Associations of protein C and protein S with serum lipid concentrations

Transcription

1 British Journal of Haematology, 1998, 102, Associations of protein C and protein S with serum lipid concentrations P. K. M ACCALLUM, J.A.COOPER, J.MARTIN, D.J.HOWARTH, T.W.MEADE AND G. J. MILLER MRC Epidemiology and Medical Care Unit, Wolfson Institute of Preventive Medicine, St Bartholomew s and the Royal London School of Medicine and Dentistry, London Received 18 November 1997; accepted for publication 16 April 1998 Summary. Difficulties in the laboratory measurement of protein C and protein S levels cause problems in the diagnosis of deficiency states in individual patients and may complicate estimation of the prevalence of these states in the general population. Some difficulties may be due to unappreciated influences affecting the measured levels of proteins C and S. We measured protein C activity and antigen, total and free protein S antigen, and serum total cholesterol, high-density cholesterol and triglyceride in a community-based study of 150 adults (73 male, 77 female), age range years. Participants were identified from the list of a single general practice by stratified random sampling within sex and decade of age. Protein C activity and antigen were strongly associated with serum lipids, mean levels increasing by approximately 0 25 u/ml as total cholesterol and triglyceride concentration each rose from the 5th to 95th centile. Total protein S antigen concentration was associated with total cholesterol, the mean rising by over 0 1 u/ml as total cholesterol increased from the 5th to the 95th centile, whilst a similar rise in triglyceride was associated with an increase in mean free protein S of more than 0 3 u/ml. Overall, physiological variation in total cholesterol and triglyceride concentration was associated with significant variation in protein C and protein S levels, independent of age and sex, suggesting that it is important to take serum lipids into account when investigating patients for protein C or protein S deficiency. Failure to do so may be misleading in some circumstances. Keywords: protein C, protein S, lipids, cholesterol, triglyceride. Protein C and protein S are vitamin-k-dependent naturally occurring anticoagulants whose physiological importance is evident from the association of heterozygous deficiencies of either with an increased risk of venous thromboembolism and of homozygous deficiencies with neonatal purpura fulminans (Lane et al, 1996a, b). A number of genetic and environment factors modulate protein C levels, including protein C promoter region polymorphisms, age, sex, ethnic group, body mass index and, possibly, the use of the combined oral contraceptive pill (Conlan et al, 1993; Tait et al, 1993; Dolan et al, 1994; Spek et al, 1995; Lowe et al, 1997; Woodward et al, 1997). Positive associations of protein C antigen with triglyceride, lowdensity (LDL) cholesterol and high-density (HDL) cholesterol were described in the population-based ARIC study (Conlan et al, 1993), and of protein C activity with total cholesterol and triglyceride in the third Glasgow MONICA Survey (Woodward et al, 1997). Both protein C antigen and activity Correspondence: Dr P. K. MacCallum, MRC Epidemiology and Medical Care Unit, Wolfson Institute of Preventive Medicine, St Bartholomew s and the Royal London School of Medicine and Dentistry, Charterhouse Square, London EC1M 6BQ Blackwell Science Ltd were associated with triglyceride and total cholesterol in patients undergoing coronary angiography for angina pectoris (ECAT Angina Pectoris Study Group, 1993). In a recent population study, cholesterol and triglyceride were the most important predictors of protein C activity (Rodeghiero & Tosetto, 1996). However, in the PROCAM study, no independent association of protein C and triglyceride was found (Assmann & Schulte, 1992). There have been few reports of associations of protein S antigen with serum lipids, although positive associations of functional