Healthy Decisions for the Love of Health

Low-Fat/High-Carb Diets vs. Low-Carb/High-Protein Diets: CHD Risk


Cumulative data from peer-reviewed studies in
the last decade casts doubt on conventional wisdom about
desirable macronutrient ratios in the diet.

Dietary Macronutrient Ratios and
their Effect on Biochemical Indicators
of Risk for Heart Disease

Comparing High-Protein/Low-Carbohydrate Diets
vs. High-Carbohydrate/Low-Fat Diets


by Loren Cordain, Ph.D.
Copyright 1999 by Loren Cordain. All rights reserved.


IMPORTANT WORD DEFINITIONS: CHD: Coronary heart disease, i.e., a form of cardiovascular disease or atherosclerosis. HUFA: Highly unsaturated fatty acids. In vitro: Biochemical effects obtained in an artificial environment outside the living organism. In vivo: Biochemical effects as they occur inside the living organism. Isocaloric: Calorie- for- calorie. Used when describing controlled replacement of one dietary macronutrient for another. Lipids: Fatty acids, i.e., dietary "fats." Mesenteric fat: Fat surrounding/ protecting the intestinal area. n3 and n6 fats: Omega-3 and omega-6 fatty acids, respectively. Perinephral fat: Fat surrounding and/or protecting the kidney. Plasma: Refers to levels of nutrients as measured in the blood. Post-prandial: After a meal. Used in describing biochemical responses in the post-meal period. Polymorphism: A variant form of a gene. Often genes have several naturally occurring variants. PUFA: Polyunsaturated fatty acids. Serum: Synonym for blood, as in measured blood levels of nutrients.

Based on and edited from postings made to the Paleodiet listgroup on 4/30/97, 5/5/97, 5/26/97, 8/28/97, 9/2/97, 10/9/97, 11/13/97, 1/26/98, and 5/8/98.

Conventional wisdom about macronutrient ratios challenged in last decade. The conventional wisdom of orthodox nutritionists for the past 20-25 years regarding macronutrient intake has been that a high- carbohydrate, low- fat diet is the optimal diet for humans and benefits virtually all pathological conditions ranging from heart disease to cancer. In the past 10-12 years, however, this concept has been seriously questioned in terms of deleterious changes (elevated triglycerides and VLDL, and lowered HDL) which occur in blood- lipid profiles from such advice. Increasingly, influential scientists (Scott Grundy, Walter Willett, Gerald Reaven) and institutions (Harvard School of Public Health) have recognized this shortcoming of high- carb, low- fat diets, and are now recommending monounsaturated fat in lieu of carbohydrate [Grundy 1986; Mensink et al. 1987].

Despite the thousands-- perhaps tens of thousands-- of clinical trials which have been conducted manipulating the macronutrient (protein, carbohydrate, and fat) content of diet, however, there are perhaps no more than a half- dozen which have examined the influence of a high- protein, low- carbohydrate diet (with varying fat levels) upon human health and metabolism. This is somewhat ironic, in that this macronutrient pattern appears to be the one which nourished humankind (members of the genus Homo) for all of our time (2.5 million years) on this planet, except for the last 10,000 years since the advent of grain- based agriculture.

The specific dietary context in which fat and the other macronutrients occur is key, but often overlooked

There is little doubt that Paleolithic man consumed (probably preferentially) the fatty portions of wild game animals. During certain times of the year (late summer and early fall) the total lipid content of large herbivorous animals was considerable, and at these times saturated fat consumption could have been high. At other times, however, it would have been modest (except perhaps for high- latitude peoples who were the most dependent on animal meat due to a colder climate less hospitable to plant life), even while the overall yearly consumption of all forms of fat could have been relatively high. However, despite the consumption of a largely animal- based diet that at times can include a high overall level of fats, most modern- day hunter- gatherers exhibit low serum cholesterol levels, low blood pressure, and low to non- existent mortality rates from coronary heart disease (CHD) [Eaton 1988; Bang and Dyerberg 1980; Leonard et al. 1994].

At first, this data seems paradoxical in light of the almost universal recommendations of a low- fat, high- carbohydrate diet in the treatment of CHD. However, there are important and subtle differences between the high- fat/ animal- based diets of pre- agricultural man and the high- fat diets of modern man which can account for this paradoxical situation:


  • Differences in trans fats and oxidation of cholesterol. There is increasing recognition that for the atherosclerotic process to occur, there must not only be elevated levels of LDL cholesterol, but that the lipids and cholesterol carried by LDL must be oxidized [Steinberg et al. 1989]. The macrophages which take up oxidized LDL molecules and eventually become the foam cells of the atherosclerotic plaque have a scavenger receptor which is different from native LDL receptors and which does not down- regulate. Consequently, continually elevated levels of oxidized LDL in the plasma tends to promote the atherosclerotic process.

    High levels of dietary linoleate increase LDL oxidizability ([Louheranta et al. 1996]. Because refined vegetable oils were not present in pre- agricultural diets, the linoleate levels would have been lower than in Western diets, wherein the vegetable fat consumption has increased 300% since 1910 and the animal- fat consumption has decreased slightly [ASCN/AIN Task Force 1996]. Thus the relatively high levels of vegetable oils consumed in the Western diet, along with relatively high levels of saturated fats, promote a lipid profile in which LDL cholesterol is elevated and more prone to oxidation and hence to the development of CHD.


  • Differences in protein intake levels. The protein content of the paleolithic diet was significantly higher than the average 12-15% of the Western diet. Recent studies [Wolfe 1995; Wolfe et al. 1991] show that isocaloric (calorie- for- calorie) replacement of carbohydrate with protein lowers total cholesterol, LDL, VLDL, and triglycerides (TG) while elevating HDL cholesterol-- all of which are favorable responses in terms of blood- lipid levels. Consequently, a high dietary protein content even in the face of increasing overall fats, or increasing saturated fat, serves to lower serum cholesterol levels and reduce the risk for CHD.


