What Genes Can’t Tell Us: An Exchange

To the Editors:

Richard Lewontin’s review of Evelyn Fox Keller’s The Mirage of a Space Between Nature and Nurture [NYR, May 26] is, as usual, full of wisdom and erudition. I must, however, ask Professor Lewontin to expand on the sentence “The study of genes for specific diseases has indeed been of limited value.” For whom has the value been limited?

It is has certainly not been of limited value to those individuals who have, as a result of genetic research, been found not to carry mutations in certain genes that have been repeatedly and convincingly associated with a very high risk for familial forms of cancer and other serious conditions. This is easy to document: one just has to ask these persons and they will tell you of the value to them, their spouses, and their offspring. For those who do carry such disease-associated mutations, there is at least the chance of prevention, early detection, and better treatment, which were not possible for those family members born one hundred years ago.

Again, analyses of pedigrees tell their own story. It is not that these conditions are rare. For example, almost one quarter of women diagnosed with ovarian carcinoma between the ages of forty and forty-nine carry a deleterious mutation in the cancer susceptibility genes BRCA1 or BRCA2,¹ and the frequency of these mutations is probably twice as high among Ashkenazi Jewish women with ovarian carcinoma diagnosed in the same age interval.² Moreover, this is not “a few classical cases.” Ovarian cancer is a common disease, affecting approximately one in seventy women in North America.

Of course, knowledge of the mechanisms of action of these genes and their products lags far behind our ability to identify these mutations, but just like for cholera, one does not always have to know the mechanism of action of a disease-associated agent to introduce effective measures against it.

Many, myself included, recognize the serious limitations of a purely reductionist genetic approach to human health, and it is easy to overstate the significance of genetic findings, but babies and bathwater come to mind when reading Professor Lewontin’s views on recent research into the cause of human disease. The value of genetic research into the causes, prevention, early diagnosis, and treatment of specific diseases inevitably depends on one’s perspective, but there are concrete examples of benefit.

William D. Foulkes
McGill University
Montreal, Canada

To the Editors:

Richard Lewontin’s review of The Mirage of a Space Between Nature and Nuture is almost perfect. But our most eminent population geneticist misunderstands clinical genetics. In saying that “the study of genes for specific diseases has…been of limited value,” that is, the genetic dissection of rare Mendelian diseases such as Marfan Syndrome or achondroplastic dwarfism, Lewontin throws the baby out with the bathwater.

If we define “value” as science that illuminates eukaryotic biology, informs treatment strategies, and significantly impacts public health, the study of just one of thousands of single-gene human disorders, familial hypercholesterolemia, would be sufficient. The work begun by Michael Brown and Joseph Goldstein to understand why certain families have very high cholesterol levels and premature death from heart disease has become a major scientific and commercial enterprise comprised of basic scientists, drug discoverers, and clinicians, all of which led to the development of “statins” used by millions of patients to mostly good effect.³

Indeed, the study of relatively rare single-gene disorders should be a national research priority, if not the only diseases studied by genome sequencing, because of what we are likely to learn, especially when compared to the studies that occupy the university genomic factories in which late-onset common diseases with little genetic contribution drive the research agenda. On the other hand, the Mendelian conditions are typically caused by one gene mutation that “overpowers” any environmental influences, presenting to the physician as recognizable syndromes and therefore read on basic biology. Alas, it is much harder work to carefully characterize such patients clinically than to collect the DNA of tumor cells for industrial sequencing but we should make that extra effort. As the record clearly shows, there is much to be learned from these natural errors of development and metabolism that will impact public health.

New York Review readers deserve a more nuanced view of current genetics and genomics. It is not the study of genetics or DNA that is a false science; what matters is the choice of traits and whose DNA we study. Like other abused technologies (cell phones, the Internet), sequencing DNA is here to stay.

Hugh Young Rienhoff Jr., MD
San Francisco, California

To the Editors:

Richard Lewontin explains why studying the relationship between DNA and disease has not been successful. This failure is indeed disappointing, but in the science of treating these diseases there is more reason to be sanguine.

