PREVENTIVE MEDICINE 19, 31-39 (1990)

Dietary Protein and Blood Pressure in Monozygotic Twins

RICHARD J. HAVLIK, M.D.,*+ RICHARD R. FABSITZ, M.A.,*? SONA KALousDrAN,*‘t M.D., NEMAT 0. BORHANI, M.D.,5 AND

JOE C. CHRISTIAN, M.D.#

*Clinical and Genetic Epidemiology Branch, fEpidemiology and Biometry Program, #Division of Epidemiology and Clinical Applications, National Heart, Lung, and Blood Institute, Bethesda,

Maryland 20892; /Department of Community Health, University of California, Davis, California 95616; and #Department of Medical Genetics, Indiana University School of Medicine

Cross-sectional studies relating blood pressure to dietary intake have shown equivocal results, in part due to the inability to take into account the strong genetic component of blood pressure. Intervention studies, using the same subject as his own control, often encounter additional problems when subjects are asked to adhere to an alternate diet. The National Heart, Lung, and Blood Institute Twin Study of middle-aged men provided infor- mation concerning the possible relationship of food-frequency-estimated nutrient intake to blood pressure while controlling for genetic effects in a free-living group of subjects. Using differences in monozygotic twins, a direct association of dietary protein intake and diastolic blood pressure was identified and persisted after adjustment for known covariates of blood pressure. Adjusting for known covariates and holding total calories constant, a 9-g differ- ence in daily protein intake was directly associated with a 1 mm Hg difference in diastolic blood pressure. For protein intake as a percentage of total calories, a 2.18% difference was directly associated with a 1 mm Hg difference in diastolic blood pressure. The co- twin-control method provides a powerful design to address the interrelationships between nutrients and blood pressure in an observational as well as an experimental setting. 619!30 Academic Press, Inc.

INTRODUCTION The effect of dietary protein intake on blood pressure (BP) is uncertain. There

are observation, intervention, and laboratory data that suggest no effect or a minimal effect at most of dietary protein on BP; however, considerations, such as the possible counterbalancing effects of specific amino acids or more complex interrelationships in certain hypertension subgroups, justify further study. The initial suggestions that animal protein might affect BP levels came from observa- tions of lower BP among vegetarians who were most diligent in abstaining from meat (1). However, various feeding experiments with vegetarian diets have not established that the meat protein component of the diet has an effect (24). Cross- sectional observational studies in Framingham, Hawaii, and Puerto Rico have given contradictory data on any possible influence of proteins (5-7). A particular problem is the strong interrelationship of protein with other dietary components. For example, an inverse relationship of dietary protein and BP is likely con- founded by the high correlation of dietary potassium with protein (7).

Besides the direct effect on BP there are other considerations involving end

i To whom reprint requests should be addressed, at National Heart, Lung, and Blood Institute, Federal Building, Room 300, Bethesda, MD 20892.

31 0091-7435/90 $3.00 Copyright 0 1990 by Academic Ress, Inc. All lights of reproduction in any form reserved.

32 HAVLIK ET AL.

organ effects. Specifically, a high protein diet has the effect of raising glomerular filtration rate secondary to increased renal blood flow. This could result in pos- sible structural damage with glomerular sclerosis (8). An observation in stroke- prone hypertensive rats suggested that the higher protein content of feed used in the United States, as compared with Japanese laboratories, might protect the vascular integrity of cerebral vessels in animals (9). Finally, the amino acids tyrosine and tryptophan have been shown to have different effects on BP depend- ing on the animal model used and the physiologic status of the animals (10, 11). Whether any of these observations will be relevant for human BP+lietary protein relationships awaits further research.

Identical twins provide a unique opportunity to isolate and study possible en- vironmental effects. Because the co-twins are identical genetically, any differ- ences in a variable of interest, such as BP, must be related to differences in environmental factors (12). Although the approach has been described (13-15), the methodology has not been used extensively. The availability of dietary data from a group of middle-aged twins allowed the current investigation of the rela- tionship of dietary intake to BP.

