P Salo 1,2 *, J Viikari 3,4, L Rask-NissilaÈ 1,MHaÈmaÈlaÈinen 5,6,TRoÈnnemaa 3, R SeppaÈnen 7 and O Simell 2

Transcription

1 European Journal of Clinical Nutrition (1999) 53, 927±932 ß 1999 Stockton Press. All rights reserved 0954±3007/99 $ Effect of low-saturated fat, low-cholesterol dietary intervention on fatty acid compositions in serum lipid fractions in 5-year-old children. The STRIP project P Salo 1,2 *, J Viikari 3,4, L Rask-NissilaÈ 1,MHaÈmaÈlaÈinen 5,6,TRoÈnnemaa 3, R SeppaÈnen 7 and O Simell 2 1 The Cardiorespiratory Research Unit, University of Turku; 2 Department of Pediatrics, University of Turku; 3 Department of Medicine, University of Turku; 4 Department of Medical Chemistry, University of Turku; 5 Department of Clinical Chemistry, University of Turku; 6 The Joint Clinical Biochemistry Laboratory of University of Turku, Turku University Hospital and Wallac Ltd, Turku; 7 The Research and Development Unit of the Social Insurance Institution, Turku, Finland Objective: To evaluate the effect of dietary low-saturated fat, low-cholesterol intervention on fat intake and fatty acid compositions in serum cholesterol ester (CE), phospholipid (PL) and triglyceride (TG) fractions in veyear-old children. Design and subjects: The STRIP project is a prospective, randomised intervention project in which 1062 sevenmonth-old infants were recruited from the well-baby clinics. 764 children participated in the 5-year follow-up; 202 of them were randomly selected for this study. Diet was assessed with 4-d dietary records. Serum CE, PL and TG fatty acid compositions were analysed with gas-liquid chromatography. Results: Saturated fat intake of intervention children (mean (con dence interval)) (girls 11.9 (11.2 ± 12.6) % of energy intake (E%); boys 12.5 (11.9 ± 13.1)) was lower than that of the control children (girls 14.4 (13.7 ± 15.2) E%; boys 15.0 (14.3 ± 15.8) E%) (P ˆ for the difference between intervention and control groups). The intake of unsaturated fat differed only slightly. Dietary ratios of polyunsaturated to saturated fatty acids (PS ratios) of the intervention and control diets were 0.44 and 0.33, respectively (P ˆ ). Furthermore, serum cholesterol concentrations of the intervention and control children differed (4.28 (4.13 ± 4.43) mmol=l vs 4.49 (4.35 ± 4.63)mmol=L; P ˆ 0.04). Relative proportion of saturated fatty acids in serum TG was lower (34.9% vs 36.3%; P ˆ 0.04) and that of n-6 polyunsaturated fatty acids higher (13.9% vs 12.4%; P ˆ ) in the intervention than in the control children, whereas serum CE and PL fatty acid compositions of intervention and control groups were closely similar. However, intake of linoleic acid correlated better with serum linoleic acid relative content in the CE fraction (r ˆ 0.36; P ˆ ) than in the PL (r ˆ 0.27; P ˆ ) or in the TG (r ˆ 0.23; P ˆ ) fraction Conclusions: Intervention resulted in decreased intake of saturated fatty acids and lowered serum total and LDL cholesterol concentrations. Of serum lipid fractions, TG fatty acid composition was the most sensitive and parallelled the ndings in dietary food records. Descriptors: children; cholesterol esters; dietary intervention; fatty acids; phospholipids; triglycerides Introduction Eating habits are formed during childhood and have a potential lifelong effect on serum lipid levels, and thus, indirectly, on coronary heart disease risk later in adulthood. STRIP project was launched to evaluate if healthy eating habits can be adopted permanently by starting dietary education early in life, that is, already during infancy. During the rst years of life small but signi cant differences have evolved in the diets and also in serum lipoprotein concentrations between intervention and control children (Lapinleimu et al, 1995; Niinikoski et al, 1996). *Correspondence: Pia Salo, Cardiorespiratory Research Unit, University of Turku, Kiinamyllynkatu 10, Turku, Finland. Guarantor: Pia Salo. Contributors: Olli Simell, Tapani RoÈnnemaa and Jorma Viikari have originally designed the STRIP study and are responsible for study coordination. Pia Salo and Jorma Viikari are responsible for the analysis and interpretation of the fatty acid compositions and writing of the paper. Leena Rask-NissilaÈ examined the children at the time point of 5 years and took the blood samples. Ritva SeppaÈnen is in charge of nutritional planning and analyses. Mauri HaÈmaÈlaÈinen is responsible for GLC equipment quality control. Received 26 February 1998; revised 1 June 1999; accepted 7 June 1999 Cholesterol ester (CE) fatty acid compositions have been evaluated of the children since the beginning of the study because cholesterol ester fatty acid compositions have been regarded as a reliable marker of the fatty acid composition of the diet (Moilanen et al, 1992; Ma et al, 1995). In the STRIP study, the association between the fatty acid composition of the diet and the fatty acid composition of serum cholesterol esters was particularly clear when the majority of the infants were still breast-fed (Salo et al, 1997), but later the effects of dietary intervention on serum CE fatty acid composition in the project children has been minimal (Salo et al, 1999). Meanwhile, serum cholesterol and nonhigh-density-lipoprotein (non-hdl) cholesterol concentrations have been lower in intervention children during the follow-up when compared with control children (Niinikoski et al, 1996; Lapinleimu et al, 1995). Besides in cholesterol esters (Glatz et al, 1989; Sarkkinen et al, 1994), fatty acid compositions can also be analysed in other lipid fractions in serum, that is, phospholipids (PL) (Gustafsson et al, 1994; Ma et al, 1995), triglycerides (TG) (Stein et al, 1982), and also in blood cell membranes (James et al, 1993; Sarkkinen et al, 1994; Mantzioris et al, 1995; Tynan et al, 1995), adipose tissue

2 928 (London et al, 1991) and even in cell membranes derived from cheek mucosal cells (McMurchie et al, 1984). Adipose tissue is regarded a reliable marker of long-term fat intake representing mean fat intake of 2 ± 3 y duration whereas the others re ect intake of fat during the preceding few weeks (CE, PL, cell membrane lipids (Katan et al, 1997; Glatz et al, 1989)) or days (TG (Moilanen, 1987)). The 5-year visit of the STRIP children was the rst time point when fasting blood samples were taken. This allowed for the rst time in the project the analysis of fasting serum triglyceride concentration and calculation of low-density lipoprotein cholesterol concentration in addition to total cholesterol and high-density lipoprotein cholesterol concentrations. Furthermore, the analysis of fasting TG fatty acid composition could now be determined. Therefore, we have analysed the fatty acid compositions of CE, PL and TG fractions to evaluate whether one of these fractions is superior to the others in re ecting and differentiating between fat intakes of the two groups in this project. Subjects and methods The STRIP project (Special Turku Coronary Risk Factor Intervention Project) has been described in detail previously (Niinikoski et al, 1996; Lapinleimu et al, 1995). In short, the project aims at decreasing exposure of the intervention children to known diet-induced and other environmental atherosclerosis risk factors. At the age of 7 months 1062 infants were randomised to the intervention (n ˆ 540) and control groups (n ˆ 522). The intervention families were seen by the counselling team at child's ages of 7, 8, 10, 13, 15, 18, 21 and 24 months and twice a year thereafter; the control families were seen continuously twice a year. At the 5-year visit, 764 children (386 intervention and 378 control) still remained in the project. The fatty acid composition in the serum cholesterol ester fraction has been analysed from the beginning of the study of a group that was selected at random at age of 13 months. The randomisation was based on successive temporal order of visits at the Cardiorespiratory Research Unit. Of those 291 children, 202 were still participating in the STRIP project at the age of 5 y (n ˆ 202; 103 intervention and 99 control); their serum samples were then analysed for CE, PL and TG fatty acid compositions. Four-day dietary records were available of 189 of them. The food records were collected prior to (that is, within two weeks of) the taking of the blood samples. The counselling team consisted of two physicians, two nutritionists and a registered nurse. Diet Intervention children. After