Dr. Andrew Mente – Diet and cardiovascular disease: Messages from the PURE study

Dr. Andrew Mente – Diet and cardiovascular disease: Messages from the PURE study

November 4, 2019 3 By William Morgan


– [Andrew] It’s a pleasure to be here
to present on diet and
cardiovascular disease,
and some of the work that we’ve conducted
over the last five years
with the global PURE Study.
Now y’all know the diet heart hypothesis,
so I’m not gonna repeat that,
so I can spend more time on PURE,
so I’ll just jump right to it.
Last year, we published three papers
that appeared in The Lancet,
The Lancet of Diabetes & Endocrinology.
Now the first of our papers,
we looked at the association
of dietary nutrients
with blood lipids and blood pressure,
looking at 18 countries around the world.
Now this is the cross-sectional
analysis of the PURE data.
So I’ll tell you more about PURE.
PURE is a large global
prospective cohort study
from 667 communities, in 18
countries around the world,
spanning five continents.
So because the study is global,
it captures broad
patterns of diet globally,
across a broad range of intakes.
So from very low levels
to very high levels.
So, in doing that, you’re
able to characterize
the shapes of relationships
between nutrients,
or foods, and healthy outcomes,
which has never been characterized,
across a broad range,
and with high statistical
precision ever before.
Now here are the countries
that are part of PURE,
by geographic regions,
including South Asia, China,
Southeast Asia, Africa,
North America and Europe,
Middle East and South America.
This is a cross-sectional
analysis of PURE.
The participants are from
the general population,
free-living populations, in
the community in 18 countries.
135,000 people aged 35 to 70 years,
without cardiovascular
disease at baseline.
Diet was assessed using country-specific
validated food frequency questionnaires,
and we measured all known
confounders like demographics,
lifestyle, like physical
activities, smoking,
alcohol intake, health history
and study center as well.
We had data on blood
pressure in 125,000 people,
blood lipids in 104,000, and
apoB and apoA in 18,000 people,
and we used standard statistical methods
to analyze the data.
So you could see, here,
we look at the results for saturated fat
versus the blood lipids.
So the top left, you see the
results for saturated fat
versus LDL cholesterol,
and you can see across a broad range,
from less than 4% of energy
from saturated fat, to 11%,
we see a positive association,
with higher saturated fat,
higher LDL cholesterol,
which is consistent
with what we know from randomized trials,
which relatively speaks
to a modest effect.
You look at the change from 2.8 to 3.2,
across a big change in saturated fat,
consistent with the Hooper meta-analysis.
But we also see an increase
in HDL cholesterol.
When you take the ratio,
here in the bottom left,
of total cholesterol to HDL,
you see an inverse association.
We know from large epidemiological studies
that total cholesterol to HDL ratio,
is a stronger risk marker of
future cardiovascular events,
compared to LDL cholesterol alone.
So this would suggest a beneficial effect.
Then on the bottom right,
you see apoB/apoA ratio,
we see higher saturated fat
related to lower apoB/apoA.
We know from the INTERHEART
and INTERSTROKE study,
two global studies of
heart attacks and stroke,
that apoB and apoA ratio was
the strongest lipid marker
of future cardiovascular events.
Stronger than LDL
cholesterol, much stronger.
It was the strongest biomarker
that related to future events.
So this would suggest a
beneficial effect of saturated fat
when we look at apoB to apoA.
So if you’re focusing on LDL,
the advice you give to the population
is to reduce saturated fat.
That’s what the guidelines do.
But when you look at
the other risk markers,
you get a completely different story.
So what can you make out of this.
Well bottom line is really,
you need to look at
actual clinical events.
Heart attacks, stroke, death.
And for carbohydrates we see the opposite.
We see with higher carb
intake, we see lower LDL,
but we see lower HDL.
An increase in total
cholesterol to HDL ratio,
would suggest a harmful effect.
When you look at apoB/apoA,
we see an increase in the ratio
with higher carbohydrate intake.
So it’s the complete flip side
of what you get for saturated fat.
Again, we’ve been
focusing on LDL, top left,
getting people to replace fats with carbs,
traditionally in dietary advice,
but when you look at the other markers,
you get a completely different story.
There’s been a series of
meta-analysis published
over the last 10 years.
These are the most recent ones,
looking at randomized controlled trials,
where they replace saturated
fat with polyunsaturated fat.
