26.08.2013 Views

Repeated-dose 90-day Oral Toxicity Studies on Whole ... - Europabio

Repeated-dose 90-day Oral Toxicity Studies on Whole ... - Europabio

Repeated-dose 90-day Oral Toxicity Studies on Whole ... - Europabio

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<str<strong>on</strong>g>Repeated</str<strong>on</strong>g>-<str<strong>on</strong>g>dose</str<strong>on</strong>g> <str<strong>on</strong>g>90</str<strong>on</strong>g>-<str<strong>on</strong>g>day</str<strong>on</strong>g> <str<strong>on</strong>g>Oral</str<strong>on</strong>g><br />

<str<strong>on</strong>g>Toxicity</str<strong>on</strong>g> <str<strong>on</strong>g>Studies</str<strong>on</strong>g> <strong>on</strong> <strong>Whole</strong><br />

Food/Feed<br />

Presenters:<br />

Kevin Leiner, Margaret Nemeth, and<br />

Keith Ward<br />

1


Introducti<strong>on</strong><br />

• EFSA has stated that its guidance <strong>on</strong> repeated-<str<strong>on</strong>g>dose</str<strong>on</strong>g> <str<strong>on</strong>g>90</str<strong>on</strong>g>-<str<strong>on</strong>g>day</str<strong>on</strong>g> oral toxicity<br />

studies in rodents <strong>on</strong> whole food/feed is “not intended to provide<br />

prescriptive experimental test protocols.”<br />

• EuropaBio Members agree there should be flexibility in the design of future<br />

studies with whole food/feed and that justificati<strong>on</strong> for the approach to the<br />

study is important.<br />

• The new EFSA guideline will:<br />

– Impact design, c<strong>on</strong>duct, analysis and interpretati<strong>on</strong> both toxicologically and<br />

statistically.<br />

– Create challenges that require additi<strong>on</strong>al discussi<strong>on</strong> with EFSA and other<br />

stakeholders in the EU.<br />

• EuropaBio Members appreciate the opportunity to discuss these new<br />

challenges, and their proposed soluti<strong>on</strong>s to them, with key stakeholders in<br />

the EU.<br />

2


Existing safety data and the<br />

weight of evidence<br />

• Prior to the c<strong>on</strong>duct of the <str<strong>on</strong>g>90</str<strong>on</strong>g>-<str<strong>on</strong>g>day</str<strong>on</strong>g> study several safety<br />

assessments have been c<strong>on</strong>ducted including:<br />

– Compositi<strong>on</strong>al analysis, agr<strong>on</strong>omic performance, and molecular analysis of the<br />

crop.<br />

– Evaluati<strong>on</strong>s of the inserted trait and its protein product:<br />

• Bioinformatic evaluati<strong>on</strong>s, in vitro digestibility and heat stability assays as well as<br />

protein toxicity studies <strong>on</strong> a case-by-case basis.<br />

• Establishing Substantial Equivalence between the GM and<br />

c<strong>on</strong>venti<strong>on</strong>al crop, and an absence of adverse findings in the<br />

indicated tests, provides a weight of evidence for the safety of<br />

the whole food (Codex, 2009).<br />

• EuropaBio Members believe the 15 + year history of safe use of<br />

GM crops evaluated and approved under this approach<br />

attests to its strength.<br />

3


Hypothesis of safety and<br />

study justificati<strong>on</strong><br />

• In such cases, EuropaBio Members believe there is no<br />

specific hypothesis to test in the <str<strong>on</strong>g>90</str<strong>on</strong>g>-<str<strong>on</strong>g>day</str<strong>on</strong>g> study and<br />

thus no justificati<strong>on</strong> for them.<br />

– EFSA published opini<strong>on</strong> appears to agree with this positi<strong>on</strong>.<br />

• EFSA Panel <strong>on</strong> Genetically Modified Organisms (GMO). Scientific opini<strong>on</strong>-<br />

Guidance for risk assessment of food and feed from genetically modified<br />

plants. EFSA Journal 2011; 9(5):2150<br />

• However, if potential indicati<strong>on</strong>s of toxicity are<br />

present in the existing data, then EuropaBio Members<br />

believe the principles of Hazard Identificati<strong>on</strong> should<br />

be followed to maximise the possibility of detecting<br />

toxic effects of the test diet.<br />

4


Hazard Identificati<strong>on</strong> and<br />

multiple <str<strong>on</strong>g>dose</str<strong>on</strong>g> levels<br />

• For Hazard Identificati<strong>on</strong> purposes, administering a <str<strong>on</strong>g>dose</str<strong>on</strong>g> lower than the<br />

highest possible (which maintains nutriti<strong>on</strong>al balance) is of disputable<br />

value.<br />

• Testing at the highest possible c<strong>on</strong>centrati<strong>on</strong> maximises exposure and thus<br />

the potential to detect unanticipated adverse effects.<br />

– C<strong>on</strong>sistent with the limit <str<strong>on</strong>g>dose</str<strong>on</strong>g> approach detailed in OECD TG 408 (1998).<br />

– Requirement of nutriti<strong>on</strong>al balance limits c<strong>on</strong>centrati<strong>on</strong> in diet.<br />

