Geographic distribution of the 20210 G to A prothrombin variant

Thromb Haemost 1998, 79 706-8 © 1998 Schattauer Verlag, Stuttgart

Ropid Communication

Geographie Distribution of the 20210 G

to A Prothrombin Variant

F. R, Rosendaal

, C. J. M Doggen

, A. Zivelin

, V. R. Arruda

, M. Aiach

,

D S. Siscovick

, A. Hillarp

, H. H, Watzke

, F. Bernardi

, A. M. Cumming

, F. E. Preston

,

P. H. Reitsma

From the1 Department of Clinical Epidemiology, Leiden University Medical Centre, Leiden The Netherlands, -2Department of Hematology, Leiden University Medical Centre, Leiden, The Netherlands,

Institute of Thrombosis and Hemostasis, Department of Hematology, Sheba Medical Centre Tel-Hashomer Israel, "Hematology-Hemotherapy Centre, State University of Campinas, Campinas, Brazil,

5Laboratory of Hemostasis, Broussais Hospital, Paris, France,

6Department of Epidemiology, Department of Infernal Medicine, University of Washington, Seattle, USA, 7Department of Clinical Chemistry, Lund University, Malmo, Sweden,

8Department of Hematology and Hemostasiology, University of Vienna, Vienna, Austna, 9Department of Biochemistry and Molecular Biology University of Ferrara, Ferrara, Italy, 10Department of Haematology Manchester Royal Infirmary Manchester, United Kingdom

"Haemophilia and Thrombosis Centre, Royal Hallamshire Hospital, Sheffield, United Kingdom

12Laboratory for Expenmental Infernal Medicine, Academic Medical Centre Amsterdam, The Netherlands

Summary

A vaiiant in prothrombm (clottmg factor II), a G to A tiansition at nucleotide position 20210, has lecently been shown to be associated with the prothiombm plasraa levels and the nsk of both venous and artenal throrabosis The purpose of this study was to mvestigate the prevalence of caniership of this mutation in vanous populations

We combmed data fiom 11 centres m nme countnes, where tests foi this mutation had been performed m groups representmg the general population We calculated an oveiall pievalence estimate, by a piecision weighted method, and, smce the distnbution of the preva-lences did not appear homogeneous, by an unweighled average of the prevalences We exarmned differences in the prevalences by geographical location and ethnic background äs a possible explanation for the heterogeneity

Among a total of 5527 mdividuals who had been tested, 111 hetero-zygous camers of the 2021OA mutation were found The prevalence estimates vaned from 07 to 4 0 between the centres The overall prevalence estimate was 2 0 peicent (CI95 l 4-2 6%) The Variation around the summary estimate appeared more than was expected by chance alone, and this heteiogeneity could be explamed by geographic differences In southern Euiope, the prevalence was 3 0 percent (0195 2 3 to 3 7%), neaily twice äs high äs the prevalence m northern Euiope (l 7%, CI95 l 3 to 2 2%) The piothrombm vanant appeared very rare m mdividuals fiom Asian and Afncan descent

The 20210A prothiombm vaiiant is a common abnormality, with a prevalence of carnership between one and four percent It is more common m southern than m northern Europe Smce this distnbution withm Europe is very diffeient to that of anothei prothrombotic muta

Correspondence to Dr F R Rosendaal, Dept of Clinical Epidemiology, Bldg l, CO-P, University Hospital Leiden P 0 Box 9600, NL 2300 RC Lei-den, The Netherlands - Tel +31715264037, FAX +31715248122, e mail Rosendaal@rullf2 leidenuniv nl

tion (factor V Leiden or factor V R506Q), foundei effects are the most likely explanation for the geographical disttibution of both mutations

Introduction

Recently, we leported a mutation m clottmg factor II (prothrombm) that was associated with plasma levels of prothrombm and with the nsk of deep-vem thiombosis [1] The smgle base Substitution (G to A) at position 20210 [2] of the 3' untranslated legion of the prothrombm gene was found in 18 percent of probands of thrombophilic famihes, six peicent of unselected consecutive patients with deep-vem thrombosis and 2 percent of healthy controls [1] Thus, carnership of this mutation led to a three-fold mcreased nsk of venous thrombosis [1]

