville MD USA

2 London School of Hygiene and Tropical Medicine LondonUK

3 Regional Cancer Center Trivandrum India

4 Office for National Statistics London UK5 Alfred I duPont Hospital for Children Wilmington DE USA

The Author 2007 all rights reserved Advance Access publication 18 December 2007 International Journal of Epidemiology 200837147ndash160

doi101093ijedym219

147

by guest on January 16 2014httpijeoxfordjournalsorg

Dow

nloaded from

IntroductionDramatic international variations exist in cancerincidence rates1 While recent efforts have estimatedcancer incidence rates for India as a whole2 cancerincidence among Asian Indians in different areas ofthe world have not been compared to any greatextent3 Only in Singapore have Asian Indians beenspecifically identified in the cancer registry Seminalobservations on Japanese and other Asian Americansled to epidemiologic research in the West examiningdietary and other factors associated with cancerrisk4ndash6 Our understanding of cancer epidemiologytherefore could be significantly broadened includingrevealing factors requiring investigation by studyingother diverse populations7

Asian Indians represent one-sixth of the worldpopulation with a global total of over one billionand approximately 20 million living outside IndiaThe population of Asian IndiansPakistanis (AIP) inthe United States (US) was 184 million in 2000and rising89 In the United Kingdom (UK) theSouth Asian (SA) population was 227 million in200110

Data are emerging regarding cancer rates amongAsian Indians in the US and the UK11ndash17 We presentestimated national incidence rates in India ratesamong Indians and other SA residing in Singaporethe UK and the US and rates among US whites

MethodsData sources

IndiaNational estimates of cases and rates for India as awhole were obtained from the IARC Globocan 2002data set2 These estimates were derived using data for1993ndash97 from the nine regional population-basedregistries (Ahmedabad Bangalore Chennai DelhiKarunagappally Mumbai Nagpur Poona andTrivandrum) included in Volume VIII of Cancer inFive Continents3 and took into account regional andurbanrural variations in rates

SingaporeData for 1993ndash97 incident cases of cancer and ratesfor the Singapore Cancer Registry were abstractedfrom Cancer Incidence in Five Continents Vol VIII3

The Singapore registry included an Indian subconti-nent population of about 220 000 (includes IndiansPakistanis and Sri Lankans)

United KingdomRates for UK SA (includes Indians Pakistanis andother SA) were derived using cancer registry andpopulation census data The National CancerIntelligence Centre (NCIC) of the Office for National

Statistics receives notification of cancer registrationsfor England and Wales from 10 regional cancerregistries As ethnicity information on cancer casessubmitted to the NCIC including persons of SAorigin is incomplete or unavailable a validatedcomputerized name algorithm South Asian Namesand Group Recognition Algorithm (SANGRA)18 wasused to identify SA cancer cases diagnosed in theperiod 1999ndash2001 SANGRA identifies persons of SAorigin by matching the names in the NCIC data to SAnames contained in its dictionaries The algorithm hasbeen shown to have over 89ndash96 sensitivity and94ndash98 specificity when validated against self-described ethnicity from a number of localities inEngland18 To improve specificity further all namesidentified as being SA were visually inspected andexcluded if they were common to other ethnic groupsAs a check on sensitivity 1000 randomly selectednames were visually inspected to assess the percen-tage not detected by the algorithm Three names outof a random sample of 1032 names that had not beenpicked by SANGRA were found to be of SA originsuggesting that 03 of cases may have been missed

Population estimates of UK SA were available fromthe 2001 Census for England and Wales Theyincluded all persons self-described as Asian or AsianBritish and of specifically East African Asian or SAorigin (nfrac14 227 million)

United StatesRates for AIP were derived using cancer registry andpopulation census data We obtained data for cancercases diagnosed among individuals of self-reportedAIP raceethnicity during the time interval 1999ndash2001from the Surveillance Epidemiology and End ResultsProgram (SEER)19 of the National Cancer Institute(using SEERStat software program) The SEER 11database combines data from cancer registries inAtlanta Connecticut Detroit Hawaii Iowa LosAngeles County New Mexico San Francisco-Oakland San Jose-Monterey Seattle-Puget Soundand Utah As SEER 11 covers a small proportion ofthe total US AIP population [19 (352 573) of the USAIP population (184 million)] we supplementedSEER 11 data with corresponding self-reportedethnicity data provided by state registries inCalifornia20 New York21 New Jersey22 Texas23

Florida24 Pennsylvania25 and Michigan26 Stateregistry data were used instead of SEER data whenthere was an overlap State registry data in conjunc-tion with SEER enabled us to cover 124 million of theUS AIP population

US AIP population data were obtained from the2000 US Census8 The US 2000 census forms allowedpersons to choose more than one ethnicity resultingin two population values (lower and upper estimates)the lower bound represents persons identifyingthemselves as Indian or Pakistani only and the

148 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

by guest on January 16 2014httpijeoxfordjournalsorg

Dow

nloaded from

upper bound these same persons plus those whoselected another ethnicity in addition to Indian orPakistani The number of persons indicating Indian orPakistani alone or in any combination was 153higher than the number indicating Indian orPakistani alone We therefore derived lower andupper estimates for cancer incidence rates for USAIP using cases as numerators and the two popula-tion estimates as denominators Cancer incidencerates for US whites were obtained from SEER 1119

