© 1999Schattauci Vcildg Stutlgait ThiombHacmosi 1999,82 1091-6
Hormonal State rather than Age Influences Cut-off Values
of Protein S: Reevaluation of the Thrombotic Risk Associated
with Protein S Deficiency
G. Liberti
1 2, R. M. Bertina
1, F. R, Rosendaal
1 3From the 'Hemostasis and Thrombosis Research Center, Department of Hematology, Leiden University Medical Center, The Netherlands, 2Hematology and BMT Unit University Hospital, Palermo, Italy 3Department of Clinical Epidemiology, Leiden University Medical Center, The Netherlands
Summary
Ot the well known nsk factois toi thiombosis piotem S deticiency is
one of the most difficult to diagnose with ceitamty Rehable estimates
tor the pievalence of piotem S deticiency m the general population aie
not available and the nsk of thiorabosis is a contioveisial issue 1t has
been shown that levels of protem S fluctuate ovei time However
the determmants of low levels of piotem S in the healthy population
aie not cleai Theielore, we evaluated the mfluence of sex, age and
hormonal state on the antigen levels ot piotem S in 474 healthy control
subjects of the Leiden Thiombophiha Study (LETS) In umvaiiate
analysis, sex, age, oral contiaceptive (OC) use and post-menopausal
state all mfluenced piotem S antigen levels In a multivanate model
for the whole sample only menopausal state and OC use lud still an
etfect on the levels of total piotem S and only menopausal state had
an independent effect on the values of fiee protem S On the basis of
this analysis we estabhshed diffeient cut-otf levels for these subgioups
and we re-evaluated in the Leiden Thrombophiha Study the nsk of
thiombosis for mdividuals with low piotem S usmg these different
reference ranges With these specific cut-off pomts, we did not
observe an mcrease in the nsk of thrombosis in patients deficient of
total piotem S (OR l 2,95% CI0 5-2 9) 01 free piotem S (OR l 3,95%
CI 0 5-3 5) When men and women weie analyzed sepaiately, the nsk
m women was l 5 (95% CI 0 4-5 4) and 2 4 (95% CI 0 6-9 2) for total
and free protem S deficiencies, respectively, and there was no
mcrease in thrombotic nsk for men We conclude that it may be helpful
to apply separate cut-off levels in the assessment of protem S levels
This does not, however explam the diffeiences between our results
and those of otheis in the estimate of thrombotic nsk of protem S
deficiency
Introduction
Protein S (PS) is a Vitamin K dependent piotem that circulates in
plasma in two forms, 60% äs a complex with C4b-bmdmg protem
(C4BP), a component of the complement cascade, and 40% äs free PS
(1-3) Its mam anticoagulant pioperties aie its function äs cofactor
of activated piotem C (APC) in the degradation of Factor Va and
Fac-tor Villa, and its activity äs an APC independent mhibiFac-tor of the
prothrombmase and tenase complexes (4-6) When bound to C4BP,
Coirespondencc to Piol F R Rosendaal, Dcpt öl Climcal Epidemiology, Bldg l, CO-P, University Hospital Leiden, P 0 Box 9600, NL-2300 RC Lei-den, The Netheilands - Tel +31715264037, Fax +3171 5248122, E mail Roscndaal@mail mcdlac Icidenuniv nlPS displays no APC cofactoi activity, but still has some of its
APC-mdependent anticoagulant activity (6)
The presence of fiee and bound PS in plasma has made the
diagno-sis of PS deticiency difficult because it is not clear which smgle
meas-uiement is most useful to diagnose this deficiency, nor which leference
lange has to be used (7-9) The diagnosis of PS deficiency is fmthei
comphcaled by a large oveilap between PS levels m noimal and in
heterozygous mdividuals, and by fluctuation of levels of PS ovei time
It has been repoited that the levels of PS aie mfluenced by sex, age,
piegnancy and hoimonal state (10-13)
Additional problems in the diagnosis of PS deficiency are related to
its heteiogeneous moleculai basis and the large spectrum ot mutations
that each may cieate different effects on the expression of its
pheno-type The peicentage of PS deficient patients (äs based on assays in
plasma), in whom a mutation has been found vanes between 50%-70%
(14-16) At present, neither the prevalence of protem S deficiency in the
population is known, nor is theie consensus on its relevance äs a nsk
factor foi thiombosis Accuiate diagnosis is theiefore of cntical
impor-tance
For families with mhented thiombophilia and PS deficiency it has
