Astrup A et al. Am J Clin - American Heart Association

Saturated Fat and Cardiovascular
Disease Risk
Patty W. Siri-Tarino, PhD

Dietary Saturated Fat and CVD Model
Dietary
Saturated Fat
LDL Cholesterol
Other CVD risk
factors
CVD

CHD Mortality is Correlated with Plasma
Cholesterol Levels
18
16
Six Year CHD Mortality from MRFIT
CHD
Death
Rate
per
1000
14
12
10
8
6
4
2
0
Desirable
Borderline
High
HIGH
140 160 180 200 220 240 260 280 300
Plasma Cholesterol (mg/dl)
LaRosa et al, 1990

Saturated Fat Replacement Scenarios
Replacement
Nutrient
TC LDL-C HDL-C TG TC/HD
L-C
PUFA ≈
MUFA ≈
CHO ≈
Mensink RP and Katan MB. Arterio Thromb 1992, 12:911-919; Mensink RP et al. Am J Clin Nutr 2003, 77:1146-55.

Atherogenic Dyslipidemia
Obesity
Insulin Resistance
Dietary Carbohydrate
- Elevated triglyceride concentrations
- Reduced HDL cholesterol concentration
- Small and dense LDL particles
Krauss RM and Siri PW. Endocrin Metab Clin N Amer 2004; 33:405-415

LDL cholesterol is comprised of different
subclasses with differing CVD risk
Large
1
Medium
Small
2
3
4
Increased CVD risk
Reduced clearance
Greater entry into artery
Greater retention
Faster oxidation
Distribution of subclasses varies widely among individuals and is
independent of total LDL cholesterol
Berneis and Krauss, JLR 43:1155, 2002

Saturated fat increases large and medium, but
not small, LDL particles
Krauss RM, et al. 2006; 83: 1025-1031; Siri-Tarino P W et al. Am J Clin Nutr 2010;91:502-509

Reduced Total/HDL Cholesterol with Both
Carbohydrate Reduction and Weight Loss
∆ Total/HDL
cholesterol
0
-.1
-.2
-.3
-.4
-.5
-.6
-.7
-.8
-.9
Higher sat
same fat
20 25 30 35 40 45 50 55
% Carbohydrate
Before weight loss
After weight loss
Krauss RM, et al. 2006; 83: 1025-1031

Association of saturated fat with serum
cholesterol is modified by body weight
Saturated fat intake was
not associated with
serum cholesterol in
obese versus lean men.
•
120 +
100-119
< 100
x Hawaii
• Japan
Kato et al., Am J Epid 97(5), 1973
Even in young, healthy non-obese men, the plasma lipid response was associated with BMI.
Jansen S et al. J Nutr 1998; 128:1144-9.
Insulin resistance has also been associated with a reduced LDL response to reductions
in saturated fat. Lefevre M et al. Am J Clin Nutr 2005; 82:957-63.

SUMMARY
LDL cholesterol is associated with CVD risk, and saturated fat
increases LDL cholesterol
Replacement of saturated fat with carbohydrate can lead to
atherogenic dyslipidemia
Dietary saturated fat increases larger and more buoyant LDL
particles
Weight loss trumps any effect of dietary composition
Body weight may modulate effects saturated fat and
carbohydrates on blood lipid levels, such that saturated fats have
weaker effects in overweight and insulin resistant persons

Saturated Fat and CVD Model
Dietary
Saturated Fat
LDL Cholesterol
Other CVD risk
factors
CVD

Saturated Fat & Other CVD Risk Factors
HYPERTENSION: No effect of high saturated fat on blood
pressure in 162 healthy persons. Rasmussen et al, AJCN 2006
INSULIN RESISTANCE:
Association of high saturated fat diets with fasting or PPL
hyperinsulinemia
Marshall et al, Diabetologia 40(4) 1997
Parker et al, AJCN 58(2), 1993
Christiansen E, et al, Diabetes Care 20(5) 1997
No association with risk of Type 2 diabetes
Salmeron et al, AJCN 73(6) 2006
van Dam et al, Diabetes Care 25(3) 2002
Effect of saturated fat on insulin sensitivity may be
modulated by the total amount of fat in the diet
Vessby et al, Diabetologia 44(3) 2001

Saturated Fat & Other CVD Risk Factors
THROMBOSIS: Stearic, but not lauric, myristic and palmitic, increased
fibrinogen levels Baer et al, AJCN 2004
INFLAMMATION:
Positive effects
Activates NFκB, Cox2 and other markers (MØ) Laine et al, BBRC 2007
Cream challenge delays PPL ROS peak Mohanty et al, AJCN 2002
No effects
CRP Baer et al, AJCN 2004; Koren et al Nutr 2006; Erlinger et al. Circ 2003
Acute PPL response Bellido et al, AJCN 2004
Cytokine response - 1 mo butter or soybean oil Han et al, JLR 2002
VASCULAR FUNCTION: HDL collected after a meal had increased
ICAM and VCAM in persons who ate coconut meal versus safflower or
unsaturated fat meal Nicholls et al, JACC 2006