Really a direct test of
the diet heart hypothesis.
You can see, from these meta-analyses,
generally when you crunch the data
from these randomized
trials, you get neutrality.
No significant benefit in
replacing saturated fat
with polyunsaturated fat.
Three years ago, we
published a meta-analysis
of prospective cohort studies,
looking at the relationship
of saturated fat
with coronary heart disease.
The advantage of looking
at cohort studies,
over randomized trials,
is you look at long-term dietary exposure.
Whereas randomized trials is
relatively short-term exposure.
Of course, there are also limitations with
epidemiological studies,
which I won’t get into today.
When you look at the data from these
prospective cohort studies,
looking at lowest and
highest intake group,
you see no relationship to saturated fat
and the various health outcomes,
including mortality,
coronary heart disease,
cardiovascular mortality, ischemic stroke,
and type-2 diabetes.
That brings us to the second of our papers
published last year,
this one looking at the clinical events.
Nutrients versus heart
attacks, stroke and mortality.
So now we’re looking at the
prospective cohort data in PURE,
18 countries, 135,000 people.
During the course of the follow-up,
we accrued more than 4,700
major cardiovascular events,
and more than 5,700 total deaths,
during a follow-up period of 7.4 years.
We used standard,
multi-variable Cox aggression
to analyze the data,
with Study Center as a random inter-some.
Here you see the results
for the risk of mortality,
shown on the left,
and major cardiovascular
events, on the right,
by macro-nutrient intake.
So first, on the left, you can
see with higher carb intake,
from 46% of energy to 77% of energy,
we see an increase in
the risk of mortality
with higher carb intake.
For total fat and protein,
we see the reverse.
We see higher intake as associated
with a lower risk of mortality.
On the right, we see that
each of the nutrients,
generally has a pretty neutral effect
on major cardiovascular events.
Then we looked at the various types of fat
versus mortality and major
cardiovascular events.
So on the left, looking
at total mortality,
we see saturated fat,
monounsaturated and polyunsaturated,
each are associated with
a lower risk of mortality.
On the right, looking at
major cardiovascular events,
we see that the results
are pretty neutral.
We see no significant association.
So then we sub-divided,
we looked at the types of death.
We looked at cardiovascular
death, on the left,
and non-cardiovascular
death, on the right.
So first we’re focusing
on cardiovascular death,
we see with higher carb intake,
we see increase in the risk
for cardiovascular disease,
mortality, that trended
toward significance.
But we see the reverse pattern
for total fat and protein,
where higher fat and protein intake
trended toward a lower risk
of cardiovascular death.
For non-cardiovascular
death, which in PURE,
is largely represented by
cancer and respiratory outcomes,
the most common, and there
are others out there as well,
but those are the most common,
cancer and respiratory.
You see that with higher carb intake,
you see an increase in
the risk of mortality,
dose-dependent.
Whereas for total fat and
protein, we see the reverse.
Fat and protein are
associated with lower risk.
Then when we look at the types of fat,
on the left, we see what
higher saturated fat,
mono and polyunsaturated fat,
there’s a pattern that
suggests a protective effect
against cardiovascular disease, death.
On the right, similarly,
we see higher intake
of saturated and polyunsaturated
and monounsaturated,
also is associated with lower risk
of non-cardiovascular death.
This suggests a protective effect
for the various types of
fats versus CVD and non-CVD,
whereas for carbs, you
get the complete opposite.
The third paper that
we published last year,
this one in The Lancet by my
PhD student, Victoria Miller,
who just graduated, by
the way, with her PhD,
we look at the fruit,
vegetable and legume intake
in cardiovascular disease
and deaths in 18 countries.
In the interest of time,
I can only shown one slide on that,
but generally what we found
was higher fruit,
vegetable, legume intake,
was associated with a
lower risk of mortality,
and a lower risk,
well a trend for a lower
risk with major CVD,
but that’s not quite clear-cut,
so more neutral than protective.
But for total mortality,
we see a protective effect,
with a maximum benefit at
three to four servings per day.
Now it’s important to point out that PURE
largely represents
countries around the world
where in large parts
of the world people eat
predominantly carbs that are
refined grains and added sugar.
Fruit and vegetables represents
only a small component
of carb intake.
I’m not showing the distribution today,
in the interest of time,
but it was less than
10% of the carb intake,
was fruit and vegetables.