– EuropaBio Members estimate the highest possible c<strong>on</strong>centrati<strong>on</strong> of soybean<br />

and maize will typically exceed human intake in the EU 10X to 100X (WHO,<br />

2003; EFSA 2011c).<br />

• It is difficult to scientifically justify the inclusi<strong>on</strong> of the low <str<strong>on</strong>g>dose</str<strong>on</strong>g> group if<br />

regulating <strong>on</strong> the basis of Hazard Identificati<strong>on</strong>.<br />

– A “clean” low <str<strong>on</strong>g>dose</str<strong>on</strong>g> level would then likely be irrelevant if the high <str<strong>on</strong>g>dose</str<strong>on</strong>g> is<br />

adverse.<br />

5


The regulatory paradigm for<br />

GM crops<br />

• Based <strong>on</strong> practical experience, EuropaBio<br />

Members have presumed the current<br />

regulatory paradigm for GM crops is based<br />

<strong>on</strong> Hazard Identificati<strong>on</strong> (or rather an<br />

absence of Hazard).<br />

• EFSA’s inclusi<strong>on</strong> of a low <str<strong>on</strong>g>dose</str<strong>on</strong>g> level could be<br />

interpreted as a willingness to regulate GM<br />

crops <strong>on</strong> the principle of Risk.<br />

– Risk = Hazard × exposure<br />

6


Secti<strong>on</strong> summary<br />

• EuropaBio Members believe:<br />

– When there is no evidence of unintended changes, the<br />

<str<strong>on</strong>g>90</str<strong>on</strong>g>-<str<strong>on</strong>g>day</str<strong>on</strong>g> study would not normally be expected to add<br />

value to the safety assessment – it is largely<br />

c<strong>on</strong>firmatory.<br />

– The current regulatory paradigm for GM crops is based<br />

<strong>on</strong> Hazard Identificati<strong>on</strong>.<br />

– The purpose of testing a <str<strong>on</strong>g>dose</str<strong>on</strong>g> below the highest dietary<br />

incorporati<strong>on</strong> level that maintains nutriti<strong>on</strong>al balance is<br />

unclear and needs to be resolved.<br />

7


Discussi<strong>on</strong><br />

• EuropaBio Members agree with EFSA’s Scientific<br />

Opini<strong>on</strong> (2011a) that an applicant could forego<br />

c<strong>on</strong>ducting a <str<strong>on</strong>g>90</str<strong>on</strong>g>-<str<strong>on</strong>g>day</str<strong>on</strong>g> study if data indicating the GM<br />

crop and its parental isoline:<br />

– Are Substantially Equivalent in regard to nutriti<strong>on</strong>al compositi<strong>on</strong>.<br />

– Dem<strong>on</strong>strate no indicati<strong>on</strong>s of unintended adverse effects in<br />

molecular, compositi<strong>on</strong>al, or phenotypic analyses.<br />

• EuropaBio Members request clarificati<strong>on</strong> <strong>on</strong> whether<br />

inclusi<strong>on</strong> of two (or more) <str<strong>on</strong>g>dose</str<strong>on</strong>g> levels is mandatory<br />

and, if so, how results from the low <str<strong>on</strong>g>dose</str<strong>on</strong>g> will be<br />

used.<br />

8


Part 1: Study design and c<strong>on</strong>duct<br />

9


Challenges in study design<br />

and c<strong>on</strong>duct<br />

1. Designing for appropriate power<br />

2. Experimental design and c<strong>on</strong>duct<br />

a) Practical aspects regarding cage locati<strong>on</strong><br />

b) Variati<strong>on</strong> c<strong>on</strong>trols<br />

3. Housing c<strong>on</strong>diti<strong>on</strong>s and impact <strong>on</strong> the<br />

Experimental Unit (ExpU)<br />

4. Statistical power investigati<strong>on</strong><br />

10


1. Defining toxicologically<br />

relevant differences<br />

• As per guidance, relevant endpoints, c<strong>on</strong>sistent with OECD TG<br />

408, will be m<strong>on</strong>itored.<br />

– TG 408 specifies that each test animal will be m<strong>on</strong>itored <strong>on</strong> >200 individual<br />

endpoints.<br />

– Ensures a thorough examinati<strong>on</strong> of the whole animal and that known<br />

mechanisms of toxicity will be evaluated.<br />

• Toxicologists will review the data for c<strong>on</strong>sistent trends in related<br />

endpoints associated with toxic mechanisms.<br />

– ↑ serum ALT, ↑ liver weight, histopathological altera<strong>on</strong>s of liver ssue.<br />

– ↑ serum ALT, in the absence of other indica<strong>on</strong>s of liver damage, is of equivocal<br />

importance.<br />

• To reach scientifically valid c<strong>on</strong>clusi<strong>on</strong>s, the weight of evidence<br />

must be c<strong>on</strong>sidered to determine if the difference observed is<br />

treatment-related.<br />

11


1. Defining toxicologically relevant<br />

differences (c<strong>on</strong>tinued)<br />

• Identifying a toxicologically-relevant difference for individual<br />

endpoints in isolati<strong>on</strong> is very difficult, especially when there<br />

is no evidence of adverse changes in related parameters.<br />

• However, to perform power calculati<strong>on</strong>s, relevant effect sizes<br />