Data analysisUsing data from each population we derived cancerincidence rates for all cancers excluding non-mela-noma skin cancers and for 19 specific primarycancers oral cavity and pharynx oesophagus sto-mach colorectum liver pancreas larynx lung andbronchus breast cervix uteri corpus and uterus nototherwise specified (NOS) ovary prostate urinarybladder kidney and renal pelvis brain and nervoussystem thyroid cancers non-Hodgkin lymphoma(NHL) and leukaemia Incidence rates were age-standardized to the 1960 Segi world population(IARC WHO)2 We deemed rates unstable if thenumber of cancer cases was less than 12 and the ratewas not shown Standard errors based on thenumbers of cases and rates were estimated27 for allthe rates except those for India for which thenumbers are estimated for the entire country andactual counts are not available2

Rate ratios (RR) were calculated comparing inci-dence rates of each population (Singapore IndiansUK SA US AIP upper and lower bounds and USwhites) with rates in India (referent group) for thetotal and for 19 specific primary cancers Age-incidence curves were plotted on semi-logarithmicgraphs using 10-year age-specific rates for each regionexcept India which used the 20-year or larger age-specific rates available from Globocan 20022

ResultsOverall ratesOverall age-adjusted cancer incidence rates amongboth males and females were lowest in India andSingapore Indians intermediate in UK SA and USAIP and highest in US whites (Table 1) Notably therates for all cancers combined were higher amongfemales than males in India and Singapore IndiansOnly in the US were rates distinctly higher amongmales than females among both AIP and whites

Leading cancers in each populationThe five most common cancers among Indian maleswere cancers of the oral cavity and pharynx lungoesophagus larynx and stomach Similarly SingaporeIndian males had high rates of oral cavity and

pharynx lung and stomach cancers but prostateand colorectal cancers replaced oesophagus andlarynx cancers in the top-five malignancies Thepatterns among UK SA and US AIP males weremore similar to US white males than their counter-parts in India or Singapore as the four most commonmalignancies were prostate lung and colorectalcancers and NHL The fifth most common canceramong UK SA and US API was leukaemia in contrastto urinary bladder cancer among US white men

Among females cervix uteri cancer was among thetop-five cancers only in India and Singapore Breastcancer was among the five most frequent cancersdiagnosed in all the populations studied Interestinglyovarian cancer was one of the top-five cancers amongall four Indian populations investigated but notamong US white females Oral cavity and pharynxcancer was among the top five for Indian females andSingapore Indians while oesophagus cancer was onlyin the top five among Indian females In contrastcolorectal cancer was among the five leading cancersin all groups except Indian females Corpus utericancer was among the top-five cancers in UK SA USAIP and US whites

Comparison of rates across populationsAmong males the rate for all cancers combined wassimilar in India and among Indians in Singapore(Figure 1 panel A) The overall rates among SA in theUK and the US were 50ndash75 higher and among USwhites more than three times those in India Relativeto rates in India prostate cancer rates were more than20 times higher in US whites more than 10 timeshigher in US AIP seven times higher among UK SAand twice as high among Singapore Indians Thesewere the largest relative differences apparentColorectal cancer rates were almost eight timeshigher in US whites about three times higheramong US AIP and UK SA than in India andalmost twice as high among Singapore IndiansSimilar though somewhat less dramatic gradientswere apparent for urinary bladder kidney and renalpelvis lung and thyroid cancers and leukaemia Incontrast rates for oral cavity and pharynx oesopha-gus and larynx cancers all were highest in India withthe rates among SA living in the UK or the US lowerthan even US white rates Stomach cancer rates werehighest among Singapore Indians and varied littleamong the other population groups

Among females the total cancer rate was lowest inIndia about 26 higher in Singapore 36ndash57 higheramong US AIP 72 higher among UK SA and4180higher among US whites (Figure 1 panel B) Thelargest relative differences were for lung cancer withthe RR relative to India being about three for SingaporeIndians four among UK SA and US AIP and 16 for USwhites Colorectal cancer RRs were around four amongIndians in Singapore the UK and the US and more

CANCER INCIDENCE RATES AMONG SOUTH ASIANS 149

by guest on January 16 2014httpijeoxfordjournalsorg

Dow

nloaded from

Table 1 Cancer incidence in India Singapore Indians UK SA US Asian Indians and Pakistanis and US whites

Indiaa 1993ndash1997 Singapore Indians 1993ndash1997 UK SA 1999ndash2001 US AP 1999ndash2001 US whites 1999ndash2001

Counts Ratesb Counts Ratesb Standard Counts Ratesb Standard Counts Ratesb and Standard errors Counts Ratesb Standard

Cancererrorc errorc Lower

bounddStandard

errorcUpper

bounddStandard

errorcerrorc

A Males

All cancers 404 309 990 649 1015 40 4919 1729 25 2508 1516 30 1758 35 205 440 3624 08

Oral cavity and pharynx 92 808 229 67 106 13 176 61 05 105 57 06 66 06 5861 107 01