  • Differences in carbohydrate intake. The carbohydrate content of pre- agricultural diets was generally lower than the 45-55% of the Western diet. Consequently, the post- prandial (after- meal) lipemic excursions (changes in blood lipid levels)-- during which LDL molecules are most prone to oxidation-- would have been reduced, since the addition of carbohydrate to a fat- rich meal exacerbates this swing [Chen et al. 1992]. Pre- agricultural eating patterns show that fat and protein were generally eaten together, whereas carbohydrate meals were eaten separately. This eating pattern would have reduced post- prandial lipemic excursions. Additionally, the reduced carbohydrate content of pre- agricultural diets would have improved the portions of the blood- lipid profile (TG, VLDL, HDL, Lp(a)) which are worsened by high- carbohydrate diets [Reaven 1995].


  • Differences in fatty-acid intake profiles. Work from our laboratory [Cordain et al. 1998], as well as that of others, has shown that the fatty- acid profiles of storage as well as structural fat are quite different when contrasting wild to domesticated animals. Of the dietary saturated fats, 12:0, 14:0, and 16:0 are known to elevate plasma cholesterol levels, whereas 18:0 is neutral or perhaps hypocholesterolemic (cholesterol- lowering). The saturated fat of bone marrow and depot fat in wild animals contains greater levels of 18:0 and lower levels of 14:0 and 16:0 when compared to domestic animals.

    Additionally, the structural (e.g., polyunsaturated) lipid content in game meat is also quite different than that in domestic meat. There generally are higher levels of all n3 ("omega 3") fats and higher levels of all 20 and 22-carbon fats of both the n3 and n6 ("omega 6") varieties in game meat. The n6/n3 ratio of beef averages about 15, whereas in wild animals it is about 4-5. Again, higher levels of n6 lipids in domestic animals, particularly linoleate, tend to increase LDL oxidizability, whereas the higher levels of n3 fats in game animals are cardio- protective.

    Consequently, the consumption of saturated fats in pre- agricultural diets occurred against a background of dietary lipids which was much different than the background fats in the modern diet. Recent evidence clearly shows that the composition of fatty acids in a meal can improve serum lipid values despite widely varying fat levels [Nelson et al. 1995].


  • Absence of dairy fats. Pre-agricultural diets by definition would not have included dairy fats. In modern Western diets, about a third of the saturated fat is contributed by dairy foods (milk, butter, cheese, ice cream). In metabolic ward studies, butter fat raised LDL cholesterol levels significantly higher than beef tallow [Denke and Grundy 1991]. Further, milk consumption is the best worldwide predictor of CHD mortality of all dietary elements [Artaud- Wild et al. 1993]. Bovine milk fat is quite low in the long- chain cardio- protective n3 fats, and has a high n6 ratio. Additionally the calcium to magnesium (Ca/Mg) ratio in milk and dairy products is quite high compared to the average 1:1 ratio in foods available to pre- agricultural man. Elevated Ca/Mg ratios have been shown to be positively related to CHD [Varo 1974]. This data suggests dairy products would be quite atherogenic, particularly when consumed in a background of other dietary elements in the Western diet.
  • Large disparity in activity levels. Modern man eating a high- fat diet is generally quite inactive compared to pre- agricultural man [Cordain, Gotshall, and Eaton 1997]. High levels of activity serve to improve insulin sensitivity and lower TG and VLDL while increasing HDL cholesterol.
In all likelihood, the dietary fat levels of pre- agricultural man could have been quite high (even by modern standards). However, because of differing types and amounts of carbohydrate, protein, and fatty acids, as well as differing levels of fiber and antioxidant vitamins and phytochemicals from a diet rich in plant foods as well as meats, these types of diets generally would not have elevated cholesterol levels (as confirmed by values seen in modern hunter- gatherers [Bang and Dyerberg 1980; Leonard et al. 1994]), nor have increased LDL oxidizability.

One final comment: Not only does the high sodium content of the Western diet predispose us to hypertension, osteoporosis, urinary tract stones, menierre's syndrome, stomach cancer, insomnia, asthma and initiation and promotion of all types of cancer, it also seems to do the same in our closest relative, the chimp [Denton et al. 1995].

Saturated/unsaturated fat composition of wild animal tissues, and consumption levels in modern vs. pre- agricultural peoples

Our data on the fatty- acid distribution in tissues of wild animals presented at a recent conference on the return of n3 fats to the food supply, held at the National Institutes of Health in Bethesda, Maryland has been recently published in World Review of Nutrition and Dietetics. [Cordain et al. 1998]. This data refutes contentions made by some that the overall PUFA in wild- animal tissues is low. To the contrary, it is relatively high in both brain (26%) and muscle (36%) as our data shows, and which corroborates earlier work of Crawford et al. [1969].

The difference in polyunsaturated fatty acids (PUFAs) between the Western diet and the so-called "paleolithic diet" is that the PUFAs in the Western diet are predominantly based upon 18-carbon lipids (vegetable oils) with huge amounts of 18:2n6 (linoleic acid) predominating. The PUFA content of the paleolithic diet is higher than that of the Western diet (19.2% vs. 12.7% [Bang and Dyerberg 1980]) with much higher levels of HUFA (>20-carbon lipids) of both the n6 and n3 families.

Once again, it should be emphasized as well that while pre- agricultural peoples certainly did consume saturated fat, it cannot compare with the levels consumed by modern Western populations. Bang and Dyerberg's data [1980] on Eskimo populations who ate a high- meat diet is particularly illustrative of this. Of the total dietary fats, saturated fats comprised 22.8% in Inuit people whereas saturated fats comprised 52.7% of the total dietary fats in a control population of Danes. To point to saturated fat consumption in pre- agricultural groups as license to eat freely of such fats ignores the ecological constraints that would have made modern levels of consumption highly unlikely for our paleolithic ancestors, and ignores as well the voluminous clinical data that shows their detrimental effects.

High levels of saturated fat consumption on a year- round basis only became possible when domesticated animals were bred and fed in a manner which allowed accumulation of depot fat on a year- round basis. Wild animals almost always show a seasonal variation in storage fat, and even the very fattest wild land mammals contain 60-75% less total fat than the average domesticated animal. Thus, until the advent of the "Agricultural Revolution" 10,000 years ago, it would have been extremely difficult, or perhaps impossible, to eat high levels of saturated fat on a daily basis throughout the year.

