« Faith | Main | What Works, What Doesn't »
January 23, 2010
MR on Saturated Fat
I mentioned a few weeks ago that MR responded to RDF's inquiry re: sat fat. I'm finally getting around to posting it... read and draw your own conclusions!
Keep in mind... MR is far from a lipophobe. He advocates a high fat diet, but coming mostly from MUFAs and PUFAs.
Please do be sure to follow the embedded links!
Hi Dr. Feinman,
As relayed by the fair hand of April:
> What does MR say about saturated fat? ... what is the
> evidence. Not a challenge, because I present to medical students my
> side of the picture, the failure of Framingham, Malmo and WHI and,
> most of all, Jakobsen's article showing that replacing saturated fat
> with PUFA, or MUFA or carbs is pretty much of a flip of the coin in
> terms of outcome.... The only answer I
> got was reference to all the health agencies opinions and the fact
> that the Jakobsen paper that I quoted showed PUFA was better than
> SFA.
By "the Jakobsen paper," you mean the AJCN meta-analysis, right? (Let me come back to WHI in a moment). Actually, I too would start there, as even on an abstract-surfer's casual analysis it's sufficient grounds to draw the *broad strokes* of the core conclusion: it shows fairly clearly, from a large number of mostly very high-quality prospective epidemiological studies in Western countries with a range of different dietary patterns, that typical North American and European intakes of *both* saturated fat and carb, on an isocaloric basis, are significant contributors to coronary events and mortality -- even now, when we have statins, metformin, thiazides, et al to control some of the adverse metabolic sequelae of such lifestyles.
> And, for sure, replacing SFA with carbs is, at best, a wash.
Yes -- which constitutes an indictment of both :) . Ie, a pox on both Atkins' and Ornish's houses! This argues for a lower-carb, lower-SFA, higher-PUFA diet -- something like "South Beach" or the Zone.
But let me drill down into those results a bit, as I think there's more to be mined therein:
> I presented that at the CR meeting and my interpretation was
> that there was so much individual variation that the mean didn't > really mean anything.
I agree, and think that we can say a bit more about this. As you know (you alluded to it at the Conference and have said so in more detail elsewhere), we actually know a fair amount about the BASIS for such variation in the population, due to the differential metabolic responses observed in response to a low-carb diet, with leaner, more insulin-sensitive subjects (and, though you didn't say it, younger ones) responding less favorably to a low-carb diet and more favorably to an old-style AHA low-fat (and, specifically, low-SFA, low-cholesterol) pattern. In the "Discussion" in the Jakobsen meta-analysis, they note that "It has been suggested that the association between major types of fat and risk of CHD is modified by sex and age. This study suggests that to prevent CHD, SFAs should be reduced and replaced with PUFAs among all middle-aged and older women and men. However, it cannot be excluded that associations may be stronger in subgroups, but our study only provides a suggestion for these possibilities."
A somewhat stronger case than the study they cite on this point, although indirect to the specific issue of SFA vs carb, comes from a comparison of studies on the outcomes for prospective studies in Europe vs America for mortality outcomes for a low-carb diet score based on intakes of *protein* vs carb:
http://ki.se/content/1/c6/04/99/84/WLH_22_Lagiou%202007.pdf
http://www.nature.com/ejcn/journal/v61/n5/full/1602557a.html
http://content.nejm.org/cgi/content/full/355/19/1991
In the Nurses Health Study, eating a lower-carb, higher-protein diet was associated with a reduced risk of cardiac and total mortality in women, whereas in the 2 European studies, such diets were associated with *higher* risk. Willet's editorial:
http://www3.interscience.wiley.com/journal/117966552/abstract
... notes the dietary differences amongst populations:
**************
We can set aside immediately that the difference in findings were due to fundamental differences in study quality. All three studies were prospective, meaning that the potentially serious biases in case–control studies were avoided; all used dietary assessment methods with documented validity; and all had quite complete ascertainment of outcomes. …
Thus, the apparently contrasting findings are likely to be due to real differences in the study populations, their diets or the methods of analysis. … Women in the Swedish study were substantially younger ... than in the US study ... and leaner ..., and for both of these reasons, the degree of insulin resistance would have been much lower in the Swedish study. One of the most important findings during the last decade has been that the adverse metabolic effects of high-carbohydrate intake are greatly magnified in the presence of underlying insulin resistance. Consistent with this evidence, we have seen that high dietary glycaemic load... has little effect on the risk of coronary heart disease in lean women, but nearly doubles the risk in overweight and obese women. Similarly, high glycaemic load was not associated with the risk of stroke in leaner women, but was significantly associated with risk in overweight and obese women. The types of diets consumed by the Swedish and US women may have contributed to the differences in findings. From many dozens of controlled feeding studies, we know that different types of fat have opposite effects on blood lipids [blah blah ;) ] ...
