Further drafting of the guidance on risk assessment and risk management of LM crops with resistance or tolerance to abiotic stress

Dear all:

Sorry for slow take-up on the initiation of the abiotic stress tolerance discussion.
As in the uploaded documents, there are draft guiding doc and guiding questions.
Also for starting discussion, please go over Annex III of the Protocol, whether the abiotic stress tolerance category needs further detailed guidance on specific parts?

Kind regards,

Kazuo Watanabe

Dear colleagues,

Thank you for the opportunity to comment on this document. I have two main comments:

1) I do not agree that different approaches to risk assessment are required for abiotic stress tolerant LMOs. At least three different types of traits have been assessed using current methodology (insect resistance, herbicide tolerance, and modified composition) and new approaches are not needed at this time.

2) I also don’t think that it’s necessary to draft separate guidance on risk assessment for  abiotic stress tolerant crops. Most of the considerations for the risk assessment are the same as for other types of traits and including them as separate guidance leads to a redundant document. All plant-trait combinations must be considered for increased weediness/invasiveness and persistence, not just those with abiotic stress tolerance. Similarly, any trait should be scientifically examined to determine if there is a reasonable likelihood that reproductive biology might be altered, or endogenous toxins/allergens. Each of these considerations must be addressed for all trait-plant combinations in a hypothesis-driven manner. For the considerations that are unique to abiotic stress tolerance (e.g. physiological responses to stress in the LMO and the comparator, unintended tolerance to other stresses), it should be enough to include some supplemental guidance within the Roadmap in the appropriate places.

Best regards,
Cheryl Corbett

Dear all,

I have some questions/comments on the draft of the guidance on risk assessment/management on resistance/tolerance to abiotic stress. I hope that they will help to improve the draft.

* As a whole, abiotic stress should be evaluate under receiving environment. In other word, it is unnecessary to conduct the risk assessment of LMOs with resistance or tolerance to abiotic stress under the condition which does not exist in the country. Therefore, this should be clearly mentioned in "General consideration 3" for the country where they are not familiar with this matter.

* How do you image or define the persistence in the agricultural land ? I think the examples, such as overwintering ability of matured plants, dormancy and germination rate of seed, can help assessors conduct the risk assessment of LMOs with resistance or tolerance to abiotic stress.

* Weeds and/or native plants are also included in the biotic stress. Do you include the biotic stress not only in agricultural land, but also natural environment?

Best regards
Yasuhiro YOGO

Dear all:

In the first round at the last week did not get many participation on the abiotic stress tolerance SWG.  Recalling in the first session early in this year after the first AHTEG meeting, we had relevant discussion and points raised on the abiotic stress tolerance on GM crops.
The discussion as indicated by the SCBD should be built on the previous sessions, and also consideration by looking at the overall roadmap of the RA.
My observation of the interventions on this subject overlaps quite a lot with the roadmap per se, and fundamental discussion on LMO RA is in the ANNEX and I hope the discussion become very specific which we shall raise points particularly on the abiotic stress tolerance issue. Otherwise, repetition should be avoided, unless specific revisit is essential. The guiding doc on abiotic stress keeps a lot of comments to reflect the previous discussion, however, these may or may not be kept all.  Please go to the draft doc for specific discussion.

Kind regards,
K. Watanabe, RA AHTEG SWG member on abiotic stress tolerance

Dear Colleagues,
I agree with the comments of Cheryl Corbett that a different approach is not required for the risk assessment of crops with abiotic stress tolerance traits, and that the points to consider raised in the draft document can be covered by the Roadmap.
More specifically:
i.   Annex III already recognises case-by-case LMO risk assessment (General Principle 6).
ii.  Annex III (Methodology and Points to Consider), together with the Roadmap (Step 1) provide for comprehensive characterisation of the novel genotype and phenotype, and receiving environment of the LMO.
iii. The environmental considerations (weediness, invasiveness, persistence) are not unique to abiotic stress tolerance traits. These risks are also assessed for other tolerance/resistance traits (i.e. herbicide/insect) that confer advantages under particular conditions. The available risk assessment information for these LMOs should be considered also.
iv.  There is an emphasis on phenotype in the draft document. To determine the effectiveness of an introduced abiotic stress trait, it would need to be tested under a range of conditions including absence of the abiotic stress. Adverse or unintended effects would be identified in the development and characterisation of the LMO. Where these occur, the LMO is unlikely to be developed further.
v.   The potential for crosstalk between abiotic and biotic stress mechanisms in the LMO should be identified in the characterisation of the novel gene products.
vi.  The draft document introduces compositional analysis, this does not apply to environmental risk assessments.
Kind regards,
Felicity Keiper

Dear Dr. Watanabe, dear colleagues,

In reviewing the comments of the earlier round of consultationas as well as those made on the new draft, I can only support that there are very few issues -if any- that seem to be specific for LM crops with a modified abiotic stress reaction. None seem to require an approach different from that which is used for LM crops in general.

