Effect of Dietary Fats and Carbohydrate on BloodPressure of Mildly Hypertensive PatientsFRANK M. SACKS, IAN L. ROUSE, MEIR J. STAMPFER, LOUISE M. BISHOP,

CORNELIA F. LENHERR, AND RAYMOND J. WALTHER

SUMMARY The effect on blood pressure (BP) of replacing dietary saturated fat with either poh/un-saturated fat (linoleic acid) or carbohydrate was studied in 21 untreated mildly hypertensive patients.In a randomized, double-blind, crossover protocol, all subjects received dietary supplements ofcream, safflower oil, and carbohydrate in random sequence, each prepared in flavored yogurt ormilk. Each supplement was administered for 6 weeks and followed by a 4-week washout period of nosupplementation. Dietary linoleic acid increased from 4.6 to 13% of energy intake when the saffloweroil replaced cream, while saturated fat decreased from 16 to 10%. Total fat intake was 37 to 38%during the cream and safflower oil periods but was 28% during the carbohydrate period. Compliancewith the diets was demonstrated by significant changes in fasting plasma fatty acid measurements.Mean clinic BP was 135 ± 9/93 ± 6 mm Hg at baseline. There were no significant differences in BPmeasured in the clinic or at home among the three dietary periods. The protocol had more than 80%power to detect a mean effect of diet of 3 mm Hg systolic or 2 mm Hg diastolic BP. Therefore, replacingdietary saturated fat with carbohydrate or with linoleic acid does not affect BP in subjects with mildhypertension. (Hypertension 10: 452-460, 1987)

KEY WORDS • dietary fats * hypertension • blood pressure • plasma fatty acidslinoleic acid • polyunsaturated fats • ambulatory blood pressure measurements

VEGETARIANS living in industrialized soci-eties and eating a diet that is relatively low intotal fat and saturated fat but high in polyun-

saturated fat have low mean blood pressure (BP), andin contrast with omnivores, their BP rises only slightlywith age.1"3 Within vegetarian groups, those eatingcomparatively more animal products have higherBP.1-4 However, the changes in BP have not beenconsistent when vegetarian diets were given to non-vegetarians or when animal products were added tovegetarian diets.5^ Two groups of researchers found

From the Channing Laboratory, Department of Medicine,Harvard Medical School and Brigham and Women's Hospital, andthe Clinical Research Center, Brigham and Women's Hospital,Boston; and Harvard University Health Services, Cambridge,Massachusetts.

Supported by Grants HL34586 and RR02635-O1. This work wasdone during the tenure of a Clinician-Scientist Award from theAmerican Heart Association to Dr. Sacks with funds contributed inpart by the Massachusetts Affiliate. Dr. Rouse is a Neil HamiltonFairley Fellow of the National Health and Medical Research Coun-cil of Australia.

Address for reprints: Frank M. Sacks, M.D., Channing Labora-tory, 180 Longwood Avenue, Boston, MA 02115.

Received January 5, 1987; accepted April 28, 1987.

that increasing the proportion of dietary polyunsaturat-ed to saturated fatty acids in nonvegetarian diets low-ered BP in both normotensive and hypertensive sub-jects.*"13 However, another group studying similardiets in normotensive subjects could not replicate theseresults.14

Differences in the dietary protocols may have con-tributed to the conflicting results of studies of vegetar-ian diets and of modifications in dietary fat intake. Insome studies, saturated and polyunsaturated fats weretested against each other, but in other studies, the fatreplaced carbohydrate or protein. Moreover, the di-etary interventions always affected nutrients in addi-tion to the fats; these nutrients could have either low-ered BP directly or perhaps blocked an effect of dietaryfats on BP. Many of these studies did not use random-ized controls and none were double-blind, leavingopen the possibility of a biased result.

Dietary linoleic acid has been proposed to lower BPby its conversion to arachidonate, the substrate forcyclooxygenase and prostaglandin synthesis.9-10 Somestudies have found that hypertensive patients havelower levels of vasodilating prostaglandins than donormal controls.15 Conceivably, a low intake of linole-

452

by guest on May 21, 2018

http://hyper.ahajournals.org/D

ownloaded from

DIETARY FAT AND BLOOD PRESSUREASacfcy et al. 453

ic acid or slowed conversion of linoleate to arachidon-ate could cause low prostaglandin levels in hyperten-sive subjects and could be corrected by augmenting thedietary intake. Therefore, we conducted a double-blind dietary trial to test whether replacement of satu-rated fats either by linoleic acid or by carbohydratelowers BP in hypertensive subjects.

MethodsSubjects

Subjects were recruited from the clinical practices ofstaff physicians at Harvard University Health Services(HUHS; Cambridge, MA, USA). Male or female pa-tients at HUHS were eligible if they had mild hyperten-sion (defined as a pretreatment diastolic BP between90 and 104 mm Hg inclusive) and were being treatedeither nonpharmacologically or with no more than twoantihypertensive drugs. Subjects were not eligible ifthey were obese (body weight > 20% of ideal weightfor height according to Metropolitan Insurance Com-pany tables), or had diabetes mellitus, hyperlipidemia(defined by the 95% limits for cholesterol and triglyc-eride at the HUHS clinical laboratory), or end-organdamage (defined as overt cerebrovascular or coronaryvascular disease), left ventricular hypertrophy, abnor-mal serum creatinine levels, or hypertensive retinalabnormalities. Each prospective subject had a detailedclinical and laboratory evaluation to determine eligi-bility and to exclude secondary forms of hypertensionaccording to standard guidelines.16

Thirty-one hypertensive subjects were referred tothe study by physicians at HUHS. Of this group, threesubjects declined to participate due to the time de-mands of the protocol and four subjects were excluded:one person, apparently misdiagnosed as hypertensive,whose diastolic BP was consistently below 90 mm Hgon repeated measurement; one with established hyper-tension, whose diastolic BP remained over 104 mm Hgafter withdrawal from medication; and two with estab-lished hypertension, who had BP below 85 mm Hgduring the initial withdrawal from medication. Afterthe remaining 24 subjects commenced the study, threedropped out for varied personal reasons unrelated tothe supplements or their BP levels. The final studygroup consisted of 21 subjects who completed allphases of the study. Of these subjects, 16 were men(age range, 26-63 years) and five were women (agerange, 30—62 years).

