Plasma lipoproteins and adrenocortical hormones in men — positive association of low density lipoprotein cholesterol with plasma cortisol concentration

Clinica Chimicu Acta, 180 (1989) 113-120 Elsevier

113

CCA 04376

Plasma lipoproteins in men of low density with plasma

and adrenocortical hormones positive association lipoprotein cholesterol cortisol concentration

M.N. Nanjee ’ and N.E. Miller 2 ’ Department of Chemical Pathology and Metabolic Disorders, St. Thomas’ Hospital Medical School, London (UK) 2 Section on Endocrinology and Metabolism, Department of Medicine, The Bowman Gray School of Medicine, Winston-Salem, NC (USA) (Received 5 July 1988; revision received 10 November 1988; accepted 22 November 1988) Key woruk Lipoprotein; Cortisol; Androstenedione; Adrenal cortex; Triglyceride; Cholesterol

sununary The associations of plasma lipoprotein lipids with the plasma concentrations of two adrenocortical hormones (sampled at 09.00-11.00 h) have been investigated in a random sample of 70 men aged 52-67 yr (mean, 59 yr). Plasma low density lipoprotein (LDL) cholesterol concentration was found to be positively correlated with plasma cortisol, independently of the concentrations of plasma triglyceride, high density lipoprotein cholesterol and androstenedione. After adjusting for covariates, plasma cortisol explained approximately twelve per cent of the variance in LDL cholesterol. These results suggest that plasma cortisol may significantly influence the metabolism of LDL in healthy humans.

Introduction

Evidence from clinical studies and from studies in vitro has suggested that there may be several interactions between plasma lipoprotein metabolism and adrenocortical function in humans. Firstly, hypersecretion of glucocorticoids in Cushing’s syndrome is associated with overproduction of very low density lipoprotein (VLDL) triglycerides in the liver and consequent hyperlipidemia [l]. Secondly, receptorCorrespondence to: N.E. Milk, Section on Endocrinology and Metabolism, Department of Medicine, The Bowman Gray School of Medicine, Winston-Salem, NC 27103, USA.

0009-8981/89/$03.50

0 1989 Elsevier Science Publishers B.V. (Biomedical Division)

114

mediated uptake of low density lipoprotein (LDL), the end-product of VLDL catabolism, appears to provide the principal source of cholesterol for cortisol synthesis in human adrenocortical cells [2]. Thirdly, hydrocortisone has been shown to inhibit the receptor-mediated endocytosis of LDL by cultured human fibroblasts [3]. In spite of these observations, the extent to which adrenocortical function influences lipoprotein metabolism, and vice versa, has been little investigated in normal humans. Therefore, we studied the cross-sectional associations of plasma lipoprotein lipids, plasma cortisol and plasma androstenedione in a random sample of seventy men. Methods Clinical procedures The subjects were members

of the SpeedwelI cohort of the Caerphilly and Speedwell Collaborative Heart Disease Studies [4], a random sample of 2348 men. A computer-generated random sample of 100 of the cohort was invited to participate in the present study. The responders attended a morning clinic (09.00-11.00 h) after a light fat-free breakfast. All were in apparent good health. From each subject 15 ml blood was collected into Na,EDTA (1 mg/ml), and the plasma separated by centrifugation at 4O C for 30 min at 1500 X g. Aliquots of plasma (1 ml) were frozen in liquid nitrogen for the hormone assays. The remaining plasma was maintained at 4O C while awaiting the lipid analyses (within 72 h). Luboratov

procedures

Each assay series (total cholesterol, total triglyceride, high density lipoprotein (HDL) cholesterol, cortisol and androstenedione) was carried out in duplicate and as a single batch. Cholesterol and triglyceride were quantified by enzymatic colorimetric procedures (Boehringer-Mannheim; cat. nos. 237574 and 166448). HDL cholesterol was assayed after precipitation of other lipoproteins with heparin and MnCl, (92 rnmol/l) [5]. LDL cholesterol was calculated according to Friedewald et al [6]. Plasma cortisol and androstenedione concentrations were quantified by radioimmunoassay using commercially available kits (RIA UK Ltd: cat. nos: A-4300L and A-6200L). The ranges of within-batch coefficients of variation for u-i-level quality control pools were 5-9% for cortisol and 4-98 for androstenedione. Complete data were obtained from 70 subjects, whose mean age was 59 yr (range, 52-67 yr). Statistical

analyses

Coefficients of linear correlation, partial correlations and regression models were calculated by standard procedures [7]. As 22 correlations were calculated, p < 0.01 was considered to be statistically significant. Plasma triglyceride and andrastenedione concentrations were analyzed after logarithmic transformation to the base e, as their frequency distributions were positively skewed.

