MAGNETIC RESONANCE IN MEDICINE 27,97-106 ( 1992)
Relationship between the Degree of Unsaturation of Dietary Fatty Acids and Adipose Tissue Fatty Acids Assessed by NaturalAbundance I3C Magnetic Resonance Spectroscopy in Man N. BECKMA”,* J.-J. BROCARD,?U. KELLER,? AND J. SEELIG*
’M R Center and Biocenter of
the University of Basel, Klingelbergstrasse 70, CH-4056 Basel, Switzerland: and 7 Division of Endocrinology and Metabolism, University Hospital, Petersgraben 2, CH-4051 Basel, Switzerland
Received June 18, 1991; revised October 18, 1991; accepted November 7, 1991 Natural-abundance I3C magnetic resonance spectroscopy was used for determining noninvasively the relative concentration of mono- and polyunsaturated fatty acids of adipose tissue in two groups of volunteers. The first consisted of subjects who had followed a fatreduced diet for at least half a year before the I3C measurements. The second were control subjects who were on a usual high-fat diet. The ratio of unsaturated to total fatty acids in adipose tissue determined by I3C MRS correlated significantly with the same ratio in fat of the diet composition estimated by a dietician according to food records. The results indicate that I3C MRS is capable of assessing the degree of unsaturation of dietary fatty acids consumed during the preceding months. o 1992 Academic Press, Inc. INTRODUCTION
Low levels of linoleic acid ( 18:2n-6) in the circulation and in adipose tissue have been linked with an increased risk of coronary heart disease (CHD) (1-3). These studies suggested that the relative amount of polyunsaturated fatty acids in adipose tissue reflect differences in dietary intake ( 3 ) . Measurement of the concentration of fatty acids in adipose tissue has been suggested to be a valid tool to assess the dietary fat intake for the preceding 1 to 3 years ( 4 , 5 ) . Indeed, based on literature data, Beynen et al. ( 4 ) demonstrated a direct relationship between the fatty acid composition of the habitual diet and of the subcutaneous adipose tissue in humans. For experimental and epidemiological nutritional surveys the average fat composition of the adipose tissue is a useful parameter for long-term dietary habits ( 4 ) . This association is not only of pathophysiological interest but also of practical importance since objective assessment of long-term dietary compliance is generally recognized to be an unresolved issue. In this paper natural-abundance 13Cmagnetic resonance spectroscopy ( MRS) was applied for determining the degree of unsaturation of fatty acids in human adipose tissue in relation to the diet. Correlations were made between the degree of unsaturation of adipose tissue determined by I3C MRS and that calculated from a questionnaire about habitual fatty acid intake of the subjects. 91
0740-3194192 $5.00 Copynght 0 1992 by Academic Press, Inc All nghts of reproduction In any form reserved
98
BECKMANN ET AL. METHODS
Subjects and Dietary History The study protocol was approved by the Ethics Committee of the University Hospital of Basel. Twenty-eight male volunteers gave their written informed consent to participate. The first group of subjects ( n = 13) was recruited from the Lipid Clinic of the University Hospital of Basel. They had been on a cholesterol-lowering, fat-reduced diet according to the recommendations of the American Heart Association ( 6 ) for at least half a year before the MRS measurements. They had been advised by a dietician to consume 50-60% of their daily caloric intake as carbohydrates, 10-20% as proteins, and no more than 30% as fats, divided in approximately 10% saturated, 10%monounsaturated, and 10% polyunsaturated fats. The second group of 15 normal subjects was on a usual diet which is relatively high in fat in Switzerland ( 7). The daily intake of carbohydrates, proteins, and saturated and unsaturated fats before the MRS examinations was estimated on the basis of food records from 4 consecutive days and using food composition data from Souci, et al. (8). Table 1 shows that both groups of subjects were of similar age, body weight, and degree of obesity. Their serum lipoproteins and lipids were not significantly different except for serum LDL cholesterol which was higher in the diet group. Total cholesterol and triglycerides were measured enzymatically, and LDL and HDL were assessed after precipitation (kits from Boehringer-Mannheim, Germany).
"C M R Spectroscopy Experiments were performed using a Siemens Helicon whole-body system operating at 1.5 T in combination with surface coils for transmission and detection of radiofrequency. The surface coils had a coplanar and concentric geometry with diameters of 8.0 cm (two turns) and 13.0 cm (one turn) for 13Cand 'H, respectively. Shimming was performed with the outer 'H coil. The individuals laid in prone position, with TABLE 1 Anthropometnc Parameters and Lipid Profile (Means i- SD)
Parameters
Diet group (n = 13)
Controls (n = 15)
Age (years) BMI" (kg/m2) Weight (kg) WHRh Total cholesterol (mmol/liter) HDL cholesterol (mmol/liter) LDL cholesterol (mmol/liter) Triglvcerides (mmol/liter)
49.1 f 9.5 24.2 f 2.1 73.1 f 10.2 0.88 f 0.06 6.1 f 1.2 1.2 f 0.2 4.6 f 1.1 1.8 f 0.9
43.3 f 9.9 25.5 f 2.2 11.9 f 5.1
0.89 f 0.06 5.9 f 1.1 1.3 -+ 0.3 3.8 i 1.0 2.2 -+ 1.2
Note. No parameter was significantly different between the two groups. Body mass index. Waist to hip circumference ratio.
Statistical significance ns.
ns. n.s. n.s. n.s. n.s. P < 0.05 n.s.
FATTY ACIDS ANALYSIS BY I3C MRS IN MAN
99
the coils under the mid of the abdomen. Spectra were taken from subcutaneous fat tissue of the abdominal region only. The acquired signals consisted of 1024 complex data points with a spectral width of 4 kHz. Before Fourier transformation the data were multiplied by a 2-Hz exponential filter. The areas were determined using standard software, and no spectral fitting method was applied. In order to avoid uncertainties introduced in the determination of relaxation times and NOE values of carbon resonances (see ( 9 ) ) , coupled 13C spectra with a long repetition time were acquired (TR = 10.0 s ) . T I measurements effectuated in vivo by Moonen el al. ( 9 )indicate that the longest TI relaxation time is 1.9 s from the carbonyl group. Thus with TR = 10.0 s all carbons could relax fully between the acquisitions. The length of the nonselective pulse was 200 ps. The power was adjusted in order that a 90O-pulse was obtained at the coil's center. The total time needed for a session, including positioning of the volunteer, shimming, and acquiring the I3C signal, was about 30 min. For determining the degree of unsaturation with I3C spectroscopy three assumptions were made: ( h l ) All the resonances in a "C spectrum of human adipose tissue are due to triglycerides. This has been demonstrated previously ( 9 ) . (h2) The concentration of carbonyl groups in human adipose tissue is nearly constant, so that the carbonyl signal can be used as an internal reference. This assumption is valid since it has been demonstrated in ( 9 )that the I3C signals of adipose tissue are dominated by triglyceride resonances. (h3) Linoleic acid is the only polyunsaturated fatty acid which is found in significant amounts in human adipose tissue. Studies have demonstrated that linoleic acid accounts for at least 85% of the polyunsaturated fatty acid content of adipose tissue in groups of individuals belonging to different races ( l o ) , sex ( I 1), and countries ( 3 ) and on different subcutaneous sites ( 12).
