Comments to ANZFA about

Applications A372, A375, A378 and A379

from the Food Legislation and Regulation Advisory Group (FLRAG) of the Public Health Association of Australia (PHAA) on behalf of the PHAA.

General Comments

According to ANZFA’s Act, the primary objective of ANZFA is the protection of public health and safety.

The Public Health Association of Australia (PHAA) is the peak public health body in Australia with almost 2000 members and more than 40 public health–related occupations represented. It has a national and multi-disciplinary perspective on public health issues and makes major contributions to the public health debate in Australia through representation on government boards, committees and other decision-making bodies.

The PHAA therefore views with considerable concern that ANZFA had ignored PHAA’s previously written concerns about flaws in the process and science of genetically engineered (GE) food safety assessments in Australia. In particular, ANZFA has ignored the nature of our concerns as expressed in a recent letter and in a detailed critique of three GE foods. These concerns included a lack of understanding of public health principles, a lack of published evidence upon which ANZFA bases its conclusions of safety of these foods and a lack of safety assessments by researchers who are independent of the applicant company.

The concerns also included statistical and methodological problems and irregularities such as inadequate sample sizes, inadequate statistical reporting (most of the mean, standard deviation, 95% confidence interval of the mean (95% CI), the nature of the statistical test (eg t-test) and the probability value (p value) are required but are routinely not given) and a lack of understanding of the role of controls.

In contrast, ANZFA has persisted with their previous approach including their previous methodology and statistical treatment. Our concerns do not even get a mention in Attachment 5: Summary of Public Submissions and Attachment 6: General Issues Raised in Public Comments, attached to the last GE food draft risk analysis reports, even though our comments were much more detailed and specific than the many other comments that are reproduced in those attachments, and they were provided to ANZFA four months prior to the release of these latest GE reports.

Sample sizes are still inadequate and the 95% CI, nature of the statistical test and p-value are still routinely not given. ANZFA is now occasionally putting standard deviations into its compositional analyses, a welcome development, but it needs to the done constantly. In addition, its methodological and statistical reporting of animal experimentation is now much worse than it even was before. Now, it is routine for ANZFA to give no actual data from these experiments, but to just assert that no differences were found. This is totally unacceptable and a most retrograde step, particularly as the data of animal experiments in previous draft risk analysis reports showed adverse effects in spite of the assertion that no differences were found. It would seem that ANZFA now does not want outside scrutiny of these experiments.

We can only conclude that the ANZFA wishes to ignore public health input from Australia’s peak public health body, when it sorely needs public health input, that it would rather have a pro-industry and a pro-GE food stance and a reductionist approach to GE food. Moreover, given the prominence in these ANZFA documents of matters relating to the World Trade Organisation compared to matters relating to public health, we can only conclude that ANZFA is more concerned about the World Trade Organisation than in ensuring public health and safety in Australia.

In this Attachment 6, ANZFA claims that it takes a cautious approach to the introduction of GE foods. However, ANZFA’s own document “GM Foods and the consumer, ANZFA’s safety assessment process for genetically modified foods” clearly states that ANZFA considers that GE food is safe until proven unsafe, the opposite of the precautionary principle. It is clear that ANZFA still relies almost entirely in substantial equivalence in its safety assessments. That is, draft risk analysis reports conclude with: “Under the current standard A18, which remains in effect until December 2001, food derived from [insert name of GE food] does not require labelling as it is regarded as substantially equivalent to food derived from non-genetically modified [insert name of food] varieties.” This is in spite of ANZFA’s repeated assurance to the NZ Royal Commission of Inquiry into Genetic Modification that substantial equivalence was only “a starting point” to assess the safety of these foods. ANZFA quotes from a 10 year-old document to support their use of substantial equivalence. The field has move on considerably in the last 10 years. For example, a report released this year by the Royal Society of Canada condemned substantial equivalence to be “scientifically unjustifiable and inconsistent with precautionary regulation of the technology”.

One problem is the definition of substantial equivalence. Is it 80% the same as the non GE plant? 90%? 95%? ANZFA need to quantify what it means by substantial equivalence, because there are often statistically significant differences in the amino acid and fatty acid compositions of the GE plant compared to the non GE plant. In a previous draft risk analysis report, eight out of eighteen amino acids were statistically significantly different for one GE corn compared to its control. It would therefore appear that ANZFA may quantify substantial equivalence to be an amino acid composition that is only 55% of its control. There are similar examples in these four GE draft risk analysis reports.

