Category Archives: Supplements

The Case of the Killer Vitamins Revisited

A lot of people have asked about the Atlantic article by Paul Offit, “The Vitamin Myth: Why We Think We Need Supplements.” Offit, a pediatrician, is best known for his defense of childhood immunizations. In arguing against benefits from vitamins, as he did in arguing against harm from vaccines, he takes a strong stand.

Before I take a look at his article, let me mention a competing publication about supplements that became available today, and might be more worthy of your time.

Examine.com’s Supplement Goals Reference Guide

A friend of the blog, Sol Orwell, has been an expert on nutritional supplements for many years and has consulted extensively for vitamin and supplement manufacturers. He and his colleagues have spent years compiling the most extensive database of peer-reviewed literature available, and have compiled the data into a 762 page reference guide that has a comprehensive overview of the supplement literature. It is an encyclopedic resource with background information about almost every supplement that has been studied, including herbal remedies, each graded for quality of evidence, and easily searchable by:

  • supplement, to view the evidence for effects, good and bad, of each supplement; and
  • health goal or biomarker, to see which supplements may help you achieve your health goal (such as, “blood glucose,” “breast tenderness,” “glycemic control,” “canker sores,” “fecal moisture,” “fat oxidation,” “free testosterone,” “food intake,” “memory,” “migraine,” “pain,” “postpartum depression,” and even, I kid you not, “penile girth”).

If this sounds interesting, read more about it here.

The Atlantic Article

OK, so what evidence does Offit compile against supplements?

Central to his case is a paper that we’ve already discussed on this blog (“Around the Web; The Case of the Killer Vitamins,” October 15, 2011): the 2011 analysis of the Iowa Women’s Health Study by Jaakko Mursu and collaborators in Archives of Internal Medicine. [1] Offit highlights it in the very first sentence of his article:

On October 10, 2011, researchers from the University of Minnesota found that women who took supplemental multivitamins died at rates higher than those who didn’t.

Apart from this reference he discusses two issues:

  • Linus Pauling’s argument that vitamin C can prevent colds and cancer has not been supported. A history of Pauling’s romance with vitamin C occupies about two-thirds of the article, and Offit seems to think the whole idea of supplementation was originated by Pauling and persists due to his influence (“What few people realize, however, is that their fascination with vitamins can be traced back to one man,” Pauling.).
  • Multiple studies have shown that supplementing vitamins A and E is often harmful.

This is hardly a comprehensive case against supplementation; it only shows that a few supplements tend to be harmful or lack benefits.

Regarding vitamins A and E, we agree; they are noted in our book as nutrients that should generally not be supplemented, or should be supplemented in low doses from natural sources. For example, we recommend eating a quarter pound of liver per week for vitamin A and other nutrients.

Regarding vitamin C, that it may not prevent colds or cancer does not mean it has no benefits. A few:

Our book has more evidence for benefits of vitamin C supplementation. Many people notice improved skin, hair, nails, gums, and teeth when supplementing with vitamin C, and faster wound healing. Vitamin C needs rise dramatically in sickness and stress.

It would be easy enough to compile further evidence of benefits of supplementation. Indeed, it was not so long ago that pellagra was rampant in the US South, and beriberi in East Asia. Many foods are subject to mandatory micronutrient fortification – a form of supplementation – to prevent iodine, folate, niacin, and thiamin deficiencies. So at least one branch of the government is convinced some supplementation is desirable.

So Offit’s case basically comes back to the Mursu et al paper. [1] Let’s revisit that one.

The Iowa Women’s Health Study

The study followed a large number of women in Iowa, and queried them several times about supplement use. In 1986, the baseline, the women had an average age of 62 (range of 55 to 69) and 66% were taking supplements. By 2004, the surviving women had an average age of 82 and 85% were taking supplements.

Here is the data on overall mortality vs supplement use:

“Cases” are instances of someone dying. “HR” or hazard ratio is the likelihood of dying if you supplement divided by the likelihood of dying if you don’t. Note that all the hazard ratio estimates are “adjusted.”

We can calculate the raw data from the Users and Nonusers columns. In general, supplements had no obvious effect – certainly no statistically significant effect. The fraction of Users and Nonusers who died was essentially identical. If we eliminate copper which only had 229 supplementers, the hazard ratio of supplementers averaged 99.8% that of non-supplementers – i.e., supplementers were very slightly less likely to die.

