sources. Gerbils. (6-16 animals per dietary group) were fed purified diets (21 with 0.01 to 0.08% cholesterol and. 38 cholesterol-free) for 4-week periods.
RESEARCH COMMUNICATIONS
Dietary
myristic,
palmitic,
cholesterolemia ANDRZEJ
Foster
PRAMOD
Research
Laboratory,
KEOSLA,
Brandeis
AND
Gerbils
(6-16
animals
per
dietary
group)
were
fed purified diets (21 with 0.01 to 0.08% cholesterol and 38 cholesterol-free) for 4-week periods. When cholesterolfree diets were fed, dietary 14:0 and 18:2 together accounted for 89% of the observed variation in TC. Although 14:0 consumption increased TC in a linear manner, the independent ability of 18:2 to lower cholesterol was nonlinear and exhibited a threshold effect at 5-6% dietary energy, above which further lowering of TC was less remarkable. In gerbils consuming cholesterolsupplemented diets, 87% of the observed variation in TC could be accounted for by a regression equation that included 14:0, palmitic acid (16:0), and a log function of 18:2 plus dietary cholesterol itself. These results demonstrate the applicability of gerbils for such studies and confirm previous observations in monkeys and humans that dietary 14:0 and 18:2 are the main fatty acids modulating plasma cholesterol under normocholesterolemic circumstances (i.e., when consuming lowcholesterol diets and lipoprotein metabolism is normal) whereas 16:0 also appears modestly hypercholesterolemic when LDL receptors are compromised (i.e., when dietary cholesterol or certain metabolic factors have encumbered lipoprotein metabolism).-Pronczuk, A., Khosla, P., Hayes, K. C. Dietary myristic, palmitic, and linoleic acids modulate cholesterolemia in gerbils. FASEB J. 8,
1191-1200
(1994)
Key Words: fatty threshold
controversy
0892-6638/94/0008-1
modulate
C.
HAYES’
Waltham,
Massachusetts
02254, USA
fatty acids and what effectdietary cholesterolhas on this relationshipand the underlying mechanism of action. In previous reports using nonhuman primates (4-8), as well as collaborative studies involving human subjects (9,
10), we have examined
the saturated
fat effect by focusing
on
specific fatty acids in the relative absence of dietary cholesterol,i.e., determining the impact of fattyacids in their own right.As a consequence of this inquiry, a retrospective analysis of data accumulated from feeding 16 different cholesterol-free fatblends to cebus monkeys (a speciesexceptionally sensitive to fatty acid effects) allowed generation of multiple regression equations [of the type originally produced by Hegsted and co-workers in human studies (11)] designed to predict the effectof individual fattyacids on the plasma cholesterol response (12). The percent of energy con-
tributed by dietary myristic acid (14:0) and linoleic acid (18:2) in the form of natural triglycerides explained almost 92% of the observed variation in cebus plasma cholesterol. Although dietary 14:0 increased the plasma cholesterol in a linear manner, dietary 18:2 decreased the plasma cholesterol concentration up to a threshold of 5-6% dietary energy, whereupon further increases in 18:2 intake were less effective. Because monkeys represent a limited resource, gerbils were rigorously examined as an alternative model because they have been recommended for such studies (13-15) and because their size and cost allow for more rapid determination of the points at issue. Our approach has been to feed gerbils edible fats and oils that provide fatty acids in concentrations normally present in whole-food diets, in part to attempt confirmation of results obtained in modeling similar studies with cebus monkeys and humans (12) or the somewhat divergent observations reported for hamsters (16). In all our studies we fed natural fats and oils, or their blends, on the assumption that natural triglyceride molecular configurations contribute substantially to the plasma cholesterol response
(17-19).
The data indicate that the gerbil, although more sensitive than either humans or cebus monkeys, responds to dietary fatty acids in a similar manner.