Limitation of the Keys equation in predicting expected serum cholesterol levels from fat and cholesterol in the diet

In our group over the last month or so, we have bandied about the idea of the ancestral macronutrient compositions (i.e., percent fat, protein, and carbohydrate) and how they influence health. Clearly, in the normal Western diet (approximately 45-50% carbohydrate, 35-40% fat, and 10-15% protein), if dietary saturated fats are reduced, then total and LDL cholesterol are also reduced. Keys [1965] has published an equation which has been used extensively to predict changes in serum cholesterol from dietary lipids and cholesterol. Others [Mensink 1992] more recently have confirmed Keys' equation.

However, in perhaps the most well- controlled, modern dietary study of Greenland Eskimos [Bang and Dyerberg 1980], it has been shown that ischemic heart disease is very uncommon in these people (3.5% vs. 45-50% mortality rate in Western countries). The dietary macronutrient content of these partially Westernized Eskimos was 38% carbohydrate, 39% fat, and 23% protein, whereas the values for the control group of Danish people were 47% carbohydrate, 42% fat, and 11% protein. Mean total cholesterol levels in the Eskimos (5.03 mmol/liter) were significantly lower than in the Danes (6.18 mmol/liter) whereas triglycerides (TG) (0.57 vs. 1.23 mmol/liter) and VLDL (0.43 vs. 1.29 mmol/liter) were much lower in the Eskimos, and HDL levels were significantly higher (4.00 vs. 3.34 mmol/liter).

Based upon the Keys et al. equation, the actual difference between the Eskimos' and Danes' total cholesterol levels should have been 0.67 mmol/liter, whereas in actuality it was 1.15 mmol/liter. This data suggests that the Keys equation may be invalid under circumstances wherein high quantities of animal products replace traditionally grain- dominated diets. Possible reasons for this discrepancy include the following characteristics of the Eskimos' diet:

  • Higher protein levels in the face of lowered carbohydrate may induce different lipoprotein transport mechanisms [Wolfe 1995], and/or
  • Differences in polyunsaturated fats between the two diets (high levels of n3 fats, and high levels of preformed long- chain fats of both n3 and n6 families).
The bottom line here is that present- day hunter- gatherers maintain quite low serum lipid levels despite high consumptions of animal- based foods.

Comment: To clarify the above, it appears that the likely reason the Keys equation fails to correctly predict cholesterol levels in situations such as the Eskimo study above is that it does not take into account the effects of carbohydrate on insulin secretion. Hyperinsulinemia now appears as if it may be one of the largest risk factors for CHD. Both high- protein and low- carbohydrate intakes, which were seen in the Eskimo study, promote inhibition of excess insulin.

Exactly. Ancestral, pre- agricultural diets were quite high in animal protein, and the carbohydrate that was consumed was generally of a low glycemic index. These populations also selectively consumed the fatty portions of the killed animal (brain, bone marrow, depot fat, perinephral fat, mesenteric fat, tongue, organs, etc.). However, available evidence from living hunter- gatherers show that these surrogates of our Stone- Age ancestors maintain low risk factors for CHD (blood lipid profiles, blood pressure, insulin sensitivity, body composition, etc.). All of this on a diet which contains an average 50-65% of its total calories derived from animal foods, which therefore necessarily entails lower carbohydrate consumption.

Clearly, the Keys equation breaks down when either the macronutrient content (high protein and low carbohydrate) or the fatty- acid composition of the diet (or both) varies beyond the range of conditions from which Keys originally derived his regression. Although there is much circumstantial evidence to indicate that the Keys equation is erroneous under these conditions, there is no empirical data that I am aware of which has specifically investigated or confirmed this concept.

Clarification of the role of saturated fats in promoting high cholesterol

Comment: Some who promote diets based on those of traditional peoples-- who may at times have eaten higher levels of saturated fat-- suggest that the modern (high) levels of CHD do not have anything to do with saturated fat from animal sources. Rather, they point to modern processing techniques as having introduced new food substances into the human diet with detrimental effects, particularly excess polyunsaturates, hydrogenated oils, and refined carbohydrates.


There is much evidence to support the second half of this sentiment, but the evidence does not agree with the first part.


  • Hydrogenated oils (trans fatty acids), refined carbohdrates, and polyunsaturates. There is now substantial evidence that hydrogenated oils (trans fats) are atherogenic via their hypercholesterolemic effects [Willett et al. 1994]. And in terms of their cholesterol- raising properties they may be worse than saturated fats, because they cause a decrease in HDL cholesterol [Ascherio et al. 1997]. Refined carbohydrate (sucrose in particular) has been known for more than 30 years [Yudkin 1972] to be implicated in its CHD- promoting effects, probably through increases in VLDL (the precursor to LDL), triglycerides, total cholesterol, and perhaps decreases in HDL [Hollenbeck et al. 1989]. Recently it has been recognized that although dietary polyunsaturates may lower serum cholesterol levels, they may actually increase the risk for CHD by increasing the susceptibility of LDL to oxidation [Louheranta et al. 1996].

    So I am in agreement that hydrogenated fats, refined carbohydrates, and excessive polyunsaturated fats (primarily linoleic acid, 18:2n6) contribute to the development of CHD via hypercholesterolemic and LDL- oxidizing mechanisms.


  • Saturated animal fats in overall dietary context. However, I cannot agree with the statement that saturated fats from animals in modern diets have nothing to do with CHD. It may be possible that the hypercholesterolemic effects of saturated fats (12:0, 14:0, 16:0) can be negated or somewhat ameliorated by extremely low levels of dietary carbohydrates (particularly in insulin- resistant subjects) or by high levels of dietary protein (>20% of total calories) via protein's VLDL- suppressing effects [Kalopissis et al. 1995].

    However, it is clear beyond a shadow of a doubt that dietary saturated fats (12:0, 14:0, and 16:0) elevate serum cholesterol levels within the context of the "average American diet." A recent meta- analysis of 224 published studies encompassing 8,143 subjects (many under metabolic ward conditions) has unequivocally demonstrated the hypercholesterolemic effect of dietary saturated fats [Howell et al. 1997]. The cellular basis for this observation stems from the regulation of low- density lipoproteins (LDLs). When the amount of cholesterol or saturated fat coming into the body is increased, there is an expansion of the sterol pools within liver cells, and to a lesser extent, peripheral cells, which causes a down- regulation of LDL receptors. As a consequence, LDL in plasma increases [Dietschy 1997].

