Phytochemical Fallacies?

[For those who already read the first version of this post, I made a mistake while editing and erased all the great commments :-(. But I made some improvements to the content anyway. :-)]

I used to believe that people would derive great benefits from a diet containing large amounts of fruits and vegetables because such foods would supply people with an abundance of “beneficial” phytochemicals. I had gotten this belief from my conventional nutrition education – you can find the claim promoted in most textbooks and popular books on nutrition. Consequently I consumed 15 or more servings of vegetables and fruits every day for many years, and advocated a “produce-dominated” diet in my last book.

However, as I integrated the concept of evolutionary nutrition and studied native diets and health in that context, I gradually came to realize that humans can live quite well without plant foods (witness the Eskimos, Nenet, Masai, and numerous other tribes). This of course led me to question the idea that people need daily doses of phytonutrients to maintain a semblance of health. After all, Eskimos had no cancer on their native diet, yet antioxidant rich vegetables and fruits played little or no role in their daily diets.

Epidemiology Not Impressive

If you examine the evidence frequently trotted out to support the frequent ingestion of phytochemicals from vegetables and fruits, you will notice that advocates of plant-dominated diets rely almost exclusively on epidemiological evidence.

For example, to support a recommendation for consumption of at least 400g of vegetables and fruits daily, the World Cancer Research Institute and the American Institute for Cancer Research cite 4500 epidemiological studies reviewed by 150 scientists as evidence for the cancer-preventive effects of vegetables and fruits.

Sounds impressive, but it does not establish that vegetables and fruits protect against cancer. These studies at most establish that people who have a higher intake of vegetables and fruits have less cancer than people who a lower intake of the same. Since people who have the habit of eating lots of produce very likely have higher income, or other habits that may also protect against cancer, like refraining from smoking, taking less alcohol, exercising regularly, getting sun exposure (vitamin D), or eating less sugar, we can’t conclude from these studies that their higher intake of produce is what is actually responsible for their reduced risk of cancer.

In science, clinical trials provide the best way to test a hypothesis. If we hypothesize that increased vegetable and fruit intake will provide health benefits to humans, our best evidence will come from studies where we get people to increase their consumption of vegetables and fruits, then measure outcomes, preferably hard endpoints like incidence of heart disease or cancer, but at least soft endpoints like changes in blood characteristics that we have reason to believe will influence health outcomes.

Clinical Trials

In clinical trials, it appears that increasing human intake of vegetables and fruits either may have little or no benefits, or could even cause injury.

As discussed by Stephan in his two part series The Diet-Heart Hypothesis: Oxidized LDL , we have good evidence to indicate that oxidized LDL causes atherosclerosis. Since vegetables and fruits supply lots of putative “antioxidants,” we could recommend increasing vegetables and fruits for prevention of heart disease if we could show that it would reduce oxidation of LDL.

Van het Hoff et al found increased intake of vegetables and fruits did raise serum levels of vitamin C and carotenoids (ie, {alpha}-carotene, ß-carotene, lutein, zeaxanthin, and ß-cryptoxanthin), but this did not result in reduced oxidation of LDL. They wrote:

“Despite the significant increases in plasma concentrations of vitamin C and carotenoids in the vegetable- and carotenoid-supplemented groups, the total antioxidant activity of plasma, measured as ferric-reducing ability, remained unchanged (Table 3). Furthermore, consumption of the diets supplemented with vegetables or carotenoids did not enhance protection of LDL against copper-induced oxidation ex vivo. Neither the changes in lag time before onset of oxidation nor the maximum rate of oxidation were significantly different from those found in the control group.”


In Changes in Dietary Fat Intake Alter Plasma Levels of Oxidized Low-Density Lipoprotein and Lipoprotein(a), Silaste et al report the results of feeding two diets to thirty-seven healthy women. Both diets contained a reduced amount of total and saturated fat, while one diet was low in vegetables and the other was high in vegetables, berries, and fruit.

As reported by Silaste et al, “During the low-fat, low-vegetable diet, the subjects consumed 2 portions of both fruit and vegetables daily. On the low-fat, high-vegetable diet, they consumed 4 to 5 portions of fruit or berries and 5 to 6 portions of vegetables.”