Oesophagus 29 652 76 20 32 07 93 34 04 30 20 04 24 04 2925 51 01

Stomach 22 650 57 57 90 12 180 64 05 74 46 05 54 06 3842 64 01

Colon and rectum 19 508 47 56 83 11 484 173 08 230 130 09 150 10 22 211 371 02

Liver 9 153 23 50 79 11 186 66 05 68 42 05 49 06 2764 43 01

Pancreas 5 711 14 16 23 06 162 60 05 57 35 05 41 05 4474 76 01

Larynx 24 216 62 33 50 09 73 27 03 29 19 04 23 04 2441 45 01

Lung and bronchus 35 495 90 68 100 12 617 224 09 252 168 11 195 12 27 425 471 03

Prostate 16 789 46 65 99 12 891 337 11 696 474 18 549 21 62 398 1131 05

Urinary bladder 12 444 32 33 54 09 186 70 05 105 68 07 79 08 14 170 231 02

Kidney and renal pelvis 4 738 12 17 28 07 128 44 04 82 47 05 54 06 6268 114 01

Brain and nervous system 12 150 26 14 27 07 169 54 04 85 42 05 49 05 3278 65 01

Thyroid 4 361 10 ndashf ndash ndash 38 12 02 52 20 03 23 03 1779 33 01

Non-Hodgkin lymphoma 13 900 32 19 31 07 347 113 06 156 83 07 96 08 9272 163 02

Leukaemia 15 062 31 28 49 09 282 93 06 141 74 06 86 07 6153 114 01

150

INT

ER

NA

TIO

NA

LJO

UR

NA

LO

FE

PID

EM

IOL

OG

Y

by guest on January 16 2014 httpijeoxfordjournalsorg Downloaded from

B Females

All cancers 447 592 1044 554 1317 56 5119 1792 25 2464 1422 29 1644 33 198 585 2956 07

Oral cavity and pharynx 39 849 96 30 77 14 174 61 05 62 34 04 39 05 2914 42 01

Oesophagus 20 805 51 ndash ndash ndash 86 33 04 28 19 04 22 04 3842 12 00

Stomach 11 743 28 23 60 13 71 26 03 44 28 04 32 05 2423 30 01

Colon and rectum 13 555 32 53 149 20 333 121 07 168 102 08 118 09 21 842 266 02

Liver 4 477 11 ndash ndash ndash 72 27 03 25 17 03 20 04 1371 15 00

Pancreas 3 506 08 ndash ndash ndash 102 39 04 35 23 04 27 05 4485 55 01

Larynx 3 157 08 ndash ndash ndash 11 04 01 ndash ndash ndash ndash ndash 660 11 00

Lung and bronchus 8 046 20 17 54 13 189 73 05 113 73 07 85 08 23 495 330 02

Breast 82 951 191 167 367 28 1894 646 15 918 506 17 582 19 64 645 1017 04

Cervix uteri 132 082 307 35 82 14 145 46 04 81 47 05 54 06 3926 70 01

Corpus and uterus NOSe 6 937 17 31 69 12 265 98 06 139 83 07 96 08 11 786 187 02

Ovary 21 146 49 45 92 14 278 95 06 135 73 06 84 07 6820 106 01

Urinary bladder 3 031 07 ndash ndash ndash 49 19 03 35 24 04 28 05 4732 60 01

Kidney and renal Pelvis 2 129 05 ndash ndash ndash 65 23 03 30 18 03 20 04 3828 58 01

Brain and nervous system 7 530 16 ndash ndash ndash 111 39 04 46 25 04 29 04 2468 44 01

Thyroid 8 686 19 22 35 07 120 36 03 130 62 05 71 06 5174 97 01

Non-Hodgkin lymphoma 7 389 17 12 25 07 247 87 06 89 55 06 64 07 8 110 39

Leukaemia 9 778 21 ndash ndash ndash 203 72 05 85 49 05 58 06 4610 72 01

aAs estimated by Globocan 2002 IARC standard errors not calculated because actual national India counts are not availablebRates per 100 000 age-adjusted to the 1960 Segi world populationcStandard error estimated as the ratesquare root(count) (3)dLower and upper bound estimates for US AIP cancer incidence rates are presented since the US 2000 Census provides two values for the US AIP population (Asian Indian orPakistani alone or in combination with another ethnicity)eNot otherwise specifiedfDash indicates statistic not shown rate based on less than 12 cases

CA

NC

ER

INC

IDE

NC

ER

AT

ES

AM

ON

GS

OU

TH

AS

IAN

S151

by guest on January 16 2014 httpijeoxfordjournalsorg Downloaded from

than eight for US whites Rates for cancers of the breastcorpus uteri urinary bladder kidney and renal pelvisand thyroid showed similar although less dramaticgradients with rates lowest in India intermediateamong Indians elsewhere and highest among USwhites As with the rates among males Indian femaleshad the highest rates of oral cavitypharynxand oesophagus cancers The cervix uteri cancer rate

was substantially higher in India than in any otherstudy region The stomach cancer rate among SingaporeIndians was twice that of any other group

Comparison of age-specific incidence ratesDistinct patterns were apparent in the age-specificcurves according to cancer (Figure 2 panels A and B)

Leukemia

Non-Hodgkin Lymphoma

Thyroid

Brain and Nervous System

Kidney and Renal Pelvis

Urinary Bladder

Prostate

Lung and Bronchus

Larynx

Pancreas

Liver

Colon and Rectum

Stomach

Esophagus

Oral Cavity and Pharynx

All Cancers

01 1 10

~

Rate Ratios(referent group India)

Data from India and Singapore (1993ndash1997) UK and USA (1999ndash2001)~ Rates in this group were based on less than 12 cases

Singapore Indians UK South AsiansUSA Asian Indians amp Pakistanis -upperboundUSA Asian Indians amp Pakistanis -lower bound

USA Whites

A Males

Figure 1 Rate ratios of directly age-adjusted (world standard) cancer incidence rates among three Indian groups (residingin Singapore UK or US) and US whites relative to rates in India A Males B Females

152 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

by guest on January 16 2014httpijeoxfordjournalsorg

Dow

nloaded from

In the first pattern US whites had the highest ratesacross most age groups with more pronounceddifferences at older ages for all cancers combinedand cancers of the colorectum pancreas lungbronchus female breast corpus uteri bladder andkidney and renal pelvis The relative difference at

older ages for lung cancer was greater among femalesthan males Also the divergence in female breastcancer rates is clearly evident in post-menopausal butnot pre-menopausal women In contrast for prostateand thyroid cancers the variation in rates betweenpopulations was most pronounced at the middle ages