Experimental ProtocolThe baseline period consisted of 1 week of measure-

ments of BP and other variables (described in a subse-quent section). Prior to the baseline period, subjectswho were receiving antihypertensive medication werewithdrawn from drugs for 8 weeks during which timetheir BP was monitored. Subjects were entered into thestudy if their baseline mean diastolic BP was between85 and 104 mm Hg inclusive.

Following the baseline period, the subjects wererandomly allocated in blocks of six, using a random

number table, into one of the six possible orders ofadministration of the three following dietary supple-ments: 1) polyunsaturated fat — 30 ml of safflower oilthat was deodorized and decolorized by standard com-mercial processing techniques (California Fats andOils, Richmond, CA, USA), 2) saturated fat — 81 mlof cream (Hood Dairy, Charlestown, MA, USA), 3)low fat, high carbohydrate — 68 g of powdered skimmilk and 40 g of polysaccharide (Sumacal; OrganonNutritional Products, West Orange, NJ, USA). Thesupplements were mixed with an emulsifier of mono-glycerides and diglycerides, 1.3 g (DUR-EM 117;Durkee Foods, Strongsville, OH, USA), in flavorednonfat yogurt or skim milk as a milk shake. Threeflavors were used and rotated. The nutrient content ofthe supplements is shown in Table 1. Each subject wasrequired to eat an 8-oz (224-ml) portion of yogurt ordrink a 12-oz (336-ml) milk shake daily. Aside fromthe dietary supplements, the subjects were free to eattheir usual diet, modified to allow incorporation ofthe dietary supplement. The team dietitian (L.M.B.)counseled the subjects to make the necessary changeswith a goal of minimizing increases in total calories orbody weight. Generally, the supplements were taken atlunchtime. Each supplement was consumed for 6weeks. The periods of dietary supplementation wereeach separated by a 4-week washout period in whichthe subjects consumed their usual diet.

Double-blind conditions were maintained through-out. No researcher who had any contact with the sub-jects knew of the treatment assignments. The supple-ments were prepared and distributed weekly, labeledonly with patients' identification numbers. No subjecthad the opportunity to compare directly the taste of thedifferent supplements.

Dietary AssessmentTo quantitate the diet at baseline and during the

three periods of supplementation, a semiquantitativefood frequency questionnaire was completed by eachsubject. This questionnaire contained 116 food-related

TABLE 1. Nutrient Content of the Dietary Supplements

Nutrient

Calories (kcal)

Protein (g)

Carbohydrate (g)

Saturated fat (g)

Polyunsaturated fat (g)

Cholesterol (mg)

Sodium (mg)

Potassium (mg)

Calcium (mg)

Magnesium (mg)

Vitamin E (mg)

Carbohydrate

484

19.2

99.1

1.1

0.1

11.3

324

836

587

70

0

Supplement

Safflower oil

486

11.4

38.1

4.2

22.1

8.2

174

525

371

48

11.5

Cream

502

14.2

40.6

17.9

1.2

89

200

597

432

52

0.6

by guest on May 21, 2018

http://hyper.ahajournals.org/D

ownloaded from

454 HYPERTENSION VOL 10, No 4, OCTOBER 1987

items, each with a specified portion size, and ninepossible categories of frequency of ingestion. The sub-jects were questioned by a dietitian (L.M.B.) whenclarification of responses was needed. This question-naire has been extensively validated against diet rec-ords and biochemical indices of nutrition in populationstudies and in clinical trials.l7~19 The questionnaire wasoptically scanned and the data analyzed at the HarvardUniversity Computing Center using a nutrient database complete for all listed nutrients.

In addition to the questionnaires, each subject com-pleted a 7-day diet record at baseline and during eachperiod of supplementation. A dietitian used these rec-ords to show each subject how to incorporate the sup-plements into the usual diet and to verify that theinstructions were followed.

BP MeasurementsClinic Readings

BP was measured on one random zero sphygmoma-nometer by a single observer (L.M.B.) throughout theprotocol. The observer was trained at the East BostonNeighborhood Health Center, a performance site of theHypertension Detection and Follow-up Program,using that program's protocol.20 BP was measured oneach of 3 consecutive days during the baseline period,at the midpoint of the supplementation periods, and inthe final week of supplementation and washout peri-ods. Before each measurement, the subjects sat quietlyin a chair for 5 minutes. Then three readings weretaken in succession from the left arm. BP was mea-sured at the same time of day in each subject through-out the protocol, either in the morning or in the earlyafternoon.

