Recommendations for design of the rat comet assay

Although the rodent comet assay is gaining acceptance as a standard technique for evaluating DNA damage in vivo, there is no internationally accepted guideline for its conduct and several aspects of its experimental design have not been optimized. For example, no standard positive control is used, there is no agreement on how tissue toxicity should be measured and sources of experimental variability have not been considered in relation to experimental design. This study showed that methylnitrosourea is a good alternative positive control inducing DNA damage in all tissues examined (stomach, liver, blood and bone marrow) over a dose range of 25–100 mg/kg at both 3 and 24 h after treatment. At the highest dose, significant toxicity was seen in all tissues using the neutral diffusion assay and also by histopathological/haematological analysis, except in the liver where no change was seen even 7 days after dosing. Analyses using control data pooled from several studies showed that, as expected, the greatest variability was seen between tissue preparations from different animals and that different numbers of animals were required to detect the same fold increases in different tissues. Power analyses showed that, preparing three gels for each tissue and scoring 50 nuclei per gel, a group of six animals allows 2-fold increases over control in the liver, bone marrow and stomach and a 3-fold increase in blood to be detected with 80% probability. It is recommended that similar investigations of experimental variability should be performed to determine optimal experimental design in any laboratory using the rodent comet assay.


Introduction
The rodent comet assay is gaining acceptance as a standard technique for evaluating DNA damage in vivo and has been proposed as an alternative to unscheduled DNA synthesis in rat liver when a second in vivo test is required for regulatory submissions with novel pharmaceutical agents.In terms of regulatory acceptance, the Food and Drug Administration Guidance on Integration of Genetic Toxicology results (2006) (1) considers that it can be useful in clarifying in vitro positive results and at least two regulatory submissions have been reported to include comet assay data (2,3).
The comet assay evaluates compounds for their ability to cause DNA strand breaks and alkali-labile sites (4)(5)(6) and has been used for the detection of DNA damage in cells exposed to chemical and physical agents under in vitro and in vivo conditions (4,(7)(8)(9)(10).Despite a general consensus on how the alkaline comet assay should be performed, there is currently no Organisation for Economic Co-operation and Development guideline for its design.Recommendations have been published (2,(11)(12)(13)(14) but factors such as group size are based on other in vivo assays, e.g. the bone marrow micronucleus test.More recently, the International Workgroup on Genotoxicity Testing (IWGT) considered areas such as the number of dose levels required, cell isolation techniques, measures of cytotoxicity, the parameters to assess DNA damage in comets and the need for historical control data (11).The issue of cytotoxicity is important since it is vital to distinguish DNA strand breakage that may be a direct consequence of genotoxic insult from DNA degradation resulting from cell death.Suggested methods for measuring cytotoxicity include tests for membrane integrity or metabolic competence, the frequency of cells with lowmolecular weight DNA [the neutral diffusion assay (NDA)] and histopathological examination.The later is concluded to be the 'Gold Standard' to assess levels of necrosis and apoptosis when positive results are seen in an in vivo comet assay (11).In addition, nuclei with a small or non-existent head and large diffused tails ['hedgehogs' or non-detectable cell nuclei (NDCN)] may also be important as these represent dead or dying cells and give an indication of the integrity of control preparations.Various aspects of the comet assay including the nature of DNA in comet tails, calibration of numbers of breaks, electrophoresis conditions, scoring methods and measures to assess viability have recently been reviewed by Collins et al. (15).
Although ethyl methanesulphonate (EMS) and 2-acetylaminofluorene (2-AAF) have been used as positive controls in several rat and mouse genotoxic assays, both have practical disadvantages.EMS induces DNA damage in all tissues examined in the rat (2), but presents safety issues because it is a volatile liquid and its use is restricted to licensed laboratories in some European countries, e.g.Sweden.2-AAF induces a response only in the liver (16) and it is also difficult to formulate consistently due to its physical properties.
The aim of the current work was to investigate three main aspects of the in vivo comet assay using the rat.First, to establish a better positive control than those currently used.Second, to compare different cytotoxicity endpoints, i.e. numbers of NDCN, the NDA and pathological and haematological examinations.Third, to optimize the experimental design for the numbers of animals per group, the number of gels required per tissue and the number of comets scored per gel; to achieve this, power calculations were performed using control data pooled from several studies.