METHODS

Population

The National Heart, Lung, and Blood Institute Twin Study recruited 514 pairs of white male twins, ages 42-56 years and living in five geographic areas, who were drawn from the National Academy of Sciences-National Research Council Veterans Twin Registry. Two hundred and fifty pairs of monozygotic twins were examined in five centers across the country during the period 1969-1973 (16). Excluded from the present analyses were 36 subjects missing the nutrition data (the San Francisco center did not collect diet data using the food frequency questionnaire), 32 subjects who were under treatment for hypertension, and 30 subjects whose co-twin was excluded. Thus, 402 of the monozygotic twins had complete data on cardiovascular risk factors and diet for both members of the twin pair.

Blood pressure was measured first by a nurse and later by a physician, at the beginning and at the end of the physical examination, using a standard sphygmo- manometer. Different physicians examined the co-twins. The participants were seated and both systolic and diastolic (Phase V) BP from the left arm were re- corded. The physician’s first BP reading was used for this analysis. Body size was recorded as weight using a calibrated scale. (Because of similar heights in identical twins, body mass index provided no additional information.) Triglycerides and other blood lipids were measured on fasting plasma and determined in standard- ized laboratories using the procedures of the Lipid Research Clinics and Kessler and Lederer (17, 18). Clinical chemistries were done by local laboratories.

Dietary Znstruments

Nutritional data were collected using a food frequency questionnaire patterned after those in use nationally and internationally at the time of the examination in

DIETARY PROTEIN AND BLOOD PRESSURE 33

1969 (19). The questionnaire was designed to gather information on the quantita- tive frequency with which food items characteristic of the usual U.S. diet were ingested. The frequency consumed per day or over a period of 1 week was re- corded. The range of seven possible responses for frequency was from “seldom or never, ” “less than once per week,” to “more than 4 times per day.” The phrasing of the question was, “How often do you eat or drink: milk, cheese, eggs, salads, fruit, etc?” Forty-four food items were used. Additional information on the number of eggs per meal and use of fats, oils, mayonnaise, and other condi- ments was obtained by trained nutritionists employing a standardized interview technique. While the same nutritionist generally interviewed both members of a twin pair, diet data were collected independently with no reference to a twin about his co-twin. A computer program was written to convert qualitative food fre- quency information to quantitative nutrient values (6). A standard serving size was assumed for the age group under study. A table of food composition was prepared, which conformed to the frequency questionnaire. The values were con- verted into an average daily nutrient intake.

Statistics

Means, medians, and standard deviations for the nutrients in grams, in percent- age calories, and adjusted for total caloric intake were calculated. Adjustment for total caloric intake followed the procedure of Willett and Stampfer (20) in which each nutrient is regressed on total calories. Residuals from the equations are added to the sample mean and used as measures of the nutrients that are inde- pendent of the total caloric intake. Because opinion varies on the advantage of this method of adjustment, analyses were repeated with each of the three nutrient measures. Results for the absolute level and adjusted level were so similar that only the adjusted and percentage nutrient results are described from the tables. Spearman correlation coefficients were calculated to describe the relationship of BP to the nutrients and selected covariates for individuals and for within-pair differences. Stepwise multiple regression analysis was used to estimate difference in BP attributed to differences in nutrient intake while adjusting for known cova- riates of BP. Possible covariates included weight, log triglycerides (TG), log glu- cose (1 hr post 50-g load), uric acid, phosphorus, serum protein, blood urea nitrogen (BUN), cholesterol, hematocrit, number of cigarettes per day, log drinks of alcohol per week, and heart rate. These covariates were selected because they have been shown to be related to BP in various populations (21). Only those covariates that were significantly related to BP in these data are included in the tables.

RESULTS

The mean and median daily nutrient intake for individual twins as estimated by the food frequency questionnaire are presented in Table 1 for the 402 individuals from 201 sets of monozygotic twins. Means are also presented for systolic and diastolic BP and for covariates that were significantly related to BP in this sample of twins. Covariates include weight, log TG, and total cholesterol (TC).