must be defined.<br />

• EFSA guidance acknowledges that this is no easy task, and<br />

does not provide specific guidance in this respect.<br />

• To determine endpoints and effect sizes with the greatest<br />

potential to identify toxicity, EuropaBio toxicologists<br />

examined numerous guidance documents.<br />

• Guidance documents <strong>on</strong> determining a Maximum Tolerated<br />

Dose (MTD) provided definitive endpoints and effect sizes.<br />

12


1. Endpoints and effect<br />

sizes<br />

Endpoint Change relative to c<strong>on</strong>trol<br />

Body weight Decrease 10%<br />

Liver weight Increase 25%<br />

Kidney weight Increase 25%<br />

Leukocyte count Decrease/increase 30%<br />

Lymphocyte count Decrease/increase 30%<br />

Cholesterol Increase 200%<br />

Blood urea nitrogen Increase 50%<br />

Creatinine Increase 50%<br />

Alkaline phosphatase Increase 200%<br />

(Foster, 2002; EPA, 2002; EPA, 2003; Rhomberg et al., 2007).<br />

13


1. Toxicological endpoints<br />

for power calculati<strong>on</strong>s<br />

• The MTD has also been defined as the <str<strong>on</strong>g>dose</str<strong>on</strong>g><br />

that, “…does not produce mortality, clinical<br />

signs of toxicity, or pathologic lesi<strong>on</strong>s…”<br />

(Dorato et al., 2008; S<strong>on</strong>tag, 1976).<br />

• Hence, it is a <str<strong>on</strong>g>dose</str<strong>on</strong>g> that elicits signs more<br />

subtle than overt toxicity.<br />

• Endpoints and effect sizes are taken from the<br />

indicated guidance documents.<br />

14


1. Toxicological endpoints for<br />

power calculati<strong>on</strong>s (c<strong>on</strong>tinued)<br />

• The toxicological relevance of any differences observed will be<br />

determined <strong>on</strong> a case-by-case basis.<br />

– Alkaline phosphatase factors: AST levels, ALT levels, liver weight, liver<br />

histopathology, etc.<br />

• All effect sizes proposed have been c<strong>on</strong>sidered <strong>on</strong> the scale of<br />

original units.<br />

– This is the scale that makes most sense to toxicologists as it provides<br />

necessary c<strong>on</strong>text for evaluati<strong>on</strong> of biological significance.<br />

• Specifying effect sizes <strong>on</strong> the original units scale removes the<br />

need to c<strong>on</strong>sider effect sizes <strong>on</strong> the standardised scale.<br />

– To be discussed in greater detail later.<br />

15


1. Secti<strong>on</strong><br />

summary/discussi<strong>on</strong><br />

• EuropaBio Members believe the endpoints<br />

and toxicologically-relevant effect sizes chosen<br />

provide a robust platform <strong>on</strong> which to base<br />

subsequent power calculati<strong>on</strong>s for the study<br />

as a whole.<br />

• All effect sizes proposed have been<br />

c<strong>on</strong>sidered <strong>on</strong> the scale of original units; the<br />

scale which makes the most sense to<br />

toxicologists.<br />

16


2. Experimental design<br />

• EFSA guidance advocates use of a randomised complete block<br />

design (RCBD) in which males and females are randomised<br />

together.<br />

• However, current toxicology study animal husbandry practices<br />

intenti<strong>on</strong>ally separate males and females (e.g., by assigning<br />

genders to different cage racks).<br />

• There is a c<strong>on</strong>cern that housing opposite genders in adjacent<br />

cages, particularly in paired housing setups, may lead to intra-<br />

MSK2 MSK3<br />

cage aggressi<strong>on</strong> in males.<br />

• EuropaBio Members’ str<strong>on</strong>g recommendati<strong>on</strong> is that the<br />

current practice of gender separati<strong>on</strong> be maintained.<br />

17


Slide 17<br />

MSK2 Follow up with technical folks.<br />

Animal behavior studies.<br />

Best practices, etc.<br />

Michael Koch; 1/10/2012<br />

MSK3 Will be stated in much str<strong>on</strong>ger language if available informati<strong>on</strong> allows. Still being researched.<br />

"Housing genders in adjacent cages is specifically avoided to prevent intra-cage aggressi<strong>on</strong>.."<br />

Michael Koch; 9/10/2012


2. Experimental design<br />

(c<strong>on</strong>tinued)<br />

• EFSA guidance states that use of designs other than<br />

RCBD “may be acceptable provided that appropriate<br />

justificati<strong>on</strong> for using them is given”.<br />

• EuropaBio Members agree that, in circumstances<br />

where a good reas<strong>on</strong> for blocking exists, an RCBD is<br />

an appropriate choice.<br />

• However, in the absence of a known blocking<br />

structure, a completely randomised design (CRD) is<br />

the most obvious choice and indeed is likely to be<br />

slightly more powerful than an RCBD.<br />

18


2. Variati<strong>on</strong> c<strong>on</strong>trols<br />

• The following variati<strong>on</strong> c<strong>on</strong>trol measures are<br />

typically utilised to minimise variati<strong>on</strong> am<strong>on</strong>gst the<br />

ExpUs:<br />

– Age-matched animals (single stock, single supplier)<br />

– A small range in initial body weight is specified.<br />

– Housed in a single, envir<strong>on</strong>mentally-c<strong>on</strong>trolled room.<br />

