Plasma plant sterols serve as poor markers of cholesterol absorption in man.

The validation of the use of plasma plant sterols as a marker of cholesterol absorption is frail. Nevertheless, plant sterol concentrations are routinely used to describe treatment-induced changes in cholesterol absorption. Their use has also been advocated as a clinical tool to tailor cholesterol-lowering therapy. Prior to wider implementation, however, the validity of plant sterols as absorption markers needs solid evaluation. Therefore, we compared plasma plant sterol concentrations to gold-standard stable isotope-determined cholesterol absorption. Plasma campesterol/TC concentrations (camp/TC) were measured in a population of 175 mildly hypercholesterolemic individuals (age: 59.7 ± 5.6 years; BMI: 25.5 ± 2.9 kg/m(2); LDL-C: 4.01 ± 0.56 mmol/l). We compared cholesterol absorption according to the plasma dual-isotope method in subjects with the highest camp/TC concentrations (N = 41, camp/TC: 2.14 ± 0.68 μg/mg) and the lowest camp/TC concentrations (N = 39, camp/TC: 0.97 ± 0.22 μg/mg). Fractional cholesterol absorption did not differ between the groups (24 ± 12% versus 25 ± 16%, P = 0.60), nor was it associated with plasma camp/TC concentrations in the total population of 80 individuals (β = 0.13; P = 0.30, adjusted for BMI and plasma triglycerides). Our findings do not support a relation between plasma plant sterol concentrations and true cholesterol absorption and, therefore, do not favor the use of these sterols as markers of cholesterol absorption. This bears direct consequences for the interpretation of earlier studies, as well as for future studies targeting intestinal regulation of cholesterol metabolism.

This breakfast contained 50 g carbohydrate, 27 g protein, 15 g fat, 440 kcal, and 84 mg cholesterol (including 50 mg of the tracer). Subjects refrained from food for 4 h following the test meal to allow for complete gastric emptying. Participants maintained their usual dietary pattern as well as physical activity and kept a dietary record throughout the study. Additional blood samples were collected at 24, 48, and 72 h postadministration of the tracers.
Plasma was isolated by centrifugation and stored at Ϫ 80°C, for subsequent cholesterol extraction and measurement of the tracer enrichments. Fractional cholesterol absorption (FCA) was calculated as the plasma ratio of D 7 and 13 C-cholesterol enrichment at T = 72 h divided by the ratio D 7 / 13 C-cholesterol species administered to the subjects ( 2 ). These enrichments were measured by means of GCMS, as described below. Dietary intake was calculated using a Dutch nutrition database program (www.dieetinzicht.nl). Plant sterol intake was calculated through the USDA National Nutrient Database for Standard Reference (ndb.nal. usda.gov).

Analytic procedures
Fasted plasma TC, HDL-C, and TG concentrations were measured using standard automated methods; LDL-C concentrations were calculated using the Friedewald formula ( 20 ). Plasma noncholesterol sterol concentrations were analyzed from nonsaponifi able plasma material with GCMS, as previously described ( 21 ). Plasma enrichments of D 7 -cholesterol and 13 C 2 -cholesterol were measured by GCMS (Center for Liver, Digestive and Metabolic Diseases, University of Groningen). Plasma free cholesterol was extracted with ethanol/acetone (1:1, v/v) for analysis of isotopic enrichment by GCMS. Cholesterol derivatives were separated with a 5975C InertXL EI/CI MSD (Agilent Technologies) in the electron ionization mode using a ZB5ms column [30 m(l) × 0.25 mm (ID) × 0.25 m (fi lm) (Phenomenex)]. Ions monitored were m/z 368-375 corresponding to the M0-M7 mass isotopomers. The measured fractional isotopomer distribution was corrected for the fractional distribution due to natural abundance of 13 C and 2 H by multiple linear regression as described by Lee et al. ( 22 ) to obtain excess fractional distribution of mass isotopomers resulting from isotope dilution and incorporation of the infused tracer. In this approach, M7 represented the orally administered label. The day-to-day coeffi cient of variance for the ratio 12 C/ 13 C was 0.06%.