the age of one year, the child's diet was planned to contain 30 ± 35 E% fat. The intervention families were counselled to modify child's fat intake so that saturated, monounsaturated and polyunsaturated fat would each comprise 10% of daily energy (E%). No xed diets were ordered but at each visit small changes were suggested to the diet, which would modify child's diet step by step towards the optimal composition. Use of vegetable oil or soft margarine in food preparation was recommended and detailed advice was give on how to replace products containing large amounts of saturated fat with products containing more unsaturated fat. The control families were advised to follow the dietary recommendations routinely given at the well-baby clinics in Finland. Detailed instructions how to record child's food intake (4- d food records) were given to the families. All records were reviewed by a nutritionist for completeness and accuracy item by item during each follow-up visit. Daily energy and nutrient intakes were calculated using the Micro Nutrica program developed at the Research Centre of the Social Insurance Institution, Turku, Finland. The program uses the Food and Nutrient Data Base of the Social Insurance Institution and calculates 62 nutrients of commonly used foods and dishes in Finland (Hakala et al, 1996). Control children. The control children were seen by the same counselling team twice a year. The families received only minimal and non-individualised dietary information according to the current practices at the Finnish well-baby clinics; the consumption of milk with at least 1% of fat was suggested, otherwise no detailed advice for selection of foods was given. Lipid analyses At the age of 5 y, the blood samples were now drawn for the rst time in the STRIP project after an overnight fast; all the 5-year visits were arranged so that the children were seen either at 0800 or 0830 h so that the duration of the fasting was 10 ± 12 h. Serum cholesterol and HDL lipoprotein concentrations were measured as described earlier (Lapinleimu et al, 1995; Niinikoski et al, 1996). Triglyceride concentrations were measured with a colorimetric GPO-PAP method (Merck, Darmstadt, Germany) in an automatic Olympus AU 510 analyser. The intra-assay and inter-assay coef cients of variation for the method were 3.1 ± 4.1% and 4.3 ± 4.5%, respectively, depending on the level of triglyceride concentrations. LDL was calculated according to the Friedewald formula (Friedewald et al, 1972). Serum CE, PL and TG fatty acid compositions were analysed using gas chromatography as described earlier (Salo et al, 1997). Statistical analyses The proportions of the individual serum CE, PL and TG fatty acids and the cholesterol concentration are expressed as means (con dence interval). Two-way ANOVA was used in comparison of intervention and control boys and girls. Pearson correlations were calculated between fatty acid intakes and the respective proportions of fatty acids in the serum lipid fractions (Table 4). Differences were regarded signi cant at P < Ethical approval The study was approved by the Joint Ethics Committee of the University of Turku and Turku University Central Hospital. Informed consent was obtained from all parents. Results Serum lipids and lipoproteins In this randomly selected group of the STRIP children, serum total cholesterol was lower in intervention children than in control children (Table 1). The difference was more marked between intervention and control boys. Furthermore, LDL cholesterol concentration was slightly lower in intervention than in control children whereas HDL cholesterol concentration, the ratio of HDL to total cholesterol or the triglyceride concentrations did not differ.

