Clinica Chimica Acta 287 (1999) 4557 www.elsevier.

com / locate / clinchim

Interdependence of serum concentrations of vitamin K 1 , vitamin E, lipids, apolipoprotein A 1 , and apolipoprotein B: importance in assessing vitamin status
a, b c Bill E. Cham *, Jeffery L. Smith , David M. Colquhoun
a

The Curacel Institute of Medical Research, 14 /1645 Ipswich Road, Rocklea, Queensland 4106, Australia b Lipid Metabolism Laboratory, Department of Surgery, The University of Queensland, Royal Brisbane Hospital, Herston, Queensland 4029, Australia c Wesley Medical Centre, Auchenower, Queensland 4066, Australia Received 10 March 1999; received in revised form 13 May 1999; accepted 22 May 1999

Abstract Vitamin E (a-tocopherol) and vitamin K 1 (phylloquinone) are fat-soluble vitamins and are important nutrients in health and disease. In this study serum concentrations of vitamin E and vitamin K 1 , lipids and apolipoproteins A 1 and B were measured in neonates, normal and hyperlipidaemic individuals in an attempt to establish their interrelationships. A high degree of correlation was observed between the concentrations of the vitamins and those of lipids and apolipoproteins (r ranged from 0.42 to 0.92; p , 0.001). Stepwise linear regression methods determined that serum concentrations of both vitamin E and vitamin K 1 could best be predicted by using equations excluding lipids but containing only apolipoprotein A 1 and B concentrations. Correlation coefcients between predicted and measured values were 0.89 for serum vitamin E, and 0.83 for serum vitamin K 1 concentrations. To test the validity of the derived formulae, measured and estimated vitamin K 1 and vitamin E concentrations in serum were determined in another group of neonates, normal adults and hypercholesterolemic adults and the comparisons were shown to be very good. These results indicate that the serum levels of both vitamins depend critically on the concentration of the lipoprotein carriers, apolipoproteins A 1 and B. Hence, in order to identify variations in serum vitamin K 1 and vitamin E concentrations, which are independent of variations in carrier concentration, it will be necessary to express these serum vitamins as ratios of vitamins to apolipoprotein A 1 and B carriers. 1999 Elsevier Science B.V. All rights reserved.

*Corresponding author. Tel.: 161-732-744-452; fax: 161-732-744-453. 0009-8981 / 99 / $ see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S0009-8981( 99 )00117-5

46

B.E. Cham et al. / Clinica Chimica Acta 287 (1999) 45 57

Keywords: Vitamin E; Vitamin K 1 ; Apolipoprotein B; Apolipoprotein A 1 ; Cholesterol

1. Introduction The fat-soluble vitamins K 1 and E do not have specic physiological carriers in plasma. They are transported by the plasma lipoproteins and are present in all lipoprotein fractions, namely, chylomicrons, very-low-density lipoproteins (VLDLs), low-density lipoproteins (LDLs) and high-density lipoproteins (HDLs) [13]. There are strong similarities and links between the metabolism of LDL and that of vitamin E. Vitamin E is secreted from the liver in VLDL and is delivered to cells chiey via the high afnity receptor for LDL [4]. As for LDL, cells may also take up the vitamin via a route, which is independent of the LDL receptor [4,5]. Serum concentrations of vitamin E are dependent on plasma lipid concentrations [6]. For example in a hyperlipidaemic state the absolute value of serum vitamin E concentration may be normal and the subject still found to be vitamin E nutritionally decient [7] Thus, the use of a relative measure of plasma vitamin E (ratio of tocopherol to total lipids or to cholesterol) has become essential in evaluating vitamin E nutritional status in individuals or in populations [6,810]. The relation between vitamin K 1 and serum lipids is less clear. Sadowski et al. assessed the vitamin K nutritional status of individuals in different age groups [11]. The authors noted that they obtained different results when they used a ratio of vitamin to lipid rather than absolute concentrations of serum vitamin as the index of nutritional adequacy [11]. Further studies in neonates found that a series of very low absolute concentrations of vitamin K 1 in serum did not reect the vitamin K status [12]; serum lipid values were however not considered. Thus, despite the limited evidence, it seemed probable that the interpretation of serum vitamin K 1 levels might require similar qualications to those applying to vitamin E [10]. The present study has examined the interrelation between the concentration of serum vitamins K 1 and E and some of their carrier lipoprotein components using vitamin E as a currently established marker.