Furthermore, the ANZFA continues to repeat their view that the only feeding studies that need to be done for the GE foods are short-term toxicological testing of the substance that is genetically engineered to appear. The hypothesis that only this substance will appear in the GE food is an untested hypothesis. In addition, ANZFA argues that feeding the whole food to experimental animals is not appropriate as it is “not possible to conduct dose-response experiments for foods in the same way that these experiments are conducted for chemicals. In addition, A key factor ... is the need to maintain the nutritional value and balance of the diet.” Such a toxicological view of food ignores the whole body of literature associated with animal feeding studies for nutritional research. Even a brief view of this literature would make it clear to ANZFA that the type of studies that they view as inappropriate are in fact routinely done. Moreover, previous risk analysis reports of GE foods published by ANZFA have contained such studies. In fact, in the nutritional literature, the usual way of testing the health effects of a food are to feed the whole food to experimental animals first, and then follow-up any adverse effects with subsequent feeding and other studies of the components, a process that is the opposite of the accepted by ANZFA for these foods. ANZFA should also note that the acute toxicity testing proposed as adequate would simply not pick-up cancer, teratology or the long-tem effects of nutrient deficiencies or increases in anti-nutrients.

Moreover, ANZFA seems to be more concerned about the ethics of subjecting “experimental animals to such a study if it is unlikely to provide meaningful information”, instead of the more pertinent ethics of feeding people these foods first without first feeding them to animals. This, in effect, makes humans the experimental animal, when informed consent, a necessary prerequisite of human experimentation, has not been obtained. Indeed, in survey after survey, the public has overwhelmingly made it clear that they do not want to eat GE food. Yet the experiment has continued on people who have withdrawn their consent. ANZFA’s assertion that animal experiments are unnecessary also ignores the evidence from their own draft risk analysis reports, where adverse effects were found in experimental animals fed the whole GE food. This result was not predicted by any of the other safety assessments that were done, and which were the type of safety assessments that ANZFA considered to be sufficient.

Furthermore, ANZFA’s views on antibiotic resistance are also remarkably out-of-step with scientific opinion and the response of their own Department of Health and Aged Care (AFFA and DHAC, 2000), which has recently chosen to implement most of the recommendations of the Joint Expert Technical Advisory Committee on Antibiotic Resistance (JETACAR) report into reducing antibiotic resistance in the human and animal populations.

In addition, ANZFA’s views on the transfer of antibiotic resistance, transfer of novel genes and viral recombination run counter to a considerable number of international experts, who’s views and evidence can be seen as part of the evidence given to the NZ Royal Commission of Inquiry into Genetic Modification (www.gmcommission.govt.nz).

Finally, in paragraph 10 of Attachment 6, ANZFA may be confusing surveillance with experimental studies (eg cohort studies). Surveillance simply counts the number of cases of disease or injury. Except for cases of injury (eg falls in the elderly), it does not attempt to ascertain causative exposure. The linking of disease with its causative exposure with anything other than injury would really require an experimental approach such as a cohort or case-control study. Such studies do not need to rely on data linkage. Variations in exposure would occur naturally in the population without resorting to temporal or geographical trends, simply because different people eat different quantities of different types of food. There are a huge number of such studies reported in the scientific and medical literature. Some examples include the Port Pirie study in Australia into lead exposure in children and their subsequent IQ, and the “nurses study” in the USA into fat intake and cancer.

Specific comments on Application A372. Oil derived from glufosinate-ammonium tolerant canola Topas 19/2 and T45 and oil derived from glufosinate-ammonium tolerant and pollination controlled lines Ms1, Ms8, Rf1, Rf2 and Rf3.

The application covers seven lines of canola. There are no published articles reported in the ANZFA document to back-up claims of safety. Moreover, ANZFA has not released information relating to the exact combination of elements present in each of the plasmids involved in the genetic engineering as ANZFA regards them as “commercial in confidence”. This has prevented health experts from fully assessing the potential health effects of these foods and therefore ANZFA has placed commercial considerations ahead of health.