Similarly, the type of carbohydrate rather than the total amount of carbohydrate is key in relation to the risk of cardiovascular disease. As noted above, high amounts of refined starch and sugar have adverse metabolic effects and increase risk; in contrast, whole grain/high-fibre forms of carbohydrate have beneficial metabolic effects and are related to reduced risks of cardiovascular disease. ... [F]rom other reports we know that the cereal fibre content (the form of fibre most strongly related to lower risk of cardiovascular disease and diabetes) is approximately three times higher in Swedish compared to US women [9]. Whilst there is little reason to believe that the type of protein influences the risk of cardiovascular disease [I'd actually disagree on this -- there is substantial evidence of a more protective effect of *vegetable* protein; references available on request ;) -MR], the protein package can include large amounts of cholesterol and saturated fat (in the case of animal proteins) or fibre and beneficial micronutrients (with plant proteins). The available data do not allow a comparison of the sources of proteins in the Swedish and US studies. ***********
Jakobsen et al also note, re: the surprising neutral-to-unfavorable results for MUFA, that "the main source of MUFAs was animal fat, whereby confounding from other dietary components in meat and dairy products cannot be excluded. " Other studies point pretty clearly to a protective effect of olive oil and other less-confounded MUFA sources with a protective effect against mortality and other outcomes.
While it's getting a bit off-topic, and while I imagine you're initially going to think that I am attempting to rouse a long-dead myth, a substantial amount of evidence has emerged just in the last few years for a divergent effect of dietary *cholesterol* (a fellow-traveller with SFA when it comes from animal sources) depending on leanness and insulin sensitivity:
http://www.crsociety.org/archive/read.php?2,189896,189896#msg-189896
(Do follow the embedded links).
What I think we're seeing is exactly the divergence within these populations that you know: that carb is really rather bad for overweight, insulin-sensitive people, such that replacing it even with SFA is relatively harmless -- whereas for lean, insulin-sensitive people, SFA (and dietary cholesterol, its fellow-traveller in omnivorous diets) is likely more *relatively* harmful, because carb is less able to derange the metabolism. We have to remember that any time we look at these studies and see only modest or borderline-significant effects: 66% of the US population is overweight, and half of that majority is obeese; Europe is somewhat better-off, at 49.8% and 13.3% in men and 36.0 & 13.5% in women per MONICA. So the deleterious effects of any nutrient with a differential effect on low-BMI, insulin-sensitive people will tend to be blunted by the much larger number of people for whom such effects are blunted by their "larger" problem.
It also means that the deleterious effects of a rise in SFA intake are at least temporarily outweighed if it is is part of a dietary shift into a lower-carb diet when it is successfully used for weight loss (as opposed to just being a person's self-selected default diet, which of course is what's going on in teh studies in Jakobsen and in the Swedish, Greek, and US Nurses low-carb/high-protein studies). But it's reasonably clear that if you're insulin-sensitive -- which, interestingly, is what one is likely to become after losing weight on a successful low-carb weight-loss diet! -- the effects of SFA become more *relatively* harmful as teh deleterious effects of carb recede.
Moreover, I think we have a fair amount of evidence for a specific (albeit widely exaggerated) benefit of omega-3 fatty acid intake, which unfortunately the Jakobsen meta-analysis couldn't evaluate. I think it's reasonable to expect that a discrimination here would have further emphasized the benefits of these fats.