'
Consequently, the relative concentrations of mono- and polyunsaturated fatty acids were calculated using the following relations: [mono](%) = 100.
(12,
-
13b)
21 1 13,
[poly](%)= loo*-. 2 II
[2 1
Equations [ 11 and [ 21 can be understood from the assignment given in Table 2 ( 9 ) . The Inb)s represent areas under the resonances of the spectrum shown in Fig. 1. 1, consists of the area under the carbonyl signal (resonance { 1 } ), which reflects the total amount of triglycerides in adipose tissue, independently of their saturation or unsaturation. IZb and 13b are calculated from resonances { 2b) and { 3b}, respectively, which originate from unsaturated carbons only. Resonances { 2 } correspond to carbons in
' Linoleic acid: CH3- (CH,),- 'CH=3CH-CH2-3CH=*CH(CH,),-COOH. The indexes 2 or 3 on the unsaturated carbons indicate to which resonance they contribute in the spectrum of Fig. I .
100
BECKMANN ET AL. TABLE 2 Assignment of the "C Resonances Indicated in Fig. 1, Corresponding to Carbonyl Groups and Unsaturated Carbons Resonance no.
Carbons
R-0-CO -R (monounsatruated) and -CH =CH -CH2-CH =CH - (polyunsaturated) -CH=CH-CH2-CH=CH(polyunsaturated only)
1
-CH =CH -
2
3
double bonds of mono- and polyunsaturated fatty acid chains. Resonances { 3 } , on the other hand, correspond exclusively to carbons in double bonds of polyunsaturated chains (Table 2 ) . The factor 2 in Eqs. [ l ] and [ 2 ] arises from the fact that there are two carbons per double bond. It follows that the degree of saturation is given by
+ [poly]).
[satur](%)= 100 -([mono]
[31
The areas were calculated using standard equipment software. Due to their better spectral resolution we used only the unsaturated resonances 2b and 3b for quantification. Resonances 2a and 3a have a somewhat poorer resolution because of longrange couplings. Evidence for this is found in the fact that decoupling improves the resolution of the signals of double-bonded carbons as well as of carbonyl groups. The long-range coupling constants JCHand JCHare from 2 to 15 Hz, and for carbonyl groups between 3 and 7 Hz [ 131. Correlation coefficients (Pearson) and unpaired t tests were calculated using BMDP Statistical Software (BMDP Inc., Los Angeles). RESULTS
Dietary Energy, Carbohydrate, and Fat Intake (Table 3 ) There were statistically significant differences in total caloric intake and in the percentage of calories from carbohydrates and from fat between the two groups. The 1
1
200
I
I
160
120
I
80
PPM FIG. 1. Partial view of a typical coupled I3C MRS spectrum of human adipose tissue. Assignment of the
resonances is indicated in Table 2.
101
FATTY ACIDS ANALYSIS BY I3C MRS IN MAN TABLE 3 Dietary History in Diet-Treated and in Control Subjects (Means +- SD)
Parameters
Diet group (n = 13)
Controls ( n = 15)
Statistical significance
Total energy intake (kcal) % Carbohydrate intake" YO Total fat intake" 7 ' 0 Monounsaturated of fat intake % Polyunsaturated of fat intake 7 ' 0 Unsaturated of fat intake % Saturated of fat intake
1628.2 +- 301.4 48.0 f 5.6 32.9 f 5.0 33.1 f 1.2 31.4 i 10.7 65.1 i 6.6 34.9 f 6.6
2079.1 f 525.2 38.0 7.6 44.4 t 8.8 31.2 4.1 19.1 t 1.9 56.9 t 6.2 43.1 i_ 6.2
* *
P < 0.002 P < 0.001 P < 0.0002 n.s. P i0.003 P < 0.002 P < 0.002
'Percentage of total caloric intake.
mean daily fat intake was 98 g in the control group and 56 g in the diet group. Diettreated subjects consummed more polyunsaturated ( P < 0.003) and less saturated fatty acids ( P < 0.002) than controls. The intake of unsaturated fat expressed as percentage of total fat intake was significantly (P < 0.002) higher in the diet group than in the control group (65.1 ? 6.6% vs 56.9 f 6.2%).
13CM R Spectroscopy of Adipose Tissue (Table 4 )
+
The average degree of unsaturation (mono poly ) of abdominal fat as determined by I3C MRS was significantly higher in the diet group than in controls (P< 0.0001 ). However, considering the mono- and polyunsaturation degrees separately, only monounsaturated fatty acids were significantly different.
Correlation between the Ratio of Unsaturated and of Saturated Fatty Acids to Total Fatty Acids in the Diet and in Adipose Tissue
r
There was a significant positive correlation [Y = 0.63 for the diet group ( n = 13); 0.52 for controls ( n = 15)] between the dietary intake of monounsaturated fats
=
TABLE 4 Concentration of Mono- and Polyunsaturated Fatty Acids (in YO of Total Fat) in Human Adipose Tissue as Determined by I3C MR Spectroscopy (Means k SD)
Unsaturation values % Mono % Poly Total YO unsaturated
970 Saturated
Diet group ( n = 13)
Controls ( n = 15)
Statistical significance
52.6 + 5.5 17.5 + 3.3 70.1 f 4.3 29.9 _+ 4.3
44.7 t 4.2 15.4 4.3 60.1 4.5 39.8 _+ 4.5
Y < 0.0002
+
*
n.s.
