By Prof. Jack Heinemann, School of Biological Sciences, University of Canterbury.
My report on assessing the risks of a form of GM wheat has sparked heated comment here and on other blog sites. The Sciblog-associated Science Media Centre (NZ) included comments from Peter Dearden and a link to the “Genetics Otago” blog along with comments made by Australian scientists.
For me, these events have raised some fundamental issues – not new ones but recurring ones – that have been confronting the scientific and regulatory communities at the forefront of developing, and critically evaluating, new technologies. I don’t pretend to have all the answers in this difficult area, and my views do and will continue to evolve. In the meantime, let’s pause to reflect on some issues.
The report being discussed was prepared for the Safe Food Institute and communicated by the associated Foundation in August 2012. Well over a decade ago, however, I took an interest in dsRNA-mediated gene silencing. In 2004 when I started actively engaging with regulators in Australia and New Zealand, I began advocating for testing for unintended dsRNA effects because of the possibility that intended and unintended dsRNAs might be transmissible through food.
Using assumption-based reasoning (that is, not using experiments to come to a conclusion), food safety regulators, including published scientists, dismissed the predictions, asserting that “the scientific evidence does not support the theory that RNA molecules in food can be transmitted” because they would be unstable during cooking, storage and digestion, unable to access mammalian cells or exert an effect on gene expression in mammalian cells (1). Our repeated calls to test these assumptions, and those made by others to the Australian Office of the Gene Technology Regulator, were met with, effectively, silence.
In 2011, a landmark paper published in Cell Research (4) proved these assumptions wrong (assumptions made, it should be noted, by the scientists serving both as regulators and as peer-reviewers for the regulator). The Cell Research paper concluded that “these results indicate that exogenous plant [dsRNAs] are able to enter the serum and organs of mammals via food intake, and that plant [dsRNA] can bind to the nucleotide sequence located in exon 4 of mammalian LDLRAP1, leading to the inhibition of LDLRAP1 expression in vivo.”
That dsRNA could cause silencing in humans was considered surprising to our community just 12 years ago, because of assumptions that it should not work in us, and because initial attempts to make it work in human tissue culture cells failed (2). It turns out that these assumptions were wrong, just like the assumptions that dsRNA would not transmit through food, and the failure in this case was due to using dsRNA molecules that were too long.
These anecdotes suggest that we shouldn’t lead with assumption-based reasoning when it comes to risk assessment.
Prof. Rick Roush said on the Australian ABC radio that he’d gladly volunteer to eat the GM wheat modified by dsRNA-mediated silencing techniques. He concluded that the potential for harm to humans was “very, very small.” Contrary to my report, which asks for safety testing to be done, entomologist Roush makes an a priori conclusion about the wheat’s safety.
This isn’t how risk assessment is done. The first step is what is sometimes called a “hazards” analysis. This analysis attempts to identify anything new, or anything produced at new concentrations, that might be the cause of an adverse effect. For the effect to happen, there must be exposure to the hazard. If exposure is a possibility, then the developer can be asked to determine whether the adverse effect is possible (or even probable), and to determine as much as possible what the severity of the consequences might be. Then the regulator makes a determination of whether the potential adverse effect can be mitigated or eliminated. If it can, or if the regulator then determines that harm to be negligible, the product may be recommended for testing or use.
That makes risk assessments at the stage of my report by definition speculative. And that is a good thing. It would be best to have identified what should be tested while the adverse effect is speculative rather than after a product causes harm to human health or the environment.
Prof. Dearden asks why, if people have been eating dsRNAs from plants (as shown by the Cell Research paper), should a test on those created in GM wheat be done. We have already answered that question in a peer-reviewed and published article in the risk assessment literature. “Likewise, as GMOs modified by regulatory RNAs (e.g., dsRNA of the siRNA, antisense, and shRNA kind) grow in number, it will be critical to determine the on and off target effects of the novel RNA” (3).
The sequence and the structure of the dsRNA intended to be produced in the GM wheat has never been eaten by humans before. We have no way of knowing if this molecule is benign, beneficial or harmful. We don’t have a history of safe use of this wheat. “Neither overall amounts of small RNA molecules, nor the presence of benign small RNAs in conventional plants are sufficient as evidence that all novel small RNAs will be safe in the food chain or environment” (3). The potential effect is caused by the sequence and the sequence is unique to the GM wheat.
This begs the question then, whether Roush and other commentators’ conclusions were risk assessments or risk conclusions. If the latter, then these should be based on open, reviewed and accessible methodology. Indeed, to say that harm is unlikely is a claim that many would feel should have gone through anonymous peer-review and have been published in a journal.
Interestingly, risk assessments are rarely if ever published in journals following an anonymous peer-review. For example, it is not FSANZ’s policy to do so. Regulators use almost exclusively unpublished material from the developer at the time they make their decisions. Why are there no objections to the lack of anonymous peer review and journal publication here, especially as these assessments, unlike my report, will actually determine what goes into the food supply?
Again, in contrast, my report on a risk assessment met the same or higher criteria for peer review that regulators and the industry use routinely. Unlike their work, my report had no potential to put an unsafe product in front of the public, and was open to complete public review. Yet it was criticised for not appearing in a journal.
Worse than this, many commentators who addressed my report in the media chose ad hominem argumentation rather than discuss the outlined exposure pathways and the nature of the experiments that are needed to test legitimate hypotheses about what and how adverse effects could arise. They made mistakes, such as failing to check facts on where the target gene had come from.
Importantly, not a single commentator to my knowledge has offered any evidence that a risk assessment on the dsRNA molecules, or secondaries that might arise, was done by anyone. Much less have they offered even a single shred of data from any testing specific to this risk of the GM wheat.
Science routinely shows prevailing assumptions, such as those made earlier about dsRNA, to have been wrong. The proper response to challenges to assumptions is further research. This, not denunciation of the challengers, is the way to maintain public trust in the regulatory system, and in science.
1. FSANZ. 2006. Draft Assessment Report Application 549 Food Derived from High Lysine Corn LY038. Food Standards Australia New Zealand.
2. Gura, T. 2000. A silence that speaks volumes. Nature 404:804-808.
3. Heinemann, J. A., B. Kurenbach, and D. Quist. 2011. Molecular profiling — a tool for addressing emerging gaps in the comparative risk assessment of GMOs. Env. Int. 37:1285-1293.
4. Zhang, L., D. Hou, X. Chen, D. Li, L. Zhu, Y. Zhang, J. Li, Z. Bian, X. Liang, X. Cai, Y. Yin, C. H. Wang, T. Zhang, D. Zhu, D. Zhang, J. Xu, Q. Chen, Y. Ba, J.-J. Liu, Q. Wang, J. Chen, J. Wang, M. Wang, Q. Zhang, J. Zhang, K. Zen, and C.-Y. Zhang. 2012. Exogenous plant MIR168a specifically targets mammalian LDLRAP1: evidence of cross-kingdom regulation by microRNA. Cell Res 22:107-126.
Correction: reference to Australian Science Media Centre Changed to NZ Science Media Centre