Materials and methods
All chemicals and reagents were purchased from Sigma (Dorset, UK) unless specified.
*To whom correspondence should be addressed.Tel: þ44 1625 232435; Fax: þ44 1625 231281; Email: catherine.smith@astrazeneca.comÓ The Author 2008.Published by Oxford University Press on behalf of the UK Environmental Mutagen Society.All rights reserved.For permissions, please e-mail: journals.permissions@oxfordjournals.org.

Animal husbandry and treatment
Male Wistar Han rats (substrain AlpkHsdRccHan-WIST, obtained from the AstraZeneca, Alderley Park breeding unit) were approximately 9-10 weeks old at dosing and were housed three o four per cage.Environmental controls were set to maintain conditions of 19-23°C and 40-70% relative humidity, with a 12-h light-dark cycle.All animals were treated in accordance with approved UK Home Office licence requirements.
Groups of three rats were given the vehicle or test compound by oral gavage in a dose volume of 10 ml/kg.EMS (300 mg/kg) was dissolved and 2-AAF (75 mg/kg) was suspended in 0.5% hydroxypropyl methylcellulose in 0.1% w/v polysorbate 80. Methylnitrosourea [(MNU) 25, 50 and 100 mg/kg) was dissolved in sterile water.All formulations were prepared no .3h prior to dosing.The rats were killed by cervical dislocation and blood, liver, bone marrow and stomach were sampled $3 and 24 h after dosing for comet analysis and NDCN.In the study with MNU, the NDA, histopathological, haematological and plasma chemistry analyses were also performed and additional groups given the highest dose or vehicle control were analysed for all parameters 7 days after treatment.

Preparation of single-cell suspensions
Liver.The left lateral liver lobe was removed and a cube (1 cm 3 ) was excised from the lower portion.The piece of liver was washed three times in ice-cold buffer 1 [Hank's balanced salts solution (HBSS; Invitrogen, Paisley, UK) containing 25 mmol/l ethylenediaminetetraacetic acid (EDTA) disodium salt and 10% dimethyl sulfoxide (DMSO), pH7.5], and then placed in 1 ml ice-cold buffer 1 and ruptured by squeezing with tweezers 10 times.The cell suspension was mixed using a 1-ml pipette.
Stomach.The stomach was cut open along the greater curvex and washed in buffer 2 (HBSS with 24 mmol/l EDTA, 1 mmol/l EGTA and 10% DMSO added freshly, pH 7.4) and pinned out on a wax dish with the inner stomach facing upwards.The fore-stomach was removed and the glandular tissue was immersed in buffer 2 and incubated in the dark on ice for 15 min.The tissue was scraped to remove the mucosal layer prior to adding 1 ml fresh buffer 2 and being scraped a further six times with a blunt scissor blade.Samples were filtered through a 150-lm bolting cloth and the cell suspension was mixed using a 1-ml pipette.
Bone marrow.The left femur was removed and the bone marrow at both ends was exposed with bone cutters.Cells were flushed out with 3 ml phosphatebuffered saline (PBS) using a needle and syringe, and the cell suspension was filtered through a 150 lm bolting cloth.
Blood.Tail vein blood was collected into lithium-heparin tubes and centrifuged at 500 g for 2 min at 4°C.The serum was removed, an equal volume of PBS was added and the cells were re-suspended using a Pasteur pipette.

