Another man's poison

The medical profession has established beyond doubt that high cholesterol contributes to heart disease. What it won't accept is that low cholesterol can be dangerous too-causing violent and suicidal behaviour in some people. Matt Ridley explains why going back to fry-ups may be good for us
January 20, 1996

One day in 1987, two scientists were walking down the street in Washington, DC, when one stopped to buy the New York Times. It carried news of the latest study of heart disease, in which 2,051 people in Helsinki had been given a drug to lower their cholesterol levels, compared with 2,030 who had been given a placebo. The results of the study were only mildly encouraging. Fourteen of those on the drug had died of heart disease in five years; 19 of those on the placebo. But what caught Stephen Manuck's eye was something else.

"Look at this bit," he said to his companion, Jay Kaplan. "Yet another study finds more people dying violent deaths after taking the cholesterol-lowering drugs." Ten people had died of "non-illness mortality" (medicspeak for accidents, violence and suicide) compared with four in the placebo group. The ensuing conversation was the start of a detective story which leads to the heart of human behaviour and has some disturbing things to teach us about following doctors' advice. Very low cholesterol levels might be just as bad for us as very high ones-but in different ways. It is a message the medical profession is extremely reluctant to let us hear.

Manuck's colleague at the University of Pittsburgh, Matt Muldoon, went back through the six large trials of cholesterol-lowering then published, and found his hunch confirmed. They were all studies in which one group of mostly middle-aged men had been given a drug or a diet, compared with another group given a placebo. In every study, more of those given the drug or diet had died violent deaths than in the control group. Only four of the six studies showed a reduction in heart-disease deaths. Taking all the studies together, cholesterol treatment reduced the risk of dying from heart disease by just 14 per cent. It raised the risk of accidents, violence, and suicide by an astonishing 78 per cent.

Two years later Muldoon added another two studies to his portfolio and found the effect still impressive. Only "secondary studies"-those in which the patients were already suffering from heart disease when the studies began-proved immune to the "violent death" effect. All primary studies showed a substantial increase in violent deaths among those taking the anti-cholesterol drug or diet. Of course, because more people die of heart disease than in accidents, the overall effect was little or no change in total mortality. The additional violent deaths almost exactly cancelled out the lives saved from heart disease. None the less, the increase in violent deaths was more than ten times as significant statistically (i.e. ten times less likely to be as a result of chance) as the fall in mortality from heart disease. And violent deaths sometimes involve innocent bystanders.

This fact should not have come as a surprise. It has been known for 20 years that impulsive, anti-social and depressed people-including prisoners, violent offenders and failed suicides-have generally lower cholesterol levels than the population at large.

Meanwhile, Kaplan went back to his laboratory at Bowman Gray Medical School in North Carolina and dug out the results of a trial he had carried out on Macaque monkeys. He had kept 15 monkeys on a rich, fatty diet for 22 months and given another 15 the sort of diet recommended by the American Heart Association. The cholesterol levels of the "luxury" monkeys soared to 468 milligrams per 100 millilitres-coronary level in man. The cholesterol counts of the "prudent" eating monkeys collapsed to 145. Kaplan had recorded the behaviour of the two groups but had never compared them. Now he did so. There was a stark difference. The "prudent" monkeys were far more aggressive than the "luxury" monkeys; much more liable to hit, bite or chase each other.

This was too much of a coincidence, argued Manuck, Muldoon and Kaplan. Perhaps some people, too, become more impulsive, aggressive or rash when their cholesterol level drops. Perhaps this leads a few of them into violence, risk-taking and suicide. The medical establishment preferred to explain away the effect, arguing variously that it was a fluke or a side effect of something else. Cholesterol reduction is the jewel in the cardiologist's crown, the best piece of biochemical advice gleaned from thousands of man hours spent on heart disease. It represents a huge vested interest, not only because so many doctors have pinned their colours to its mast, but because cholesterol-lowering drugs and cholesterol tests are big business. Kaplan was actually abused over the telephone by one drug company executive.