The mutation also affects the nsk of artenal disease among young women we found a four fold mcreased nsk of myocardial mfarction among camers of the factor II20210A allele [3], while among men the nsk was mcreased l 5-fold [4]

Several recent reports, mamly m the foim of meetmg abstracts and piesentations, have addiessed the venous and artenal nsk mciease m other populations [5-11] One of the pioblems with the studies is that the relatively low prevalence of the mutation m the general population leads to large uncertamties m the usk estimates Moreover, it is possible that the pievalence depends on ethmcity and geography, äs was found for the factor V Leiden mutation

Therefore, we decided to combme our data on the frequency of the factoi II 20210A vanant m the general population, first, to arnve at a summaiy allele frequency estimate, and secondly, to examme whether there is heterogeneity in its distnbution

Methods

At the 11 centres from nme different countnes, testmg was performed for prothiombm 20210 G to A carneiship in healthy mdividuals The ongin of each study group is hsted m Table l Most centres were in Euiope and the Middle East, and ranged fiom Scandmavia to the Mediterranean In addition, we mcluded a study from South-Amenca (Campinas, Brazil) and from North

Rosendaal et al Distribution of PT 20210 G to A

Table l Numbers and ongm of the tested mdividuals per centre

centre ongin of tested mdividuals Europe/Middle Hast. Leiden [l 4] 1760 Tel Aviv [17] Amsterdam [5] Paris [6] Malmo [7] Vienna [8] Ferrara [9] Manchester [10] Sheffield [11] 1081 400 398 282 213 176 164 150

medical Controls (n=646 n-291)1 population Controls

(n=474) blood donors (n-249) Hospital statf(n 100) population controls1

Hospital staff (n 272) population controls (n-13 8) medical controls3

blood donors (n~l 16) medical controls (n=166)4

population controls (n=l 11) newborns (n=102) population controls

blood donors

blood donors (n-96) population controls (n~54)

America

Brazil[18] 522 population controls (n=83) newborns (n-295) hospital staff (n=144)!

United States [3] 384 population controls

' orthopedic surgery patients (n-646) rheumatoid arthntis patients (n~29l)

2 from vanous ethmc groups Ashkenazi North African Middle Eastern (Iraqi and Iraman) Yememte and Ethiopian Jews and Arabs

3 medical checkup centre visitors excludtng those with cardiovascular disease

4 mdividuals who underwent orthopedic surgery (n~142) or gave blood samples for vanous

reasons (n-24) excluding those with a history of cardiovascular disease

5 from three regions with different ethmc background Amazoman Indians (n~83) Caucasians

(n=295 Campinas south-east) and with a mainly African descent (n-144 Bahia north east)

America (Seattle, United States) These latter two centres and the Tel Aviv center also mcluded non-whites, while all other solely or mamly compnsed Caucasian mdividuals

The selection of mdividuals vaned between the vanous studies In most cases, an effort was made to include healthy mdividuals, äs a sample from the population These samples ranged from hospital employees, to blood donors, newborns, and vanous groups of mdividuals visitmg health care providers ('medical control groups') and, fmally formal control groups from the general population (Table 1) In none of the centres, mdividuals were selected m a way that would lead to preferential mclusion of mdividuals with thrombotic disease, while m most centres those with a history of cardiovascular disease were excluded Approximately equal numbers of both sexes were mcluded

12 10 6 4 2 -prevalence ( Br US Mo Ma Sri Am Le Pa Vi Fe TA All

Fig l Prevalence of carnership (m percentage pomts) of prothrombm

2021OA, per centre, m geographical order America US = United States (Seattle), Br = Brazil, northern Europe Mo = Malmo, Ma = Manchester, Sh = Sheffield, Am = Amsterdam, Le = Leiden, southern Europe/Middle-East Pa = Paris, Vi = Vienna, Fe = Ferrara, T A = Tel Aviv The European centres are m order of decreasmg latitude On the nght (All) the overall summary estimate is given (straight average) All estimates are shown with a 95% confidence mterval

