Current Models of Risk Assessment Used in Biotechnology Regulation

Charles L. Elkins

Charles L. Elkins is Director, Office of Toxic Substances, U.S. Environmental Protection Agency.

I am not a biotechnology expert, but I will provide my personal views on the subject, arising from my involvement with the U.S. Environmental Protection Agency's (EPA) biotechnology regulation, through its Office of Toxic Substances.

First, I would like to give you my own perspective on the current status of EPA's biotechnology regulation. Second, I will consider an example of risk assessment, and finally, I will discuss the future direction of EPA's regulation.

What is the problem we face? Although in one sense biotechnology is a new economic endeavor, in another sense it is as old as the human race. There is some risk to biotechnology, but it is hard to quantify because there are so many unknowns.

Some groups outside the government, as well as some inside, worry that policymakers will overreact to potential risks; others are concerned that the policymakers will underreact. From the policymaker's perspective, these concerns might seem irrelevant to scientific work. But while the risks the public perceives from this new endeavor may or may not match reality, the policymaker must still consider them.

"Risk" is a term that the general public does not understand well, and a topic on which professionals and nonprofessionals do not communicate very well. In my office we are in charge of a massive new social experiment under the Superfund Amendments and Reauthorization Act (SARA) n1, called "Community-Right-To-Know."² On July 1, about 1.5 million pages of documents will flow into our office, which we will then put onto a computer so that the public can learn what all manufacturers in this country are emitting into the environment.

The Chemical Manufacturers Association (CMA) has put much effort into improving risk communication. CMA realizes that society needs better risk communication; it is an inexact science. We should invest enough time in risk communication and in understanding public perception of biotechnology, rather than dismissing it as the sole responsibility of public relations officials.

One must not only make the right decision, but one must communicate it well. Keeping public perception in mind, consider the radiation issue, with which this country has had a great deal of experience. Nuclear power is no longer a budding technology. Yet there is an excess of public fear about nuclear power, perhaps out of proportion to the risks.

The Three Mile Island incident certainly brought great credibility problems to both the industry and the regulators, [19 ELR 10497] since both had assured the public that this problem would never arise. What does this have to do with risk? The nuclear power industry will never be the same because of this incident. The biotechnology industry cannot sustain a Three Mile Island-type incident, particularly this early in its history.

This illustrates that, although they sit on different sides of the table in any discussion of risk, industry, government, and the public actually have a common interest. They share a strong interest in sound science, open decisionmaking, and good risk communication. None can afford the attitude that they should only tell the public those things it needs to know, and that the public does not need to know much about biotechnology. That will not work. We need to find a way not only to make good decisions, but to communicate them.

How does a regulatory agency like EPA carry out its mission to protect public health and the environment without over- or under-regulating? EPA could strangle biotechnology without even knowing it, or could easily cause undue public alarm about the technology. EPA could become inflexible and unable to regulate new risks; by structuring the agency to regulate yesterday's risks, EPA could be boxed in when faced with new ones. The agency could exempt a whole class of activity and then find itself helpless to control it aggressively.

So what do we use as a model as we forge a regulatory program to meet our goals? One of the characteristics of government is that it builds upon past experiences. We already have a group of laws in our federal structure governing products, which we are adapting to fit biotechnology's products. This is a good idea, but it has its strengths and weaknesses.

Under the TSCA³ and premanufacture notification (PMN) process,⁴ which has been in operation for ten years, EPA receives notification any time a company wants to market a new chemical. The agency then has 90 days to decide whether the new chemical is unreasonably risky.

Over the years, EPA has developed an efficient system to administer the PMN program. The agency now receives about 2,500 to 3,000 PMN notices a year. Most of the notices are "dropped" within a few weeks. We have gained a great deal of experience in distinguishing risk, given the nature of the chemical and the way in which it will be used. This allows us to make risk decisions and to give concentrated attention to a very few of those 3,000 chemicals a year. We end up with about 75 regulatory orders a year for new chemicals.

