Risk assessment and risk management of transgenic microorganisms and viruses
by Kaare Magne Nielsen, Department of Pharmacy, University of Tromsø, Tromsø, Norway. E-mail: knielsen@farmasi.uit.no
The basis for risk assessments (RA) is the knowledge available and its quality denominators, the use of inferences where necessary, and a transparent treatment of the identified uncertainties and knowledge gaps. A key challenge is to communicate how these conditions define the strength of the risk conclusions drawn; as conclusions cannot be stronger than the evidence behind. Often, uncertainties are of a qualitative nature and it is scientifically flawed to present risk conclusions on a quantitative scale when the underlying biological processes and uncertainties are not understood numerically. The influence of assumptions and inferences in RA outcomes needs to be more transparently communicated.
Scientific knowledge production and, hence, RA may be value influenced (e.g. subjective bias in risk hypothesis formulation, question framing and data interpretation). Stringent measures need to be developed to effectively identify values inherent in RA. A recent editorial in Nature Biotechnology (2008, p. 1051) points to that researcher biases/motivations influence study outcomes and, hence, the unmet need for independent research and information sources. RA procedures lack efficient tools to adequately deal with study biases. The question is not how to discard values from the process, but how to appropriately recognize them in a transparent manner so that the globally diverse approaches to GMMV can be most effectively understood, harmonized and made operational.
The initial risk characterization process consists of the following steps:
a. hazard identification,
b. hazard characterization, and
c. exposure assessment.
In the on-line discussions, it is necessary to explicitly communicate the step and process that is discussed. It will also be useful to distinguish between the RA components of intended versus unintended effects, and considerations of sources of uncertainty, indeterminacy and ignorance.
Categorically, the GMMV products can be divided into:
i) purified products derived thereof,
ii) inactivated products that have no replication or DNA transfer potential, and
iii) products with replication potential.
These categories of products generate different biosafety questions that can initially be structured using the concepts of familiarity and substantially equivalence. The comparative approach requires that suitable comparators can be found. Identifying a relevant comparator may, however, be challenging or impossible. This on-line discussion offers an opportunity for the identification and development of alternatives to the concepts of familiarity and substantial equivalence.
Some key biosafety considerations of live GMMVs are the likelihood of:
i) undesired ecosystem interactions/impacts due to unintended survival, spread and persistence of GMMV,
ii) undesired heritable biological system impacts caused by horizontal gene transfer (HGT) of the novel genetic modification to new recipients, or caused by HGT/recombination of the GMMV with wildtype genomes leading to e.g. altered host range and dispersal dynamics of the GMMV itself.
A crucial point is potential irreversibility. The limited understanding of key environmental processes presents a challenge to RA of GMMV. For instance, to fully understand the genetic impact of novel GM traits in live MVs, questions such as those listed below must be answered: What are the drivers of natural MV diversity, dynamics, and evolution? Do we know how the various drivers interact to shape genetic diversity in a given environment? Which existing genes in the current gene pool will survive or are on the way to extinction? How will these processes affect future community structure and composition? These questions can perhaps be addressed for most GM-plants but not for GMMVs. A mechanistic knowledge of the cell cytoplasm can be constructed today by using “omics” technologies but the ecosystem roles and interactions of live GMMV remain fragmented, descriptive and with little functional contextualization.
The overall key determining potential unintended impacts of live GMMV is selection; the ultimate determinant of long-term survival of GMMV or disseminated GM DNA. Relative and absolute fitness considerations are therefore required in RAs along with relevant tempo-spatial population dynamic considerations.
The knowledge gaps introduced above call for the use of minimal sized GM inserts, broader knowledge of selection, and understanding of the fitness changes caused by the GM trait when present in the intended recipient or in new recipients after HGT. Depending on the category and products of the GMMV assessed, careful considerations of relevant biological, ecological and evolutionary scales are needed. By doing so, questions will arise as to if quantitative descriptors can be found and to what extent researcher motivation and biases influence data quality through study design, outcome interpretation and reporting.
Return to the
discussion page.