    Some have argued that increases in total plasma cholesterol and LDL may not necessarily have a direct relationship to mortality from CHD [Stamler et al. 1986]. Clearly, there are a wide variety of independent risk factors for CHD including hypertension, homocysteine (increased by deficiencies primarily in folate, vitamin B-6, and secondarily in B-12), catecholamines, n6/n3 fatty- acid ratio, antioxidant status (vitamins E, C, beta- carotene, phytochemicals, etc.), dietary fiber, cigarette smoking, and ethanol (alcohol) consumption, which influence a variety of physiological systems involved with CHD. However, there is powerful evidence (n = 356,222) to indicate that the relationship between serum cholesterol levels and the risk of premature death from CHD is, nevertheless, continuous and graded [Stamler et al. 1986].

    Therefore, the recommendation by some that it is harmless to consume high levels of dietary saturated fats within the context of the "average American diet" appears to not only be erroneous, but probably deadly. Our hunter- gatherer ancestors consumed high levels of animal food (probably >55% of their total daily calories); however, the context under which this was done was much different than present- day conditions.

    As I have previously mentioned, the carbohydrate content of the diet was low (~<35% of total calories) and composed of plant foods with high soluble fiber and low starch content. The protein content of the diet would have exceeded 20% and may have been as high as 30-40%. The polyunsaturated fats consumed would have had a low n6/n3 ratio, and there would have been both ample levels of 20 and 22-carbon fats of both the n6 and n3 variety. Since marrow contains 70-75% monounsaturated fats and was a favored food, it is likely that although the fat content of the diet may have been as high as 40%, it was composed of not only a much more favorable n6/n3 polyunsaturated fat ratio, but higher levels of monounsaturated fats and non- atherogenic saturated fats such as stearic acid (18:0) as well.

The role of essential fatty acids (EFAs) and the balance of omega-6 to omega-3 fats.

Question: Regarding saturated fats in context, is it not the case that:


  • A low-fat diet that is deficient in (polyunsaturated) essential fatty acids (EFAs) will cause heart disease, whereas


  • A diet high in saturated fat but containing plenty of polyunsaturated n3 fats (i.e., omega-3s, found in animal foods) will keep arteries clear, such as in the Eskimo?


The first part of this is essentially correct. Indeed, levels of EFAs (essential fatty acids-- particularly the chain- elongated 20 and 22-carbon forms of both n6 and n3 families) are inversely related to levels of coronary heart disease (CHD). Paradoxically (at least in terms of the American Heart Association dietary recommendations), Hindu vegetarians from India whose diet is composed largely of low-fat grains and pulses (legumes) maintain CHD rates equal to [Begom et al. 1995] or higher [Miller et al. 1988] than those in the USA and countries of Europe, despite their diets' lower total fat content when compared to American and European diets.

Indian populations have consistently exhibited high plasma n6/n3 ratios, low levels of 20:5n3 and 22:6n3, and high levels of 18:2n6 when compared to Western populations [Miller et al. 1988; Reddy et al. 1994; Ghafoorunissa 1984; McKeigue et al. 1985]. All of these EFA profiles are conducive to CHD and occur because of the lack of an appropriate balance of n6/n3, and because of the almost total lack of 20 and 22-carbon fatty acids in commonly consumed plant- based foods.

Regarding the second part of the above comment, it is partially correct to say that omega-3 (n3) fats provide protection against CHD, but it has little to do, directly, with keeping the arteries clear (i.e., atherosclerosis). N3 fats provide protection from CHD in that they lower triglycerides and perhaps VLDL; additionally, they reduce platelet adhesitivity and decrease thrombotic tendencies as well as reducing cardiac arrhythmias [Leaf et al.1988]. However, recent large- scale meta- analyses [Harris 1997] show that n3 fats actually cause a 5-10% rise in LDL cholesterol and a small rise (1-3%) in HDL. Eskimo populations indeed do consume higher levels of both saturated fat and polyunsaturated n3 fats than do Western populations; they also exhibit significantly lower serum LDL and total cholesterol levels than Europeans [Bang and Dyerberg 1980].

Thus, logic (derived from the meta- analytical data) dictates that the n3 fats are not the element responsible for the lower total and LDL serum cholesterol in these populations. Careful analysis of Bang and Dyerberg's data [1980] reveals a much higher protein intake (26% of total calories) compared to the 11% value in Danes. High protein intakes are known to cause drastic inhibition of hepatic VLDL synthesis [Kalopissis et al. 1995] (VLDLs are the source of LDLs), and high- protein diets in humans have been clinically shown to reduce total cholesterol, LDL cholesterol, and triglycerides while simultaneously increasing HDL [Wolfe 1995]. Further, acute consumption of high levels of low-fat (6.5%), lean- beef protein is not associated with a post- prandial rise in insulin but rather an increase in glucagon levels [Westphal et al. 1990].

Consequently, the major reason why Eskimo diets keep serum cholesterol levels low and atherosclerosis at bay is because of their high protein content primarily. There is no doubt that n3 fats also contribute to lowering CHD, but it is not directly mediated by a lowering of LDL cholesterol but rather by other mechanisms previously outlined.

Effect of fat, protein, and carbohydrate on glucagon levels

Follow-up question: To clarify the above statement that low-fat (6.5%) beef stimulates higher glucagon levels, isn't it the protein content of beef rather than the fact it is also low in fat that stimulates glucagon? My impression was that fat intake level has little effect on the insulin/ glucagon response to food. (Editorial note: Release of insulin not only causes uptake of glucose into cells, but also promotes fat storage. Glucagon is the other side of the equation, causing mobilization of body fat for conversion to energy; that is, it causes fat to be burned.)