METHODS acids
.
plasma
cholesterol
.
regression analysis
THE EFFECT OF DIETARY FAT SATURATION on plasma cholesterol (TC)2 has been studied continuously for almost 40 years. From the earliestobservations to the more recent findingsof the Lipid Research Clinics Coronary Primary Prevention Trial (involving more than 6000 subjects),ithas been clearly shown that saturated fats raise plasma cholesterolwhereas polyunsaturated fatslower it(1-3).Despite thisvast body of
data,
K.
University,
ABSTRACT Previous studies with cebus monkeys and review of published human data indicated that 85% to 90% of the variation in plasma cholesterol (TC) could be explained on the basis of dietary myristic (14:0) and linoleic (18:2) acid intake in the absence of cholesterol, and that 16:0 contributed to cholesterolemia as dietary cholesterol was increased. Because monkeys are a limited resource, a more convenient, sensitive model was sought for investigating these dietary fatty acid and plasma lipid relationships. Accordingly, this report describes the results of multiple regression analysis of the TC response to dietary fatty acids based on 319 young, male Mongolian gerbils fed a total of 59 purified diets supplying about 40% energy as fat from single or blended fat
sources.
acids
in gerbils
PRONCZUK,
Biomedical
and linoleic
persists
concerning
191/$01.50. © FASEB
the effects of specific
The database used for these analyses represents a summary of results from a series of feeding trials (1990-1993) involving 319 male (50-90 g body wt)
‘To whom correspondence should be addressed, at: Foster Biomedical Research Laboratory, Brandeis University, Waltham, MA 02254, USA. 2Abbreviations: HLSO, high-linoleic acid safflower oil; HOSO, high-oleic acid safflower oil; GLC, gas-liquid chromatrography; I, intercept term; SATS, saturated fatty acids; POLYS, polyunsaturated fatty acids; MONOS, monounsaturated fatty acids; TGSR, triglyceride secretion rates; TC, plasma cholesterol.
1191
RESEARCH
COMMUNICATIONS
Mongolian gerbils (Tumblebrook Farm, West Brookfield, Mass). Gerbils (6 ± 3 months of age; mean ± SD) were fedpurified diets(4 wk per trial) (Table 1) containing 20% (w/w) of a single fat or a blend of naturally occurring fats, fed either without dietary cholesterol (38 diets) or with 0.01-0.08% added cholesterol (21 diets). Animals were housed in groups of five or six and were kept in a controlled environment with a 12 h light-dark cyde (light on 18:00 h). Based on preliminary experiments, diets met a predetermined caloric requirement for growth and maintenance, but not to excess, as the latter can result in severe hypertriglyceridemia in this species. After 4 wk of feeding, total plasma cholesterol was determined by enzymatic assay (Sigma Diagnostic kit, procedure 352, St. Louis, Mo.) on plasma samples obtained by cardiac puncture under light anesthesia after overnight fasting (16 h) in hanging wire-bottomed cages. All protocols and procedures were approved by the Brandeis University Animal Care and Use Committee. The total data set from experiments without dietary cholesterol consisted of 363 plasma cholesterol values generated from gerbils fed a total of 38 different diets (Table 2). The diets included a wide range of the most common fatty acids representing a percentage of dietary energy (% en) as follows: 12:0 (0-21% energy); 14:0 (0-10%); 16:0 (3-24%); 18:0 (1-13%); 18:1 (3-30%); 18:2 (0.7-30%), and 18:3 (0-0.3%). To ensure that the diets were essentially cholesterol-free, vegetable oils and cholesterol-stripped fats were fed. Dietary fat represented a single oil or fat blend. The following fats and oils were used: coconut oil, corn oil, cocoa butter, cottonseed oil, olive oil, soybean oil, palm oil, palm stearin, palm olein, high-oleic acid safflower o,l (HOSO), high-linoleic acid safflower oil (HLSO), cholesterol-stripped lard, cholesterol-stripped beef tallow, and cholesterol.stripped milk fat (cholesterolstripped fats provided by