protein S with cholesterol and triglyceride were described in the third Glasgow MONICA Survey (Woodward et al, 1997). The existence of modulators of protein C and protein S levels may result in difficulties in the diagnosis of deficiency states in individual patients and may complicate estimation of the prevalence of these states in the general population. In a community-based study we have examined the associations of protein C and protein S with serum total cholesterol, HDL cholesterol and triglyceride and have found strong relationships with important practical implications for the diagnosis of protein C and protein S deficiency. 609

2 610 P. K. MacCallum et al SUBJECTS AND METHODS Participants and blood sampling. 166 adults were identified from the list of a general practice in South-East England by stratified random sampling on gender and decade of age. 150 subjects (73 males, 77 females) aged years agreed to take part (90% of the total). The mean age (range) was 55 7 (23 79) years for males and 52 1 (26 80) years for females. They attended the surgery at different times through the day, the majority (63%) during the morning between and hours. Venepunctures were performed using minimal stasis. Blood was collected into Vacutainer tubes containing no anticoagulant (for serum) or M citrate (9:1 vol:vol) and, for measurement of vitamin-k-dependent proteins, kept at room temperature until centrifugation. Serum and citrated plasma samples were centrifuged at room temperature for 10 min at 1500 g and stored at ¹70 C. Laboratory methods. Protein C activity was measured in microtitre plates by chromogenic assay (substrate S-2366, Chromogenix, Sweden) using Protac (Alpha Laboratories, U.K.) as the activator and an endpoint detection method. Protein C antigen was measured by modified Laurell electroimmunoassay (Tripodi & Mannucci, 1991) using antibodies from Dako, U.K. Modified Laurell electroimmunoassay (Tripodi & Mannucci, 1991), using antibodies from Dako, U.K., was also used to determine total and free protein S antigen concentration, the latter following 3 75% PEG 8000 precipitation at 4 C of C4BP-bound protein S. The protein C standard for activity and antigen was obtained from Immuno (U.K.) and standard curves were constructed using neat plasma and dilutions of 80%, 60%, 40%, 20% and 0%. A normal plasma pool obtained from 50 donors was used for the protein S standard and neat plasma and 1/2 and 1/4 dilutions were used for the standard curves. Tests were performed using neat plasma and all standards and tests were performed in duplicate. Total cholesterol, triglyceride and HDL cholesterol were measured using an Ektachem 250 (Johnson and Johnson, U.S.A.). Statistical methods. Variables were log-transformed where necessary to normalize the distributions and associations with age and sex were established. Associations of protein C and protein S with serum total cholesterol, HDL cholesterol and triglyceride were then individually examined after adjustment for age and sex. In the case of free protein S antigen, total protein S was also examined as an explanatory variable because free protein S reflects the molar excess of total protein S over C4BP (Griffin et al, 1992; Garcia de Frutos et al, 1994) and an association might therefore be expected. In the case of protein C, since there were significant univariate associations with more than one lipid variable, age, sex, total cholesterol, triglyceride and HDL cholesterol were all entered as independent explanatory variables in multiple regression models, and regression coefficients (standard errors) and partial correlation coefficients for each lipid variable were obtained. Five of the participants were taking the combined oral contraceptive pill and one was receiving hormone replacement therapy; they were excluded from the analysis although their results were not significantly different from the other females. RESULTS Protein C In females, protein C activity (r ¼ 0 37, P ¼ 0 002) and antigen (r ¼ 0 48, P < ) increased with age, though this was limited to those over 50 years, suggesting a menopausal effect. Thus, mean values in females for protein C activity were 0 89 u/ml under the age of 50 years and 1 04 u/ml for those over 50 years (P ¼ ). In males there was no significant association with age (r ¼ 0 14, P ¼ 0 25). Below the age of 50 years protein C levels tended to be lower in females than males, whereas the reverse was true in the over-50s. Protein C was log normally distributed and mean values are shown in Table I. Table I. Geometric mean [approx. SD] and (95% CI) levels in males and females (results adjusted to age 55 years). Males Females Protein C activity (u/ml) 0 97 [0 17] 0 99 [0 17] ( ) ( ) Protein C antigen (u/ml) 0 97 [0 15] 1 06 [0 21] ( ) ( ) Total protein S antigen (u/ml) 1 02 [0 11] 0 99 [0 16] ( ) ( ) Free protein S antigen (u/ml) 1 12 [0 29] 0 99 [0 30] ( ) ( ) Cholesterol (mmol/l) 5 3 [0 8] 5 74 [0 88] ( ) ( ) Triglyceride (mmol/l) 1 58 [0 76] 1 31 [0 68] ( ) ( ) HDL cholesterol (mmol/l) 1 22 [0 33] 1 48 [0 32] ( ) ( ) We then examined associations of protein C with serum lipids after adjustment for age and sex. Highly significant associations with total cholesterol and triglyceride were seen for both activity (Fig 1) and antigen. In multiple regression models which included age, sex and serum lipids as independent explanatory variables, protein C activity (Fig 2) and antigen were significantly and independently associated with total cholesterol and triglyceride but not with HDL cholesterol. A 1 0 mmol/l rise in serum total cholesterol (study normal range mmol/l) was associated with a 0 07 u/ml increase in both protein C activity and antigen. Thus, as cholesterol rose from the 5th to the 95th centile, mean protein C activity and antigen increased by 0 24 u/ml. A 1 0 mmol/l rise in triglyceride (study normal range mmol/l) was associated with a 0 09 u/ml increase in protein C activity and a 0 07 u/ml increase in protein C antigen. As triglyceride rose

3 Associations of Proteins C and S with Lipids 611 Fig 1. Scatter plots showing associations of protein C activity with (a) total cholesterol (r ¼ 0 57, P < ) and (b) triglyceride (r ¼ 0 55, P < ) after adjustment for age and sex (log/log scale). from the 5th to the 95th centile, mean protein C activity increased by 0 29 u/ml and mean protein C antigen by 0 22 u/ml. This slight difference was not significant (P ¼ 0 45). After adjustment for serum lipids the positive associations of protein C activity and antigen with age in females were no longer significant (r ¼ 0 02, P ¼ 0 84 for activity, r ¼ 0 15, P ¼ 0 21 for antigen). This indicated that the association of protein C with age largely represented confounding by levels of cholesterol and triglyceride. Total protein S antigen Total protein S antigen was normally distributed. Mean values are shown in Table I. There was an increase with age of borderline significance (r ¼ 0 23, P ¼ 0 06) in both males and females. Associations of total protein S antigen with serum lipids were examined after adjustment for age and sex. The only statistically significant association was with total cholesterol (Fig 3), a 1 0 mmol/l rise in total cholesterol being associated with a 0 04 u/ml rise in total protein S. As cholesterol rose from the 5th to the 95th centile, mean total protein S increased by 0 12 u/ml. Fig 2. Multiple regression model showing association of protein C activity with total cholesterol at the mean level (1 4 mmol/l) and 2 SD of triglyceride (log/log scale). For the association with total cholesterol, the partial correlation coefficient ¼ 0 34 and the slope of the regression line (b) ¼ 0 40 (SE 0 09), P < For the association with triglyceride, the partial correlation coefficient ¼ 0 36 and b ¼ 0 15 (0 03), P< Protein C activity can be calculated from the regression equation: log e protein C ¼¹0 65 þ 0 40 log e total cholesterol þ 0 15 log e triglyceride þ 0 09 log e HDL cholesterol ¹ 0 01 sex (if female) ¹ age.