Comet assay
End-frosted slides were pre-coated with 0.5% normal melting point agarose and allowed to dry at room temperature overnight.For each tissue, cell suspensions were mixed with 0.5% low melting point agarose and 40 ll aliquots were added to three slides and covered with a 24 Â 24 mm glass coverslip.Two gels were made on each slide from different tissue samples.The gels were allowed to set on a cold plate and then the coverslips were removed.
Slides were immersed in lysis solution (2.5 mol/l NaCl, 100 mmol/l EDTA disodium salt, 10 mmol/l, pH 10, Tris buffer, 10% DMSO and 1% Triton X-100) and stored at 4°C overnight.Slides were placed on an electrophoresis platform, covered with electrophoresis buffer (1 mmol/l EDTA disodium salt and 0.3 mol/l NaOH) and DNA was allowed to unwind for 20 min before electrophoresis (0.7 V/cm, 300 mA) for a further 20 min.DNA unwinding and electrophoresis were performed in the dark in a cold unit set at 4°C.Slides were immersed in three changes of neutralizing buffer (0.4 mol/l, pH 7.5, Tris-HCl) for 5 min at room temperature and then stained with 50 ll propidium iodide (20 lg/mL).Twenty min later, slides were scored, 50 nuclei per gel, using a comet IV capture system (Perceptive Instruments).Tail intensity (TI) was used as the measure of damage and was defined as the percentage of DNA that had migrated from the head of the comet into the tail.For the liver samples, two cell types were analysed: medium-sized nuclei (between 30 and 40 lm head length), classed as hepatocytes (parenchymal cells), and small-sized nuclei (,30 lm head length), classed as non-parenchymal cells.Due to the high level of damage observed, it was impossible to differentiate between small-and medium-sized nuclei from the livers of rats given 50 and 100 mg/kg MNU after 3 h; therefore, 50 nuclei irrespective of the size were scored from each gel in the MNU study.

Cytotoxicity measures
NDCN.For all gels, 100 nuclei were assessed for the number of highly damaged cells with NDCN (hedgehog in shape).
Histopathology.Tissue samples from the same part of the left lateral lobe of the liver used for comet analysis and part of the stomach and the sternum were fixed and preserved in buffered formalin.Liver and stomach tissue were embedded in paraffin wax, sectioned at 5 lm and stained with haematoxylin and eosin for microscopic examination.Femurs were decalcified in 10% formic acid until soft enough to trim, and then processed as the liver and stomach.
Haematology and blood biochemistry.Tail vein blood was collected into EDTA tubes and analysed using the Bayer Advia 120 Haematology System.The following parameters were recorded: percentage of erythrocytes, haemoglobin, haematocrit, mean corpuscular haemoglobin concentration, mean red cell volume, red cell distribution width, reticulocytes, platelets, leucocytes, neutrophils, lymphocytes, monocytes, basophils, eosinophils and large unstained cells.
Bone marrow from the right femur was analysed by flow cytometry (17) based on a combination of the differential expression of leukocyte common antigen (CD45) on different cell lineages and the expression of transferrin receptor (CD71).
Plasma was isolated from tail vein blood collected in lithium-heparin tubes by centrifugation at 1100 g and analysed on a Roche P-module instrument.The following parameters were recorded: albumin, total protein, cholesterol, globulin, glutamate dehydrogenase, total bilirubin, alanine amino transferase, aspartate amino transferase and alkaline phosphatase.
NDA. Single-cell suspensions from all tissues were processed as in the comet assay with the omission of DNA unwinding and electrophoresis (18).The nuclei were scored in batches within 20 min after neutralization.The nuclei were scored qualitatively into two categories-(i) condensed: highly condensed to slight diffusion of the nuclei or (ii) Halo: large or extended halo.