2. Subjects and methods

2.1. Subjects
Blood samples (10 ml) were taken from an antecubital vein from 34 normal adult subjects: twenty males, aged 2349 (mean 42 years); fourteen females,

B.E. Cham et al. / Clinica Chimica Acta 287 (1999) 45 57

47

aged 2553 (mean 45 years), whose serum total cholesterol concentrations were , 5.5 mmol / l, and 40 hypercholesterolemic (serum cholesterol . 5.5 mmol / l) subjects: twenty males, aged 3062 (mean 51 years); twenty females, aged 3768 (mean 53 years) who had fasted for 12 h, or from the umbilical cord vein from 32 neonates supplied by the Royal Womens Hospital, Brisbane. Twelve male and seventeen female hypercholesterolemic patients were on hypolipidemic (Zocor) drug therapy. The rational for including neonates was to establish whether the observations were limited or more generalized over a wide range of lipoprotein constituents. The procedures used were in accord with the Helsinki Declaration of 1975 as revised in 1983.

2.2. Measurements
The blood samples were protected from the light and were centrifuged at 48C. Serum samples were extracted immediately or stored at 2 708C in disposable polystyrene tubes with polyethylene stoppers. No chylomicrons were present in any of the specimens. Serum concentrations of vitamin K 1 and vitamin E [13], of total cholesterol, HDL-cholesterol and triglycerides [14], and of apolipoproteins A 1 and B [15], were measured according to established published methods. The concentration of LDL-cholesterol was calculated from the values for total cholesterol, HDL-cholesterol and triglycerides using the Friedewald formula [16]. VLDL-cholesterol was calculated from the difference of total cholesterol, HDL-cholesterol and LDL-cholesterol.

2.3. Statistical methods

Preliminary inspection of the data of 20 neonates, 22 normal adults and 28 hypercholesterolemic subjects indicated that all the measured variables were highly correlated with one another and that not all variables would be required in models to describe the relationships between vitamin K 1 and vitamin E and the other variables. Stepwise linear regression methods were used to determine the subsets of variables with greatest predictive power and the best tting models based on these variables. The patterns of variability in the data indicated that the regression analyses should be based on the logarithmic values of measured variables. The resulting derived models were then simplied by rounding off the values of estimated regression coefcients (without loss of predictive power) and expressed on the scale of the original measurements rather than in terms of their logarithmic values. An additional set of multiple regression analyses was carried out to determine the extent to which the equations relating vitamin K 1 and vitamin E to the other variable might differ between the three subject groups (neonates, normal adults and hypercholesterolemic individuals). These analyses included a nominal

48

B.E. Cham et al. / Clinica Chimica Acta 287 (1999) 45 57

variable in the regression models. The variable took one of three values and represented the subject groups. Samples of a further 12 neonates, 12 normal adults and 12 hypercholesterolemic patients were randomly chosen and the serum vitamins K 1 and E were measured. The measured values of these vitamins were then compared with the estimated values (derived from the models) using linear regression analysis. These results were also expressed as ratios of individual measured vitamin E / estimated vitamin E and individual measured vitamin K 1 / estimated vitamin K 1 values.

3. Results

3.1. Range of serum lipids and apolipoproteins

Table 1 shows the measured concentrations of the mean, standard deviation (SD) and range of total cholesterol, total triglyceride, HDL-cholesterol, Apo A 1 , Apo B, vitamin K 1 , vitamin E, and the calculated concentrations of VLDLcholesterol and LDL-cholesterol in serum from neonates, normal adults and hypercholesterolemic subjects. In addition, individual cholesterol standardised vitamin E and vitamin K 1 levels in serum are presented. The mean concentration of total cholesterol in serum of normal adults was three-fold that of neonates, and the hypercholesterolemic patients had a mean level which was four-fold that of neonates. A similar pattern was seen for most of the other parameters in serum (Table 1), except for HDL-cholesterol and Apo A 1 which were found to be similar in concentrations in normal adults and in hypercholesterolemic subjects. When compared, the differences in concentrations of the other parameters were highly signicant in sera of neonates, normal adults and hypercholesterolemic subjects (Table 1). In contrast, cholesterol standardised individual values for vitamins E and K 1 are not signicantly different in the three groups.