But – and this is a key point – supplement use increased with age throughout the study. Roughly, 66% of 62 year olds took supplements and 82% of 82 year olds took supplements. But mortality at age 82 is about five times higher than mortality at age 62. So the high-mortality 82 year olds were supplementing more than the low-mortality 62 year olds, but supplementers had the same mortality as non-supplementers! This indicates that with age adjustment, supplementation would have shown a clear benefit.

Did Mursu et al offer an age-adjusted analysis? No, they did not. The next column in the table is age-and-energy-adjusted. “Energy” means calories of daily food intake. But the purpose of eating is to supply our body with nutrients, and supplementing nutrients reduces appetite and energy intake. (This is discussed in Chapter 17 of our book.) Lower energy intake is associated with better health, largely because a high proportion of the elderly are diabetic: 27% of those over age 65 or older are diabetic, and 50% are diabetic or prediabetic; diabetics and prediabetics benefit from lower energy intake. By adjusting for energy, they are removing credit from the supplements for the health improvements due to reduced energy intake.

Nevertheless, after age-and-energy adjustment, we find that supplements generally decreased mortality. Nine of the fifteen supplements decreased mortality, five increased mortality. At the 95% confidence interval, five supplements decreased mortality, only one increased mortality.

Making the Elephant Wiggle His Trunk

The mathematician John von Neumann gave us the insight we need to understand this paper’s analysis:

With four parameters I can fit an elephant, and with five I can make him wiggle his trunk.

Mursu et al used multivariable adjustments with 11 parameters and 16 parameters respectively to obtain their “results.” Using so many parameters lets the investigators generate whatever results they want.

I don’t think it’s a coincidence that both multivariable adjustments substantially increased the hazard ratio of every single one of the 15 supplements. The 11-variable adjustment increased hazard ratios by an average of 7%, the 16-variable adjustment by an average of 8.2%.

Rest assured, it would have been easy enough to find multivariable adjustments that would have decreased hazard ratios for every single one of the 15 supplements by 7 or 8 percent.

I may as well quote my earlier analysis:

I believe it verged on the unethical that the variables chosen include dangerous health conditions: diabetes, high blood pressure, and obesity. These three health conditions just happen to be conditions that are improved by supplementation.

Anyone familiar with how regression analyses work will immediately recognize the problem. The adjustment variables serve as competing explanations for changes in mortality. If supplementation decreases diabetes, high blood pressure, and obesity, and through these changes decreases mortality, the supplements will not get credit for the mortality reduction; rather the decreased diabetes, blood pressure, and obesity will get the credit.

It’s appropriate to ask: if it’s proper to include health conditions like diabetes as variables in the regression, why not include other health conditions like cancer? The likely answer: Supplementation does not generally help conditions such as cancer, so including such conditions as adjustment factors would not have made the supplements seem more harmful. Rather, by giving greater weight to diabetes and obesity – conditions supplementation benefits – it would have made supplements look more beneficial.

It is impossible to take seriously studies that provide 11- and 16-variable adjustments, with arbitrarily chosen adjustment factors and no sensitivity analysis showing how alternative choices of adjustment factors would have altered the results.

Conclusion

The great economist Ronald Coase (in his essay “How should economists choose?”) said, “If you torture the data enough, nature will always confess.”

The Mursu paper was an exercise in torturing data until it declared, “Supplements are harmful!”

The Offit piece is a polemical exercise pretending that an unsettled part of biology – our nutrient needs, and the circumstances in which food fails to meet them – is a settled subject with a simple answer.

Now, it’s quite difficult to establish the healthfulness of supplementation in general, because you can always get too much of a nutrient, nutrient needs differ among persons depending on their health and age, and whether a person will benefit from a nutrient depends on whether the rest of the diet is deficient in that nutrient. So any given supplement is going to be harmful in some circumstances, beneficial in others.

A proper scientific approach would be to try to determine the circumstances under which supplements (or dietary changes eliminating the deficiency) would be beneficial.

Offit’s piece doesn’t attempt that. Our book does, and would make a much better resource to those considering supplementation. So would the Examine supplement goals reference guide.

References

[1] Mursu J et al. Dietary supplements and mortality rate in older women: the Iowa Women’s Health Study. Arch Intern Med. 2011 Oct 10;171(18):1625-33. http://pmid.us/21987192.