In line with the previous comments:
- A case by case risk assessment (as indicated in General considerations 3.) is a general principle 6 of Annex III.
- Comparing performance under different relevant conditions tailored to the specifics of the trait is already a standard way of performing studies (e.g. herbicide tolerant crops are typically evaluated and compared with controls under different herbicide treatment regimes).
- Non-permissive environments: again this is not a new issue. Taking the same example, applying the selective herbicide on non-tolerant entries would not yield comparable material. Still developers have develope approaches to make meaningful comparisons.
- All of the question raised as "of particular relevance" as well as the potential adverse effects are already addressed in a standard risk assessment.

On this basis, I do not see the advantage of having a separate document.

I would also like to point out that although searching the references may eventually lead to interesting information, the relevance for the paper is not always clear and certainly not direct.

Kind regards, Patrick

I agree with Patrick here that there is really no need for a separate document on the risk assessment of LM crops tolerant to abiotic stress. The approach will not differ from what is used in assessing risks of LM crops in general.The only difference is that such crops maybe targeted for hostile environments, i.e. for areas not previously planted with the crop. It will be a challenge for regulators to ask the right questions but a different guidance documents is not going to help. What would help are studies on similar situations such as new crop introductions and what these have doen to the environment.

If there is to be a sepatate document for abiotic stress 9and i understand the reasons that are offerred to support such an approach0, I would make the following comments to the current draft:
* the "comparitor" should be specuified as the closely related conventional parent; the earlier formulation which required at least 50% genetic identity should be re-instated.
*the second step, focusing on the consequences of the use of the LMO in  environmental conditions that are not suitable for the conventional organism being compared, is the crux of the matter. This is where most of the assessment energies must be focussed and where creativity on the part of the assessors will be needed. By definition, the comparitor does not grow under the stress condition central to the development of the LMO.  This fact exposes the limitations of the fairly mechanistic approaches to risk assessment  contained in many documents (which often just state that a comparison should be made to another organism, etc).  I have commented on this as well in the thread on the basic Roadmap.
*Research and experimental data may be necessary, not just a literature search,etc. in particular, animal feeding studies are likely to be necessary, otherwise the comparison is just superficial and does not serve the goal--which is to assess the risks in the real world.

Philip L. Bereano
Professor Emeritus, University of Washington
for the Washington Biotechnology Action Council

While I understand the merits of a comparison of closely related genotypes, I want to point out that most risk assessors recommend an approach of comparing new traits with different controls, typically including a range of genotypes that represent the natural variation in the species and/or commercial varieties. Several guidance documents are available to support this (e.g. from the EFSA GMO Panel).

Also the previous comment seems to suggest that the comparator may not grow at all under the specific stress condition. This assumes a very strong trait that overcomes a limiting stress factor. I would rather expect that most stress "tolerances" will yield a relative improvement rather than a black or white change in behaviour. Nevertheless, this only underlines the importance of not limiting the comparison to a single genotype.

I agree however that simple comparisons, even with a range of genotypes, has its limitations. How will we react when a line is developed that is clearly performing outside of the normal range? When developed via traditioanl breeding, some might say that this is just exploiting the limits of the natural range. Will these be acceptable without concern?

Again, I have not seen any convincing argument to claim that abiotic stress tolerance poses challenges that are not addressed by the general Roadmap.

Patrick Rüdelsheim

Dear All

Abiotic stress is unlike herbicide and pest tolerance, and modified composition, in important ways. Addressing these, as Kazuo Watanabe (#1523) reminds us, is what the document should do without being repetitive. The goal as I see it is to outline what would constitute adequate and appropriate science for the task (Roadmap 2(b)). In setting these guidelines, we would not want to be too prescriptive and thus prevent the innovative regulator and developer from using improved tools and methods that may come in the future, but we also want to establish some uniformity to the approach for harmonisation.