But the heretic scientists were not arguing against the link between high cholesterol and heart disease. Nor am I. There is no longer any doubt that cholesterol-lowering drugs prevent heart disease. The new generation of anti-cholesterol drugs, the statins, are much more effective than previous ones. In two big recent trials they established beyond doubt their ability to save lives. Deaths from coronary heart disease were 42 per cent and 28 per cent lower in those treated, as against in those taking the placebo.

Furthermore, neither study showed any increase in violent death. The establishment heaved a collective sigh of relief. "We can now forget about the increase in non-cardiovascular mortality," said the unofficial dean of cholesterol studies, Professor Michael Oliver of London's Royal Brompton Hospital. He faxed Muldoon a message to that effect.

Up to a point, professor. If two inconclusive studies could invalidate a theory, the link between heart disease and cholesterol would have been abandoned long ago. Besides, one of the two studies featured only those men already suffering from heart disease-where the violence effect is not predicted-and the other, carried out in Scotland, featured only people with high cholesterol levels. It is among normal, healthy people that the effect shows up-presumably because some people with low cholesterol are made violent by having it lowered still further.

Even adding the new studies to the previous ones, the increase in violent death remains statistically stronger than the reduction in heart disease mortality. As the data now stand, if you wish to accept that the studies prove that cholesterol lowering prevents heart attacks, then you have to accept that it also kills about as many people in other ways.

Professor Oliver responds that you would never tell somebody with low cholesterol to lower it further in any case, and certainly would not put him on a drug to do so. Nobody with low cholesterol is ever likely to take a statin. So, in that sense, the drugs are as safe and efficacious as you could want them to be.

But hold hard. Are we not all under constant bombardment to change our diets to lower cholesterol, whatever our age, weight and state of ignorance? Eating less cholesterol is one of those things experts tell us repeatedly and stridently to do. Their advice should carry a government health warning: low cholesterol could get you violently killed.

The medics' riposte to this is ironic: there's nothing wrong with recommended diets because diets don't have any effect in any case. "The usual recommendation for these men at lower risk is to decrease dietary unsaturated fat intake and increase the ratio of polyunsaturated to saturated fat, but this is often ineffective outside clinic conditions," wrote Professor Oliver in a recent issue of the Lancet. Read carefully between the lines and you can even find this admission in official warnings. The Department of Health recently put out a document which said: "Everyone should follow the general guidelines for healthy eating, which will help to reduce the population average level of serum total cholesterol." Note that subtle reference to the population average: no promises that everyone's cholesterol will come down.

So mild low-cholesterol dieting has almost no effect on most people's cholesterol levels; and severe dieting is almost impossible for most people to stick to. If your cholesterol level is below 210, then it is unlikely to fall by more than 2 per cent if you eat a low-cholesterol diet. Only those with high cholesterol levels in the first place have much to gain from diets and drugs.

What goes down with difficulty also goes up with difficulty. If you feed people very high-fat and very high-cholesterol diets, in 90 per cent of cases it has no effect at all. Their cholesterol levels do not budge. But in the other 10 per cent there is a marked increase. The difference is probably genetic. The truth is that heart disease is looking almost more like a genetic problem than an environmental one. The 10 per cent of people who react to high-fat diets by raising their cholesterol levels are the ones who need treating and telling-not the rest of us.

Sauce for the goose, sauce for the gander. If the population as a whole needs warning about high cholesterol, just to catch the vulnerable few at the top end, then the whole population needs warning about low cholesterol-just to catch the vulnerable few at the bottom end. Remember that the death rate in middle-aged men was roughly the same with or without treatment for cholesterol. Violence and heart attacks cancelled each other out. In young men, with generally lower cholesterol and generally higher risk-seeking behaviour, low cholesterol might be far more dangerous than high cholesterol.

