Now PURE is this enormous study over five continents, 18 different countries, over 135,000 different individuals, that has had some pretty profound research evidence as it comes to saturated fats, cholesterol fat in general and their implication on overall mortality. It has a number of data on salt intake and mortality. And a lot of it is contrary to conventional wisdom and guidelines.
Now all that being said, this is an epidemiological study and we definitely talk about the strengths and benefits of epidemiology versus randomized controlled trial and he is a good perspective on how we really need both to further research and affect policy. So there's a lot of data here going over the PURE study and it has some pretty profound impacts on the way we should make recommendations and how we should see older recommendations and their falling.
So I hope you enjoy this interview with Prof. Andrew Mente and learn a lot about the PURE study and understand how we can use that data in our daily lives. Prof. Andrew Mente thank you so much for joining me on the DietDoctor podcast.
That's important because that allows us to assess shapes of relationships between dietary variables and health outcomes. Which has never been characterized before with any high degree of statistical precision.
Of course with a large randomized controlled trials we can better assess causative effects. The problem with large randomized trials is they're very difficult to conduct with diet and it's very difficult for people to sustain a particular diet for a long term. And so there's a challenge there.
On the other hand when you have weak effects, it's harder to assess weak effects in observational studies, because you don't know if the result is true or due to residual confounding. So we tend to think of different designs as complementing one another.
So not one design being the best considering feasibility and also, you know, what is the cleanest design. But using different designs complementary to one another and capitalizing on the strengths of each is the ideal way to go forward.
So our data suggest that the conventional way of thinking of diet focusing on higher intake of carbohydrates may actually backfire, which supports what has actually happened. And so higher carb intake, and remember many parts of the world consume very high amounts of carbohydrates, low and middle income countries, and largely it's refined carbohydrates and added sugar. And we find that higher carb is related to more cardiovascular events and mortality particularly all-cause mortality, whereas for fats we see the opposite.
We see higher fat intake related to lower risk of mortality and saturated fat related to lower risk of stroke. So this kind of challenges conventional wisdom on diet, but it is consistent with the trials, because you look at the randomized trials that replaced saturated fat with polyunsaturated fat, that haven't really panned out. Largely neutral effects. And other observational studies too have shown neutrality looking at relationship with saturated fat and clinical outcomes. So our findings, if anything, are supportive of previous studies.
The red meat leading to colon cancer at 1.17 is a small hazard ratio. So the hazard ratio here was small at 1.17 and 1.28. So how do you help us interpret that in terms of it was a fact because of how many patients there were, but yet the hazard ratio was small and it's sort of against what the guidelines say. So how do you incorporate all that into how we should interpret that data?
So that's a challenge with diet, studying diet in observational studies, but if anything, you look at the data of other cohort studies and if you focus on the studies that looked at carbohydrates versus mortality as a percent of energy, you also there see that higher carb intake shows an increase in the risk of mortality.
Now some studies have looked at calculated diet scores or carbohydrates scores and so what goes into that is different coating of our carbohydrate foods. So you can select almost any food you want to go into a carbohydrate score and you'll get different results, but the studies that looked at percent of energy from carbs, you see a positive association with mortality.
Now there aren't that many free living populations with very low carbohydrate intake. So don't misread me, I'm not saying that going as low as possible would be beneficial because that has yet to be demonstrated but certainly appears that there is an optimal range between 50% to 55% of energy from carbs that appears to be associated with lowest risk. At the low-end it's a little more murky, we really don't know.
So not a surprise that a higher level of carbohydrate increased mortality risk. Now what maybe was a surprise that the higher level of fat intake decreased mortality risk, I think that's where the real headline is, that is so counter to what we're being told. And now you broke that up into monounsaturated fats, polyunsaturated fats and saturated fats in terms of their mortality risk; so tell us how those varied.
Now what this suggests is that when you go to low levels below 10% and further do you actually see the increase in mortality? Which is actually what the guidelines recommend; to go to those lower-levels. Now we are not saying that our data supports consuming 20% or 25% of energy from saturated fat, only because that's not captured by the natural distribution of saturated fat in free living populations.
And certainly some societies you see consumed 3 to 4 decades ago much higher amount of saturated fat. So our data is not capturing that high level of saturated fat, but up to about 13% or 14% of energy we see a lower risk of mortality compared to people consuming lower amounts of saturated fat.
The Women's Health Initiative trial which compared the low-fat diet to a higher fat diet again found no significant change in risk for cardiovascular events and mortality. So that was another large study, cost half $1 billion. So if anything our results are consistent with that.