Home ReadingsAll subjects were instructed to measure their own

BP for 7 days during the midpoint and last week of thedietary supplementation periods and during the lastweek of the two washout periods using a fully automat-ed machine (Lumiscope, Edison, NJ, USA). Home BPreadings were not obtained during the baseline period.On each day during these week-long assessment peri-ods, subjects were requested to make duplicate deter-minations in the morning, late afternoon, and evening.One machine was allocated to each subject for theentire study and was used to obtain all of the readingsin that subject. The machine weighs 28 oz (784 g). Itautomatically inflates the cuff to a level that is 20 mmHg over the subject's previous systolic BP reading,decreases the pressure by 2 mm/sec while displayingthe pressure that is in the cuff, and provides a printedrecord of date, time, systolic and diastolic BPs, andheart rate. Each machine was validated in the follow-ing manner before it was allocated to a subject. First,the pressure transducer of each machine was tested byconnecting a Y tube from an inflatable arm cuff to amercury manometer and to the automated machine andobserving pressure readings from 0 to 200 mm Hg.Then, BP measurements of each automated machinewere compared with auscultatory measurements ob-

tained with a standard mercury sphygmomanometer(W.A. Baum, Copiague, NY, USA) by taking threeconsecutive simultaneous readings with both devicesfrom the left arm of four normal subjects. Each ma-chine was retested at the different assessment periodsagainst the mercury sphygmomanometer. The meandifference between measurements made with the auto-mated machine and the mercury sphygmomanometerwas 0.15 ± 3.53 (SD) mm Hg for systolic BP and-1 .55 ±3 .45 mm Hg (p<0.001) for diastolic BPbased on 198 simultaneous readings.

Urine MeasurementsTwo consecutive 24-hour urine collections were

made at the end of each period of supplementation.Sodium, potassium, calcium, and creatinine were de-termined by autoanalyzer at the Core Laboratory of theClinical Research Center, Brigham and Women'sHospital. Results of the two collections were averagedfor each subject.

Plasma Fatty Acid MeasurementsA fasting sample of venous blood anticoagulated

with 3 mM edetic acid was taken at baseline and on 1day during the last week of each of the three periods ofdietary supplementation. Plasma fatty acids were mea-sured by extracting and transmethylating the lipidswith methanolic HC1.2' The solvent was evaporated,and the fatty acid methyl esters were redissolved inhexane and quantitated by gas-liquid chromatographyas follows: fused silica capillary column, 30 m x 0.24mm, SP 2330 liquid phase, 0.22-fj.m film thickness(Supelco, Belefonte, PA, USA); split injection, 100:1;hydrogen carrier gas at 48 cm/sec; flame ionizationdetector; Hewlett-Packard Model 5880A gas chro-matograph (Palo Alto, CA, USA) with peak area inte-gration; temperature program of 140 to 170°C at10°C/min, 170°C for 4 minutes, 170 to 190°C at10°C/min, 190°C for 2 minutes, 190 to 240°C at10°C/min. Peak retention times were identified by in-jecting known standards (NuCheck Prep, Elysium,MN, USA). Results are expressed as percentage of thetotal fatty acid methyl esters.

Body weight was measured in light clothing withoutshoes at baseline and at the midpoint and end of eachsupplement period using one portable scale.

Data AnalysisStandard methods for analysis of crossover trials

were followed for the analysis of changes in all of thevariables.22-23 Clinic BP for each period of measure-ment is defined as the mean of nine readings, three oneach of 3 days, whereas home BP is defined as themean of 42 readings, two at each of three times per dayfor 7 days, as described previously. The significanceof effects of treatment (diet) and of period (nonspecificor seasonal effects unrelated to the specific treatment)was tested using repeated-measures analysis of vari-ance (ANOVA).24 Where there was a significant effectof diet, differences between the effects of any twodietary supplements were examined with least-signifi-

by guest on May 21, 2018

http://hyper.ahajournals.org/D

ownloaded from

DIETARY FAT AND BLOOD PRESSURE/Sacfcs et al. 455

cant-difference tests.24 To evaluate whether there wereany carryover effects of one treatment onto the nexttreatment (order or interactive effects), we adopted thestrategy of Armitage and Hills,23 which models thechange in BP during the supplementation periods as afunction of period, diet, and their interaction. To deter-mine whether changes in BP were related to changes inother variables, multiple regression analysis was per-formed using the General Linear Models Procedures ofthe Statistical Analysis System.23 In all analyses, two-sided p values were used.

ResultsDiet

During supplementation with safflower oil as com-pared with cream, intake of polyunsaturated fat in-creased from 5 to 14% of total energy, whereas saturat-ed fat decreased from 16 to 10% and monounsaturatedfat decreased from 13 to 11 % (Table 2). There were nodifferences among these periods in intake of protein,total fat, carbohydrate, potassium, calcium, magne-sium, vitamin C, or vitamin E.

During both high fat periods, total fat constituted 37to 38% of total calories and carbohydrate 42 to 43%, ascompared with 28 and 52%, respectively, during con-sumption of the low fat carbohydrate supplement (see

Table 2). In comparison to the carbohydrate period,the higher dietary fat in the polyunsaturated phase wasaccounted for entirely by linoleic acid. Two thirds ofthe increase in fat intake during the saturated fat periodwas accounted for by saturated fat (myristic, palmitic,and stearic acids) and one third by monounsaturated fat(oleic acid). During ingestion of the carbohydrate sup-plement, there were small but statistically significantincreases in intake of energy, protein, potassium, cal-cium, and magnesium as compared with the high fatsupplements. Most of these increases resulted from thenonfat milk powder added to the low fat supplement torender it isocaloric to the high fat supplements.

