Cholesterol Levels in Vegans: EPIC-Oxford Study
I have updated the blood lipids section of the VeganHealth.org article, Disease Markers of Vegetarians, with a new report from EPIC-Oxford measuring the cholesterol levels of 422 vegans living in the UK (1).
Vegan men had an average total cholesterol level of 170 mg/dl compared to 204 mg/dl for non-vegetarians. Vegan women had an average cholesterol level of 172 mg/dl compared to 195 mg/dl for non-vegetarians. More results, including those for lacto-ovo and pesco-vegetarians, can be seen in Table 1 (link).
Vegans also had a significantly lower amount of apolipoprotein B which is thought to promote fat deposits in the arteries.
The authors of the study suggest that vegans have lower cholesterol levels due to a lower body mass index, replacement of saturated fats with polyunsaturated fats, and higher fiber intakes.
References
1. Bradbury KE, Crowe FL, Appleby PN, Schmidt JA, Travis RC, Key TJ. Serum concentrations of cholesterol, apolipoprotein A-I and apolipoprotein B in a total of 1694 meat-eaters, fish-eaters, vegetarians and vegans. Eur J Clin Nutr. 2013 Dec 18. [Epub ahead of print] | link
December 22nd, 2013 at 11:00 pm
For me it’s definitely not the low BMI. I went from thin to obese (eating disorder up-and-down) but my cholesterol stayed low. My doctor actually said “i’ve never seen such a low cholesterol!” Does the study support his hypothesis or does he just “suggest” it?
December 23rd, 2013 at 9:34 am
Lea,
I’m glad Dan answered you because I didn’t really understand your question.
Dan,
Interestingly, the levels in the EPIC report were not fasting. I thought this was odd so I did a quick search and found some sources saying that fasting for a cholesterol test is possibly not necessary:
http://www.webmd.com/cholesterol-management/news/20121108/fasting-may-not-be-needed-before-cholesterol-test
I didn’t track the study down to look into it. The authors didn’t address whether this could affect their results.
December 23rd, 2013 at 7:07 am
These data are fascinating.
From the INTERHEART study of heart attacks in 52 countries, the strongest predictor of the risk of myocardial infarction was the apoB:ApoA1 ratio, with no threshold below which risk did not decrease – at least down to an apoB:apoA1 ratio of 0.43 (and probably lower, as these are based on deciles).
In the Bradbury paper (EPIC-Oxford), the mean and 95% CI apoB:apoA1 ratios for male and female vegans respectively were: 0.57 (0.54–0.59) and 0.49 (0.47–0.51).
In INTERHEART, the apoB:apoA1 ratio had the highest population-attributable risk for heart attacks out of any risk factor tested (54%).
Thus in EPIC-Oxford, the vegans are not fully protected from heart attacks, as they are clearly not achieving the median apoB:apoA1 ratio of the first decile (0.43). This is disappointing, but it should also be noted that they likely have other protective factors (BMI, smoking, hypertension, diabetes) – although all of these are adjusted for in the INTERHEART study with measurement of the population-attributable risk.
It should be noted that risk increases in a graded fashion, from the first decile at 0.43 to the highest decile at 1.28.
References:
1) Lipids, lipoproteins, and apolipoproteins as risk markers of myocardial infarction in 52 countries (the INTERHEART study): a case-control study.
McQueen MJ, Hawken S, Wang X, Ounpuu S, Sniderman A, Probstfield J, Steyn K, Sanderson JE, Hasani M, Volkova E, Kazmi K, Yusuf S; INTERHEART study investigators. Lancet. 2008 Jul 19;372(9634):224-33. doi: 10.1016/S0140-6736(08)61076-4.
2) Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study.
Yusuf S, Hawken S, Ounpuu S, Dans T, Avezum A, Lanas F, McQueen M, Budaj A, Pais P, Varigos J, Lisheng L; INTERHEART Study Investigators.
Lancet. 2004 Sep 11-17;364(9438):937-52.
3) Bradbury KE, Crowe FL, Appleby PN, Schmidt JA, Travis RC, Key TJ. Serum concentrations of cholesterol, apolipoprotein A-I and apolipoprotein B in a total of 1694 meat-eaters, fish-eaters, vegetarians and vegans. Eur J Clin Nutr. 2013 Dec 18. [Epub ahead of print]
December 23rd, 2013 at 9:26 am
Lea,
Like most physicians, he or she was probably ordering a fasting cholesterol which is not physiologic and really represents the carrot in the snowman, rather than the snowman itself. Meaning, fasting lipids are the tip of the iceberg. Postprandial/postchezial/post-meal cholesterol is where the money’s at. Unless one is measuring apoB/apoA1, which does not change with fasting or recent meal intake….
We spike our cholesterol, triglycerides and chlyomicrons every time we eat, but we don’t capture these surges with an overnight fasting level. Hence we are always just measuring the carrot rather than the snowman.
December 23rd, 2013 at 10:17 am
Agreed. However, for people whose triglycerides levels go up after a bolus of fat, risk for a short-term cardiovascular event is increased by 3-4-fold (Homburg Sugar and Spice Study). But for apoB and apoA1 it probably doesn’t matter, as these are relatively insensitive to fasting state
December 23rd, 2013 at 10:19 am
Dan,
But in your comment to Lea, you say, “Postprandial/postchezial/post-meal cholesterol is where the money’s at,” which sounds like you think there is a big difference between fasting and post-meal cholesterol levels.
December 23rd, 2013 at 10:47 am
Well cholesterol is notoriously unreliable, whether fasting or post-meal. Even LDL cholesterol is unreliable as it’s a calculation (not a measurement) that relies on guesswork assumptions about the status of other lipids. There is up to 20% variability in measurement depending on fasting vs non-fasting status.
