Lewontin’s balanced review of books on genetically modified food [“Genes in the Food!” NYR, June 21] fails to explain that such foods may become vital for the survival of large sections of the rising world population. Since the early 1960s, food available per head in the developing world has risen, despite a doubling of the population. This outstanding success has been achieved by the introduction of crops improved by crossing and by intensive application of fertilizers, pesticides, and weed killers. Even so, there are 800 million hungry people and 185 million seriously malnourished preschool children in the developing world. It is unlikely that the methods that have raised cereal yields hitherto will allow them to be raised again sufficiently to reduce these distressing numbers. In fact, agricultural production in India can already no longer keep up with the continuing population rise.

Since most fertile land is already intensively cultivated, scientists are trying to introduce genes into crops that would allow them to be grown on poorer soils and in harsher climates, and to make existing crops more nutritious. In the tropics, fungi, bacteria, and viruses still cause huge harvest losses. Scientists are trying to introduce genes that will confer resistance to some of these pests, enabling farmers to use fewer pesticides. Genetically modified plants offer our best hope of feeding a world population that is expected to double again in the next fifty years. It will be tragic if the present outcry over genetically modified foods in the rich countries were to discourage further research and development in this field. As Lewontin stresses, no one has yet been harmed by eating such foods.

M.F. Perutz

Medical Research Council

MRC Laboratory of Molecular Biology

Cambridge, England

To the Editors:

The review “Genes in the Food!” states that “despite the fears about the humaningestion of the products of genetic engineering, no one appears to be worried about the large number of diabetics who are injecting bacterially produced insulin twice a day. As far as anyone knows, no one has been harmed by this product of genetic engineering….”

There are a number of errors in this excerpt of your review. In 1995, Humulin insulin, manufactured by Eli Lilly, was the eighth-most-reported drug for adverse effects in the United States. In Canada, rDNA insulin brands are responsible for at least 460 reports of adverse drug reactions, including eight deaths linked to the use of synthetic insulin. In Britain, the British Diabetics Association (now Diabetes UK) reported that up to 20 percent of diabetics in that country were unable to use synthetic brand insulins efficaciously. The UK-based Insulin Dependent Diabetes Trust (IDDT) and the US-based Diabetics International Foundation (DIF) both were founded to advocate in favor of “insulin choice” after animal-sourced products were removed from the markets in their respective countries. In Canada, where I live, over two hundred people have joined together to investigate the possibility of a class-action suit against Eli Lilly and Novo Nordisk because they have experienced serious health problems after converting to the synthetic insulin brands manufactured by the two companies.

Synthetic or “human” insulin was introduced to the world in the early 1980s. In the case of Humulin, the clinical trial history leaves a lot to be desired. The first molecule was cloned in 1979 by Genentech, clinical trials began in 1981, and the product, Humulin, was on the market in early 1983. A few weeks after approval was granted, Dr. Solomen Sobell of the FDA reported that in clinical testing of the synthetic insulin, human antibodies formed despite expectations that Humulin would not cause an immune response. To my knowledge, although Sobell said that follow-up clinical trials were needed (and that these would take years), Phase III and IV trials were not conducted.

Thus, your assertion that “no one appears to be worried” is incorrect. There are literally many thousands around the world, in the UK, Australia, New Zealand, India, the US, Canada, Switzerland, Germany, France, Finland, Brazil, Argentina, and I’m sure many, many other countries who are demanding the continuance of animal insulin products because they have experienced negative reactions to the synthetic brands. These reactions range in severity from hypoglycemia unawareness and memory loss, to severe weight loss, diarrhea, rashes, edema, and other negative experiences up to and including death. Many attribute these reactions to problems in the process of manufacturing the synthetic product itself, while others are pointing to an immune response that may have been responsible for the development of diabetes in the first place (e.g., an autoimmune response to “human” insulin).

I would suggest that to expand your research on this subject you visit the Web sites of IDDT at www.iddtinternational.org and of DIF at members.tripod.com/diabetics_world. Both Web sites provide links to a great many other groups who are organizing to support “insulin choice.”


Finally, the days when diabetics took two injections a day are long gone, except for those who continue on beef or beef/pork insulin. Synthetic insulin, among many of its lower-quality characteristics, does not last as long as the bovine variety.

Colleen Fuller

Vancouver, British Columbia


To the Editors:

Richard Lewontin had a number of interesting and important things to say in his review of current controversies over genetically modified foods. Unfortunately, his characterization of my own contribution (Appendix A of the National Research Council report) is wrong. He calls that appendix a cost-benefit analysis and asserts that I “clearly felt that the sorts of regulation recommended by the report were not worthwhile.” He goes on to object generally to the use of cost-benefit analysis, arguing that it fails to take into account the fact that costs and benefits accrue to different parties and that health effects and some environmental effects can’t be expressed in monetary terms.

I signed the report because I agreed with its findings and recommendations. From an economist’s perspective, though, the report’s exclusive focus on potential benefits of regulation, without regard for cost, makes it quite one-sided. My appendix attempts to provide a degree of balance by looking at one question: whether stricter pre-market regulatory review could impede the entry of competitors to the big multinational agrichemical firms dominating the pesticide and seed industries. I conclude that the costs of regulatory testing could be prohibitive for public-sector researchers, biotech startups, and small breeders and seed companies but are unlikely to deter big agrichemical companies. This analysis provides the basis for the report’s final two recommendations.