Around the Web; The Case of the Killer Vitamins

I’d like to thank Patrick Timpone for a very enjoyable interview on The Morning Show at One Radio Network. Here is the MP3; I’m on for the second half of the show. You can find a zip file at the archive for October 13. Patrick’s producer Sharon tells me that she’s already benefited from our book:

I was following The Primal Diet and since I read the book, I’ve been allowing myself potatoes and rice and doing very very well on them among doing some other things you recommend.

Also, I’d like to thank Jimmy Moore once more for hosting his highly entertaining “safe starch” symposium (Jimmy’s original post; my response, here and at Jimmy’s). It was great to get the opportunity to explain ourselves to so many people in the low-carb and Paleo movements.

Jimmy is planning to try our diet for a week in November, which will be a good occasion for us to publish a 7-day meal plan. We’ll invite anyone who’s curious to try the diet along with Jimmy, and compare notes.

[1] Interesting posts this week:

Angelo Coppola on Latest in Paleo wonders if Denmark’s saturated fat tax will apply to mother’s milk. If so, it’s bad news for unemployed infants! (He also discusses the “safe starch” debate.)

I once knew a French astronomer who died from snorting cocaine while observing at 14,500 feet. Emily Deans makes me wonder:  Did he have Crisco for dinner?

Stan the Heretic offers his mitochondrial dysfunction theory of diabetes. Peter Dobromylskyj and JS Stanton are also developing ideas along this line. Speaking of JS, his post this week has some great photos of Sierra wildflowers and reflections on the state of the Paleo community.

CarbSane partially confirms Dr. Ron Rosedale: eating carbs does raise leptin levels compared to eating fat, but it is a mild rise over an extended period of time, not a “spike.”

Beth Mazur explains why her bathroom door is always closed.

Chris Kresser discusses why chronic illness often generates a form of hypothyroidism, low T3 syndrome.

Joshua Newman knows how to flatter.

How solid is the case against Andrew Wakefield? Autism is certainly characterized by intestinal dysfunction, and Age of Autism notes that distinguished scientists are citing Wakefield’s work.

Richard Nikoley claims he doesn’t know the words to “Kumbayah.”

Seth Roberts points out that the Specific Carbohydrate Diet has been curing Crohn’s for 80 years, but still no clinical trial.

Jamie Scott, That Paleo Horse Doctor, asks: Why do horses get laminitis?

We’ve quoted vegetarian Dr. Michael Greger’s concerns about arsenic in eggs. I’m more concerned about soy protein in eggs.

Following Steve Jobs’s death, Tim asked for an opinion about the unconventional cancer therapies of Dr Mercola’s friend Nicholas Gonzalez. David Gorski, toward the end of a detailed examination of Jobs’s medical condition and treatment, links to his own claim that the Gonzalez protocol is “worse than useless.”

[2] Music to read by:

[3] Cute animal photo:

[4] Notable comments this week:

PeterC’s dad, who has diabetes, is doing well on our diet. Daniel’s stepdad had a similar experience.

Helen informs us that sweet potato intolerance may be due to raffinose.

Mario Iwakura gives us his infectious theory of diabetes. I think a lot of the cases of disrupted glucose regulation, where people get frequent hyperglycemic and hypoglycemic episodes, may be due to occult infections.

Dr Jacquie Kidd (who blogs at drjacs.com) has gotten some great advice from Jamie Scott.

Ellen tells us of cases of iodine supplementation controlling diabetes.

Ned is looking for grass-fed cowbells.

[5] Do Vitamins Kill?: An analysis of the Iowa Women’s Health Study came out this week, and it purported to show that nearly all supplements except calcium and vitamin D increased mortality, with iron being the worst. Oskar asked us to look into it, so we did.

The study followed a large number of women in Iowa, and queried them several times about supplement use. In 1986, the baseline, the women had an average age of 62 (range of 55 to 69) and 66% were taking supplements. By 2004, the surviving women had an average age of 82 and 85% were taking supplements.

Here is the data on overall mortality vs supplement use:

“Cases” are instances of someone dying. “HR” or hazard ratio is the likelihood of dying if you supplement divided by the likelihood of dying if you don’t. Note that all the hazard ratio estimates are “adjusted.”

Unadjusted Hazard Ratios

The left columns of the table give us death statistics and allow us to calculate raw hazard ratios, with no adjustment whatsoever. Seven of the supplements have unadjusted HRs below 1.00, eight have unadjusted HRs above 1.00. The 15 HRs average to 1.01. Without copper, which has an unadjusted HR of 1.17, they average to 0.998. In short, death rates among supplementers were almost identical to death rates among non-supplementers.