Overall, I think the epidemiology is consistent with your view that the effects of carb are differential, and that we need to pay attention to that. They're also consistent in particular with the widely-accepted view that there is 'good carb/bad carb' as well as 'good fat/bad fat'. Again, this argues to me that that the thing to do is to lose weight, by whatever means, is a Good Thing, but that the best way to do this -- and, more importantly, the *maintenance* diet for people who are of normal weight and below and who are insulin-sensitive -- is a *relatively* (vs population average) low-carb, low-saturated-fat, high-PUFA (and probably plant-derived MUFA), high-(vegetable)-protein diet, with the carb coming from vegetables and whole grains rather than starchy crap.
Ie, a pretty boringly mainstream viewpoint :) with a bias toward the Zone (or South Beach, if Agatson's recipes and detailed dietary guidance actually lived up to his high-level advice and the diet's reputation).
Now, that's the epidemiology. Interventionally, you mention the lack of reported benefit from a low-fat diet in WHI. I would first note, of course, that I don't advocate for a low-fat diet, but for a low-SFA, moderate- to high-fat diet. But also, I must draw your attention to the fact that the trumpeted "No benefit of a low-fat diet!" headline is based on (wait for it!) an intention-to-treat analysis ;). The intervention group self-reported a 29% fat diet, vs the 20% study goal, and as multiple studies have clearly shown, people systematically underreport almost everything they're eating, but especially fat (because it's easier to see that you've eaten a whole vs a half an apple than that you've added a whole vs a half tablespoon of lard to your frying pan); meanwhile, the controls reduced their fat intake almost as much, and both groups modestly (almost indistinguishably) increased consumption of fruits and vegetables, and neither group significantly changed intake of grains. In biomarkers, despite the fact that lower SFA intake will, pretty unambigiously, lower LDL levels (whatever else it does -- and more on THAT below), LDL decreased only 2.6% more in the low-fat group than in the controls.
And, we don't know *which* fats these people were eating. We not only don't really know if they lowered SFA relative to other fats: we don't know if, as part of following widely-publicized "low fat" advice, they replaced SFA with margarine, which was still trendy through most of the intervention period, going from bad to what I'm sure you already agree was worse; we also don't know if they rendered themselves n3 deficient. I'll say it again: I wouldn't say that there is an evidence base for supporting a merely low-fat diet!
And was there even a REAL (even if tiny!) reduction in even fat intake?? If they fell prey to the "Snackwell effect," not actually LOWERING their total fat or SFA intake, but increasing their intake of simple carb (as we know from CDC and others that the US population as a whole did following the propagation of the low-fat oversimplification by the public health authorities in teh late 70s thru' the early 2000s) the result would be reported as a percentage reduction of SFA when they've actually just added high-GI carb Calories onto already-toxic Standard American Diets ("WHI! The New and Improved SAD -- with Added Metabolic Syndrome!").
So what is GOOD evidence? I'd say the clearest intervention data we have is also the most robust: the Lyon Diet Heart Study:
http://archinte.ama-assn.org/cgi/content/full/158/11/1181
http://www.circ.ahajournals.org/cgi/content/full/99/6/779
... which as I'm sure you know reported that a dietary intervention with multiple components to the *guidance* , but which in the end based on actual shifts in dietary intake (by self-report and by biomarkers) largely amounted to a modest reduction of SFA in vavor of a higher intake of ALA (and, modestly, MUFA) led to quite astonishing reductions in cardiovascular and total mortality (as well as cancer, although the n was so small as to not make this a strong specific outcome, esp as it was a post-hoc outcome analysis). I know that there were initially methodological criticisms of the study, but in my view the biomarker data rebut a lot of that, and certainly it's the best overall evidence we have for a dietary intervention trial. Similar results, with more rigor, were reported by Singh et al:
http://linkinghub.elsevier.com/retrieve/pii/S0140673602114723
... despite using mustard oil, which is problematic (inferior n6:n3, may contain inflammatory constituents, erucic acid content seems to have been absolved of previous suspicion still somewhat controversial); unfortunately Dr. Singh's research is now under a substantial cloud of suspicion: http://dx.doi.org/10.1016/S0140-6736(05)67005-5
I'm sure you can come up with a lot of criticisms on methods or analysis of these studies, but it looks to me to be the best evidence that's available on the subject, and what I give above appears to be the best analysis, based on the results, of its implications in terms of how one should eat for one's health. That is: this is the balance of evidence from prospective epidemiology and clinical trials, and there doesn't to my knowledge seem to be evidence of similar quality (large-scale, well-characterized populations, followed up for sufficient time to expect to see a difference in outcomes) that is equally suggestive of a contrary conclusion.