3.2. Linear correlation coefcients

Table 2 shows the linear correlation coefcients of vitamin K 1 and vitamin E with total cholesterol, total triglyceride, HDL-cholesterol, VLDL-cholesterol, LDL-cholesterol, Apo A 1 and in Apo B in sera of neonates, normolipidemic adults and hypercholesterolemic patients. The results of Table 2 lead to the following models for best predicting the concentrations of vitamins K 1 and E as described in the methods.

B.E. Cham et al. / Clinica Chimica Acta 287 (1999) 45 57

49

Table 1 Mean, standard deviation, and range of serum lipids and apolipoproteins Neonates n520 VLDL-cholesterol mmol / l LDL-cholesterol mmol / l HDL-cholesterol mmol / l Total cholesterol mmol / l Total triglyceride mmol / l Apo B g / l Apo A 1 g / l Vitamin K 1 ng / l Vitamin E mg / l Vitamin K 1 nmol / l ]]]]]]] Total cholesterol mmol / l Vitamin E mmol / l ]]]]]]] Total cholesterol mmol / l 0.2160.06 a (0.100.35) 0.6660.25 a (0.191.01) 0.6760.11 a (0.420.84) 1.5460.34 a (0.82.2) 0.4560.14 a (0.210.77) 0.3160 08 a (0.180.48) 0.7260.12 a (0.450.89) 82.86120.3 a (24571) 2.5161.01 a (0.35.3) 0.1360.18 a (0.040.85) 3.861.4 a (0.57.2) Normal adults n522 0.3860.15 b (0.180.85) 3.0660.51 b (2.073.99) 1.3360.38 b (0.662.05) 4.7660.56 b (3.75.5) 0.8260.33 b (0.391.86) 0.6260.14 b (0.410.88) 1.6360.35 b (0.962.15) 3436155 b (49590) 7.9061.50 b (5.511.7) 0.1660.08 a (0.020.33) 3.960.6 a (3.05.0) Hypercholesterolemic patients n528 0.7660 36 c (0.101 33) 4.3761.07 c (2.527.28) 1.1860.42 b (0.682.10) 6.3161.07 c (5.59.2) 1.7660.73 c (0.462.89) 1.0760.19 c (0.791.69) 1.5260.39 b (1.012.20) 5416197 c (204910) 11.1263.00 c (6.017.7) 0.1960.06 a (0.090.35) 4.160.9 a (2.46.1)

Values with different superscripts in the different columns are signicantly different ( p,0.001) from each other for that particular parameter as determined by Students t test. Values with similar superscripts in the different columns for that particular parameter are not signicantly different from each other.

3.3. Model for vitamin K1

The best tting stepwise regression model was: Log e vitamin K 1 (ng / l) 5 5.624(0.150) 1 2.124(0.193) log e Apo A 1 (g / l) 1 0.884(0.178) log e (Apo B /Apo A 1 ), with R 2 5 0.697. Values in brackets are parameter standard errors. The equation may be simplied to:

50

B.E. Cham et al. / Clinica Chimica Acta 287 (1999) 45 57

Table 2 Linear correlation coefcients between vitamin K 1 and vitamin E concentrations in serum and the concentrations of lipids and apolipoproteins in serum of neonates (n520). Normal adults (n522) and hypercholesterolemic patients (n528) Vitamin K 1 Variable VLDL-Ch LDL-Ch HDL-Ch Total-Ch Triglycerides Apo A 1 Apo B Vitamin E r 0.253* 0.744 0.451 0.749 0.423 0.654 0.745 0.695 Vitamin E Variable VLDL-Ch LDL-Ch HDL-Ch Total-Ch Triglycerides Apo A 1 Apo B Vitamin K 1 r 0.416 0.916 0.495 0.920 0.589 0.680 0.866 0.695

* P50.049, P,0.001 for all other values.