Regarding fat intake, “The total fat intake was 70 g per day at baseline and decreased to 56 g (low-fat, low-vegetable diet) and to 59 g (low-fat, high-vegetable diet). The saturated fat intake decreased from 28 g to 20 g and to 19 g, and the amount of polyunsaturated fat intake increased from 11 g to 13 g and to 19 g (baseline; low-fat, low-vegetable; low-fat, high-vegetable; respectively).”

What happened? “Surprisingly, we found that in response to the low-fat, low-vegetable and the low-fat, high-vegetable diets, the average plasma levels of OxLDL-EO6 increased compared with the baseline diet.” In fact, the median plasma OxLDL-EO6 –oxidized LDL—increased by 27% in response to the low-fat, low-vegetable diet and 19% in response to the low-fat, high-vegetable diet. Also, the Lp(a) concentration was increased by 7% and 9%, respectively.

The study design used a randomized crossover design for each of the study diets, which increased confidence in the conclusion that both of the interventions – cutting fat and increasing produce – resulted in increased oxidation of LDL.

To summarize the results of this study by Silaste et al:
1. Reducing saturated fat intake by 8 to 9 grams, and replacing it with 2 to 8 grams of polyunsaturates, resulted in an increase of 27% in oxidized LDL.
2. Taking this (#1) as a second base, increasing fruit and vegetable intake from four portions daily (low fat, low vegetable and fruit diet) to nine to eleven portions daily (low fat, high vegetable and fruit diet) countered the effect of increasing polyunsaturates, dropping the increase in oxidation of LDL from a 27% increase to a 19% increase compared to the "high fat" control diet.
3. Doubling fruit and vegetable intake did not completely counter the increased oxidation brought about by decreasing total and saturated fat intake and increasing polyunsaturated fat intake.
4. The group with the highest intake of fruits and vegetables (nine to eleven portions daily) had the greatest (9%) increase in Lp(a).

Research suggests that increased Lp(a) increases risk for coronary heart disease, cerebrovascular disease, atherosclerosis, thrombosis, and stroke. Marcovina et al found that vegetarian Bantus have 40% higher levels of Lp(a) than fish-eating Bantus. Combining these two studies (Marcovina and Silaste), we get the impression that eating more fish will decrease your Lp(a) levels and reduce your risk of coronary heart disease, cerebrovascular disease, atherosclerosis, thrombosis, and stroke, while eating nine to eleven servings of fruits and vegetables may increase your Lp(a) and very slightly increase your risk of those diseases!

In the WHEL trial, Pierce et al randomly assigned about 3000 women previously treated for early stage breast cancer to two groups: one received “a telephone counseling program supplemented with cooking classes and newsletters that promoted daily targets of 5 vegetable servings plus 16 oz of vegetable juice; 3 fruit servings; 30 g of fiber; and 15% to 20% of energy intake from fat” while the other received literature describing the "5-A-Day" dietary guidelines.

Over a four year period, they found that this intensive education resulted in increases in intake of fruits and vegetables and decreases in fat intake: “servings of vegetables, +65%; fruit, +25%; fiber, +30%, and energy intake from fat, −13%.” They measured plasma carotenoid concentrations and these validated changes in fruit and vegetable intake.

After 7.3 years of follow-up care, the two groups had almost identical invasive cancer incidence, 16.7% in the intervention group, and 16.9% in the control group. Increasing intake of produce appeared to have no effect on cancer recurrence despite changes in blood levels of antioxidants. It neither reduced nor increased the incidence of recurrence. Since the incidence of recurrence was the same in both groups, this leaves open the possibility that the intake of fruits and vegetables increased the risk of recurrence. Unfortunately, subjects in both arms of this study, so we can't tell if the intake of produce contributed to recurrence. They needed an arm with no intake of vegetables and fruits to find out whether fruits and vegetables might have had an adverse effect.