4 612 P. K. MacCallum et al Fig 3. Association of (a) total protein S antigen with total cholesterol (r ¼ 0 23, b ¼ 0 20 (0 07), P ¼ 0 01) and (b) free protein S antigen with triglyceride (r ¼ 0 30, b ¼ 0 16 (0 04), P < ) after adjustment for age and sex (log/log scale). After adjustment for serum lipids, the association of total protein S antigen with age was no longer significant (r ¼ 0 10, P ¼ 0 23). Free protein S antigen The mean value of free protein S antigen (Table I) was significantly higher in males than females (P ¼ 0 007). Levels decreased with age in males (r ¼¹0 29, P ¼ 0 01). In females there was no significant linear effect of age (r ¼ 0 13, P ¼ 0 30). Since free protein S antigen is determined by the molar excess of total protein S over b-chain containing C4BP (Griffin et al, 1992; Garcia de Frutos et al, 1994), we examined the relationship between total and free protein S and found a strong positive association (r ¼ 0 30, P ¼ ) before and after adjustment for age and sex. As total protein S rose from the 5th to the 95th centile there was a 0 28 u/ml increase in mean free protein S antigen. We next examined the associations of free protein S with serum lipids after adjustment for age and sex. Free protein S was significantly associated with triglyceride (Fig 3) but not total or HDL cholesterol. A 1 0 mmol/l rise in triglyceride was associated with a 0 10 u/ml increase in free protein S antigen. As triglyceride rose from the 5th to the 95th centile, there was a 0 33 u/ml increase in mean free protein S antigen. The association between free protein S and triglyceride was independent of the association between total and free protein S antigen since it persisted (r¼ 0 28, P¼ 0 001) after adjustment for total protein S. Similarly, the association between total and free protein S was independent of their individual associations with serum lipids. Therefore the associations of triglyceride and total protein S with free protein S antigen were independent and very similar in overall effect. After adjustment for the association with lipids, the inverse association of free protein S antigen with age in males persisted (r ¼ ¹0 30, P ¼ 0 01). There was a significant difference in the associations of free protein S with age in the two sexes (P for interaction < 0 05). Serum lipids Mean levels are shown in Table I. Serum total cholesterol was log normally distributed. Cholesterol levels were higher in females (P ¼ 0 005) and increased with age (r ¼ 0 46, P < ). The age effect differed by sex, levels increasing more steeply with age in females than in males (P for

5 interaction ¼ 0 01). Serum triglyceride was log normally distributed and was higher in males (P ¼ 0 04). In females there was a significant increase with age (r ¼ 0 47, P < ) but in males there was no age effect (P ¼ 0 79). HDL cholesterol was log normally distributed. The mean level was significantly higher in females than in males (P < ), but did not change with age in either sex. DISCUSSION The prevalence of protein S deficiency in the general population is unclear and estimates for protein C deficiency have varied between studies (Miletich et al, 1987; Gladson et al, 1988; Tait et al, 1995). This uncertainty partly reflects difficulties in the laboratory diagnosis of deficiencies of these naturally occurring anticoagulants. Some of these problems may be due to hitherto unappreciated influences affecting the measured levels of proteins C and S. Difficulties may also occur in the clinical setting where levels of protein C or protein S are usually determined because there is an individual or family history of thrombosis. The problem in classifying individuals in family studies as having heterozygous protein C deficiency or not on the basis of their protein C levels was reported by Allaart et al (1993). They observed that 15% of genetically determined heterozygotes for protein C deficiency had protein C levels within the normal range whereas some of those without the identified genetic defect had reduced levels. To improve the separation of those with and without hereditary protein C deficiency, Tait et al (1993) recommended the use of age and sex related reference ranges for protein C in a study of nearly blood donors. Although an association between triglyceride and protein C levels has not clearly been shown in all population studies (Assmann & Schulte, 1992), data from the VITA project suggest that total cholesterol and triglyceride are the most important determinants of protein C activity (Rodeghiero & Tosetto, 1996). Our