3 Table 1 Serum lipids in the 5-year-old STRIP project intervention (48 girls, 55 boys) and control (45 girls, 54 boys) children. Mean (95% con dence interval) 929 Intervention (n ˆ 103) P Control (n ˆ 99) Total cholesterol (mmol=l) all 4.28 (4.13 ± 4.43) (4.35 ± 4.63) girls 4.42 (4.20 ± 4.65) ns 4.52 (4.30 ± 4.74) boys 4.15 (3.95 ± 4.35) (4.27 ± 4.66) LDL-cholesterol (mmol=l) all 2.73 (2.60 ± 2.86) (2.78 ± 3.04) girls 2.87 (2.68 ± 3.06) ns 2.98 (2.79 ± 3.17) boys 2.61 (2.44 ± 2.78) (2.68 ± 3.04) HDL-cholesterol (mmol=l) all 1.21 (1.16 ± 1.26) ns 1.24 (1.19 ± 1.30) girls 1.19 (1.13 ± 1.26) ns 1.21 (1.14 ± 1.27) boys 1.22 (1.14 ± 1.30) ns 1.27 (1.19 ± 1.35) Ratio HDL=total cholesterol all 0.29 (0.27 ± 0.30) ns 0.28 (0.27 ± 0.29) girls 0.27 (0.26 ± 0.29) ns 0.27 (0.26 ± 0.29) boys 0.30 (0.28 ± 0.32) ns 0.29 (0.27 ± 0.31) Triglycerides (mmol=l) all 0.67 (0.63 ± 0.71) ns 0.67 (0.62 ± 0.71) girls 0.71 (0.65 ± 0.78) ns 0.67 (0.61 ± 0.73) boys 0.64 (0.59 ± 0.69) ns 0.66 (0.60 ± 0.73) LDL ˆ low density lipoprotein; HDL ˆ high density lipoprotein; ns ˆ not signi cant. Dietary fatty acids Total intake of fat was 31.2E% in intervention and 33.7 E% in control children (P ˆ ). The main difference in fatty acid intake was in the intake of saturated fat (Table 2); intervention children consumed less saturated fat (12.2 E%) than control children did (14.7 E% P ˆ ). The same was true for the main saturated fatty acid, palmitic acid. The intakes of polyunsaturated fatty acids including linoleic acid differed only slightly between groups. The mean PS ratios were 0.44 and 0.33 (P ˆ ) in intervention and control children's diets, respectively. Intakes of cholesterol did not differ between groups. In Table 2 the data is given separately for girls and boys. The fatty acid compositions of serum CE, PL and TG The percentual distributions of fatty acids differed between the lipid fractions (Table 3). The only lipid fraction that differed between intervention and control groups was the TG fraction. The TG palmitic and total saturated fatty acids were lower and linoleic, g-linoleic and total n-6 fatty acids as well as a-linoleic, docosahexaenoic and total n-3 polyunsaturated fatty acids higher in intervention than in control children. Also the PS ratio of serum TG fatty acids was higher in intervention than in control children whereas CE and PL fractions showed no signi cant differences (Table 3). No differences were observed between girls and boys in respect to the fatty acid compositions (data not shown). Correlations of serum fatty acid compositions with dietary intake data Intake of saturated fat correlated best with the proportion of myristic acid in the CE fraction (r ˆ 0.27; P ˆ ) (Table 4). Intake of monounsaturated fat was best re ected in CE and PL linoleic acid (r ˆ 0.24; P ˆ and 0.24; P ˆ 0.001) but not in oleic acid of any serum fraction although oleic acid is the main monounsaturated fatty acid in diet and in serum lipids. Overall, the best correlations between diet and serum lipid fractions were between polyunsaturated fat intake and CE linoleic acid (0.36; P ˆ ); the respective correlations were 0.26 in the PL fraction (P ˆ ) and 0.25 in the TG fraction (P ˆ ). Discussion This study shows that in 5-year-old participants of the STRIP project 2.5 E% difference in saturated fat intake, 0.5 E% difference in polyunsaturated fat intake, a difference of 0.11 in the PS ratio and 5% difference in serum cholesterol concentration between the intervention and control children was associated with only minor differences in the relative fatty acid concentrations in serum CE, PL and TG fractions. No difference was observed in cholesterol intake between intervention and control children. The CE and PL fatty acid compositions were close to identical Table 2 Comparison of dietary fat intake by the intervention and the control children at the age of 5 years. Mean (95% con dence interval) Intervention Control Girls Boys Girls Boys P P Diet n ˆ 46 n ˆ 49 n ˆ 44 n ˆ 50 effect of group effect of gender Saturated fat (E%) 11.9 (11.2 ± 12.6) 12.5 (11.9 ± 13.1) 14.4 (13.7 ± 15.2) 15.0 (14.3 ± 15.8) ns Palmitic acid (E%) 5.5 (5.2 ± 5.9) 6.0 (5.8 ± 6.3) 6.8 (6.4 ± 7.1) 7.1 (6.7 ± 7.4) Monounsaturated fat (E%) 10.4 (9.9 ± 10.9) 10.8 (10.3 ± 11.2) 10.8 (10.2 ± 11.3) 11.3 (10.7 ± 11.9) ns ns Polyunsaturated fat (E%) 5.2 (4.9 ± 5.5) 5.3 (5.0 ± 5.5) 4.8 (4.3 ± 5.3) 4.8 (4.5 ± 5.1) ns ns Linoleic acid (E%) 3.8 (3.5 ± 4.1) 4.1 (3.8 ± 4.4) 3.6 (3.2 ± 4.0) 3.8 (3.5 ± 4.0) ns ns PS ratio 0.45 (0.42 ± 0.49) 0.44 (0.40 ± 0.47) 0.35 (0.30 ± 0.39) 0.33 (0.30 ± 0.36) ns Cholesterol mg=1000kcal 127 (113 ± 141) 117 (107 ± 125) 124 (112 ± 137) 128 (117 ± 140) ns ns E% ˆ % of daily energy; ns ˆ not signi cant.