EPA has declared that microorganisms are chemicals as defined under TSCA,⁵ and no one really wants to debate this decision. Industry, the public, and government agree that we need an oversight function here, and it makes sense to use existing statutory authority.

We could evaluate the risks of microorganisms on an ad hoc basis. To some extent, that approach is necessary with the first few applications anyway. But industry and the public eventually need procedures, rules, and policy. They and we both need certainty in the regulatory structure, because nothing can destroy a new industry, or discourage meaningful participation by the public, more than uncertainty.

Certainty is difficult in this area because we do not know enough about biotechnology. The recent Congressional Office of Technology Assessment (OTA) report⁶ acknowledges that it is too soon to be able to determine that a whole class of microorganisms poses no risk. But the report does state, and EPA agrees, that a rough screening of microorganisms is possible to decide which cases deserve the most scrutiny. The challenge for regulators is to put this concept into regulatory language.

To design a screening program, in theory one would first hire experts and have them waiting in the wings as the PMN applications arrived. Keep in mind that industry can submit an application whenever it wishes. EPA must be ready to begin its review at any time, because the 90-day period starts the day the application is received.

The second step would be to develop a screening process based on the depth of governmental review necessary for each organism. Early on, EPA proposed a classification keyed to the method of recombinant DNA construction. Based on public comment, the agency concluded that this is not an adequately risk-based approach. Instead, we must look at a more difficult subject — the nature of the organism and the use to which it will be placed.

Since the marketplace regulates most things in this country, we also need a measured, flexible program that finds the optimum balance between government regulation and the marketplace. In the United States, regulation historically has intervened only in those relatively few instances where the marketplace is not working.

The program I have just described is not very different from the regular EPA program for new chemicals. Biotechnology is not really different in kind, but it is tougher to regulate because more scientific uncertainty is involved. Society, including EPA professionals, lacks expertise in biotechnology. We lack a pool of skilled personnel — not just biotechnology experts but also experts to evaluate the environmental impacts of releasing genetically engineered organisms.

Much more than the regular chemical program, biotechnology operates within constraints of public perception. This brings up the old nemesis, "NIMBY," or "not in my backyard." No matter how much they might believe that biotechnology is safe, people still do not want it nearby.

Where is EPA's place in this regulatory scheme? The United States Department of Agriculture (USDA) has jurisdiction over some agricultural uses. The Food and Drug Administration (FDA) has jurisdiction over foods, drugs, medical devices, and cosmetics, including the microorganisms used to produce them. EPA's Pesticides Program has jurisdiction over pesticides, no matter what type or origin.

Under TSCA, EPA also has jurisdiction over remaining applications of biotechnology. There are some cases, such as fish, where the agency has chosen not to regulate. But EPA has asserted its authority over the remaining applications of microorganisms, such as in metal mining, waste degradation, and biomass for energy. This makes it hard to find experts for our staff; they have to evaluate many diverse uses of biotechnology.

As a case study in risk assessment, I would like to consider EPA's decision on a Rhizobium designed to enhance the nitrogen-fixing capability of alfalfa roots. This is a good [19 ELR 10498] example of the risk question, because we know a lot about Rhizobium. In fact, Rhizobium have been used in agriculture for the last 100 years.

In our risk assessment of the Rhizobium, EPA looked at effects on humans and other animals. The agency also wanted to know what the effects would be on indigenous Rhizobium, because these organisms naturally occur in soil. Further, we considered effects on alfalfa, the target organism, and on nontarget plants. One concern was that the Rhizobium might cause weedy legumes to grow faster than one might want. In sum, we looked for any change in ecology, as opposed to micro-changes confined to the immediate test site. We concluded that the Rhizobium would not have an adverse effect on the nitrogen cycle.

Once a microorganism is field-tested, it is impossible to confine it entirely to the test site. No one can guarantee that there will be no exposure to the organism. So EPA looked at, first, potential releases from the production site, and second, potential releases from the field test site.