been reported that in heteiozygous subjects, the probabihty of
expen-encmg a thrombotic event befoie the age of 45, is around 50% (17,18)
In selected groups of thiombophihc patients, PS deficiency has been
reported to occui m l 5%-7% (19-21) In a recently reported Itahan
case-control study the relative nsk of thrombosis associated with low
levels of free PS was 2 4 (9) In our previous lepoit on the results of a
population-based case-control study we could not find an association
between reduced levels of total or free PS and thrombotic nsk (22)
It has been suggested that this might be due to the global cut-off levels
that were used to identify mdividuals with a reduced PS We reanalyzed
our data to assess the thiombotic nsk of mdividuals with a reduced
total and fiee PS usmg diffeient diagnostic cntena and diffeient cut-off
levels We performed this analysis m two phases First, in the contiol
gioup, we assessed the effects of sex, age, oral contraceptive use and
menopausal state on PS levels With these data we constructed cut-off
levels specific for these variables and subsequently compaied the
number of mdividuals with PS levels below these cut-off levels in
patients and controls, thus assessmg the nsk associated with leduced
levels of PS
Subjects, Materials and Methods
SubjeitsandMeaiuiemenfi
Thromh Hacmost 1999; 82: 1093-6
wcre sclectcd from thc Computer filcs of thc Anticoagulation Clinics m Leiden, Amsterdam, and Rotterdam, which monitor coumarin trcalmenl in virlually all patients with venous thrombosis in a defmed geographic area. These cases were 474 unselected and consecutive patienls undcr the age of 70, with a first, objec-tively confirmed episodc of deep vein thrombosis but without an undcrlying malignant disease. The thrombotic patients were asked to find thcir own heallhy control subject according to predefined criteria; in cases m which no conlrol was found, thc partners of other patients were used äs controls.
To assess the effect of sex, age and hormonal state on plasma levels of total and free PS, we measured the levels of free PS m 426 of 474 healthy controls, 172 men and 254 women (forty-cight controls were excluded becausc the meas-urement of free PS was not performcd). Data for total PS were available for all controls, 202 men and 272 women.
Blood was collectcd from the antecubital vcin into 0.106 mol/l trisodium citrate. Plasma was prepared by centrifugation for 10 min at 2000 g at room temperature and stored at -70° C, in l .5 ml volumes. Total PS antigen was measured by polyclonal enzyme-linked immunosorbent assay (ELISA) (23), and free PS was measured directly in plasma by ELISA using two monoclonal anlibodies specific for free PS (Asserachrom frcc Protein S, Stago Diagnostica, Asnieres-sur-Seine, France). The results are expressed in %, where 100% refers to the activity or antigen present in l ml poolcd normal plasma.
Statistical analysis. We made comparisons between groups by analysis of
variance and regression analysis. In the control group we analyzed the effect of age in three age classes (tertiles, ranged 15-37, 37-47, 47-73 years) äs well äs contmuous variable, the effects of sex (indieator variable 0 men, l women), of oral contraceptive users (indieator variable 0 for none OC users, l for OC users) and of menopausal state (indieator variable postmenopausal state 0, pre-menopausal stale l) on the levels of total and free PS by multiple linear regres-sion techniques. The regresregres-sion cocfficicnt obtained shows the increase in thc level of PS per % increase in the factor studicd, adjusted for the effect of other variables in thc model. For the identification of individuals with a rcduced total and free PS wc set the cut-off level of PS al thc 2.5 pcrcentiles of the distribu-tion that we found in our healthy controls. We calculated separate cut-off values for men and women, oral contraceptive users and non users, and pre-menopausal and postpre-menopausal women, all by the same method of pciccntilcs among healthy controls. Odds ratios (OR) for levels bclow thc cut-off values were calculated äs a measure of relative risk in thc Standard unmatchcd fashion. A 95% confidence interval (CI) was constructed according lo Woolf. Gcncral-ly, thc OR estimates the risk of thrombosis whcn a risk factor is present relative to thc reference category.
