Teaching overenthusiastic CAM advocates a little bit about gene expression profiling

Yesterday, I was depressed. Today I'm a little irritated.

I'm irritated because I came across a study from a couple of weeks ago that's actually a really cool study that applies actual science to the question of how diet and lifestyle changes might alter biology to improve health. It's exactly the sort of study that can apply help understand how diet affects health. It's a study by Dean Ornish, who's widely known for his advocacy of a lifestyle-driven approach to treating atherosclerotic coronary artery disease and producing evidence in the early 1990s that such a lifestyle alteration could not only halt disease progression but actually partially reverse it. This lifestyle intervention included a low-fat vegetarian diet, smoking cessation, regular exercise, and meditation. Although his approach is interesting, attempts to replicate it have been mixed. For example, Dr. Frank Sacks, a nutrition professor at Harvard Medical School and the Harvard School of Public Health. Dr. Sacks, tried to replicate Dr. Ornish's results with a grant from the NIH but found it difficult to recruit patients because few could stick it out for the whole program. In other words, Dr. Ornish's treatment is very difficult to follow and not suitable for many, if not most people.

A couple of weeks ago, Dr. Ornish published a study in the Proceedings of the National Academy of Sciences of the USA that examined the effect of his particular set of dietary and lifestyle interventions on the development of prostate cancer. The study, entitled Changes in prostate gene expression in men undergoing an intensive nutrition and lifestyle intervention. As I've discussed before, prostate cancer is in most men a relatively indolent cancer. Indeed, in autopsy series, most men who die after the age of 80 of something other than prostate cancer have evidence of cancer in their prostate glands. Many prostate cancers never progresses fast enough to endanger a man's life, especially when it's detected through PSA screening. The problem is, we don't know how to differentiate between dangerous prostate cancer that needs to be treated and indolent cancer that does not, nor do we yet have any good treatment strategies to keep such tumors in check.

Before I give you my take on this study and its findings, let me first show you a quote.This one is by our favorite advocate of a "more fluid" concept of evidence, Dr. David Katz:

But we have tended to think in terms of "Nature versus nurture," as lifestyle and genetic influences on health as independent, and potentially competing forces. This study changes the game. It suggests that lifestyle and genetics are not independent after all, but interact. Even our genes are influenced by lifestyle choices.

[...]

The lifestyle intervention in GEMINAL was rather intense, allowing only 10% of calories from dietary fat, and requiring more than an hour and a half of exercise and meditation daily. We don't yet know if less intensive lifestyle approaches would influence genes as this program did. And the current work is, admittedly, only a pilot study, limited to 30 men with prostate cancer. More research will be required to prove what it suggests. But what it suggests is quite provocative enough for now: take good enough care of yourself, and even your genes will get a makeover.

Healthful behaviors are not a mere attempt to bluff our way through whatever genetic hand we were dealt. They are, it turns out, an opportunity to reshuffle the genetic deck in our favor.

No, no, no, no.

Nothing in this study--and I mean nothing--demonstrates that any deck has been reshuffled, genetic or otherwise, or that this diet gives anyone's genes a "makeover." I can't believe this sort of ignorant statement came from a physician! Clearly, Dr. Katz does not understand the difference between a gene and gene expression, or if he does he's dumbed it down far more than necessary to make it sound as though diet and exercise are doing something magical and never before seen. They're not. He also argues against a straw man in that I bet he'd be hard-pressed to find a physician who actually says that lifestyle and genetic influences on health are "independent." They're clearly not, although they can certainly be competing, as in a person with a genetic predisposition to atherosclerosis who exercises. In this case, that person's healthy lifestyle is indeed to some extent competing with or fighting against his body's natural tendency to develop that disease.

But, irritated or not, Orac is a benevolent skeptic, at least most of the time. That is why an explanation of this study is right up my alley. Let's look at what the study actually found. In this study, called the Gene Expression Modulation by Intervention and Lifestyle (GEMINAL) study, thirty-one men (not thirty, as is usually reported--more on that later) were enrolled in the study, in which they underwent a prostate biopsy before and after a three month intervention in which they adhered to Dr. Ornish's rather rigorous diet and lifestyle changes. They were also followed for a number of health measures, such as weight, blood pressure, and a number of other parameters. The diet itself was almost vegan, with a radically decreased intake of meat and fat, plus increased doses of fruits and grains, supplemented with vitamins C and E. In addition, the men participated in physical exercise and stress-reduction intervention, walking and/or working out for 30 minutes six times a week. They also did yoga stretches, breathing and meditation sessions for an hour every day, and every week they also took part in one hour group support sessions.

