Detailed comment on Carman et al (2013): study design and conduct

Given the authors claim that effects of the GM diet include gastroenteric effects and effects on the female reproductive system, it is surprising that the mycotoxin assays did not include trichothecenes, which could cause gastroenteric inflammation, and zearalenone, which has oestrogenic effects. Trichothecenes and zearalenone are common contaminants in grain-based animal feeds in the Midwest of the United States.

There is a lack of information on the composition of the control (non-GM) and GM diets. This does not allow the impact of other dietary factors, unrelated to the GM trait, to be excluded.

Only a single GM diet was used which precludes the possibility of determining if a dose-response relationship exists.

No analyses were done to confirm that the particle size of the diets was equivalent. This is surprising given the authors' themselves noted that the gastric mucosa of pigs is very sensitive to the particle size of the diet.

The group sizes are very large (14 followed by 42) which makes any calculation of feed intake subject to large uncertainty.

Mortalities in both groups are extremely high by industry standards. This suggests there may have been confounding stressors affecting the pigs.

There is no apparent reason as to why the intestines were not weighed. Failure to examine the mucosa of the intestines, and the intestinal contents, is also a major deficiency. If the pigs had been suffering blood loss from gastric ulcers for some time, as the authors seem to believe, then this might be evident in rectal contents, so these should have been examined.

The authors claim that the stomachs showed 'inflammation' based on the presence of hyperaemia (reddening) but have failed to establish that inflammation was present because there is no histopathology. Inflammation can only be confirmed by demonstrating the infiltration of inflammatory cells (leukocytes such as polymorph neutrophils, lymphocytes and macrophages). The rugae (internal folds) of the stomach of the GM-fed pigs do not appear to be swollen relative to those in the stomach of the non-GM pig. If there was genuine inflammation, oedema, leukocyte infiltration and fibrosis would cause obviously thickened rugae.

The mean stomach-to-body weight ratio of the GM-fed pigs is reported to be comparable to that of the non-GM fed pigs, yet if the GM-fed pigs had been suffering gastric inflammation for weeks, oedema, infiltration of inflammatory cells (leukocytes) and fibrosis would be expected, and these changes are likely to lead to a significant increase in stomach weight, relative to body weight.

Given that the pigs identified as B15, D22 and C34 (stomachs photographed in Figure 1) were given the same diet of GM food, the difference in gross appearance between their stomachs is considerable. Acute stress can cause hyperaemia of the gastric mucosa therefore this study may have been confounded by the stress of fasting and slaughter. Pigs become very agitated and stressed if they see or hear other pigs in the same pen being stunned for slaughter. Thus it would be important to know the order of slaughter relative to the severity of gastric hyperaemia, and whether the pigs had seen other pigs stunned before they were themselves stunned, but this information is not provided.

It is unfortunate that regional lymph nodes were not collected since if the stomachs are really inflamed, the draining lymph nodes should also be enlarged and reactive, relative to those of control pigs.

It is surprising that the observed 'inflammation' did not affect feed intake, feed conversion ratio or final body weight. Trichothecenes, for example, cause a dramatic decline in feed intake in association with gastrointestinal erosions.

Given that the authors attribute ulceration to the GM feed, it is surprising that haematological parameters (haematocrit, red cell count, reticulocyte count, MCV, MCH, MCHC, RDW, WBC and differential count) were not determined. The authors acknowledge in the discussion that haematology could be informative, but do not explain why it was not done on this study.

FSANZ disagrees with the authors' statement that standard haematology and serology provide 'poor measure[s] of inflammation'. On the contrary, white cell count and differential are very sensitive measures of inflammation while fibrinogen, total proteins and albumin:globulin ratio are very frequently informative and sufficiently sensitive.