All centres except one used a strategy for the detection of the 20210 A allele similar to the one from the original report [1] Leiden, Campinas, Amsterdam, Seattle Malmo, Vienna, and Manchestei used amphfication primers exactly äs descnbed before [1] Tel-Aviv, Paris, and Sheffield used primers that differ from the Originals but that still use the concept of mtroducmg a Hmd III recogmtion site with a mutagemc primer The sequences of these primers and the reaction conditions can be obtamed from the respective authois Each centre reported that their local protocol was robust, which mdicates that pnmer choice is not cntical m the assay for the prothrombm mutation Ferrara used allele specific amphfication (ASA) to discnmmate the prothrombm alleles Two allele specific primers, 5'-CTGGGAGCATTGAAGCTC 3' (nucleotides 20227-20210) and 5' CTGGGAGCATTGAAGCTT 3' (nucleo tides 20227-20210) were used äs reverse primers m combmation with the forward primer 5'GGGAAATATGGCTTCTACA-3' (nucleotides ^20035-20054) The presence of an additional mismatch m the ASA primers mcreased selectivity of the ASA

We categonzed the prevalence of carners of the prothrombm 20210 A allele per centre, and pooled the estimates by two methods first, by calculat-mg an average weighted for the number of carriers per centre This average gives most weight to the more precise estimates, smce the vanance, under Poisson assumptions, depends on the numbei of carneis This is the appropnate method when it is assumed that the 'true' allehc prevalence is the same m all study groups, and that all observed Variation m prevalences is random This precision-weighted average is not the appropnate appioach when the allehc prevalences are not constant We assessed homogeneity of the prevalences by Poisson modelmg of the prevalence rates and comparmg the model with only the constant, to the saturated model with a dummy vanable for centre, with the hkelihood ratio statistic As a second approach we calculated an unweighted average of the prevalence estimates (le, the aveiage with equal weights, which is the straight mean of the prevalence estimates) This unweighted straight average is more appropnate when homogeneity cannot be assumed Prompted by the reports on a north south differencc m the prevalence of another common prothrombotic mutation, factor V Leiden [ 12], wc stratifled the analysis by geogiaphy mto northern and southern European countnes (the latter mcludmg the Middle East)

Foi an analysis by ethmcity, we grouped togethei African Amcncans fiom the United States and Brazil, and Ethiopian Jews from Israel äs mdividuals with an Afncan descent Native-Amencans, le, American Indians fiom north western United States and the Amazon region m Brazil, were grouped together äs mdividuals with an Asian background

Confidence mtervals for mdwidual prevalence estimates were based on exact Poisson hmits [13] and for pooled estimates on the normal approxima tion of the Poisson distribution

Results

In the 11 cenlres, a total of 5527 mdividuals were tested The total number of mdividuals per centre vaned from 150 to 1760 Among these 5527 mdividuals, 111 heterozygous carners of prothrombm 20210A were identified No homozygous mdividuals were found The number of camers vaned from l to 33 per centre, and the prevalence of carnership ranged from 0 7 percent to 4 0 percent (Fig l)

The overall prevalence was 2 3 peicent (CI95 l 9 to 2 6%), when a weighted average was calculated The vanation of the prevalence estimates appeared to be more than would be expected by random Variation around one underlymg mean (LRS = 16 9, p = 0 08) There fore, we calculated the average with equal weights, this straight average of prevalences was 2 0 percent (CI95 l 4 to 2 6%)

There was a clear relationship of the prevalence with the geographic latitude m Europe (mcludmg Israel) in the centres north of 50° N (Malmo, Manchester, Sheffield, Amsterdam, Leiden) there were 45 carriers among 2756 mdividuals tested, yieldmg a precision-weighted prevalence of l 7 percent (CI95 l 3 to 2 2%), m the centres located

Thiomb Haemost 1998, 79. 706-8

south of 50°N (Paris, Vienna, Ferrara, Tel Aviv) we found 55 carriers among a total of 1868, yielding a prevalence of 3.0 percent (CI95 2.3 to 3.7%). So, the prevalence of the prothrombm vanant was nearly twice äs common in the southern äs in the northern regions (χ2 = 9.05, df = l, p = 0.003). As can be seen in the figure, where the centres are shown in geographical order, in each of the four centres south of 50° N the prevalence estimate was higher than in each of the five centies north of50°N.