The difference between abiotic stress and other modifications to date is that unintended or unanticipated adverse effects may occur only under environmental conditions for which the scientifically strong control (including at least the comparator as defined by Philip Beraeano (#1555), but also potentially other genotypes as noted by Patrick Rudelsheim (#1556)) may not be available or, if it is, it may not be physiologically equivalent. This is generally not the problem for pest and herbicide tolerance since the plants are still growing within their normal physiological parameters or other techniques may be used to mimic tolerance.

The scientific standard for comparison requires not just controls, but the use of the comparator. The comparator provides baseline information for comparison between all trials, replications and environments because it is grown at the same time and location as the LMO in every experiment. If this were specified in the Roadmap, then it would not be required in the abiotic stress document. But I cannot see this in the Roadmap. Having said this, there is nothing that precludes using additional genotypes provided that the data from these more distantly related genotypes is not used to dilute the statistics with noise, especially if the additional genotypes are not used as consistently as the comparator.

The proper comparator will be necessary to measure any differences (Roadmap Step 1) under conditions where the abiotic stress is absent. As Patrick Rudelsheim reminds us, abiotic stress may not kill conventional counterparts, it may only reduce performance. In addition, we can expect that stress tolerant LMOs will not always be grown under conditions that the conventional counterpart would find stressful (e.g. a high water year for drought tolerant plants) and thus the use of the proper comparator for these assessments will be critical.

The added complexity of abiotic stress is that a difference in composition (that might be relevant to some important environmental interactions between the LMO and microbes, plants or animals in its intended environment) or behaviour may only manifest in an environment where the proper comparator will not serve us well. Imagine the case where a conventional plant accumulates an anti-nutrient under water stress, but the drought tolerant version accumulates a potential toxin in the intended environment. If compared through studies on insects provided these two plants as food, the anti-nutrient effect of the conventional may mask the potential toxin in the LMO, giving the impression that there are no unique hazards in the latter, and possibly even benefits that would be overestimated.

These kinds of effects may not be seen by comparison alone and this fact, I suggest, makes it worth emphasising the need for the hazard identification methods of Roadmap  Step 1. Different regulators may be satisfied with a different combination of data of the types suggested in Step 1. They may do so because they have what they feel is comparable or redundant sources of information that can substitute for some of that information. [For example, Felicity Keiper (#1533) and I disagree on the need for compositional studies. Thus, the Roadmap has not resolved a particular ambiguity arising in the consideration of abiotic stress tolerance. This example illustrates why there may be value in retaining this document in whatever final form it will take.] In the case of abiotic stress, the LMO may be taken outside of any past experience and thus the regulator could be reminded of the higher importance of collecting comprehensive hazard ID data for these LMOs.

Potential “blind spots” in abiotic stress tolerant LMO hazard ID won’t be completely compensated by using additional genotypes which may only complicate the statistical analysis without necessarily increasing confidence in the hazard characterisation. Addressing these concerns will require methods that some would not normally invoke, such as compositional studies. Compositional studies are important since they may provide some indication of compounds that possibly accumulate only in the LMOs and that might be anti-nutrients or toxins of relevance for interactions with non-humans. Feeding studies (#1555) would need to mean studies on the toxic or anti-nutritive content with regard to insects, mammals and other relevant members of the food web.

If the Roadmap were amended to include the important specifics that I believe are necessary for abiotic stress assessment, as Cheryl Corbett (#1512) suggests, then I could agree with others that see no need for a separate document. I can also see disadvantages to making the Roadmap bigger and more detail laden, especially when some of the most important distinctions are not in general advice about looking for adverse effects, but in how to look for them in a scientifically meaningful way.

All the best
Jack

POSTED ON BEHALF OF HECTOR QUEMADA
(this message was received by email shortly after closing of this discussion group)

-----

The point of a roadmap, it seems to me, is to to provide clarity on where to go.  I think the comments of the experienced risk assessors during this discussion reflect the concern that the roadmap has not accomplished this goal.  I believe that the discussion on abiotic stress exemplifies this concern best.

I think the suggestions made by Andrew Roberts about the roadmap, especially the one calling for links to examples, would be the best approach in the case of abiotic stress, rather than an extensive list of additional points to consider and a lengthy discussion of proper comparators and conditions of testing.  Examples of risk assessments of abiotic stress tolerant LMO plants ought to be available in the near future as those plants are submitted to regulatory agencies.  The agencies that will see these submissions first will be those that have already conducted numerous risk assessments, and therefore will know how to apply the principles and points to consider in Annex III to these newer types of LMOs.  These examples will then serve as useful signposts for the roadmap and would be helpful guidance to those regulatory agencies who may be considering these LMOs at a later time.
When asking someone for the way to go from point A to point B, I would rather ask someone who has traveled the road before.