As far as I know, there are no good, recent data evaluating the effects of varying protein/ fat mixtures upon insulin/ glucagon responses in humans. Most of the data involves manipulating carbohydrate, with varying amounts of fat; protein is usually held constant. The Westphal et al. [1990] paper evaluates protein/ carbohydrate mixtures on serum glucagon responses. Pure dietary carbohydrate (50g glucose) shows no rise in plasma glucagon, whereas pure protein (actually 93.5% lean beef, 6.5% fat, as mentioned above) causes the greatest rise in glucagon after 1 hour; with roughly equal areas under the curve after 3 hours when comparing pure protein to protein/ carbohydrate mixtures (50g glucose/ 50g protein).

Thus, there appears to be a dose/ response effect on glucagon with protein/ carbohydrate mixtures, and from the data, it can probably be interpreted that there is a dose/ response effect with pure protein. As far as insulin response goes (as opposed to glucagon response), fat/ carbohydrate mixtures cause a greater rise than carbohydrate meals alone, presumably because of the stimulatory effect of fat upon glucose- dependent insulinotropic peptide (GIP) [Collier et al. 1988].

Thus, as indicated by the question, there is a dose- dependent effect of dietary protein upon glucagon secretion which is largely independent of either carbohydrate or fat.

Protein levels and their effect on blood lipids

Some would dismiss the idea that dietary protein can have any influence upon cardiovascular disease with the argument that there is no difference in CHD incidence in populations consuming high vs. low- protein diets. However, a serious problem with this argument is the lack of much substantial variability in current protein consumption levels worldwide to produce support for this line of reasoning via epidemiological comparisons.

Global surveys of the world's populations indicate a remarkably limited range of protein consumption that varies from about 10 to 15% of total calories [Speth 1989]. Further, except for reports of Inuit and Eskimo diets, I know of no references showing any contemporary populations consuming 15-20% of their calories as protein, much less high- protein diets in the 30-40% range of consumption such as our ancestors or recent hunter- gatherers have sometimes eaten.

Speth [1989] has extensively studied protein intakes in contemporary worldwide populations and notes that most human populations today obtain between 10-15% of their total energy requirements from protein. For Americans the value is 14%, for Swedes it is 12%; for Italian shipyard workers it is 12.5-12.8%; for Japanese it is 14.4%, and for West Germans it is 11.1%. Even among athletes, values rarely exceed 15%. Speth [1989] shows that Italian athletes consumed between 17-18% of their caloric intake as protein; Russian athletes consumed 11-13%; and Australian athletes competing at the 1968 Olympic Games consumed 14.4% of their daily calories as protein. This data clearly demonstrates the relative homogeneity amongst contemporary global populations in their protein consumption levels.

That protein consumption may have anything to do with the atherosclerotic process and hence CHD is an obscure topic which has been rarely examined by the medical and nutritional communities. It is not surprising that few are aware of the literature which supports this concept. However, there are now at least three human clinical trials [Wolfe et al. 1991; Wolfe et al. 1992; Wolfe 1995] demonstrating that isocaloric (calorie- for- calorie) substitution of protein (ranging from 17-27% of total daily calories) for carbohydrate reduces triglycerides, VLDL, LDL, and total cholesterol while increasing HDL cholesterol. Further, acute consumption of high levels of beef protein without carbohydrate evokes an extremely small rise in serum insulin levels and a concomitant substantial rise in glucagon [Westphal et al. 1990]. Both of these acute responses would tend to be associated with a reduced risk for CHD.

Lastly, in animal models, high levels of protein are known to dramatically inhibit hepatic VLDL synthesis [Kalopissis et al. 1995]. VLDLs are the precursor molecules for LDL cholesterol. In their classic study of Inuit, Bang and Dyerberg [1980] have shown that the serum cholesterol levels of the Inuit were 0.48 mmol/liter lower than what would have been predicted by the Keys equation, which estimates plasma lipid levels from dietary saturated fats, polyunsaturated fats, and cholesterol. At the time (1980), it was suggested that the paradoxically low serum cholesterol levels may have resulted from the higher omega-3 (n3) fats found in the Eskimo's seafood- based diet.

However, after almost 30 years of research, meta- analytical studies have shown that n3 fatty acids slightly elevate (by 5-10%) LDL cholesterol concentrations, but do not materially affect total cholesterol [Harris 1997]. Consequently, it may have been the higher dietary protein intake (23-26% of total calories) in the Inuit compared to the Danish controls (11% of total calories as protein) which accounted for these differences. However, since the Keys equation considers dietary monounsaturated fats as neutral (which more recent research indicates is not the case [Gardner et al. 1995]), it is possible that the higher monounsaturated fat content (57.3% of total fat) in the Inuit diet (vs. 34.6% in the Danes) may have also contributed to the plasma cholesterol differences.

Low-carbohydrate diets by themselves do not eliminate the cholesterol- raising effects of high- saturated- fat diets

Many people who have been influenced by the recent interest in low- carbohydrate dieting would argue that the cholesterol- lowering effect of the Eskimo diet stemmed from its low carbohydrate content. However, in one of the few (and best- controlled) metabolic ward trials of a carbohydrate- free (<20 gm/day) diet, Phinney and colleagues [Phinney et al. 1983] demonstrated a rather large rise in serum cholesterol (159 to 208 mg/dl) in nine lean, healthy males who participated in this 35-day in-patient trial.

The protein content of the diet was estimated to be 15%, whereas the fat content of the diet represented between 83-85% of total daily calories. Consequently, during the dietary trial, the protein content remained similar to the average daily intake in the U.S. and was not increased. This experiment shows that a carbohydrate- free diet composed of "ground beef, breast of chicken, water- packed tuna fish, powdered egg solids, and cheddar cheese with mayonnaise, heavy cream, sour cream, and cream cheese as primary lipid sources" was definitely hypercholesterolemic.

In a less- well- publicized but highly controlled clinical research center (CRC) study, Gray et al. showed similar results in a 3-week study of 10 healthy males who consumed a diet composed of 73-75% fat, 7-9% carbohydrate, and 16-20% protein. Compared to their standard (normal- carbohydrate) diet, the high- fat diet increased total cholesterol from 156.5 mg/dl to 167.6 mg/dl, and LDL cholesterol increased from 46.6 mg/dl to 55 mg/dl. The total cholesterol/HDL ratio, however, improved on the high- fat diet, going from 3.36 to 3.20.