115

Descriptive statistics of the data appear in Table IA, and a correlation matrix in Table IB. Plasma triglyceride was unrelated to either adrenocortical hormone. Both LDL cholesterol and HDL cholesterol were positively correlated with plasma cortisol. There was also a positive correlation between HDL cholesterol and androstenedione concentration. Plasma triglyceride and HDL cholesterol were negatively correlated, and plasma cortisol and androstenedione were positively correlated. Partial correlations are shown in Table II. Adjusting for other lipids had little effect on the strengths of the associations of LDL cholesterol and HDL cholesterol with the hormones. When adjustment was made for other lipids and the other hormone, the associations of HDL cholesterol with cortisol and androstenedione were weakened to an extent that neither was statistically significant. In contrast, the correlation between LDL cholesterol and cortisol was strengthened and remained highly statistically significant (p = O.OOll), after adjusting for HDL cholesterol, triglyceride and androstenedione.

TABLE

IA

Plasma lipoprotein lipid and hormone concentrations

Total triglyceride Total cholesterol LDL cholesterol HDL cholesterol COrti.Wl Androstenedione

a Geometric TABLE

units

Mean

SD

Range

mmol/l IlUIlOl/l mmol/l mmol/l nmOl/l nmol/l

0.65 ’ 5.86 3.63 1.21 4% 1.28 ’

0.52 0.91 0.93 0.43 195 0.34

0.21-2.15 3.81-8.34 1.19-6.29 0.20-2.55 186-1218 0.41-1.99

mean

IB

Correlation

matrix between plasma lipids and hormones Triglyceride

LDL cholesterol HDL cholesterol Cortisol Androstenedione

- 0.228 (0.057) - 0.450 (0.001) - 0.124 (0.307) - 0.055 (0.654)

Plasma triglyceride and androstcnedione

valuesare given ill pprenthcscs.

LDL cholesterol

HDL cholesterol

- 0.005 (0.967) 0.320 (0.007) - 0.001 (0.993)

(0.006) 0.284 (0.017)

Cortisol

0.323

were analyzed after logarithmic

transformation

0.499 (0.001) to the base e.

p

116 TABLE II Partial correlations between lipids and hormone concentrations Triglyceride

LDL chol

Adjusting for the other lipids Cortisol

0.125 (0.311) Androstenedione 0.089 (0.472) Adjusting for the other lipids and hormone Cortisol 0.094 (0.449) 0.033 Androstenedione (0.788)

Triglyceride base e. TABLE

and androstenedione

concentrations

HDL chol

0.359 (0.003) 0.023 (0.850)

0.359 (0.003) 0.291 (0.016)

0.390 (0.001) -0.165 (0.182)

0.253 (0.039) 0.129 (0.297)

were analysed

after logarithmic

transformation

to the

III

Regression models with LDL cholesterol as the dependent variable Variables

in the regression

In Tg

mode1 In And

HDL

X X

X

X

X

X

X

X

X

X

Adjusted

0.000 0.102 0.137 0.067 0.067 0.187 0.209

0.147 0.089 0.111 0.039 0.025 0.150 0.160

r2

Cortisol

x

X

r2

X X X

X

r2 gives the proportion of the variation in LDL cholesterol that is explained by the variables model. Adjusted r2 has been corrected to reflect the number of variables in the model. In Tg, log, triglyceride; HDL, HDL cholesterol; In And, log, androstenedione.

*1 i

0

.

200

Fig. 1. Relationship

400

600 600 1000 CortJsol (nmol/l)

1200

1400

of LDL cholesterol concentration

to plasma cortisol concentration

in 70 men.

in the

117

Multiple regression models with LDL cholesterol as the dependent variate are summarized in Table III. It can be seen that cortisol explained 9-148 of the variation in LDL cholesterol concentration in this group of men, whereas androstenedione made no significant contribution. The relationship between LDL cholesterol and plasma cortisol is illustrated in Fig. 1. From this it can be seen that there was a single outlying value. Omission of this result from the data had no significant effect on the outcome of the statistical analyses.

