The GMO Salmon Struggle
FDA May for First Time Ever Approve Genetically Modified Animal for Human Consumption
I met Elliott Entis, the entrepreneur and former president of AquaBounty, more than 15 years ago at a symposium on food products derived from genetically engineered plants and animals. Even then he was sure that regulatory approval of his fast-growing transgenic salmon, AcuAdvantage, was “just around the corner.” After a decade and a half, federal approval of these fish has recently met opposition from a group of eight senators from salmon fishing states on food safety and environmental grounds. I will review each of these claims in turn, but first, a quick review.
Transgenic salmon are not the first animal product derived from genetic engineering. That would be the transgenic mouse, an animal developed for biomedical research and now widely utilized in many different custom formulations. It is not even the first fish. That would have been the “Glowfish,” a zebrafish transformed to fluoresce under ultra-violet light.
AquaBounty salmon have been genetically transformed to grow faster than natural salmon, provided that copious amounts of feed are readily available. They are among a handful of genetically engineered products meant for human consumption, and they have long been projected to be the first edible transgenic animal that will receive regulatory approval.
Food safety concerns
When the Food and Drug Administration held public meetings on the AquAdvantage salmon last fall, the briefing document prepared for the Veterinary Medical Advisory Committee concluded that there was no basis for regarding the salmon’s genetic transformation as a risk factor for food safety. Not surprisingly, some critics are not satisfied, and have called for animal studies and clinical trials. But such studies are of little value in the context of food safety. Here is a quick review of why that is so.
Animal studies on drugs or on food additives (such as coloring agents), probe the relationship between dose (e.g. the amount of a substance) and response (e.g. observed toxic or other harmful effects). The methodology of animal studies assumes that the response varies with the dose. This allows researchers to probe for toxicity or carcinogenicity by increasing the dose well beyond normal, and making the inference that a proportionally lower dose would have a proportionally smaller response.
Although researchers realize this methodology can overstate risk, they nonetheless view the result as appropriately precautionary. Administering an unrealistically large dose of a particular additive allows them to err on the side of caution because they can detect risks that would not be observed even in large populations of lab animals receiving a normal dose. If, for example, you are looking for a response in 1 out of 100,000 individuals exposed to the normal dose, you would need to observe several million rats in order to obtain a statistically significant result. If the dose response relationship is consistent you can find it by raising the dose 1000 times and observing several dozen rats. But, unlike additives, researchers cannot feed an animal 1,000 times the amount of food it would normally eat. It would explode. So this method is unavailable for identifying potential hazards accruing from unsafe foods such as salmon.
Clinical studies have also proven to be of limited value in probing the effects of consuming a particular food. Clinical studies of drugs, for example, involve one or more groups who are administered the substance under study, and one or more groups who are not. This latter group, the “control,” may be administered a placebo, or undergo other procedures (such as visits with a physician or treatments with other established drugs) to ensure they are relevantly similar to those who receive the treatment.
But food testing is more difficult because people tend to eat lots of different things and eat them in combinations that make effective control of a clinical population impossible. It would be like every person in the test was taking dozens of different drugs in addition to the one being tested, making it impossible to determine whether safety (or efficacy) could really be traced to the drug in question. And people are not particularly adept at recording and reporting what they have eaten, in any case. As a result, decades of clinical studies on the effects of eating salty or fatty foods are still less than fully conclusive in their results.
So how do regulators decide whether a food is safe? One part of the answer is that they actually don’t—at least, not in a long-term way. While it may sound surprising, the FDA conducts zero testing for long-term health effects of the food products it approves for consumption. While specific chemicals or drugs added into foods are tested, foods more generally are not tested in the same way. At the risk of rankling the overly-cautious-about-their-language regulators at FDA, I would characterize their approach to food safety in terms of using common sense: if food were highly toxic, animals (or people) who ate it even once would get visibly ill. FDA classifies food items we have been eating for a long time under the heading Generally Recognized As Safe, or GRAS, but this does not mean that there has ever been a scientific certification of their safety.