The health establishment is caught in a bind of its own making: either it argues that eating less cholesterol is ineffective, or it admits that it could be dangerous for just as many people as it could be life-saving. "I took my kids off skimmed milk," says Kaplan.

Having failed to fault the heretics' science, some doctors question their motives. To be sure, there might be mischief makers about, employed by the meat industry whose interest in the debate is vested and untrustworthy. Jay Kaplan firmly denies any cause for bias. "I'm tenured," he says, "I really don't care where my research takes me." The truth is that if even half as much money had gone into investigating the effects of low cholesterol as that spent on investigating high cholesterol, the evidence might well look different.

Moreover, in the last three years a new piece of evidence has transformed the debate. Until 1992, no study had found a link between low cholesterol (as opposed to falling cholesterol) and behaviour. Since then three huge studies of ordinary people not being treated for cholesterol (a total of 700,000 people) have come in. The results could not be clearer. Those with low cholesterol were found to be much more likely to commit suicide or suffer some other form of violent death. The 25 per cent of men with the lowest cholesterol count were four times as likely to commit suicide as the 25 per cent of men with the highest cholesterol. (There is no similar pattern in women.) The most convincing of these results is the so-called MrFit trial, in which 351,000 people from seven countries have been followed for six years. The graph of total mortality against cholesterol levels in these people is as clear as a bell: people with very low or very high cholesterol are both almost twice as likely to die as people with medium cholesterol. So it is not just lowering cholesterol that violently risks lives, but having low cholesterol in the first place.

one reason why doctors are so reluctant to believe that low cholesterol causes violent death is that it seems crazy to suggest that somebody crashed a car or jumped off a bridge because of his diet. A con- nection between diet and heart disease is much easier to envisage. Psychological problems must be cured by talk, not food. Brains are not considered parts of bodies.

Yet brains are parts of bodies, and bodies are full of cholesterol. Cholesterol, to most people, is like dioxin or arsenic: a nasty poison. But not only is it one of the commonest chemicals in our bodies, an essential component in every cell membrane; it is also manufactured on site. Most of the cholesterol in your body is made, not eaten. That is why it is so hard to affect blood cholesterol by keeping it out of your diet. Your body simply makes more of the stuff to compensate. The ingredients, after all, are the simplest imaginable: carbon, hydrogen and oxygen.

It is also the raw material for the manufacture of some rather interesting hormones. Testosterone, oestrogens and the stress hormone cortisol are all just modified forms of cholesterol. Cholesterol, in other words, is a rather special and intriguing substance. It lives with fats but is not one; it is made into hormones but is not one; it kills in excess but kills in deficit, too.

And it seems to have an accomplice. Detectives sniffing out the trail of cholesterol in the monkey's brain have come across a familiar substance: serotonin. Serotonin is just about the most fashionable chemical in human biology right now, a sort of biochemical Princess of Wales. You cannot keep it off the medical front pages.

Serotonin is a neurotransmitter, a chemical which passes signals between nerve cells in the brain. It is a vital component of mood. As an absurd generalisation, the more serotonin in your brain, the better you feel. Or at least, so goes the reasoning behind Prozac, the main effect of which is to keep serotonin levels high outside nerve cells by preventing them from reabsorbing it.

The uncanny thing about serotonin is that when supplies run low it produces many of the same symptoms as low cholesterol: depression, suicide, aggression and irritability. It is now generally accepted that people with low serotonin levels are more impulsive, while people with high serotonin levels are more compulsive. You cannot measure serotonin directly, but a sample of cerebrospinal fluid contains a chemical derived from it known as 5-hydroxyindoleacetic acid (mercifully abbreviated to 5-HIAA), and its concentration is a good proxy for serotonin itself.