So the other carbs largely explain
the harmful effect that we see for carbs.
So to summarize, a high carbohydrate diet
exceeding about 50, 55% of energy
is associated with
increase risk of mortality.
Various types of fats, can
be saturated and unsaturated,
are associated with
lower risk of mortality.
No association between
total fat, types of fat,
and cardiovascular events.
We did see that saturated fat
was associated with a
lower risk of stroke,
which is also consistent with
prospective cohort studies.
So the current advice to limit total fat
to less than 30% of
energy in saturated fat,
to less than 10%, is not
supported by this global study.
We recommend that people
eat fruit, vegetables, nuts,
legumes, dairy and meats.
And to eat less refined
grains, sugar, processed meats,
sweetened drinks, and to
avoid industrial trans fat.
The rest of my talk will focus on salt.
We’ve done a lot of work on salt,
which is the other so-called,
holy grail nutrient
for reducing cardiovascular
disease in populations.
Just real briefly, salt intake is related,
the hypothesis states,
that salt intake is related
to raised blood pressure.
Raised blood pressure is a risk factor for
heart attack, stroke and death.
So if we reduce salt
intake in the population,
we’ll also reduce heart
attacks, stroke and death
in the population as well.
Of course, this assumes
that sodium affects
only blood pressure
and has no other biological
effects in the body.
Of course it doesn’t
quite work out that way.
That’s why I’m here.
So just briefly, the current
guidelines on salt by the WHO,
also other guidelines, like AHA,
recommend a very low amount of salt,
between less than two grams
per day to 2.4 grams per day.
Which is only about one
teaspoon of table salt.
That’s a very low amount of salt.
Very difficult for most people
to sustain in the short-term,
let alone in the long-term.
Some guidelines, like AHA,
recommend even lower amounts
for high risk people,
less than 1.5 grams per day,
which is only 0.7
teaspoons of salt per day.
In the interest of time,
I can’t really get much
into the literature,
but in general, studies
of general populations,
looking at sodium versus blood pressure,
have found weak relationships
between sodium and blood pressure.
The most widely cited study is with the
INTERSALT Study, published BMJ, 1988,
they found about a one
millimeter mercury increase,
systolic pressure per gram.
But an equally-well
conducted study in Scotland,
published in the same issue
of the BMJ’s INTERSALT,
found no relationship between
sodium and blood pressure.
But it’s INTERSALT that
gets widely quoted.
More recently, INTERMAP,
study of four countries,
or 17 centers and four countries,
also found no relationship
between sodium and blood pressure.
These are studies that use
24-hour urine collections
to measure sodium.
The DASH trial is the primary basis
for the current AHA guidelines,
and the dietary guidelines on sodium.
But it’s important to point
out that the DASH trial
was a proof-of-concept study,
as to whether changes in
multiple aspects of the diet,
including sodium, reduce blood pressure.
Meals were provided to participants
over the course of five weeks.
This was a short-term feeding study,
to look at short-term
effects on blood pressure.
It was not designed to assess of sodium,
also reduces cardiovascular disease
and mortality in general populations.
There were also other issues
with the DASH trial that
I won’t get into today,
but are covered in many
of our review papers.
The bread and butter of
epidemiology is measurement,
the measurement of the dietary exposure.
Now 24-hour urine collection
is the reference method
for assessing sodium
intake in individuals,
but it’s not feasible
in large scale studies.
And under-collection is a
problem, and so is selection bias,
because of the high-subject burden
in doing 24-hour urine collections,
you get high non-participation rates.
We need simpler methods to assess
intake in large populations studies.
Fasting morning urine has been validated
verus 24-hour urine collections,
and also has been shown
to be associated with blood pressure,
with slopes identical
to randomized trials.
For example, here’s data in 133,000 people
that we published, pulling
data from four studies,
spanning more than 50 countries.
We applied our method,
because we had morning
fasting urines collected
in all these individuals
versus systolic blood pressure.
Red is people with hypertension.
Blue is people without hypertension.
You can see that you see
a positive association
between sodium and
systolic blood pressure,
in both the hypertensives
and non-hypertensives,
with a slope that’s twice
as steep in hypertensives.
Consistent with what we know.
And slopes of association
that are consistent
with randomized trials.
For use in large populations
to stratify individuals,
to look at relationships
with clinical events,
this is a very valid method
that could be applied, widely.