High- fat, low- carbohydrate diets-- as in the Phinney [1983] and Gray studies-- characteristically induce other beneficial lipid profiles such as increased HDL levels and decreased triglyceride levels. These blood lipid changes (increased HDL and reduced triglycerides) have also been frequently demonstrated in reduced- carbohydrate diets [Jeppesen et al. 1997; Coulston et al. 1983] in which carbohydrate has been reduced, but not as drastically as in the Phinney and Gray trials.

So in summary, the animal foods of our Stone- Age ancestors were probably non- atherogenic because they contained high levels of protein (>20% of total calories), lower levels of saturated fats, higher levels of monounsaturated fats, higher levels of n3 polyunsaturated fats, little or no trans fats, and higher levels of HUFA (>18-carbon) fats of both the n6 and n3 varieties than modern Western meat- based diets. The higher consumption of animal- based foods would have necessarily reduced the carbohydrate content of the diet, and this would have also benefited certain aspects of the lipid profile as just enumerated.

Glycemic response of fat combined with carbohydrate

In a previous comment, I suggested that meals of pre- agricultural peoples tended to produce less of a glycemic response than do modern Western meals. This was based on the observation that hunter- gatherer meals generally were not the elaborate mixtures of fat/ carbohydrate/ protein that are typical of Western meat/ potato meals. Hunter- gatherers quite often would eat only the animal killed for a meal without added plant courses. Thus, protein/ fat macronutrient mixtures were the norm. Carbohydrates generally were consumed as they were collected, or separate from animal- based meals. It has been well- established that by mixing fat with carbohydrate, the glycemic response worsens [Collier et al. 1988].

What is the relevance of genetic differences in individual blood lipid response to high and low-fat diets?

In view of recent discussions about low- carbohydrate diets and reevaluation of the effects of high- carbohydrate diets, there have been speculations regarding human blood- lipid responses to high and low- carbohydrate diets, and whether or not there is a genetic basis for differential responders. To follow up on this, there is substantial evidence to show that blood- lipid response to variation in dietary fat and cholesterol intake varies widely among individuals [Mistry et al. 1981; Jacobs et al. 1983; Katan et al. 1988], and that this variability is likely attributable to genetic factors with polymorphisms [variant forms of a gene] at several genetic loci, including genes for apolipoproteins and for low- density lipoprotein (LDL) particle size and density [Dreon et al. 1992].

There is an LDL subclass called pattern "B" which is characterized by a preponderance of small, dense LDL particles, elevated triglycerides, low high- density cholesterol (HDL), and increased coronary heart disease (CHD) risk. LDL pattern "B" occurs in approximately 30% of the male population [Austin et al. 1988]. LDL subclass pattern "A" is characterized by larger, more buoyant LDL particles. Low- fat, high- carbohydrate diets induce a reduction in the atherogenic, small, dense LDL in individuals displaying pattern "B", and also cause reductions in LDL cholesterol greater than in subjects displaying pattern "A" [Dreon and Krauss 1997]. These data clearly suggest that low- fat, high- carbohydrate diets may be more effective in lowering LDL cholesterol and small, dense LDL in about 30% of the population, and less effective in 70% of the population.

LDL subclass pattern "B" is influenced by a major gene or genes with a prevalence in the American population estimated to be 25% [Austin et al. 1988]. The specific gene or genes responsible for this trait have not been identified, but there is evidence to show linkage to polymorphic markers near the LDL receptor gene on chromosome 19p [Nishina et al. 1992].

To date, there are no experimental data evaluating the effects of quite low- carbohydrate diets (<30% of total energy) upon blood lipid responses in LDL subclasses "A" or "B". However, Krauss et al. [1995] have clearly shown that all subjects (n = 105), whether subclass "A" or "B", responded to a high- fat diet (46% energy) by substantial increases in LDL cholesterol, and responded to a low- fat diet (23.9% energy) by decreases in LDL cholesterol. The difference was simply in the magnitude of the negative effect experienced, not whether it occurred or not.

This information does not support the contention by some that differential responders to high and low- fat diets bias the interpretation of dietary intervention trials, nor does it lend support to the proposal that high- fat diets can improve blood- lipid profiles. I contend that any improvement in total cholesterol or LDL cholesterol by uncontrolled, self- administered low- carbohydrate diets are an artifact of:

  • Reductions in total caloric intake,


  • Increases in total protein,


  • Unknowing changes in the dietary polyunsaturated/ monounsaturated/ saturated fat (P:M:S) ratio, and


  • Combination of the three.


Further, improvements in triglycerides, VLDL, and HDL can be mainly attributed to reductions in carbohydrate.

Under isocalorically (calorie- for- calorie) controlled conditions in which dietary saturated fat is increased at the expense of any other lipid or macronutrient, there will be a characteristic increase in LDL cholesterol, as shown time and again with meta- analyses [Howell et al. 1997], under metabolic ward conditions [Phinney et al. 1983], and corroborated by in vitro and in vivo data showing that LDL receptors are down- regulated by dietary saturated fat [Brown and Goldstein 1976].


As I hope the foregoing has demonstrated, further studies that have been performed in the years since the low-fat, high- carbohydrate viewpoint first became standard have revealed additional factors affecting blood lipids, and that the previous view has been too simplistic. Serious drawbacks have become apparent in the conventional wisdom about low-fat, high- carbohydrate diets. No longer does this view adequately explain what we have come to know about the effects of macronutrient content with increasing resolution at the biochemical level over the last decade.

We are entering an era of dietary research where the details of underlying biochemical processes that govern lipid responses are being increasingly well- understood. Certain of these details validate the positive health effects that may accrue from the dietary pattern suggested by recently emerging studies of diet in human evolution. Hunter- gatherers who eat high levels of protein, lower levels of carbohydrate, and similar or even higher levels of fat (but with a much different lipid profile) compared to modern Western diets exhibit extremely positive blood lipid profiles and quite low rates of CHD. This presents a serious challenge for researchers, since this result would not be predicted by previous theories about fat in the diet.