Results
Variables Affecting PS Levels in Healthy Controls
We observed higher levels of total and free protein S in men com-pared to women: 108.2% vs. 98.1%, (p<.0001) for total PS and 104.5% vs. 91.8%, (p <.0001) for free PS. In women the concentra-tions of total and free PS increased with age, which effect was most striking for total PS. In men, the effects related to age were less clear: although men in the youngest age group had the lowest PS levels (free and total), the levels remained largely unchanged after the age of 37 (total and free PS). The influence of age on the levels of total and free PS are shown in Table l. In a linear regression model, age and sex both had an effect; for each year of age total PS increased 0.5% and free PS 0.3%. We performed the regression analysis also separately for men and women. In women we still observed an effect of age, in men we found little effect both on total and free PS antigen levels. In men we found for total PS, a regression coefficient for age in tertiles of 4.0 (95% CI 0.6 to 7.4) (constant 102); for free PS this was 1.3 (95% CI -2.5 to 5.2) (constant 102).
To assess the effect of hormonal Status in women we looked at menopausal state and use of oral contraceptives. The levels of total and
Table l Values of total and free Protein S in men and women according to
threc age groups
age 16-37 (n) age 37 - 47 (n) age 47 - 73 (n) all (n) total PS* men 1012(209) (33) 1108(168) (46) 109 1 (162) (123) 1082(173) (202) women 8 5 9 ( 1 4 8 ) (84) 9 8 3 ( 1 8 1) (73) 1070(205) (115) 98 1 (20 2) (272) free PS* men 980(22 1) (26) 1095(200) (41) 1045(190) (172) 96 1 (18 4) (103) women 8 4 9 ( 1 6 2 ) (79) 93 3 (22 7) (72) 104 5 (21 5) (172) 948(197) (254) * mean values (SD) m %
free PS in premenopausal women were lower than in postmenopausal women (mean total PS 95.7% vs. 111.2%, free PS 88.7% vs. 99.8%). In women taking oral contraceptives the concentration of total PS was lower than in women not taking oral contraceptives (mean 83.3% vs. 95.7%). For free PS the difference was much less clear (mean 85.1% vs. 88.7%). Multiple regression analysis (simultaneously including age, oral contraception and menopausal state) showed clear effect of meno-pausal state and oral contraceptive use on total PS antigen levels, whereas age no longer had an effect: OC users had -10.1% (95% CI -16.2 to -4.1) lower protein S levels (adjusted for age); premenopausal women had -12.9% (95% CI -19.6 to -6.1) lower levels of protein S than postmenopausal women (adjusted for age and OC use). For free PS antigen levels we only found an effect for menopausal state and not for oral contraceptive use; premenopausal women had -9.8% (95% CI -17.4 to -2.2) lower free PS levels. In these models, age no longer had an effect on either total or free PS levels, i.e. the apparent effect of age is explained fully by the (age-dependent) factors menopausal state and OC
use.
Risk of Thrombosis Associated with Low Protein S L·vels: Comparison of Patients and Controls
On the basis of our analysis we have found sex, menopausal state and use of oral contraceptives to be main determinants of PS levels. We defined separate cut-off levels for these variables, using the 2.5 percen-tiles from the healthy control group äs the lower limit of normal. In this way, for total PS we found a cut-off level in men of 78%, in all women of 62%, in women taking oral contraceptive of 53%, in premenopausal women not taking oral contraceptive (OC) of 66% and in postmeno-pausal women not using OC of 80%; for free protein S the cut-off levels were in men 65%, in all women 54%, in women taking OC 47%, in premenopausal women not taking OC 53 % and in post-menopausal women 58%. A PS deficiency is then defined äs having a PS level below the cut-off level. When we apply these cut-off levels we ob-served 11/474 patients (5 men and 6 women) and 9/474 controls (5 men and 4 women) wilh levels of total PS below the corresponding normal reference ränge. For free PS, we found 9/424 cases (2 men and 7 women) and 7/424 controls (4 men and 3 women). Application of these specific cut-off levels led to a relative risk for total PS deficiency of 1.2 (95% CI: 0.5-3.0) and for free PS deficiency of 1.3 (95% CI: 0.5-3.5). When men and women were analyzed separately, the risk for women was 1.5 (95% CI 0.4-5.4) for total PS deficiency and 2.4 (95% CI 0.6-9.2) for free PS deficiency, while no increase in risk for men for either of these deficiency states was found.
