Let’s talk about rats.
That’s right, rats.
Specifically, Sprague-Dawley rats.
It turns out that Sprague-Dawley rats are very susceptible
to getting spontaneous cancers.
All kinds of cancers—pituitary, mammary, pancreatic, adrenal, liver,
thyroid, ovarian, uterine, prostate, skin, kidney, bladder, stomach, brain . .
. you get the idea. This is
the reason that these rats are commonly used for carcinogenic studies. Makes sense—if you want to find out if a
compound is carcinogenic, you might as well start out with an animal that is
prone to getting cancer. Even if
you do absolutely nothing to these rats, between 50% and 70% of them will
develop some kind of neoplasm (any abnormal growth of tissue, whether benign or
lethal).
But if you knew that 50% would develop a neoplasm of some
type, how many cancers above 50% would you need in order to “prove” that your
chemical caused cancer? Well, this
is the kind of statistical problem that scientists face all the time: how do you show that the effect of the
chemical is not due to chance alone, like flipping a coin?
First of all, you need LOTS of rats. I think that’s intuitively obvious,
especially if you’ve ever flipped coins.
If you flip only 10 times, you might get 10 heads in a row, but if you
flip 1,000 times, you’ll get a lot closer to the statistically-expected 50/50
split between heads and tails.
So how many rats do you need? Well, the scientific
world has decided that 50 rats per treatment is a minimum for these kinds of
carcinogenic studies. The Organization for Economic Co-operation and Development
recommends “at least” 50 animals per group. The Environmental Protection Agency states: “Current standardized carcinogenicity studies in rodents test at
least 50 animals per sex per dose group in each of three treatment groups and
in a concurrent control group, usually for 18 to 24 months, depending on the
rodent species tested . . . .”
By now, you are probably wondering why I’m going
through all of this. The answer is that a 2012* paper
published in a respected scientific journal DID NOT follow
these guidelines.
They didn’t even follow the guidelines for a
well designed high school science project.
And yet their paper is being used by the
press to “prove” that genetically engineered crops (“genetically engineered
organisms” or “GMOs”) are undesirable.
This has caused hysteria around the world—and some of
that hysteria is on the part of scientists who want the paper retracted.
Before we get into the nitty-gritty of the
experiment, we need to know what the term “genetic engineering” really means. There are some people who would say
that ANY plant breeding constitutes genetic engineering, even the old-fashioned
kind that involves transferring pollen by hand from one plant to another. But in the context of the GMO
debate, most people would say genetic engineering consists of
moving individual genes from one organism to another using molecular biology
techniques developed since the 1970’s. And for our purposes here, all we need to know is that
DNA, the chemical that makes up genes, can be extracted from any known organism
and transferred to nearly any other known organism. Genes can be moved from bacteria to humans, for example—with
a high degree of precision. They
can be “engineered.”
And with that little bit of background, here’s the experiment that caused all the furor:
The test subjects were Sprague-Dawley
rats that were fed a diet including various percentages of corn. The “normal” (control) group received
33% non-GMO corn, and the remaining rats received either 11%, 22%, or 33% GMO
corn. The experiment ran for two
years, during which time the researchers kept track of the number of rats that
died and those that got one or more cancerous growths.
There were 10 rats in each treatment group for
each sex. When you do the math, it
turns out that with an expected 50% baseline Sprague-Dawley mortality rate, it
takes 9 rats in a single group to die (or survive) in order to “prove” with
statistical significance that the cause was something other than chance.
Here is the mortality data for each treatment group in the
study—
Number of deaths
(male rats)
|
Non-GMO Corn
|
11% GMO
|
22% GMO
|
33% GMO
|
|
3
|
5
|
1
|
1
|
Number of deaths (female rats)
|
Non-GMO Corn
|
11% GMO
|
22% GMO
|
33% GMO
|
|
2
|
3
|
7
|
4
|
So, in six of the eight treatment groups, the number
of deaths was no different than what you would expect get a coin toss—remember
that because these rats are bred to get cancer, an average of 5 in each group would
die regardless of their diet. And what
about the other two groups, the ones that had only one death apiece? Statistically, those results ARE
significant because fewer rats in these groups died than predicted by chance
alone. The conclusion? Since these rats were fed the two highest
rates of GMO corn, it must have been good for them. HA!
Now, the baseline mortality rate I used for
these calculations was 50%, but the mortality rate for Sprague-Dawley
rats over the course of two years can be as high as 70%. If we do the same calculations again but
with a 70% baseline mortality rate, it turns out that ALL the rats in a given
treatment group would have to die or survive in order for the results to have any
statistical significance.
The point here is that the number of rats assigned to each
treatment group in this study is entirely too small for the results to be at
all meaningful. This is
highlighted by the fact that the data show no relationship between the
mortality rate for a given treatment group and the “dose” of GMO corn in that
group’s diet. After all, if a particular
substance actually caused the rats to die, you would expect the rats receiving
the most of that substance to have the highest mortality rate—but just the
opposite was observed in this study.
So what this means is that all of the study data is apparently
just random statistical noise.
Which raises an interesting question—why did the researchers
use such small numbers of rats? Were they really incapable of designing a robust, meaningful
study? Or did they just WANT to generate mortality in treatment groups as
cheaply as possible in order to raise an alarm?
I must say that upon a quick glance, the
original paper is scary. The raw
data is arresting (35% of the rats in the GMO groups DIED), the accompanying
photographs of gigantic tumors are lurid, and reports of pituitary cancer are
enough to frighten anybody. It is
only after doing some homework that you begin to realize that these same results
are found in a high percentage of ALL Sprague-Dawley rats.