Glufosinate ammonium is a structural analogue of glutamate, an amino acid. Thus, glutimate is a building block of proteins in animals and plants. Glufosinate ammonium inhibits the plant enzyme glutamine synthetase, an essential enzyme in nitrogen metabolism and amino acid biosynthesis in plants, leading to the accumulation of ammonia in the plant and the death of the plant. It is also toxic to mammals, resulting in central nervous system problems (it can cross the blood-brain barrier to cause unconsciousness, convulsions, apnoea and amnesia), haematological and haemodynamic changes, increase levels of creatine kinase, gastro-intestinal symptoms and a syndrome similar to diabetes insipidus (Tanaka et al, 1998; Watanabbe and Sano, 1998; Takahashi et al, 2000). It also causes considerable problems in embryos of mice and offspring of rats, including functional brain abnormalities, growth retardation, increased embryo lethality and very high rates of morphological defects (Watanabe and Iwase, 1996; Fujii, 1997).

Comments about the pat gene and its associated PAT enzyme are given in comments about Application A375.

It is reported in ANZFA’s own draft risk analysis report for this food that the barnase gene produces a non-specific ribonuclease that destroys the cells in which it is expressed. This ribonuclease has been found to cause nephrotoxicity to rat kidney in vitro and functional impairment of isolated perfused rat kidney, probably due to RNA degradation (Ilinskaya and Vamvakas, 1997), while Prior et al (1996) established its cytotoxicity to a variety of human cell lines. Even when this ribonuclease was decomposed into six modules, it retained RNase activity. That is, three of the six modules still had this ability.

This GE plant has this gene in every cell of the plant and in the seeds produced by the plant. As it is such a comprehensive cell poison, it is of concern that it is being placed in plants that enter the human and animal food supply. (Canola oil is consumed directly by people and canola meal is fed to animals that people eat.) There is no guarantee that it will remain unexpressed in these plants or during digestion by animals or humans.

Moreover, ANZFA’s assertion that all DNA and protein will be processed-out of canola oil, ignores the higher protein and DNA concentrations in cold-pressed canola oil, a substance which is sold in Australia.

Only the proteins expected to be produced by the plant were tested-for. Moreover, it is assumed that any health problems would come from the production of proteins. The production of fat-soluble substances that may be present in the oil fraction is not considered. It is of concern that compositional data for phosphorus, sterol, chlorophyll, tocopherol, specific gravity and the smoke point of the oil were not given. The fatty acid composition of the canola oil of some of these GE canola lines was given, but no mean, standard deviation, 95% CI, statistical test or p-value of these results, making it extremely difficult to assess these results. Moreover, the data about the composition of canola meal including protein and individual glucosinolates are not given at all but are just stated as being “within the ranges observed for non-transformed canola lines”, when data from the GE plant should have been compared to its control. Animal feeding studies using oil from these GE canola lines were not done, even though this fraction is consumed by people, because it may “cause nutritional and biochemical imbalances”. Yet the scientific literature in nutrition contains thousands of these types of studies. So why does ANZFA insist that they are not viable?

Animal feeding studies using canola meal were undertaken but the results of these are not provided by ANZFA on the basis that this is not consumed by people, even though meal from a previous canola (GT73), when fed to animals showed unexpected adverse results that were not predicted by the other safety testing. Instead, two feeding studies using the whole canola seed are given, when humans do not eat these either. One of these studies was on chickens and the other was on rabbits. The chicken study had good sample sizes (280 chickens in total), but only fed the birds one of the seven canola lines (glufosinate-ammonia tolerant Topas 19/2) and only measured body weight, feed intake and mortality during the study. At the end of the study, only chilled carcass weight and yield of deboned breast meat as a percent of carcass weight was measured. No actual data were given, only a declaration that no significant difference was found for body weight, feed intake or mortality. The statistical difference or otherwise of the other variables was not given. Results of post-mortem examinations were not given.

In the rabbit study, ten animals per group were fed only one of the GE canola lines (a cross between Ms1 and Rf1), were only fed for four days and only faecal samples were measured for dry matter, ash, nitrogen, fat, crude fibre and gross energy (by an adiabatic bomb calorimeter). No actual data were given, only a declaration that seed from this canola exhibited “at least similar zootechnical performance as seed from the original Drakkar variety”. When this statement is combined with the fact that chickens and rabbits are rather unusual models for human health studies and that the measurements taken (eg abdominal fat pad weight and total deboned breast meat yield) are rather unusual measures of human health, it can only be assumed that these studies were done more for reassuring primary producers of the ability of animal feed containing this GE food to permit farmed animals to grow, than for assessing the risk to human health.

On the basis of these very limited animal tests on two GE canola lines, all seven canola lines were considered to be safe for human consumption.