Three centres included non-Caucasians: Brazil, Seattle and Tel-Aviv, which allowed a separate analysis of individuals with a different ethnic background. Among individuals with an African background, 2 carriers of factor II 2021OA were found among 300 individuals (0/9 m Seattle, 2/144 in Bahia, Brazil, and 0/147 among Ethiopian Jews in Israel). Taken together, this accounts for a prev-alence of carriership of only 0.67 percent (CI95: 0.08 to 2.4%, comparison with Caucasians χ2 =3.02, p = 0.08 ). Among 103 Native Americans, ie, individuals with an Asian background (20 from Seattle, 83 from the Amazon region in Brazil) no carrier was detected (prevalence CI95: 0 to 3.6%, comparison with Caucasians χ2 =2.24, p = 0.13).

Discussion

We have combined the data on the prevalence of carriership of the prothrombin 20210A variant fiom 11 centres, including a total of 5527 healthy individuals among whom 111 carried the vanant. The estimate of the overall prevalence, under the assumption of homogeneity, was 2.3 percent (CI95 1.9 to 2.6%). The stiaight average of the prevalences was 2 percent (0195 1.4-2.6%). We did not find any homozygous individuals This is not surprising, since with an allele frequency of l percent, we would expect homozygotes to occur in only l in 10,000. The assumption of homogeneity did not appear true. Withm Europe, there was an association with geographical location, with highei prevalences in the southern than in the northern countnes. In southem Europe/Middle Hast, the prevalence was 3 0 percent (CBS 2.3 to 3.7%), and m northern Europe it was 1.7 peicent (CI95 1.2 to 2.2%). This distribution m Europe is different to that reported for factor V Leiden, which has a high prevalence in northern Europe, especially in southern Sweden, and a low prevalence in Italy [12]. It has been hypothesized that such geographical differences might be the result of geographical differences in other risk factors, and thus by differences in genetic fitness and selection associated with gene-environment inter-action. The fmdmg of a very different geographical distribution for the prothrombin variant renders these explanations less likely. Probably, founder effects have played a much more important role. Interestingly, both factor V Leiden and prothrombin 2021 OA appear common in the Middle-East [14] and m Cyprus [9,15].

We have combined data from centres that used different cnteria to recruit individuals, which raises questions about the comparability. Most centres excluded individuals with a history of thrombosis, or recruited blood donors and volunteers amongst whom such a history is unlikely. Although this would lead to lower estimates than obtained from a theoretically ideal study group of consecutive newborns, this difference is immatenal. Selection bias leading to overestimation would have occurred if for some reason individuals with thrombotic disease had been preferentially included. This bias did not occur, since the medical controls either comprised patients with disorders that are not associated with prothrombotic mutations (rheumatoid arthritis, orthopedic trauma), or, individuals without a history of thrombosis.

Factor V Leiden is extremely rare among non-Caucasians, which is in accordance with the estimated age of the mutation of 21,000 to 34,000 years [16], ie, occurrmg after the divergence of Africans from non-Afncans and Caucasians from Asians [16]. Although the number of non-Caucasians in our collaborative study was relatively small, we found only 0.67 percent prevalence of carriers among individuals with an African background, and no camers among Native Americans. In another recent report, no carriers were found among 40 Somalians [9]. These findings suggests a low pievalence of prothrombin 20210A among Africans and Asians, äs has been reported also for factor V Leiden [15,16].

We conclude that prothrombin 2021 OA is a common prothrombotic genetic variant, that has a pievalence of carriership between one and four percent. There are geographic differences m this prevalence, with a higher frequency m southern than m northern Europe. In areas with a high prevalence this variant may contribute significantly to the occur-rence of thrombotic disease in the population.