Statistical analysis
Determination of responses in individual tissues.The TI data were log transformed (natural logs; value þ0.0001) and the mean of the log-transformed data was used as a summary measure for the nuclei on the gel.The analysis was performed using a mixed model, fitting the group as a fixed effect and animal as a random effect (19).One-sided pairwise comparisons between the test and the vehicle control group were performed at the 5% level since only increases in response were of interest.The model assessed differences in the average of the 50 nuclei counted from each gel against the between-animal variability and least squares means were calculated (20).Each tissue and each time point was analysed using a separate model.
Power analyses.In order to calculate the power of the study, a mixed model was used to obtain the variance components for the variabilities between animals, within an animal (variability between gels) and within gels (variability for different numbers of nuclei scored) for each tissue analysed.Study and time point were fitted as fixed effects, so that any differences due to these factors were removed.Data from vehicle controls from several studies were used for these analyses in order to determine the intrinsic variability without any effect of inter-individual variation in response to genotoxic agents.The resulting variance components were used to calculate the power of a one-sided test at the 5% level for different fold changes, with different numbers of slides per animal and different numbers of nuclei scored per gel.The analysis was set theoretically to detect a 2-, 2.5-and 3-fold change to compare different numbers of animals per group, numbers of gels prepared per tissue and scoring 50 or 100 nuclei per gel.In total, the analysis was based on TI data from 43 animals for liver, 36 animals for blood and 24 animals for stomach and bone marrow.The data were pooled from six studies performed over 2 years.

Control levels of DNA damage
For each tissue, the levels of DNA damage in the controls were very similar at both 3 and 24 h after dosing; therefore, the two time points were combined from the studies with EMS, 2-AAF and MNU (Tables I and III).The mean %TI levels (AEstandard deviation) were as follows: stomach, 10.1 AE 1.67% (n 5 18); non-parenchymal liver cells, 2.4 AE 0.36% (n 5 12); hepatocytes, 1.8 AE 0.34% (n 5 12); liver overall, 2.2 AE 0.29% (n 5 18); blood, 2.3 AE 0.30% (n 5 18) and bone marrow, 2.1 AE 0.31% (n 5 18).The replicate numbers for hepatocytes and non-parenchymal liver cells are less than the liver overall because the two cell types could not be differentiated in the study with MNU.These data indicate that nuclei were prepared satisfactorily from all tissues but it should be noted that the control level of DNA damage was substantially higher in the stomach.It is possible that this reflects greater levels of endogenous DNA damage in the stomach environment or that more damage was introduced during the isolation procedure or both.Because of the differences in control levels, increase in %TI was considered to be a better measure than fold increase to compare induced damage in different tissues, although both are included in Tables I, III and V.
DNA damage and toxicity induced by 2-AAF 2-AAF induced significant DNA damage only in the liver (Table I).Three hours after treatment, TI increased by 4.4% from 1.6 to 6.0% in hepatocytes with no significant increases in non-parenchymal cells.A similar increase, 4.9%, was seen in hepatocytes after 24 h and some evidence of a smaller response was also seen in non-parenchymal cells (1.5%, 1.6-fold, P , 0.05).There was no observable increase in NDCN for any of the tissues at either 3 or 24 h after treatment (Table II).