Among the questions that arose were: What is the possibility for survival of the organism? How does it compete with the indigenous organisms? What about dissemination? Here, our data indicated that there would be very low dissemination outside the test plot, so long as a good buffer was used. The data also indicated that human exposure and gene transfer were not expected.

Another question was: what is the potential for releases in situations other than this field test? We concluded that significant exposure could occur in uncontrolled tests if they were carried out on a much more expanded basis. Hence, a field test made sense if one hoped eventually to use this organism on a larger scale. A field test is the logical next step after the laboratory and the greenhouse.

As the OTA report indicates, in some cases field tests may be the only way to adequately evaluate an organism. This assumes, however, that one must be willing to live with any possible adverse effects.

Although the applicant needed information from this field test, EPA believed that the federal government and society could also benefit from the information. Therefore, the agency stipulated that the data from the field test should be made publicly available and not kept confidential by the company.

Among the statistics that EPA considered significant for this test were: the population trends of both the engineered organism and the controlled organism; the relative nodulation success of the controlled versus the new organism; the horizontal and vertical dissemination within the test plot for both; and dissemination beyond the test plot for both. We were also concerned about engineered plasma transfer from this organism to an indigenous organism; plasmic loss from the test organism; and the association between the organism and the alfalfa yield. This last correlation is an indication of the Rhizobium's success: based on early tests, the company was hoping for a 17 percent increase in yield. Finally, we wanted several additional greenhouse studies done before we would allow a larger scale application.

As always, since costs and benefits are traded off under TSCA, we had to determine costs. Costs were relatively minor in this case compared to some of theother actions we take.

I hope you find that this is a common-sense way to evaluate an application for an organism. We did not ask outlandish questions, nor did we give all factors equal weight. The process was a matter of logically evaluating risk. This enabled us to make a decision about whether this field test should take place without any controls or whether some controls were needed. In the end, we required some fairly modest controls and asked for the data to be returned to the government.

Finally, I would like to address EPA's direction in future biotechnology regulation. Recall that the problem with a new economic endeavor like biotechnology is that we lack sufficient experts to assist in environmental assessments of releases of these organisms. Many of the experts do not want to work for the government — they would rather work where they can do research, make more money, or both.

Nonetheless, the federal government cannot shirk its responsibility to undertake these reviews. One solution has been to expand the understanding of these releases on a national basis. Thus, everyone learns, and, as the regulatory decisionmaker, EPA has the benefit of more than just the expertise on the EPA payroll or on its immediate advisory committees.

As part of a rule to be proposed soon, Once established, an EBC will interact with researchers. This interaction usually will be on the research and development level of biotechnology, since that is currently the context for most environmental releases. Researchers will submit proposals either to the EBC or directly to EPA. The EBC will then decide whether the controls for a particular environmental release are appropriate.

If the EBC rejects the proposal, the researcher may submit a Toxic Environmental Release Application (TERA) to EPA. The research proposal itself is not foreclosed at this point, but must be reviewed by EPA.

If the EBC approves the proposal, the researcher must determine if the organism is intergeneric. In simple terms, an intergeneric organism combines elements of two genera. TSCA treats new chemicals, including new intergeneric microorganisms, differently from existing chemicals. New chemicals are subject to a great deal more review prior to their entrance into the marketplace.

If the organism is not intergeneric, the researcher may proceed to a field test. In these cases, the EBC has reviewed the risk and determined that it is acceptable. If the organism is intergeneric, the researcher must submit to EPA a TERA and the EBC opinion. Since an EBC already has reviewed the proposal, EPA does not conduct a de novo review at this point. Rather, EPA reviews the EBC review, which allows for quick decisionmaking.