Libciti cl al Cul oll Valucs Ιοί Piotcm S and Risk öl Thiombosis
Subsequently, we applied the stuctest uitena toi total PS deti ciency (total PS <61%) and tree PS deiiciency (tiee PS <45%) Foi total PS we still found more control subjects than patients with levels below this cut off pomt, which was the case foi any conceivable cut off pomt Foi free PS we found 3 cases (l male and 2 temales) and 0 contiol
Discussion
This study confums that levels of total and tree PS die mfluenced by several factors We tound that women, geneially, have lower levels of PS than men Futtheimoie piemenopausal women have the lowest levels, especially when they use oial contracepüves The effect of age that we obseived in women in umvanate analysis was entnely due to then hormonal state, smce the menopausal state and OC use are both stiongly age dependent In men an age effect was obseived only in the youngest age group The mtluence ot OC-use was appaient mamly on the levels of total PS and only m pait on the level of ftee PS This could be due to a direct mfluence of estiogen on the C4BP component, äs others have hypothesized (24, 25) We have not been able to assess if the use ot hormonal leplacement theiapy (postmenopausal oestrogens) could also have some efiect on the level of protein S smce this theiapy was used only m one patient Oui data only m pait support the hypoth-esis that m oldei patients the diagnosis of PS deficiency might be missed, if based on total PS measuiement, due to total PS levels, being age-dependent m contrast with tiee PS levels (7, 26), m our analysis, we found an age-effect on PS levels but only m women, and mediated by hormonal state This mdicates that theie is no mdependent age effecl on PS levels We agiee with others (8, 9) that different noimal values could be used toi men and women However, in women the hormonal Status should be taken mto account to avoid undeidiagnosing or ovei-diagnosmg in some subgroups, äs has also been undeilmed by Henkens et al (10) We cannot confirm, however, that this would lead to differ-ent risk estimates in etiologic studies, while formal studies mdicatmg that this would impiove sensitivity and specificity of diagnostic testmg for PS deficiency aie lacking
When we le evaluated the nsk of thrombosis usmg different new cut-off values, we found nsk estimates veiy similar to those of oui previous analysis usmg global cut-off values (22) (0 7 vs 13 for total PS and l 7 vs 1 3 m free PS), with a possible excepüon in women for free PS deficiency, where we observed an odds ratio of 2 4
Oui data differ from those of Faiom et al (9) who recently have found that PS deficiency is a nsk factoi for venous thrombosis with an estimated relative risk of thrombosis of 2 4 for individuals with a free PS deficiency and an increased risk for men compared to women (OR 5 0 for men and l 6 foi women) Contiary to then Suggestion, application of specif ic or stnctei cut oft values did not explam this dif-ference between studies The difdif-ference between these two population based studies (Faiom's and ours) seems minimal when compaied to recent family studies (27, 28) usmg free PS levels, we found 2[%P5 deficiencies among cases and l 6% among controls (relative risk l 3), whereas Faiom et al (9) found 3 0% free PS deficiencies among cases and l 3% among controls (relative nsk 2 4) Family studies among selected individuals with thrombophiha and a PS deficiency suggest a much higher lelative risk of at least 10 (28) The two case control stud les, when considered togethei, pomt to fiee PS deficiency äs a common abnoimahty (~\% of the population) with a moderate nsk potential (l 3-2 4) The tamily studies, äs do pievious lepoits, suggest a stiong nsk tactor (17, 27, 28) There aie two possibilities to reconcile these contiastmg lesults First, äs has been shown for piotem C deficiency,