Plasma Fatty AcidsTo verify that the subjects ate the supplements, fast-

ing plasma fatty acids were measured. Since in a pre-vious study,26 dietary supplements of safflower oil in-creased linoleate in all of the plasma lipid classes, wemeasured fatty acids in unfractionated plasma lipids.Compared with the saturated fat period or with thecarbohydrate period, intake of safflower oil enrichedthe plasma fatty acids with linoleate (18:2) from 31 to38%, whereas the proportion of most saturated fattyacids (14:0, 15:0, 16:0), oleic acid (18:1), and eico-sapentaenoic acid (20:5a>3) diminished significantly(Table 3). In all but one subject, plasma linoleate in-

TABLE 2. Daily Dietary Intakes and Body Weight at Baseline and During Dietary Supplementation in 21 Subjects

Nutrient

Energy (kcal)

Carbohydrate (g)% of kcal

Protein (g)% of kcal

Total fat (g)% of kcal

Saturated fat (g)% of kcal

Polyunsaturated fat (g)% of kcal

P/S ratio

Linoleate (g)% of kcal

Monounsaturated fat (g)% of kcal

Cholesterol (mg)

Dietary fiber (g)

Potassium (mg)

Calcium (mg)

Magnesium (mg)

Vitamin C (mg)

Vitamin E (mg)

Body weight (kg)

Baseline

2023 ±728

232 ±7746

75 ±2615

74 ±3433

27 ±1312

14±86.0

0.55

13±75.7

27±1312

312±118

19±8

3174±1105

849 ±367

217±73

294 ±243

39 ±95

77.9± 17.6

C

2482 ±585

321±6452

94 ±2115

78 ±2828

30±1211

14±65.0

0.49

12±54.4

28+1210

338+144

17±5

3796 ±835

1529 + 476

305 ±72

316 + 277

32±86

78.8±17.4

Supplement

P

2198±431

235 ±6043

76±1714

92+1737

25±810

34±314

1.42

32±313

27 ±711

268 ±11515±5

3297 ±726

1112 ± 380258 ±79

258±219

23 ±10

79.3+18.0

S

2297 ±571

241+6542

82±2114

98 + 2438

42±1116

14±65.4

0.33

12±54.6

33±1O13

375 ±120

17±6

3354 ±725

1191 ±334265 ±77

243 ±182

12±11

78.2± 17.7

Comparisons (p-value)*

C vsP

<0.001

<0.001

<0.001

0.001

0.011

<0.001

<0.001

>0.10

0.005

0.004

<0.001

<0.001

<0.001

>0.10

C vsS

0.008

<0.00I

0.002

<0.001

<0.001

>0.10

>0.10

0.003

>0.10

>0.10

<0.001

<0.001

<0.001

>0.10

P vsS

>0.10

>0.10

0.056

>0.10

<0.001

<0.001

<0.001

<0.001

<0.001

0.031

>0.10

>0.10

>0.10

0.008

Values are means ± SD. C = carbohydrate; P = polyunsaturated fat; S = saturated fat.•Comparisons of different supplements were made by least-significant-difference testing when the significance level

of the overall repeated-measures ANOVA was <0.10.

by guest on May 21, 2018

http://hyper.ahajournals.org/D

ownloaded from

456 HYPERTENSION VOL 10, No 4, OCTOBER 1987

TABLE 3. Composition of Fasting Plasma Fatty Acids During Dietary Supplementation with Carbohydrate and Fat in21 Subjects

Fatty acid(% of total)

C14:0

C15:0

C16:0

C18:0

C18:lco9

C18:2a>6

C18:3o>6

C18:3o)3

C20:2&>6

C20:3o>6

C20:4<o6

C20:5co3

Baseline

0.28 + 0.43

0.18+0.26

17.2±2.8

7.6±0.9

17.7±2.1

33.5±4.1

0.53±0.15

0.45 ±0.12

0.27 ±0.10

1.56±0.28

9.09±1.77

0.88 ±0.70

C

0.27 + 0.25

O.13±0.O6

17.4 + 3.2

7.9±1.0

18.9±2.7

31.3±4.5

O.63±0.17

0.46 + 0.13

O.28±0.10

1.87±0.36

8.80±1.57

0.65 + 0.25

Supplement

P

0.16±0.14

0.10±0.04

16.2±2.2

8.0±0.8

15.8 + 2.2

38.2±4.3

0.57±0.19

0.38±0.08

0.29±0.09

1.51 ±0.33

9.25+1.78

0.42±0.19

S

0.26±0.20

0.16±0.12

17.8 + 2.4

7.8±0.8

18.5±2.2

31.5±3.9

0.53 ±0.20

0.43 ±0.12

O.23±O.O6

1.64±0.28

8.81± 1.68

0.66±0.27

Comparisons (p-value)*

C vs P

0.002

>0.10

0.036

<0.001

<0.001

>0.10

0.012

>0.10

<0.001

<0.001

C vsS

>0.10

>0.10

>0.10

>0.10

>0.10

0.011

>0.10

0.034

0.004

>0.10

P vsS

0.004

0.013

0.005

<0.001

<0.001

>0.10

>0.10

0.013

0.081

<0.001

Values are means ± SD. C = carbohydrate; P = polyunsaturated fat; S = saturated fat.•Comparisons of different supplements were made by least-significant-difference testing when the significance level

of the overall repeated-measures ANOVA was <0.10.

creased during the safflower oil period over levels dur-ing the ingestion of cream or carbohydrate (Figure 1).Arachidonate (20:4), the precursor to prostaglandinsof the 2 series, and other desaturation and elongationproducts of linoleate, such as 18:3 and 20:3, did notincrease during ingestion of linoleate. Paradoxically,20:3 decreased during ingestion of linoleic acid, as wehad found previously in normotensive subjects.26 Thedietary saturated fatty acids and oleic acid in cream didnot raise the respective plasma fatty acid levels overlevels present during the carbohydrate period. Unlikesafflower oil, which is composed predominantly oflinoleic acid, cream contains moderate amounts of14:0, 16:0, 18:0, and 18:1, and the increments inthese individual dietary fatty acids may not have been

50

PLASMALINOLEATE(% of totaltatty acid

methyl esters)

40

30

20Saturated Linoleic Carbohydrate

TYPE OF DIETARY SUPPLEMENT

FIGURE 1. Changes in fasting plasma linoleate level in re-sponse to ingestion of safflower oil.

enough to raise the respective plasma levels. Alterna-tively, since humans can synthesize saturated fattyacids and oleic acid (but not linoleic acid, an essentialnutrient), plasma levels of the nonessential fatty acidsare likely to be less sensitive than essential fatty acidsto increments in the diet.

