Andrew Apel
Posted: Friday, December 5, 2008 10:52 am

Scientists around the world are scrutinizing the Zentek mouse study, and we are sure to hear from more of them in due course. Politicians are a far less reticent crowd, but they're being uncharacteristically quiet about their deliberations -- despite Austria's efforts to promote the study. The press are mostly copying each other's coverage, while inventing claims regarding 'impotence' and 'infertility'. In sum, the study appears to be a bombshell that won't go off -- possibly because a bit of scrutiny shows the report is either largely inconclusive, or a blow to the anti-biotech camp. The main problem? Conventional maize has far more impact on mice than engineered maize.

The promotional effort

On November 13, Dr. Eva Claudia Lang of Austria's Federal Ministry of Health, Family and Children (Bundesministerium für Gesundheit, Familie und Jugend) circulated an email to over 400 public officials in the European Commission and EU member states urging them to download Zentek's study, "Biological effects of transgenic maize NK603 x MON810 fed in long term reproduction studies in mice", from the ministry's website.

This was two days after Greenpeace first broke the story of Zentek's study, which was revealed on Nov. 11 at a seminar in Vienna. With publicity from Greenpeace's media experts, Austria's promotion, and "potential unknown long-term side-effects" near the top of the activist wish list, the impact of the Zentek study should have been on a par with similarly-promoted studies by Ermakova, Pusztai and others.

Promotional effort fails

It's possible that skepticism is emerging over research findings that are not peer-reviewed, at least in the field of GMOs. There may even be an emerging consensus that dodging peer review, at least in the field of GMOs, is an admission that the paper would fail peer-review, and that the authors have a keener interest in sensationalism than in science.

In this case, it's far more likely that scrutiny is silently at work -- because Dr. Zentek's preliminary findings, freely available online, paint a picture quite opposite to what Austria and Greenpeace would prefer regulators and the world at large to perceive.

The main problem? Conventional maize has far more impact on mice than engineered maize.

Problems with the paper - multi-generation study

a. comparison with GM feed.

In the multi-generation study, mice gave issue to successive generations of offspring -- i.e., parents gave birth to infant mice, who, in their turn, became parents for the next generation. Regardless of diet, the report finds "[n]o differences were seen in performance of the parental mice in all generations", and [n]o statistically significant differences were seen in reproduction data between the two feeding groups".

b. comparisons between non-GM feeds.

When the scientists compared results between the non-GM equivalent of the mouse feed (ISO), and a mouse feed based on maize grown in Austria (A REF), a startling picture emerged.

Differences in parental performance of females at delivery in the F1 and F3 generation where females from the ISO group were significantly lighter than females from the A REF group. In those generations, the body mass of females 3 weeks after delivery was significantly different and again females from the ISO group were significantly lighter than females from the A REF group.

Differences in body mass of males was seen in the F2 generation and males from the ISO group were significantly lighter than males from the A REF group. In the ISO group body mass of females and males at mating, females at delivery, and 3 weeks after delivery, differed significantly over several generations. In the A REF body mass differed significantly at mating in the F2 > F1 in females and F3>F1 in males.

In other words, conventional maize appears to have been an unhealthy diet for the mice.

d. comparing GM organ weights.

Comparing GM feed to feed made from ISO, the scientists looked at organ weights. The best they could say was, "Relative organ weights showed significant differences between groups that were not consistent through the investigation."

Consistent inconsistency is another way of saying "random", or "inconclusive".

Things were not quite so inconclusive when it came to comparing the non-GM ISO and A REF results.

e. comparing non-GM organ weights.

The authors downplayed the significance of their findings. "Although a number of significant differences concerning relative organ weights had been found", they said, "these differences could not be corroborated by the microscopic comparisons between the feeding groups in any of the above mentioned organs."

In this portion of the study, they concluded: "[t]here was no evidence of diet related changes in the tissues of the gastrointestinal tract, liver, pancreas, kidneys, spleen, lungs and testes."

f. immunochemistry.

Immunohistochemistry was equally inconclusive.  "The differences are inconsistent between the two sexes and were not found in all segments. For the CD3+ immune population the impact of feed seems rather low", the scientists wrote. The same was true of CD20 + B-lymphocytes and macrophages: there was either too much "inter-individual variability of the results", or "[n]o statistically significant differences were seen between the groups."

g. ultrastructural investigation.

Ultrastructural investigation showed dramatic differences. Comparing the ISO to its GM version, the scientists could find "[n]o significant divergences ... in the spleen and pancreatic cells."

The dramatic difference emerged when comparing ISO and A REF feeds. [em added] "A significant variation regarding the nuclear shape irregularity was only ascertained in liver cells of female mice", they said, "which was lower (0.025) in the A REF group in comparison to the ISO group (Table 47) The pore density of lymphocyte nuclei in the spleen was significantly lower in males (p=0.026) in the A REF group than in the ISO Group."

h. gene micorarrays.

The authors admit that the value of their gene microarray analyses are questionable, but their results are nonetheless suggestive.

When comparing ISO to GM feed, they found "439 genes were ... expressed differentially". Comparing ISO to A REF, they said: "In total, 1016 genes were found to be differentially expressed".

Problems with the paper - continuous breeding.

In continuous breeding, a set of parent mice give birth to successive litters, and these litters are compared and evaluated.

The proponents of the Zentek study point almost solely to this section of the paper as proof that GM feed damages mice.

As a slap in the face to interpretations rendered by Greenpeace, the Daily Mail, et. al. regarding offspring performance, the scientists actually said, "all the consecutive average litter weights failed to meet the level of significance by a small margin".

Before authors explain observed differences between mice fed GM and non-GM feed, they will need to explain the far greater observed differences between mice fed two versions of conventional maize.

Coda.

On p. 14, the authors explain that "A difference was considered statistical[ly] significant at p < 0.05." At this level of significance, the author(s) of a paper should expect to discover at least one 'false positive' among twenty tests.

This paper lacks a working hypothesis, i.e., a plausible statement of fact which tests would verify or falsify. Instead, it describes a program to establish a low level of significance, and then conduct so many tests that something was bound to emerge "significantly".

With such a low criterion for "significance", and so many tests, "significant" results were were nearly guaranteed.

Even so, the results they got suggest the most "significant" impacts on the mice come from conventional feed -- leading to the silence of credible media and cautious politicians about whether this paper means anything at all.