Professor Lewontin’s dismissal of cost-benefit analysis in general is too hasty. Tradeoffs between private profit and protecting health and the environment aren’t the only ones involved in policy decisions about genetically modified plants or other forms of environmental regulation. Requiring more extensive regulation of, say, genetically modified plants without significantly expanding staffing and funding for regulatory agencies like the EPA and FDA inevitably means less time and money spent on other ways of protecting the environment and human health. For instance, diverting FDA staff time and funds from microbial pathogens to genetically modified foods will almost certainly not lead to a net improvement in public health and safety.

More generally, cost-benefit analysis is useful as a means of understanding the potential consequences of policies—which is fundamental to making informed choices. At bottom, benefits are simply consequences that are good in some way while costs are those that are somehow bad. Assessments of benefits, costs, and (as Professor Lewontin rightly notes) how those benefits and costs are distributed among members of society are essential to our ability to hold informed debates about what course of action to take. Being able to express all benefits and costs in monetary terms is not essential. Saying that cost-benefit analysis is misguided because of a belief that some benefits and costs are incommensurable misses another important point. Making a choice implies in and of itself that the benefits and costs involved have been weighed against each other. Again, whether they are weighed in money terms is immaterial.

Professor Lewontin’s depiction of technical change as a treadmill leading independent family farms to ruin is also questionable. In his view, technical change forces farmers to rely increasingly on purchased farm inputs, leading to a squeeze between rising costs and falling prices, mounting debts, bankruptcies, and growing dependence on off-farm work. The evidence does not support such a stark perspective. Because of technical progress, farming generates more income today than in the past. Farmers may spend more per acre on chemicals and other inputs, but yields have grown fast enough that costs per unit of output have fallen more than prices. Except for the mid-1980s, rates of financial distress have not been abnormally high; even during that period, small farmers weren’t more likely to experience financial distress than large ones. Exit from farming has slowed to a crawl: while the number of farms in the US fell by almost 60 percent between 1950 and 1974, it has remained roughly constant at around two million since then. Fifty years ago, small operators working off the farm (as a quarter of all farmers did) were much poorer than the US population as a whole. The average household income of the bottom 20 percent of farmers in 1950 was only about 40 percent that of the overall US average, compared to 90 percent of the US average today—despite the fact that this group apparently loses several thousand dollars a year from farming. It seems more likely that these small operators stay in farming as a lifestyle choice rather than being forced to work off-farm to make ends meet.

Erik Lichtenberg

Professor, Department of Agricultural and Resource Economics


University of Maryland

College Park, Maryland

Richard Lewontin replies:

Max Perutz is certainly right that third-world agricultures should not be deprived of the possible gains in productivity that might accrue from the use of recombinant DNA methods. It is not clear, however, that these are the only methods by which such gains can be made. The immense effort put into agricultural research in rich countries in the past is immeasurably greater than that expended on third-world agriculture and it seems likely that conventional breeding methods still have a good deal to offer Asian and African agriculture. Moreover, as I pointed out, recombinant DNA methods may be very destructive of those economies by allowing the present production of special value-added commodities like fats and oils or caffeine in the third world to be replaced by their synthesis in American and European crops. Whether, on balance, genetic engineering will make poor countries better or worse off remains to be seen.

Colleen Fuller (as well as two other correspondents who wrote me directly) has called attention to a situation of which I was completely unaware when I wrote my review. It has a special interest because it illustrates how the introduction of a technology may, through a chain of circumstances, lead to harmful consequences even when the immediate product of the technology is not itself harmful. Ever since the introduction of insulin therapy for diabetes about seventy years ago, diabetics have had the problem of adjusting the dosage of injected insulin in order to reduce their blood sugar to an acceptable level while not reducing it to such a low level that they suffer from insulin shock. The consequences of abnormally low levels of blood sugar are a general depression of metabolism, drowsiness, loss of balance, and, in extreme cases, coma and even death. To keep a balance, diabetics may monitor their blood sugar level as many as five times a day and respond with insulin injections when necessary. It appears that the problems of keeping a balance when so-called “human insulin” is used are considerably greater for some people than was the case when insulin was extracted from pigs and cows. Whether the proportion of diabetics who have severe difficulties with the new insulin is, in fact, greater than with the old has not yet been determined from a large statistical study, but cow and pig insulin is no longer available as an alternative, so it is not clear what is to be done for those who suffer from severe hyperinsulism.

It might be thought that the new “human” insulin ought to be better than the old, since the amino acid sequence of the molecule being produced by the bacteria, using the human insulin gene, is identical to normal human insulin while cow and pig insulins differ by a few amino acids from the human. The problem, however, is that genes do not make proteins. They only specify the sequence of amino acids in the protein, while the protein itself is a folded molecule and the same amino acid sequence can be folded in many different ways. When “human” insulin was first produced by bacteria, it turned out that the bacteria folded the molecule incorrectly, producing an insulin that was physiologically inactive and therefore useless as a drug. To solve this problem, the manufacturers harvest the incorrectly folded protein from the bacteria, treat it with heat to unfold it, and then refold it by an industrial process that is a trade secret.