Blood PressureAt baseline, mean clinic BP was 134.9 ±9.5/

92.6 ± 5.7 mm Hg. There were no significant differ-ences in clinic BP among the three periods of dietarysupplementation (Table 4). In fact, BPwas remarkablystable throughout the protocol, including baseline andwashout periods (Figure 2), with no evidence of adownward drift in BP during the study, such as mightbe caused by regression to the mean or by subjectsbecoming accustomed to the procedure of measure-ment. The pooled intraindividual standard deviation ofclinic BP over the entire study was 4.2 mm Hg forsystolic BP and 3.0 mm Hg for diastolic BP. There-fore, with the use of ANOVA, the study had over 80%power to detect a treatment effect of 3 mm Hg systolicand 2 mm Hg diastolic BP with ap value of less than0.05.

The effect of replacing saturated fat with either poly-unsaturated fat or carbohydrate on home BP, like clinicBP, was strikingly null (see Table 4 and Figure 2).Home BP was significantly lower than clinic BPthroughout the study.

Body WeightIrrespective of the type of supplement eaten, weight

increased during the first supplement period and re-mained stable thereafter. Increments of body weightover baseline were 0.9 kg for the first supplementperiod, 1.1 kg for the second period, and 0.9 kg for thethird period. Analysis of changes in body weight by

by guest on May 21, 2018

http://hyper.ahajournals.org/D

ownloaded from

DIETARY FAT AND BLOOD PRESSURE/Sacfa et al. 457

TABLE 4. BP at Clinic and Home During Dietary Supplementation in 21 Subjects

BP (mm Hg)

SystolicClinicHome

DiastolicClinicHome

Baseline

134.9±9.5

92.6±5.7

Carbohydrate

135.9±7.4133.6±9.6

92.3 ±6.585.8±9.0

Supplement

Polyunsaturated

135.7±7.2134.8 + 8.0

91.8±6.487.5±7.1

Saturated

137.0±8.4133.7 + 12.6

93.6±6.386.7±8.5

SD*

4.35.1

2.93.6

RANOVA

^2.40

0.580.37

2.221.24

P

0.560.69

0.120.30

Values are means ± SD. RANOVA = repeated-measures ANOVA.*Within-person SD of differences in BP among the three dietary periods.

Systolic 1 4 ° "BP

(mmHg) 1 3 5 .

130

• = clinic BP O = home BP

IIl U

Diastolic 9 0 -

BP(mmHg) 8 5-

II* {I

I 2 j j

1 l

2 2

DIET Baseline Saturated Carbohydrate LlnolelcFat Acid

WEEKS ON 0 3 6 36 36SUPPLEMENT

B

Systolic 1 4 0

BP(mmHg)

130

I1

O • homt I

DiastolicBP

(mmHg)

9 0 -

8 5 -

t I

1

I

ho

I

2

I

ho

I

2

I

2

I

1

I

2

PEfUOO

WEEKS M

STUOT

• a*««n« tvqpUmmt 1 Without 1 luwtumn) 2 Washout 2 Suppl«fn«nt 3

0 1 • 10 11 l« » 21 M

FIGURE 2. A. Changes in BP during con-

sumption of dietary supplements of fat and car-bohydrate in a crossover study. B. Changes inBP in temporal sequence during the crossoverstudy.

by guest on May 21, 2018

http://hyper.ahajournals.org/D

ownloaded from

458 HYPERTENSION VOL 10, No 4, OCTOBER 1987

type of supplement showed that weight was signifi-cantly higher by 1.1 kg during the polyunsaturated fatsupplement than during the saturated fat supplement(see Table 2). There was no significant difference inweight between the carbohydrate period and either ofthe two fat periods.

Regression analysis showed no relationship betweenchanges in body weight and BP, and including changein body weight as a covariate of the effect of dietarysupplements on BP did not change the null result.

Urinary AnalysesThe average urinary excretion of sodium, potas-

sium, calcium, and creatinine in two 24-hour collec-tions was not significantly different among the dietaryperiods (Table 5).

Period and Order EffectsThe nutritional data, plasma fatty acids, BPs, and

urinary analytes were analyzed by sequence of supple-mentation (as well as by type of supplement) to assessdrift in these variables (period effects) over the courseof the study. No such effect was found for any of thevariables except body weight, as discussed previously.

In a crossover study, the possibility that one treat-ment may affect the results of the subsequent treatment(order or carryover effect) could seriously bias theoverall estimate of the treatment effects.22'n General-ly, the results of crossover trials are not examined forcarryover effects, but in this study we used the ap-proach of Armitage and Hills23 and found no evidencefor a carryover effect on BP from one supplement tothe next (p>0.10 for all analyses).

DiscussionThe present study was conducted to resolve some of

the inconsistencies in previous work on dietary fats andBP. The goal was a nutritional study patterned after therigorous principles often used in contemporary drugtrials. Thus, the study design was a double-blind, con-trolled, crossover trial with biochemical validation ofself-reported dietary compliance. This design mini-mizes the possibility of observing an effect on BPattributable to variables other than the dietary fats.