This is interesting because supplement usage rose rapidly with age. It was 66% at age 62 and 85% at age 82. Supplement users were, on average, older than non-supplement users. But mortality rises rapidly with age. So there should have been a lot more deaths among the supplement users, just because of their more advanced age.

The paper should have, but didn’t, report age-adjusted hazard ratios. Adjusting for age is very important, since mortality depends strongly on age, and so does supplement use. However, it’s obvious what the result of age-only adjustment would have been. Supplement usage would have shown a substantial reduction in the risk of dying.

Hazard Ratios Adjusted for Age and Energy Intake

The least-adjusted hazard ratios reported in the paper are adjusted for age and energy intake.

The energy intake adjustment is disappointing, because energy intake is affected by health: healthier people are more active and eat more, and obese people also eat more. Including indices of health as independent variables in a regression analysis will tend to mask the impact of the supplements on health, creating misleading results.

However, let’s go with what we have. Based on “Age and Energy Adjusted” hazard ratios, supplements generally decrease mortality. Nine of the fifteen supplements decreased mortality, five increased mortality. At the 95% confidence interval, five supplements decreased mortality, only one increased mortality.

Looking at the specific supplements, results are mostly consistent with our book analysis. Let’s start with the five that showed harm:

  • Folic acid and iron – two nutrients we regard as dangerous and recommend not supplementing – both elevate mortality, as we would expect. Iron is particularly harmful, and should generally be avoided by women once they have stopped menstruating.
  • Multivitamins slightly increase mortality, a result that has been found before and that we acknowledge in the book. This is probably due to (a) an excess of folic acid, (b) an excess of iron (if the women are taking iron-containing multis after menopause), (c) an excess of vitamin A (this is no longer the case – multi manufacturers have reduced the A content of vitamins in response to data – but in 1986-2004 most multis contained substantial amounts of A) which is harmful in women with vitamin D and/or K2 deficiencies (both extremely common, and D deficiency in this cohort is supported by the benefits of D and calcium in the study and the northerly latitude of Iowa) or (d) imbalances in other nutrients; for reasons of bulk multis tend to lack certain minerals, notably magnesium and calcium.
  • Vitamin B6 is an anomaly, as we wouldn’t expect B6 to be harmful in moderation. I’m guessing B6 would have been taken to reduce high homocysteine and for this purpose would often have been taken along with folic acid, a harmful supplement. Also, B6 should be balanced by vitamin B12 and biotin, and may not have been. Perhaps people with cancer were unaware that B6 promotes tumor growth; (UPDATE: See comments; I was misremembering studies, B12 and folic acid can promote tumor growth, but in other studies B6 looks protective against cancer) indeed, in the breakdown by cause of death in Table 3, B6 increases cancer mortality by 6%, but CVD mortality by only 1%. (Folic acid and vitamin A were other cancer-promoting supplements.) The harm from B6 was not statistically significant and I wouldn’t read much into it.
  • Copper is another anomalous result, but this was the least popular supplement, taken by only 229 women or 0.59%. Copper’s hazard ratios were dramatically affected by adjustment: in the raw data, mortality is only 17% higher among copper supplementers, but after age and energy adjustment it is 31% higher, and multivariable adjustment increases it substantially again. Clearly the effect of copper is highly sensitive to adjustment factors, indicating that copper was being taken by an unusual population. I think the hazard ratio for copper is impossible to interpret without knowing why these women were supplementing copper. If we knew their situation, there would probably be an appropriate adjustment that would make a huge difference in mortality. I would say the numbers are too small, the population too skewed, and the information too limited to draw any conclusion here.

Overall, I would interpret the nine that showed benefits as being highly supportive of micronutrient supplementation. The fact that vitamin A, vitamin B complex, vitamin C, vitamin D, vitamin E, calcium, magnesium, selenium, and zinc all reduced mortality suggests that a well-formulated multivitamin would likely have reduced mortality.

Hazard Ratios After Multivariable Adjustment

Now, what about the “Multivariable Adjusted” results, which were responsible for the headlines?

We have to keep in mind a famous aphorism from the mathematician John von Neumann:

With four parameters I can fit an elephant, and with five I can make him wiggle his trunk.