4 930 Table 3 interval) Fatty acid composition of serum CE, PL and TG in 5-year-old intervention (n ˆ 103) and control (n ˆ 99) children. Mean (95% con dence Cholesterol esters Phospholipid fatty acids Triglyceride fatty acids Fatty acid % Intervention Control P Intervention Control P Intervention Control P Saturated 13.4 (13.2 ± 13.7) 13.3 (13.1 ± 13.6) ns 45.5 (45.1 ± 46.0) 45.6 (44.9 ± 46.2) ns 34.9 (33.8 ± 36.0) 36.3 (35.5 ± 37.2) : (0.79 ± 0.86) 0.87 (0.83 ± 0.92) ns 0.39 (0.37 ± 0.41) 0.41 (0.38 ± 0.43) ns 2.01 (1.84 ± 2.18) 2.18 (2.01 ± 2.34) ns 16: (11.1 ± 11.5) 11.2 (11.0 ± 11.4) ns 28.2 (27.9 ± 28.5) 28.5 (28.0 ± 29.0) ns 27.4 (26.6 ± 28.2) 28.4 (27.8 ± 29.1) : (1.30 ± 1.41) 1.28 (1.23 ± 1.33) (16.7 ± 17.2) 16.6 (16.3 ± 16.9) ns 5.51 (5.21 ± 5.80) 5.72 (5.43 ± 6.01) ns Monounsaturated 25.0 (24.4 ± 25.5) 25.4 (24.9 ± 26.0) ns 15.0 (14.7 ± 15.4) 15.2 (14.9 ± 15.5) ns 49.5 (48.6 ± 50.5) 50.2 (49.6 ± 50.8) ns 16: (1.95 ± 2.24) 2.22 (2.10 ± 2.34) ns 0.42 (0.39 ± 0.45) 0.42 (0.42 ± 0.45) ns 2.69 (2.54 ± 2.84) 2.78 (2.62 ± 2.93) ns 18: (22.4 ± 23.3) 23.2 (22.8 ± 23.7) ns 14.6 (14.3 ± 15.0) 14.8 (14.5 ± 15.1) ns 46.9 (45.9 ± 47.9) 47.4 (46.8 ± 48.0) ns Polyunsaturated, n (58.8 ± 60.1) 59.1 (58.5 ± 59.7) ns 34.5 (33.9 ± 34.8) 34.6(33.8 ± 35.2) ns 13.9 (13.2 ± 14.6) 12.4 (11.8 ± 13.0) :2n ± (52.0 ± 53.4) 52.2 (51.5 ± 52.9) ns 22.7 (22.3 ± 23.2) 22.8 (22.3 ± 23.4) ns 11.8 (11.1 ± 12.5) 10.2 (9.58 ± 10.7) :3n ± (0.58 ± 0.68) 0.68 (0.62 ± 0.74) ns 0.06 (0.05 ± 0.06) 0.06 (0.05 ± 0.07) ns 0.24 (0.21 ± 0.27) 0.20 (0.17 ± 0.22) :3n ± (0.59 ± 0.66) 0.65 (0.60 ± 0.69) ns 2.88 (2.77 ± 2.99) 2.85 (2.71 ± 2.98) ns 0.20 (0.18 ± 0.23) 0.17 (0.15 ± 0.19) ns 20:4n ± (5.25 ± 5.75) 5.59 (5.35 ± 5.83) ns 8.79 (8.44 ± 9.14) 8.84 (8.45 ± 9.23) ns 1.54 (1.42 ± 1.66) 1.66 (1.52 ± 1.81) ns Polyunsaturated, n ± (2.00 ± 2.26) 2.12 (2.00 ± 2.25) ns 4.99 (4.69 ± 5.28) 4.63 (4.31 ± 4.95) ns 1.86 (1.68 ± 2.04) 1.60 (1.48 ± 1.73) :3n ± (0.85 ± 0.92) 0.86 (0.82 ± 0.91) ns 0.34 (0.32 ± 0.36) 0.34 (0.32 ± 0.37) ns 1.16 (1.03 ± 1.28) 0.94 (0.86 ± 1.02) :5n ± (0.68 ± 0.84) 0.81 (0.73 ± 0.89) ns 0.91 (0.83 ± 0.99) 0.87 (0.80 ± 0.93) ns 0.19 (0.16 ± 0.22) 0.19 (0.15 ± 0.22) ns 22:6n ± (0.44 ± 0.53) 0.45 (0.41 ± 0.49) ns 3.73 (3.49 ± 3.98) 3.42 (3.15 ± 3.69) ns 0.45 (0.38 ± 0.51) 0.35 (0.28 ± 0.41) 0.04 PS ratio 4.65 (4.51 ± 4.77) 4.66 (4.53 ± 4.78) ns 0.87 (0.85 ± 0.89) 0.87 (0.84 ± 0.90) ns 0.49 (0.45 ± 0.53) 0.42 (0.39 ± 0.45) ns ˆ not signi cant. Table 4 Pearson correlations of dietary intake of fat with serum fatty acids (CE, PL, TG), n ˆ 189 Dietary intake Monounsaturateunsaturated Poly- Saturated Linoleic Serum fatty acid fat fat fat acid 14:0 CE 0.27*** ns ** ** PL 0.20* ns * * TG 0.23** ns ** ** 16:0 CE ns ns ns ns PL 0.07 ns ns ** * TG 0.13 ns ns ** *** 18:1 CE 0.09 ns ns ** ** PL 0.08 ns 0.02 ns ns ns TG 0.03 ns 0.05 ns 0.12 ns 0.17* 18:2n ± 6 CE ns 0.24** 0.36*** 0.36*** PL 0.07 ns 0.24** 0.26** 0.27** TG *** ns 0.25** 0.23** *P < 0.05; **P < 0.01; ***P < CE ˆ cholesterol ester; PL ˆ phospholipid; TG ˆ triglyceride. between intervention and control groups whereas TG fatty acids showed differences between the two groups that were in line with dietary intake data. The 2.5 E% smaller intake of saturated fat by the intervention children led to a 1.4% lower proportion of saturated fatty acids in serum TG fraction in intervention group when compared with control group. Similarly, 0.4 E% higher intake of polyunsaturated fatty acids by the intervention children was re ected by 1.6% higher proportion of n-6 polyunsaturated fatty acids in TG (P < 0.001). In addition, in serum TG fraction, the proportions of the parent fatty acids of the n-6 and n-3 series, linoleic acid and a-linoleic acid, respectively, were higher in the intervention children than in the control children. The strictly controlled dietary experiments are usually planned so that the change in only one type of fat or fatty acid can be evaluated at a time (for example Mantzioris et