Design: Follow up of patients with coronary heart disease in whom baseline ratios were related to major coronary events.

Setting: Four universities in Finland.

Subjects: A subgroup of 868 patients with coronary heart disease selected from the Scandinavian simvastatin survival study.

Intervention: Treatment with simvastatin or placebo.

Main outcome measures: Serum concentrations of low density lipoprotein and high density lipoprotein cholesterol, total triglyceride concentration, and cholesterol:cholestanol ratio. Major coronary events.

Results: With increasing baseline quarter of cholestanol distribution the reduction in relative risk increased gradually from 0.623 (95% confidence interval 0.395 to 0.982) to 1.166 (0.791 to 1.72). The risk of recurrence of major coronary events increased 2.2-fold (P<0.01) by multiple logistic regression analysis between the lowest and highest quarter of cholestanol. The ratio of cholestanol was related inversely to the body mass index and directly to high density lipoprotein cholesterol and triglyceride concentrations but their quarters of distribution were not related to risk reduction.

Conclusions: Measurement of serum cholestanol concentration revealed a subgroup of patients with coronary heart disease in whom coronary events were not reduced by simvastatin treatment. Thus, patients with high baseline synthesis of cholesterol seem to be responders whereas those with low synthesis of cholesterol are non-responders.

The Scandinavian simvastatin survival study (4S) showed that simvastatin clearly decreases serum cholesterol concentration and considerably reduces mortality from all causes and coronary events and major coronary events in patients with coronary heart disease.¹ As the reduction in the relative risk of major coronary events was not associated with the baseline serum lipid concentrations in the survival study,² however, we considered that reduction in relative risk may be related to baseline intestinal absorption or to endogenous synthesis of cholesterol. From among the serum non-cholesterol sterols, the cholesterol precursor sterols—that is, lanosterol and other methyl sterols and demethylated cholestenol, desmosterol, and lathosterol—are directly related to cholesterol synthesis,^3–9 especially in the liver,¹⁰ while cholestanol and plant sterols—that is, campesterol and sitosterol—are directly related to cholesterol absorption.^9,11 Accordingly, we would expect quantification of these sterols to reveal subgroups with high or low absorption or synthesis of cholesterol. To this end we examined whether measurement of baseline serum cholestanol concentration in the Finnish coronary subpopulation of the survival study predicts the extent to which simvastatin could reduce the risk of major coronary events.

The whole study population of the survival study and the methods used have been reported previously.^1,12 The present study population included the Finnish subgroup of 868 patients with coronary heart disease selected from 1374 candidates for the 4444 participants of the original study and were randomised to placebo (n=434) or simvastatin 20-40mg/day (n=434) for 5 years and 3 months. Distributions of age, sex, and lipid concentrations were similar in the placebo and simvastatin subgroups and were comparable with those in the main study. The secondary end point of the survival study—major coronary events, including coronary deaths, non-fatal myocardial infarctions, and revascularisation procedures—was used for analytical purposes.

Concentrations of total and low and high density lipoprotein cholesterol and triglycerides were analysed at the central laboratory from serum obtained after an overnight fast.¹² Concentrations of cholesterol and non-cholesterol sterol were measured in two saponified baseline serum samples by gas-liquid chromatography on a 50m long SE-30 capillary column (polydimethylsiloxane), with 5α-cholestane as an internal standard^13,14 by running the samples from a single patient in a single batch from frozen stored serum samples. The non-cholesterol sterols are transported in serum by lipoproteins, about 70% by low density lipoprotein, so that the decrease in concentration of low density lipoprotein cholesterol by simvastatin also changes the serum concentration of cholestanol and other non-cholesterol sterols. We have therefore expressed the values of these sterols in terms of mmol/mol of cholesterol—that is, as ratios to cholesterol. The values given are means of two baseline determinations.

Statistical analysis —The patients were ranked according to quarters of baseline distribution of body mass index (kg/m²), serum concentrations of total and low density and high density lipoprotein cholesterol, total triglycerides, and cholestanol to cholesterol ratios. We counted major coronary events during the follow up of 5 years 3 months in the quarters of these variables and calculated relative risks (95% confidence intervals) between simvastatin and placebo groups.¹⁵ Analysis of variance was used to compare means of continuous variables in the quarters. Multiple logistic regression analysis was used to test the associations between different prognostic variables and the occurrence of major coronary events. The goodness of fit and consistency with logistic function were evaluated with Hosmer-Lemeshow and C C Brown tests, respectively. The statistical analyses were carried out with bmdp statistical software package.¹⁶

In the Finnish subgroup the mean changes in lipid concentrations caused by simvastatin—that is, −28, −35, 8, and −15% for total, low density lipoprotein, and high density lipoprotein cholesterol and triglycerides, respectively—were similar to those seen in the main study population.¹