While the detrimental role of high levels of saturated fat by itself has been increasingly well- validated, the overall picture of the various other types of fat is turning out to be more complex. Fat is as essential a nutrient as the other macronutrients. More important than the overall level of fat in the diet are the roles and ratios of specific types of fat, such as the positive role of monounsaturated fats and a high n3/n6 polyunsaturated ratio, and the negative effects of trans fatty acids and deficiencies in EFAs.

Where the polyunsaturated fats are concerned, modern diets contain excessive amounts of the n6 fat linoleic acid (that would have been present in lesser amounts in preagricultural diets), which promotes oxidation of cholesterol and consequently formation of atherosclerotic plaque. Also, what saturated fats are consumed by pre- agricultural peoples come from wild animal tissues. Compared to modern domesticated animals, these animal tissues are much higher in the non-atherogenic saturated fat stearic acid and lower in the 14:0 and 16:0 fats that promote high cholesterol.

At the same time, as the roles of various fats in the diet are becoming more well- understood, attention has recently begun to turn to investigation of the biochemical effects on blood lipids of the other macronutrients. By comparison with the voluminous studies performed in recent decades on fatty acids, these have been relatively ignored. However, only by devoting the same detailed attention to the effects of carbohydrate and protein on blood lipid response will we fully understand the role of all the macronutrients on health in relation to each other. As previously mentioned, the hyperinsulinemic effect of excess carbohydrates is looming large as a subject warranting much further study. And what studies have been performed initially on higher protein consumption levels show that they exert very positive effects on blood- lipid profiles.

In this ongoing investigation, the "paleolithic" picture of the foods and macronutrient ratios that would have prevailed during human evolution provides a valuable template: One that can yield key insights for guiding future study into the food consumption patterns to which the human species is genetically best adapted.

--Loren Cordain, Ph.D.

For further Paleodiet research from Dr. Cordain, you can download printable PDFs of his research group's peer-reviewed papers (more than 20 at last count) at his website, plus get information about his 2002 book, The Paleo Diet.



Artaud-Wild SM et al. (1993) "Differences in coronary mortality can be explained by differences in cholesterol and saturated fat intakes in 40 countries but not in France and Finland. A paradox." Circulation, vol. 88, pp. 2771-2779.

Ascherio A et al. (1997) "Health effects of trans fatty acids." American Journal of Clinical Nutrition, vol. 66, pp. 1006s-1010s.

ASCN/AIN Task Force on Trans Fatty Acids. (1996) "Position paper on trans fatty acids." American Journal of Clinical Nutrition, vol. 63, pp. 663-670.

Austin MA et al. (1988) "Inheritance of low- density lipoprotein subclass patterns: results of complex segregation analysis." American Journal of Human Genetics, vol. 43, pp. 838-846.

Bang HO, Dyerberg J. (1980) "Lipid metabolism and ischemic heart disease in Greenland Eskimos." In: Draper HH (ed.) Advances in Nutrition Research, vol. 3, New York: Plenum Press, pp. 1-22.

Begom R et al. (1995) "Prevalence of coronary artery disease and its risk factors in the urban population of south and north India." Acta Cardiologica, vol. 50, pp. 227-240.

Brown MS, Goldstein JL. (1976) "Receptor-mediated control of cholesterol metabolism." Science, vol. 191, pp. 150-154.

Chen YDI et al. (1992) "Effect of acute variations in dietary fat and carbohydrate intake on retinly ester content of intestinally derived lipoproteins." J Clin Endocrin Metabolism, vol. 74, pp. 28-32.

Collier GR et al. (1988) "The acute effect of fat on insulin secretion." J Clin Endocrin Metabolism, vol. 66, pp. 323-326.

Cordain L, Gotshall RW, Eaton SB. (1997) "Evolutionary aspects of exercise." World Review of Nutrition and Dietetics, vol. 81, pp. 49-60.

Cordain L, Martin C, Florant G, Watkins BA. (1998) "The fatty acid composition of muscle, brain, marrow and adipose tissue in elk: evolutionary implications for human dietary lipid requirements." World Review of Nutrition and Dietetics, vol. 83, p. 225.

Coulston AM et al. (1983) "Plasma glucose, insulin and lipid responses to high- carbohydrate, low-fat diets in normal humans." Metabolism, vol. 32, pp. 52-56.

Crawford MA et al. (1969) "Linoleic acid and linolenic acid elongation products in muscle tissue of syncerus caffer and other ruminant species." Biochem J, vol. 115, pp. 25-27.

Denke MA, Grundy SM. (1991) "Effects of fats high in stearic acid on lipid and lipoprotein concentrations in men." American Journal of Clinical Nutrition, vol. 54, pp. 1036-1040.

Denton D et al. (1995) "The effect of increased salt intake on blood pressure of chimpanzees." Nature Medicine, vol. 1, pp. 1009-1016.

Dietschy JM. (1997) "Theoretical considerations of what regulates low- density lipoprotein and high- density lipoprotein cholesterol." American Journal of Clinical Nutrition, vol. 65, pp. 1581S-1589S.

Dreon DM et al. (1992) "Gene-diet interactions in lipoprotein metabolism." Monographs in Human Genetics, vol. 14, pp. 325-349.

Dreon DM, Krauss RM. (1997) "Diet-gene interactions in human lipoprotein metabolism." J Am Coll Nutr, vol. 16, pp. 313-324.

Eaton S. Boyd; Konner M; Shostak M. (1988) "Stone-agers in the fast lane: chronic degenerative diseases in evolutionary perspective." American Journal of Medicine, vol. 84, pp. 739-749.

Gardner CD et al. (1995) "Monounsaturated versus polyunsaturated dietary fat and serum lipids. A meta analysis." Arterioscler Thromb Vasc Biol, vol. 15, pp. 1917-1927.

Ghafoorunissa. (1984) "Essential fatty-acid nutritional status of apparently normal Indian men." Human Nutrition: Clinical Nutrition, vol. 38C, pp. 269-278.

Gray CG et al. (year unavailable; likely early 1980s) "The effect of a three- week adaptation to a low- carbohydrate/ high- fat diet on metabolism and cognitive performance." Naval Research Center Report No. 90-20. Naval Health Research Center, San Diego, California.