Table 2 shows the Spear-man rank correlation coefficients for various nutrient

34 HAVLIK ET AL.

TABLE 1 MEDIAN AND MEAN BLOOD PRESSURES, COVAIUATES, AND NUTRIENT INTAKE ESTIMATED BY

FOOD FREQUENCY QUESTIONNAIRE FOR INDIVIDUALS FROM MONOZYGOTIC TWIN PAIRS

Nutrient Median Mean Standard deviation

Total calories (kcal) Total protein (g) Total carbohydrates (g) Saturated fat (g) Polyunsaturated fat (g) % Protein (% cal) % Carbohydrates (% cal) % Saturated fat (% cal) % Polyunsaturated fat (% Cal) ADJ protein (g) ADJ carbohydrates (g) ADJ saturated fat (g) ADJ polyunsaturated fat (g) Systolic blood pressure (mm Hg) Diastolic blood pressure (mm Hg) Weight (kg) Log triglycerides (log meqfliter) Total cholesterol (mg/dl) Number of individuals

1954 71.7

225.6 34.2 7.8

14.7 45.5 15.6 3.4

74.0 229.0 35.3 8.0

125.0 80.0 77.3 5.94

220.0 402

2027 610 74.7 22.1

228.6 75.2 35.5 13.6 0.0 3.6

15.0 2.7 45.2 7.0 15.6 3.1 3.6 1.3

74.7 12.0 228.6 35.0 35.5 6.7 8.0 2.7

127.6 17.1 81.7 10.8 77.7 10.9 5.96 .53

219.0 35.1

TABLE 2 SPEARMAN CORRELATION COEFFICIENTS OF SYSTOLIC AND DIASTOLIC BLEND PRESSURE WITH

COVARIATES FOR INDIVIDUALS, AND PAIRED DIFFERENCES FROM MONOZYG~TIC TWINS

Monozygotic Within-pair individuals differences

SBP DBP SBP DBP

Calories Protein Total CHO Saturated fat Polyunsaturated fat % Protein % Carbohydrates % Saturated fat % Polyunsaturated ADJ protein ADJ carbohydrates ADJ saturated fat ADJ polyunsaturated fat Weight Log triglycerides Total cholesterol Number

-0.10 - 0.08 -0.10 -0.10 -0.01

0.05 -0.02 -0.03

0.06 0.01

-0.02 -0.02

0.08 0.17* 0.11 0.07

-0.14* -0.06 -0.16* -0.12 -0.03

0.11 -0.07

0.02 0.08 0.08

- 0.08 0.02 0.09 0.26* 0.20* 0.12

402

0.08 0.08 0.12 0.00 0.06 0.01 0.10

-0.13 -0.03

0.09 0.10 0.04 0.08 0.30* 0.18* 0.11

0.09 0.17* 0.02 0.08 0.06 0.14*

-0.06 0.03

-0.01 0.14* 0.05 0.09 0.08 0.35* 0.20* 0.18*

201

* Significance P < 0.05.

DIETARY PROTEIN AND BLOOD PRESSURE 35

intakes with systolic and diastolic BP. The first two columns represent correla- tions for twins as individuals. Weight, log TG, and TC are also shown because they are the major correlates for BP in this sample. The values for weight, log TG, and TC are slightly higher if differences between monozygotic twins are corre- lated with differences in BP (columns 3 and 4). Among the differences in nutrition variables, only the protein measurements (absolute, adjusted, and percentage) were significantly related to diastolic BP. Diastolic BP was unrelated to protein when the twins were analyzed as individuals. Correlation coefficients for systolic BP with the nutrients were not significantly different from 0 for either individuals or monozygotic paired differences.

Table 3 presents the results of a multiple regression analysis to estimate the magnitude of the relationship of diastolic BP and protein intake after adjustment for known covariates. Differences in diastolic BP were regressed on differences in nutrients and percentage nutrients, separately, after differences in weight, log TG, and/or TC entered the model. In all analyses 21% of the variance in diastolic BP was explained. About a 9-g difference, or a little more than a 2% difference, in protein intake between monozygotic co-twins was associated with a 1 mm Hg difference in diastolic BP. Differences in systolic BP were not related to protein intake or any other measure of nutrient intake after weight and log TG were included in the model (R2 = 0.12).