Where does this fit into the overall scheme of EPA's future regulation? To begin with, EPA plans to retain intergeneric organisms within the TSCA definition of new chemicals, [19 ELR 10499] subjecting them to heightened review. The agency wants to carry through its 1986 policy commitment to regulate biotechnology research and development.⁷ However, this policy only applies to commercial biotechnology research and development; non-commercial biotechnology research and development is not regulated by anyone.

Under the rules we will be proposing, for commercial research and development involving intergeneric organisms, researchers must submit a TERA. A TERA requires less data than does a PMN application. EPA would encourage researchers to consult an EBC before coming to EPA. Commercial uses going beyond the research and development stage are reviewed directly by EPA.

Naturally occurring organisms are at the other end of the spectrum. EPA is proposing that commercial research and development for naturally occurring organisms be exempt from any review. However, commercial use of the organisms will be reviewed by EPA if the organism is not found on a special list of exempt organisms.

Besides regulating new chemicals, the other basic authority that EPA has under TSCA is to issue a Significant New Use Rule (SNUR).⁸ If an existing chemical is used in a new way, that use would be subject to review as if it were a new chemical. For biotechnology, we are proposing that new uses of naturally occurring organisms will also be subject to Significant New Use Review.

The other regulation that EPA is proposing will cover the remaining cases. This regulation will apply to organisms that are not new, but for which EPA will require review for research and development uses. Organisms will be exempt from this review if they appear on the exemption list or if the proposal has been reviewed by an EBC.

The exemption list will be an important element in the regulatory scheme. The list will be comprised of the types of organisms and the specific uses exempted. If a proposal covers the same organism and same use as on the exemption list, then it will not be subject to review. If the proposal is for a different organism, same use, or the same organism, different use, then it will be subject to review.

Our aim is to devise a flexible system that can grow over time as EPA, the EBCs, and society learn more about this technology. We may learn that organisms are safe for certain uses and that further review is not required. We may be able to exempt certain organisms from review for any end uses, or to subject them to a lower level of review.

To address that general point, the OTA report asserts that it is too early to define categories for low, medium, and high review. EPA believes that it is not too early to enunciate some general rules, however. Using pathogens as an example, we tried to make our rule very specific by stating that if the organism is a pathogen or has elements from a pathogen, then we wanted to review it. This definition turned out to be too narrow, and did not take in some of the other possible cases we wanted to review. So we broadened the circle of our definition, but developed the exemption list concept so as not to have to review everything in that circle.

To conclude, let me return to my basic point about the essential elements of the biotechnology regulatory system. First of all, we need experts. EPA cannot hire all the necessary experts, so we need a national network of experts to study environmental release and learn with us. EPA believes that the EBCs provide such a mechanism.

Second, we need a screening process. We need to categorize organisms by risk and not by method of construction. We need a flexible system allowing us to discuss the rules and the risks with the public and with the regulated entities. We need to allow the regulatory system to evolve over time. We need to gain some confidence that the system works.

Finally, we need wisdom. Here I have slightly modified a saying which I think is appropriate for this topic: "God give us the will to leave those things the same that should not be changed, the courage to change the things that should be changed, and the wisdom to know the difference." This applies to microorganisms, to government policies, and to good common sense.

DISCUSSION

MICHAEL BAGDASARIAN: When you use the term "intergeneric transfer," do you use it differently from the NIH Guidelines,¹ which exempt organisms that naturally exchange genetic information between themselves?

ELKINS: I am not certain — that is an issue we debated while writing this rule, and I am not confident that I could tell you the exact scope of our final resolution.

RICHARD B. STEWART: When you look at the systemic effects on the nitrogen cycle, how far do you go? Are you equipped to look at whether a new agricultural product will change farming practices in a major way, with secondary effects on the environment?

ELKINS: Your first question is, do we look at it? Although we have not been confronted with this issue yet, you are asking whether we would look at it if it were within the jurisdiction of TSCA. I think we would, because we see ourselves as the agency responsible for making this decision. Someone has to make the decision of whether the experiment should proceed, if it is to be made by someone other than the researcher.