P < 0.000 1 P < o.Oo01
102
BECKMANN ET AL.
expressed as percentage of total fat consumption and the degree of monounsaturation of abdominal fat determined by MR spectroscopy. There was also a significant correlation between the dietary polyunsaturated fatty acid fraction and the polyunsaturated
50
a
-
I 20
20
30
50
40
% sat (food records)
b
80
-2
70
E
v Y
!i
1
60
50
50
60
70
80
% unsat (food records) FIG.2. ( a ) Relationship between average dietary saturated fat intake expressed as percentage of total fat intake and the average degree of saturation of abdominal fat determined by I3C MRS in the two groups of subjects. ( b ) The same for the degree of unsaturation. 0, Low-fat diet group: controls.
*,
FATTY ACIDS ANALYSIS BY "C MRS IN MAN
103
fatty acids in the abdominal fat in the diet group ( r = 0.56, n = 13). Figure 2 shows the average degrees of saturation and unsaturation in both groups. The graph demonstrates a clear separation of the diet-treated and the control subjects. DISCUSSION
The purpose of the present study was to investigate the ability of natural-abundance 13C MRS to discriminate the ratio of unsaturated versus saturated fatty acids in abdominal subcutaneous fat in individuals with different dietary habits regarding fatty acid saturation. The results demonstrated that there was a significant difference in the degree of saturation ( P < 0.0001 ) and monosaturation ( P < 0.0002) of adipose tissue as determined by 13CMRS between the diet and the control group. In contrast, no statistically significant difference between the two groups was seen in the degree of polyunsaturation of adipose tissue whereas the difference in the total degree of unsaturation (mono polyunsaturation) was highly significant ( P < 0.000 1 ). In agreement with the adipose tissue fatty acid composition, the evaluation of the dietary fatty acids demonstrated significant differences in the consumption of saturated and of unsaturated fatty acids between the two groups. The only disagreement was observed in the distinction between mono- and polyunsaturated fatty acids. Thus, the present study demonstrated a clear separation of the content of saturated and total unsaturated fatty acids in the diet group and the controls. The reason for the failure to distinguish the diet and the control groups regarding polyunsaturated fatty acid content is not obvious; due to the lower relative content of polyunsaturated fatty acids in adipose tissue, the I3C MRS technique may not have been sensitive enough to detect a difference. The advantages of using I3C MRS are that: ( 1 ) it is noninvasive and can thus be applied repeatedly on the same individual; ( 2 ) it permits the determination of the relative concentration of saturated and unsaturated fatty acids; ( 3 ) the determined concentration represents a local average, i.e., it is applicable to the study of various sites of adipose tissue in the body. I3CMRS has recently been used to determine the degree of unsaturation of adipose tissue fatty acids in normal volunteers ( 9 )and in patients with cystic fibrosis ( 14). Moonen et al. observed in two normal subjects individual relative contributions of 18.4 and 21.5% polyunsaturated, and 40.8 and 45.1% monounsaturated fatty acids, respectively ( 9 ) ;these results are in close agreement with the present data of control subjects. However, these authors ( 9 , 1 4 ) used decoupling for the assessment of the degree of unsaturation of fatty acids, i.e., two separate radiofrequency channels were required for 'H and 13Cwhich are not yet standard equipment of most whole-body scanners. In contrast, our procedure needs only one channel tuned to the 13Cfrequency. In addition, acquiring coupled spectra avoids determination of NOE factors. Although this study uses long repetition times and only a few acquisitions, the signal-to-noise ratio of the spectra is sufficient for quantification of the spectral lines (see Fig. 1 ). Our data are also in good agreement with fatty acid compositions determined with gas chromatography ( 3 , 1 2 ) . Since all the unsaturated carbons contribute signals to resonances 2 and 3 (see Fig. 1 and Table 2), the total degree of unsaturation can be determined from the I3C spectra without making any assumption about the fatty acid composition. However,
+
104
BECKMANN ET AL.
to distinguish between mono- and polyunsaturated fatty acids, the additional hypothesis is put forward that linoleic acid is the only significant polyunsaturated fatty acid in human adipose tissue, as discussed under Methods. This assumption is necessary since the in vivo I3C MRS spectra have a limited resolution, not providing separate information about all the different fatty acyl chains in adipose tissue. The consequence of this assumption is that the degree of polyunsaturation determined by Eq. [ 21 is to a certain extent larger than the real values, the error being proportional to the concentration of other polyunsaturated fatty acids like linolenic or arachidonic acid. It was of interest to note that the average consumption of fat in the control subjects of the present study was 44% (Table 3). This was close to the value of 43% reported previously ( 7) in a study about the dietary habits of the Swiss population. The controls consumed amounts of saturated fatty acids similar to that of the American population ( 15-20%) according to the NCEP report (15).Their average carbohydrateconsumption (38.0 k 7.6%) was similar to the average value of 44.0% reported for the Swiss population 10 years ago ( 7). In the diet group the average fat intake was reduced to 32.9% of total calories, and 11.7% of the consumed calories were saturated fats which was close to the recommendation (6). The major factor determining the time required for a change in the dietary fat type to be reflected in the fatty acid profile of adipose tissue is the turnover rate of individual fatty acids of adipose tissue. Our patients were following a diet half a year or more before the measurements, and a statistically significant difference between the groups could be seen in the degree of saturation, of monounsaturation, and of total unsaturated fatty acids as determined by MRS. Beynen et al. found that for a longer-term diet, there was a linear relationship between the degree of polyunsaturation of adipose tissue fatty acids using biopsies and the polyunsaturated fatty acid content of a longterm diet ( 4 ) . As these authors pointed out ( 4 ) , carbohydrate intake, alcohol consumption, energy expenditure, and conversion of saturated into monounsaturated fatty acids may all influence