al 1995). The STRIP project, in contrast, is based on dietary recommendations given to free-living children and families with the intention to in uence all types of fatty acids, that is, to reduce the intake of saturated fat and increase the intake of unsaturated fat. Since the increased intake of one type of fatty acid increases the respective proportion of the fatty acid of total fatty acids, inevitably there must be a concomitant reduction in the proportion of some other fatty acid(s), which complicates evaluation of true intakes. Furthermore, the dietary changes induced in short-term intervention studies are generally much larger than those seen between the STRIP intervention and control children. For example, the mean PS ratios were 0.44 and 0.33 in the STRIP intervention and control children, respectively whereas in the short-term studies the ratios may be changed even 10-fold (Glatz et al, 1989). The relatively small differences in dietary fat quality are probably the reason why CE and PL fatty acid compositions in the two groups of children were so closely similar. In general, the best biomarkers of dietary fatty acids are those fatty acids that can not be synthesised endogenously. Linoleic acid and a-linoleic acid are such fatty acids. Saturated fatty acids, however, can be synthesised de novo from carbohydrates and proteins and can further be desaturated and elongated into longer-chain saturated and unsaturated derivatives. This endogenous synthesis makes the estimation of dietary-derived saturated and monounsaturated fatty acids by biomarkers dif cult. Hence, a larger dietary difference is required for saturated and monounsaturated fatty acid intakes than for linoleic, or a-linoleic acids to have an effect on biomarkers. This was demonstrated in studies by Zock et al, 1997) where an increase of 10E% of linoleic acid resulted in an increase of 9.3% of linoleic acid in the cholesterol ester fraction whereas equal increases in oleic acid and saturated fatty acids resulted in increases of 6.5% and 2.2%, respectively. The correlations between dietary intake of fatty acids and the fatty acids in serum lipid fractions were best for linoleic acid. The best marker of intake of polyunsaturated fatty acids (and of linoleic acid) was serum CE linoleic acid whereas the best marker of saturated fatty acid intake was serum CE myristic acid. The intake of monounsaturated fat did not correlate with either of the analysed

5 monounsaturated fatty acids, oleic or palmitoleic acid, in any of the lipid fractions whereas signi cant correlations were observed between monounsaturated fatty acid intake and CE and PL linoleic acid. Since the food records were collected within two weeks prior to taking of the blood samples for fatty acid analysis, the poor correlations between intake and serum fatty acids in the TG fraction are probably due to the sensitivity of this fraction to day-today variations in fat intake. Metabolic regulation ensures that no large variations take place in serum fatty acid compositions despite of daily uctuations in intakes. There are some major differences in the extent that individual fatty acids are incorporated in the different lipid fractions in serum. For example, saturated fatty acids constitute approximately 15%, 45% and 35% of CE, PL and TG fatty acids whereas polyunsaturated fatty acids make up for about 60%, 40% and 15% and monounsaturated fatty acids 25%, 15% and 50% of the fatty acids in CE, PL and TG, respectively. Furthermore, changes in dietary intake of fatty acids can change the relative proportion of tissue fatty acids only to a limited extent. This is well demonstrated for example in the study by Mantzioris et al, (1995) where the increase in the consumption of linoleic acid from 10 to 20 g=d resulted in an increase in PL linoleic acid from 21 to 22% ( 5% increase) whereas an equal increase in the intake of a- linoleic acid resulted in an increase in PL a-linoleic acid from 0.8 to 