Thanks to all for a very stimulating discussion.

Hector

POSTED ON BEHALF OF WAYNE PARROTT
(this message was received by email shortly after closing of this discussion group)

-----

Dear all,

I also do not see a need for a different approach for stress-tolerant LMOs.

Thus far, I have not seen discussion on the type of traits that would be engineered to obtain stress tolerance, and I think that could be useful to provide context, and help guide revisions of the road map so as to provide clarity on where to go.

First, all crops show a stress response unless they were growing under most idyllic of situations, when it is never too hot, cold, wet or dry.
However, agronomically, this stress response is often too little or too late. Thus, one way to engineer for stress tolerance is to use transcription factors that start the gene expression cascade necessary for stress resistance, such that it is in place before the start of the stress itself.

Barring temporal differences, the response to stress of the LMO should be the same as that of the non transgenic version. Hence the appropriate comparator is the nontransgenic version grown under whatever level of stress it will tolerate.

Another way to get stress tolerance is to over-express *one* component of the general stress response; e.g., the production of fructans or other compounds involved in membrane protection/stabilization.

What happens next from a risk evaluation stand point depends on whether the one component is endogenous to that particular crop or not. If it is, then that one compound should be found in the non-transgenic comparator grown under the stress conditions it will tolerate. If the one compound has never existed in that crop before, perhaps a more extensive evaluation might be warranted, for example, to ensure that it does not serve as a substrate for the production of some undesirable compound. Having said that, it would be useful to ask the question, to what extent can a compound like fructan be used to derive a toxic compound? If it cannot be used, then there is little sense in carrying out additional evaluations.

In general, it is important to maintain perspective– I am not aware of any case where a crop produced a previously unknown toxin when moved from one environment to another. If a crop is able to produce a toxin or other negative compound, it will already have done so under some cultivation conditions.

Likewise, stress-tolerant crops are primarily designed to help the crop overcome a period of stress where it is currently grown, and thus help stabilize yields. Such an intended use would be different from trying to engineer a crop so that it grows where it has never been able to grow before.

So, all this fits in to the existing procedures. First, the differences are assessed. When a difference is detected, the risk associated with the difference is assessed in more detail in a case-specific manner. In the end, it is important to base decisions on what is known, using the best available information for the regulatory decisions, and not set up a research program for every risk assessment.

Even when the best available information is limited (for example, some new energy crops), there are still other sources that can be drawn upon.
For example, experience for assessing the potential for a species to be invasive might use the experience from the International Plant Protection Convention. Again, the guidance should be simple.

Thanks to all for permitting me to share these thoughts with you.

Wayne Parrott

POSTED ON BEHALF OF HIROSHI YOSHIKURA
(this message was received by email shortly after closing of this discussion group)

-----

Dear all
My comment is on ‘comparator’ for abiotic stress-resistant LMOs. As Prof. Bereano wrote in this forum, it would be “the closely related parent” or “a related organism/variety(, its environmental behavior for which there is experience of establishing safe use) (phrase in the parenthesis is my own, modified from the codex definition)”. However, as pointed out by Jack Heinemann, the LMO developed for abiotic stress may not have an appropriate comparator as its parental organism will not survive well under the stress condition. One approach could be to use as the second comparator the conventional living organism(s) that are indigenous to the abiotic stress condition and are quite “like” the parent organisms. The word “like” used here can be interpreted broadly as “relatedness in phylogeny, sharing of the key gene(s), similarity in anatomy or in physiology, or any other traits suitable as comparator. The choice will depend upon the LMO, the receiving environment, or the combination of them. The similar approach was taken for plant oil with a composition that is significantly different from its conventional counterpart (paragraph 51, Guideline for the conduct of food safety assessment of foods derived from recombinant-DNA plants). With the two comparators, Bereano-san’s and the one proposed here, the comparative safety assessment could be conducted fairly reasonably in the normal and stress conditions.
Hiroshi

Hiroshi Yoshikura
Adviser, Food Safety Division,
Ministry of Health Labour and Welfare
1-2-2　Kasumigaseki Chiyoda-ku, Tokyo　１００－８９１６
FAX:+81-3-3503-7965
Tel: +81-3-3595-2142/+-81-3-5253-1111 (2408)
E-mail:yoshikura-hiroshi@mhlw.go.jp