TABLE 5. Urinary Excretion of Cations and Creatinine DuringDietary Supplementation Determined from 48-Hour Collections

Urinaryexcretion

Creatinine(mg/day)

Sodium(mmol/day)

Potassium(nunol/day)

Calcium(mg/day)

Carbohydrate

1440 + 339

125+47

65 + 19

217+108

Supplement

Polyunsaturated

1530 + 471

123±43

60+19

223+109

Saturated

1390±410

116±60

60±24

192±81

Values are means ± SD of 21 subjects. Differences among themeans were not significant by repeaied-measures ANOVA.

The principal comparison in the present study is ofBP during the ingestion of supplements of saturated fatand linoleic acid. Dietary data show that adding thedietary supplements produced substantial alterations inoverall daily intake of saturated fat and of linoleic acid,covering the practical dietary range. The magnitude ofthe resulting change in fasting plasma linoleate is con-sistent with many previous studies.26"29 The total di-etary intake during these two dietary periods was verysimilar in nutrients other than fatty acids. However,BP, whether measured at the clinic or at home, wasvirtually unchanged by intake of the supplements ofsaturated and polyunsaturated fatty acids. The multi-plicity of individual readings of BP taken on eachsubject during each period dampened the inherentwithin-person variability in BP and provided sufficientpower to detect very small changes in BP.

We studied a third dietary supplement that was lowin fat and high in carbohydrate as an additional nutri-tional control to help distinguish the effects of eachtype of fat. Clinically, a low fat, high carbohydratediet is relevant since it is widely recommended to low-er blood cholesterol and risk of cardiovascular dis-ease.30"32 The carbohydrate supplement caused majorchanges in overall intake of fat and carbohydrate andsmall but statistically significant increases in the calcu-lated dietary potassium, calcium, magnesium, andprotein. Whereas the calculated increase in dietary po-tassium was 12 to 13 mmol/day, urinary potassiumexcretion over 48 hours was only 4 to 5 mmol/dayhigher on carbohydrate than on either supplement offat (p>0.10) , suggesting that a change in potassiumintake, if it did occur, was very small. Taken together,the existing literature would predict at most a hypoten-sive effect of the added cations of 1 to 2 mm Hg.33"36

Dietary protein has not been found to affect bloodpressure.3738 Moreover, ingestion of these nutrientswas unaffected during the exchange of saturated forpolyunsaturated fat. Therefore, the stability of BP dur-ing all three dietary periods suggests that exchanginglinoleic acid or carbohydrate for saturated fat does notaffect BP in mildly hypertensive subjects during a 6-week period of study.

Over the past 10 years, many studies have reportedthat removing some saturated fat from the diet andreplacing it with either polyunsaturated vegetable oilor carbohydrate lowers BP.9"13 39 Most of the experi-mental designs involved a baseline period of a usualhigh saturated fat diet, a test diet, and a final period ofthe usual diet.9"13-M The diets of the three periods wereeither self-selected, with instruction for the purchaseand use of the foods for the test diet,9' "" '3 'w or com-pletely controlled.10 Except for one study12 that will bediscussed subsequently, the apparent hypotensive ef-fects of diets low in saturated fat or high in polyunsat-urated fat were based on comparing BP at the baselineand/or the final period with BP measured at the end ofthe test period. Comparisons of treatment period withbaseline BP when there is no randomized control groupare difficult to interpret since the changes in BP frombaseline could be the result of many influences unrelat-

by guest on May 21, 2018

http://hyper.ahajournals.org/D

ownloaded from

DIETARY FAT AND BLOOD PRESSURE/5acfcy et al. 459

ed to the specific treatment. None of these studiesutilized double-blind conditions.

In two studies, a section of the protocol allowed acomparison of two controlled diets that differed indietary fat composition (see References 10 and 13), butthe negative findings were disregarded in the conclu-sions. For example, completely controlled diets of 35and 25% fat distributed among the saturated and un-saturated classes of fatty acids, sequentially adminis-tered, produced no difference in BP (see Study 1 inReference 10). A randomized study of two diets con-taining 25% fat, one high in linoleic acid and the otherhigh in saturated fats, showed no change in BP.13 Butwhen compared with baseline, all of these test dietslowered BP.

Only one study that used a randomized controlgroup suggested that changes in dietary fat alteredBP.12 Subjects were randomized to either a low fat dietor a low salt diet with high fat intake or asked tocontinue.their usual diet that was high in fat and salt.Each diet was eaten for 6 weeks and followed by areturn to their usual diets. Compared with pretreatmentand posttreatment levels, BP decreased significantlyonly in the low fat group. Unfortunately, the analysisdid not make use of the randomized study design to testwhether the decrease in BP of 8/6 mm Hg on the lowfat diet was significantly greater than the decrease of2/2 mm Hg that occurred in the no intervention controlgroup. Moreover, since the low fat diet was a complexdietary intervention that caused changes in intake ofother nutrients, including vitamins and minerals, it isdifficult to attribute the decrease in BP exclusively tothe lowered intake of saturated fat.

Several studies that changed intake of fat using acomplex dietary program of intervention did not findeffects on BP. The multicenter National Diet HeartStudy randomized about 1000 participants to one oftwo experimental diets that lowered saturated fat andincreased linoleic acid or to a high saturated fat controldiet.27 The BP responses after 1 year of the experimen-tal diets were within 1 % of those of the control group.At one of the centers that enrolled 211 patients, afourth randomized group received a very high (22%)linoleic acid diet that also had no effect on BP. TheBritish Medical Research Council randomized con-trolled trial of soybean oil in 377 patients with myocar-dial infarction found no tendency toward lower BP inthe intervention group as compared with the controlgroup over 5 years of observation.40 In a series ofsmall, controlled dietary studies on cholesterol metab-olism that used a variety of combinations of dietarysaturated fat, polyunsaturated fat, and carbohydrate,Brussaard et al.M found no significant changes in BP inany of the studies in spite of substantial changes inplasma lipids.