Source Food Technology Corp., Bumsville, Minn.). Experiments with dietary cholesterol generated 208 plasma cholesterol values from gerbils fed a total of 21 diets containing 0.01-0.08% cholesterol accompanied by fat blends representing a wide range of fatty acids (Table 3). Dietary fats included either a single oil or fat blends from edible coconut oil, milk fat, lard, and beef tallow (stock and cholesterol-stripped), beef tallow stearin and olein, lard olein, palm stearin, HOSO, HLSO, and olive oil. Because analysis of the dietary oils/fats revealed differences in the fatty acid composition from different batches (e.g., see Table 1, diets 2, 28, and 29), the fatty acid profile of all diets was analyzed by gas-liquid chromatography (GLC) using the one-step transestenfication method of Lepage and Roy (20). Fatty acid methyl esters were separated on a Hewlett-Packard 5790A GLC equipped with a flame ionization detector (Hewlett-Packard Co., Avondale, Pa.) and a 30-meter fused silica column (SP-2330, Supelco Inc., Bellefonte, Pa.). To generate the simplest regression equation to predict the quantitative effect of dietary fatty acids on plasma cholesterol, the observed plasma cholesterol (group mean for a given diet) was regressed against the dietary energy (% of total) contributed by the specific fatty acid (or acids) and (in the case of diets with cholesterol) against the concentration of dietary cholesterol as described previously (11, 12). Calculations were carried out on a Macintosh Plus computer (Apple Systems Inc., Cupertino, Calif.) using Statview 512’ (Brain Power Inc., Calabasca, Calif.) and Cricket Graph (Cricket Software Inc., Philadelphia, Pa.) statistical packages. Although it is possible to generate a multiple regression from a minimal number of diets, the applicability of such an equation to real-world physiology of a larger population is questionable (21). Accordingly, in these experiments we fed a relatively large number of diets covering a wide range of intakes for the key fatty acids. Selection of the most physiologically meaningful regression equations depended on three criteria that have been discussed previously (12). Briefly, the intercept term (I) obtained for the regression equation should approximate the average total plasma cholesterol obtained in response to the dietary fats fed; r2 (an estimate of the total variance explained by the regression equation) for any given multiple regression equation should be greater (i.e., better) than the value obtained with a regression equation based solely on one fatty acid; and the standard deviation about the regression line (SD) should be as small as possible. A ‘simple’ equation refers to one that fulfilled all of these criteria (i.e., generated the best equation) using the fewest number of descriptors (individual fatty acids, dietary cholesterol) to define a regression with a low SD.
TABLE
1. Composition
of pursfied gerbil diets’
% energy
Ingredient
g/l00 g
Casein
22.2 23.0
21.4 24.2
13.3 15.0
10.9
20.0
43.5
Cornstarch Glucose
Cellulose Fat (different
for each diet)b
5.0
Mineral mix’ Vitamin mix’1 Choline chloride Cholesterol ‘Diets
1.2
0.3
(varied were
with study)’
fed as starch
0 or 0.01-0.08
gel blocks,
prepared
by withholding
60 g of
cornstarch (per kg of mix) from the formulation and premixing it with 800 ml of simmering water to form a gelling to which the remaining ingredients were blended. #{176}Diets contained fat or fat blends as detailed in the Methods section and tables. ‘Ausman-Hayes mineral mix, BioServ, Frenchtown, N.J. dHayesCathcart vitamin mix (g/kg of mix): dl-atocopheryl acetate (500 IU/g), 15; inositol, 5; niacin, 3; calcium pantothenate, 1.6; retinyl palmitate (500,000 IU/g), 1.5; cholecalciferol (400,000 IU/g), 0.1; menadione, 0.200; biotin, 0.020; folic acid, 0.200; riboflavin, 0.700; thiamin, 0.600; pyridoxine HCI, 0.700; cyanocobalarnin, 0.001; and dextrin, 972. ‘Diets ranged from 0.013 to 0.08% cholesterol, when added.