That brings us to the first
paper that we published
in the New England Journal, 2014.
We looked at sodium and
potassium versus blood pressure.
In the interest of time,
I’m going to skip a few of these slides.
This one here, I wanted to point out.
This slide here shows the
frequency distribution
of sodium intake around the
world, in 100,000 people,
spanning five continents.
The red curve shows
observed sodium excretion
based on a single urine measurement.
The blue curve represents usual intake
after adjusting for random error.
You can see, overall the
average sodium intake
in the global population
is 4.9 grams per day.
You’ll also note that the U.S. average is
markedly lower than the
global population average.
So it’s not like we’re
consuming higher amounts
than other countries.
If anything, we’re markedly lower
than the global average,
of 3.5 grams per day.
You’ll also note that,
looking at the 2.3 gram recommendation,
less than 3% of the global sample
meets that 2.3 gram recommendation.
And when you adjust for random error,
less than 1% meets the recommendation.
When we look at the 1.5
gram per day cutoff,
specified by the AHA, less than
1% meet that recommendation.
When we adjust for random error,
nobody meets that recommendation.
So in other words, what
we currently recommend,
is what nobody eats.
(audience laughing)
So the human experience for
low sodium is extremely sparse.
The other important point is that China
has a markedly higher level of sodium
compared to most other
countries of the world.
That’s an important point to make,
as you’ll see with the data later.
So China’s average is 5.6 grams per day,
whereas, most other countries
are below five grams per day.
An average of about 4.4 grams per day.
Potassium intake, though,
is pretty similar.
Sodium is an essential nutrient.
We know every essential nutrient
follows a U-shaped
relationship with health.
Sodium is no different.
It’s an essential nutrient.
The body needs sodium.
It’s not like tobacco,
where we know the optimal level is zero.
Sodium ya need.
We know at high levels you get toxicity,
at low levels you get deficiency.
There’s always a sweet spot in the middle
for every essential nutrient.
That brings us to the first of our papers
where we looked at clinical events.
This is on target.
People with vascular disease,
followed up for about five years.
28,000 people, more than
4,700 clinical events
accrued during the follow-up.
You can see that, on the right,
you see the cubic spline curve.
When intake exceeds
about six grams per day,
we seen an increase in risk.
At the low end, when intake
is below three grams per day,
we also see an increase risk.
So there’s a sweet spot
between about three to six grams per day
associated with the lowest risk,
which is consistent with
any essential nutrient.
Note that the U.S. average
is currently at the lower
end of the sweet spot,
which means that we’re
eating about the right amount
that we should be.
If we were to reduce intake
to the currently recommended
level for vascular patients,
we would actually put
people at increased risk.
On target, is people
with vascular disease.
What about general populations?
So that brings us to
the second of our papers
published in New England Journal, 2014.
Here we look at sodium and potassium
versus cardiovascular events.
Over 100,000 people,
from the 18 countries,
we accrued more than 3,300
major cardiovascular events
during the follow-up.
At that time, follow-up was 3.7 years.
This slide summarizes the results
for the various outcomes of
the primary composite outcome
of major cardiovascular disease and death.
Total deaths, on the top right,
and at the bottom, we have
major cardiovascular disease.
You can see a similar pattern of results.
Again, a sweet spot in the middle.
High intake, you seen an increase risk,
about six grams per day and higher.
Low intake you see an
increased risk, even more.
Less than three grams per day
associated with increased risk.
So between three to six
grams, we found to be optimal.
Note the U.S. average again
is at the lowest part of that sweet spot.
I don’t have time to get into
the extensive analyses that we did,
but I’ll just say that no
matter how you analyze the data,
every humanly way possible,
the results would not go away.
We did extensive adjustment
for all know confounders.
We excluded people with
cardiovascular disease, cancer,
people with an event
in the first 24 months.
Again we say that low sodium
is associated with increase risk
compared to moderate sodium intake.
For potassium we see,
clearly a protective effect.
Higher potassium
associated with lower risk.
We got very favorable commentaries
from various journals,
including the New England Journal,
Suzanne Oparil,
who was the former head of the
American Heart Association,
had a very positive
commentary on our paper.
Despite what you might see,
some individuals out there
are trying to portray that it’s only PURE.
Other studies have shown
exactly the same thing,
including studies that use
24-hour urine collections.