• Temperature, humidity, photoperiod, and fresh air exchanges<br />

closely m<strong>on</strong>itored.<br />

– In some cases, cage-rack positi<strong>on</strong> is periodically rotated<br />

within the animal room to further minimise potential<br />

envir<strong>on</strong>mental variability.<br />

19


2. Secti<strong>on</strong><br />

summary/discussi<strong>on</strong><br />

• EuropaBio Members seek agreement<br />

from EFSA that:<br />

– Maintaining the current practice of<br />

physically separating males and females is<br />

acceptable.<br />

– The absence of obvious blocking factors is<br />

sufficient justificati<strong>on</strong> for the use of a CRD.<br />

20


3. Group or single housing<br />

• EFSA’s Scientific Opini<strong>on</strong> document states that “In accordance with<br />

the European Directive 2010/063, the test animals should be housed<br />

socially”, and recommends that animals of the same sex are housed in<br />

pairs.<br />

• EuropaBio Members understand this requirement for studies<br />

c<strong>on</strong>ducted within the EU.<br />

• For studies performed in the U.S., housing animals singly is acceptable<br />

from an animal welfare perspective and is comm<strong>on</strong>place.<br />

• EuropaBio Members believe group and single housing are equally valid<br />

and should remain opti<strong>on</strong>s.<br />

– Existing historical c<strong>on</strong>trol data based <strong>on</strong> single housing remains relevant.<br />

– It acknowledges different animal welfare requirements exist in testing<br />

locati<strong>on</strong>s utilized for whole food toxicity testing.<br />

21


3. Group or single housing<br />

(c<strong>on</strong>tinued)<br />

• In cases where animals are housed two (or more)<br />

per cage, the guidance makes clear that cage<br />

should be regarded as the ExpU.<br />

• The guidance then explains how it is possible to<br />

test for cage effects and, in cases where the<br />

between-cage variati<strong>on</strong> is not significantly greater<br />

than the between-animal variati<strong>on</strong>, the statistical<br />

analysis can be based <strong>on</strong> individual animals.<br />

22


3. Group or single housing<br />

(c<strong>on</strong>tinued)<br />

• While the analysis can be d<strong>on</strong>e in this way, the majority of<br />

EuropaBio statisticians believe it is more appropriate to<br />

c<strong>on</strong>duct tests of main effects at the cage level regardless of<br />

whether the cage effect is significant or not (in which case<br />

testing the significance of the cage effect is unnecessary).<br />

• EuropaBio members seek assurance that the statement<br />

“statistical analysis can be based <strong>on</strong> the individual animals” is<br />

not prescriptive and that an approach in which main effects are<br />

always tested at the cage level is acceptable.<br />

23


3. Secti<strong>on</strong><br />

summary/discussi<strong>on</strong><br />

• EuropaBio members believe single or pair housing<br />

arrangements are valid for toxicology studies with<br />

whole foods and request that EFSA permit flexibility<br />

in animal housing c<strong>on</strong>diti<strong>on</strong>s c<strong>on</strong>sistent with local<br />

jurisdicti<strong>on</strong>.<br />

• EuropaBio members seek assurance that an<br />

approach in which main effects are always tested at<br />

the cage level is acceptable (in which case testing<br />

the significance of the cage effect would be<br />

unnecessary).<br />

24


4. Statistical power<br />

investigati<strong>on</strong><br />

• EuropaBio Members have carried out an extensive<br />

power investigati<strong>on</strong> based <strong>on</strong> existing <str<strong>on</strong>g>90</str<strong>on</strong>g>-<str<strong>on</strong>g>day</str<strong>on</strong>g> feeding<br />

data.<br />

– Mean resp<strong>on</strong>se levels and estimates of variati<strong>on</strong> based <strong>on</strong> 13 studies<br />

for corn-based diets and 10 studies for soy-based diets.<br />

– Some studies based <strong>on</strong> <strong>on</strong>e animal per cage and others based <strong>on</strong> two<br />

or more animals per cage.<br />

• The endpoints and effect sizes used in this exercise<br />

were those discussed previously, that are widely<br />

c<strong>on</strong>sidered by experts in the field to be<br />

toxicologically-relevant.<br />

25


4. Endpoints and effect<br />

sizes<br />

Endpoint Change relative to c<strong>on</strong>trol<br />

Body weight Decrease 10%<br />

Liver weight Increase 25%<br />

Kidney weight Increase 25%<br />

Leukocyte count Decrease/increase 30%<br />

Lymphocyte count Decrease/increase 30%<br />

Cholesterol Increase 200%<br />

Blood urea nitrogen Increase 50%<br />

Creatinine Increase 50%<br />

Alkaline phosphatase Increase 200%<br />

26


4. Measures of variati<strong>on</strong><br />

• Given that effect sizes are expressed in percentage terms, the most<br />

appropriate measure of variati<strong>on</strong> is the coefficient of variati<strong>on</strong> (CV).<br />

• Cage was regarded as the ExpU; CV at the cage level is dependent<br />

<strong>on</strong> the number of rats per cage.<br />

• Evidence of differences in CV between males and females was<br />

detected.<br />

• For each of the endpoints c<strong>on</strong>sidered, CV’s for a given gender and<br />

number of rats per cage were found to be fairly c<strong>on</strong>sistent across<br />

studies.<br />

– Also, no evidence to suggest that CV differed between corn and<br />

soy-based diets.<br />

• So, for each endpoint, an average CV was generated for each<br />

gender / housing policy combinati<strong>on</strong>, to be used in subsequent<br />