treme sensitivity(threefoldrange) of thisspecies to changes in the dietary fatty acid profile in the absence of dietary
cholesterol. The sensitivity was evident at both ends of the scale, i.e., extremely low plasma cholesterol values were generated tremely
by high elevated
intakes values
of 18:2 (safflower resulted from
oil) whereas consumption
exof
14:0-rich fats (coconut oil and stripped milk fat). As a first step in regression analysis, the role of total saturated (SATS), polyunsaturated (POLYS), and monounsaturated (MONOS) fatty acids was considered (Table 4). Based on the criteria outlined in Methods, the most superior regression equation was:
plasma
cholesterol
in mg/dl (TC) 0.39
-
=
63
+
3.28
ESATS
EPOLYS
(Eq. 1)
where
ESATS
(% en)
supplied as saturated and EPOLYS
energy
from polyunsaturated
The predictive
denotes
power
the
percentage
of dietary energy
the percentage of fatty acids for any given diet.
of the correlation
(r2) for Eq.
1 (0.748)
was relatively high (P = 0.0001). However, the equation assigned minimal cholesterol-lowering power to polyunsaturates and an exaggerated cholesterol-elevating power to saturates, especially when compared to that described previously for humans (11, 22). In addition, the intercept term, I, was only half the expected value. Multiple
regression of individual
fatty
acids
RESULTS A totalof 127 multiple regression equations were generated Response
to fatty
acids
in the absence
of dietary
cholesterol Multiple
regression
polyunsaturated
of total
fatty
saturated,
monounsaturate4
and
acids
Group mean plasma cholesterol concentrations for gerbils in response to all 38 cholesterol-free dietsranged from 67 to 209 mg/dl (mean ± SE, 118 ± 2, n = 363), indicating the ex-
1192
Vol. 8
November
1994
The
to assess the impact of individual fatty acids as variables. When evaluated for the predictive power of a single fatty acid, the most hypercholesterolemic among all the saturated fatty acids was myristic acid (14:0) (Fig. 1A). Using only the dietary concentration of 14:0 explained 62% of the observed variation in plasma cholesterol (Table 5, Eq. 2). Linoleic acid (18:2) had the only appreciable cholesterol-lowering effect, and although the correlation between TC and this fatty acid was nonlinear, 18:2 exerted the strongest predictive
FASEB Journal
PRONCZUK
E AL.
RESEARCH COMMUNICATIONS TABLE
2. Percent energy
from
dietaryfatty
acids
and plasma
cholesterol
concentrations
Dietary Diet’ (n)b
12:0
14:0
16:0
frd cholesterol-free diets for
of gerbils
fatty acids
18:0
18:1
Percent
4 weeks
dietary
18:2
18:3
TC’
energy