Statistical analyses
Sample size. Assuming a common standard deviation of cholesterol absorption of 10%, a sample size of 10 subjects in each group has more than 80% power to detect a difference in means of 20%. An additional power calculation showed that this sample size of 20 would have 80% power to detect a signifi cant and clinically acceptable correlation ( R = 0.60) between two continuous variables, such as plasma plant sterol concentrations and cholesterol absorption rates. However, as plasma plant sterol levels were previously validated in a population of 17 ( 5 ) and 63 ( 12 ) subjects, we chose to validate these markers in an equally large population of 80 subjects; i.e., 40 subjects with high plasma campesterol/TC concentrations versus 40 with low campesterol/ TC concentrations.
Study outcomes. Normally distributed data are presented as means and standard deviations (SD). Skewed data were log-transformed prior to testing and are presented as median with the range. Differences in parameters between subjects with high and low campesterol/TC concentrations were analyzed with independent sample t -tests. Associations between plasma plant sterol We revalidated plasma plant sterol concentrations as markers of cholesterol absorption in a population of 80 mildly hypercholesterolemic subjects. Our results do not favor their use as valid markers of human cholesterol absorption.

Study design
Our cross-sectional study consisted of a cholesterol absorption measurement according to the plasma dual-isotope method ( 20 ) in 80 mildly hypercholesterolemic adults, who were selected for this measurement based on their plasma campesterol/TC concentrations. This study was approved by the Institutional Review Board of the Academic Medical Center, Amsterdam, the Netherlands. Each participant gave written informed consent.

Selection of subjects
Subjects who participated in previous studies at our department were invited for screening or were recruited via advertisements in local newspapers. Caucasian subjects 18 years of age or older with plasma low-density lipoprotein cholesterol (LDL-C) concentrations between 3.0 and 5.0 mmol/l were considered eligible if they did not meet the following exclusion criteria: use of statins or consumption of plant sterol-or stanol-enriched food products; a history of arterial disease, including unstable angina, myocardial infarction, transient ischemic attack, or a cerebrovascular accident; diabetes mellitus; thyroid illness; uncontrolled hypertension; familial hypercholesterolemia (FH), diagnosed either by genotyping or by WHO diagnostic criteria; plasma triglyceride (TG) concentrations > 3.0 mmol/l; body mass index (BMI) > 30 kg/m 2 ; or excessive alcohol consumption. Women had to be postmenopausal without hormone-replacement therapy.
According to our power calculation described below, we set out to include at least 160 mildly hypercholesterolemic subjects in whom plasma noncholesterol sterol concentrations were measured. These subjects were stratifi ed according to their plasma cholesterol-adjusted campesterol (campesterol/TC) concentrations. Participants were invited for the cholesterol absorption measurement, starting with those with the highest and the lowest campesterol/TC concentrations, respectively. Inclusion was ceased after the fi rst 80 subjects with a successfully performed plasma dual-isotope measurement. Study design and the selection of participants are outlined in Fig. 1 .

Measurements of cholesterol absorption
Cholesterol absorption was measured according to the plasma stable dual-isotope method ( 2 ). In short, after an overnight fast, blood was drawn (T = 0 h) and subjects received an intravenous dose of 30 mg [3,4-13 C 2 ] cholesterol (Isotec, Miamisbury, OH) dissolved in 22 ml 10% Liposyn III (Hospira Inc., Lake Forest, IL) over a period of 20 min. The 13 C 2 -cholesterol was prepared for injection by dissolving the tracer in warm USP ethanol into a clear solution under sterile conditions. Thereafter, the isotope/ ethanol mixture was added to 10% Liposyn III for a total administrable volume of 22 ml. Immediately after infusion of the intravenous tracer dose, subjects ingested 50 mg of [25,26,26,26, 27,27,27-D 7 ] cholesterol (Cambridge Isotope Laboratories, Andover, MA), administered in a stomach-soluble gelatin capsule together with a standardized breakfast, consisting of 2 slices of whole wheat bread with 15 g of margarine, 40 g cheese, 150 ml semi-skimmed milk, and coffee with a small slice of gingerbread. and with plasma triglyceride concentrations ( R = Ϫ 0.19, P = 0.014 for campesterol/TC; R = Ϫ 0.27, P < 0.001 for sitosterol/TC; R = Ϫ 0.194, P = 0.010 for cholestanol/TC; and R = 0.28, P < 0.001 for lathosterol/TC ).