Humans display substantial variability in the genes that orchestrate production of drug-metabolizing enzymes, in some cases resulting in dramatic alterations in drug response. For example, the commonly used drug codeine is converted to its active metabolite (morphine) by the isozyme cytochrome P4502D6 (CYP2D6), and people who are genetically deficient in CYP2D6 usually have reduced analgesic response to codeine.

A more dangerous situation occurs, however, when codeine is used in people with genetically increased CYP2D6 activity. The increased activity is due to gene multiplication, and such people (called “ultra-rapid” metabolizers) are at increased risk of morphine toxicity when they take codeine. Indeed, in one tragic and oft-cited case, a healthy breastfeeding newborn developed fatal morphine toxicity, because his mother was an ultra-rapid metabolizer of CYP2D6 and was taking codeine. Her milk was tested and found to contain toxic concentrations of morphine.

Another drug for which genetic polymorphisms can be critical is tamoxifen, commonly used in the treatment of breast cancer. Tamoxifen, like codeine, is a prodrug that must be converted by CYP2D6 in the body to an active metabolite. As one might expect, women taking tamoxifen for breast cancer who are genetically deficient in CYP2D6 have a higher cancer recurrence rate. Unfortunately, many commonly used medications inhibit CYP2D6, so even people with no genetic deficiency in CYP2D6 are at increased risk of breast cancer recurrence if they take such drugs with tamoxifen.

Numerous other examples could be cited of drug toxicity or reduced drug efficacy due to genetic variability in drug-metabolizing enzymes (and endogenous drug transporters). So while the search for genes that cause common diseases has not been fruitful, genetic polymorphism research has become catnip for those in the fields of pharmacokinetics and pharmacodynamics. If the results of this research were applied consistently in health care settings (not yet a reality for a variety of reasons), enhanced safety and efficacy of medications would be the result.

Philip D. Hansten
Professor Emeritus
University of Washington
Port Ludlow, Washington

Richard Lewontin replies:

All three of our correspondents are concerned with my claim that the genetic studies of human diseases and tests carried out to determine whether individuals carry certain gene mutations have been of “limited value.” Dr. Foulkes asks, appropriately, “For whom has the value been limited?”

I do not doubt that a woman’s discovery that she carries a mutation for the BRCA1 or BRCA2 gene is of considerable value to her since she then knows that she has a high probability of developing ovarian or breast cancer and may want to choose a traumatic surgical intervention. The “value” discussed in my review, however, was the extent to which a knowledge of genetic mutations has led to an understanding of the root metabolic and developmental causes of various diseases and to new and more effective and less traumatic preventive or curative measures.

The answer is “very little.” The reason is that, in technologically advanced countries, the main pathologies leading to mortality are heart disease, cancer, and stroke, which are the result of complex developmental and physiological causes that involve many interacting processes. Even though someone with a mutated BRCA1 gene has a very high probability of developing breast or ovarian cancer, only about 20 percent of women affected by those diseases turn out to have a mutated gene. What are all the other causes? Moreover, the very high risk of developing certain cancers for those women who do carry certain mutations is not the usual situation, although a few other cases are known. Typically, as documented in the issue of The New England Journal of Medicine cited in my review, where some association is found between a gene mutation and a disease, the added disease risk is less than 1 or 2 percent. The fault lies not in our genes but in our whole selves.

I am not familiar with the literature described by Professor Hansten but it is quite plausible and reasonable and, indeed, expected on a straightforward biochemical basis. Genetic variation in the structures and amounts and even presence of various enzymes certainly exists in populations and would often arise from single mutations in the relevant genes. Such mutations would indeed have the effects on the reaction to and efficacy of drug treatments noted by Hansten. It is the confusion between this sort of simple molecular variation in single molecules and the etiology of most diseases that has given rise to the ineffectual search for the “genes for disease.”