Grundy SM. (1986) "Comparison of monosaturated fatty acids and carbohydrates for lowering plasma cholesterol." N Engl J Med, vol. 314, pp. 745-748.

Harris WS. (1997) "n3 fatty acids and serum lipoproteins: human studies." American Journal of Clinical Nutrition, vol. 65(supp), pp. 1645s-1654s.

Hollenbeck CB et al. (1989) "Effects of sucrose on carbohydrate and lipid metabolism in NIDDM patients." Diabetes Care, vol. 12, pp. 62-66.

Howell WH et al. (1997) "Plasma lipid and lipoprotein responses to dietary fat and cholesterol: a meta- analysis." American Journal of Clinical Nutrition, vol. 65, pp. 1747-1764.

Jacobs DR et al. (1983) "Variability in individual serum cholesterol response to change in diet." Arteriosclerosis, vol. 3.

Jeppesen J et al. (1997) "Effects of low-fat, high- carbohydrate diets on risk factors for ischemic heart disease in postmenopausal women." American Journal of Clinical Nutrition, vol. 65, pp. 1027-1033.

Kalopissis AD et al. (1995) "Inhibition of hepatic very- low- density lipoprotein secretion in obese Zucker rats adapted to a high- protein diet." Metabolism, vol. 44, pp. 19-29.

Katan MB et al. (1988) "Congruence of individual responsiveness to dietary cholesterol and to saturated fat in humans." Journal of Lipid Research, vol. 29, pp. 883-892.

Keys A et al. (1965) "Serum cholesterol response to changes in the diet. IV. Particular saturated fatty acids in the diet." Metabolism, vol. 14, pp. 776-787.

Krauss RM, Dreon DM. (1995) "Low-density lipoprotein subclasses and response to a low-fat diet in healthy men." American Journal of Clinical Nutrition, vol. 62, pp. 478s-487s.

Leaf A et al. (1988) "Cardiovascular effects of n3 fatty acids." New England Journal of Medicine, vol. 318, pp. 549-557.

Leonard WR, et al. (1994) "Correlates of low serum lipid levels among Evenki herders of Siberia." American Journal of Human Biology, vol. 6, pp. 329-338.

Louheranta AM et al. (1996) "Linoleic acid intake and susceptibility of very- low- density and low- density lipoproteins to oxidation in men." American Journal of Clinical Nutrition, vol. 63, pp. 698-703.

McKeigue PM et al. (1985) "Diet and risk factors for coronary heart disease in Asians in northwest London." Lancet, 1985;ii, pp. 1086-1090.

Mensink RP et al. (1987) "Effects of monounsaturated fatty acids versus complex carbohydrates on high- density lipoproteins in healthy men and women." Lancet, 1987;1, pp. 122-125.

Mensink RP, Katan MB. (1992) "Effect of dietary fatty acids on serum lipids and lipoproteins. A meta- analysis of 27 trials." Arterioscler Thromb, Aug. 1992, vol. 12, no. 8, pp. 911-919.

Miller GJ. et al. (1988) "Dietary and other characteristics relevant for coronary heart disease in men of Indian, West Indian and European descent in London." Atherosclerosis, vol. 70, pp. 63-72.

Mistry F et al. (1981) "Individual variation in the effects of dietary cholesterol on plasma lipoproteins and cellular homeostasis in man." J Clin Invest, vol. 67, pp. 493-502.

Nelson GJ et al. (1995) "Low-fat diets do not lower plasma cholesterol levels in healthy men compared to high- fat diets with similar fatty acid composition at constant caloric intake." Lipids, vol. 30, pp. 969-976.

Nishina PM et al. (1992) "Linkage of atherogenic lipoprotein phenotype to the low- density lipoprotein receptor locus on the short arm of chromosome 19." Proc Natl Acad Sci, vol. 89, pp. 708-712.

Phinney SD et al. (1983) "The human metabolic response to chronic ketosis without caloric restriction: physical and biochemical adaptation." Metabolism, vol. 32, pp. 757-768.

Reaven GM. (1995) "Pathophysiology of insulin resistance in human disease." Physiol Rev, vol. 75, pp. 473-486.

Reddy S. et al. (1994) "The influence of maternal vegetarian diet on essential fatty acid status of the newborn." European Journal of Clinical Nutrition, vol. 48, pp. 358-368.

Speth JD. (1989) "Early hominid hunting and scavenging: the role of meat as an energy source." Journal of Human Evolution, vol. 18, pp. 329-343.

Stamler J et al. (1986) "Is relationship between serum cholesterol and risk of premature death from coronary heart disease continuous and graded?" JAMA, vol. 256, pp. 2823-2828.

Steinberg D et al. (1989) "Beyond cholesterol: modifications of low- density lipoprotein that increase its atherogenicity." New England Journal of Medicine, vol. 320, pp. 915-924.

Varo P. (1974) "Mineral element balance and coronary heart disease." Internat J Vit Nutr Res, vol. 44, pp. 267-273.

Vega GL et al. (1996) "Hypoalphalipoproteinemia (low high- density lipoprotein) as a risk factor for coronary heart disease." Curr Opin Lipidology, vol. 7, pp. 209-216.

Westphal SA et al. (1990) "Metabolic response to glucose ingested with various amounts of protein." American Journal of Clinical Nutrition, vol. 52, pp. 267-272.

Willett WC et al. (1994) "Trans fatty acids: are the effects only marginal?" American Journal of Public Health, vol. 84, pp. 722-724.

Wolfe BM et al. (1991) "Short-term effects of substituting protein for carbohydrate in the diets of moderately hypercholesterolemic human subjects." Metabolism, vol. 40, pp. 338-343.

Wolfe BM et al. (1992) "High-protein diet complements resin therapy of familial hypercholesterolemia." Clin Invest Med, vol. 15, pp. 349-359.

Wolfe BM. (1995) "Potential role of raising dietary protein intake for reducing risk of atherosclerosis." Can J Cardiol, vol. 11(supp G), pp. 127G-131G.

Yudkin J. (1972) "Sucrose and cardiovascular disease." Proc Nutr Soc, vol. 31, pp. 331-337.

placeholder image
placeholder image