Predicted vitamin K 1 (ng / l) 5 369 3 Apo B (g / l) 3 Apo A 1 (g / l) with R 5 0.683 and an approximate standard error for a predicted vitamin K 1 level of 0.673 predicted vitamin K 1 .
2

3.4. Models for vitamin E

Two models, with equally good ts, resulted from the stepwise regression analyses. These equations were: Log e vitamin E (mg / l) 5 1.084(0.346) 1 0.173(0.184) log e cholesterol (mmol / l) 1 0.429(0.208) log e Apo B (g / l), with an R of 0.802 and Log e vitamin E (mg / l) 5 2.168(0.090) 1 0.535(0.145) log e Apo A 1 (g / l) 1 0.916(0.107) log e Apo B (g / l) with an R of 0.798.
2 2

(1)

(2)

These equations may be simplied with very little loss of t by rounding the coefcients of both log e cholesterol and log e Apo B to 0.5 in Eq. (1) and rounding the coefcients of log e Apo A 1 to 0.5, and log e Apo B to 1.0 in Eq. (2). R 2 for the modied equations were 0.785 and 0.795 respectively.

B.E. Cham et al. / Clinica Chimica Acta 287 (1999) 45 57

51

The simplied version of Eq. (2) may be expressed as: Predicted vitamin E (mg / l) 5 9.78 3 Apo B (g / l) 3 Apo A 0.3 (g / l) 1 The standard error of a predicted vitamin E determination using the above equation is 0.363predicted vitamin E.

3.5. Model for vitamin K1 with vitamin E included in the set of explanatory variables
A slightly better tting model is derived if vitamin E is one of the explanatory variables. Log e vitamin K 1 (ng / l) 5 4.697(0.444) 1 1.461(0.353) log e Apo A 1 (g / l) 1 0.529 (0.236) (Apo B /Apo A 1 ) 1 0.446(0.201) log e vitamin E (mg / l), with R 5 0.722. In its simplied form the above equation becomes: Predicted vitamin K 1 (ng / l) 5 115 3 Apo A 1 (g / l) 3 Apo B 3 vitamin E 0.5 (g / l), with an R 2 of 0.706 and an approximate standard error for predicted vitamin K 1 50.6453vitamin K 1 .
2

3.6. Subject group differences

There were small but statistically signicant improvements in best tting regression models when differences between subject groups were considered. For example, allowing for subject differences the model for log e vitamin K 1 becomes: Log e vitamin K 1 (ng / l) 5 C 1 0.684 (0.343) log e Apo A 1 (g / l) 1 0.354(0.200) log e vitamin E (mg / l) where: C 5 4.01(0.25) for neonates 5 4.67(0.44) for normal adults 5 5.13(0.50) for hyperlipidemics

This model has an R 2 of 0.741 which is only a small increase on the R 2 (50.722) for the equivalent model without subject group effects (see section on

52

B.E. Cham et al. / Clinica Chimica Acta 287 (1999) 45 57

Model for vitamin K 1 with vitamin E included in the set of explanatory variables). However, it should be noted that the above model no longer contains a term in log e Apo B. Similar small improvements in degree of t and changes in model structure occurred with other models when subject group effects were included.

3.7. Calculation of serum concentrations of vitamins K1 and E

To conrm the reliability of the equations for vitamins K 1 and E, viz; Vitamin K 1 (ng / l) 5 369 3 Apo B (g / l) 3 Apo A 1 (g / l), and Vitamin E (mg / l) 5 9.78 3 Apo B (g / l) 3 Apo A 0.3 (g / l) 1 we calculated the concentrations of these two vitamins of another set of subjects consisting of 12 neonates, 12 normolipidemic adults and 12 hypercholesterolemic subjects and compared the individual predicted values to the measured individual values. The results expressed as the ratios of measured vitamin E / estimated vitamin E and the ratios of measured vitamin K 1 / estimated vitamin K 1 of each individual are shown in Table 3 and they conrm the reliability of the above equations.