Leukemia

Non-Hodgkin Lymphoma

Thyroid

Brain and Nervous System

Kidney and Renal Pelvis

Urinary Bladder

Ovary

Corpus and Uterus NOS

Cervix Uteri

Breast

Lung and Bronchus

Larynx

Pancreas

Liver

Colon and Rectum

Stomach

Esophagus

Oral Cavity and Pharynx

All Cancers

01 1 10

~

~~

~

~

~

~

~

Rate Ratios

(referent group India)

Data from India and Singapore (1993ndash1997) UK and USA (1999ndash2001)~Rates in this group were based on less than 12 cases

Singapore IndiansUK South Asians

USA Asian Indians amp Pakistanis-upper bound USA Asian Indians amp Pakistanis-lowerbound

~

USA Whites

B Females

Figure 1 Continued

CANCER INCIDENCE RATES AMONG SOUTH ASIANS 153

by guest on January 16 2014httpijeoxfordjournalsorg

Dow

nloaded from

0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100

0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100

0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100

0 20 40 60 80 100 0 20 40 60 80 100

04

1

10

100

1000

4000

A Males

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

02

1

10

100

1000

2000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

Data from India and Singapore (1993ndash1997) UK and USA (1999ndash2001) Age-specific curves are not shown when age-adjusted rate was

based on less than 12 cases

Rat

e pe

r 10

000

0 pe

rson

-yea

rs

India

Singapore Indians

UK South Asians

USA Asian Indians amp Pakistanis -upper bound

USA Asian Indians amp Pakistanis -lower bound

USA Whites

All Cancers Oral Cavity and Pharynx Esophagus Stomach Colon and Rectum

Rat

e pe

r 10

000

0 pe

rson

-yea

rs

Pancreas Larynx Lung and Bronchus Prostate Urinary Bladder Kidney and Renal Pelvis

Rat

e pe

r100

000

per

son-

year

s

Brain and Nervous System Thyroid

Liver

Age at diagnosis Age at diagnosisLeukemiaNon-Hodgkin Lymphoma

Age at diagnosis Age at diagnosis Age at diagnosis Age at diagnosis

Figure 2 Age-specific incidence rates for specific cancers in India three Indian groups (residing in Singapore UK or US) and US whites A Males B Females

154

INT

ER

NA

TIO

NA

LJO

UR

NA

LO

FE

PID

EM

IOL

OG

Y

by guest on January 16 2014 httpijeoxfordjournalsorg Downloaded from

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100

0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100

0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 10004

1

10

100

1000

4000

B Females

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

01

1

10

100

1000

Rat

e pe

r 10

000

0 pe

rson

-yea

rs

India

SingaporeIndians

UK South Asians

USA AsianIndiansamp Pakistanis - upperbound

USA AsianIndiansamp Pakistanis - lowerbound

USA Whites Data from India and Singapore (1993ndash1997) UK and USA (1999ndash2001) Age-specific curves are not shown when age-adjusted rate wasbased on less than 12 cases

Oral Cavity and Pharynx Esophagus Stomach Colon and Rectum Liver Pancreas

Rat

e pe

r 10

000

0 pe

rson

-yea

rs

Lung and Bronchus Breast Cervix Uteri Corpus and Uterus NOS

Rat

e pe

r 10

000

0 pe

rson

-yea

rs

Brain and Nervous System Thyroid Non-Hodgkin Lymphoma Leukemia

All Cancers

Larynx Ovary Urinary Bladder

Kidney and Renal Pelvis

Age at diagnosis

0 20 40 60 80 100

Age at diagnosis

0 20 40 60 80 100

Age at diagnosis

0 20 40 60 80 100

Age at diagnosis

0 20 40 60 80 100

Age at diagnosis

Age at diagnosis Age at diagnosis

Figure 2 Continued

CA

NC

ER

INC

IDE

NC

ER

AT

ES

AM

ON

GS

OU

TH

AS

IAN

S155

by guest on January 16 2014 httpijeoxfordjournalsorg Downloaded from

Incidence rate patterns among groups remainedrelatively constant across all ages for NHL and braincancers

In a second pattern rates were highest in Indiaacross most or all age groups for cancers of the oralcavitypharynx oesophagus and cervix uteri In athird pattern remarkably little geographic variabilitywas apparent across groups cancers of the stomachlarynx and ovary Childhood peaks for kidney cancerbrain cancer and leukaemia are not apparent in Indiadue to the availability of only grouped data for ages0ndash14 years

In both the RR bar graphs and the age-specificcurves the US AIP rates varied little according toestimates of population at risk (upper and lowerbounds) and had little impact on the overall patternsobserved

DiscussionAmong both males and females the incidence rate forall cancers combined and cancers of the prostatefemale breast corpus uteri colorectum lung urinarybladder and kidney were lowest in India highestamong US whites and intermediate among Indians inSingapore the UK and the US In contrast rates werehighest in India for oral cavitypharynx oesophagusmale larynx and cervix uteri cancers Stomach andovarian cancer rates varied little geographicallyexcept that stomach cancer rates were high inSingapore Indians and the ovarian cancer rate waslow in India