the nsk m thiombophihc famihes may be oveiestimated due to jomt piesence of additional risk tactois (29, 30, 31) Second, PS deficiency may be much moie laie than estimated based on plasma PS assays in the case contiol studies, wheie many if not all individuals diagnosed äs PS deficiency m fact had levels m the extreme of the normal distnbu-tion - which may confei a small risk of thiombosis, but is not the phe notypic expiession of a tiue deticiency In the absence of a venfication ot the genetic delect by DNA analysis, or extensive family studies the piopoition of individuals consideied abnormal depends to a large degiee on the cut-off pomt used And so, if the 2 5th peicentile, or the mean minus 2SD aie used, 2 5% of the population will be classified äs abnoimal It should be obvious that these figuies beai no relation to the pievalence of PS deficiency Piobably, if PS deficiency is raie, the likehhood to detect a diffeience between patients and controls is very low, and piobably, true PS deficiencies aie much more raie than the piopoition of the population below the cut oft level, and so, the major ity, it not all of those with abnoimal levels do not cany a defect in the PS gene Only with stnctesl cut off pomts tor fiee PS well below the lowei limit of the laboiatoiy refeience ränge we found 3 patients and 0 controls that fulfilled cntena for free PS deficiency These thiee pa tients might have a true piotem S deficiency and genetic analysis should be able to detect this and confirm that the measurement of free PS is moie useful than the measurement of total PS in detecting PS deficiency (9, 26)
In conclusion, the use of separate cut off levels may well be helpful in the diagnosis of protein S deficiency, but does not explam the lack of excess nsk associated with low levels of PS m oui pievious population based study Both total and free piotem S assays should be mcluded for the evaluation An estimate of the pievalence of PS deficiency com-bmed with DNA analysis in the normal population is necessaiy to assess a correct relative nsk of thiombosis foi tiue PS deficiency pa-tients
References
1 Dahlback B, Stenflo J High molecular weight complex m human plasma between vitamm k dependent piotem S and complement C4b binding piolcin PiocNatlAcadSciUSA 1981 78 25126
2 Dahlback B Protein S and C4b-bmdmg proiem components mvolved m the legulation oi the protein C anticoagulant System Thiomb Haemost
1991,66 49-61
3 Gntfm J, Grubei A, Femandcz A Re evaluation ot total fiee and bound protein S and C4b binding protein levels m plasma anticoagulaled with citrate or hirudin Bloodl992,79 3203-11
4 Walker FJ Piotem S and the regulation of activated protein C Semin Thromb Hemost 1984 10 131-8
5 Shen L, Dahlback B Factoi V and piotem S äs synergistic cofactors to activated protein C m degradation öl factoi Villa J Biol Chcm 1994 269
18735-8
6 Heeb MJ MesteisRM Tans G, Rosmg J GutfinJH Bmdmg of piotem S to factoi Va associated with Inhibition ot prothrombmase that is mde pendent oi activated protein C J Biol Chem 1993, 268 2872 7
7 Zollci B Gaicia de Fuitos P, Dahlback B Evaluation of the relationship between piotem S and C4b binding piotem isoforms in hereditary protein S deficiency demostiatmg type I and type III deficiencies to be phenotypic vanants ot the same genetic disease Blood 1995, 85 3524 31
8 Gau M FalkonL U n u t i a T VallvcC BoncIlM Fontcubeila J Themflu cnce öl low piotem S levels m young women on the delmition of the noi mal lange Thiomb Rcs 1994, 73 149 52
9 Faiom EM Valsecchi C, Palla A Taioh E Ra//an C Mannucci PM Fiee proiem S deiiciency is a nsk lactoi loi venous thiombosis Thromb Haemost 1997 78 13436
Thromb Hacmost 1999; 82: 1093-6
10. Hcnkens CMA, Born VJJ, van der Schaaf W, Pclsma PM, Smit SibingaCT, De Kam PJ, van der Meer J. Plasma levcls of protein S, protein C, and Faclor X: effccts of sex, hormonal state and agc. Thromb Hacmosl 1995; 74: 1271-5.
11. Comp PC, Thurnau GR, Wclsh J, Esmon CT. Functional and Immunologie protein S levcls are decrcased during pregnancy. Blood 1986; 68: 881-5. 12. Boerger LM, Morris PC, Thurnau GR, Esmon CT, Comp PC. Oral
contra-cepüves and gcnder affect protein S Status. Blood 1987; 69: 692-4. 13. Malm J, Laurell M, Dahlbäck B. Changcs in the plasma Icvels of vilamin
k-dependent proteins C and S and C4b-binding protein during pregnancy and oral conlraception. Br J Haematol 1988; 68: 437-43.
14. Gandrille S, Borgel D, Eschwege-Gufflet V, Aillaud M, Dreyfus M, Matheron C, Gaussen P, Abgrall JE, Jude B, Sie P. Identification of 15 dif-ferent candidate causal point mutations and three polymorphisms in 19 pa-tients with protein S deficiency using a scanning method for the analysis of the protein S active gene. Blood 1995; 85: 130-8.
15. Gomcz E, Poort SR, Bertina RM, Reitsma PH. Identification of eight point mutations in protein S deficiency type I- analysis of 15 pedigrees. Thromb Haemost 1995; 73: 750-5.
16. Reitsma PH, Ploos van Amstel H, Bcrtina RM. Three novel mutations in five unrelated subjects with hereditary protein S deficiency type I. J Clin Investl994;93:486-92.