It is this chemically engineered molecule that is sold under the name of Humulin. As Fuller says, we do not know whether there are adverse reactions because there are impurities introduced in the manufacture, whether the refolded molecule is too different from normal human insulin, or, paradoxically, because it is too much like the real thing and that some diabetes is actually the consequence of an allergy to insulin. Whichever of these explanations is correct, it is not the product of the genetically engineered bacteria that is causing the problem, but the result of an industrial reprocessing of a physiologically inactive biologically produced protein.

There is yet another possibility that is more fundamental in its implications for replacement drug therapies. The production by the body of physiologically active molecules, such as hormones, is under constant feedback control through complex pathways of metabolism so that the level of these molecules in the body is in constant flux in response to the internal demand for them. When, in an abnormal individual, such a molecule is not produced and we supply it externally at arbitrary moments and in amounts that do not correspond closely to the demand for it, we will sometimes cause an overshoot of the metabolic process involved. That has always been a problem with insulin therapy, for example. Suppose that Humulin is, in fact, just as active, or, worse still, more active than the insulin produced normally in our bodies. Then insulin shock would be more common than it was when less active, physiologically imperfect pig and cow insulins were used. We might make it a therapeutic rule of thumb that it is always better to use a less active rather than a more active drug on the grounds that it is easier to avoid harmful overdoses.

Professor Lichtenberg did indeed sign the NRC report, which indicates that, on balance, he was willing to be part of a consensus. His fellow members of the committee, however, did not reciprocate, having gone out of their way to distance themselves from his cost-benefit analysis by relegating it to an appendix which “was not part of the committee’s consensus report” but was “authored by an individual committee member” with whom “the committee as a whole did not necessarily agree.” Unfortunately, Lichtenberg’s letter does not offer us any more insight into what must have been an interaction of great interest to the student of the political process.

Professor Lichtenberg’s measured description of his analysis and of cost-benefit analysis in general must leave the uninformed reader completely perplexed as to what conceivable objection the committee could have had to his work. The reasoning process he describes is indistinguishable from the kind of weighing of advantages and disadvantages through which we must all go in trying to make a decision about a course of action. The economist’s claim for cost-benefit analysis, however, is rather more ambitious. It is supposed to offer us a quantitative tool for making a balance sheet of cost and benefit so that we can decide which is larger and by how much. Presumably in the Gradgrindian world of narrow rationality that has made this brand of economics infamous, we will choose our action based on whether the resultant number is positive or negative. But to make such a quantitative judgment requires a common numeraire that can be added and subtracted to produce the net worth of the action.

This is usually, although not necessarily, money, so that one is reduced to finding a dollar value of pain, suffering, and social power. Moreover it must be the net worth to an identifiable party who will bear the costs and reap the benefits. Economists will often identify the party as “society” (Professor Lichtenberg refers to “public” health), but that mystical abstraction hides the real heterogeneity of interest among real people and the impossibility of reaching a consensus on how much each per-son’s interest is worth. This is the rock on which Bentham’s utilitarian ship long ago foundered. The model of cost-benefit analysis is the calculation of whether it is worthwhile for a corporation to close a factory. One can certainly calculate the dollar value to the company. The impact on the town or workforce is left out of the calculation since they are conveniently regarded as “externalities.”

I do not know what to make of Professor Lichtenberg’s apparent initial disagreement with my analysis of the economics of agriculture. If farmers are doing better than ever, then why has the proportion of farmers working off the farm risen in the last fifty years from 25 percent (his figure) to the present 60 percent and why, in his words, do farmers have to “work off-farm to make ends meet”? Again he reports an increase in total household income, yet seems to acknowledge that this is irrelevant to the issues since it confounds income from farm product marketings, income from outside employment, and a significant income from federal subsidy payments. The question is why more farmers than ever are losing money as farmers and find it necessary to work on assembly lines in order to hold on to their farming “life-style.” The answer is that there is an increasing crisis of overproduction in farming as a consequence of the historical changes in agricultural technology, as I explained in my review.

I have another letter, not reproduced here, from Belinda Martineau that raises the issue of food labeling, one that I did not discuss. Martineau was a member of the team that developed the genetically engineered Flavr Savr tomato and she has written a very revealing book on her experience, First Fruit: The Creation of the Flavr SavrTM Tomato and the Birth of Biotech Food (McGraw-Hill, 2001). She writes in her letter:

And working from the logical premise that engineered foreign proteins added to our food are food additives begs another question: shouldn’t these food additives be listed on US food labels?… What better way to be transparent to a wary public than to label these foods?… Regulating genetically engineered proteins as food additives and labeling foods containing them would allow the public to vote on that policy with their pocketbooks. Isn’t that the least we should expect in a democratic, capitalistic society?

Indeed. But, in addition, we need to try to explain what genetic engineering is really all about.

This Issue

October 4, 2001