None of these considerations apply to apoB:apoA1 measurement, which in my view is the premier lipid variable. I would be interested to know whether other studies directly measured apoB or apoA1 in vegans or vegetarians.
December 23rd, 2013 at 10:50 am
> I would be interested to know whether other studies directly measured apoB or apoA1 in vegans or vegetarians.
I don’t recall ever seeing it before.
For what it’s worth, my cholesterol levels have been measured fairly consistently over the years among many different labs. I think most of them were fasting.
December 23rd, 2013 at 2:07 pm
I seem to have completely contradicted myself in the above series of quotes. Not unusual, as my brain is foggy today.
There is not much variation between fasting and postprandial lipids (with the exception of triglycerides), but strictly speaking, for cholesterol, LDL, HDL, etc, there is little variation. However, what variation there is tends to have prognostic importance, and this is especially so for a group of individuals known as cholesterol hyperabsorbers (of course if you are on a strict vegan diet with no processed foods containing things like liquid whole eggs then the cholesterol hyperabsorption issue is moot). Triglyceride spikes after meals are also very strong predictors of cardiovascular events.
Saturated fat is the other stimulator of serum cholesterol and is a more potent provocateur than dietary cholesterol. If you think about it, all those little postprandial spikes in serum fats do add up over time, and may in fact be missed by a fasting sample. Most people eat 3 meals per day plus snacks, and that is a much more physiological state than a 12-14 hour overnight fast. A person who has small postprandial increases in serum lipids would be expected to have a greater risk for atherosclerosis than a person who has no increase in postprandial lipids, because atherosclerosis is a gradual, incremental process and not caused by any one meal, and because small differences in lipids even in the fasting state appear to have marked effects on long-term risk. This is especially true at the population level. The same is also true for postprandial glucose surges, which typically show up about 10 years before fasting glucose is elevated in type 2 diabetes.
Thus while I seem to have contradicted myself, I may not have entirely refuted what I first said about the carrot and the snowman. That phrase is from a memorable editorial in the Canadian Journal of Cardiology about postprandial lipids (in case you thought I was making all this up!).
December 24th, 2013 at 3:56 am
Dan, do you have references for RCT that show SFA (or fat in general) increases the amount of postprandial lipoproteins?
December 24th, 2013 at 9:23 am
Here are only the 3 latest studies indexed in Medline. There are hundreds.
1) Short-term, high-fat diet increases the expression of key intestinal genes involved in lipoprotein metabolism in healthy men.
Tremblay AJ, Lamarche B, Guay V, Charest A, Lemelin V, Couture P.
Am J Clin Nutr. 2013 Jul;98(1):32-41. doi: 10.3945/ajcn.113.060251. Epub 2013 May 29.
“After the 3-d high-fat diet, plasma cholesterol, LDL cholesterol, and HDL cholesterol concentrations were significantly higher than concentrations observed after the low-fat diet was consumed.”
2) J Atheroscler Thromb. 2013;20(6):591-600. Epub 2013 Apr 19.
Simultaneous ingestion of fructose and fat exacerbates postprandial exogenous lipidemia in young healthy Japanese women.
Saito H, Kagaya M, Suzuki M, Yoshida A, Naito M.
“The simultaneous ingestion of fructose and fat markedly enhances postprandial exogenous lipidemia in young healthy Japanese women.”
3) Increased blood cholesterol after a high saturated fat diet is prevented by aerobic exercise training.
Ortega JF, Fernández-Elías VE, Hamouti N, Mora-Rodriguez R.
Appl Physiol Nutr Metab. 2013 Jan;38(1):42-8. doi: 10.1139/apnm-2012-0123. Epub 2012 Nov 1.
PMID: 23368827 [PubMed – indexed for MEDLINE]
” Increases in T(C) and LDL-C concentration induced by 14 days of HSFAD can be prevented by concurrent aerobic exercise training, which, in addition, improves cardiorespiratory fitness.”
December 24th, 2013 at 12:37 pm
Thanks Dan, but none of these studies is what I am looking for. (1) and (3) did not measure postprandial response. Study (2) does measure it, but it has a major flaw in my opinion: the test meals are not isocaloric. Also, even though that fat meal contained more calories that the glucose/fructose meals, there was no statistically significant difference in Delta-AUC-apoB (which I believe is the most important measure).
So, to be more precise, I am looking for RCT comparing either SFA to Carb/other fats or Fat to Carb, and uses isocaloric meals. So far I have seen one such study (http://onlinelibrary.wiley.com/doi/10.1002/mnfr.201200452/abstract) and in this study there was no statistically significant difference between the high SFA meal vs the high carbs meal in either TG or apoB response.
December 27th, 2013 at 9:02 pm
By my calculation, the effect of the difference in SFA/PUFA/fiber consumption does not explain the entire difference in non-HDL cholesterol: It explains 15.5 out of the 26.3 (adjusted) difference when comparing meat eaters vs. vegan, and 6.5 out of 19.3 when comparing vegetarian vs. vegan.
So where does the remaining difference come from? Possible explanations are:
1) Dietary cholesterol (but the effect of dietary cholesterol is relatively small if you consume a “normal” amount of it).
2) The reported consumption of SFA/PUFA/fiber is not accurate.
3) The meta-analysis I used for computing the effect of SFA/PUFA/fiber are not accurate. For example, it is possible that the studies on which these meta-analysis are based were too short.
4) Chance.
5) Selection bias. It is possible that vegan that know they have high cholesterol were less likely to join the study.