DNA damage and toxicity induced by EMS
EMS induced significant responses in all tissues examined at both 3 and 24 h after treatment (Table I).After 3 h, the rank order of increases in TI above control was stomach 12.4%, hepatocytes 10.8%, non-parenchymal cells 9.4%, bone marrow 6.8% and blood 3.8%.The number of NDCN in the stomach was also increased from 9.7 to 24.3% but there was no increase in NDCN in any other tissue (Table II).
After 24 h, the greatest effect was again seen in stomach with an increase in TI of 15.6%, followed by hepatocytes 6.0%, non-parenchymal cells 6.9%, bone marrow 5.6% and blood 6.8% (Table I).Similar to 3 h after treatment, the number of NDCN in the stomach preparations was again high (28.8%)and there were also slight increases in the blood (2.8%), bone marrow (3%) and liver (1.8%) (Table II).
DNA damage and toxicity induced by MNU Stomach.Highly significant, dose-related increases in DNA damage were seen at all dose levels at both 3 and 24 h after treatment, with increases in TI of 52-69% and 28-38%, respectively (Table III).No significant increase was seen in the group sampled after 7 days (Table V).Changes in NDCN were seen at all doses after 3 h (89-100%) and 24 h (35-21%) (Table IV) but no increase was observed after 7 days (Table V).
Significant cytotoxicity was indicated by the NDA with dosedependent decreases in the percentage of condensed nuclei after 3 and 24 h (Table IV).Consistent with this, histopathological examination showed scattered apoptotic cells within the epithelium of the fundic region at 3 and 24 h in all treatment groups.After 7 days, more severe toxicity was evident in the mucosa and submucosa of the fundic region, i.e. acute inflammatory cell infiltration, haemorrhage, oedema and epithelial cell basophilia but the NDA showed no significant difference from the controls.
Liver.Hepatocytes and non-parenchymal cells could not be distinguished because of the high levels of damage at all doses and sampling times.Highly significant, dose-related increases in DNA damage were seen at all dose levels at both 3 and 24 h after treatment, with increases in TI of 38-70% and 26-43%, respectively (Table III).A statistically significant increase in TI, 26.9%, was still detectable after 7 days (Table V).Changes

Rat comet assay design
in NDCN were seen after 3 and 24 h with 100 mg/kg (100 and 66.9%, respectively) and with 50 mg/kg only after 3 h (66.8%) (Table IV).No changes in NDCN were seen after 7 days (Table V).
Toxicity measured by the NDA showed a significant dosedependent decrease in the percentage of condensed nuclei with similar responses at 3 and 24 h (Table IV); no effect was seen after 7 days.Perhaps surprisingly, there was no histopathological change at any time point, and only minor changes in globulin, total protein and albumin were seen with no change in any other liver enzyme.
Blood.Highly significant, dose-related increases in DNA damage were seen at all dose levels at both 3 and 24 h after treatment, with increases in TI of 17-40% and 6-19%, respectively (Table III).An increase of 5.7% (P , 0.05) was seen in the group sampled after 7 days (Table V).Minimal changes in NDCN were seen after 3 and 24 h only with 100 mg/kg (2.8% at both time points) (Table IV).After 3 h, the NDA showed a decrease in the percentage of condensed nuclei only at 100 mg/kg (75.5%) but, after 24 h, there were dosedependent decreases (Table IV).There was no effect 7 days after dosing (Table V).Haematological analysis showed a marked decrease in total white cell count (42%) 3 h after 50 and 100 mg/kg and decreases in reticulocytes (45%), neutrophils (42%) and lymphocytes (15-28%) were seen at all dose levels after 24 h.Reductions in white cell count (37%) and reticulocytes (24%) were still apparent after 7 days.
Bone marrow.Highly significant, dose-related increases in DNA damage were seen at all dose levels at both 3 and 24 h after treatment, with increases in TI of 15-41% and 5-22%, respectively (Table III).No significant increase was seen after 7 days (Table V).Changes in NDCN were seen only with 100 mg/kg after 3 h (47.7%) and, minimally, after 24 h (3.8%) with 100 mg/kg only (Table IV).The NDA showed a significant dose-dependent decrease in the percentage of condensed nuclei 3 h after dosing but, at 24 h, there was a significant decrease only at 100 mg/kg (33.2%) (Table IV) and by 7 days there was no observable effect (Table V).Clinical pathology showed a marked decrease in the ratio of nucleated to non-nucleated cells 3 h after treatment with 50 and 100 mg/kg and at all dose levels after 24 h; a significant decrease was still apparent after Rat comet assay design 7 days.There was no change in the total number of nucleated cells 3 h after dosing but, at 24 h, there were marked decreases at all doses (33, 67 and 47%, respectively) and this effect was still apparent after 7 days (40%).Histopathological examination showed hypocellularity of the bone marrow in all treated groups 24 h after dosing and, although reduced, was still present after 7 days.