1.7% ( 113% increase). The time scale of changes in the fatty acid composition varies between serum lipid fractions. Changes in the fatty acid composition of TG take place rapidly in response to changes in the diet, that is, signi cant changes can be detected within one to two days, whereas changes in CE and PL fatty acids occur more slowly, that is, within one to two weeks (Katan et al, 1997; Vessby et al, 1980). Consequently, it is easy to understand that the long-term reproducibility of TG fatty acid composition is poorer than that of CE or PL (Moilanen, 1987). This was also demonstrated in the STRIP children as the correlation of dietary fatty acids with the fatty acids in TG was signi cantly better when blood samples were drawn for determination of TG fatty acid compositions immediately after the four-day food records than when there was an interval of 5 to 15 days between food records and drawing of the blood sample (LagstroÈm et al, 1998). Thus, the fatty acid composition of the TG fraction provides us objectively the information that differences did exist between intervention and control diets during the days preceding the follow-up visits. The disadvantage in using the TG fraction is, however, that it is very sensitive to day-to-day variations in intake. Furthermore, since the lling of the food records per se may in uence the dietary intake during the days of the recording, TG cannot be used as a biomarker in a long-term study like the STRIP project in which it is more relevant to get information of the level of long-term fat intake. Conclusion Small though signi cant changes were accomplished by dietary counselling on the dietary fat composition of the intervention children. The bene t of this was a lower serum total cholesterol and LDL-cholesterol level in intervention children when compared with control children. The fatty acid composition in the TG fraction replicated the dietary difference between intervention and control groups in close agreement with the food records. However, since the correlations between fatty acid intake and the respective serum fatty acid proportions were better in the CE fraction, CE fraction is the fraction of choice as a biomarker of dietary fatty acids in the long-term follow-up of the STRIP project children. Acknowledgements ÐThis study was supported by grants from Medical Research Foundation of Turku University Central Hospital (formerly Turku University Hospital Research Fund), Mannerheim League for Child Welfare, Finnish Cardiac Research Foundation, Medical Council of the Academy of Finland, YrjoÈ Jahnsson Foundation, Foundation for Pediatric Research, Finland, Piltti Foundation, Finnish Medical Society Duodecim, Juho Vainio Foundation, Turku University Foundation, Farmos Research and Science Foundation, Orion Company Research Foundation, Finnish Cultural Foundation, and Van den Bergh Foods Company. References Friedewald WT, Levy RI, & Fredrickson DS (1972): Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin. Chem. 18, 499 ± 502. Glatz JF, Soffers AE, & Katan MB (1989): Fatty acid composition of serum cholesteryl esters and erythrocyte membranes as indicators of linoleic acid intake in man. Am. J. Clin. Nutr. 49, 269 ± 276. Gustafsson IB, Vessby B, Ohrvall M, & Nydahl M (1994): A diet rich in monounsaturated rapeseed oil reduces the lipoprotein cholesterol concentration and increases the relative content of n-3 fatty acids in serum in hyperlipidemic subjects. Am. J. Clin. Nutr. 59, 667 ± 674. Hakala P, Marniemi J, Knuts L, Kumpulainen J, Tahvonen R, & Plaami S (1996): Calculated vs analyzed nutrient composition of weight reduction diets. Food Chemistry, 57, 71 ± 75. James MJ, Gibson RA, D'Angelo M, Neumann MA, & Cleland LG (1993): Simple relationships exist between dietary linoleate and the n- 6 fatty acids of human neutrophils and plasma. Am. J. Clin. Nutr. 58, 497 ± 500. Katan MB, Grundy SM, & Willett WC (1997): Should a low-fat, highcarbohydrate diet be recommended