The multivariable adjustments use 11 parameters and 16 parameters respectively. Using so many parameters lets the investigators generate whatever results they want.

I don’t think it’s a coincidence that both multivariable adjustments substantially increased the hazard ratio of every single one of the 15 supplements. The 11-variable adjustment increased hazard ratios by an average of 7%, the 16-variable adjustment by an average of 8.2%.

Rest assured, it would have been easy enough to find multivariable adjustments that would have decreased hazard ratios for every single one of the 15 supplements.

I believe it verges on the unethical that the variables chosen include dangerous health conditions: diabetes, high blood pressure, and obesity. These three health conditions just happen to be conditions that are often improved by supplementation.

Anyone familiar with how regression analyses work will immediately recognize the problem. The adjustment variables serve as competing explanations for changes in mortality. If supplementation decreases diabetes, high blood pressure, and obesity, and through these changes decreases mortality, the supplements will not get credit for the mortality reduction; rather the decreased diabetes, blood pressure, and obesity will get the credit.

Imagine we had a magic pill that completely eliminated diabetes, obesity, and high blood pressure, and reduced mortality by 20%, with no negative health effects under any circumstances. But if regression analysis showed that non-diabetic, non-obese, and non-hypertensive people had 25% less mortality, then a multivariable adjusted analysis would show that the magic pill increased mortality. Why? Because the elimination of diabetes, obesity, and hypertension should have decreased mortality by 25% (the regression analysis predicts), but mortality was only decreased 20%, so adjusted for diabetes, obesity, and hypertension the magic pill must be credited with the additional 5% dead. The multivariable adjusted HR for the magic pill becomes 0.8/0.75 = 1.067.

Of course, what ordinary people want to know is: Will this magic pill improve my health? The answer to that would be yes.

What (too many) scientists want to know is: Which methodology for analyzing this magic pill data will get me grant money? That depends on whether the funding authorities are positively or negatively disposed toward the magic pill industry. Once you know that, you search for the 16-variable multivariable regression that generates the hazard ratios the authorities would like to see.

My take? Judging by the data in Table 2 plus corroborating evidence from clinical trials reviewed in our book, I would say that a well-formulated supplement program, begun at age 62, may increase the odds of survival to age 82 by something on the order of 5% to 10%. Perhaps not a magic pill; but worthwhile.

[6] Not the weekly video: An exceptional magic show:

[7] Shou-Ching’s Photo Art:

[8] Weekly video: A new tool for stroke recovery:

Answer Day: What Causes High LDL on Low-Carb Paleo?

First, thank you to everyone who commented on the quiz. I enjoyed reading your thoughts.

Is High LDL Something to Worry About?

Perhaps this ought to be the first question. Jack Kronk says “I don’t believe that high LDL is necessarily a problem” and Poisonguy writes “Treat the symptoms, Larry, not the numbers.” Poisonguy’s comment assumes that the LDL number is not a symptom of trouble. Is it?

I think so. It helps to know a little about the biology of cholesterol and of blood vessels.

When cells in culture plates are separated from their neighbors and need to move, they make a lot of cholesterol and transport it to their membranes. When cells find good neighbors and settle down, they stop producing cholesterol.

The same thing happens in the body. Any time there is a wound or injury that needs to be healed, cholesterol production gets jacked up.

When people have widespread vascular injuries, cholesterol is produced in large quantities by cells lining blood vessels. Now, to repair injuries cells have to coordinate their functions. Endothelial cells are the coordinators of vascular repair: they direct other cell types, like smooth muscle cells and fibroblasts, in the healing of vascular injuries.

To heal vascular injuries, these cells not only need more cholesterol for movement; they also need to multiply. It turns out that LDL, which carries cholesterol, also causes vascular cells to reproduce (“mitogenesis”):

The best-characterized function of LDLs is to deliver cholesterol to cells. They may, however, have functions in addition to transporting cholesterol. For example, they seem to produce a mitogenic effect on endothelial cells, smooth muscle cells, and fibroblasts, and induce growth-factor production, chemotaxis, cell proliferation, and cytotoxicity (3). Moreover, an increase of LDL plasma concentration, which is observed during the development of atherosclerosis, can activate various mitogen-activated protein kinase (MAPK) pathways …

We also show … LDL-induced fibroblast spreading … [1]