Power analyses to determine optimal experimental design
The aim was to determine the optimal design for a rat comet assay performed in this laboratory by taking into account the possible sources of experimental variability.For each tissue, the following parameters were investigated: the number of nuclei scored per gel, the number of gels per tissue and the number of rats per group.For every tissue, increasing the number of nuclei scored per gel from 50 to 100 resulted in virtually no increase in sensitivity.This is illustrated in Figure 1 for stomach and blood, the tissues showing the least and most variability, respectively.From this, it was concluded that 50 nuclei per gel gives sufficient accuracy and, consequently, all subsequent calculations assumed 50 nuclei scored per gel.
Increasing the number of gels from three to five, as expected, increased the sensitivity (Figure 2), but when the group size was less than or equal to six, this increase in sensitivity was not as great as that achieved by increasing the group size by one.
From these analyses it became apparent, as might have been expected, that the biggest source of variability for each tissue was between preparations from different animals.Also, variability was different for the individual tissues and, perhaps surprisingly, was smallest for the stomach where the control DNA damage was highest.The overall variability for all tissues is illustrated in Figure 3 showing the number of rats required to detect 2-, 2.5and 3-fold increases over control with 80% probability.The effect of increasing the number of gels from three to five is also summarized in Table VI.
Although the precise design of an individual study may need to be modified depending on the target organ being examined, the data presented here show that for a multi-organ study with a practicable experimental design (3 gels per tissue, 50 nuclei per gel), a group of six rats should allow a 2.5-fold increase over control in the liver, bone marrow and stomach and a 3-fold increase in blood to be detected with 80% probability.