One thing that should also be noted about these men is that they are a highly select group. 273 men from the GEMINAL study were screened for this pilot study, of which 146 did not meet inclusion criteria, and 96 declined to participate. In other words, only 31/127 eligible men. Whatever the reasons were (probably the two prostate biopsies combined with the major change in lifestyle), less than a quarter of those eligible agreed to participate. Another thing that should be noted was that these patients were also a highly select group from another perspective. They all had small, low grade tumors that were deemed safe for "watchful waiting," rather than surgery or radiation therapy. These tumors all had low PSA levels and low Gleason scores. (The Gleason score is the standard measure used by pathologists to estimate how aggressive a prostate cancer is.) hese men were also monitored very carefully to make sure that their tumors were not progressing during the study. It should also be noted that tumors did not regress, nor did overall PSA levels change on average.

Consistent with many prior studies, the intervention had a positive effect on several measures of health, including blood pressure, weight, and lipid profile. PSA levels even tended to stabilize. That, of course, was not the "sexy" part of the study that got it news coverage. What got this study in the news was the part of the study where each biopsy was subjected to gene expression analysis. Here's where Dr. Katz's statement grates. Dr. Katz makes it sound as though this lifestyle intervention somehow "reshuffled the genetic deck." That implies a major change in the genes. What was really found was not so dramatic, although nonetheless interesting, but to understand it you need to understand what gene expression is. Gene expression is how much transcript (or messenger RNA, abbreviated mRNA) each gene makes, which ultimately directs the production of the functional protein that carries out a function in the cell. Gene expression is thus an estimate of whether a gene is "turned on" and making its RNA product or not and how "turned on" it is. (I say "estimate" because although mRNA levels usually correlate with the level of the protein whose production the gene and RNA direct, such is not always the case. There are genes whose mRNA level don't always correlate with protein levels.) Consequently, what a gene expression profile, such as the ones done for this study, measures is the simultaneous level of expression or activity of thousands of genes, It's a powerful tool that over the last decade has revealed a number of fascinating insights into the biology of tumors, particularly breast cancer.

What this study showed is that 48 genes were upregulated (mRNA level increased) and that 453 were downregulated (mRNA level decreased). Moreover, the genes that were downregulated tended to be associated with cancers (such as oncogenes, or cancer-causing genes), and the ones upregulated tended to be associated with tumor suppression. Because expression profiling can produce numerous false positive "hits" for genes that change with an intervention, the levels of some of the transcipts were "spot-checked" as a means of verification. The implication of this study is that diet can produce profound changes in gene expression in the prostate. That such changes would be observed was not in itself particularly surprising, but what would have been difficult to foresee is how many genes potentially important to tumor initiation and progression were downregulated.

This study was a pilot study, of course, and had a number of limitations that aren't apparent from the news reports. The first (and probably most minor one) is that one 1/31 patients showed evidence of tumor progression while on the study and thus underwent surgery. Inexplicably, he was not included in the analysis, even though it would have been very easy to get another biopsy at the time of surgery for the post-treatment biopsy. No good explanation is given for why this patient wasn't included.