17. Engesser L, Brockmans AW, Briet E, Brommer EJP, Bertina RM. Heredi-tary protein S deficiency: Clinical manifestations. Ann Intern Mcd 1987; 106: 677-82.
18. De Stcfano V, Leone G, Mastrangelo S, Tripodi A, Rodeghiero F, Casta-man G, Barbui T, Finazzi G, Bizzi B, Mannucci PM. Clinical Casta- manifesta-tions and managcment of inherited trombophilia: Retrospectivc analysis and follow-up aller diagnosis of 238 palients with congenital deficiency of anlilhrombin III, protein C, protein S. Thromb Hacmost 1994; 72: 352-8. 19. Ben-Tal 0, Zivelin A, Seligsohn U. The relative frequcncy of hereditary
thrombotic disorders among 107 paticnts wilh thrombophilia in Israel. Thromb Haemost 1989; 61: 50-3.
20. Broekmans AW, van der Linden IK, Jansen-Koeter Y, Bertina RM. Prcva-lence of protein C (PC) and protein S (PS) deficiency in palients with ihromboembolic disease. Thromb Rcs Suppl 1986; 6:135a.
21. Pabinger I, Kyrie PA, Heistingcr M, Eichinger S, Wittman E, Lechner K. The nsk of thromboembolism m asymptomatic patients with protein C and protein S deficiency: a prospcctive cohort study. Thromb Haemost 1994; 71:441-5.
22. Kostcr T, Roscndaal FR, Briet E, van der Meer FJM, Colly LP, Trienekcns PH, Poort SR, Reitsma PH, Vandcnbrouckc JP. Protein C deficiency m a controlled scrics of unselccted outpatients: an inlrequent but clcar risk factor for vcnous thrombosis (Leiden Thrombophilia study). Blood 1995; 85:2756-61.
23. Deutz-Tcrlouw PP, Ballering L, van Wijngaarden A, Bertina RM. Two ELISA's for measuremenl of protein S, and their use in the laboralory diagnosis of protein S deficiency. Clin Chim Acta 1990; 186: 321-34. 24. Carr ME, Stcingold KA, Zekert SL. Protein S Icvels during normal
men-strual cyclc and drug estrogcn therapy for prematurc ovarian failurc. Am J Medicine 1993; 306: 212-7.
25. Bremmc K, Wramsby H, Andersson 0, Wallin M, Blombäck M. Do lowcred factor VII levels at cxtremcly high cndogenous ocstradiol Icvels protect against thrombin formation? Blood Coagul Fibrinolysis 1994; 5: 205-10.
26. Simmonds RE, Zöller B, Ireland H, Thompson E, Garcia de Frutos P, Dahlbäck B, Lane DA. Gcnetic and phcnotypic analysis of a largc (122-mcmber) protein S-deficient kindrcd provides an explanation for ihc familial coexistencc of type I and type III plasma phenotypcs. Blood 1997; 89: 4364-70.
27. Martinclli I, Mannucci PM, De Stefane V, Taioli E, Rossi V, Crosti F, Paciaroni K, Leone G, Faiom EM. Diffcrent risks of thrombosis in four coagulation dcfects associated wilh inherited thrombophilia: a study of 150 familics. Blood 1998; 92: 2353-8.
28. Simmonds RE, Ircland H, Lane DA, Zoll« B, Garcia de Frulos P, Dahlbäck B. Clanficalion of the risk for venous thrombosis associated with heredilary prolcin S deficiency by mvestigation of a large kindrcd with a charactcrizcd gcne dcfect. Ann Intern Med 1998; 128: 8-14.
29. Zoller B, Bernsdotter A, Garcia de Frutos P, Dahlbäck B. Resistance to activated protein C äs an additional gcnetic risk factor in hereditary defi-ciency of protein S. Blood 1995; 12:3518-23.
30. Kocleman BP, van Rumpt D, Hamulyak K, Reitsma PH, Bcrtina RM. Fac-tor V Leiden: an additional risk facFac-tor for thrombosis in protein S dcficicnt families? Thromb Haemost 1995; 74: 580-3.
31. Mustafa S, Mannhalter C, Rintclcn C, Kyrie PA, Knobl P, Lechner K, Pabinger I. Clinical features of thrombophilia in families with gcnc defects in protein C or protein S combincd with lactor V Leiden. Blood Coagul Fibrinolysis 1998; 85-9.
Rcccivcd November 9, 1998 Acccptcd alter resubmission April 26, 1999