Plasma dual-isotope method
Population characteristics and plasma noncholesterol sterol levels. Thirty-nine subjects with low campesterol/TC concentrations and cholesterol absorption in the entire study group were analyzed by means of multiple regression analysis, with corrections for BMI and triglyceride concentrations, as those differed between subjects with high and low campesterol/TC concentrations. Analyses were performed by use of SPSS 15.0 for Windows software (SPSS Inc., Chicago, IL).

Baseline characteristics
Of 244 subjects, 175 met the inclusion and exclusion criteria ( Table 1 ). Plasma campesterol/TC concentrations ranged from 0.45 to 4.43 g/mg; sitosterol/TC from 0.31 to 2.95 g/mg; and cholestanol/TC from 0.31 to 2.55 g/mg. Lathosterol/TC levels varied from 0.32 to 2.75 g/mg. We found signifi cant negative correlations between plasma absorption markers on the one hand and plasma lathosterol/TC concentrations on the other ( R = Ϫ 0.

DISCUSSION
The present study demonstrates that plasma plant sterol concentrations are not associated with measured cholesterol absorption in a population of 80 mildly hypercholesterolemic subjects. These markers were unable to distinguish between "high and low cholesterol absorbers," as we found comparable FCA rates in individuals despite 3-fold varying plasma campesterol/TC concentrations. We chose to select "high and low absorbers" based on cholesterol-adjusted plant sterol levels and campesterol in particular, since this sterol is more readily absorbed ( 16 ).
concentrations and 41 subjects with high concentrations were subjected to the plasma dual-isotope measurement ( Table 2 ). By defi nition, plasma noncholesterol sterol concentrations differed significantly between the two groups ( There were no signifi cant differences between men and women (data not shown).