DISCUSSION

A direct relationship between dietary protein and diastolic BP is indicated by this analysis and suggests the hypothesis that a reduction in the proportion of calories from protein may reduce diastolic BP. This statement is based on the

TABLE 3 REGRESSION RESULTS FOR DIASTOLIC BLOOD PRESSURE ON PROTEIN, ADJ PROTEIN, AND

PERCENTAGE PROTEIN WITH SIGNIFICANT COVARIATES USING DIFFERENCES IN MONOZYGOTIC TWINS

Independent variables Coefficient P value

Protein 0.110 0.020 Total calories - 0.003 0.114 Weight 0.336 0.001 Log triglycerides 3.355 0.020 Total cholesterol 0.049 0.036

ADJ protein 0.110 Total calories -0.006 Weight 0.336 Log triglycerides 3.355 Total cholesterol 0.049

Percentage protein 45.906 Weight 0.353 Log triglycerides 3.523 Total calories 0.046

0.020 0.043 0.001 0.020 0.036 R2 = 0.21

0.030 0.001 0.014 0.047 R2 = 0.21

R2 = 0.21

36 HAVLIK ET AL.

strength of the co-twin methodology which provides more appropriate informa- tion than studies of individuals because of the inherent degree of control of en- dogenous and extraneous factors. First, this method eliminates the influence of genetic factors, because the twins are genetically identical. (The only qualification is the possibility that there could be a genetic-environmental interaction operating such that different protein-BP relationships would be observed in those pairs with particular genes and not in others.) Such control of confounders is important since there is a definite genetic influence on the variation of BP differences among people (12), as well as evidence of a possible genetic influence on nutrient intake, and specifically on protein concentration (22, 23). Second, the method minimizes some of the unmeasured environment, since co-twins are acknowledged to share common environment. Thus, the method should have a greater potential to en- hance or identify the influence of more subtle environmental factors.

The co-twin method is somewhat analagous to individuals serving as their own controls to eliminate the effects of genetic variation, e.g., crossover design. Such studies are limited by the need for the subject to adhere to an alternate diet. One study of non-twin individuals showed results similar to these with a 1% drop in systolic BP and a significant 4.5% drop in diastolic BP when a group of omnivo- rous volunteers were subjected to a 6-week vegetarian diet (3). However, a short- term modification study in a group of young students did not show any effect of modifying animal or vegetable protein (2). Also, other studies have not suggested a major BP effect with major increases in protein intake (4). Recently, Slattery et al. (24) did a similar analysis on a group of 71 pairs of male and female monozy- gotic twins ranging in age from 22 to 66 years. Regression analyses of differences in diastolic BP on differences in dietary components, without covariates, indi- cated a significant coefficient for only alcohol intake. Results for protein intake provided a similar regression coefficient of 0.11 for protein intake but it was not significant in that study.

The fact that only protein intake showed a consistent relationship in this study may be important new information or could be the result of a chance occurrence due to the multiple factors considered. It must be emphasized that high intercor- relations among the major nutrients were found in this study, as they have been in others (7). In particular, protein was highly correlated with saturated fats and carbohydrates as well as with total calories. Consequently, the problem of mul- ticollinearity could account for some of the relationship between protein and BP. Correlations between the nutrients were at least 0.6, much higher than the cor- relations between protein and BP.

A factor that may have worked to lessen, if not confound, the relationship was the methodology used to ascertain dietary intake which did not obtain specific information on the size of food portions eaten. Some have stated that the results by most food frequency methodologies are misleading, especially for cholesterol (25). One study validated a semiquantitive food frequency for nutrients including protein but the emphasis of that study was on fats and vitamins in the diet (15). Another group claims success with a 20-item list for protein (26). iJnfortunately, although the food frequency used in the current study gave reasonable results for

DIETARY PROTEIN AND BLOOD PRESSURE 37

calories and protein as estimated by NHANES I (27), it has not been validated against some other objective measure.

Analyses were also done by individual items from the food frequency question- naire to minimize the effect of estimation methods. For differences in the monozy- gotic twin pairs no correlations exceeded 0. Il. The six items with the highest correlations were pork, cheese, ice cream, wine, alcohol, and beef. In a multivariate analysis, none of these was significant alter inclusion of weight, TG, and TC.