4. Discussion True deciency of vitamin E, associated with neurological sequelae [7] is uncommon and conned to a small number of subjects with major disturbances of vitamin E absorption or distribution [17,18]. The evaluation of such individual patients relies less on serum vitamin E levels but rather on functional assessments of vitamin E activity, such as peroxidative haemolysis tests [7]. For population studies, such cumbersome tests to assess vitamin E nutrition are unsuitable. Serum assays of vitamin E have the advantage of simplicity, and the capacity for storage and batch analysis. It is now established that the most valuable measurement of vitamin E nutritional status is the lipid-standardised serum vitamin E (mmol / l tocopherol / mmol / l cholesterol) [6,8,9]. In this measurement cholesterol serves as a simple marker of lipoprotein concentration, however the present study indicates that a combined function of Apo A 1 and B may be more accurate. The utility of this index has been well demonstrated by Gey in his evaluation of vitamin E as a risk factor in ischaemic heart disease (IHD) [9]. The study demonstrated an inverse correlation between lipid-standardised vitamin E and survival from IHD, independent of the inuence of cholesterol itself. The values of vitamin E involved were within the accepted

B.E. Cham et al. / Clinica Chimica Acta 287 (1999) 45 57

53

Table 3 Comparisons of measured and estimated vitamin K 1 and vitamin E concentrations in serum Neonates (n512) Vitamin K 1 Measured ng / l a Estimated ng / l b Ratio measured / estimated c r-value d Measured mg / l e Estimated mg / l f Ratio measured / estimated g r-value d 51199 38244 1.0860.10 0.981 1.45.5 1.55.3 1.0360.10 0.944 Normal adults (n512) 296572 281561 1.0460.04 0.989 5.99.8 5.810.0 0.9960.10 0.954 Hypercholesterolemic adults (n512) 3362180 3582591 1.0060.10 0.998 9.847.9 10.551.0 1.0060 10 0.997

Vitamin E

a b

Range of Vitamin K 1 concentrations as measured by an HPLC procedure. Range of vitamin K 1 concentrations as estimated according to the formula: Vitamin K 1 (ng / l) 5 369 3 Apo B (g / l) 3 Apo A 1 (g / l).

Ratios of corresponding individual measured / estimated values of vitamin K 1 in serum. r-Values were obtained by comparing individual measured values with the corresponding estimated value using a linear regression analysis; p,0.001 for all correlations. e Range of vitamin E concentrations as measured by an HPLC method. f Range of vitamin E concentrations as estimated according to the formula:
d

Vitamin E (mg / l) 5 9.78 3 Apo B (g / l) 3 Apo A 0.3 (g / l). 1

g

Ratios of corresponding individual measured / estimated values of vitamin E in serum.

reference range and well above those identied as indicating vitamin E deciency [8,9]. Thus, the concept of a relative vitamin E deciency with adverse pathological and clinical consequences has emerged, even though the mechanisms involved in the pathogenesis remain unclear. Although the overall sample size of the various sera in the present study is relatively small, the statistical observations of the study have underlined the necessity of quoting serum vitamin E and K 1 levels adjusted for not only lipids but apolipoprotein concentrations. The three groups of subjects examined had highly signicantly different serum levels of lipids, apolipoproteins and vitamins but these differences disappeared when vitamin E and vitamin K 1 levels were related to total cholesterol (Table 1) or to apolipoproteins (data not shown). The means and ranges for lipid-standardised vitamin E of the two adult groups encompassed those published by Gey for a number of regional European populations [9]. The present work has identied a degree of correlation between serum vitamin K 1 and lipid parameters which is similar to that for vitamin E. In consequence, the interpretation of absolute levels of vitamin K 1 in serum is also problematical. Quite low serum concentrations of the vitamin K 1 have been