1(8)
1.11
4.83
13.50
6.09
10.66
1.24
0.10
142 ± 7
2 (8)
0.00
0.92
8.40
7.00
16.70
0.70
0.12
159
3 (8) 4 (8) 5 (9)
0.00 0.10
0.36 0.24
8.40 10.50
5.32 13.00
18.80 14.10
17.76 0.10 0.00 0.00
10.08 0.76
5.64 23.10
1.64 1.90
3.68 11.00
4.50 1.50 0.95
0.20 0.00 0.00
99 ± 4 120 ± 7 201 ± 18
2.90
0.00
2.90 3.50 3.70
0.84 0.92 1.80
30.50 5.90 6.50
5.60 29.40 26.00
0.10 0.00 0.00
130 ± 11 81 ± 3
6(7)
±
8
9 (9)
0.20
0.16 0.16 0.32
10 (9)
0.20
0.21
3.70
1.10
5.90
28.40
0.00
67
11(9)
0.04
1.60
5.70 13.56
3.20 4.76
12.20 10.64
16.20 3.04
0.00 0.24
85 ± 4
12 (7) 13 (7) 14 (7) 15 (7) 16 (16) 17 (15) 18 (15) 19 (14)
0.38 4.12
1.16
2.96
11.16
4.36
10.80
1.20
2.60 1.50
8.80 6.34
2.92 2.16
8.48 7.20
7.84
4.00
1.32
2.92
0.00
121 ± 4 93 ± 7 81 ± 6 71 ± 5 196 ± 8
6.72 3.04
3.92 3.20
1.24 1.12
3.48 4.64
5.96 19.88
0.00 0.00
105 ± 4
7(12)
8 (12)
0.64 19.40
15.96 7.20
8.44
75 ± 5 87 ± 4
0.01
14.90 21.90
0.16 0.00
1.08
163
2.20
1.00
2.68
1.08
5.16
27.32
0.00
1.88 0.20
0.50 0.13
2.88 2.80
1.04 1.00
5.16 5.40
27.73 30.16
0.00 0.00
0.00 0.00 0.00
1.00 0.96 0.84
10.00 9.76 9.32
7.16 7.04 6.56
17.72 16.52 14.48
2.00 3.30 6.40
0.08 0.12 0.16
25 (6) 26 (7) 27 (7) 28 (7) 29 (10)
0.00 19.88 1.27
0.68 8.24 4.32
7.12 4.45 14.04
5.68 1.31 4.88
14.56 3.09 11.32
10.60 0.96 1.04
0.24 0.06
1.08
10.16
8.04
18.08
0.80
0.08 0.00
133
0.00 0.00
1.00
9.60
7.40
19.16
0.72
0.00
30 (10)
0.00
0.92
8.72
6.64
17.96
3.72
0.00
20 (15) 21(16) 22 (7) 23 (7)
24 (6)
87
± 3
±
5
±
4
79 ± 5 79 ± 3 132 ± 10 110 ± 3 92 ± 9 87 ± 4 209 ± 14 ±
13
140 ± 11
31(10)
0.00
0.64
7.64
7.86
19.52
2.96
0.10
32 33 34 35
0.00 0.00 0.05
0.84 1.08 0.52
9.20
10.88 7.43
6.08 6.16 3.01
18.04 18.40 22.37
3.64 1.48 4.81
0.32 0.00 0.20
155 ± 107 ± 114 ± 107 ± 122 ± 101 ±
(8)
36
(13)
1.90 0.80 2.80
9.74 2.50 9.20
4.45 4.20 2.90
0.07 0.00
111 165
± 5 ± 6
0.20
135
±
0.80
13.20
3.00
0.30
120 ± 8
(9) (12) (7)
0.40
0.76
21.00
7.40
4.40
4.00
22.54 3.90 13.10
12.70
4.10
4.30
37 (9)
38 (10)
9 11 8 8 8 5
10
‘Dietswere fed with fat contributing 40% energy. All diets were devoid of cholesterol (see Table I). The fattyacid composition of each diet was determined by GLC. Dietary fatswere formulated (usingeithera singleoil/fat or blends of oils/fats) as follows(dietnumbers appear in boldface):1, 27, milk fat(chol.-stripped); 2, 28, 29, beef tallow(chol.-stripped); 3, lard(chol.-stripped); 4, cocoa butter;5, 26, coconut oil;6, palm stearin;7, higholeicsaffloweroil (HOSO); 8, high-linoleic safflower oil (HLSO); 9, 18% HLSO + 2% beef tallow (chol.-stripped); 10, 19% HLSO + 1% palm oil; 11, 7% beef tallow(chol.-stripped) + 13% corn oil;12, 19% milkfat(chol.