The mechanisms for the relationship of protein intake and BP are likely to be complex. For example, if increased protein intake was related to BP through an effect on the kidneys, impaired renal function might be expected to enhance the relationship (28). An analysis recoding twins on BUN being above or below the 75th or 90th percentile within each laboratory did not reveal an enhanced rela- tionship in the higher BUN group. Another possibility is that the protein relation- ship to diastolic BP is mediated through calcium. It has been shown that increased protein intake causes the kidney to deplete calcium from the blood (29). Several studies have shown that calcium is indirectly related to the level of BP (13, 23). Thus, the direct relationship of protein intake has a possible biological basis.

The need to confirm these results, preferably in another group of monozygotic twins of the same age and sex but using an optimal dietary intake instrument, is evident. Other twin populations or special populations, such as those with renal insufficiency, might also provide useful information. In addition to further obser- vational studies, this twin model also provides advantages for possible interven- tion studies. It is possible to administer a particular dietary regimen, such as a low protein diet, to one co-twin and use the other co-twin as a control (13). This strategy would be a very powerful approach to resolving some of the dietary questions, especially since BP is rather responsive to acute changes. For variables such as BP, which have a significant genetic component, twin studies offer a variety of study designs to focus on environmental effects while controlling for known genetic effects.

REFERENCES 1. Sacks FM, Rosner B, Kass EH. Blood pressure in vegetarians. Am .I Epidemiol 1974; 100:390-

398. 2. Brussaard JH, van Raay JMA, Stasse-Wolthuir M, Katan MB, Hauvart JGAJ. Blood pressure and

diet in normotensive volunteers: Absence of an effect of dietary fiber, protein, or fat. Amer J Clin Nutr 1971; 34~2023-2029.

3. Burstyn P. Effect of meat on blood pressure: Letter to the editor. JAMA 1982; 248(1):29-30. 4. Sacks FM, Wood PG, Kass EH. Stability of blood pressure in vegetarians receiving dietary

protein supplements. Hypertension 1984; 6:199-201. 5. Garcia-Palmieri MR, Costas R Jr, Cruz-Vidal M, Sorlie PD, Tillotson J, Havlik RJ. Milk con-

sumption, calcium intake and decreased hypertension in Puerto Rico. Hypertension 1984; 6~322-328.

6. Kannel WB, Dawber TR. Hypertension cardiovascular disease. In: Onesti G, Kim KE, Mayer JH, Eds. The Framingham Study in Hypertension: Mechanisms and Management. New York: Grune and Stratton, Inc. 1973: 93.

7. Reed D, McGee D, Yano K, Hankin J. Diet, blood pressure, and multicollinearity. Hypertension 1985: 7:405-410.

38 HAVLIK ET AL.

8. Brenner BM, Meyer TW, Hostetter TH. Dietary protein intake and the progressive nature of kidney disease: The role of hemodynamically mediated glomerular injury as the pathogenesis of progressive glomerular sclerosis in aging, renal ablation, and intrinsic renal disease. N Engl J Med 1982; 307:652&&I

9. Yamori Y, Horie R, Tanase H, Fujiwara K, Nara Y, Lovenberg W. Possible role of nutritional factors in the incidence of cerebral lesions in stroke-prone spontaneously hypertensive rats. Hypertension 1984; 6349-53.

10. Sved AF, van Itallie CM, Femstrom JD. Studies on the antihypertensive action of L-tryptophan. J Pharmacol Exp Ther 1982; 221:329-333.

11. Sved AF, Femstrom JP, Wurtman RJ. Tyrosine administration reduces blood pressure and en- hances brain norepinephrine release in spontaneously hypertensive rats. Proc Nat1 Acad Sci USA 1979; 76:3511-3514.

12. Feinleib M, Garrison RJ. The contributions of family studies to the partitioning of population variation of blood pressure. In: Sing CF, Scolnick M, Eds. Genetic analysis of common disease: Applications to Predictive Factors of Coronary Disease. Proceedings Workshop, Snowbird Utah. New York: A. R. Liss, 1979: 653.