The low rates in India compared with US whites andSA in UK and US may be due partially to under-diagnosis but may also be due to lifestyle andenvironmental factors In India there exists widevariability in dietary patterns physical activity levelsand environmental exposures7 There are uniqueaspects of the diet28 ranging from high intakes ofthe spice turmeric containing curcumin with anti-carcinogenic properties29 to the common practice ofvegetarianism primarily for religious and not healthreasons The higher cancer rates overall observedamong Indians living in the UK and US may be dueto acculturation as seen for other ethnic groups whenthey migrate or to economic changes This patternhas been noted for other US Asian groups56 forexample colorectal cancer rates among JapaneseAmericans approached rates among US whites6

within one generation This is already apparent forbreast cancer incidence the leading cancer in allIndian-SA women outside India13 this has stimulatedsystematic investigation of environmental factorssuch as the potential protective effect of SA dietsagainst breast cancer14 Moreover increases over thepast several decades in breast cancer incidence havebeen observed in Bangalore Chennai MumbaiNagpur and Pune registries These changes mayreflect the effects of rapid transition towards

industrialization and urbanization as well as adoptionof semi-Western diets and lifestyles including child-bearing patterns among higher socioeconomic urbansubgroups30ndash32

The higher incidence of cancers of the oral cavityoesophagus and larynx in India compared with SAgroups outside India probably is due to the use ofmultiple tobacco products which is less common inthe West The majority of these cases in India forexample have been attributable to chewing tobaccoand the common practice of chewing betel quids(with and without tobacco)33ndash37 Decreases in oralcancer in Mumbai have been reported38 In contrastlung cancer rates which are quite modest amongIndians everywhere still are higher in the UK and theUS than in India likely reflecting the differences inthe prevalence of cigarette smoking

Cervical cancer the most common cancer in womenin India is causally associated with exposure to thehuman papilloma virus (HPV)3940 The age-adjustedincidence rates of cervical cancer among the differentregistries in India varied from 109 in Trivandrum to655 in Ambillikai a rural area High literacy rates andhealth-care awareness among women in Trivandrummay have contributed to the lower rates41 Declines incervical cancer rates have been observed in Mumbaiand Chennai the long-standing urban registriesThe lower rates in Singapore Indian UK SA andUS AIP females similarly may reflect better healthknowledge and access to screening includingpap smears and appropriate management of cervicalpre-cancer

There was however little geographic variability instomach cancer incidence with low rates among allstudied groups except Singapore Indians This trend isopposite to what has been observed among USJapanese Korean and Chinese migrants wherestomach cancer rates were lower in these groupscompared with their native countries4243 This may berelated to the lower intake in the UK and the US ofsalty and nitritenitrate-rich foods such as preservedfoods and meats which are associated with gastriccancer incidence44ndash46 In contrast Indians mayincrease their intake of local foods when they moveto Singapore In Singapore the stomach cancer ratesamong Indians were intermediate between those ofthe Chinese (higher) and Malays (lower)3

Ovarian cancer rates varied little among Indians inSingapore the UK and US and US whites The ratefor India as a whole was notably lower althoughwithin India the rates vary considerably with astrong urbanrural gradient3 In fact rates in theurban areas were similar to those among Indians inSingapore the UK and the US The patterns suggestpossible underdiagnosis as well as a potential role ofbehavioural or environmental factors

Prostate cancer rates varied more than 20-foldbetween India and US whites with a pronouncedgradient among Indians in Singapore the UK and

156 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

by guest on January 16 2014httpijeoxfordjournalsorg

Dow

nloaded from

the US All the registry-specific rates in India3 weresubstantially lower than the rates among Indians inthese other areas As yet prostate cancer rates havenot increased in Mumbai47 in contrast to many otherareas of the world48 Part of the variation is likely dueto diagnosis and screening practices but the rolesof environmental behavioural and genetic factors areunknown

This study adds valuable new information on cancerincidence rates among US AIP which until now wasunavailable The US AIP population is large (184million) and the numbers of cases were sufficient toyield fairly precise estimates of the rates High-qualitydata from SEER were supplemented with data fromregistries of states with large US AIP populationsincluding California New York New Jersey TexasFlorida Pennsylvania and Michigan The newly derivedincidence rates among UK SA were obtained from 10regional cancer registries and are nationally represen-tative Data on Indians in Singapore were obtainedfrom Cancer Incidence in Five Continents (VolumeVIII) and thus have met quality standards As cancerrates in the various regions of India vary considerablywe used the Globocan national rates for India that wereestimated using data from the nine population-basedregistries contributing to Cancer Incidence in FiveContinents (Volume VIII) taking into account regionaland urbanrural patterns This should be preferable tousing the simple aggregated data which would bedominated by the data for Mumbai (Bombay) thelargest Indian registry Rates were higher for all cancerscombined colon and rectum lung and bronchusbreast prostate and many other cancers in Mumbaithan in the Globocan total India estimates andconsiderably lower for cervix uteri cancer Althoughthe Mumbai population includes native-Mumbai resi-dents and people who have moved there from all overIndia in varying proportions they are not a nationallyrepresentative sample The Globocan estimates thus aremore likely to reflect the national Indian experienceFurthermore the SA in Singapore the UK and the UShave migrated from different parts of India so it may bemore appropriate to use a nationally representativesample for India until additional information on cancerincidence among the specific migrant groups in thesecountries is available

Due to variability in coding ethnicity among thenational cancer registry programs we were unable toidentify SAIndian cancer cases in regions outsideIndia using the same methodology Though we usedata for India as our referent group we were not ableto limit our analyses in the UK and US to onlyIndians we had to include persons of other SA originbecause the cases were not identified separately ineach of the countries However Indians are themajority of the US AIP population (92 Indian and8 Pakistani) and the largest UK SA group (46Indian 31 Pakistani 12 Bangladeshi and 11other) In the UK data we cannot exclude the

potential for inaccurate identification of SA cases(numerator) with use of the name algorithm how-ever the reliability of this algorithm was shown to behigh18 Even though the cancer rates were estimatedby unavoidably dissimilar methods in the variouscountries the observed cancer patterns are strikingand may provide impetus for various cancer registriesto collect more detailed information by ethnicities Aswell we were unable to present a comparison ofcancer incidence rates for exactly the same timeperiods across groups (India and Singapore 1993ndash97UK SA US AIP US whites 1999ndash2001) Thegeographic variations observed however appear sub-stantially larger than what could be accounted for bytemporal changes

