Agriculture Reference
In-Depth Information
process, but had exploratory goals. Taken
together, these results validate the step-by-
step approach, i.e. there is no indication that
long-term studies should be performed
mandatorily. Further support showing that
the concepts used up to now to assess the
safety of GM food and feed are sound has
also been provided recently by the use
of technologies such as metabolomics,
proteomics and transcriptomics (see Ricroch
et al
., 2011, for a review).
parameters, but should not replace the use
of a well-characterized isogenic line.
Other weaknesses in the studies
examined here are the absence of repetitions
(see below), over-interpretation of dif-
ferences, which are often within the normal
range of variation, and poor toxicological
interpretation of the data.
h e major l aws in some papers underline
the need to improve the reviewing process
before publication of papers addressing this
subject, in order to avoid confusion in the
general press. Many unfounded allegations in
the media regarding the health hazard of GM
food and feed could have been avoided if non-
specialized scientii c reviews had not
published the results of experiments which
did not meet the internationally recognized
criteria. h us, qualii ed scientists (e.g. from
the EFSA, etc.) would not lose valuable time
in analysing and refuting these publications.
Very few published long-term feeding
studies of genetically modii ed organisms
(GMOs) use the same animal model and the
same plant model, and do not consider the
same parameters. Hence, no trial has been
carried out twice in the same conditions by
dif erent research teams. h is wide diversity
of models makes it hard to perform analysis
of the results on a large scale. h erefore,
improvement in the protocols should be
made, particularly focusing on repro-
ducibility.
As discussed before (Snell
et al
., 2012),
although long-term and multi-generational
studies would rarely be used in a step-by-
step assessment of the safety of GM whole
food, they could play an important role for
its future improvement (e.g. to assess the
ef ects of a particular diet, such as a specii c
amount of crop, or to i nd out which amount
of GM material per diet is the most
appropriate) and to validate new method-
ologies.
studies and harmonization of protocols
h e standard procedures outlined in the
OECD Test (1998) recommend the use of at
least ten animals per sex and per group, with
three doses of the test substance and a
control group. Two multi-generational
studies used a reduced size sample of three
animals (Brake and Evenson, 2004; Brake
et
al
., 2004). In some studies, the number of
animals is correct per treatment, while the
number per sex is not clearly mentioned.
Inadequate experimental design in these
studies has disabling ef ects on the statistical
analysis (see internationally agreed
statistical methods: EFSA, 2011). A balance
should be found between an experimental
design allowing robust toxicological inter-
pretations and a reasonable cost.
h e plant material and its description
constitute another major problem. Out of
the 33 studies examined, 15 did not state
the use of isogenic lines as a control. h is has
been the case in the studies by Malatesta
et
al
. (see above), which have been severely
criticized (Williams and DeSesso, 2010) due
to six methodological errors. In addition, it
is a general weakness of toxicological studies
where the feed being tested and compared to
a near-isogenic line (i.e. the best comparator
available) may not provide feed with fully
identical composition. h erefore, if changes
are observed, they can be caused by the
dif erences between cultivars and not
specii cally by the transgene. Inclusion of
commercial cultivars can help to establish
whether the observed values fall within the
range of variation observed for dif erent
Up to now, long-term and multi-generational
studies have been performed as exploratory
fundamental research projects. To date,
none of these studies has proven the need to