In addition to the present study, three double-blindstudies have compared the effects on BP of ingestionof different fatty acids. These studies involved simpleinterventions that minimized changes in other nutri-ents. Neither substitution of saturated fats for linoleicacid (present study)41 nor substitution of oleic acid

(monounsaturated fat) for linoleic acid26-42 producedsignificant differences in BP of normotensive sub-jects2641 or of hypertensive subjects (present study).42

To summarize the results of trials of dietary fats andBP, studies that used a randomized control group,blind or open, generally did not find significant effectson BP.7'8-14> M' "•4<M2 Uncontrolled observational stud-ies that used baseline and posttreatment follow-up val-ues for comparison often found small decreases in BPwhen the diet was enriched with linoleic acid.*""-13-39

Such study designs may suffice for certain biologicalvariables, but BP is notoriously and variably subject toplacebo effects, observer influences, diurnal variation,and seasonal drift. Substantial effort needs to be takenin studies of BP to eliminate these extraneous effects.We believe that the rigor required of drug treatmenttrials should, as much as possible, be applied to nutri-tional studies.

It is likely that when the overall diet is changed tomodify the content of fat, ingestion of one or morenutrients other than fats sometimes may increase tolower BP. For example, studies of two types of vege-tarian diets obtained divergent results on changes inBP.6 '7 Although the dietary fat content was altered inboth diets, the hypotensive result was obtained withthe diet that concomitantly increased dietary fiber, po-tassium, calcium, magnesium, and several vitamins.7

These results, together with the consistent findings oflow BP in vegetarians,1"' suggest that dietary fats perse are not directly involved in the regulation of BP butthat changes in their intake may be a marker for an-other nutrient or set of nutrients that are hypotensive.Since modified fat diets have in common the ability tolower blood cholesterol and are widely recommend-ed,31"2 the identification of associated nutrients thatmay also reduce BP should be a high priority for futureresearch.

AcknowledgmentWe thank Steven Cioli for supervising and assisting in prepara-

tion of the dietary supplements.

References1. Sacks FM, Rosner B, Kass EH. Blood pressure in vegetarians.

Am J Epidemiol 1974;10O:39O-3982. Rouse IL, Armstrong BK, Beilin LJ. The relationship of blood

pressure to diet and lifestyle in two religious populations. JHypertens 1983;1:65-71

3. Ophir O, Peer G, Gilad J, Blum M, Aviram A. Low bloodpressure in vegetarians: the possible role of potassium. Am JClin Nutr 1983;37:755-762

4. Armstrong B, Van Merwyk AJ, Coates H. Blood pressure inSeventh-Day Adventist vegetarians. Am J Epidemiol 1977;105:444-449

5. Sacks FM, Donner A, Castelli WP, Gronemeyer J, et al. Effectof ingestion of meat on plasma cholesterol of vegetarians.JAMA 1981;246:640-644

6. Sacks FM, Marais GE, Handysides G, et al. Lack of an effectof dietary saturated fat and cholesterol on blood pressure innormotensives. Hypertension 1984;6:193-198

7. Rouse IL, Beilin LJ, Armstrong BK, Vandongen R. Bloodpressure lowering effect of a vegetarian diet: controlled trial innormotensive subjects. Lancet 1983;1:5—10

8. Margetts BM, Beilin LJ, Vandongen R, Armstrong BK. Vege-

by guest on May 21, 2018

http://hyper.ahajournals.org/D

ownloaded from

460 HYPERTENSION VOL 10, No 4, OCTOBER 1987

tarian diet in mild hypertension: a randomised controlled trial.BrMedJ 1986;293:1468-1471

9. Comberg HU, Heyden S, Hames CG, Vergroesen AJ, Fleisch-man AI. Hypotensive effect of dietary prostaglandin precursorin hypertensive man. Prostaglandins 1978; 15:193—197

10. Iacono JM, Judd JT, Marshall MW, et al. The role of dietaryessential fatty acids and prostaglandins in reducing blood pres-sure. Prog Lipid Res 1981;20:349-364

11. Iacono JM.PuskaP, Dougherty RM.etal. Effect of dietary faton blood pressure in a rural Finnish population. Am J Clin Nutr1983;38:86O-869

12. PuskaP, Iacono JM.Nissinen A, etal. Controlled, randomisedtrial of the effect of dietary fat on blood pressure. Lancet1983;l:l-5

13. Puska P, Iacono JM, Nissinen A, et al. Dietary fat and bloodpressure: an intervention study on the effects of a low-fat dietwith two levels of polyunsaturated fat. Prev Med 1985;14:573-584

14. Brussaard JH, van Raaij JM, Stasse-Wolthuis M, Katan MB,Hautvast JG. Blood pressure and diet in normotensive volun-teers: absence of an effect of dietary fiber, protein, or fat. Am JClin Nutr 1981;34:2023-2029

15. Smith MC, Dunn MI. The role of prostaglandins in humanhypertension. Am J Kidney Dis 1985;5:A32-A39

16. The Joint National Committee on Detection, Evaluation, andTreatment of High Blood Pressure. The 1984 report of the JointNational Committee on Detection, Evaluation, and Treatmentof High Blood Pressure. Arch Intern Med 1984;144:1045-1057

17. Willett WC, Sampson L, Stampfer Ml, et al. Reproducibilityand validity of a semiquantitative food frequency question-naire. Am J Epidemiol 1985;122:51-65