Discussion The major finding of this study was that in a random sample of apparently healthy men, the plasma concentration of LDL cholesterol was positively correlated with that of cortisol. This association was independent of the plasma triglyceride and HDL cholesterol concentrations. In contrast, LDL cholesterol was unrelated to the concentration of a second adrenocortical steroid, androstenedione. A positive correlation between plasma cortisol and plasma total cholesterol concentration (presumably reflecting LDL) has been observed in men with coronary artery disease or Type A behavior by Schwertner et al. [8]. To our knowledge, however, our results provide the first evidence that a similar relationship exists in the general male population. Plasma LDL cholesterol was not measured directly in our study, but was estimated from plasma total cholesterol, triglyceride and HDL cholesterol using the Friedewald formula, which had originally been constructed using data from fasted subjects [6]. Although our subjects was permitted to have a light breakfast on the day of venepuncture, the possibility that this produced an artefactual correlation between LDL cholesterol and cortisol can be discounted. Ingestion of lo-20 g of fat by any participant would have raised plasma triglyceride on’ average by 0.1-0.5 mmol/l after 2-4 h [9], due mostly to the presence of chylomicrons. It can be calculated that this would have resulted in an underestimation of LDL cholesterol by up to 0.20 mmol/l. Even if such errors had been non-randomly distributed among the 70 men, they could have explained no more than about ten percent of the slope of the regression line shown in Fig. 1. There was no visible chylomicronemia in any sample of plasma, and no difficulty was experienced in precipitating apo B-containing lipoproteins. Cortisol secretion is known to be increased in obesity [lo], and obesity.is also a cause of hyperlipidemia [ll]. Although relative body weight was not measured in the present subjects, the association between LDL cholesterol and plasma cortisol is unlikely to be explained by a confounding effect of adiposity for two reasons. Firstly, plasma cortisol concentration is not increased in obese individuals, oversecretion being compensated by increased elimination [lo]. Secondly, the major effect of obesity on plasma lipids is not on LDL, but on plasma triglycerides (raised) and HDL cholesterol (lowered) [ll], neither of which was significantly correlated with plasma cortisol in this study.

118

Since LDL cholesterol is utihzed by the human adrenal cortex for cortisol synthesis [2], one possible explanation for the association between LDL and plasma cortisol is that the circulating LDL concentration can, under some circumstances, become rate-limiting in cortisol synthesis. This seems unlikely within the normal range of LDL concentration, however, as heterozygous hypobetahpoproteinemic individuals have been shown to have normal adrenocortical function under basal and ACTH-stimulated states [12]. Only when there was complete absence of circulating LDL, as in abetalipoproteinemia and homozygous hypobetalipoproteinemia, was the response of the adrenal cortex to ACTH found to be subnormal [12,13]. Other evidence that the rate of delivery of LDL to adrenal cortical cells is unlikely to influence the rate of cortisol synthesis in vivo has been provided by demonstrations that adrenocortical function is normal, or only slightly diminished, in patients with homozygous familial hypercholesterolemia, whose cells have no functioning LDL receptors [14,15]. A second possibility is that circulating cortisol stimulates LDL production and/or reduces LDL catabolism. Glucocorticoid administration has been found to increase hepatic triglyceride secretion in rats [Ml, and Taskinen et al. [l] have shown that the production rate of VLDL triglyceride is increased in patients with Cushing’s syndrome. Taslcinen et al. speculated that this might lead to increased conversion of VLDL to LDL, the concentration of which was also raised in their patients. However, in both Cushing’s syndrome and patients taking exogenous glucocorticoids for immunosuppression [17-191 hypertriglyceridemia is usually much more prominent than hypercholesterolemia. In contrast, in the present men the rise in LDL cholesterol that was associated with increasing plasma cortisol concentration occurred without any increase in plasma triglycerides, suggesting that the rise in LDL was not secondary to an increase in VLDL synthesis. Henze et al. [3] found that hydrocortisone decreased the internalization (and hence the degradation) of LDL by cultured human fibroblasts and arterial smooth muscle cells, by reducing the endocytosis of receptor-bound LDL particles. A similar phenomenon occurring in vivo might reduce the fractional rate of clearance of LDL from the circulation. A further possibility is that the association between LDL cholesterol and cortisol reflects a relation of LDL metabolism not with cortisol activity, but with the concentration of cortisol-binding protein. Alternatively, the primary relationship of LDL metabolism might be with ACTH secretion, rather than with cortisol concentration. As no measurements were made of cortisol-binding protein or of ACTH in our study, these possibilities will need to be examined in a future investigation. In view of the well recognized circadian rhythm of cortisol secretion, measurement of the 24 h urinary cortisol excretion rate would also be informative. Although we found no clear relationship between HDL cholesterol and either cortisol or androstenedione that was independent of the other hormone, positive trends with both hormones were present (particularly with cortisol), which may be of physiological significance. The possibility that cortisol influences HDL metabolism has been raised by demonstrations of raised HDL cholesterol levels in patients with Cushing’s syndrome [l], and in hypercholesterolaemic patients given ACTH or