54

B.E. Cham et al. / Clinica Chimica Acta 287 (1999) 45 57

reported to occur in patients with fractures, with or without osteoporosis. Values quoted were 113 pg / ml (mean) for individuals with traumatic fractures [19], and 98 pg / ml (mean) for those with osteoporosis and fractures [20]. These serum concentrations were considered pathological because they were signicantly lower than age-matched control values, but it should be noted that they do not differ from absolute values in healthy neonates (Table 1). It is not possible to conclude whether the observed changes in vitamin K 1 were a consequence of changes in lipoprotein metabolism, vitamin K metabolism or both. The data in Table 1 also indicates that a 6.5 fold difference in mean serum vitamin K 1 concentrations between subject groups attenuate markedly when normalized for serum cholesterol concentration. Deciency of vitamin K will result in defective gamma-carboxylation of vitamin K dependent proteins, and will be manifested by the failure of these proteins to function normally [21]. Traditionally, the prothrombin time has been used as an index of function of vitamin K dependent clotting proteins but it is known to be a crude and very insensitive measure [21]. Radioimmunoassay of decarboxylated prothrombin, before and after vitamin K administration, has proved to be a much more sensitive index of subclinical vitamin K deciency [22]. Using this technique, 31% of patients with chronic bowel disease had mild vitamin K deciency, without overt clotting defects, but neither serum vitamin K nor lipids were monitored in this study [22]. Osteocalcin or bone-Gla-protein (BGP) is a well-studied vitamin K-dependent protein secreted by osteoblasts [23]. A previous study has suggested that post-menopausal women may develop a vitamin K decient state which is manifested by sub-optimal gamma-carboxylation of BGP, detected by calcium binding studies [24]. No defects in nutrition or bowel function were identied in these women. However, measurable effects on calcium metabolism, reversible on treatment with vitamin K 1 were noted. These results, viewed in the light of the low absolute serum vitamin K 1 levels in osteoporotic subjects with hip fractures, have raised the question of a possible role of vitamin K in the pathogenesis of post-menopausal or senile osteoporosis. The formation of other Gla-proteins is also vitamin K dependent. Such Gla-proteins are found in atherosclerotic plaques and are associated with calcication; an early step in plaque evolution. Calcium phosphate (hydroxyapatite) precipitates by a mechanism similar to active bone formation and is vitamin K dependent [25]. The Gla-proteins only known function is to bind calcium and do so with very high afnity [26,27]. These proteins do not interfere with normal calcium homeostasis and paradoxically may inhibit precipitation of calcium salts. The association of vitamin K and elevated lipid serum levels could lead to a synergistic development of the atheromatous plaque and acute coronary syndromes [28]. Lipid and calcium are co-localised in plaques and rupture of plaques that precipitates acute events tends to occur at

B.E. Cham et al. / Clinica Chimica Acta 287 (1999) 45 57

55

sites of calcium deposition [29,30]. Thus in the context of atherosclerosis there is a nexus with lipoproteins, vitamin K and vitamin E. Two relative deciency states of vitamin K have been identied by assays that detect reduced gamma-carboxylation of vitamin K dependent proteins, correctable by vitamin K administration [24]. In neither case has the deciency state been correlated with absolute or lipoprotein-standardised levels of serum vitamin K 1 . There is a clear need to relate lipid-standardised serum vitamin K 1 concentrations to other indices of bone metabolism in both post-menopausal women and in individuals with chronic bowel disease in the hope of identifying those at high risk of osteoporotic fractures, and in patients with atherosclerosis with a view to preventive intervention. At this stage it is unknown how the interdependence of serum concentrations of vitamin K 1 , vitamin E, lipids, apolipoprotein A 1 and apolipoprotein B will be affected by induced acute changes in lipid removal with retention of apolipoproteins in serum, as is the case for the lipid apheresis procedure currently being investigated as a possible treatment for atherosclerosis in our laboratory [3136]. The present study has provided a range for lipoprotein standardised serum vitamin K 1 and vitamin E levels in 50 adults and 22 neonates. The observed correlation between serum vitamin K 1 with serum cholesterol followed a similar pattern as has previously been shown for serum vitamin E and serum cholesterol. Importantly, this communication indicates that the use of apolipoprotein levels appear to yield information beyond simple lipid sub-fractions to the extent that the atherogenic apolipoprotein B and antiatherogenic apolipoprotein A 1 , both being non-lipid components of lipoprotein fractions, best described the relative status of vitamins E and K 1 in such a manner that reliable equations using concentrations of such apolipoprotein components of lipoproteins could be used to calculate the serum concentrations of these vitamins. Previous studies with haemodialysis patients have shown that vitamin K 1 was related to apolipoprotein E and strongly modulated by the apo E genotype [37,38]. It is clear that further groups of both healthy and diseased subjects need to be studied to dene reference ranges and to identify abnormalities linked to disease processes.