power calculati<strong>on</strong>s.<br />

27


4. Average coefficients of<br />

variati<strong>on</strong><br />

Endpoint<br />

1 animal/cage<br />

Male Female<br />

2 animals/cage<br />

Male Female<br />

Body weight 9.4% 8.3% 4.9% 4.9%<br />

Liver weight 12.4% 10.3% 7.9% 6.9%<br />

Kidney weight 10.8% 10.5% 6.4% 6.5%<br />

Leukocyte count 24.3% 28.5% 16.6% 20.3%<br />

Lymphocyte count 27.5% 34.7% 18.1% 21.7%<br />

Cholesterol 22.6% 22.2% 12.9% 16.7%<br />

Blood urea nitrogen 14.0% 14.9% 10.0% 9.9%<br />

Creatinine 19.5% 16.9% 7.5% 7.5%<br />

Alkaline phosphatase 21.8% 28.0% 12.8% 20.8%<br />

28


4. Design scenarios<br />

c<strong>on</strong>sidered<br />

Three design scenarios were c<strong>on</strong>sidered:<br />

• 1 animal/cage; 3 treatment groups x 12 cages per treatment<br />

per gender = 72 animals in total<br />

– high <str<strong>on</strong>g>dose</str<strong>on</strong>g>, low <str<strong>on</strong>g>dose</str<strong>on</strong>g>, c<strong>on</strong>trol<br />

• 2 animals/cage; 3 treatment groups x 6 cages per treatment<br />

per gender = 72 animals in total<br />

– high <str<strong>on</strong>g>dose</str<strong>on</strong>g>, low <str<strong>on</strong>g>dose</str<strong>on</strong>g>, c<strong>on</strong>trol<br />

• 2 animals/cage; 4 treatment groups x 5 cages per treatment<br />

per gender = 80 animals in total<br />

– GM high incorporati<strong>on</strong> rate, n<strong>on</strong>-GM high incorporati<strong>on</strong> rate,<br />

GM low incorporati<strong>on</strong> rate, n<strong>on</strong>-GM low incorporati<strong>on</strong> rate<br />

29


4. Analysis details<br />

• For the 3-treatment group designs, power was based <strong>on</strong> two<br />

analysis scenarios:<br />

– the comparis<strong>on</strong> of individual treatments,<br />

– the comparis<strong>on</strong> between c<strong>on</strong>trol and mean of high and low <str<strong>on</strong>g>dose</str<strong>on</strong>g>s.<br />

• For the 4-treatment group design, power was based <strong>on</strong> the<br />

comparis<strong>on</strong> of treatments (i.e. GM vs n<strong>on</strong>-GM) averaged over<br />

incorporati<strong>on</strong> rates, in line with EFSA guidance.<br />

• In all cases, cage was regarded as the ExpU.<br />

• Males and females were c<strong>on</strong>sidered separately.<br />

• All results assume a two-sided alternative hypothesis.<br />

30


4. Results<br />

1 animal/cage, 3 treatment groups, 12 replicate<br />

cages/group; individual treatment comparis<strong>on</strong>s<br />

Endpoint Effect Size<br />

Predicted Power<br />

Males Females<br />

Body weight 10% 72% 82%<br />

Liver weight 25% >99% >99%<br />

Kidney weight 25% >99% >99%<br />

Leukocyte count 30% 84% 71%<br />

Lymphocyte count 30% 74% 54%<br />

Cholesterol 200% >99% >99%<br />

Blood urea nitrogen 50% >99% >99%<br />

Creatinine 50% >99% >99%<br />

Alkaline phosphatase 100% >99% >99%<br />

31


4. Results<br />

1 animal/cage, 3 treatment groups, 12 replicate<br />

cages/group; c<strong>on</strong>trol vs. mean of low and high <str<strong>on</strong>g>dose</str<strong>on</strong>g><br />

Endpoint Effect Size<br />

Predicted Power<br />

Males Females<br />

Body weight 10% 83% 91%<br />

Liver weight 25% >99% >99%<br />

Kidney weight 25% >99% >99%<br />

Leukocyte count 30% 92% 82%<br />

Lymphocyte count 30% 85% 66%<br />

Cholesterol 200% >99% >99%<br />

Blood urea nitrogen 50% >99% >99%<br />

Creatinine 50% >99% >99%<br />

Alkaline phosphatase 100% >99% >99%<br />

32


4. Results<br />

2 animals/cage, 3 treatment groups, 6 replicate<br />

cages/group; individual treatment comparis<strong>on</strong>s<br />

Endpoint Effect Size<br />

Predicted Power<br />

Males Females<br />

Body weight 10% 91% 91%<br />

Liver weight 25% >99% >99%<br />

Kidney weight 25% >99% >99%<br />

Leukocyte count 30% 83% 67%<br />

Lymphocyte count 30% 77% 61%<br />

Cholesterol 200% >99% >99%<br />

Blood urea nitrogen 50% >99% >99%<br />

Creatinine 50% >99% >99%<br />

Alkaline phosphatase 100% >99% >99%<br />

33


4. Results<br />

2 animals/cage, 3 treatment groups, 6 replicate<br />

cages/group; c<strong>on</strong>trol vs. mean of low and high <str<strong>on</strong>g>dose</str<strong>on</strong>g><br />