study, though smaller, is consistent with this finding and suggests that the previously described association of protein C with age in bigger cross-sectional studies largely reflects confounding by age-related changes in serum lipids. Variation in levels of serum lipids may therefore contribute to the difficulty in diagnosing heterozygous protein C deficiency. Thus, the overall study normal range (i.e. 2 SD from the mean) for protein C activity was u/ml. However, for those with cholesterol levels < 4 5 mmol/l (n ¼ 22) the range was u/ml, whereas for those with cholesterol > 6 5 mmol/l (n ¼ 18) it was u/ml. Associations of free and total protein S antigen include age, sex, pregnancy and use of the oestrogen-containing oral contraceptive pill (Boerger et al, 1987; Malm et al, 1988; Henkens et al, 1995). Determinants of protein S measured by a functional assay have been reported to include, amongst others, age, sex, use of the oral contraceptive pill or hormone replacement therapy (Lowe et al, 1997), total cholesterol, triglyceride and body mass index (Woodward et al, 1997). None of our participants was pregnant and the five taking the combined oral contraceptive pill and one on hormone replacement therapy were excluded from the analysis, Associations of Proteins C and S with Lipids 613 although their results were not significantly different from those of the remaining female participants. Inclusion of body mass index in the multiple regression models did not alter the associations of total and free protein S with serum lipids (data not shown). Total protein S antigen was significantly associated with cholesterol whereas free protein S antigen was associated with triglyceride levels. Total and free protein S antigen levels were strongly associated, as has been previously described (Griffin et al, 1992). The associations of triglyceride and total protein S with free protein S antigen were independent and similar in degree, with mean free protein S increasing by about 30% as triglyceride and total protein S each rose from the 5th to 95th centile. The normal range for males with triglyceride < 2 0 mmol/l was u/ml, whereas it was u/ml for those with triglyceride > 2 0 mmol/l. For females the ranges were u/ml and u/ml when the triglyceride levels were < and > 2 0 mmol/l, respectively. Classification of protein S deficiency is complicated by the distinction between free and bound forms of protein S and by lack of a reliable functional test. Deficiencies of both total and free (type I deficiency) and free protein S alone (type III deficiency) have been described within the same families (Zoller et al, 1995) and may be phenotypic variants of the same genetic disorder due to an age-related increase of total protein S (Simmonds et al, 1997). In our population we found a less striking increase of total protein S with age, which was largely due to confounding by the age-related increase in cholesterol. However, we also observed in males a reduction in free protein S with age which was independent of serum lipids. Conceivably part of the explanation for difficulties in classification of protein S deficiency in some instances could be the different associations of total and free protein S antigen levels with cholesterol and triglyceride as we have observed in our study of individuals without hereditary protein S deficiency. We collected samples from non-fasting subjects and the question arises as to whether the observed associations reflect acute post-prandial effects, a more chronic effect of lipid levels, or measurement artefact. Non-fasting samples may be more physiologically relevant than fasting samples because most time is spent in the non-fasting state and most samples in routine laboratories are collected from nonfasting individuals. Unpublished observations from our laboratory show no effect of a high fat meal on protein C and protein S antigen levels, despite elevation of triglyceride to concentrations as high as 8 mmol/l. This suggests that the associations are chronic rather than either an acute response to dietary fat or measurement artefact. As a consequence, the cross-sectional associations of protein C and protein S with triglyceride may have been underestimated, suggesting that there may be an advantage for diagnostic purposes in measuring levels in the fasting state. Triglyceride did not appear to have any significant spurious effect on the assay measurement of protein C activity because the increases in protein C activity and antigen with triglyceride were very similar and there was no change in the ratio of activity to antigen in the upper compared to the lower tertile of triglyceride (1 0 and 0 97 respectively,