Furthermore, a 2012 review paper that looked at
GMOs in 24 studies for maize, soybeans, potato, rice, and triticale found NO papers
showing that GMO crops have any negative impact on health. Perhaps we shouldn’t be surprised that
the authors of the Sprague-Dawley rat study do not cite any of these papers.
A related issue has to do with the
“composition” of GMOs. There has
been a concern ever since the introduction of GMO crops that their protein
and/or carbohydrates and/or fats had somehow been altered by the introduction
of new genes. This has never made
any sense to me—I mean, is it really likely that an herbicide-resistance gene would
affect the fat content of soybeans? But under the theory that the nutritional content of GMOs might
somehow be different from that of varieties produced by traditional means, the
“compositional equivalence” of GMOs has been examined since 1993 (thus adding
about $1 million dollars to the cost of producing a new GMO variety).
In fact, in Europe at least eight field sites
must be used, with each site to include both GMO and non-GMO lines for
comparison purposes.
Let’s be clear here: “traditional” plant
varieties, produced by conventional techniques, do NOT require any type of
compositional testing—even though traditional plant breeding has the potential to
cause radical changes in the genome.
For example, if ancestral parents are crossed with modern varieties,
very strange progeny can result; a virtual “earthquake” of random genetic
effects can be induced, including newly-introduced genes that have not been
seen in the modern varieties for thousands of years.
In my opinion, GMOs are subjected to additional
requirements not because of rational scientific concerns, but rather because of
fear on the part of the general public—these additional tests are just
roadblocks designed to prevent or delay the introduction of new GMO crops.
This is borne out by a 2013 paper that reviewed
20 years’ worth of GMO studies covering corn, soybeans, cotton, canola, wheat,
potato, alfalfa, rice, papaya, tomato, cabbage, pepper, raspberry, and
mushrooms. Guess what they found? GMOs are compositionally indistinguishable
from non-GMOs. In fact, plants
produced by “traditional methods” show more variation in nutritional
composition than GMOs do. Which
actually makes sense given the various ways genetic variation is introduced
using conventional techniques (radiation, chemical mutagenesis, somaclonal
variation, wide crosses).
And in the same vein as the Sprague-Dawley rat
study, there is yet another 2013 paper that is being used by some parties to
show that GMO crops do not have a yield benefit. Now, if this is in fact the case, why are farmers growing GMOs? I mean, millions of acres of GMO crops
are planted worldwide, including 90% of the corn, cotton and soybean acreage in
the United States alone. If there
is no benefit in terms of yield, I would have to conclude that all of these
farmers are just plain dumb. Since
that seems unlikely, one has to wonder if perhaps there is another explanation.
So, once again, we need to look at the paper to
see what the data actually says.
For this study, 4,748 hybrid corn varieties
were grown across Wisconsin from 1990 to 2010. 2,653 of these varieties were conventional hybrids
(non-GMO), and 2,095 were GMOs. (My
first thought is to question how 4,748 hybrids could possibly be developed for a
minor corn-producing state like Wisconsin and consequently, how well any of
these hybrids were adapted to growing conditions there. I would rather have seen this study
performed in Ohio or Illinois.)
The GMO varieties were divided into 12 groups:
one group was tolerant to Round Up herbicide; another was tolerant to
glufosinate herbicides; another produced Bacillus thuringiensis (Bt) toxin
against European corn borer; another produced Bt toxin against corn root
worm. Some hybrids had two-way gene
combinations, such as resistance to both Round Up and corn rootworm, and others
had three-way combinations.
The study data show that three of the GMO
groups had yields that were statistically lower than the conventional hybrids,
three of the GMO groups were statistically superior to the conventional hybrids,
and the remaining six groups were statistically the same as the conventional
hybrids. So on average, it appears
that the GMO yields are equivalent to the non-GMO yields, and this is what is
being touted by the GMO naysayers.
But what they ignore is that ALL of the GMO crops had greater STABILITY than
the conventional hybrids.
Why is this important? Because when a crop has “stability”, it consistently gives
the same yield from one year to the next and from one environment to
another. This means less risk for
the farmers because they can plan on getting a certain yield from their crop every
year, even if it is grown in a different place or under different conditions. The risk-reduction benefits of crop stability
are so predictable that agricultural economists attach an actual value to it
(called a “risk premium”). In
other words, they can predict the added economic benefit that will accrue to
the farmer from growing a particular high-stability, low-risk variety. That benefit is expressed in terms of bushels/acre
because it is equivalent to the increase in profit that would result if yields increased
by a certain amount.
In the Wisconsin corn study, the researchers
showed that the reduced risk due to the stability of the GMO varieties was equivalent
to an increase of 0.78 - 4.19 bushels/acre over the actual yield.
So even the studies that were apparently designed to show genetically-engineered
crops in a bad light were unable to do so. This is due to the fact that genetic engineering is closely
aligned with a phenomenon known as “horizontal gene transfer,” which is the
transfer of genetic information between different species in the absence of
mating—something that nature has been doing for, oh, about 500 million years. And horizontal gene transfer is not
limited to the transfer of genes between closely related organisms. A 2012* paper showed that a particular moss
had acquired 57 different families of nuclear genes from bacteria, fungi, and viruses. These genes are related to vascular
development, cuticle and epidermis, hormones, stomata pattern, herbivore
resistance, plastid development and pathogen resistance.
Now, that is genetic engineering on a grand scale!
My point is that plants/animals/bacteria/viruses/fungi have
been mixing it up for millions of years, and genetic engineering by the hand of
man is not inherently different from genetic engineering by Mother Nature.
Useful references:
**http://research.sustainablefoodtrust.org/wp-content/uploads/2012/09/Final-Paper.pdf
http://pubs.acs.org/doi/full/10.1021/jf400135r
No comments:
Post a Comment