the percentage of saturated and monounsaturated fatty acids in adipose tissue. On the other hand Beynen et al. pointed to the fact that the content of polyunsaturated fatty acids in tissue was often lower than that of the diet ( 4 ) . Indeed, dietary fatty acids, before being incorporated into adipose tissue, are considerably diluted with de n o w synthesized fatty acids, especially when the absolute fat intake is low. The study described in ( 1 6 ) shows that the concentration of polyunsaturated fatty acids in adipose tissue of six different groups of subjects was lower than that of the diet. Four consecutive days of food records were considered acceptable as a reasonable compromise for minimal, reliable monitoring of diet compliance in outpatients ( 17). However, for achieving a 95% confidence limit 7 consecutive days of food recording were recommended ( 17). Consequently, the correlation between the food records and the MRS data in our study could be improved by extending the food recording period from 4 to 7 consecutive days. On the other hand, it must be emphasized that the control subjects followed no standard diet which might be one of the reasons that no significant correlation could be found between the degree of polyunsaturation of the fatty acids of the diet and of adipose tissue in this group, in contrast to the diet group. Recently MR imaging has been applied for measuring adipose tissue distribution in the human body ( 1 8 ) . It has been emphasized that there are large differences of
FATTY ACIDS ANALYSIS BY I3C MRS IN MAN
I05
fat metabolism in various regions of the human body. Therefore it would be of interest to compare the degree of unsaturation of fatty acids in different regions of the human body with I3CMRS and to assess the time course of changes of fatty acid composition after the dietary content of fatty acids has been changed. Furthermore, by combining imaging techniques with 13C spectroscopy it should be possible to obtain an assessment of both fatty acid composition and distribution in the human body. CONCLUSIONS
In this paper the application of 13C MRS for the noninvasive assessment of the relative concentration of mono- and polyunsaturated fatty acids of adipose tissue in two groups of volunteers was described. One group of subjects had followed a fatreduced diet for at least half a year before the 13C measurements. The second group consisted of controls who were on a high-fat but nonstandard diet. MRS data were compared with the fatty acid composition estimated by a dietician according to food records. The principal results were as follows: There was a highly significant difference between the two groups in the degree of saturation and monounsaturation of adipose tissue determined with ' 3C MRS No statistically significant difference was seen in the degree of polyunsaturation determined with MRS, although the diet-treated subjects ingested proportionally more polyunsaturated fatty acids than the controls. However, the difference in the total degree of unsaturation (mono + polyunsaturation ) was highly significant There was a significant positive correlation between the dietary intake of monounsaturated fats and the degree of monounsaturation of abdominal fat determined by 13CMRS. For the polyunsaturated fatty acids such a correlation could be found only in the diet group. Our data show that I3CMRS is capable of detecting differences in the total unsaturation degree of fatty acids between the groups. That no significant difference was found in the polyunsaturation degree, despite that this value for the controls agreed well with data from the literature (obtained with gas chromatography), might be explained by the fact that the proportion of polyunsaturated fatty acids incorporated into adipose tissue is usually smaller than their content in the diet. Therefore, I3C MRS might not have been sensitive enough to detect a difference in the polyunsaturation degree. The fact that the controls followed no standardized diet might be one of the factors for the observation of no correlation between the polyunsaturation degree of the diet and of adipose tissue. ACKNOWLEDGMENTS The contributions of Ms. A. Biichi and P. Joho (dieticians) are gratefully acknowledged. We also thank Dr. S. Muller, MR Center, for advice. Financial support was obtained from the Swiss National Science Foundation (Grant No. 4.889.85.18 to J.S.). REFERENCES
I. R. I. LOGAN,R. A. RIEMERSMA, M. THOMSON,M. F. OLIVER,A. G . OLSSON,G. WALLDIUS, S. ROSSNER,L. (1978).
KAIJSER,
E. CALLMER, L. A. CARLSON, L. LOCKERBIE, AND w . LUTZ, Lance1 1, 949
106
BECKMANN ET AL.
2. D. A. WOOD. S. BUTLER,S, R. A. RIEMERSMA, M. THOMSON,M. F. OLIVER,M. FIJLTON,A. BIRTHWHISTLE, AND R. ELTON,Lancet 2, 117 (1984). 3. R. A. RIEMERSMA, D. A. WOOD, S. BUTLER, R. A. ELTON, M. OLIVER,M. SALO.T. NIKKARI,E. P. PUSKA,F. GEY,P. RUBBA,M. MANCINI,AND F. RDANZA, Brit. Med. J. 292, 1423 VARTIAINEN, (1986). 4. A. C. BEYNEN,R. J. J. HERMUS,AND J. G. A. J. HAUTVAST,Am. J. Clin. Nzitr. 33,81 ( 1980). 5. M. B. KATAN, W. A. VANSTAVEREN, P. DEURENBERG, J. BARDENDEUSE-VAN LEEUWEN, C. GERMINGNOUWEN,A. SOFFERS,J. BERKEL,AND A . C. BEYNEN,Prog. Lipid Res. 25, 193 ( 1986). 6. American Heart Association. Arteriosclerosis 8, 218A ( 1988). 7. H. AEBI AND A. BLUMENTHAL, “Zweiter Schweizerischer Ernahrungsbericht,” Hans-Huber Verlag, Bern, 1984. 8. S. W. Soucr, W. FACHMANN, AND H. KRAUT,“Die Zusammensetzung der Lebensmittel,” 4th ed., Wissenschaftliche Verlagsanstalt, Stuttgart, 1989. 9. C. T. W. MOONEN,R. J. DIMAND,AND K. L. COX,Mugn. Reson. Med. 6, 140 (1988). 10. W. INSULLAND G. E. BARTSCH,Am. J. Clin. Nutr. 20, 13 (1967). 11. S. J. LONDON,F. M. SACKS,J. CAESAR, M. J. STAMPFER, E. SIGUEL,A N D W. C. WILLETT,Am. J . C h . Nutr. 54, 340 ( I99 1 ). 12. G . T. MALCOM,A. K. BHATTACHARYYA, M. VELEZ-DURAN, M. A. GUZMAN,M . C. OALMANN, AND J. P. STRONG,Am. J. Clin. Nutr. 50, 288 (1989). 13. F. W. WEHRLIAND T. WIRTHLIN,“Interpretation of Carbon-13 Spectra,” Wiley Heyden, London, 1983. 14. R. J. DIMAND,C. T. W. MOONEN,S. C. CHU, E. M. BRADBURY,G. KURLAND.A N D K. L. Cox, Pediatr. Rex 24, 243 ( 1988 ) . 15. Report of the National Cholesterol Education Program Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Arch. Intern. Med. 148, 36 ( 1988). 16. R. F. SCOTT,K. T. LEE, D. N.KIM, E. S. MORRISON,AND F. GOODALE,A m . J . C h . Nuir. 14, 280 (1964). 17. B. JACKSON, C. A. DUJOVNE,S. DECOURSEY, P. BEYER,E. F. BROWN,AND K. HASSANEIN, J . Am. Diet. Assoc. 86, 1531 (1986). 18. J. C. SEIDELL, C. J. G. BAKKER, AND K. VANDER KOOY, Am. J. C h . Nutr. 51, 953 ( 1990).