119

cortisone acetate [20]. Furthermore, as apo E~n~g subclasses of HDL are known to bind with high affinity to LDL receptors 1211, the possibility that such subclassescontribute to the provision of cholesterol for steroid hormone synthesisin humans cannot be discounted, Acknowledgements

We thank D. Bainton, I. Baker and P. Sweetnam (Medical Research Council, UK) for assistance with the recruitment of subjects for the study, and D. Case (Bowman Gray school of Medicine, USA) for assistance with the statistical analyses. The work was supported by funds from St. Thomas’ Charitable Trust. References 1 Taskinen M-R, Nikkil8 EA, Pelkonen R, Sane T. Plasma lipoproteins, lipolytic enzymes, and very low density lipoprotein triglyceride turnover in Gushing’s syndrome. J Clin Endocrinol Metab I983;57:619-626. 2 Brown MS, Kovanen E*I; Gohlstein JL. Receptor-mediated uptake of ~~~~~01~~01 and its utilization for steroid synthesis in the adrenal cortex. Recent Prog Horm Res 1979;35:215-257. 3 Henze K, Chait A, Albers JJ, Bierman EL. Hydnrortisone decreasm the intemalization of low density lipoprotein cultured human fibroblasts and arterial smooth muscle cells. Eur J Clin Invest 1983;13:171-177. 4 Caerphilly and Speedwell Collaborative Group. Caerphilly and Speedwell Collaborative Heart Disease Studies. J Epidemiol Commun Hlth 19&4,38:259-262. 5 Warnick GR, Albers JJ. A comprehensive evaluation of the ~~-~~ precipitation procedure for estimating high density lipoprotein cholesterol. J Lipid Res 1978;19:65-76. 6 Friedewald WI’, Levy RI, Fredrickson DS. Estimation of the conoentration of LDL cholesterol in plasma without the use of the preparative ultracentrifuge. Clin Chem 1972;18:499-502. 7 Snedecor GW, Co&an WG. Statistical Methods. Iowa State University Press, Ames, Iowa, sixth ed. 8 Schwertner HA, Troxler RG, Uhl GS, Jackson WG. Relationship between cortisol and cholesterol in men with coronary artery disease and type A behavior. Arteriosclerosis 1984;4:59-64. 9 Lewis B, Chait A, February AW, Mattock M. Functional overlap between chylomicra and very low density lipoproteins of human plasma during alimentary lipemia. Atherosclerosis 1973;17:455-462. 10 Glass AR, Beerman KD, Dahms WT. et al. Endocrine function in human obesity. Metabolism 1981;30:89-104. 11 Lewis 8. Influence of diet, energy balance and hormones on serum lipids. In: The Hyperlipidaemias. Clinical and Laboratory practice. Oxford: Blackwell scientific Publ., 131-180. 12 Illingworth DR, Kenny TA, Connor WE, Grwoll ES. Adrenal function in heterozygous and homozygous h~~~rot~~a. J Clin Endocrinol Metab 1982;54:27-33. 13 Illingworth DR, Kenny TA, Commr WE Grwoll ES. Corticosteroid production in abetalipoproteinemia: evidence for an impaired response to ACTH. J Hab Clin Med 1982;100:115-226. 14 Allen JM, Thompson GR, Myant NB. Normal adrenocortical reqonse to adrenocorticotrophic hormone in patients with homozygous familial hypercholesterolaemia. Clin !Ici 1983;65:99-101. 15 Illingworth DR, Lees AM, Lees RS. Adrenal cortical function in homozygous familial hypercholesterolemia. Metabolism 1983;M:104S-1052. 16 Bagdade D, Ya E, Albers J, et al. Glucocorticoids and triglyceride transport: effects on triglyceride secretion rate, lipoprotein lipase and plasma lipoproteins in the rate. Metabolism 197@25:533-542. 17 Ibels LS, Alfrey AC, Weil R. Hyperlipidemia in adult, pediatric and diabetic renal transplant recipients. Am J Med 1978;64:634-642. 18 El-Shaboury AH, Hayes TM. Hyperlipidaemia in asthmatic patients receiving long term steroid therapy. Br Med J 1973;2:85-86.

120 19 Casaretto A, Goldsmith R, Marchioro TL, Bagdade JD. Hyperlipidaemia after successful renal transplantation. Lmcet 1974;1:481-484. 20 Oliver MF, Boyd GS. Endocrine aspects of coronary sclerosis. Lancet 1956;2:1273-1276. 21 Mahley RW, Weisgraber KH, Bersot TP, Innerarity TL. Effects of cholesterol-feeding on human and animal high density lipoproteins. In: Gotto AM, Miller NE, Oliver MF, eds. High density lipoproteins and atherosclerosis. Amsterdam: Elsevier, 149-176.