Endpoint Effect Size<br />

Predicted Power<br />

Males Females<br />

Body weight 10% 97% 97%<br />

Liver weight 25% >99% >99%<br />

Kidney weight 25% >99% >99%<br />

Leukocyte count 30% 92% 79%<br />

Lymphocyte count 30% 87% 73%<br />

Cholesterol 200% >99% >99%<br />

Blood urea nitrogen 50% >99% >99%<br />

Creatinine 50% >99% >99%<br />

Alkaline phosphatase 100% >99% >99%<br />

34


4. Results<br />

2 animals/cage, 4 treatment groups, 5 replicate cages/group;<br />

GM vs. n<strong>on</strong>-GM averaged over incorporati<strong>on</strong> rates<br />

Endpoint Effect Size<br />

Predicted Power<br />

Males Females<br />

Body weight 10% 99% 99%<br />

Liver weight 25% >99% >99%<br />

Kidney weight 25% >99% >99%<br />

Leukocyte count 30% 97% 87%<br />

Lymphocyte count 30% 94% 83%<br />

Cholesterol 200% >99% >99%<br />

Blood urea nitrogen 50% >99% >99%<br />

Creatinine 50% >99% >99%<br />

Alkaline phosphatase 100% >99% >99%<br />

35


4. Provisos for power<br />

analyses<br />

• Power investigati<strong>on</strong> for 3-treatment group scenario<br />

comparing c<strong>on</strong>trol with mean of high and low <str<strong>on</strong>g>dose</str<strong>on</strong>g>s<br />

assumes that, in reality, mean of low <str<strong>on</strong>g>dose</str<strong>on</strong>g> = mean<br />

of high <str<strong>on</strong>g>dose</str<strong>on</strong>g>.<br />

• Power investigati<strong>on</strong> for the 4-treatment group<br />

design assumes that there is no interacti<strong>on</strong> between<br />

treatment (i.e. GM vs. n<strong>on</strong>-GM) and incorporati<strong>on</strong><br />

rate.<br />

– Analysis of previous data indicates that this<br />

assumpti<strong>on</strong> should typically hold in practice.<br />

36


4. Should males and females<br />

be analysed together?<br />

• Physically separating males and females does not<br />

necessarily preclude analysing data from both genders<br />

together but there are good reas<strong>on</strong>s for doing so:<br />

– Analysis across genders for some variables is problematic due<br />

to well-established gender-specific differences in variance,<br />

which experience suggests cannot be readily overcome by the<br />

use of a data transformati<strong>on</strong>.<br />

– Large differences in gender means (e.g., for body and organ<br />

weights) may make results of a combined analysis more<br />

complicated to interpret (Gad, 2006).<br />

– Analysis across genders would be more complex and could be<br />

difficult for some CRO’s to implement correctly.<br />

37


4. Should males and females<br />

be analysed together?<br />

• Reas<strong>on</strong>s for analysing genders separately (c<strong>on</strong>tinued):<br />

– Analysis by gender is the standard approach in toxicology.<br />

– Certain endpoints (e.g., testes or uterus weights) are limited to<br />

<strong>on</strong>e gender and thus necessitate separate analyses by default.<br />

– Results from the power investigati<strong>on</strong> show that powers based<br />

<strong>on</strong> gender-specific analyses are generally more than adequate<br />

for the designated purpose.<br />

• EuropaBio Members therefore recommend analysing<br />

genders separately.<br />

38


4. Secti<strong>on</strong><br />

summary/discussi<strong>on</strong><br />

• The power profile as a whole is very positive for each of the<br />

three proposed design scenarios.<br />

– Each scenario strikes a reas<strong>on</strong>able balance between<br />

power and ethical c<strong>on</strong>siderati<strong>on</strong>s.<br />

• Findings dem<strong>on</strong>strate that the power inherent in previous<br />

studies of this size (i.e., 10-12 animals per treatment group<br />

per gender, in line with OECD TG 408) should in general be<br />

similarly adequate.<br />

• While power is important, the toxicological relevance of all<br />

observed differences are discussed in any case regardless of<br />

statistical significance.<br />

39


Part 2: Study analysis and<br />

interpretati<strong>on</strong><br />

40


Challenges in study analysis<br />

and interpretati<strong>on</strong><br />

1. Toxicological interpretati<strong>on</strong><br />

2. Standardised effect sizes (SES)<br />

3. Statistical Analysis Plan & summary<br />

statistics and raw analysis outputs<br />

41


1. Toxicological<br />

interpretati<strong>on</strong><br />

• EuropaBio Members appreciate that EFSA has commented <strong>on</strong> the<br />

imperfect relati<strong>on</strong>ship between statistical significance and<br />

toxicological relevance in the final versi<strong>on</strong> of the guidance.<br />

• However, without feedback from EFSA regarding the flexibility that<br />

will be permitted in overall design, some c<strong>on</strong>cerns about the<br />

prominence of the statistical analysis persist.<br />

• EFSA (2011b) recently stated, “Many researchers incorrectly c<strong>on</strong>clude<br />

that any statistically significant effect is biologically relevant...”<br />

– This gives EuropaBio Members c<strong>on</strong>fidence that EFSA c<strong>on</strong>siders statistical analysis<br />

<strong>on</strong>e tool of many to be used to identify toxicologically-relevant differences.<br />