Relationship between the Degree of Unsaturation of Dietary Fatty Acids and Adipose Tissue Fatty Acids Assessed by NaturalAbundance I3C Magnetic Resonance Spectroscopy in Man N. BECKMA”,* J.-J. BROCARD,?U. KELLER,? AND J. SEELIG*
’M R Center and Biocenter of
the University of Basel, Klingelbergstrasse 70, CH-4056 Basel, Switzerland: and 7 Division of Endocrinology and Metabolism, University Hospital, Petersgraben 2, CH-4051 Basel, Switzerland
Received June 18, 1991; revised October 18, 1991; accepted November 7, 1991 Natural-abundance I3C magnetic resonance spectroscopy was used for determining noninvasively the relative concentration of mono- and polyunsaturated fatty acids of adipose tissue in two groups of volunteers. The first consisted of subjects who had followed a fatreduced diet for at least half a year before the I3C measurements. The second were control subjects who were on a usual high-fat diet. The ratio of unsaturated to total fatty acids in adipose tissue determined by I3C MRS correlated significantly with the same ratio in fat of the diet composition estimated by a dietician according to food records. The results indicate that I3C MRS is capable of assessing the degree of unsaturation of dietary fatty acids consumed during the preceding months. o 1992 Academic Press, Inc. INTRODUCTION
Low levels of linoleic acid ( 18:2n-6) in the circulation and in adipose tissue have been linked with an increased risk of coronary heart disease (CHD) (1-3). These studies suggested that the relative amount of polyunsaturated fatty acids in adipose tissue reflect differences in dietary intake ( 3 ) . Measurement of the concentration of fatty acids in adipose tissue has been suggested to be a valid tool to assess the dietary fat intake for the preceding 1 to 3 years ( 4 , 5 ) . Indeed, based on literature data, Beynen et al. ( 4 ) demonstrated a direct relationship between the fatty acid composition of the habitual diet and of the subcutaneous adipose tissue in humans. For experimental and epidemiological nutritional surveys the average fat composition of the adipose tissue is a useful parameter for long-term dietary habits ( 4 ) . This association is not only of pathophysiological interest but also of practical importance since objective assessment of long-term dietary compliance is generally recognized to be an unresolved issue. In this paper natural-abundance 13Cmagnetic resonance spectroscopy ( MRS) was applied for determining the degree of unsaturation of fatty acids in human adipose tissue in relation to the diet. Correlations were made between the degree of unsaturation of adipose tissue determined by I3C MRS and that calculated from a questionnaire about habitual fatty acid intake of the subjects. 91
0740-3194192 $5.00 Copynght 0 1992 by Academic Press, Inc All nghts of reproduction In any form reserved
98
BECKMANN ET AL. METHODS
Subjects and Dietary History The study protocol was approved by the Ethics Committee of the University Hospital of Basel. Twenty-eight male volunteers gave their written informed consent to participate. The first group of subjects ( n = 13) was recruited from the Lipid Clinic of the University Hospital of Basel. They had been on a cholesterol-lowering, fat-reduced diet according to the recommendations of the American Heart Association ( 6 ) for at least half a year before the MRS measurements. They had been advised by a dietician to consume 50-60% of their daily caloric intake as carbohydrates, 10-20% as proteins, and no more than 30% as fats, divided in approximately 10% saturated, 10%monounsaturated, and 10% polyunsaturated fats. The second group of 15 normal subjects was on a usual diet which is relatively high in fat in Switzerland ( 7). The daily intake of carbohydrates, proteins, and saturated and unsaturated fats before the MRS examinations was estimated on the basis of food records from 4 consecutive days and using food composition data from Souci, et al. (8). Table 1 shows that both groups of subjects were of similar age, body weight, and degree of obesity. Their serum lipoproteins and lipids were not significantly different except for serum LDL cholesterol which was higher in the diet group. Total cholesterol and triglycerides were measured enzymatically, and LDL and HDL were assessed after precipitation (kits from Boehringer-Mannheim, Germany).
"C M R Spectroscopy Experiments were performed using a Siemens Helicon whole-body system operating at 1.5 T in combination with surface coils for transmission and detection of radiofrequency. The surface coils had a coplanar and concentric geometry with diameters of 8.0 cm (two turns) and 13.0 cm (one turn) for 13Cand 'H, respectively. Shimming was performed with the outer 'H coil. The individuals laid in prone position, with TABLE 1 Anthropometnc Parameters and Lipid Profile (Means i- SD)
Parameters
Diet group (n = 13)
Controls (n = 15)
Age (years) BMI" (kg/m2) Weight (kg) WHRh Total cholesterol (mmol/liter) HDL cholesterol (mmol/liter) LDL cholesterol (mmol/liter) Triglvcerides (mmol/liter)
49.1 f 9.5 24.2 f 2.1 73.1 f 10.2 0.88 f 0.06 6.1 f 1.2 1.2 f 0.2 4.6 f 1.1 1.8 f 0.9
43.3 f 9.9 25.5 f 2.2 11.9 f 5.1
0.89 f 0.06 5.9 f 1.1 1.3 -+ 0.3 3.8 i 1.0 2.2 -+ 1.2
Note. No parameter was significantly different between the two groups. Body mass index. Waist to hip circumference ratio.
Statistical significance ns.
ns. n.s. n.s. n.s. n.s. P < 0.05 n.s.
FATTY ACIDS ANALYSIS BY I3C MRS IN MAN
99
the coils under the mid of the abdomen. Spectra were taken from subcutaneous fat tissue of the abdominal region only. The acquired signals consisted of 1024 complex data points with a spectral width of 4 kHz. Before Fourier transformation the data were multiplied by a 2-Hz exponential filter. The areas were determined using standard software, and no spectral fitting method was applied. In order to avoid uncertainties introduced in the determination of relaxation times and NOE values of carbon resonances (see ( 9 ) ) , coupled 13C spectra with a long repetition time were acquired (TR = 10.0 s ) . T I measurements effectuated in vivo by Moonen el al. ( 9 )indicate that the longest TI relaxation time is 1.9 s from the carbonyl group. Thus with TR = 10.0 s all carbons could relax fully between the acquisitions. The length of the nonselective pulse was 200 ps. The power was adjusted in order that a 90O-pulse was obtained at the coil's center. The total time needed for a session, including positioning of the volunteer, shimming, and acquiring the I3C signal, was about 30 min. For determining the degree of unsaturation with I3C spectroscopy three assumptions were made: ( h l ) All the resonances in a "C spectrum of human adipose tissue are due to triglycerides. This has been demonstrated previously ( 9 ) . (h2) The concentration of carbonyl groups in human adipose tissue is nearly constant, so that the carbonyl signal can be used as an internal reference. This assumption is valid since it has been demonstrated in ( 9 )that the I3C signals of adipose tissue are dominated by triglyceride resonances. (h3) Linoleic acid is the only polyunsaturated fatty acid which is found in significant amounts in human adipose tissue. Studies have demonstrated that linoleic acid accounts for at least 85% of the polyunsaturated fatty acid content of adipose tissue in groups of individuals belonging to different races ( l o ) , sex ( I 1), and countries ( 3 ) and on different subcutaneous sites ( 12).