References

1 Poort SR, Rosendaal FR, Reitsma PH, Bertina RM A common genetic Variation in the 3'-untranslated region of the prolhrombm gene is associated with elevated plasma piothrombm levels and an mcrease m venous thrombosis Blood 1996, 88 3698-701 2 Degen SJ, Davie EW Nucleotide sequence of the gene for human prothrombin

Bio-chemtslry 1987,26 6165-77

3 Rosendaal FR, Siscovick DS, Schwanz SM, Psaty BM, Raghunathan TE, Vos HL A common piothrombm vanant (20210 G to A) mcrcases the risk of myocardial mfarclion m youngwomen Blood 1997, 90 1747-50

4 Doggen CJM, Manger Cals V, Bertina RM, Rosendaal FR Interaclion of coagulation de-fects and cardiovascular nsk factors increased risk of myocardial infarction associated wilh faclor V Leiden orprothiombm 2021 OA Circulation 1998 (mprcss)

5 Franco RF, Tup MD, ten Cate H, Prms MH, Kastelem JJP, Reitsma PH The prevalence of the 20210 G>A mutation m the 3'-unlianslated legion of the prothrombin gene in palienK wilh premature coronary aiteiy disease (abstract) Thromb Haemost 1997, Suppl 769-70

6 Alhenc-Gelas M, Le Cam Duchez V, Einmeiich J, Frcbourg T, Fiessmgei JN, Borg JY. Aiach M The 20210A allele of the prothrombin gene is not frequently associated with Ihe tactor V Arg 506 to Gin mulation m Ihrombophihc families Blood 1997, 90 1 7 1 1 7 Hillarp A, Zoller B, Svcnsson PJ, Dahlback B The 20210 A jllele of the prothrombin

gene is a common risk factoi among Swedish outpatients with venfied deep venous thiorabosis Thromb Haemost 1997,78 990-2

8 Watzke HH, Schuttrumpt ], Gral S, Hubei K, Panzer S Increased prevalence of a polymorphism in the gene codmg for human prothrombin m patients with coronary heart disease Thiomb Rcs 1997,87 521-6

9 Ferraresi P, Marchctti G, Legnam C, Cavallan E, Castoldi E, Mascoli F, Ardissmo D, Palareli G, Bemardi F The heterozygous 20210 G/A prolhrombm genolype is associated with early venous Ihrombosis in inhented thiombophilia and is not increased m frequency m artenal disease Artenoscler Thromb Vase Biol 1997 (m prcss)

10 Cummmg AM, Keeney S, Salden A, Bhavnam M, Shwe KH, Hay CRM The prothrombin gene G20210A vanant prevalence m a U K anticoagulant climc population Br J Hacmatol 1997,98 353-5

11 Makns M, Preston FE, Bcauchamp NJ, Hampton KK, Daly ME, Coopei P, Bayhss P, Peake IR Co-mhentance of the 2021 OA allele of the prothrombin gene mcreases the thrombotic nsk in subjecls with famihal thrombophiha (abstract) Thiomb Haemost 1997, Suppl 165

12 Cox MJ, Rces DC, Martmson JJ, Clcgg JB Evidence tor a singlc ongm of factoi V Leiden BiIHaematol 1996,92 1022-5

13 CIBA-GEIGY AG Wissenschaftliche Tabellen Geigy Teilband Statistik Basel, CIBA-GEIGY, 1980

14 Seligsohn U, Zivehn A Thiombophilia äs a multigemc disorder Thromb Haemost 1997, 78 297-301

15 Recs DC, Cox M, Clegg JB World distnbution of factor V Leiden Lancet 1995, 346

ΙΠ3-4

16 Zivehn A, Gnffm JH, Xu X, Pabmger I, Samama M, Conard J, Brennei B, Eldor A, Seligsohn U A single genetic ongin for a common Caucasian nsk factoi foi venous thiombosis Blood 1997, 2 397-402

17 Zivehn A, Rosenbeig N, Dardik R, Amit Y, Kenet G, Kornbrol N, Fndman A, Seligsohn U A low frequency of Iwo genetic nsk factors for thrombosis m Yememte-jews who manifest a decreased mcidence of coronary artery disease (abstract) Thromb Haemost 1997, Suppl 222-3

18 Anuda VR, Von Zuben PM, Chiapanni LC, Annichmo-Bi/acchi JM, Costa FF The mutation Ala677>Val in the methylene letrahydrofolate reduclase gene a nsk factor for arlenal disease and venous thiombosis Thiomb Haemost 1997,77 818-21

Received November 4,1997 Accepled Decembei 18,1997