However, that was relatively minor compared to the major problem with this study. One that I noticed right away is that only 1/3 of the samples contained actual tumor; the rest were all noncancerous prostate tissue. This wasn't reported in the Results section and was only mentioned briefly in the Discussion section. It's a good thing that this was mentioned in the discussion but not enough information was given to allow a good assessment of whether this shortcoming introduced significant bias. Indeed, it would be rather important to know what proportion of matched tumors contained no tumor in the pre-treatment biopsy or no tumor in the post-treatment biopsy. If, for example, more matched pre and post sets contained tumor in the pre-treatment biopsy but no tumor in the post-treatment biopsy than contained tumor in the post-treatment biopsy but no tumor in the pre-treatment biopsy, that could introduce an artifact that would bias the results towards the sort of results seen. The reason would be because in essence normal prostate in the post-treatment result would be compared with cancerous prostate in the pre-treatment biopsy, thus giving the appearance that the intervention shifted the gene expression profile of a cancer towards normal prostate. In any case, this is a huge problem with the study. The authors try to argue that the effects seen could be beneficial on normal prostate tissue and thus prevent the development and progression of cancer from the prospective of the tumor microenvironment (the local environment in the vicinity of the tumor). However, the whole hypothesis of this study was that this intervention would inhibit the progression of the tumor itself.

On another level, the entire design of this study is a really messy affair, a hodge-podge of a lot of different simultaneous interventions. Many of the gene changes observed could very well be a consequence of weight loss and decreased fat intake more than anything else. Whether the extra exercise and all the supplements, including vitamin E, selenium, vitamin C 2 gm/day (a big dose), fish oil (3 gm/dy), and soy (tofu at 1 meal/day, plus 58 gm/dy of soy protein), have anything to contribute to this effect is impossible to tell. My guess is probably not. It would have been much less difficult to interpret if the most likely part of the intervention to be of benefit, namely the diet, were tested first without all that extra supplementation. Another problem with the study is that, other than the one patient whose tumor progressed, the results of PSA levels were reported in aggregate rather than individually. Thus, we have no way of knowing if some men's PSA drifted up (but not enough to provoke intervention after three months) and some men's PSA drifted down, thus producing a wash. There was also no attempt apparently made to correlate clinical status with changes in gene expression in individuals, although in fairness the numbers were probably too small for such an analysis to be particularly meaningful.

Getting back to Dr. Katz, there is nothing in this study that implies a "reshuffling" of the genetic destiny of these men. What is happening, if this study is repeated and validated, is that an exogenous influence has altered gene expression according to the preexisting genetic predisposition of the individual. No change in nature versus nurture need be postulated any more than a reshuffling. We've known that drugs cause changes in gene expression like this, why not dietary alterations that can result in the change of a number of hormones in the body?

Oddly enough, Dr. Ornish himself was fairly reasonable about the results of the study:

"We found that simple changes have a powerful impact on gene expression," Dean Ornish, founder and president of the Preventive Medicine Research Institute and clinical professor at the University of California, San Francisco (U.C.S.F.), said during a news conference. "People say, 'Oh, it's all in my genes, what can I do?' That's what I call genetic nihilism. This may be an antidote to that. Genes may be our predisposition, but they are not our fate."

[...]

"People say, 'Why bother?' But when they see that in just three months these changes can make a difference, they may change their minds," he said. "It is not really so much about risk-factor reduction or preventing something bad from happening. These changes can occur so quickly you don't have to wait years to see the benefits."

I would point out that most physicians would not accept that one's genetic makeup is the be-all and end-all of health or that there is nothing people can do to improve their health using lifestyle and diet. Given the impressive advances in our understanding of the genetic basis of disease and chronic health problems in the last two decades, it may well be that the general public sometimes gets that impression, although I think Ornish may be overplaying his hand in using the claim of "genetic nihilism" to promote his study. On the other hand, unlike Dr. Katz, he at least appears to have some understanding of the difference between "genetic destiny" and maximizing one's chances of avoiding disease based on one's genetic background. It's all a matter of doing what you can with what you have, and, even if this study is completely accurate in recapitulating the biology of prostate cancer, that's all it says. It says nothing about how one's genetic background affects a man's risk of developing prostate cancer, only that maybe--just maybe--diet and lifestyle interventions may have a significant impact on just how that genetic predisposition plays out.

REFERENCE:

Ornish, D., Magbanua, M.J., Weidner, G., Weinberg, V., Kemp, C., Green, C., Mattie, M.D., Marlin, R., Simko, J., Shinohara, K., Haqq, C.M., Carroll, P.R. (2008). Changes in prostate gene expression in men undergoing an intensive nutrition and lifestyle intervention. Proceedings of the National Academy of Sciences, 105(24), 8369-8374. DOI: 10.1073/pnas.0803080105