Specific comments on Application A375. Food derived from glufosinate ammonium tolerant corn line T25 (“Liberty Link corn”)

This application covers one line of GE corn. There are no published papers reported in the ANZFA document to back-up claims of safety.

The pat gene is transferred by genetic engineering into the plant. It produces the enzyme phosphinothricin acetyltransferase, which acetylates phosphinothricin, the active moity of the herbicide, thereby inactivating it. No information is given as to the enzyme’s specificity to just this substance. As active enzyme is found throughout the plant, including the kernels (Table 1 of ANZFA’s document), human and farm animals will be exposed to it. If this enzyme is not specific to this substrate, it may acetylate or deacetylate other proteins in humans and farm animals. The ANZFA document states that processing to produce processed corn products would inactivate the enzyme. However, this ignores the minimal cooking of corn kernels or whole corn cobs as a vegetable, and the use of raw corn kernels in salads.

This corn also contains a gene for resistance to a number of b-lactam antibiotics including the moderate spectrum penicillin, ampicillin. In one part of ANZFA’s document, it states that this gene was deactivated by removing the 3’ end of the gene sequence to make it non-functional to the corn. However, in another section of the document, this was contradicted by stating that 25% of the 5’ end had been removed.

These were transferred into a proprietary inbred corn line, B73 to produce the GE corn line T25. The only novel protein that the applicant company therefore expects to be produced is therefore the PAT enzyme.

The details of the method of transformation and the proprietary plasmid used in the transformation have not been released by ANZFA for scrutiny by health experts as they are regarded as “confidential commercial information” by ANZFA. This is unacceptable as commercial considerations have been placed before public health considerations.

The DNA digestibility study presented was just a repeat of the results form the earlier canola application (A372). This is not acceptable. Plant leaf material from canola is different to material from corn.

The toxicity of the PAT protein was assessed using a single oral gavage to mice (5/sex/group) of 51% PAT protein in carboxymethylcellulose and various control groups. The PAT protein used did not come from the corn, but from bacteria and was purified before use. Mice were observed for 14 days before being sacrificed and body weight and “gross pathology” were done. There was no description of what was actually done in the gross pathology. One treated male mouse died. His pathology results were not supplied, but it was declared that “as no other clinical signs were observed in animals of any group, these signs are not considered to be treatment related”. No biochemistry, immune function, neurology, liver function, kidney function, gut function, complete autopsy, cancer or teratological measurements were taken.

In the compositional studies, the GE corn and its control were grown in a randomised block experimental design in both the USA and Canada. Therefore, the composition of the GE plant can be directly compared to its non-GE control as both types of plants were grown in the same site and hence received the same rainfall, temperature, etc. Corn from the USA site used the T25-2 hybrid. Two of five (40%) of the proximate measures (protein and carbohydrate), two of six (33%) of the fatty acids and seven of eighteen (39%) of the amino acids measured were statistically different in the GE corn compared to its control. Calcium and phosphorus concentrations were also statistically different. The calcium concentration in the GE plant was only 36% of its control. Corn from the Canadian site used a different hybrid, T25-5. Here, two of five (40%) of the proximate measures (protein and carbohydrate), two of six (33%) of the fatty acids and one of eighteen (6%) of the amino acids were statistically different. When data from all field trials were combined, a process that would be expected to reduce differences between the GE plant and its control, one of five (20%) of proximate measures (fat), four of six (67%) of fatty acids and one of eighteen (6%) of amino acids were still statistically different.

Although analyses were done comparing sprayed GE corn with non-sprayed GE corn, no data were given, just a statement that no significant differences were found.

The composition of kernels was also assessed from the USA field trials. Here, three of five (60%) of proximate measures (protein, carbohydrate and ash), two of five (40%) of fatty acids and eight of eighteen (44%) of amino acids were statistically different. Calcium and phosphorus results were not given. No sample sizes were given for any of these results.

On the basis of these results, it is clear that the genetic engineering of this corn has produced significant unexpected and unexplained differences in the plant and its kernels as measured by compositional analyses. Therefore, the GE plant cannot be regarded as “comparable” or “substantially equivalent” as ANZFA claims and should undergo thorough safety testing before consumption.