Dietary Saturated Fat and CVD Model
Dietary
Saturated Fat
LDL Cholesterol
Other CVD risk
factors
CVD

What is the evidence for an association
between saturated fat and CVD?
Randomized clinical trials
Epidemiological studies
Prospective cohort studies
Case control studies
Population/cross-sectional studies

Dietary saturated fat has been positively
associated with increased CVD
The Seven Countries
Study by Ancel Keys was
one of the first to link
dietary saturated fat with
increased CVD risk
Caveat: Inability to prove
causality; Other
population lifestyle and
covariates may be
relevant
Keys et al., Acta Med Scan 1966

RCTs with high PUFA and reduced SFA
intake: beneficial effects on CVD
Veteran Affairs; n=846; randomization to 40% fat diet, low
SFA (9%) and cholesterol, high PUFA; reduction in total
CVD
Dayton et al. Circulation 1969.
Finnish Mental Hospital Study; n=676 men; 6 yr crossover
study with low SFA (9.1%) and cholesterol, high PUFA;
reduced cholesterol, CHD and coronary death
Turpeinen et al. Int J Epidem 1979
Oslo Diet Heart; n=412 men post-MI, rand to 40% fat
(SFA=8.5%), high PUFA diet; reduced coronary death
Leren P. Acta Med Scan 1966, Circ 1970

RCTs with high PUFA and reduced SFA
intake: No effects on CVD
British Medical Research Council Soy Oil Intervention;
n=393 men post-MI; randomization to soybean oil
supplementation for 4 yrs; lower cholesterol but no
reduction in CVD Morris et al, Lancet 2:693, 1968
Minnesota Coronary Survey; n= 4393 men and 4664
women; randomization to 38% fat diet (9 vs 15% SFA; 5
vs 15% PUFA); lower cholesterol but no differences in
CVD events, deaths or total mortality Frantz et al, Arterioscler 9:129,
1989

Women’s Health Initiative:
Replacing saturated fat with carbohydrate
Randomization of > 48,000 post-menopausal women to low-fat or control diet to
evaluate effects on cancer and CVD risk.
Howard BV et al; JAMA 2006; 295: 655-66

Women’s Health Initiative:
Replacing saturated fat with carbohydrate
No differences in CVD risk were observed.
Caveats: dietary compliance; magnitude of fat difference achieved was less than
desired; reduction in PUFA along with decreased SFA and increased CHO
Howard BV et al; JAMA 2006; 295: 655-66

Lessons from RCTs
Replacement of saturated fat with polyunsaturated fat, but
not carbohydrate, was associated with CVD benefit
Figure: Hu FB et al. Am J Clin Nutr 1999 70: 1001-8
Calculated P:S ratios
ranged from 1.4 to 2.4,
values much higher than
a previously reported
threshold of 0.49. The
presumed benefit of
reductions in saturated
fat may therefore rely
on a significant increase
in polyunsaturated fat.
Siri-Tarino P W et al. Am J Clin Nutr
2010 91:535-546

Meta-analysis: Replacement of SFA with PUFA was
associated with reduced CHD risk
Mozaffarian D et al; PLoS Med. 2010 Mar 23;7(3):e1000252.

Siri-Tarino P W et al. Am J Clin Nutr 2010 91:535-546

Objectives
Primary: To summarize the evidence related to the
association of dietary saturated fat with risk of CHD,
stroke and CVD in prospective epidemiological studies
Secondary: To consider modulation of the association of
saturated fat and CHD, stroke and CVD by:
Age and gender
Replacement nutrient
Total energy intake
Siri-Tarino P W et al. Am J Clin Nutr 2010 91:535-546

Study Selection Process
Siri-Tarino P W et al. Am J Clin Nutr 2010 91:535-546

Baseline characteristics of participants in 21 unique
prospective epidemiological studies of SFA & CVD risk
Siri-Tarino P W et al. Am J Clin Nutr 2010 91:535-546

Saturated fat was not associated with an elevated
risk of CHD, stroke, or CVD as a composite outcome
RR CHD = 1.07 (0.96, 1.19)
RR stroke = 0.81 (0.62, 1.05)
RR CVD = 0.81 (0.62, 1.05)
Siri-Tarino P W et al. Am J Clin Nutr 2010;91:535-546

Subgroup analyses evaluating association of
saturated fat and CVD risk
1) Gender
2) Age (< or ≥ 60 yrs)
3) Gender & age
4) Gender & age
5) Total energy; n=15 studies
6) Replacement with carbohydrate; n=6 studies
7) Replacement with polyunsaturated fat; n=5 studies
Siri-Tarino P W et al. Am J Clin Nutr 2010;91:535-546

Caveats and Considerations

Dietary Assessment Methods
Diet records > FFQ > diet history > 24 hour recall
The accuracy of nutrient assessment can depend on the
method used.
The method of nutrient assessment was incorporated into a
quality score that also considered the number of dietary
assessments and the number of adjusted established CVD risk
factors.
Evaluation of the studies on the basis of this quality score did
not change the findings of this meta-analysis.
Siri-Tarino P W et al. Am J Clin Nutr 2010;91:535-546

Modulation by Replacement Nutrient
Mozaffarian D et al; PLoS Med. 2010 Mar 23;7(3):e1000252; Jakobsen MU et al; 89:1425-32; Howard BV et al; JAMA 2006; 295: 655-66
Hu FB, et al. NEJM 1997 337: 1491-99.