While these ecologic data cannot be used to assesscausal associations they can stimulate further morefocused studies into cancer aetiology Indians inSingapore the UK and the US have come fromdifferent regions of India Future investigationsshould compare rates among the same Indian ethnicsubgroup residing in India and in other geographicregions Detailed studies could also compare lifestyleschanges in rates between first- and second-generationgroups variability in stage of diagnosis betweenpopulations and the impact of screening on cancerrates Cancer registration in India has expandedgreatly since it was established in Mumbai in 1964by the Indian Cancer Society Since 1982 the IndianCouncil for Medical Research has made considerableefforts to expand the national cancer registry pro-gram which now includes 18 of the urban and 1of the rural populations Although the areas andpopulations covered by the registries are smallnonetheless the data have helped us to understandcancer patterns in India and allowed the developmentof Cancer Atlas49 This information is crucial fordeveloping cancer-control programs in India as wellas for international comparisons

In this study we evaluated cancer incidence amongAsian Indian populations living in diverse geographiclocales The total cancer incidence rates were lower inIndia than among Indians in Singapore the UK andthe US as were the rates for many cancers with astrong gradient for several Rates for certain cancerswere higher in India than among Indians livingelsewhere These observations suggest the role ofenvironmental and lifestyle factors as well as possiblediagnostic and screening practice differencesFuture research should focus on these patterns toadvance our understanding of cancer aetiology and todevelop the means of prevention

AcknowledgementsThis research was supported in part by the IntramuralResearch Program of the National Institutes of HealthNational Cancer Institute (NCI) The Division ofCancer Epidemiology (NCI) group appreciates the

CANCER INCIDENCE RATES AMONG SOUTH ASIANS 157

by guest on January 16 2014httpijeoxfordjournalsorg

Dow

nloaded from

efforts of Kenneth Chu (Center to Reduce CancerHealth Disparities NCI) John Lahey (IMS Inc) andLynn Ries (SEER DCCPS Surveillance ResearchProgram Cancer Statistics Branch NCI) We alsoappreciate the contribution of state cancer registriesMark Allen (California Cancer Registry) ColleenMcLaughlin and Serban Negoita (New York StateCancer Registry) Lisa Paddock (New Jersey StateCancer Registry) Paul Betts (Texas Cancer Registry)James Carreira (Florida Cancer Data System) AlanFleegal (Pennsylvania State Cancer Registry) andGeorgia Spivak (Michigan Cancer SurveillanceProgram) From the California Cancer Registry thecollection of cancer incidence data used in this studywas supported by the California Department of HealthServices as part of the statewide cancer reportingprogram mandated by California Health and SafetyCode Section 103885 the National Cancer InstitutersquosSurveillance Epidemiology and End Results Programunder contract N01-PC-35136 awarded to theNorthern California Cancer Center contract N01-PC-35139 awarded to the University of SouthernCalifornia and contract N02-PC-15105 awarded to

the Public Health Institute and the Centers forDisease Control and Preventionrsquos National Programof Cancer Registries under agreement U55CCR921930-02 awarded to the Public HealthInstitute The ideas and opinions expressed hereinare those of the author(s) and endorsement by theState of California Department of Health Servicesthe National Cancer Institute and the Centers forDisease Control and Prevention or their contractorsand subcontractors neither is intended nor shouldbe inferred From the Pennsylvania State CancerRegistry these data were provided by the Bureau ofHealth Statistics Pennsylvania Department of HealthThe Department specifically disclaims responsibilityfor any analyses interpretations or conclusions TheEngland and Wales Cancer Group (Michel ColemanMike Quinn) gratefully acknowledge the work ofcolleagues in the cancer registries in England andWales who collected the original cancer patient datafor collation by the National Cancer Registry at theUK Office for National Statistics

Conflict of interest None declared

References1 Parkin DM Bray F Ferlay J Pisani P Global cancer

statistics 2002 CA Cancer J Clin 20055574ndash1082 Ferlay J Bray F Pisani P Parkin DM GLOBOCAN 2002

Cancer Incidence Mortality and Prevalence WorldwideIARC Cancer Base No 5 version 20 Lyon IARC Press2004

3 Parkin DM Whelan S Ferlay J Teppo L Thomas DBCancer Incidence in Five Continents Volume VIIIIARC Scientific Publication No 155 Lyon FranceInternational Agency for Research on Cancer (IARC)Press 2002

4 Parkin DM International variation Oncogene2004236329ndash40

5 Yu H Harris RE Gao YT Gao R Wynder EL Comparativeepidemiology of cancers of the colon rectum prostateand breast in Shanghai China vs the United States Int JEpidemiol 19912076ndash81

6 Dunn JE Cancer epidemiology in populations of theUnited Statesndashwith emphasis on Hawaii and Californiandashand Japan Cancer Res 1975353240ndash45

7 Rastogi T Hildesheim A Sinha R Opportunities forcancer epidemiology in developing countries Nat RevCancer 20044909ndash17

8 US Census Bureau Census 2000 Summary File 2 (SF2)100-Percent Data using American FactFinderQuickTables Available at httpfactfindercensusgovAccessed June 10 2004 US Department of CommerceEconomics and Statistics Administration UCB (ed) 2004

KEY MESSAGES

International migration studies where the cancer incidence rates of the same genetic populationresiding in different geographic regions are compared have highlighted the role of environment andlifestyle in cancer aetiology