For another example, consider “Green tea extract only affects markers of oxidative status postprandially: lasting antioxidant effect of flavonoid-free diet.” This title tells all – a flavonoid-free diet produced a lasting antioxidant effect. The abstract reported: “Since no long-term effects of GTE [green tea extract] were observed, the study essentially served as a fruit and vegetables depletion study. The overall effect of the 10-week period without dietary fruits and vegetables was a decrease in oxidative damage to DNA, blood proteins, and plasma lipids, concomitantly with marked changes in antioxidative defence.”

So giving up fruits and vegetables for 10 weeks reduced oxidative damage to DNA, blood proteins, and plasma lipids; that means that eating fruits and vegetables increases damage to DNA, blood proteins, and plasma lipids.

How can that be?

For anyone convinced that flavonoids are just beneficial “antioxidants,” this evidence produces cognitive dissonance. How could eating vegetables and fruits possibly increase damage to DNA, blood protein, and blood lipids?

Now, taking an evolutionary tact, let us ask, why do plants produce flavonoids? Well, they provide photoprotection, necessary if you sit out in the sun all day.

But that's not all. Plants have no claws or teeth, nor can they run from predators, so they have chemical defenses instead. From a natural selection standpoint, any plant that produces and stores in its tissues chemicals that will injure, sterilize, or kill predators will have a marked selective reproductive advantage over less-equipped plants. Well, it turns out that flavonoids also function as pesticides!

In “Dietary pesticides (99.99% all natural)” Ames et al “estimate that Americans eat about 1.5 g of natural pesticides per person per day, which is about 10,000 times more than they eat of synthetic pesticide residues.” This natural pesticide intake includes compounds in coffee (caffeine, chlorogenic acid, neochlorogenic acid, caffeic acid), phenols (several hundred mg daily, in tea, berries, etc.), flavonoids and glucosinolates (another several hundred mg), potato and tomato toxins (a hundred mg), and saponins from legumes (peanuts included).

Naturally, they state, “The human intake of these toxins varies markedly with diet and would be higher in vegetarians.” Hmmm….food for thought.

Many people seek out “organic” foods to avoid pesticides. They may not know that plants grown “organically” actually have higher levels of these natural pesticides than plants grown conventionally. You will likely increase your total natural pesticide exposure by eating organic produce.

These natural phytochemicals don't appear benign – as Ames et al note, of the 52 of these natural pesticides have been tested in high-dose animal tests, 27 have proven carcinogenic, and these occur in many common foods. In isolation, in high enough doses, they have toxicity – and carcinogenicity – comparable to similar doses of synthetic chemicals.

Some will argue that the synergistic effect of multiple ingredients in plants reduces this toxicity. But I doubt that. Remember, the plants that survived evolution were those that either poisoned or created symbiotic relationships with predators. These chemicals may make plants excellent medicines -- many have antiviral, antibacterial, antifungal, and general antimicrobial effects, just what a plant needs to survive.

With this knowledge, you may no longer feel surprised that eating too many servings of vegetables and fruits – especially raw vegetables – might actually cause problems such as increased Lp(a), oxidized LDL, or damaged DNA. These natural pesticides might not kill you outright as they might an insect or worm, but they sure could cause digestive distress, cellular damage, or even neurological effects (like caffeine). You can learn more about that at Plant Poisons and Other Rotten Stuff.

The dose makes the poison. Most likely, we don't consume carcinogenic levels of these poisons when eating typical quantities of fruits and vegetables. Some hunter-gatherer tribes do eat substantial amounts of fruits and vegetables (wild types, which have even higher concentrations of poisons), yet remain free of cancer and heart disease. Significantly, hunter-gatherers tend to consume more fruits than vegetables. Fruits have lower toxin levels and enable a plant to work symbiotically with animals/predators to increase reproductive success.

Natural selection also favored the survival of plant predators (animals) that have the ability to detoxify these plant poisons. Our liver does this; but one must wonder, at what intake of produced -- especially high toxin vegetables like leaves -- does one start burdening the liver's detoxification systems?

In many cases, these poisons do have a bad taste (usually bitter) and adverse effects on the GI tract. Traditional cultures with plant based diets got around these toxins with cooking and other processing techniques. I consider unwise the modern raw-food vegetarian who does not follow their example.