The only animal feeding study of the whole corn was done on male broiler chickens. Chickens were fed ad libitum on either the GE corn line T25 or a control, for 42 days, after which eight birds from each group was sacrificed. Food intake and body weights were measured on all birds. Carcass weight, abdominal fat pad weight and total deboned breast meat yield were measured on the sacrificed birds, an inadequate sample size to determine statistical significance. A “normal” mortality rate of 7% was recorded for the birds. No other data are given but it was stated that no statistical differences were found. Although birds who died underwent post mortems, no data on what was done, what was found or any statistical differences were given.

As chickens are an unusual model for human health studies and the measurements taken (eg abdominal fat pad weight and total deboned breast meat yield) are rather unusual measures of human health, it can only be assumed that these studies were done more for reassuring primary producers of the ability of animal feed containing this corn to permit chickens to grow, than for assessing the risk to human health.

Specific comments on Application A378. Food derived form glyphosate-tolerant sugarbeet line 77 (GTSB77) (“Roundup Ready Sugarbeet)

This application covers one line of GE sugarbeet. There are no published articles reported in the ANZFA document to back-up claims of safety.

This glyphosate-tolerant sugarbeet has the cp4-epsps gene which codes for the CP4-EPSPS enzyme which permits the plant's aromatic amino acid synthesis to function normally in the presence of glyphosate, thereby making the sugarbeet resistant to glyophosate. This plant also has the uidA gene which codes for the enzyme b-D-glucuronidase (GUS) which acts as a marker for plant transformation. Both of these proteins were found throughout the sugarbeet plant. b-D-glucuronidase catalyses a catabolic reaction in plants, animals and people. It breaks down a group of substances known as b-D-glucuronidases, which also occur in vertebrates, to D-glucuronate. Therefore, this enzyme is able to find substrates to breakdown in the tissues of animals and people.

The main products of sugarbeet are refined sugar, molasses and dietary fibre. While neither of these proteins were found in the first two products, the belief that “neither of these proteins would be expected to be present in refined dietary fibre”, another product of sugarbeet, should be verified.

The GE sugarbeet also has part of a gox gene which codes for the glyphosate oxidoreductase enzyme to degrade glyophosate. Only 69% of the gene was transferred during transformation and it is fused to sugarbeet DNA, resulting in a chimeric gox sequence. Messenger RNA transcripts of this gene are made by the plant, which makes a new protein, called “Protein 34550”, which is comprised of 89 amino acids of the CTP1 sequence, 299 amino acids from the N-terminus of the GOX protein and 43 amino acids from sugarbeet DNA.

[to be completed]

Specific comments on Application A379. Oil and linters form Broxynil-tolerant cotton transformation events 10211 and 10222

ANZFA has relied on the US EPS’s evaluation of the toxicity of bromoxynil as posing a “negligible human health risk”. No information is given as to its fat solubility although its chemical structure suggests that it would be quite fat soluble. ANZFA has ignored the following literature, obtained from a simple literature search. Pregnant rats and mice had a higher frequency of supernumary ribs (SNR) (any degree of ossification lateral to the first lumbar vertebrae) that was 4.4 times higher in bromoxynil-teated rats and 4.1 times higher in treated mice than controls (Chernoff et al, 1991). In another study, bromoxynil was associated with low weight and higher incidence of supernumary ribs in mice foetuses and maternal mortality, while in rats, it caused an increased incidence in supernumary ribs in foetuses and an increased liver to body wieght ratio in dams (Rogers et al, 1991). In yet another study, bromoxynil reduced plasma TT3 and TT4 levels in rats, indicating that this herbicide may lower plasma thyroid hormone levels through interference with hormone transport carriers (Van den Berg et al, 1991). Consequently, we view with concern the planting of bromoxynil-tolerant cotton as it will increase the use of bromoxynil.

Cotton trash is also used as a fodder for cattle, etc. The effects of this GE cotton trash on these animals or on the people who eat these animals has not been assessed.

This application is for two “transformation events”. It should be clear how many GE plants this refers-to.

[to be completed]

References

Takahashi H, Toya T, Matsumiya N, Koyama K (2000). A case of transient diabetes insipidus associated with poisoning by a herbicide containing glufosinate. Journal of Toxicology – Clinical Toxicology. 38(2):153-6.

Commonwealth Department of Health and Aged Care and the Commonwealth Department of Agriculture, Fisheries and Forestry - Australia (2000). The Commonwealth Government Response to the Report of the Joint Expert Technical Advisory Committee on Antibiotic Resistance (JETACAR). Commonwealth Department of Health and Aged Care and the Commonwealth Department of Agriculture, Fisheries and Forestry, Canberra, Australia.