Incidence rates in India for all cancers combined and 19 specific cancers were compared with thoseamong South Asian groups residing in Singapore UK and US and among US whites

Overall cancer incidence rates and rates of colorectal prostate thyroid pancreas lung and breastcancer and NHL were lowest in India and highest in US whites with intermediary rates observed inSouth Asians residing outside India (Singapore UK and US)

In contrast incidence rates were highest in India for cancers of the oral cavity oesophagus larynx andcervix uteri compared with other geographic regions

UK and US SA appear to adopt cancer incidence patterns of their host country Overall differences incancer rates may be attributed to variations in lifestyle factors including tobacco use diet andinfection as well as health care and knowledge

158 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

by guest on January 16 2014httpijeoxfordjournalsorg

Dow

nloaded from

9 Reeves T Bennet C We the people Asians in the UnitedStates Census 2000 Special Reports CENSR-17 2004US Department of Commerce Economics and StatisticsAdministration US Census Bureau Available at httpwwwcensusgovandjournalscancerpdf

10 UK Census Available at httpwwwstatisticsgovukcensusdefaultasp Accessed April 8 2004

11 Lin SS Clarke CA Prehn AW Glaser SL West DWOrsquoMalley CD Survival differences among Asiansubpopulations in the United States after prostatecolorectal breast and cervical carcinomas Cancer2002941175ndash82

12 Jain RV Mills PK Parikh-Patel A Cancer incidence inthe south Asian population of California 1988ndash2000 JCarcinog 2005421ndash34

13 Parikh-Patel A Mills PK Jain RV Breast cancer survivalamong South Asian women in California (United States)Cancer Causes Control 200617267ndash72

14 dos Santos Silva I Mangtani P McCormack V Bhakta DSevak L McMichael A Lifelong vegetarianism and risk ofbreast cancer a population-based case-control studyamong South Asian migrant women living in EnglandInt J Cancer 200299238ndash44

15 Harding S Rosato M Cancer incidence among firstgeneration Scottish Irish West Indian and South Asianmigrants living in England and Wales Ethn Health1999483ndash92

16 Moles DR Fedele S Speight PM Porter SR The unclearrole of ethnicity in health inequalities the scenario oforal cancer incidence and survival in the British SouthAsian population Oral Oncol 200743831ndash34

17 Keegan TH Gomez SL Clarke CA Chan JK Glaser SLRecent trends in breast cancer incidence among 6 Asiangroups in the Greater Bay Area of Northern California IntJ Cancer 20071201324ndash29

18 Nanchahal K Mangtani P Alston M dos Santos Silva IDevelopment and validation of a computerized SouthAsian Names and Group Recognition Algorithm(SANGRA) for use in British health-related studies JPublic Health Med 200123278ndash85

19 National Cancer Institute Surveillance Epidemiologyand End Results (SEER) Program SEERStat DatabaseIncidencendashSEER 11 RegsthornAK Public-Use Nov 2003 Sub(1973ndash2001 varying) Released April 2004 based on theNovember 2003 submission Available at httpwwwseercancergov Accessed June 10 2004 NationalCancer Institute DCCPS Surveillance Research ProgramCancer Statistics Branch

20 California Cancer Registry Analysis Created in July 2004using Statewide Cancer Database Released May 2004California Department of Health Services Cancer ControlBranch Cancer Surveillance Program Sacramento CA2004

21 New York State Cancer Registry Analysis Created in July2004 Albany NY New York State Department of Health2004

22 New Jersey State Cancer Registry Analysis created inAugust 2004 using SEERStat database NJ Incidence byRace (WBO) amp County 1979ndash2001ndash(Use custom Race)V91 Bladder In situ recoded as Inv (Created 2604by XN) Trenton NJ New Jersey Department ofHealth and Senior Services Cancer EpidemiologyServices 2004

23 Texas Cancer Registry Analysis created in July 2004using SEERStat database Texas Cancer Registry1995ndash2001 Incidence File as of 12022003 Austin TXTexas Department of State Health Services CancerEpidemiology and Surveillance Program 2004

24 Florida Cancer Data System Analysis Created in July2003 Miami FL University of Miami Miller School ofMedicine 2004

25 Pennsylvania Department of Health lsquoAnnual CancerIncidence Files 1999ndash2001rsquo Released July 2004Harrisburg PA Pennsylvania Cancer Registry 2004

26 Michigan Resident Cancer Incidence File Includes casesdiagnosed in 1999ndash2001 and processed by the MichiganDepartment of Community Health Vital Records andHealth Data Development Section by 14 November 20032004

27 Miettinen OS Nurminen M Comparative analysis of tworates Stat Med 19854213ndash26

28 Sinha R Anderson DE McDonald SS Greenwald PCancer risk and diet in India J Postgrad Med200349222ndash28

29 Weber WM Hunsaker LA Abcouwer SF Deck LMVander Jagt DL Anti-oxidant activities of curcumin andrelated enones Bioorg Med Chem 2005133811ndash12

30 Yeole BB Kurkure AB An epidemiological assessment ofincreasing incidence and trends in breast cancer inMumbai and other sites in India during the last twodecades Asian Pac J Cancer Prev 2003451ndash56

31 Murthy NS Agarwal UK Chaudhry K Saxena S A studyon time trends in incidence of breast cancerndashIndianscenario Eur J Cancer Care 200616185ndash86

32 Colditz GA Epidemiology and prevention of breastcancer Cancer Epidemiol Biomarkers Prev 200514768ndash72

33 Moore SR Johnson NW Pierce AM Wilson DF Theepidemiology of mouth cancer a review of globalincidence Oral Dis 2000665ndash74