-stripped) + 1% HLSO; 13, 16% milkfat(chol.-stripped) + 4% HLSO; 14, 12% milk fat(chol.-stripped) + 8% HLSO; 15,8% milkfat(chol.-stripped) + 12% HLSO; 16, 19.8% coconut oil + 0.2% HLSO; 17, 17% coconut oil + 3% HLSO; 18, 8% coconut oil + 12% HLSO; 19, 3% coconut oil + 17% HLSO; 20, 1% coconut oil + 19% HLSO; 21, 0.2% coconut oil + 19.8% HLSO; 22, 19% beef tallow(chol.-stripped) + 1% HLSO; 23, 18% beef tallow(chol.-stripped) + 2% HLSO; 24, 16% beef tallow(chol.stripped)+ 4% HLSO; 25, 14% beef tallow(chol..stripped) + 6% HLSO; 30, 18% beef tallow(chol.-stripped) + 2% corn oil;31, 18% beef tallow (chol.-stripped) + 2% cottonseedoil;32, 18% beef tallow(chol.-stripped) + 2% soybean oil; 33, 18% beef tallow (chol.-stripped) + 2% palm olein; 34, 14% oliveoil/6% tailow(chol.-stripped); 35, palm stearin 18.4%/l.6% HLSO; 36, 17.5% coconut oil/ 2.5% HLSO; 37, 1% oliveoil/2% HLSO/3% coconut oil/4.0% palm stearin/ll% milk fat (chol.-stripped); 38, 9% olive oil/11% coconut oil. ‘Number of gerbils. ‘mg/dl totalplasma cholesterol. Mean ± SEM.
power
overall,
cholesterol
alone
variation
predicting
64%
of
the
plasma
(Table 5, Eq. 6; Fig. IB). The combina-
tion of 14:0 and 18:2 explained 89% of the variation (Eq. 8), which did not change after inclusion of palmitic acid (16:0) (Eq. 9) and only increased to 92% by including oleic acid (18:1) (Eq. 10). However, intercept terms became aberrant (equations less acceptable) when 18:1 was included. No satu-
FATTY ACIDS AND
PLASMA
CHOLESTEROL
rated fatty acid other than cholesterolemic,
either
alone
14:0 or
was consistently hyperin combination with other
fatty acids. Neither dietary oleic acid (18:1) alone (Eq. 5) nor linolenic acid (18:3) alone (Eq. 7) revealed a significant relationship with plasma cholesterol. From the final assessment of all multiple regression analyses that used two or more fatty acids
to predict
the gerbil total cholesterol
response
in the ab-
1193
RESEARCH COMMUNICATIONS TABLE
3. Pe rcent energy fro m dietary fatty
acids, concentra lion of cholest erol in diets, and plasma
Dietary Diet’(n)’
12:0
14:0
1(8) 3(8)
0.96 0.88 0.04
4(8)
0.02
5(8)
0.03
3.64 3.48 0.76 0.92 1.36
10.80 10.12 9.56 8.36 10.64
6(8) 7(8) 8(8)
0.03
1.00
8.32
0.02
0.44
9(10) 10(11) 11(10) 12(12) 13(11) 14(12)
0.02 0.04
0.36 0.40
9.04 8.40 8.20
0.12 0.12 0.12 19.72 19.72
0.60 0.60 0.60 7.52 7.52
15(12) 16(11)
19.72 0.00
17(10)
18(11) 19(12)
2(8)
16:0
18:0
chotes terol of gerbils fed diets with choles terolfor
4 weeks
fatty acids
18:1
18:2
18:3
1.88
2.00 0.68 0.72 0.68
0.12 0.12 0.12 0.12 0.08
Diet chol.’