Type of dietary carbohydrate
Heart disease (n=2) 1.25 (1.00,
1.56)
Stroke (n=1) 1.02 (0.86,
1.21)
Barclay
All diseases
AW et al. Am
(n=27)
J Clin Nutr 2008;87:627–37
1.14 (1.09,
1.19)
GI RR p GL RR p
0.050 1.57 (0.87,
2.84)
0.805 1.28 (0.83,
1.98)
<
0.001
1.09 (1.04,
1.15)
0.14
0.27

Summary
There is insufficient evidence from prospective
epidemiologic studies to conclude that saturated fat is
associated with an increased risk of CHD, stroke or CVD.
However, this study lacked the statistical power to
consider age and gender effects, the appropriate
replacement nutrient, and the role of types of
carbohydrate.
Siri-Tarino P W et al. Am J Clin Nutr 2010;91:535-546

Addressing the Controversy
Assertion:
Inclusion of serum cholesterol in models can obscure the
saturated fat and CVD association.
Response:
Subgroup analysis in 9 CHD studies and 6 stroke studies that
did not adjust for serum cholesterol (n=291,126) yielded results
that did not differ significantly from those we reported for all
21 studies (n=347, 747).
RR CHD = 1.13 (0.96, 1.33)
RR stroke = 0.84 (0.63, 1.10)
RR CVD = 1.02 (0.86, 1.19)

Addressing the Controversy
Assertion:
The saturated fat and CHD relation may differ for “hard” versus
“soft” endpoints (RR ”hard” = 1.13 versus RR ”soft” = 0.99)
Response:
Our evaluation of this hypothesis yielded a non-significant pooled
Our evaluation of this hypothesis yielded a non-significant pooled
RR estimate of 1.18 (0.99-1.42) (n=7 studies that considered fatal
CHD as the outcome variable). Siri-Tarino PW et al. Curr Athero Rep 2010; 12(6):384-90

Addressing the Controversy
Assertion:
We were “incorrect” in suggesting that clinical trial data or
epidemiologic data are lacking for an association of saturated
fat with CVD below 9% of energy.
Response:
Evaluation of the RCTs (VA, Oslo, Finnish) and prospective
epidemiological studies (included in our meta-analysis) that
showed associations of saturated fat with CVD all had diets in
which the lowest quantile of saturated fat ≥ ~9% of energy.

Addressing the Controversy
Assertion:
We “ignore findings” that support CVD benefit of diets with
low saturated fat content (DASH and OmniHeart diets are
cited specifically).
Response:
We do not dispute the beneficial effects of the DASH diet on
CVD risk factors. However, the DASH diet was not an outcomes
study. Furthermore, OmniHeart showed that replacing 10% of
carbohydrates with polyunsaturated or monounsaturated fat led
to even greater benefit. Replacement of carbohydrate with
saturated fat was not evaluated.

Saturated Fat & CVD in Context
SCIENTIFIC/ACADEMIC CONSIDERATIONS:
Intrinsically a replacement issue
The lipid and clinical endpoint data that indicate benefits of
The lipid and clinical endpoint data that indicate benefits of
replacing saturated fat with polyunsaturated fat do not rule out
the possibility that the beneficial effects are due specifically to
the polyunsaturated fat content of the diet.

Saturated Fat & CVD in Context
PUBLIC HEALTH IMPLICATIONS:
Replacement of SFA with refined carbohydrates can:
1) Exacerbate atherogenic dyslipidemia (already an issue given the
current epidemics of obesity and insulin resistance)
2) Adversely affect other features of the metabolic profile, ie.
glucose, inflammatory and thrombotic markers Astrup A et al. Am J Clin
Nutr. 2011 Jan 26. [Epub ahead of print]

Saturated Fat:
Future Directions & Considerations
Need to consider:
Specific macronutrient sources to which SFA is compared
Foods or context in which SFA is consumed
Whether effect of replacing SFA with CHO has changed
with a population that is more obese
The nutrient based approach may not be as relevant for
the prevention and treatment of chronic diseases. Is a
move towards foods and food patterns more
appropriate?
Astrup A, et al. Am J Clin Nutr 2011 Jan 26 [epub ahead print]; Mozaffarian D, et al. JAMA 2010: 304: 681-682