If endothelial cells are the coordinators of vascular repair, and LDL particles their messengers to fibroblasts and smooth muscle cells, then ECs should be able to generate LDL particles locally. Guess what:  ECs make a lipase whose main effect is to decrease HDL levels but can also convert VLDL and IDL particles into LDL particles and remove fat from LDL particles to make them into small, dense LDL:

Endothelial lipase (EL) has recently been identified as a new member of the triglyceride lipase gene family. EL shares a relatively high degree of homology with lipoprotein lipase and hepatic lipase …

In vitro, EL has hydrolyzed phospholipids in chylomicrons, very low density lipoprotein (VLDL), intermediate density lipoprotein and LDL. [2]

Immune cells, of course, are essential for wound healing and they should be attracted to any site of vascular injury. It turns out that immune cells have LDL receptors and these receptors may help them congregate at sites of vascular injury. [3]

I don’t want to exaggerate the state of the literature here:  this is a surprisingly poorly investigated area. But I believe these things:

1.      Cholesterol and LDL particles are part of the vascular wound repair process.

2.      Very high LDL levels are a marker of widespread vascular injury.

Now this is not the “lipid hypothesis.” Compare the two views:

  • The lipid hypothesis:  LDL cholesterol causes vascular injury.
  • My view:  LDL cholesterol is the ambulance crew that arrives at the scene of the crime to help the victims. The lipid hypothesis is the view that ambulance drivers should be arrested for homicide because they are commonly found at murder scenes.

So, to Poisonguy, on my view high LDL numbers are a symptom of vascular injury and are a cause for concern.

Big-Picture View of the Cause of High LDL

So, on a micro-level, I think vascular damage causes high LDL. But what causes vascular damage?

Here I notice a striking difference in commenters’ perspectives and mine. I tend to take a big-picture, top-down view of biology. There are three basic causes of nearly all pathologies:

1.      Toxins, usually food toxins.

2.      Malnutrition.

3.      Pathogens.

The whole organization of our book is dictated by this view. It is organized in four Steps. Step One is about re-orienting people’s views of macronutrients away from high-grain, fat-phobic, vegetable-oil-rich diets toward diets rich in animal fats. The other steps are about removing the causes of disease:

1.      Step Two is “Eat Paleo, Not Toxic” – remove food toxins.

2.      Step Three is “Be Well Nourished” – eliminate malnutrition.

3.      Step Four is “Heal and Prevent Disease” – address pathogens by enhancing immunity and, where appropriate, taking advantage of antibiotic therapies.

So when someone offers a pathology, any pathology, my first question is: Which cause is behind this, and which step do they need to focus on?

In Larry’s case, he had been eating low-carb Paleo for years. So toxins were not a problem.

Pathogens might be a problem – after all, he’s 64, and everybody collects chronic infections which tend to grow increasingly severe with age – but Larry hadn’t reported any other symptoms. More to the point, low-carb Paleo diets typically enhance immunity, yet Larry had fine LDL numbers before adopting low-carb Paleo and then his LDL got worse. So it wouldn’t be infectious in origin unless his diet had suppressed immunity through malnutrition – in which case the first step would be to address the malnutrition.

Step Three, malnutrition, was the only logical answer. The conversion to Paleo removes a lot of foods from the diet and could easily have removed the primary sources of some micronutrients.

So I was immediately convinced, just from the time-course of the pathology, that the cause was malnutrition.

Micronutrient Deficiencies are Very Common

In the book (Step Three) we explain why nearly everyone is deficient in micronutrients. The problems are most severe for minerals:  water treatment removes minerals from water, and mineral depletion of soil by industrial agriculture leads to mineral deficiencies in farmed plants and grain-fed animals.

This is why our “essential supplements” include a multimineral supplement plus additional quantities of five minerals – magnesium, copper, chromium, iodine, and selenium. Vitamins get a lot of attention, but minerals are where the big health gains are.

Copper Deficiency and LDL

Some micronutrient deficiencies are known to cause elevated LDL.

Readers of our book know that copper causes vascular disease; blog readers may be more familiar with an excellent post by Stephan, “Copper and Cardiovascular Disease”, discussing evidence that copper deficiency causes cardiovascular disease. As I’ve just argued that cardiovascular disease causes high LDL, it shouldn’t be a surprise that copper deficiency also causes hypercholesterolemia:

Copper and iron are essential nutrients in human physiology as their importance is linked to their role as cofactors of many redox enzymes involved in a wide range of biological processes, as well as in oxygen and electron transport. Mild dietary deficiencies of both metals … may cause long-term deleterious effects in cardiovascular system and alterations in lipid metabolism (3)….

