Having access to embargoed news from the Annals of Internal Medicine is really fun. I get to report on important studies at the same time as the news media. But this week, I got my hands on a study that I'm not sure will be widely reported (Mozaffarian et al. Trans-palmitoleic Acid, Metabolic Risk Factors, and New-Onset Diabetes in US Adults. Ann Internal Med. 2010). Why? Because it suggests that dairy fat may protect against diabetes.
The lead author is Dr. Dariush Mozaffarian, whose meta-analysis of diet-heart controlled trials I recently criticized (1). I think this is a good opportunity for me to acknowledge that Dr. Mozaffarian and his colleagues have published some brave papers in the past that challenged conventional wisdom. For example, in a 2005 study, they found that postmenopausal women who ate the most saturated fat had the slowest rate of narrowing of their coronary arteries over time (2). It wasn't a popular finding but he has defended it. His colleague Dr. Walter Willett thinks dietary fat is fine (although he favors corn oil), whole eggs can be part of a healthy diet, and there are worse things than eating coconut from time to time. Dr. Willett is also a strong advocate of unrefined foods and home cooking, which I believe are two of the main pillars of healthy eating.
Let's hit the data
Investigators collected two measures of dairy fat intake in 3,736 Americans:
- 24 hour dietary recall questionnaires, six times. This records volunteers' food intake at the beginning of the study.
- Blood (plasma phospholipid) content of trans-palmitoleate. Dairy fat and red meat fat are virtually the only sources of this fatty acid, so it reflects the intake of these foods. Most of the trans-palmitoleate came from dairy in this study, although red meat was also a significant source.
Even though certain blood fatty acids partially represent food intake, they can also represent metabolic conditions. For example, people on their way to type II diabetes tend to have more saturated blood lipids, independent of diet (3, 4)*. So it's reassuring to see that dietary trans-palmitoleate intake was closely related to the serum level. The investigators also noted that "greater whole-fat dairy consumption was associated with lower risk for diabetes," which increases my confidence that serum trans-palmitoleate is actually measuring dairy fat intake to some degree. However, in the end, I think the striking association they observed was partially due to dairy fat intake, but mostly due to metabolic factors that had nothing to do with dairy fat**.
Here's a nice quote:
Our findings support potential metabolic benefits of dairy consumption and suggest that trans-palmitoleate may mediate these effects***. They also suggest that efforts to promote exclusive consumption of low-fat and nonfat dairy products, which would lower population exposure to trans-palmitoleate, may be premature until the mediators of the health effects of dairy consumption are better established.
*Probably due to uncontrolled de novo lipogenesis because of insulin resistance. Many studies find that serum saturated fatty acids are higher in those with metabolic dysfunction, independent of diet. They sometimes interpret that as showing that people are lying about their diet, rather than that serum saturated fatty acids don't reflect diet very well. For example, in one study I cited, investigators found no relationship between dietary saturated fat and diabetes risk, but they did find a relationship between serum saturated fatty acids and diabetes risk (5). They then proceeded to refer to the serum measurements as "objective measurements" that can tease apart "important associations with diabetes incidence that may be missed when assessed by [food questionnaires]." They go on to say that serum fatty acids are "useful as biomarkers for fatty acid intake," which is true for some fatty acids but not remotely for most of the saturated ones, according to their data.
**Why do I say that? Because mathematically adjusting for dairy and meat fat intake did not substantially weaken the association between phospholipid trans-palmitoleate and reduced diabetes risk (Table 4). In other words, if you believe their math, dairy/meat fat intake only accounted for a small part of the protective association. That implies that healthy people maintain a higher serum phospholipid trans-palmitoleate level than unhealthy people, even if both groups eat the same amount of trans-palmitoleate. If they hadn't mentioned that full-fat dairy fat intake was directly associated with a lower risk of diabetes, I would not find the study very interesting because I'd have my doubts that it was relevant to diet.
***I find it highly doubtful that trans-palmitoleate entirely mediates the positive health outcomes associated with dairy fat intake. I think it's more likely to simply be a marker of milk fat, which contains a number of potentially protective substances such as CLA, vitamin K2, butyric acid, and the natural trans fats including trans-palmitoleate. In addition, dairy fat is low in omega-6 polyunsaturated fat. I find it unlikely that their fancy math was able to tease those factors apart, because those substances all travel together in dairy fat. trans-palmitoleate pills are not going to replace butter.