'
Consequently, the relative concentrations of mono- and polyunsaturated fatty acids were calculated using the following relations: [mono](%) = 100.
(12,
-
13b)
21 1 13,
[poly](%)= loo*-. 2 II
[2 1
Equations [ 11 and [ 21 can be understood from the assignment given in Table 2 ( 9 ) . The Inb)s represent areas under the resonances of the spectrum shown in Fig. 1. 1, consists of the area under the carbonyl signal (resonance { 1 } ), which reflects the total amount of triglycerides in adipose tissue, independently of their saturation or unsaturation. IZb and 13b are calculated from resonances { 2b) and { 3b}, respectively, which originate from unsaturated carbons only. Resonances { 2 } correspond to carbons in
' Linoleic acid: CH3- (CH,),- 'CH=3CH-CH2-3CH=*CH(CH,),-COOH. The indexes 2 or 3 on the unsaturated carbons indicate to which resonance they contribute in the spectrum of Fig. I .
100
BECKMANN ET AL. TABLE 2 Assignment of the "C Resonances Indicated in Fig. 1, Corresponding to Carbonyl Groups and Unsaturated Carbons Resonance no.
Carbons
R-0-CO -R (monounsatruated) and -CH =CH -CH2-CH =CH - (polyunsaturated) -CH=CH-CH2-CH=CH(polyunsaturated only)
1
-CH =CH -
2
3
double bonds of mono- and polyunsaturated fatty acid chains. Resonances { 3 } , on the other hand, correspond exclusively to carbons in double bonds of polyunsaturated chains (Table 2 ) . The factor 2 in Eqs. [ l ] and [ 2 ] arises from the fact that there are two carbons per double bond. It follows that the degree of saturation is given by
+ [poly]).
[satur](%)= 100 -([mono]
[31
The areas were calculated using standard equipment software. Due to their better spectral resolution we used only the unsaturated resonances 2b and 3b for quantification. Resonances 2a and 3a have a somewhat poorer resolution because of longrange couplings. Evidence for this is found in the fact that decoupling improves the resolution of the signals of double-bonded carbons as well as of carbonyl groups. The long-range coupling constants JCHand JCHare from 2 to 15 Hz, and for carbonyl groups between 3 and 7 Hz [ 131. Correlation coefficients (Pearson) and unpaired t tests were calculated using BMDP Statistical Software (BMDP Inc., Los Angeles). RESULTS
Dietary Energy, Carbohydrate, and Fat Intake (Table 3 ) There were statistically significant differences in total caloric intake and in the percentage of calories from carbohydrates and from fat between the two groups. The 1
1
200
I
I
160
120
I
80
PPM FIG. 1. Partial view of a typical coupled I3C MRS spectrum of human adipose tissue. Assignment of the
resonances is indicated in Table 2.
101
FATTY ACIDS ANALYSIS BY I3C MRS IN MAN TABLE 3 Dietary History in Diet-Treated and in Control Subjects (Means +- SD)
Parameters
Diet group (n = 13)
Controls ( n = 15)
Statistical significance
Total energy intake (kcal) % Carbohydrate intake" YO Total fat intake" 7 ' 0 Monounsaturated of fat intake % Polyunsaturated of fat intake 7 ' 0 Unsaturated of fat intake % Saturated of fat intake
1628.2 +- 301.4 48.0 f 5.6 32.9 f 5.0 33.1 f 1.2 31.4 i 10.7 65.1 i 6.6 34.9 f 6.6
2079.1 f 525.2 38.0 7.6 44.4 t 8.8 31.2 4.1 19.1 t 1.9 56.9 t 6.2 43.1 i_ 6.2
* *
P < 0.002 P < 0.001 P < 0.0002 n.s. P i0.003 P < 0.002 P < 0.002
'Percentage of total caloric intake.
mean daily fat intake was 98 g in the control group and 56 g in the diet group. Diettreated subjects consummed more polyunsaturated ( P < 0.003) and less saturated fatty acids ( P < 0.002) than controls. The intake of unsaturated fat expressed as percentage of total fat intake was significantly (P < 0.002) higher in the diet group than in the control group (65.1 ? 6.6% vs 56.9 f 6.2%).
13CM R Spectroscopy of Adipose Tissue (Table 4 )
+
The average degree of unsaturation (mono poly ) of abdominal fat as determined by I3C MRS was significantly higher in the diet group than in controls (P< 0.0001 ). However, considering the mono- and polyunsaturation degrees separately, only monounsaturated fatty acids were significantly different.
Correlation between the Ratio of Unsaturated and of Saturated Fatty Acids to Total Fatty Acids in the Diet and in Adipose Tissue
r
There was a significant positive correlation [Y = 0.63 for the diet group ( n = 13); 0.52 for controls ( n = 15)] between the dietary intake of monounsaturated fats
=
TABLE 4 Concentration of Mono- and Polyunsaturated Fatty Acids (in YO of Total Fat) in Human Adipose Tissue as Determined by I3C MR Spectroscopy (Means k SD)
Unsaturation values % Mono % Poly Total YO unsaturated
970 Saturated
Diet group ( n = 13)
Controls ( n = 15)
Statistical significance
52.6 + 5.5 17.5 + 3.3 70.1 f 4.3 29.9 _+ 4.3
44.7 t 4.2 15.4 4.3 60.1 4.5 39.8 _+ 4.5
Y < 0.0002
+
*
n.s.