Plasma noncholesterol sterols and cholesterol absorption.
Mean fractional cholesterol absorption as measured by the plasma dual-isotope method was 24% ± 14, ranging from 1 to 73%. As average cholesterol absorption was lower compared with the literature [i.e., 24% instead of 40 to 50% ( 1, 2 )], we evaluated whether this difference might be due to lower bioavailability of the oral D 7 -cholesterol tracer caused by our method of administration. We could not substantiate this (see supplementary data).
Measured FCA did not differ signifi cantly between subjects with high and low campesterol/TC concentrations ( Fig. 2 ). There were no signifi cant differences between men and women (data not shown). Measured FCA was not associated with plasma campesterol/TC concentrations: ␤ = 0.13, P = 0.30, adjusted for BMI and plasma triglyceride concentrations ( Fig. 3 ). This lack of association with Data are presented as means ± SD, median [range] or number (n). Analyses were performed with independent sample t -tests. Triglyceride data were skewed and log-transformed prior to testing; however, untransformed medians and range are presented. * P < 0.001; ** P = 0.05.  to lower bioavailability of the oral D 7 -cholesterol tracer, caused by the method of administration (see supplementary data). Furthermore, it is probably not due to population-specifi c characteristics, as we found similar FCA rates in another population of 15 mildly hypercholesterolemic Dutch men (unpublished data). Measured FCA was associated with plasma cholestanol/TC concentrations and negatively with plasma lathosterol/TC concentrations, which is also in line with previous fi ndings ( 6,25 ).
Despite the aforementioned affi rmative fi ndings, a sample size large enough to detect a clinically relevant association, and the fact that our study was designed to answer the specifi c question whether cholesterol absorption differs between subjects with high and low plasma campesterol/TC concentrations, we failed to demonstrate any association with measured cholesterol absorption. This absence of a direct correlation between plasma plant sterol levels and cholesterol absorption may not be surprising when the metabolic routes of the different sterols are compared. Despite their high degree of structural similarity ( 37 ), there are numerous differences in the metabolic fate of plant sterols compared with cholesterol within the human body. Plant sterols are not endogenously synthesized and derive strictly from the diet. They are absorbed via the Niemann-Pick C1 Like 1 transporter in the proximal small intestine, but they are a very poor substrate for ACAT2 and, hence, are not incorporated in chylomicrons. The resulting accumulation of the plant sterols in the enterocyte activates effl ux into the lumen via the ABCG5/G8 heterodimer transporter ( 38 ). Instead of following the chylomicron route, plant sterols may be transported to the blood via ABCA1 (39)(40)(41). The small amounts that reach the liver are more rapidly secreted into the bile compared with cholesterol, probably due to a low plasma and hepatic esterification rate ( 42,43 ). This complex, but efficient effl ux machinery results in plasma plant sterol levels that are approximately 200-fold lower compared with cholesterol in normal individuals ( 8,43 ). These low levels are not only determined by intestinal absorption but also, or possibly even more, by hepatic clearance. For instance, a sitosterolemia patient, characterized by 100-fold elevated plasma plant sterol concentrations due to a mutation in the ABCG5/G8 transporter, exhibited completely normalized plasma plant sterol levels after liver transplantation, despite a persistent intestinal defective ABCG5/G8 transporter ( 44 ). Clearly, all steps involved in plant sterol transport may control their plasma concentrations and considering the fact that kinetic parameters for both NPC1L1 and ABCG5/G8 differ between plant sterol and cholesterol, good correlations between plant sterol and cholesterol absorption and metabolic turnover are not to be expected.
To investigate potential relations between activity of plant sterol transporter metabolism and their plasma levels, we considered several factors contributing to variations in plasma plant sterol concentrations, as recently reviewed ( 31 ). These include dietary intake of plant sterols, gender, polymorphisms in ABCG5/G8 , presence of metabolic syndrome, and apoE phenotypes. Dietary intake of plant In addition, we also evaluated associations between cholesterol absorption and other cholesterol-adjusted noncholesterol sterols, including cholestanol concentrations and lathosterol/campesterol ratios, as well as absolute values. Plasma cholestanol/TC and lathosterol/TC concentrations were signifi cantly associated with measured FCA.
Our fi ndings are in stark contrast to current clinical practice, in which plasma plant sterol concentrations are used as reliable markers of cholesterol absorption. This was fi rst suggested by a study of 17 individuals from two families with familial hypertriglyceridemia ( 5 ). A signifi cant association was found between plasma campesterol/TC concentrations and fractional cholesterol absorption rates ( R = 0.73). A similarly high correlation was found in 37 siblings of low and high absorbers ( 23 ). However, these were both relatively small-scaled studies in related individuals, which might have strengthened the observed associations, as it was shown that plasma plant sterol concentrations are heritable to a certain degree ( 24 ). Evaluation of these markers in larger populations of unrelated individuals yielded signifi cantly weaker correlations: R = 0.32 for plasma cholestanol levels in 61 middle-aged men ( 6 ), and