for everyone? Beyond low-fat diets. N. Engl. J. Med. 337, 563 ± 566. LagstroÈm H, Jokinen E, SeppaÈnen R, Salo P, Viikari J, RoÈnnemaa T, Helenius H, & Simell O (1998): Fatty acids in serum lipid fractions as indicators of fat intake in 5-year-old children in the STRIP project. Scand. J. Nutr. 42, 140 ± 145. Lapinleimu H, Viikari J, Jokinen E, Salo P, Routi T, Leino A, RoÈnnemaa T, SeppaÈnen R, VaÈlimaÈki I, & Simell O (1995): Prospective randomised trial in 1062 infants of diet low in saturated fat and cholesterol. Lancet, 345, 471 ± 476. London SJ, Sacks FM, Caesar J, Stampfer MJ, Siguel E, & Willett WC (1991): Fatty acid composition of subcutaneous adipose tissue and diet in postmenopausal US women. Am. J. Clin. Nutr. 54, 340 ± 345. Ma J, Folsom AR, Eckfeldt JH, Lewis L, Chambless LE and the Atherosclerosis Risk in Communities (ARIC) Study Investigators (1995): Short- and long-term repeatability of fatty acid composition of human plasma phospholipids and cholesterol esters. Am. J. Clin. Nutr. 62, 572 ± 578. Mantzioris E, James MJ, Gibson RA, & Cleland, LG (1995): Differences exist in the relationships between dietary linoleic and alpha-linolenic acids and their respective long-chain metabolites. Am. J. Clin. Nutr. 61, 320 ± 324. McMurchie EJ, Margetts BM, Beilin LJ, Croft KD, Vandongen R, & Armstrong BK (1984): Dietary-induced changes in the fatty acid composition of human cheek cell phospholipids: correlation with changes in the dietary polyunsaturated=saturated fat ratio. Am. J. Clin. Nutr. 39, 975 ± 980. Moilanen, T (1987): Short-term biological reproducibility of serum fatty acid composition in children. Lipids, 22, 250 ± 252. Moilanen T, RaÈsaÈnen L, Viikari J, AÊ kerblom HK, & Nikkari T (1992): Tracking of serum fatty acid composition: a 6-year follow-up study in Finnish youths. Am. J. Epidemiol. 136, 1487 ± Niinikoski H, Viikari J, RoÈnnemaa T, Lapinleimu H, Jokinen E, Salo P, SeppaÈnen R, Leino A, Tuominen J, VaÈlimaÈki I, & Simell O (1996): Prospective randomized trial of low-saturated-fat, low-cholesterol diet during the rst 3 years of life. The STRIP baby project. Circulation, 94, 1386 ±

6 932 Salo P, Viikari J, HaÈmaÈlaÈinen M, Lapinleimu H, Routi T, RoÈnnemaa T, SeppaÈnen R, Jokinen E, VaÈlimaÈki I, & Simell O (1999): Serum cholesterol ester fatty acids in 7- and 13-month-old children in a prospective randomized trial of low-saturated fat, low-cholesterol diet. The STRIP baby project. Acta Paediatrica, in press. Salo P, Viikari J, RoÈnnemaa T, HaÈmaÈlaÈinen M, Jokinen E, VaÈlimaÈki I, & Simell O (1979): Milk type during mixed feeding: contribution to serum cholesterol ester fatty acids in late infancy. J Pediatr. 130, 110 ± 116. Sarkkinen ES, AÊ gren JJ, Ahola I, Ovaskainen ML, & Uusitupa MI (1994): Fatty acid composition of serum cholesterol esters, and erythrocyte and platelet membranes as indicators of long-term adherence to fat-modi ed diets. Am. J. Clin. Nutr. 59, 364 ± 370. Stein EA, Shapero J, McNerney C, Glueck CJ, Tracy T, & Gartside P (1982): Changes in plasma lipid and lipoprotein fractions after alteration in dietary cholesterol, polyunsaturated, saturated, and total fat in free-living normal and hypercholesterolemic children. Am. J. Clin. Nutr. 35, 1375 ± Tynan MB, Nicholls DP, Maguire SM, Steele IC, McMaster C, Moore R, Trimble ER, & Pearce J (1995): Erythrocyte membrane fatty acid composition as a marker of dietary compliance in hyperlipidaemic subjects. Atherosclerosis, 117, 245 ± 252. Vessby B, Gustafsson IB, Boberg J, KarlstroÈm B, Lithell H, & Werner I (1980): Substituting polyunsaturated for saturated fat as a single change in a Swedish diet: effects on serum lipoprotein metabolism and glucose tolerance in patients with hyperlipoproteinaemia. Eur. J. Clin. Invest. 10: 193 ± 202. Zock PL, Mensink RP, Harryvan J, de Vries JH, & Katan MB (1997): Fatty acids in serum cholesteryl esters as quantitative biomarkers of dietary intake in humans. Am. J. Epidemiol. 145, 1114 ± 1122.