Including people with type-1
diabetes, type-2 diabetes,
people from the general
population in Europe,
and people with chronic kidney disease.
Either a J-shaped relationship
or an inverse association.
Actually an inverse association.
Higher sodium related to lower risk
of cardiovascular events or death.
This is consistent across cohort studies
where you compare moderate
sodium to low sodium.
Each and every time,
moderate sodium does better.
High sodium, when you go
above five grams per day,
compared to moderate, we
see an increased risk.
So Grotto and his meta-analysis
identified a sweet spot
between 2.7 to five grams per day.
If you go above that, or below that,
you see an increased risk.
Another paper we published
in 2016 in The Lancet,
we sub-divided people with hypertension,
and without hypertension.
The point I want to make here is that
overall you see again a U-shaped curve
in 133,000 people, pulling
data across four studies
from the general population
in people with cardiovascular
disease history.
On the left and right here you see
hypertension and no hypertension.
The point here is that,
regardless of blood pressure status,
again you see low sodium is
associated with increase risk
of major cardiovascular events and death,
compared to moderate sodium.
So regardless of whether
you have high blood pressure
or normal blood pressure,
low sodium is associated with harm.
Remarkable consistency
across sub-populations.
Why would low sodium be harmful?
We know that low sodium is a predictor
of higher renin and aldosterone levels.
Intervention trials have also shown
that lowering sodium intake,
increases renin and aldosterone,
and activates the renin
angiotensin system.
We know these are vascular
damaging substances.
They even have a class
of blood pressure drugs
directed at blocking this mechanism.
Low sodium has been shown
to increase these hormones.
Last week we published a new paper,
this one also published in The Lancet.
This time we look at the community level.
Over 300 communities around the world.
First we looked at the center level,
urban and rural sub-centers,
community and individual level.
You see similar slopes of relationship
between sodium and blood pressure.
Again reaffirming the accuracy
of our sodium measurements.
Now in the interest of time,
I can only point a few of
the community level results.
On the top left, you see
major cardiovascular disease.
We see generally an increase risk
when go above five grams per
day, but not below five grams.
The effect is driven largely
by the increase in stroke.
But the increase in stroke occurs
when sodium goes above five grams per day.
Remember, that’s levels they see in China.
Where most of the other
communities are below five grams,
in the other countries you see it’s flat.
You see no increase in risk.
The one last slide I wanted
to show you, which is key.
Then we sub-divided China,
versus the other countries,
looking at the biggest
signal, which was stroke,
and we see that China,
most of the communities
have an intake above five grams per day,
whereas in the other countries,
we see most communities are
below five grams per day.
In China, we see a positive association
between sodium and stroke.
In the other countries,
we see, if anything,
an inverse association.
So it’s China that we need to focus on
getting their communities
from above five grams,
down to below five grams,
in that three to five gram sweet spot.
Potassium was beneficial.
We see higher potassium
related to lower risk.
So higher potassium intake was beneficial.
Sodium greater than five grams
is associated with increased risk.
Such high levels are
found mainly in China,
less common in other countries.
Low sodium is associated with
higher risk and mortality.
Potassium is associated …
So rather than a population-wide strategy,
we need to target
individuals and communities
above five grams per day.
Most people fall
in that three to five gram per day range,
which is what they should be eating.
More people increasingly
are starting to agree
that large randomized controlled trials
are needed to test low sodium
versus moderate sodium intake.
World Heart Federation called for more
randomized controlled trials to be done.
Last slide, to conclude.
First and foremost,
Hippocrates said, “Do no harm”.
So the public health has the potential
to do great good when it’s right,
but great harm when it’s incorrect.
This is a picture of our PURE Study team
in our last meeting in New Dehli, India,
in November of last year.
Thank you very much.
(clapping)
– [Announcer] Thank you Andrew.
Very excellent talk on
a very important study.
It’s my great honor now to
introduce our next speaker
who is Dr. Ron Krauss.
Where is Dr. Krauss?
Oh, there you are.
He’s a legend in the
area of lipid research.
He’s based in San Francisco.
If you read anything in lipid metabolism
over the last two or three decades,
you’re gonna be familiar
with Dr. Krauss’s work.
He was a pioneer in the
elucidation of particle size
in terms of lipoproteins.
It’s my pleasure to
welcome you, Dr. Krauss.
(clapping)