• The scientific training of toxicologists, and their experience in<br />

interpreting data, must ultimately determine the relati<strong>on</strong>ship<br />

between the finding and treatment with the test substance.<br />

42


2. Presentati<strong>on</strong> of results <strong>on</strong><br />

the standardised scale<br />

• Given that all endpoints are to be analysed and presented <strong>on</strong><br />

the scale of original units, EuropaBio Members questi<strong>on</strong> the<br />

need to also present results <strong>on</strong> the standardised scale.<br />

– “If the original units of measurement are meaningful, the presentati<strong>on</strong><br />

of unstandardised effect statistics is preferable over those of<br />

standardised effect statistics”. (Nakagawa and Cuthill, 2007)<br />

• Original units are meaningful to trained toxicologists.<br />

– “If researchers are familiar with their study systems, or abundant<br />

previous research <strong>on</strong> a topic of interest exists, effect sizes in original<br />

units are more readily interpretable than standardised effect statistics”.<br />

(Nakagawa and Cuthill, 2007)<br />

• The study system is familiar to trained toxicologists and the basis of<br />

decades of previous research.<br />

43


2. Presentati<strong>on</strong> of results <strong>on</strong><br />

the standardised scale<br />

• When results are presented <strong>on</strong> the standardised<br />

scale all studies of the same design will appear to<br />

be equally sensitive regardless of any differences in<br />

underlying variability.<br />

• The guidance states that “Changes in endpoints<br />

…must also be evaluated with respect to normal<br />

biological variati<strong>on</strong> of the endpoints”; this can be<br />

d<strong>on</strong>e <strong>on</strong> the scale of original units but not <strong>on</strong> the<br />

standardised scale.<br />

44


Original Scale Standardised Scale<br />

Lab 1 Lab 2 Lab 1 Lab 2<br />

Mean C<strong>on</strong>trol 9.77 9.77 -1.45 -0.59<br />

Mean GM low 12.44 12.44 0.84 0.34<br />

Mean GM high 12.17 12.17 0.61 0.25<br />

SD 1.17 2.85 1.00 1.00<br />

CV 10% 25% NA NA<br />

p-value GM high vs C<strong>on</strong>trol 0.003 0.165 0.003 0.165<br />

GM high - C<strong>on</strong>trol 2.40 2.40 2.06 0.84<br />

95% CI<br />

GM high - C<strong>on</strong>trol<br />

0.97 - 3.83 -1.11 - 5.92 0.83 - 3.29 -0.39 - 2.07<br />

Width of 95% CI 2.86 7.03 2.46 2.46<br />

Comparis<strong>on</strong> with historical<br />

ranges possible?<br />

2. Example of original vs.<br />

standardised scaling<br />

yes yes no no<br />

45


2. Secti<strong>on</strong><br />

summary/discussi<strong>on</strong><br />

• EuropaBio Members see the presentati<strong>on</strong> of<br />

results <strong>on</strong> the standardised scale as<br />

providing no additi<strong>on</strong>al benefit, and<br />

respectfully request that this aspect of the<br />

guidance be relaxed from mandatory to<br />

opti<strong>on</strong>al.<br />

46


3. Statistical Analysis<br />

Plan<br />

• From the Scientific Opini<strong>on</strong> document it is unclear<br />

whether it is mandatory to provide a separate SAP<br />

in additi<strong>on</strong> to the details c<strong>on</strong>cerning the proposed<br />

statistical analysis that are included in the study<br />

protocol.<br />

• EuropaBio Members wish to avoid unnecessary<br />

duplicati<strong>on</strong>, and believe that the level of detail that<br />

will be provided in the study protocol should<br />

suffice.<br />

47


3. Summary statistics and<br />

raw analysis outputs<br />

• EuropaBio Members’ understanding is that EFSA wishes to see<br />

results presented both in terms of means and p-values, and<br />

differences and 95% c<strong>on</strong>fidence intervals.<br />

• However, some requirements are less clear, and others (e.g., the<br />

median) seem to be of questi<strong>on</strong>able value in most cases.<br />

• It is the opini<strong>on</strong> of EuropaBio Members that the following<br />

summary statistics would generally be appropriate:<br />

– Mean, min, max and SD for each treatment group.<br />

– P-values for all fixed effects and specific c<strong>on</strong>trasts.<br />

– Point estimate of the difference between GM and n<strong>on</strong>-GM together with<br />

95% c<strong>on</strong>fidence interval for the difference.<br />

• EuropaBio Members seek c<strong>on</strong>firmati<strong>on</strong> from EFSA as to whether<br />

the above list is acceptable.<br />

48


3. Summary statistics and<br />

raw analysis outputs<br />

• With regard to the provisi<strong>on</strong> of raw data in electr<strong>on</strong>ic<br />

format, the guidance document is inc<strong>on</strong>sistent<br />

regarding whether this is a default requirement or “<strong>on</strong><br />

request”.<br />

– Please clarify.<br />

• The guidance document states that “The statistical<br />

analysis programs, logs and outputs should be<br />

provided for the purposes of review”.<br />

– To provide all this informati<strong>on</strong> by default seems excessive.<br />

– Please clarify whether this is a default requirement or “<strong>on</strong><br />

request”.<br />

49


3. Secti<strong>on</strong><br />

summary/discussi<strong>on</strong><br />

• Please clarify if a separate Statistical Analysis Plan is<br />

mandatory or if the level of detail that can be<br />

presented in the study protocol is adequate.<br />

• Does EFSA regard the summary statistics proposed by<br />

EuropaBio Members as suitable?<br />

• Please clarify whether raw data in electr<strong>on</strong>ic format is<br />

a default requirement or “<strong>on</strong> request”.<br />

• Please clarify whether the statistical analysis<br />

programs, logs and outputs are a default requirement<br />

or “<strong>on</strong> request”.<br />

50


References<br />

• Codex Alimentarius. 2009. Foods derived from modern biotechnology. Codex<br />

Alimentarius Commissi<strong>on</strong>, Joint FAO/WHO Food Standards Programme, Food and<br />