By this measure, people with "severe personality disorder" are characteristically impulsive and have low serotonin levels-and the lower their levels, the more aggressive they are. A series of studies by Markku Linnoila and others have linked low serotonin with violence and impulsiveness in criminals. Murderers and those who have attempted murder have low serotonin levels, especially those whose crimes were impulsive rather than premeditated. Other violent offenders have low serotonin levels-again excepting those who had premeditated their acts. Arsonists, too, have low serotonin levels, even excluding those intent on suicide. And attempted suicides themselves are characteristically low in serotonin.

In other words, if everybody had to display their serotonin level on their foreheads at all times, we could tell who should be avoided, incarcerated or protected from themselves. Fortunately, perhaps, this idea is as wrong as it is offensive. The reason is that serotonin levels are not innate or inflexible, but are themselves a barometer of experience. Serotonin is, roughly speaking, a measure of esteem. Low serotonin leads to low self-esteem; and low self-esteem leads to low serotonin. The arrow goes both ways. Telling people they have low serotonin could become a self-fulfilling prophecy.

back to monkeys. Male vervet monkeys are hierarchical animals with a strict pecking order in the troop. The alpha male gets his way over the beta; and so on down the pecking order. High-ranking males have high serotonin levels and are less impulsive than lower-ranking males. As a male rises up the hierarchy, its serotonin level increases, and when it falls from grace so does its serotonin.

Extrapolating from monkeys to men is fraught with problems. For one thing, male hierarchies are much more complicated in apes than in monkeys and still more so in naked apes. But the link between serotonin and impulsive actions seems to hold up pretty well in chimpanzees and people as well as in monkeys. It has been shown, for instance, that senior members of college fraternities have higher levels of serotonin; and Prozac famously makes people less irritable and explosive.

So to convince the jury of the dangers of low cholesterol, we need to establish three links: the link between low cholesterol and low serotonin; the link between low serotonin and impulsive behaviour; and thus the link between low cholesterol and impulsive behaviour. In monkeys Kaplan has demonstrated all three links. His last study compared eight monkeys on a low cholesterol diet with nine on a high cholesterol diet. Both groups had similar, generous amounts of total fat in their food. As predicted, the low-cholesterol monkeys were 40 per cent more likely to be aggressive and anti-social than the high-cholesterol ones. They also had serotonin levels nearly half as high as the high-cholesterol monkeys.

Kaplan's latest study, not yet published, is even more convincing. It shows that the same monkeys become more aggressive when put on low cholesterol diets and less aggressive when put on high cholesterol diets, however stable their social groups. And, unlike in human beings, the effect is equally strong in females as in males.

Quite how cholesterol affects serotonin is still mysterious. The best the boffins can say is that serotonin has to get in and out of cell membranes through special channels to do its job, and cell membranes change their properties drastically when starved of one of their essential components: cholesterol. But this applies to all the nerve cells in the brain, not just those communicating by serotonin.

We do not know one vital fact. Is the link between low cholesterol and impulsive behaviour something which can affect us all, or is it just a problem for a minority with a genetic predisposition to low cholesterol or to violence? Common sense suggests the latter-but the monkey experiments are less reassuring. Kaplan can turn almost any monkey violent by giving it too little cholesterol in its food. But then monkeys, which normally eat fruit, are poor models for humans, the most carnivorous primate.

A murder conviction requires a corpse, a suspect, a witness and a motive. The corpse we have: low cholesterol increases the risk of violent death, and so does artificially lowered cholesterol-in men and monkeys. The suspect is a sinister chemical with strange connections. The witness is that chemical's already convicted accomplice, serotonin. And the motive? Why should the body react impulsively to low cholesterol? Kaplan points the finger at evolution. Back in the stone age a low and falling cholesterol level was a fairly good indicator that starvation was imminent. Fighting harder for a greater share of resources-taking risks-would be a sensible strategy to adopt. After all, in early days our ancestors ate much wild meat: a cholesterol-rich but fat-poor diet. But you can see the doctors' dilemma. How can they tell us that low cholesterol is potentially bad for us without sparking a rush back to bacon and eggs and a steady rise in coronary heart disease?