On the other hand, do we have the competence to do so? We are limited to 90 days to make our decision, which in the government is more like one week once you get the paper moved around. What we have found we can do in the chemical regulation business is, first, ask the company for a voluntary suspension of the clock. Second, we can arbitrarily extend the period for another 90 days if we wish. Finally, if we reach the 90th day and all else fails, we can declare that we do not know enough to approve the chemical. We do not usually do that, but if we do, it is essentially a denial of the application.

So, in cases where a risk assessment keeps turning over rocks and we find more things that we do not like under the rocks, EPA will keep turning.

It is important to have these rules on the books — right now we have a voluntary system for R&D which is not very enforceable. We need regulations on the books to enable us to get to the courthouse door; that is what we are racing toward at this point.

REBECCA J. GOLDBURG: Will your new regulations define what it takes to make two organisms different, or are you planning to let that concept evolve over time?

[19 ELR 10500]

ELKINS: It is one thing to write a policy in fairly general terms, but regulatory language has to be more specific. We have been debating whether to base our regulations on the distinctions between same organism/different organism or same use/different use, because then we would need to be able to define both concepts fairly exactly. I think that if we get through the review process, we will draw a line and then decide how the line applies to specific cases.

While we need to learn over time, we need to have a system in place that guarantees certainty for the public and the researchers. We cannot allow great risks to go unchecked, but at the same time we need room to learn — we do not know enough about this subject. What is intimidating about biotechnology is not that it is so new or different, but that there are so many uncertainties connected with environmental release. Ideally, we should have all the data on the table when we make these decisions.

TSCA generally does not require a company to send any more data to EPA than the company already possesses. In this regard, TSCA is different from FIFRA.² But for biotechnology, our policy is that if companies do not provide certain data, we will deny applications on the basis that there are too many uncertainties.

GEOFFREY M. KARNY: Can you give us an idea of the time frame for the rules you have discussed?

ELKINS: These rules have taken longer than we anticipated, because fairly late in the development process we radically changed the concepts that we started with in 1986. The EPA Administrator has indicated that he wants to sign a final biotechnology rule while he is still at the agency. That gives us a fairly definite point in time — my target is December 31, 1988 for a final rule.

Normally, it takes one to two years to go from a proposed rule to a final rule. Since we have yet to propose, it will be a challenge to meet this target date. We hope to issue the proposed rule within the next 30 to 45 days. We have already begun to brief our fellow agencies and the Office of Management and Budget on the proposal. I think it is in everyone's interest to have these rules issued, particularly if they are flexible and can be changed as we go.

1. Pub. L. No. 99-499, 100 Stat. 1613 (1986).

2. SARA §§ 300-330 [Emergency Planning and Community Right-to-Know].

3. Toxic Substances Control Act, 15 U.S.C. §§ 2601-2654, ELR STAT. TSCA 001.

4. 15 U.S.C. § 2604, ELR STAT. TSCA 008.

5. 15 U.S.C. § 2602(2)(A), ELR STAT. TSCA 002.

6. U.S. CONGRESS, OFFICE OF TECHNOLOGY ASSESSMENT, NEW DEVELOPMENTS IN BIOTECHNOLOGY — FIELD-TESTING ENGINEERED ORGANISMS: GENETIC AND ECOLOGICAL ISSUES, OTA-BA-350 (1988) [hereinafter OTA REPORT].

7. See 51 Fed. Reg. 23,313-23,336 [EPA Policy Statement, Coordinated Framework].

8. 40 C.F.R. Part 721.

1. Department of Health and Human Services, National Institutes of Health, Guidelines for Research Involving Recombinant DNA Molecules, 51 Fed. Reg. 16,958 (May 7, 1986).

2. Federal Insecticide, Fungicide, and Rodenticide Act, 7 U.S.C. §§ 136-136y, ELR STAT. FIFRA 001.