34 Babu KG Oral cancers in India Semin Oncol200128169ndash73

35 Sankaranarayanan R Oral cancer in India an epidemio-logic and clinical review Oral Surg Oral Med Oral Pathol199069325ndash30

36 Rahman M Sakamoto J Fukui T Calculation ofpopulation attributable risk for bidi smoking and oralcancer in south Asia Prev Med 200540510ndash14

37 Jacob BJ Straif K Thomas G Ramadas K Mathew BZhang ZF et al Betel quid without tobacco as a risk factorfor oral precancers Oral Oncol 200440697ndash704

38 Sunny L Yeole BB Hakama M et al Oral cancer inMumbai India a fifteen years perspective with respect toincidence trend and cumulative risk Asian Pac J CancerPrev 20045294ndash300

39 Bekkers RL Massuger LF Bulten J Melchers WJEpidemiological and clinical aspects of human papillomavirus detection in the prevention of cervical cancer RevMed Virol 20041495ndash105

40 Franco EL Schlecht NF Saslow D The epidemiology ofcervical cancer Cancer J 20039348ndash59

41 Murthy NS Chaudhry K Saxena S Trends in cervicalcancer incidencendashIndian scenario Eur J Cancer Prev200514513ndash18

42 Kamineni A Williams MA Schwartz SM Cook LSWeiss NS The incidence of gastric carcinoma in Asian

CANCER INCIDENCE RATES AMONG SOUTH ASIANS 159

by guest on January 16 2014httpijeoxfordjournalsorg

Dow

nloaded from

migrants to the United States and their descendantsCancer Causes Control 19991077ndash83

43 Kolonel LN Nomura AM Hirohata T Hankin JHHinds MW Association of diet and place of birth withstomach cancer incidence in Hawaii Japanese andCaucasians Am J Clin Nutr 1981342478ndash85

44 Tsugane S Sasazuki S Kobayashi M Sasaki S Salt andsalted food intake and subsequent risk of gastric canceramong middle-aged Japanese men and women Br JCancer 200490128ndash34

45 Lee JK Park BJ Yoo KY Ahn YO Dietary factors andstomach cancer a case-control study in Korea Int JEpidemiol 19952433ndash41

46 Ji BT Chow WH Yang G et al Dietary habits andstomach cancer in Shanghai China Int J Cancer199876659ndash64

47 Sunny L Yeole BB Kurkure AP et al Cumulative risktrends in prostate cancer incidence in Mumbai IndiaAsian Pac J Cancer Prev 20045401ndash5

48 Hsing AW Devesa SS Trends and patterns ofprostate cancer what do they suggest Epidemiol Rev2001233ndash13

49 Nandakumar A Gupta PC Gangadharan P VisweswaraRN (eds) Development of an atlas of cancer inIndia First all India Report 2001ndash2002 BangaloreIndia National Cancer Registry Program (ICMR) 2004

Published by Oxford University Press on behalf of the International Epidemiological Association

The Author 2008 all rights reserved Advance Access publication 7 January 2008

International Journal of Epidemiology 200837160ndash161

doi101093ijedym249

Commentary Cancer incidence among AsianIndians in India and abroadRengaswamy Sankaranarayanan

Accepted 19 November 2007

Increased migration is a reality in the world andmigration has social cultural political and economicconsequences impacting the health of migrant com-munities Studying the impact of migration on healthin general on cancer and other chronic diseases inparticular can offer important information on the roleof environmental and ethnic factors in diseaseaetiology Although there are some difficulties indifferentiating the influences of environmental riskfactors from those of genetic racial factors theinformation can be used to develop aetiologicalstudies and cancer control interventions Studies oncancer incidence rates among Japanese in Japan andJapanese and Caucasians in the US revealed that therisks of cancers of the prostate corpus uteri colonthyroid breast ovary and testis among the Japanesemigrants and their descendants were elevatedwhereas those of the stomach oesophagus andcervix uteri were decreased among them in the USas compared with the Japanese in Japan1 Althoughthere have been past studies among Hispanics2

Japanese13 Chinese Filipino Vietnamese Koreanethnicities3 and European migrants4 in the UScancer incidence among Asian Indians living in

other countries has not been widely studied Thearticle by Rastogi and colleagues5 describing thecancer profile among Indians in India and personsof Indian origin in Singapore UK and the US is animportant addition to the literature on changingcancer risks and patterns among migrants

Rastogi and colleagues used the GLOBOCAN esti-mates for India for 1993ndash1997 and cancer incidencedata from the Singapore Cancer Registry for 1993ndash97the data for 1999ndash2001 for the UK were derived fromthe National Cancer Intelligence Centre (NCIC) andthat for the US from Surveillance Epidemiology andEnd Results (SEER) programme for 1999ndash2001 Thusthe comparison periods used are different which isbasically due to the non-availability of reliable datafor corresponding periods Inspite of the data limita-tions due to the possible discrepancies in thecompleteness of the counts of cancer cases and thepopulation estimates for South Asian Indians inthe UK and US and use of estimated data forIndians in India the findings are consistent withexisting knowledge about the importance of poten-tially modifiable environmental and behaviouraldeterminants of cancer risk The comparison bringsforth the importance of cancer registration earlydetection programmes development and access tohealth services and health care coverage in cancercontrol and prevention The higher rates of prostate

Screening Group World Health Organization-InternationalAgency for Research on Cancer 150 cours Albert Thomas69372 Lyon Cedex 08 France E-mail sankariarcfr

160 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

by guest on January 16 2014httpijeoxfordjournalsorg

Dow

nloaded from