6 614 P. K. MacCallum et al P ¼ 0 35). Likewise, the increases in protein C activity and antigen with cholesterol were virtually identical. Moreover, Tosetto et al (1997) for the VITA project recently provided data suggesting that the associations between lipids and protein C activity measured by chromogenic assay are unlikely to be due to a spurious laboratory effect of triglyceride or cholesterol. Even were such findings to be due to an in vitro effect they would still have important implications for individual patient diagnosis. In conclusion, physiological variations in levels of cholesterol and triglyceride are associated with clinically significant variations in measurements of protein C and protein S. Observed correlations based on single measurements are likely to be underestimates because of the regression dilution effect (MacMahon et al, 1990), reflecting the imprecision in the measurement of the variables due to biological and laboratory variability. The associations we have observed are nevertheless of sufficient size as to potentially influence the classification of protein C or protein S status in individuals. Indeed, failing to take account of such associations may not only be unhelpful but potentially misleading since, for example, a level of protein C activity of 0 80 u/ml would be well within our normal range for those with a cholesterol < 4 5 mmol/l, but below the normal range in those with a cholesterol of > 6 5 mmol/l. Likewise, a free protein S antigen level of 0 55 u/ml would be within the normal range for females with triglyceride < 2 0 mmol/l, but below that for those with triglyceride above this level. In the absence of facilities for genetic testing, which are not routinely available, it is important to be aware of these associations and, as has been suggested for protein C activity (Rodeghiero & Tosetto, 1996), to take them into account when interpreting results of individual testing or family screening in patients with possible deficiencies of protein C or protein S. Finally, we cannot exclude the possibility that the increases observed in protein C and protein S levels with increased serum lipid concentration are a response to changes in other variables rather than a direct effect of lipids. However, the association of these and other vitamin K- dependent proteins with cholesterol and triglyceride (Hoffman et al, 1994; Tracy et al, 1996;Woodwardet al, 1997) suggests some overlap in the regulation of their antigen concentrations and activities by lipids which may be of physiological and clinical significance. REFERENCES Allaart, C.F., Poort, S.R, Rosendaal, F.R., Reitsma, P.H., Bertina, R.M. & Briët, E. (1993) Increased risk of venous thrombosis in carriers of hereditary protein C deficiency defect. Lancet, 341, Assmann, G. & Schulte, H. (1992) The importance of triglycerides: results from the Prospective Cardiovascular Münster (PROCAM) Study. European Journal of Epidemiology, 8, (Suppl. 1), Boerger, L.M., Morris, P.C., Thurnau, G.R., Esmon, C.T. & Comp, P.C. (1987) Oral contraceptives and gender affect protein S status. Blood, 69, Conlan, M.G., Folsom, A.R., Finch, A., Davis, C.E., Sorlie, P. & Wu, K.K. (1993) Correlation of plasma protein C levels with cardiovascular risk factors in middle-aged adults: the Atherosclerosis Risk in Communities (ARIC) Study. Thrombosis and Haemostasis, 70, Dolan, G., Neal, K., Cooper, P., Brown, P. & Preston, F.E. (1994) Protein C, antithrombin III and plasminogen: effect of age, sex and blood group. British Journal of Haematology, 86, ECAT Angina Pectoris Study Group. (1993) ECAT angina pectoris study: baseline associations of haemostatic factors with extent of coronary arteriosclerosis and other coronary risk factors in 3000 patients with angina pectoris undergoing coronary angiography. European Heart Journal, 14, Garcia de Frutos, P., Alim, R.I.M., Härdig, Y., Zöller, B. & Dahlbäck, B. (1994) Differential regulation of a and b chains of C4b-binding protein during acute-phase response resulting in stable plasma levels of free anticoagulant protein S. Blood, 84, Gladson, C.L., Scharrer, I., Hach, V., Beck, K.H. & Griffin, J.H. (1988) The frequency of type I heterozygous protein S and protein C deficiency in 141 unrelated young patients with venous thrombosis. Thrombosis and Haemostasis, 59, Griffin, J.H., Gruber, A. & Fernandez, J.A. (1992) Reevaluation of total, free, and bound protein S and C4b-binding protein