13. Miller JZ, Nance WE, Norton JA, Wolen RL, Griffith RS, Rose RJ. Therapeutic effect of vitamin C: A control study. JAMA 1977; 237:248-251.

14. Nance WE. The relevance of twin studies to cardiovascular research. In: Rao DC, Elston RC, Kuller LH, Feinleib M, Carter C, Havlik R, Eds. Genetic Epidemiology of Coronary Heart Disease: Past, Present and Future. New York: A. R. Liss, 1984: 325.

15. Willett WC, Sampson L, Stampfer MJ, Rosner B, Bain C, Witschi J, Hennekens CH, Speizer FE. Reproducibility and validity of a semiquantitative food frequency questionnaire. Am J Epide- miol 1985; 122:51-65.

16. Feinleib M, Garrison RJ, Fabsitz R, Christian JC, Hrubec Z, Borhani NO, Kannel WB, Rosenman R, Schwartz JT, Wagner JO. The NHLBI Twin Study of cardiovascular disease risk factors: Methodology and summary of results. Am J Epidemiol 1977; 106:28&295.

17. Kessler G, Lederer H. Fluorometric measurement of triglyceride. In: Skaggs LI Jr et al., Eds. Automation in Analytical Chemistry: Technicon Symposium, 1%5. New York: Medaid, 1966.

18. Manual of Laboratory Operations, Lipid Research Clinics Program. Washington DC, GPO 1974. DHEW Publication No. (NIH) 75-628.

19. Hjortland MC. The effect of heredity and environment on nutrient intake of adult monozygotic and diazygotic twins. Thesis, University of Minnesota, 1972.

20. Willett WC, Stampfer MJ. Total energy intake: implications for epidemiologic analyses. Am J Epidemiol 1986; 124:17-27.

21. Havlik RJ, Garrison RJ, Feinleib M, Padgett S, Castelli WP, McNamara PM. Evidence for addi- tional blood pressure correlates in adults 20-56 years old. Circulation 1980; 61:710-715.

22. Fabsitz R, Garrison R, Feinleib M, Hjortland M. A win analysis of dietary intake: Evidence for a need to control for possible environmental differences in MZ and DZ twins. Behav Genet 1978; 8: 15-25.

23. Wade J, Mimer J, Krondl M. Evidence for a physiological regulation of food selection and nutrient intake in twins. Am J Clin Natr 1981; 34:143-147.

24. Slattery ML, Bishop DT, French TK, Hunt SC, Meikle AW, Williams RR. Lifestyle and blood pressure levels in male twins in Utah. Genet Epidemiol 1988; 5:277-287.

25. Lee J, Kolonel LN, Hankin JH. Cholesterol intake as measured by unquantified and quantified food frequency interviews: Implications for epidemiological research. Znt J Epidemiol 1985; 141249253.

26. Byers T, Marshall J, Fiedler R, Zielezny M, Graham S. Assessing nutrient intake with an abbre- viated dietary interview. Am J Epidemiol 1985; 12241-50.

27. National Center for Health Statistics: Abraham S, Johnson CL, Cat-roll MD. Dietary Intake Findings, United States, 1971-1974. Vital and Health Statistics, Series 11, No. 202. DHEW Publication No. (HRA) 77-1647. Public Health Service, U.S. Government Printing Offtce, 1977.

DIETARY PROTEIN AND BLOOD PRESSURE 39

28. Lyle RM, Melby CL, Hyner GC, Edmondson JW, Miller JZ, Weingerger MH. Blood pressure and metabolic effects of calcium supplementation in normotensive white and black men. JAMA 1987; 257: 17721776.

29. Lutz J, Linkswiler HM. Calcium metabolism in postmenopausal and osteoporotic women con- suming two levels of dietary protein. Am J Clin Nutr 1981; 34:2178-2186.

30. Sempos C, Cooper R, Kovar MG, Johnson C, Drizd T, Yetley E. Dietary calcium and blood pressure in National Health and Nutrition Examination Surveys I and II. Hypertension 1986; 8:1067-1074.