Now if you look at cardiovascular disease mortality and non-cardiovascular mortality, directionally we did see that different types of fats were beneficial, although it wasn't statistically significant but directionally. And in carbs directionally was harmful versus cardiovascular death and non-cardiovascular death. It was the nonfatal events were as large and neutral.
We don't have enough event rates right now to characterize cancer or respiratory events alone or individual types of cancer. But as the cohort gets older the event rates will pick up and we'll have more events. So that's why it's very important in PURE to do the follow-up over the next 10 years than we can assess individual types of cancer and diet.
What we're consuming else, fine, we could even consume a little bit more. We're not saying consume unlimited amounts, we need data for that still, but what we consume now appears to be right and we don't have to put in stringent cutoffs to get people to lower their saturated fat.
Now the downside with these questionnaires of course is random measurement error. And so that adds noise but that more dilutes associations toward the null, and that is a factor in every epidemiological study. So it is the best tool we have at the moment for large epidemiological studies and that's what we use.
Again that's why I say complementing randomized trials, focusing on risk markers would be optimal. So the major strength however is the fact that we cover again a broad range of intake across different parts of the world, again characterizing those extreme ranges as much as they are represented by human consumption and that's really where the advantage of PURE is.
So then you looked at outcome data in terms of what these markers all met. And what did you find in terms of the difference between LDL cholesterol, the ApoB to ApoA...? Share that data with us.
And when you look at ApoB to ApoA, which remember in INTERHEART and INTERSTROKE, two large international studies, was the strongest lipid predictor of heart attacks and stroke, we found that the ratio goes down with higher saturated fat, which again suggests a beneficial effect since that is the strongest risk marker and that goes down with higher saturated fat.
And then what we did is we modeled... we said okay assuming we don't have any data on clinical events, let's model and use the lipid markers to project what the effects of diet on cardiovascular risk would be. And then we did that, we modeled using LDL and we found a positive association as you would expect.
After all saturated fat is positively associated with LDL. But then when we map that versus actual events, we found that LDL was a poor predictive marker of future events when you look at the observed associations. On the other hand ApoB to ApoA ratio was much better at projecting the effects of diet on health outcomes.
So this suggests that if we focus on LDL we may be largely misinforming diet for populations. ApoB to ApoA ratio, which is a measure of small dense LDL particles that are more atherogenic than LDL appears to be the much better predictive marker to project the effects of diet on health outcomes.
Whereas with LDL they diverged in opposite directions. So the projected estimates showed an increase in risk, whereas the actual effects of saturated fat on the events go down slightly. So they diverged in different directions. It would suggest that LDL is not very good for projecting dietary effects. It may be very good for projecting statin effects on health outcomes, but not for diet.
So my hope is that this study would cause just a huge snowball effect of people realizing that LDL-C is not the marker we should be following. Yet I don't feel like I've heard enough about that in the media and in the scientific circles. Is that just because old dogma dies hard and people aren't ready to hear it? Why do you think that's the case?
They contain zinc and magnesium. So this is all thrown out and we treat food almost like it's a single nutrient saturated fat that's infused into our veins. And that's used to project the effects and really is an absurd way of thinking if you really think about it deeply. So for diet we have to think much more multidimensionally than that.
So I think that's why this evidence is so powerful and we need to be seeing this from the rooftops more to say we need to reevaluate dietary changes and their effects on cholesterol. And quick, to point that out... It may not be the same for drugs, it may not be the same for genetics, but for dietary changes that's what we need to look at.
So it's very important to look at that to see what the effect on the risk markers here are for very low-carb intake. Which PURE does not capture, because it's largely representing parts of the world that consume from moderate to high carb. So that's why Sarah's work is very important.
And I think it's a sign of somebody maybe being a little more health conscious, because that's what we're told as a healthy way of eating, but was there any parsing out individually of how vegetables are different than fruits and different than legumes individually?
Because interesting, with increased fruit intake, if somebody was diabetic or had metabolic disease, you would think that would have a deleterious effect, but over the whole sample, fruit intake was beneficial.
But because sodium is an essential nutrient it doesn't quite work out that way. So we agree that at high levels you get toxicity and increase in blood pressure, but at low levels you get deficiency. And so what that does is it activates certain mechanisms that are built into our bodies since salt is an essential nutrient. So you get renin angiotensin system activation at low levels.