Several studies showed a clear correlation among copper deficiency and dyslipidemia. The main alterations concern higher plasma CL and triglyceride (TG) concentrations, increased VLDL-LDL to HDL lipoproteins ratio, and the shape alteration of HDL lipoproteins.  [4]

The essentiality of copper (Cu) in humans is demonstrated by various clinical features associated with deficiency, such as anaemia, hypercholesterolaemia and bone malformations. [5]

Over the last couple of decades, dietary copper deficiency has been shown to cause a variety of metabolic changes, including hypercholesterolemia, hypertriglyceridemia, hypertension, and glucose intolerance. [6]

Copper deficiency is, I believe, the single most likely cause of elevated LDL on low-carb Paleo diets. The solution is to eat beef liver or supplement.

So, was my advice to Larry to supplement copper?  Yes, but that was not my only advice.

Other Micronutrient Deficiencies and Elevated LDL

Another common micronutrient deficiency that causes elevated LDL cholesterol is choline deficiency that is NOT accompanied by methionine deficiency. That is discussed in my post “Choline Deficiency and Plant Oil Induced Diabetes”:

Choline deficiency (CD) by itself induces metabolic syndrome (indicated by insulin resistance and elevated serum triglycerides and cholesterol) and obesity.

A combined methionine and choline deficiency (MCD) actually causes weight loss and reduces serum triglycerides and cholesterol …

I quote both these effects because it illustrates the complexity of nutrition. A deficiency of a micronutrient can present with totally different symptoms depending on the status of other micronutrients.

Julianne had a really nice comment, unfortunately caught in the spam filter for a while, with a number of links. She mentions vitamin C deficiency and, with other commenters, noted the link between hypothyroidism and elevated LDL. As one cause of hypothyroidism is iodine or selenium deficiency, this is another pathway by which mineral deficiencies can elevate LDL.

UPDATE: Mike Gruber reduced his LDL by 200 mg/dl by supplementing iodine. Clearly iodine can have big effects!

Other commenters brought up fish oil. They may be interested to know that fish oil not only balances omega-6 to modulate inflammatory pathways, it also suppresses endothelial lipase and thus moderates the LDL-raising and HDL-lowering effect of vascular damage:

On the other hand, physical exercise and fish oil (a rich source of eicosapentaenoic acid and docosahexaenoic acid) suppress the activity of EL and this, in turn, enhances the plasma concentrations of HDL cholesterol. [7]

Whether this effect is always desirable is a topic for another day.

My December Advice to Larry

So what was my December advice to Larry?

It was simple. In adopting a low-carb Paleo diet, he had implemented Steps One and Two of our book. My advice was to implement Step Three (“Be well nourished”) by taking our recommended supplements. Eating egg yolks and beef liver for copper and choline is a good idea too.

Just to cover all bases, I advised to include most of our “therapeutic supplements” as well as all the “essential supplements.”

Since December, Larry has been taking all the recommended supplements and eating 5 ounces per week of beef liver. As I noted yesterday, Larry’s LDL decreased from 295 mg/dl to 213 mg/dl, HDL rose from 74 mg/dl to 92 mg/dl, and triglycerides fell from 102 to 76 mg/dl since he started Step Three. This is all consistent with a healthier vasculature and reduced production of endothelial lipase.

Conclusion

Some people think there is something wrong with a diet if supplements are recommended. They believe that a well-designed diet should provide sufficient nutrition from food alone, and that if supplements are advised then the diet must be flawed.

I think this is quite mistaken. The reality is that Paleolithic man was often mildly malnourished, and modern man – due to the absence of minerals from treated water and agriculturally produced food, and the reduced diversity and higher caloric density of our foods – is severely malnourished compared to Paleolithic man.

We recommend eating a micronutrient-rich diet, including nourishing foods like egg yolks, liver, bone broth soups, seaweed, fermented vegetables, and so forth. But I think it’s only prudent to acknowledge and compensate for the widespread nutrient depletion that is so prevalent today. Even when nutrient-rich food is regularly eaten, micronutrient deficiencies are still possible.

Eating Paleo-style is not enough to guarantee perfect health. Luckily, supplementation of the key nutrients that we need for health and that are often missing from foods will often get us the rest of the way.

