Polyunsaturated Logic

I recently started reading Metabolic Regulation: A Human Perspective, by Keith Frayn, because I wanted to dig in further to the science discussed in Gary Taubes' Good Calories, Bad Calories. Not that I don't believe what Gary is saying, but it's always better to get your information from the proverbial horse's mouth. Anyway, Metabolic Regulation is turning out to be a good book, explains things reasonably well, and I'll be posting interesting tidbits in the future. The science of metabolism is mostly big words. There aren't really any hard concepts, so if you can get past the vocabulary, you can learn a lot about how your body works (or at least the current understanding at the time the book was written).

This is the second time I've gotten into an advanced book on this topic, and in both cases have noticed something very interesting. Despite going into incredible detail about metabolic processes, qualifying the conclusions appropriately (e.g. "this result holds in a test tube, but may not hold in a complete living organism), they still push certain dogmatic ideas without any comparable level of evidence. It amazes me that the authors, who otherwise present information in excruciating molecular detail, don't choke when they start preaching dogma without evidence. Weird.

The major piece of dogma is the connection between saturated fat, cholesterol, and heart disease, the so-called lipid hypothesis. It's not difficult to make arguments that these connections are not causal (e.g. saturated fat does not cause heart disease), but are rather erroneous inferences from experiments or observations of varying design quality. Malcolm Kendrick lays out a pretty solid case against the lipid hypothesis in his book The Great Cholesterol Con; you can also read about it here. When I get into the textbooks, I expect them to support the lipid hypothesis with similar detail, perhaps even greater, providing insight as to the key molecular mechanisms by which saturated fat raises serum LDL, how that LDL then exerts a direct causal effect in the development of heart disease.

But instead, I get nothing. Just recitation of the hypothesis, and no evidence beyond the usual epidemiological handwaving, really not much more than you would get from a newspaper article. Metabolic Regulation at least acknowledges the weakness of the evidence (still almost all epidemiological), though still proceeds as though the lipid hypothesis were proven, even giving a mathematical expression for the relationship between serum cholesterol and dietary fats of different saturation indices.

Given the widespread belief that saturated fatty acids (SFA) raise LDL, and that polyunsaturated fatty acids (PUFA) lower LDL, you'd think somebody would have nailed the biochemistry by now. Talk about a killer drug target: "Take Satfatium, eat all the butter and bacon you want, and keep your cholesterol normal!" They could even use Robert Jarvik as the spokesperson! Talk about a cash cow. Despite the obvious and overwhelming motivations to nail down the SFA->LDL mechanism, I haven't been able to find diddly-squat, other than some weak hypotheses having to do with SFA binding to LDL or cholesterol receptors. That seems unlikely, given the relative lack of chemical and physical similarities between SFA and cholesterol (both are lipids, but that's about as far as it goes), and the relevant parts of LDL for receptor binding are proteins, not lipids. But could it happen anyway? Possibly, but it seems like an easy thing to study in a test-tube. Scientists study all kinds of cellular receptors in vitro, why not this one? Maybe it has been studied, showed bupkus, and never got published. Most scientists are averse to publishing negative results, especially when they contradict the prevailing dogma.

It's easy to play mental games and generate hypotheses; however, that your hypothesis supports some other well-believed hypothesis doesn't make it more true than one which contradicts widely believed ideas. That's backward thinking. Scientific truth is always conditional: belief in one thing is always conditioned on the truth of other information. Thus, the degree of belief in the lipid hypothesis is conditioned on the hypothesis that SFA raises LDL and that LDL exerts a causal effect on the development of heart disease: the lipid hypothesis is only true GIVEN that SFA raises LDL AND LDL causes heart disease. If any of the conditioning statements are false, the rest of it falls apart. This sort of conditional truth is generally ignored in science, partly because of the ass-backwards way most scientists do statistics, where they actually consider the evidence that the data would have been observed GIVEN that the hypothesis were true, rather than the evidence that the hypothesis is true given the data. And no, those are not mathematically the same thing.

I'll finish this post with a little brain-bubble that occurred to me. This is total speculation, and really just shows how one can create nice-sounding hypotheses from limited information. Anyway, eating more PUFA supposedly lowers total serum cholesterol. Eating more SFA supposedly raises total serum cholesterol. I don't believe either of these statements has anywhere near unequivocal experimental proof (feel free to post evidence otherwise), as the results seem to be all over the place; but let's pretend they're true, like nearly everybody else does. Now, one hypothesis as to why PUFA lowers serum is that PUFA displaces monounsaturated fatty acids (MUFA) in cell membranes. Cell membranes contain both SFA and MUFA, along with a little PUFA. The ratios of these are thought to govern the "fluidity" of the cell membrane, because as we know, SFA can pack more closely together, and thus tend to be more solid at higher temperatures. Cholesterol plays a similar role, also promoting membrane rigidity.

Now the PUFA theory we're discussing holds that eating a lot of vegetable oil raises the PUFA content of the cell membrane. It's not good if the membrane is too fluid, so in the absense of saturated fat the body reacts by driving cholesterol into the cells to increase their rigidity, thus decreasing serum cholesterol. So maybe the opposite happens as well: eat lots of SFA, cell membranes become too rigid, so they release cholesterol into the blood, and total serum cholesterol rises. Beautiful, logically consistent, and also has the "benefit" of supporting the prevailing dogmatic belief in the lipid hypothesis.

Of course, it's all a house of cards. My hypothesis depends on the PUFA/cell membrane hypothesis, for which there is little or no proof. Unfortunately, scientists are incredibly bad at sorting out these logical relationships, for whatever reason, and hand-waving hypotheses like mine often wind up widely considered as "true". I think what happens is that people neglect that actual evidence. They look at the PUFA hypothesis, think "that seems reasonable", and follow the logical chain that "proves" the SFA hypothesis. "Seems reasonable" is not evidence, though, and that distinction is too often lost on the very people who should know better.