Notably, the critiques of these broad conclusioins that seem most widely cited by low-carb laypeople in reply (Ravnskov, Taubes, etc) clearly don't meet this standard: they spend most of their efforts bitching about the limitations of what IS available, beating dead horses, and emphasizing *lower* -quality studies while breezing over even better data. They spend, eg, an absurd amount of time criticizing Keys, who of course got the whole Diet-Heart Hypothesis going on EXTREMELY shaky grounds in the first place -- but pointing out the multiple flaws in Seven Countries doesn't take a whit away from the quality of SUBSEQUENT studies of much greater rigor, like the Nurses' Health Study, Health Professionals Follow-up Study, Physicians' Health Study, etc, which they instead ignore entirely or breeze over.
At the same time, these critics spend a lot of time emphasizing studies of much lower quality -- including, ironically, spinning the low-quality Keys data in a different direction (as if that could lead to a more reliable conclusion), and simiilarly making much of other purely cross-sectional, population-level ecological data. Ravnskov, eg, is quite enamored of what he claims is observed in the Masai; it's bad enough that this is rebutting a low-quality ecological study with an even lower-quality, even smaller ecological study (as if such studies were worthy of consideration to begin with!), but to make things worse, he misrepresents the data itself, and ignores available and more plausible explanations even for what he DOES accurately review! Taubes makes many of the same mistakes, and then does things like blame the rise in obesity in the late 20th century on the "fact" that "the percentage of fat in the American diet has been decreasing for two decades", when as noted above their actual intakes of total fat remained static and carb intake progressively increased (which (surprise!) leads to overweight and metabolic syndrome), and then asserts that over the same period, "Our cholesterol levels have been declining [true! -- due to an actual reduction in SFA -- MR], and we have been smoking less [also true -MR], and yet the incidence of heart disease has not declined as would be expected." In fact, as you may know, coronary disease and mortality declined quite remarkably in the late 20th century, DESPITE the rise in obesity, metabolic syndrome, and junk carb intake:
http://content.nejm.org/cgi/content/full/343/8/530
http://www.ncbi.nlm.nih.gov/pubmed/15308430
And all of them engage in much hypothesis-spinning from the biochemistry, which is a fine basis for generating new hypotheses but not for rebutting actual, prospective study outcomes.
At the end of the day, we have to go with what we've got. Until we get a couple of thousand healthy twenty-year-olds locked up in metabolic wards for sixty years or so for a really vigorous diet trial, I think saturated fat AND carbs (especially starchy carbs) stand out as things to reduce in the diet, in exchange for vegetables, fruit, lean protein, and PUFA (and probably MUFA) as things to maintain or increase. And most people should lose weight!
Again, boring ol' me ;) .
(A postscript, with relevance to the wider subject: at the Press Conference, I'd mentioned a meta-analysis finding that the epidemiology couldn't actually support the differential effect of small, dense LDL on cardiovascular risk, as giving additive information to the basic risk profile. It's available here:
http://www.annals.org/content/150/7/474.full
Live long -- live young!
-Michael
Posted by april at January 23, 2010 9:23 AM
Comments
Well, now I know what I'm doing for the rest of the afternoon -- reading this post! [MR, you do rather go on, don't you?] MoMR :-)
Posted by: Judith at January 23, 2010 12:03 PM
After reading this, I think I need to eat a dozen eggs.
Posted by: DAVE at January 23, 2010 4:01 PM
Hi, another occasional lurker.