P < 0.000 1 P < o.Oo01
102
BECKMANN ET AL.
expressed as percentage of total fat consumption and the degree of monounsaturation of abdominal fat determined by MR spectroscopy. There was also a significant correlation between the dietary polyunsaturated fatty acid fraction and the polyunsaturated
50
a
-
I 20
20
30
50
40
% sat (food records)
b
80
-2
70
E
v Y
!i
1
60
50
50
60
70
80
% unsat (food records) FIG.2. ( a ) Relationship between average dietary saturated fat intake expressed as percentage of total fat intake and the average degree of saturation of abdominal fat determined by I3C MRS in the two groups of subjects. ( b ) The same for the degree of unsaturation. 0, Low-fat diet group: controls.
*,
FATTY ACIDS ANALYSIS BY "C MRS IN MAN
103
fatty acids in the abdominal fat in the diet group ( r = 0.56, n = 13). Figure 2 shows the average degrees of saturation and unsaturation in both groups. The graph demonstrates a clear separation of the diet-treated and the control subjects. DISCUSSION
The purpose of the present study was to investigate the ability of natural-abundance 13C MRS to discriminate the ratio of unsaturated versus saturated fatty acids in abdominal subcutaneous fat in individuals with different dietary habits regarding fatty acid saturation. The results demonstrated that there was a significant difference in the degree of saturation ( P < 0.0001 ) and monosaturation ( P < 0.0002) of adipose tissue as determined by 13CMRS between the diet and the control group. In contrast, no statistically significant difference between the two groups was seen in the degree of polyunsaturation of adipose tissue whereas the difference in the total degree of unsaturation (mono polyunsaturation) was highly significant ( P < 0.000 1 ). In agreement with the adipose tissue fatty acid composition, the evaluation of the dietary fatty acids demonstrated significant differences in the consumption of saturated and of unsaturated fatty acids between the two groups. The only disagreement was observed in the distinction between mono- and polyunsaturated fatty acids. Thus, the present study demonstrated a clear separation of the content of saturated and total unsaturated fatty acids in the diet group and the controls. The reason for the failure to distinguish the diet and the control groups regarding polyunsaturated fatty acid content is not obvious; due to the lower relative content of polyunsaturated fatty acids in adipose tissue, the I3C MRS technique may not have been sensitive enough to detect a difference. The advantages of using I3C MRS are that: ( 1 ) it is noninvasive and can thus be applied repeatedly on the same individual; ( 2 ) it permits the determination of the relative concentration of saturated and unsaturated fatty acids; ( 3 ) the determined concentration represents a local average, i.e., it is applicable to the study of various sites of adipose tissue in the body. I3CMRS has recently been used to determine the degree of unsaturation of adipose tissue fatty acids in normal volunteers ( 9 )and in patients with cystic fibrosis ( 14). Moonen et al. observed in two normal subjects individual relative contributions of 18.4 and 21.5% polyunsaturated, and 40.8 and 45.1% monounsaturated fatty acids, respectively ( 9 ) ;these results are in close agreement with the present data of control subjects. However, these authors ( 9 , 1 4 ) used decoupling for the assessment of the degree of unsaturation of fatty acids, i.e., two separate radiofrequency channels were required for 'H and 13Cwhich are not yet standard equipment of most whole-body scanners. In contrast, our procedure needs only one channel tuned to the 13Cfrequency. In addition, acquiring coupled spectra avoids determination of NOE factors. Although this study uses long repetition times and only a few acquisitions, the signal-to-noise ratio of the spectra is sufficient for quantification of the spectral lines (see Fig. 1 ). Our data are also in good agreement with fatty acid compositions determined with gas chromatography ( 3 , 1 2 ) . Since all the unsaturated carbons contribute signals to resonances 2 and 3 (see Fig. 1 and Table 2), the total degree of unsaturation can be determined from the I3C spectra without making any assumption about the fatty acid composition. However,
+
104
BECKMANN ET AL.
to distinguish between mono- and polyunsaturated fatty acids, the additional hypothesis is put forward that linoleic acid is the only significant polyunsaturated fatty acid in human adipose tissue, as discussed under Methods. This assumption is necessary since the in vivo I3C MRS spectra have a limited resolution, not providing separate information about all the different fatty acyl chains in adipose tissue. The consequence of this assumption is that the degree of polyunsaturation determined by Eq. [ 21 is to a certain extent larger than the real values, the error being proportional to the concentration of other polyunsaturated fatty acids like linolenic or arachidonic acid. It was of interest to note that the average consumption of fat in the control subjects of the present study was 44% (Table 3). This was close to the value of 43% reported previously ( 7) in a study about the dietary habits of the Swiss population. The controls consumed amounts of saturated fatty acids similar to that of the American population ( 15-20%) according to the NCEP report (15).Their average carbohydrateconsumption (38.0 k 7.6%) was similar to the average value of 44.0% reported for the Swiss population 10 years ago ( 7). In the diet group the average fat intake was reduced to 32.9% of total calories, and 11.7% of the consumed calories were saturated fats which was close to the recommendation (6). The major factor determining the time required for a change in the dietary fat type to be reflected in the fatty acid profile of adipose tissue is the turnover rate of individual fatty acids of adipose tissue. Our patients were following a diet half a year or more before the measurements, and a statistically significant difference between the groups could be seen in the degree of saturation, of monounsaturation, and of total unsaturated fatty acids as determined by MRS. Beynen et al. found that for a longer-term diet, there was a linear relationship between the degree of polyunsaturation of adipose tissue fatty acids using biopsies and the polyunsaturated fatty acid content of a longterm diet ( 4 ) . As these authors pointed out ( 4 ) , carbohydrate intake, alcohol consumption, energy expenditure, and conversion of saturated into monounsaturated fatty acids may all influence