Acknowledgements The authors acknowledge the excellent technical work provided by Mr John Saville, Mr T. Kamst and Mr N. Caterer. The assistance with the statistical methods by Dr Anthony Barnes and the collaboration of Dr Peter Roeser are gratefully acknowledged. We also thank Annette Miles for her secretarial assistance. This work was supported by grants from the National Health and Medical Research Council of Australia and from Curacel International.

56

B.E. Cham et al. / Clinica Chimica Acta 287 (1999) 45 57

References
[1] Shearer MJ, Barkhan P, Webster GR. Absorption and excretion of an oral dose of tritiated vitamin K 1 in man. Br J Haematol 1970;18:297308. [2] Traber MG, Kayden HJ. Preferential incorporation of alpha-tocopherol vs. gamma-tocopherol in human lipoproteins. Am J Clin Nutr 1989;49:51726. [3] Traber MG, Burton GM, Ingold KU, Kayden HJ. RRR- and SRR-a-tocopherols are secreted without discrimination in human chylomicrons, but RRR-a-tocopherol is preferentially secreted in very low density lipoproteins. J Lipid Res 1990;31:67585. [4] Traber MG, Kayden HJ. Vitamin E is delivered to cells via the high afnity receptor for low-density lipoprotein. Am J Clin Nutr 1984;40:74751. [5] Meddings JB, Spady DK, Dietschy JM. Kinetic characteristics and mechanisms of regulation of receptor-dependent and receptor-independent LDL transport in the liver of different animal species and humans. Am Heart J 1987;113:47581. [6] Horwitt MK, Harvey CC, Dahm CH, Searcy MT. Relationship between tocopherol and serum lipid levels for the determination of nutritional adequacy. Ann NY Acad Sci 1972;203:22336. [7] Sokol RI. Vitamin E deciency and neurologic disease. Ann Rev Nutr 1988;8:35173. [8] Thurnham DI, Davies JA, Crump BI, Situnayake RD, Davis M. The use of different lipids to express serum tocopherol lipid ratios for the measurement of vitamin E status. Ann Clin Biochem 1986;23:51420. [9] Gey KF. Inverse correlation of vitamin E and ischaemic heart disease. Int J Vit Nutr Res 1989;30(Supplement):22431. [10] Cham BE, Smith IL, Colquhoun DM. Correlations between cholesterol, vitamin E and Vitamin K 1 in serum: paradoxical relationships to established epidemiological risk factors for cardiovascular disease. Clin Chem 1998;44:17535. [11] Sadowski IA, Hood SI, Lallal GE, Garry PJ. Phylloquinone in plasma from elderly and young adults: factors inuencing its concentration. Am J Clin Nutr 1989;50:1008. [12] Mandelbrot LM, Guillaumont M, Leclercq M, Lefrere JJ, Gozin D, Daffos F et al. Placental transfer of vitamin K and its implications in fetal haemostasis. Thromb Haemostas 1988;60:3943. [13] Cham BE, Roeser HP, Kamst TW. Simultaneous liquid-chromatographic determination of vitamin K 1 and vitamin E in serum. Clin Chem 1989;35:22859. [14] McNamara JR, Schaefer EJ. Automated enzymatic standardized lipid analyses for plasma and lipoprotein fractions. Clin Chim Acta 1987;166:18. [15] Pruvot I, Flevet C, Durieux C, Vu Dac N, Fruchart JC. Electroimmuno- and immunonephelometric assays of apolipoprotein A-I by using a mixture of monoclinical antibodies. Clin Chem 1988;34:204852. [16] Friedewald WT, Levy RI, Fredrickson DS. Estimation of plasma low density lipoprotein cholesterol concentration without use of the preparative ultracentrifuge. Clin Chem 1972;18:499504. [17] Muller DPR, Harries JT, Lloyd JK. The relative importance of the factors involved in the absorption of vitamin E in children. Gut 1974;15:96671. [18] Traber MG, Soko RI, Burton GW, Ingold KU, Papas AM, Huffaker JE, Kayden HI. Impaired ability of patients with familial isolated vitamin E deciency to incorporate tocopherol into lipoprotein secreted by the liver. J Clin Invest 1990;85:397407. [19] Bitensky L, Hart IP, Catterall A, Hodges SI, Pilkington MI, Chayen I. Circulating vitamin K levels in patients with fractures. J Bone Joint Surg 1988;70:6634. [20] Hart JP, Shearer MI, Klenerman L et al. Electrochemical detection of depressed circulating levels of vitamin K 1 in osteoporosis. J Clin Endocrin Metab 1985;60:12689.