And this has been shown repeatedly in intervention trials. And so you have dual competing mechanisms, which is consistent with an essential nutrient. Toxicity at high levels, deficiency at low levels, sweet spot in the middle. And our findings reaffirm that and other studies as well reaffirm that.
There is not one single study ever that has shown that low-sodium at currently recommended levels is better than average sodium, that sweet spot of 3 to 5 g per day, versus cardiovascular events and mortality. High levels above 5 g per day, certainly, we should get those populations down to moderate levels, but there's absolutely no evidence to support low levels versus moderate levels and yet that's what we currently recommend again based on an assumed benefit, looking at blood pressure.
Because there are a number of limitations we have to point out. One is that we have to remember that this was largely a salt sensitive group of people, a lot of hypertensives and pre-hypertensives, and we also need to remember that potassium intake was low at baseline.
So when you put someone on a very low potassium diet, lowering or changing their blood pressure will result in changes- changing their sodium will result in changes in blood pressure. But when you give people higher amounts of potassium, put them on an all-around healthy diet, like the DASH diet, that contains many of high potassium foods, then the effects of sodium will be largely mitigated.
So that's what DASH found. That when we consume a low potassium diet you see large changes in blood pressure, which doesn't surprise really anyone given that, but when you give them a high potassium diet then sodium becomes less important and so the important point is DASH is only 30 days. So we look at the long-term effects, we need studies with longer follow-up to look at the effects in the long-term.
So some studies like TOPP have looked at longer term follow-up. TOPP originally was designed to look at blood pressure, so people were followed up for a period of 36 months, but what TOPP found was that people initially... they never reached the 1.8 g per day target, they lowered their sodium a little bit to down to 2.5 g per day, but then by around a year they migrated back to their original sodium intake.
And so even though they followed people up over time, we don't even know what people were eating during the course of the extended follow-up. But there's every reason to believe they were not even following the low-sodium recommendation.
So we really don't have any data from randomized trials, so we have to look at the data on long-term clinical events and that's where the cohort studies come into play and there's consistency across a dozen cohort studies showing low-sodium is either associated to harm versus moderate sodium or there's no change in risk. But no study is suggesting or showing a lower risk with low-sodium compared to average intake.
The official recommendation based on your studies says you should be following a sodium intake that is going to worsen your health. Why is there not a public outcry about this? I mean that's unbelievable.
So as a supplies, the low-carb community, if someone is eating their broccoli and their cauliflower and their spinach and they're putting their Himalayan salt on it and having it with their know, chicken, meat, fish, eggs and cheese, that's a perfectly reasonable diet where you can be having the higher end of sodium and based on the DASH study you would say would have no effect. Is that a fair statement?
So we have to consider within the context of the dietary pattern. And DASH is important in that respect because it shows that salt sensitivity is not an immutable trait. You can mitigate it by eating an all-around healthy diet. And when you do that we find that salt becomes less important. So the messages just concentrate on consuming an all-around healthy diet and you don't need to be worried about individual nutrients like salt and saturated fat.
But at the higher end if you didn't have hypertension then that risk was mitigated, the risk did not go up as much. So would that suggest there may not be much of an upper limit if you don't already have hypertension?
So this would suggest that rather than a population wide strategy, best we target people with hypertension who also consume high amounts of sodium exceeding 5 g per day and get them down to moderate levels. At the low-end, what's interesting, we see an increased risk, as you said, irrespective of blood pressure.
So whether you have high blood pressure or normal blood pressure, you still see the increased risk at the low-end versus clinical events, cardiovascular disease and mortality. And what that suggests is another mechanisms that play here. And again consistent with other data, showing activation of the renin angiotensin system, which we know is vascular damaging.
And you get exponential rise in these hormones with low levels of sodium and therefore you see the consistent results across different subpopulations. It's been shown repeatedly in people with hypertension and without hypertension, people with diabetes and without diabetes and people with vascular disease and without vascular disease. It's a consistent finding.
And looking at heart failure patients there are some trials that are ongoing right now looking at low-sodium versus average sodium in heart failure patients so we will have to see what the results are for that.
And then at what levels you break that down, at what level of renin angiotensin activation because most of these people are on ACE inhibitors or ARBs, which are really angiotensin blockers, there's definitely a lot of other factors to incorporate for the heart failure patients.
So this is what we are going to publish in the next year or two and also we'll do more dietary assessments during follow-up and that also helps improve the precision and accuracy of the estimates of diet and then continue the follow-up as much as we can to look at effects on less studied outcomes like cancers and respiratory events and infectious disease as well.
Transcript pdf