the percentage of saturated and monounsaturated fatty acids in adipose tissue. On the other hand Beynen et al. pointed to the fact that the content of polyunsaturated fatty acids in tissue was often lower than that of the diet ( 4 ) . Indeed, dietary fatty acids, before being incorporated into adipose tissue, are considerably diluted with de n o w synthesized fatty acids, especially when the absolute fat intake is low. The study described in ( 1 6 ) shows that the concentration of polyunsaturated fatty acids in adipose tissue of six different groups of subjects was lower than that of the diet. Four consecutive days of food records were considered acceptable as a reasonable compromise for minimal, reliable monitoring of diet compliance in outpatients ( 17). However, for achieving a 95% confidence limit 7 consecutive days of food recording were recommended ( 17). Consequently, the correlation between the food records and the MRS data in our study could be improved by extending the food recording period from 4 to 7 consecutive days. On the other hand, it must be emphasized that the control subjects followed no standard diet which might be one of the reasons that no significant correlation could be found between the degree of polyunsaturation of the fatty acids of the diet and of adipose tissue in this group, in contrast to the diet group. Recently MR imaging has been applied for measuring adipose tissue distribution in the human body ( 1 8 ) . It has been emphasized that there are large differences of
FATTY ACIDS ANALYSIS BY I3C MRS IN MAN
I05
fat metabolism in various regions of the human body. Therefore it would be of interest to compare the degree of unsaturation of fatty acids in different regions of the human body with I3CMRS and to assess the time course of changes of fatty acid composition after the dietary content of fatty acids has been changed. Furthermore, by combining imaging techniques with 13C spectroscopy it should be possible to obtain an assessment of both fatty acid composition and distribution in the human body. CONCLUSIONS
In this paper the application of 13C MRS for the noninvasive assessment of the relative concentration of mono- and polyunsaturated fatty acids of adipose tissue in two groups of volunteers was described. One group of subjects had followed a fatreduced diet for at least half a year before the 13C measurements. The second group consisted of controls who were on a high-fat but nonstandard diet. MRS data were compared with the fatty acid composition estimated by a dietician according to food records. The principal results were as follows: There was a highly significant difference between the two groups in the degree of saturation and monounsaturation of adipose tissue determined with ' 3C MRS No statistically significant difference was seen in the degree of polyunsaturation determined with MRS, although the diet-treated subjects ingested proportionally more polyunsaturated fatty acids than the controls. However, the difference in the total degree of unsaturation (mono + polyunsaturation ) was highly significant There was a significant positive correlation between the dietary intake of monounsaturated fats and the degree of monounsaturation of abdominal fat determined by 13CMRS. For the polyunsaturated fatty acids such a correlation could be found only in the diet group. Our data show that I3CMRS is capable of detecting differences in the total unsaturation degree of fatty acids between the groups. That no significant difference was found in the polyunsaturation degree, despite that this value for the controls agreed well with data from the literature (obtained with gas chromatography), might be explained by the fact that the proportion of polyunsaturated fatty acids incorporated into adipose tissue is usually smaller than their content in the diet. Therefore, I3C MRS might not have been sensitive enough to detect a difference in the polyunsaturation degree. The fact that the controls followed no standardized diet might be one of the factors for the observation of no correlation between the polyunsaturation degree of the diet and of adipose tissue. ACKNOWLEDGMENTS The contributions of Ms. A. Biichi and P. Joho (dieticians) are gratefully acknowledged. We also thank Dr. S. Muller, MR Center, for advice. Financial support was obtained from the Swiss National Science Foundation (Grant No. 4.889.85.18 to J.S.). REFERENCES
I. R. I. LOGAN,R. A. RIEMERSMA, M. THOMSON,M. F. OLIVER,A. G . OLSSON,G. WALLDIUS, S. ROSSNER,L. (1978).
KAIJSER,
E. CALLMER, L. A. CARLSON, L. LOCKERBIE, AND w . LUTZ, Lance1 1, 949
106
BECKMANN ET AL.
2. D. A. WOOD. S. BUTLER,S, R. A. RIEMERSMA, M. THOMSON,M. F. OLIVER,M. FIJLTON,A. BIRTHWHISTLE, AND R. ELTON,Lancet 2, 117 (1984). 3. R. A. RIEMERSMA, D. A. WOOD, S. BUTLER, R. A. ELTON, M. OLIVER,M. SALO.T. NIKKARI,E. P. PUSKA,F. GEY,P. RUBBA,M. MANCINI,AND F. RDANZA, Brit. Med. J. 292, 1423 VARTIAINEN, (1986). 4. A. C. BEYNEN,R. J. J. HERMUS,AND J. G. A. J. HAUTVAST,Am. J. Clin. Nzitr. 33,81 ( 1980). 5. M. B. KATAN, W. A. VANSTAVEREN, P. DEURENBERG, J. BARDENDEUSE-VAN LEEUWEN, C. GERMINGNOUWEN,A. SOFFERS,J. BERKEL,AND A . C. BEYNEN,Prog. Lipid Res. 25, 193 ( 1986). 6. American Heart Association. Arteriosclerosis 8, 218A ( 1988). 7. H. AEBI AND A. BLUMENTHAL, “Zweiter Schweizerischer Ernahrungsbericht,” Hans-Huber Verlag, Bern, 1984. 8. S. W. Soucr, W. FACHMANN, AND H. KRAUT,“Die Zusammensetzung der Lebensmittel,” 4th ed., Wissenschaftliche Verlagsanstalt, Stuttgart, 1989. 9. C. T. W. MOONEN,R. J. DIMAND,AND K. L. COX,Mugn. Reson. Med. 6, 140 (1988). 10. W. INSULLAND G. E. BARTSCH,Am. J. Clin. Nutr. 20, 13 (1967). 11. S. J. LONDON,F. M. SACKS,J. CAESAR, M. J. STAMPFER, E. SIGUEL,A N D W. C. WILLETT,Am. J . C h . Nutr. 54, 340 ( I99 1 ). 12. G . T. MALCOM,A. K. BHATTACHARYYA, M. VELEZ-DURAN, M. A. GUZMAN,M . C. OALMANN, AND J. P. STRONG,Am. J. Clin. Nutr. 50, 288 (1989). 13. F. W. WEHRLIAND T. WIRTHLIN,“Interpretation of Carbon-13 Spectra,” Wiley Heyden, London, 1983. 14. R. J. DIMAND,C. T. W. MOONEN,S. C. CHU, E. M. BRADBURY,G. KURLAND.A N D K. L. Cox, Pediatr. Rex 24, 243 ( 1988 ) . 15. Report of the National Cholesterol Education Program Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Arch. Intern. Med. 148, 36 ( 1988). 16. R. F. SCOTT,K. T. LEE, D. N.KIM, E. S. MORRISON,AND F. GOODALE,A m . J . C h . Nuir. 14, 280 (1964). 17. B. JACKSON, C. A. DUJOVNE,S. DECOURSEY, P. BEYER,E. F. BROWN,AND K. HASSANEIN, J . Am. Diet. Assoc. 86, 1531 (1986). 18. J. C. SEIDELL, C. J. G. BAKKER, AND K. VANDER KOOY, Am. J. C h . Nutr. 51, 953 ( 1990).