References

[1] Dobreva I et al. LDLs induce fibroblast spreading independently of the LDL receptor via activation of the p38 MAPK pathway. J Lipid Res. 2003 Dec;44(12):2382-90. http://pmid.us/12951358.

[2] Paradis ME, Lamarche B. Endothelial lipase: its role in cardiovascular disease. Can J Cardiol. 2006 Feb;22 Suppl B:31B-34B. http://pmid.us/16498510.

[3] Giulian D et al. The role of mononuclear phagocytes in wound healing after traumatic injury to adult mammalian brain. J Neurosci. 1989 Dec;9(12):4416-29. http://pmid.us/2480402.

[4] Tosco A et al. Molecular bases of copper and iron deficiency-associated dyslipidemia: a microarray analysis of the rat intestinal transcriptome. Genes Nutr. 2010 Mar;5(1):1-8. http://pmid.us/19821111.

[5] Harvey LJ, McArdle HJ. Biomarkers of copper status: a brief update. Br J Nutr. 2008 Jun;99 Suppl 3:S10-3. http://pmid.us/18598583.

[6] Aliabadi H. A deleterious interaction between copper deficiency and sugar ingestion may be the missing link in heart disease. Med Hypotheses. 2008;70(6):1163-6. http://pmid.us/18178013.

[7] Das UN. Long-chain polyunsaturated fatty acids, endothelial lipase and atherosclerosis. Prostaglandins Leukot Essent Fatty Acids. 2005 Mar;72(3):173-9. http://pmid.us/15664301.

It’s easy to be malnourished

I recently came across a paper analyzing the micronutrient deficiencies in four popular diets – the South Beach Diet, the Atkins for Life Diet, the DASH diet, and the Best Life Diet.

According to this paper [1], none of these diets provides the RDA for more than 15 of the 27 micronutrients studied.  Some nutrients were lacking in all four diets. The diets averaged:

  • 58% of the RDA for pantothenic acid (vitamin B5).
  • 29% of the RDA for biotin.
  • 34% of the RDA for vitamin E.
  • 56% of the RDA for choline.
  • 9% of the RDA for chromium.
  • 34% of the RDA for iodine.
  • 73% of the RDA for potassium.

Of these I would take the chromium and iodine deficiencies very seriously.  These would certainly merit supplementation – indeed, in our book they are among the 8 micronutrients we strongly recommend supplementing.

Cases can also be made for supplementation of others on this list:

  • Biotin and pantothenic acid are harmless in very high doses, and extremely cheap – less than 4 cents for 500 mg pantothenic acid or 5 mg biotin.
  • Choline deficiency is widespread and there is evidence that choline supplements are very helpful for pregnant women.

More important than any specific deficiency, however, is the implication for ordinary diets.  If these comparatively healthy diets are deficient, then nearly any modern diet is likely to produce micronutrient deficiencies.

Agriculture – planting the same crops, year after year, in the same fields – tends to deplete the soil of nutrients, and hence both plant crops and crop-fed animals tend to be low in nutrition. Water treatment also removes minerals like calcium and magnesium from drinking water. 

A British study, for instance, found that copper levels in UK foods have declined by 76% in vegetables, 90% in dairy foods, and 55% in meat. [2] Some other nutrients have declined nearly as much.

It’s important to take some care, therefore, to eat a nourishing diet.  Especially nourishing foods include seaweed, green leafy vegetables, organ meats including liver, and seafoods.

Cooking style is also important:  Cooking should be done in a way that doesn’t throw away drippings from foods, but rather preserves them as a sauce or soup.  Also, very high cooking temperatures which can destroy or denature nutrients should be avoided. 

Even with these steps, supplementation is probably necessary for optimal health.  We recommend taking a daily multivitamin plus a few other supplements, notably including iodine, magnesium, vitamin D, and vitamin K2.

[1] Calton JB. Prevalence of micronutrient deficiency in popular diet plans. J Int Soc Sports Nutr. 2010 Jun 10;7(1):24. http://pmid.us/20537171. Full text:  http://www.jissn.com/content/pdf/1550-2783-7-24.pdf.

[2] Thomas D. The mineral depletion of foods available to us as a nation (1940-2002)–a review of the 6th Edition of McCance and Widdowson. Nutr Health. 2007;19(1-2):21-55. http://pmid.us/18309763. Hat tip Robert Andrew Brown, http://wholehealthsource.blogspot.com/2010/04/copper-and-cardiovascular-disease.html.