B.E. Cham et al. / Clinica Chimica Acta 287 (1999) 45 57

57

[21] Suttie JW. Recent advances in hepatic vitamin K metabolism and function. Hepatology 1987;7:36776. [22] Krasinski SD, Russell RM, Furie BC, Kruger SF, Jacques PF, Furie B. The prevalence of vitamin K deciency in chronic gastrointestinal disorders. Am J Clin Nutr 1985;41:63943. [23] Lian JB. Osteocalcin: functional studies and postulated role in bone resorption. In: Suttie JW, editor, Current advances in vitamin K research, New York: Elsevier, 1988, pp. 24557. [24] Knapen MHJ, Hamulyak K, Vermeer C. The effect of vitamin K supplementation on circulating osteocalcin (bone Gla protein) and urinary calcium excretion. Ann Int Med 1989;111:10015. [25] Bostrom KK, Watson E, Horn S, Wortham C, Herman IM, Demer LL. Bone morphogenic protein expression in human atherosclerotic lesions. J Clin Invest 1993;91:18009. [26] Vermeer C. Gamma-carboxyglutamate-containing proteins and the vitamin K-dependent carboxylase. Biochem J 1990;266:62536. [27] Price PA. Gla-containing proteins of bone. Connect Tissue Res 1989;21:5160. [28] Wexler L, Brumdage B, Crouse I et al. Coronary artery calcication: pathophysiology, epidemiology, imaging methods, and clinical implications. Circulation 1996;94:117592. [29] Demer LL, Watson KE, Bostrom K. Mechanism of calcication in atherosclerosis. Trends Cardiovasc Med 1994;4:459. [30] Hirsch D, Azoury R, Sarig S, Kruth HS. Colocalization of cholesterol and hydroxyapatite in human atherosclerotic lesion. Calcif Tissue Int 1993;52:948. [31] Cham BE, Knowles BR. A solvent system for delipidation of plasma or serum without protein precipitation. J Lipid Res 1976;17:17681. [32] Cham BE, Knowles BR. In vitro partial relipidation of apolipoproteins in plasma. J Biol Chem 1976;251:636771. [33] Cham BE, Kostner KM, Dwivedy AK et al. Lipid apheresis: an in vivo application of plasma delipidation with organic solvents resulting in acute transient reduction of circulating plasma lipids in animals. J Clin Apheresis 1995;10:619. [34] Cham BE, Kostner KM, Dwivedy AK et al. Lipid apheresis in an animal model causes in vivo changes in lipoprotein electrophoretic patterns. J Clin Apheresis 1996;11:6170. [35] Cham BE, Smith IL. Lipid apheresis in an animal model causes acute reduction in plasma lipid concentrations and mobilization of lipid from liver and aorta. Pharmacol (Life Sci Adv) 1994;13:2532. [36] Kostner KM, Smith IL, Dwivedy AK et al. Lecithincholesterol acyltransferase activity in normocholesterolaemic and hypercholesterolaemic roosters: effect of lipid apheresis. Eur J Clin Invest 1997;27:2128. [37] Saupe I, Shearer MI, Kohlmeier M. Phylloquinone transport and its inuence on gcarboxyglutamate residues of osteocalcin in patients on maintenance hemodialysis. Am J Clin Nutr 1993;58:2048. [38] Kohlmeier M, Saupe I, Drossel HI, Shearer MI. Variation of phylloquinone (vitamin K 1 ) concentrations in hemodialysis patients. Thromb Haemost 1995;74:12524.