For example, we know that the green parts of potato plants are toxic because cattle and humans have gotten sick eating them. The long and short of it is that potato breeders have to be careful, and they generally are. They will feed new varieties to animals and try small amounts themselves to see if toxic bits have gotten into the tubers. Since Entis was reporting that he had eaten his transgenic fish when I met him back in the ‘90s, we can presume that this kind of gross testing has been done. Another part of the answer is that we have a long list of the chemical constituents that are normally in a given food, and it is possible to run assays that would highlight a major difference (though they wouldn’t pick up anything they weren’t looking for).
The transgenic salmon case is tricky because FDA does not claim authority to regulate transgenic fish as a food, but as a veterinary drug. For normal drugs that are prescribed in small amounts it would be possible to run classic toxicological and clinical tests. So critics are right to state that the FDA is not requiring tests for the salmon that would be required for other animal drugs. FDA could conceivably require such testing on the active ingredient the genetic modification produces, salmon growth hormone, which is present in all salmon.
The documents released last fall summarize the scientific basis for FDA’s not requiring those tests. I would summarize that reasoning thus: humans have centuries of experience eating salmon growth hormone (whenever they eat salmon). The amount of salmon growth hormone in a transgenic fish is actually not much greater than a regular salmon; it’s the timing of growth hormone production that accounts for faster growth. Hence, there’s no hazard to look for and no point in doing animal studies or clinical trials.
While it is unlikely Aquabounty’s salmon will have negative health effects, the potential for environmental risk from fast growing fish is a more complicated story. One complication concerns the environmental risks of ocean fish farming in general, without regard to whether fish are genetically engineered or not. The combination of feed, feed additives (including pharmaceuticals), and fecal waste associated with ocean fish pens can have aquatic effects not unlike those of a Concentrated Animal Feeding Operation, or CAFO, on land. Activists who oppose fish farming oppose transgenic fish because they fear that it will only make a bad situation worse. But U.S. regulators have consistently held that such arguments are too broad to provide a basis for differentially regulating a transgenic organism.
Bill Muir, a geneticist at Purdue University, identified another complication: a scenario in which the gene for rapid growth spreads and results in the sudden and irreversible collapse in the population of wild relatives. The mix of genetics and ecology needed to sort out the likelihood of Muir’s “Trojan gene hypothesis” is quite tricky, but Muir himself is satisfied that his scenario is exceedingly unlikely to occur in the case of AquAdvantage salmon.
Here’s a nontechnical version of that story. Fast-growing salmon are voracious eaters. The genetic transformation means that they need access to a large and stable food supply in order to grow. Wild salmon grow more slowly and can tolerate much greater variability in food supply. So when the modified salmon escape their ocean net pens (and some inevitably will), they are singularly ill equipped to survive in the wild where food is more scarce. They would have to be especially wanton breeders (and salmon are not) to overcome this debility and successfully spread their genes in the wild population. To combat this issue, AquaBounty has taken additional precautions to ensure sterility in these transgenic fish, and established physical barriers to prevent their escape. But it’s this fitness argument that convinces me of their environmental safety.
FDA approval and the future of transgenic foods
There are always lingering questions after a risk assessment, and reasonable disagreement about whether those lingering questions should be subjected to further scrutiny before a project moves ahead. AquAdvantage’s critics in the NGO community have been especially attentive to the fact that the fish may serve as precedent for other products of genetic engineering. To help address some of these concerns, the FDA posted a summary of the information collected for the review of AquAdvantage Salmon at last fall’s meeting of their Veterinary Medicine Advisory Committee as well as a website of “Frequently Asked Questions.”
The fact that such painstaking work has been done to assess the risks of fast-growing salmon has led biotech industry insiders to predict that the FDA will approve the AquaBounty product on the grounds that it lacks any credible reason to deny it. As noted, these insiders have been predicting approval for a long, long time. But this does not mean that people want to eat fast-growing transgenic salmon, and the salmon industry fears that the combination of “yuck factor” responses, lingering fears, and general confusion will not be good for sales of any salmon, transgenic or not.
This is not so much a scientific story. Consumer reaction is always a blend of emotion, economics, and incomplete information. No one has developed a reliable test for determining how volatile that blend will be in any given situation. The interesting unwritten story will be to see whether those economic uncertainties trump the food safety and environmental risk assessments in the political process. It is an open question as to whether they should.
Paul B. Thompson is the W.K. Kellogg Chair in Agricultural, Food and Community Ethics at Michigan State University.
Comments on this article