Agriculture Organizati<strong>on</strong> of the United Nati<strong>on</strong>s: Rome, Italy.<br />

• Dorato, M.A., McMillian, C.L., Vodicnik, M.J. “The Toxicologic Assessment of<br />

Pharmaceutical and Biotechnology Products”. Principles and Methods of<br />

Toxicology. Ed. A.W. Hayes. New York: Informa Healthcare USA, Inc., 2008. pp 325-<br />

368.<br />

• EFSA. 2008. Safety and nutriti<strong>on</strong>al assessment of GM plants and derived food and<br />

feed: The role of animal feeding trials. Food and Chemical Toxicology 46:S2-S70.<br />

• EFSA. 2011a. Guidance for risk assessment of food and feed from genetically<br />

modified plants. EFSA Journal 9:2150.<br />

• EFSA. 2011b. Scientific opini<strong>on</strong>: Statistical significance and biological relevance.<br />

EFSA Journal 9:2372.<br />

51


References<br />

• EFSA. 2011c. Guidance of EFSA. Use of the EFSA Comprehensive European Food<br />

C<strong>on</strong>sumpti<strong>on</strong> Database in Exposure Assessment. EFSA Journal 2011;9:2097.<br />

Supplemental informati<strong>on</strong>:<br />

http://www.efsa.europa.eu/en/datexfooddb/datexfooddbspecificdata.htm.<br />

• EPA. 2002. Health effects Divisi<strong>on</strong> (HED) Guidance Document. # G0201.<br />

Hepatocellular Hypertrophy. October 21, 2002.<br />

• EPA. 2003. Health effects Divisi<strong>on</strong> (HED) Interim Guidance Document. # G2003.02.<br />

Rodent Carcinogenicity <str<strong>on</strong>g>Studies</str<strong>on</strong>g>: Dose selecti<strong>on</strong> and evaluati<strong>on</strong>. July 1, 2003.<br />

• Foster, P.M.D. 2002. EPA. 23 July 2002.<br />

<br />

• Gad, S. 2006. Statistics and Experimental Design for Toxicologists and<br />

Pharmacologists. Boca Rat<strong>on</strong>: Taylor & Francis Group.<br />

• Nakagawa, S., and I.C. Cuthill. 2007. Effect size, c<strong>on</strong>fidence interval and statistical<br />

significance: a practical guide for biologists. Biol Rev Camb Philos Soc 82:591-605.<br />

52


References<br />

• OECD. 1998. Guideline for the Testing of Chemicals – <str<strong>on</strong>g>Repeated</str<strong>on</strong>g> Dose <str<strong>on</strong>g>90</str<strong>on</strong>g>-<str<strong>on</strong>g>day</str<strong>on</strong>g> <str<strong>on</strong>g>Oral</str<strong>on</strong>g><br />

<str<strong>on</strong>g>Toxicity</str<strong>on</strong>g> Study in Rodents, 408.<br />

<br />

• OECD. 2012. Guidance Document 116 <strong>on</strong> the C<strong>on</strong>duct and Design of Chr<strong>on</strong>ic<br />

<str<strong>on</strong>g>Toxicity</str<strong>on</strong>g> and Carcinogenicity <str<strong>on</strong>g>Studies</str<strong>on</strong>g>, Supporting Test Guidelines 451, 452 and453<br />

2nd Editi<strong>on</strong>.<br />

• Rhomberg, L.R., K. Baetcke, J. Blancato, J. Bus, S. Cohen, R. C<strong>on</strong>olly, R. Dixit, J.<br />

Doe, K. Ekelman, P. Fenner-Crisp, P. Harvey, D. Hattis, A. Jacobs, D. Jacobs<strong>on</strong>-Kram,<br />

T. Lewandowski, R. Liteplo, O. Pelk<strong>on</strong>en, J. Rice, D. Somers, A. Turturro, W. West,<br />

and S. Olin. 2007. Issues in the design and interpretati<strong>on</strong> of chr<strong>on</strong>ic toxicity and<br />

carcinogenicity studies in rodents: approaches to <str<strong>on</strong>g>dose</str<strong>on</strong>g> selecti<strong>on</strong>. Crit Rev Toxicol<br />

37:729-837.<br />

• S<strong>on</strong>tag, J.M., Page, N.P., and Safiotti, V. 1976. Guidelines for Carcinogen Bioassays<br />

in Small Rodents, DHHS Publ. (NIH) 76-801. Nati<strong>on</strong>al Cancer Institute, Bethesda,<br />

MD.<br />

53


References<br />

• WHO. 2003. GEMS/FOOD Regi<strong>on</strong>al Diets: Regi<strong>on</strong>al per Capita C<strong>on</strong>sumpti<strong>on</strong> Of<br />

Raw and Semi-processed Agricultural Commodities. Revisi<strong>on</strong> September 2003.<br />

54


C<strong>on</strong>tributing EuropaBio<br />

• Bayer CropScience<br />

• BASF<br />

• Dow AgroSciences<br />

• DuP<strong>on</strong>t Pi<strong>on</strong>eer<br />

• M<strong>on</strong>santo<br />

• Syngenta<br />

Members<br />

55

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!