Michael said "... which as I'm sure you know reported that a dietary intervention with multiple components to the *guidance* , but which in the end based on actual shifts in dietary intake (by self-report and by biomarkers) largely amounted to a modest reduction of SFA in vavor of a higher intake of ALA (and, modestly, MUFA) led to quite astonishing reductions in cardiovascular and total mortality (as well as cancer, although the n was so small as to not make this a strong specific outcome, esp as it was a post-hoc outcome analysis)."
this above quote confused me as I always thought the experiment group were told to increase consumption of root vegetables, green veg, fish, bread and poultry and cut down on beef lamb and pork as well as eat fruit daily and replace butter with olive oil and a canola-based margarine that was higher in monounsaturated fats and omega3. And the blood work of the experimental group showed similar cholesterol readings then before the study but increased concentrations of omega3 and reduced concentrations of omega 6 as well as increased levels of vitamin C, E and Vitamin A (other vits/antioxidants not being measured but presumably also increasing in the experiment group). So from all of that how can Michael still feel strongly enough to single out saturated fat given all the other interventions are likely to be as powerful or more powerful in relation to heart disease prevention?
Does not look like the recipient of this email would be convinced at all by Michaels' data given the Lyon study being the best evidence against saturated fats.
Posted by: Goeff at January 25, 2010 8:14 AM
Perhaps relevant, posted today:
"Saturated fat and heart disease: studies old and new"
http://is.gd/769OD
Posted by: Scott Miller at January 26, 2010 12:00 PM
Hello Amy,
Apply for a grant instead of and/or doing a fund raiser. Or research the amount of grant money there is out there and then supplement with a fund raiser. Keep up your commitment. Health is earned and enjoy the process of trying different foods and amounts of food intake then you will settle on what is right for you. My moto is,
"There are no rules just tools"
I am wondering why as a society we are comfortable with consuming copious amounts of food and expand our bodies to morbid obesity while consuming animal flesh 3 times a day and more. Instead of being comfortable maintaining a healthy weight and if that means no animal products so be it. I remember my skinny minny neice had her cholesterol checked when she was 13 and it was over 300. She has gravitated to chosing healthy meal planning and is conscious of a healthy cholesterol level. Health is defined by the individual for the individual by freedom to chose.
Stay well, Jan
Posted by: janet at January 27, 2010 4:22 AM
I really enjoyed reading this: very thought-provoking discussion, mostly clear; with my limited internet time I'm following through a few of the embedded links as I can. A couple of edits, because I can't help it :) And a couple of questions too.
MR said "What I think we're seeing is exactly the divergence within these populations that you know: that carb is really rather bad for overweight, insulin-sensitive people, such that replacing it even with SFA is relatively harmless..."
I think this should read 'overweight, insulin-resistant people.'
Then, MR said: "Taubes makes many of the same mistakes, and then does things like blame the rise in obesity in the late 20th century on the "fact" that "the percentage of fat in the American diet has been decreasing for two decades", when as noted above their actual intakes of total fat remained static and carb intake progressively increased..."
These two statements are not inconsistent. If their _total fat intake_ remained static and their carb intake progressively increased, then _as a percentage_ of the diet, the fat intake went down!
I would like to see MR's critique of Taubes' work more clearly expounded: I'm sure that there are useful emendations to learn, despite the non-validity of that particular piece of criticism.
I hope I'm not sending in this comment too late when it's no longer 'current:' first chance I've had, and I kept it in my head.
Finally, my question. What of the difference between SFs from plant sources versus animal sources? I have concluded that I need to avoid animal SFs but have had a good experience with coconuts and coconut oil for some time now, in terms of their satiation ability as well as heat stability. The sweet taste of coconut products is also a bonus when you're not eating carbs. I have noticed the claimed anti-fungal benefits also, and I think that the MCTs are supposed to be necessary for various processes. Cacao butter is another one I think of using occasionally, but it is mostly stearic acid, I think, so less obviously beneficial and probably more of a frivolity when I use it...
Would appreciate your thoughts!
thanks,
Ela
Posted by: Ela at February 6, 2010 11:50 AM
Dissolve Away those Pesky Bones with Corn Oil
CO [corn oil] is known to promote bone loss, obesity, impaired glucose tolerance, insulin resistance and thus represents a useful model for studying the early stages in the development of obesity, hyperglycemia, Type 2 diabetes [23] and osteoporosis. We have used omega-6 fatty acids enriched diet as a fat source which is commonly observed in today's Western diets basically responsible for the pathogenesis of many diseases [24].
http://wholehealthsource.blogspot.com/2010...bones-with.html
Posted by: Scott Miller at February 18, 2010 4:43 PM