The reduction in the risk of major coronary events in the whole subgroup studied (144 v 114 events in placebo v simvastatin group, respectively) was 21% (0.790; 95% confidence interval 0.642 to 0.971). When we analysed gradually increasing body mass index and serum lipid concentrations from the first to the fourth quarter only cholestanol:cholesterol ratio exhibited percentages of recurrent events, gradually increasing from 21% to 35% (values shown only for cholestanol) with the increasing baseline sterol quarters in the simvastatin but not in placebo group (footnote to table 1). The corresponding relative risk increased gradually from 0.623 (0.395 to 0.982) to 1.166 (0.791 to 1.720). Thus, the relative risk of major coronary events was increased by 16.6% (−20.9% to 72.0%) in the highest quarter of the cholestanol ratio and significantly reduced by 37.7% (−60.5% to −1.8%) in the lowest quarter. The concentrations of cholestanol showed inconsistent changes in the relative risk ratios. Age and the concentrations of total and low density lipoprotein cholesterol were similar in all the quarters separated according to the ratios of cholestanol (table 2). The serum concentrations of high density lipoprotein cholesterol increased whereas the serum concentrations of triglycerides and body mass index values decreased. The ratios of cholestanol to cholesterol were negatively related to those of cholesterol precursor sterols (cholestanol, desmosterol, and lathosterol) and strongly positively related to those of plant sterols (campesterol and sitosterol; data not shown).

Table 1 Relative risk of major coronary events by simvastatin in patients defined by baseline quarters of total, low density lipoprotein and high density lipoprotein cholesterol and triglyceride concentrations, body mass index, and cholestanol proportion

Table 2 Mean (SD) baseline values for different variables defined by quarters of distribution of cholestanol at baseline (<107, 107-126, 127-148, >148)

As in stepwise logistic regression analysis the continuous cholestanol ratio was the only variable of table 1 significantly entered into the model (odds ratio 1.01; 1.00 to 1.02), the association between cholestanol and the occurrence of major coronary events was tested by using a fixed logistic model with three variables: treatment group, cholestanol as an ordinal scale variable based on quarters, and treatment group*cholestanol interaction. In this model both the goodness of fit (Hosmer-Lemeshow test, P value 0.983) and the consistency of logistic function (C C Brown test, P value 1.000) were good. For the treatment group the odds ratio was 3.29 (1.53 to 7.06) and the Wald’s test value 3.05 (P<0.001). For cholestanol the odds ratio was 1.31 (1.07 to 1.61). The risk of recurrence of major coronary events increased 2.2-fold between the lowest and highest quarters of the distribution of cholestanol ratios by multiple logistic regression analysis (P<0.01; Wald’s test value 2.63). For the interaction term the odds ratio was 0.73 (0.56 to 0.86) and Wald’s test −2.29 (P<0.02).

Multivariate logistic regression analysis indicated that the baseline lipid concentrations did not contribute to the results shown by the cholestanol ratio. The higher the quarter of the cholestanol ratio the greater was the risk of major coronary events. The risk of recurrence of major coronary events was 2.2-fold between the lowest and highest quarters of cholestanol, a finding not applicable to total, low density lipoprotein, or high density lipoprotein cholesterol. Thus, simvastatin treatment of patients with a low baseline cholestanol ratio predicts a clear reduction in risk of major coronary events, this improvement not being seen in patients with high baseline cholestanol ratio. Similar findings were also observed when the cholestanol ratios were used as continuous or dichotomised variables in statistical analysis. Thus, quantification of baseline cholestanol in patients with coronary heart disease would give important new information to clinicians or practitioners for evaluation of the future success of the statin treatment. The ability to predict the ineffectiveness of the relatively expensive statins in the treatment of coronary patients supports the use of such measurements. Today, automated instrumentation allows comparatively rapid gas liquid chromatographic measurements, each analysis measuring concentrations of cholesterol and several non-cholesterol sterols; the most predictive sterol for evaluation of treatment seems to be cholestanol.

The results of this study were presented at the 11th international symposium on atherosclerosis in Paris, 1997. An abstract is published in Atherosclerosis 1997;134:48(150). The Finnish 4S investigators were TA Miettinen, H Vanhanen, TE Strandberg, K Hölttä, H Luomanmäki, T Pekuri, A Vuorinen (Helsinki University Hospital); A Pasternack, H Oksa, L Siitonen, R Rimpi (Tampere University Hospital); YA Kesäniemi, M Lilja, T Korhonen, A Rantala, M Rantala, M Savolainen, O Ukkola, L Laine, L Virkkala (Oulu University Hospital); K Pyörälä, S Lehto, A Rantala, H Miettinen, A Salokannel, R Räisänen (Kuopio University Hospital).