levels in plasma anticoagulated with citrate or hirudin. Blood, 79, Henkens, C.M.A., Bom, V.J.J., van der Schaaf, W., Pelsma, P.M., Sibinga, C.Th.S., de Kam, P.J. & van der Meer, J. (1995) Plasma levels of protein S, protein C, and factor X: effects of sex, hormonal state and age. Thrombosis and Haemostasis, 74, Hoffman, C.J., Lawson, W.E., Miller, R.H. & Hultin, M.B. (1994) Correlation of vitamin K-dependent clotting factors with cholesterol and triglycerides in healthy young adults. Arteriosclerosis and Thrombosis, 14, Lane, D.A., Mannucci, P.M., Bauer, K.A., Bertina, R.M., Bochkov, N.P., Boulyjenkov, V., Chandy, M., Dahlbäck, B., Ginter, E.K., Miletich, J.P., Rosendaal, F.R. & Seligsohn, U. (1996a) Inherited thrombophilia: Part 1. Thrombosis and Haemostasis, 76, Lane, D.A., Mannucci, P.M., Bauer, K.A., Bertina, R.M., Bochkov, N.P., Boulyjenkov, V., Chandy, M., Dahlbäck, B., Ginter, E.K., Miletich, J.P., Rosendaal, F.R. & Seligsohn, U. (1996b) Inherited thrombophilia: Part 2. Thrombosis and Haemostasis, 76, Lowe, G.D.O., Rumley, A., Woodward, M., Morrison, C.E., Philippou, H., Lane, D.A. & Tunstall-Pedoe, H. (1997) Epidemiology of coagulation factors, inhibitors and activation markers: the Third Glasgow MONICA Survey. I. Illustrative reference ranges by age, sex and hormone use. British Journal of Haematology, 97, MacMahon, S., Peto, R., Cutler, J., Collins, R., Sorlie, P., Neaton, J., Abbott, R., Godwin, J., Dyer, A. & Stamler, J. (1990) Blood pressure, stroke, and coronary heart disease. Part 1. Prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet, 335, Malm, J., Laurell, M. & Dahlback, B. (1988) Changes in the plasma levels of vitamin K-dependent proteins C and S and of C4b-binding protein during pregnancy and oral contraception. British Journal of Haematology, 68, Miletich, J., Sherman, L. & Broze, G., Jr (1987) Absence of thrombosis in subjects with heterozygous protein C deficiency. New England Journal of Medicine, 317, Rodeghiero, F. & Tosetto, A. (1996) The VITA project: populationbased distributions of protein C, antithrombin III, heparincofactor II and plasminogen: relationship with physiological variables and establishment of reference ranges. Thrombosis and Haemostasis, 76, Simmonds, R.E., Zöller, B., Ireland, H., Thompson, E., de Frutos, P.G., Dahlbäck, B. & Lane, D.A. (1997) Genetic and phenotypic analysis of a large (122-member) protein S-deficient kindred provides an

7 explanation for the familial coexistence of type I and type III plasma phenotypes. Blood, 89, Spek, C.A., Koster, T., Rosendaal, F.R., Bertina, R.M. & Reitsma, P.H. (1995) Genotypic variation in the promoter region of the protein C gene is associated with plasma protein C levels and thrombotic risk. Arteriosclerosis, Thrombosis and Vascular Biology, 15, Tait, R.C., Walker, I.D., Islam, S.I.A.M., McCall, F., Conkie, J.A., Wight, M., Mitchell, R. & Davidson, J.F. (1993) Protein C activity in healthy volunteers: influence of age, sex, smoking and oral contraceptives. Thrombosis and Haemostasis, 70, Tait, R.C., Walker, I.D., Reitsma, P.H., Islam, S.I.A.M., McCall, F., Poort, S.R., Conkie, J.A. & Bertina, R.M. (1995) Prevalence of protein C deficiency in the healthy population. Thrombosis and Haemostasis, 73, Tosetto, A., Gatto, E. & Rodeghiero, F. (1997) Influence of cholesterol, triglycerides and fibrinogen on AT III, PC, HC-II and plasminogen measurement. Thrombosis and Haemostasis, 77, 804. Tracy, R.P., Losito, R., Howard, P., Cushman, M. & Bovill, E.G. (1996) The association of plasma lipids and vitamin K-dependent Associations of Proteins C and S with Lipids 615 coagulation factors. Nutrition, Genetics and Heart Disease (ed. by G. Bray and D. Ryan), pp Louisiana State University Press, Baton Rouge. Tripodi, A., & Mannucci, P.M. (1991) Laboratory diagnosis of proteins C and S. Blood Coagulation and Haemostasis: a Practical Guide (ed. by J. M. Thomson), 4th edn, pp Churchill Livingstone, Edinburgh. Woodward, M., Lowe, G.D.O., Rumley, A., Tunstall-Pedoe, H., Philippou, H., Lane, D.A. & Morrison, C.E. (1997) Epidemiology of coagulation factors, inhibitors and activation markers: The Third Glasgow MONICA Survey. II. Relationships to cardiovascular risk factors and prevalent cardiovascular disease. British Journal of Haematology, 97, Zöller, B., de Frutos, P.G. & Dahlbäck, B. (1995) Evaluation of the relationship between protein S and C4b-binding protein isoforms in hereditary protein S deficiency demonstrating type I and type III deficiencies to be phenotypic variants of the same genetic disease. Blood, 85,