R = 0.53, P = 0.01 for plasma cholestanol in a population of men with both high and low glucose concentrations ( 25 ). Although plasma plant sterols were measured, associations with measured cholesterol absorption were not reported. Similar correlations were found in 63 middle-aged men ( R = 0.42 for campesterol and R = 0.30 for sitosterol) ( 12 ), as well as in a combined population of 22 lean and obese subjects ( R = 0.48 for campesterol and R = 0.44 for sitosterol) ( 26 ). In other words, merely 10 to 25% of cholesterol absorption could be explained by plasma plant sterol concentrations in these populations. Despite the modest associations, plasma plant sterol concentrations have been used in numerous publications as a measure for FCA ever since. For example, in the past two years, 37 studies used plasma plant sterols as indirect markers of cholesterol absorption (pubmed.gov).
Negative studies are scarce and only concern specifi c, small-sized populations: serum plant sterol levels did not refl ect cholesterol absorption in mildly hypercholesterolemic males consuming plant stanol-enriched food products ( 27 ), in postmenopausal women with coronary artery disease ( 28 ), or in children with Smith-Lemli-Opitz syndrome ( 29 ). We report the fi rst adequately powered, negative study in a randomly selected population of middle-aged, mildly hypercholesterolemic but otherwise healthy subjects. Baseline characteristics of our subjects were similar to previous study populations in which positive associations between plasma plant sterols and cholesterol absorption were found ( 6,12 ). This also applies to ingested dietary plant sterols ( 30 ), the observed range of plant sterol concentrations ( 12,23,24,31 ), as well as their negative correlations with plasma lathosterol levels, BMI, and triglyceride levels ( 6,(32)(33)(34)(35). Measured cholesterol absorption rates ranged from 1 to 73%, which is in line with literature (reviewed in Ref. 36 ), although mean cholesterol absorption was lower as compared with other studies; FCA averaged 24% instead of 40 to 50%. Of note, the lower mean cholesterol absorption could not be attributed in the interpretation and correct use of plasma plant sterol concentrations, as well as differences between cholesterol and plant sterol metabolism, disqualify their current use as reliable markers of cholesterol absorption. sterols did not differ between the high and low campesterol/TC groups and none of our subjects consumed plant sterol-enriched food products, although the relation between dietary intake of plant sterols and serum plant sterol concentrations is not very strong (45)(46)(47)(48)(49). Furthermore, there were no differences in gender distribution between the high and low campesterol/TC group. The ABCG8 D19H variant has been associated with decreased plasma plant sterol concentrations, possibly mediated by a gain of function of the ABCG8 transporter ( 21,50 ). As each of our subjects had participated in a previous study in which we investigated associations between fi ve nonsynonymous polymorphisms in ABCG5/G8 and plasma sterol concentrations ( 21 ), we are able to state that we found no differences in distribution of these variants between the high and low campesterol/TC groups (data not shown), although our current study population was too small for a genetic association study. We did not fi nd any relation between these variants and measured cholesterol absorption, and to the best of our knowledge, there are no reports showing hampered cholesterol absorption in subjects carrying this phenotype. We cannot rule out the possibility that a skewed distribution of apoE phenotypes between the high and low campesterol/TC group might have infl uenced our results. We did not genotype subjects for apoE phenotype, as reports on the associations between apoE variants and cholesterol absorption have been inconsistent ( 2,31,51 ), the prevalence of the apoE4 genotype in the Netherlands is low ( 52 ), and apoE polymorphisms also affect serum cholesterol and triglyceride concentrations ( 53 ), thereby complicating the implication of certain associations with plasma plant sterols. Finally, although none of our subjects had type 2 diabetes or met the clinical criteria for metabolic syndrome, subjects with low plasma campesterol/TC concentrations did express higher BMI and plasma triglyceride concentrations in our study, which is in line with the literature ( 6,(32)(33)(34)(35). We corrected our results for this difference. Moreover, signifi cant weight loss over a period of two years has been shown to signifi cantly alter FCA rates and plasma plant sterol concentrations ( 54 ), illustrating that both cholesterol absorption and plant sterol concentrations are dynamic parameters, infl uenced by modifi able patient characteristics, such as obesity or insulin resistance. At present, plasma plant sterols are often handled as stable parameters in a given individual ( 24 ), often with unjustifi ed disregard toward possible dynamic confounders. Lack of consensus throughout the literature regarding the correct use of plant sterols as markers of cholesterol absorption further underlines the scientifi c inconclusiveness. This was recently acknowledged by the founders of these markers who, following review of the literature on this subject, stated that "it is worth using several instead of only one absorption sterol marker for making conclusions of altered absorption of cholesterol, and even then, the presence of at least some absolute measurement is valuable" ( 13 ).
In conclusion, plasma plant sterol concentrations do not refl ect cholesterol absorption in a relatively large population of mildly hypercholesterolemic subjects. Diffi culties