18. Willett WC, Stampfer MJ, Underwood BA, Speizer FE,Rosner B, Hennekens CH. Validation of a dietary question-naire with plasma carotenoid and alpha-tocopherol levels. AmJ Clin Nutr 1983;38:631-639

19. Sacks FM, Handysides G, Marais GE, Rosner B, Kass EH.Effects of a low-fat diet on plasma lipoprotein levels. ArchIntern Med 1986;146:1573-1577

20. Hypertension Detection and Follow-up Program CooperativeGroup. Variability of blood pressure and the results of screen-ing in the Hypertension Detection and Follow-up Program. JChronic Dis 1978;31:651-667

21. Lepage G, Roy CC. Improved recovery of fatty acid throughdirect transesterification without prior extraction or purifica-tion. J Lipid Res 1984;25:1391-1396

22. Brown BW. The crossover experiment for clinical trials. Bio-metrics 1980;36:69-79

23. Armitage P, Hills M. The two-period crossover trial. The Stat-istician 1982;31:119-131

24. Glantz SA. Primer of biostatistics. New York: McGraw-Hill,1981:229-260

25. Freund RJ, Littell RC. SAS for linear models: a guide to theANOVA and GLM procedures. Cary, NC: SAS Institute,1981. 231 p

26. Sacks FM, Stampfer MJ, Munoz A, McManus K, Canessa M,Kass EH. Effect of linoleic and oleic acids on blood pressure,blood viscosity and erythrocyte cation exchange. J Am CollNutr 1987;6:179-185

27. National Diet Heart Study Research Group. The National DietHeart Study Final Report. Circulation 1968;37,38(suppl I):I-141-I-169; I-228-I-230

28. Dayton S, Pearce ML, Hashimoto S, Dixon WJ, Tomiyasu U.A controlled clinical trial of a diet high in unsaturated fat inpreventing complications of atherosclerosis. Circulation 1969;40(suppl II):II-l-II-63

29. Shepherd J, Stewart JM, Clark JG, Carr K. Sequential changesin plasma lipoproteins and body fat composition during poly-unsaturated fat feeding in man. Br J Nutr 1980;44:265-271

30. Recommendations for the treatment of hyperlipidemia inadults: a joint statement of the Nutrition Committee and theCouncil on Arteriosclerosis of the American Heart Associ-ation. Arteriosclerosis 1984;4:443A-468A

31. World Health Organization Expert Committee. Prevention ofcoronary heart disease. WHO Tech Rep Ser 1982;678:1-53

32. Council on Scientific Affairs. Dietary and pharmacologic ther-apy for the lipid risk factors. JAMA 1983;250:1873-1879

33. limura O, Kijima T, Kikuchi K, et al. Studies on the hypoten-sive effect of high potassium intake in patients with essentialhypertension. Clin Sci 1981 ;61:77s-80s

34. MacGregor GA, Smith SJ, Markandu ND, Banks RA, Sag-nella GA. Moderate potassium supplementation in essentialhypertension. Lancet 1982;2:567-570

35. McCarron DA, Morris CD. Blood pressure response to oralcalcium in persons with mild to moderate hypertension. AnnIntern Med 1985;103:825-831

36. Cappuccio FP, Markandu ND, Beynon GW, Shore AC, Samp-son B, MacGregor GA. Lack of an effect of oral magnesium onhigh blood pressure: a double blind study. BrMedJ 1985;291:235-238

37. Hatch FT, Wertheim AR, Eurman GH, Watkin DM, FroebHF, Epstein HA. Effects of diet in essential hypertension: III.Alterations in sodium chloride, protein and fat intake. Am JMed 1954;17:499-513

38. Sacks FM, Wood PG, Kass EH. Stability of blood pressure invegetarians receiving dietary protein supplements. Hyperten-sion 1984;6:199-201

39. Fleischman AI, Bierenbaum ML, Stier A, Somol SH, WatsonP, Naso AM. Hypotensive effect of increased dietary linoleicacid in mildly hypertensive humans. J Med Soc NJ 1979;76:181-183

40. Controlled trial of soya-bean oil in myocardial infarction: re-port of a research committee to the Medical Research Council.Lancet 1968;2:693-700

41. Margetts BM, Beilin LJ, Armstrong BK, et al. Blood pressureand dietary polyunsaturated and saturated fats: a controlledtrial. Clin Sci 1985;69:165-175

42. Rao HR, Rao UB, Srikantia SG. Effect of polyunsaturate-richvegetable oil on blood pressure in essential hypertension. ClinExpHypertens 1981;3:27-38

by guest on May 21, 2018

http://hyper.ahajournals.org/D

ownloaded from

F M Sacks, I L Rouse, M J Stampfer, L M Bishop, C F Lenherr and R J Waltherpatients.

Effect of dietary fats and carbohydrate on blood pressure of mildly hypertensive

Print ISSN: 0194-911X. Online ISSN: 1524-4563 Copyright © 1987 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Hypertension doi: 10.1161/01.HYP.10.4.452

1987;10:452-460Hypertension. 

http://hyper.ahajournals.org/content/10/4/452the World Wide Web at:

The online version of this article, along with updated information and services, is located on

  http://hyper.ahajournals.org//subscriptions/

is online at: Hypertension Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document.

Permissions and Rights Question and Answer Further information about this process is available in therequested is located, click Request Permissions in the middle column of the Web page under Services.not the Editorial Office. Once the online version of the published article for which permission is being

can be obtained via RightsLink, a service of the Copyright Clearance Center,Hypertensionpublished in Requests for permissions to reproduce figures, tables, or portions of articles originallyPermissions:

by guest on May 21, 2018

http://hyper.ahajournals.org/D

ownloaded from