TCd
6.20 5.88 7.48 6.96
11.44 12.60 17.20 18.68
8.36 5.44
15.28 19.48
1.08
0.12
51
190 ± 13 174 ± 10 192 ± 6 205 ± 7 174 ± 7 150 ± 7
18.20 18.76 20.04 12.32 12.32 12.32 3.76 3.76
4.44 4.52 6.00 2.84 2.84 2.84 2.80 2.80
0.20
34
111
0.24
106
128
21.76 21.76 21.76 4.42 4.42
5.12 5.31 2.40 1.81 1.81 1.81 1.40 1.40
0.28 0.05 0.05 0.05 0.03 0.03
31 47 94 189 47 94
100 122 139 166 147 149
7.52 0.30
4.42 5.65
1.40 3.01
3.76 26.01
2.80 2.94
0.03 0.06
189 47
166 7 93 ± 5
0.00
0.30
5.65
3.01
26.01
2.94
0.06
94
0.00
0.30
5.65
3.01
26.01
2.94
0.06
189
0.00
0.39
5.07
2.70
9.01
21.66
124 106 72 106 35
0.02
0.00
0.39
5.07
2.70
9.01
21.66
0.02
94
21(12)
0.00
0.39
5.07
2.70
9.01
21.66
0.02
189
4 5 6
± ± 69 ± 71 ± 92 ±
6
91 121
47
20(10)
± ± ± ± ± ± ± ± ±
5 5 7 6 4
10 3 3 4
‘Dietswere fed to gerbilswith fatcontributing40% energy and variouslevels(0.01-0.08%) of cholesterol (dietcomposition,Table 1.).The fatty acid composition of each dietwas determined by GLC. Dietary fats were formulated (using either a single oil/fat or blends of oils/fats) as follows (diet number appears in boldface): 1, milk fat (stock); 2, milk fat (chol.-stripped) + chol.; 3, beef tallow (stock); 4, beef tallow (chol.-stripped) + chol.; 5, beef tallow stearin; 6, beef tallow olein; 7, lard (stock); 8, lard (chol.-stripped) + chol.;9, lardolein;10-12, palm stearin + chol.; 13-15, 18.6% coconut oil + 1.4% HLSO + chol.; 16-18, 14% olive oil + 6% beef tallow (chol.-stripped) + chol.;19-21, 15% HLSO + 5% beef tallow (chol.-stripped) + chol. ‘Number of gerbils. mg cholesterol/l000 kcal. dmg/dl total plasma cholesterol. Mean ± SEM.
sence of cholesterol, porated the dietary and
the simplest and most descriptive incorconcentrations (% energy) of only 14:0
18:2: TC
=
126
+
7.6
(±
0.8)
E14:#{248}-
equation
40 (± 4) log E18:2 (Eq.
nary effectiveness of using the dietary 18:2/14:0 % en ratio (for all 38 diets tested) to predict the total plasma cholesterol under these circumstances. Figure 2 compares the observed and predicted TC values for all diets using the simplest final
(Eq.
8).
8)
The correlation coefficient for this equation was 0.944, indicating its ability to explain 89% of the observed variation (r2) in plasma cholesterol in the absence of dietary
cholesterol. The standard deviation about the regression was 12 mg/dl. The intercept value of 126, predicting the average plasma cholesterol from all dietary fats consumed, was close average (118 mg/dl). This equation was substantially better, i.e., more predictive than that based on total SATS and POLYS (Table 4, Eq. 1).
4. Characteristics of regression equations for plasma cholesterol based on total saturated monounsaturated, and polyunsaturated fatty acids in gerbils fed 38 dietary fats without cholesterol TABLE
Reg ression
coefficients
to the actual
Test of predictability
A best and simplest
regression
equation
was calculated
ini-
tiallyupon 1-15). This was rather loped from
completion of the first15 diets (Table 2, diets equation (PC = 128 + 7.9 E10 37 log E18:2) similar to the final equation (Table 5, Eq. 8) deveall38 diets.Predictabilityof thisinitialequation was tested by feeding six additional diets (Table 2, diets
16-21). Blends of coconut oil and high-linoleic safflower oil were fed to young gerbils to deliberately vary the intake of 14:0 and 18:2. Table 6 reveals that the observed cholesterol values
P