Discussion
The techniques used here for preparing single-cell suspensions from liver, stomach, blood and bone marrow had previously been optimized in this laboratory.Cells were easily obtained from blood and bone marrow by centrifugation and filtration, respectively, but mechanical rupture was necessary for stomach and liver.For these tissues, buffers containing EDTA to chelate calcium and magnesium were used to prevent endonuclease activation, DMSO was included as a radical scavenger to limit oxidative DNA damage being introduced and it was essential for pH to be controlled accurately to minimize the incidence of NDCN.The quality of the preparations in this study was confirmed by the low incidence of DNA damage and NDCN in the controls; TI and NDCN were 1.5-2.7% and ,1%, respectively, in all tissues except the stomach where the corresponding values were $10 and 8%.It is possible that the higher values in the stomach reflect greater levels of endogenous damage, more damage introduced by the extraction procedure or a combination of both.However, it should be noted that although the control %TI was high in the stomach in comparison with the other tissues, huge dose-related increases could still be seen in response to EMS and MNU.There are various methods that can be used to score the comet assay and the consensus view from the recent IWGT workgroup (11) was that image analysis is preferred and that the amount of DNA in the tail, %TI, was the most objective.The IWGT also suggested that controls should exhibit measurable DNA migration but did not define 'measurable'.Since the extraction procedure can affect the amount of DNA in the tail, in this study, background levels were kept as low as practicable in order to be able to detect small increases in response to treatment.From the results in this study, it is proposed that %TI in control cultures should be used as a quality control measure within and between laboratories and historical control ranges should be recorded.
One of the main objectives of this work was to determine the suitability of MNU as an alternative positive control to EMS and 2-AAF.It has previously been shown to induce DNA adducts in all tissues in the rat (21) and to induce comet damage in a range of tissues in the mouse (R. Durward, SafePharm Laboratories; personal communication).Clear, statistically significant increases were seen in all tissues examined at both 3 and 24 h after treatment at all doses of MNU, 25, 50 and 100 mg/kg.Because the highest dose elicited significant signs of clinical toxicity, it is probable that 50 mg/kg will be used in future studies in this laboratory with the Wistar Han rat.However, since clear responses were seen over a 4-fold dose range, it is likely that MNU will be suitable for use in other laboratories and in other strains of rat.MNU has clear advantages over EMS in terms of safety and over 2-AAF by inducing DNA damage in all tissues so far examined and, by being soluble in water, in ease of formulation.
MNU induced toxicity in the blood and bone marrow, with decreases in white cell counts 3 h after treatment that were more severe after 24 h and still apparent after 7 days.DNA damage and cytotoxicity estimated by NDCN or NDA were greatest after 3 h and not detectable at 7 days.This is consistent with haematological toxicity being a direct consequence of the genotoxic insult.Similarly, the stomach showed significant DNA damage shortly after treatment with some evidence of histopathological changes at 3 and 24 h and significant damage to the fundic mucosa and submucosa after 7 days but with no evidence of DNA damage.In contrast, the liver showed huge increases in DNA damage and cytotoxicity estimated by NDCN or NDA but no evidence of histopathological change at any time point and only minor changes in blood biochemistry.Although this might be surprising, it is consistent with the carcinogenicity of MNU which has been shown to induce tumours in the stomach and nervous system after oral administration, but not the liver in any species by any route (22).It has been claimed that the failure of MNU to induce liver tumours in rats is due to the efficient repair of alkylated sites in DNA of the liver compared with brain (21).
Previous work with the NDA in this laboratory showed that, although it is relatively easy to produce slides when preparing slides for comet analysis, scoring is very laborious and subjective (18).Although methods have been investigated to store slides for subsequent analysis, in practice the NDA must be scored with or shortly after comet analysis.Since target organ toxicity is only an issue if positive results have been seen in the comet assay, the most practical solution would be to preserve relevant tissues for histopathological analysis if required.
The power analyses showed that, for each tissue, the biggest variability was seen between different animals and that for a practical experimental design (3 gels per tissue, 50 nuclei per gel), a group of six rats should allow a 2.5-fold increase over control in the liver, bone marrow and stomach and a 3-fold increase in blood to be detected with 80% probability.This is  Rat comet assay design consistent with the conclusions of Wiklund and Agurell (19) that increasing the number of nuclei scored .50showed no reduction in variability and that it was not practicable to increase the numbers of gels from three to five in a study examining several tissues.It has been suggested (13,19) that four to five animals should be used for the in vivo comet assay for regulatory submissions.The analyses presented here indicate that these numbers are adequate if only hepatocytes and bone marrow are being examined, but at least six are required to investigate responses in the blood or stomach.The power calculations are also directly comparable with the conclusions of Wiklund and Agurell (19) that with four or five animals per group and three slides per animal, the power would be 61-65% and 73-77%, respectively.However, in the present, the study design to detect 2-to 3-fold increases and 80% power was chosen for two reasons: (i) to minimize the number of equivocal studies and (ii) to be comparable with the rat bone marrow micronucleus test performed in this laboratory as the rat comet assay is likely to be used as a second in vivo assay.Further, it should be noted that these power calculations are based solely on experimental variability in control animals and do not consider possible 'responders' and 'non-responders' following a toxic challenge as has been shown for the rodent bone marrow micronucleus test (23,24).
In conclusion, results presented here show that MNU is a suitable positive control for use in the rat comet assay with practical advantages over both EMS and 2-AAF.There are insufficient data to make any firm conclusion on the best estimate of cytotoxicity, but it would seem sensible to sample target tissues for histopathological or haematological examination if evidence of DNA damage is found.Statistical analyses have established the contribution of various factors to the overall experimental variability using the Wistar Han rat in this laboratory, but it cannot be assumed that they will be the same for other strains of rat, other species or in different laboratories.

Fig. 1 .Fig. 2 .
Fig. 1.Power analysis of stomach and blood to compare scoring 50 nuclei (solid points) or 100 nuclei (open points) per gel to determine a 2.5-fold change when preparing three gels per tissue.

Fig. 3 .
Fig. 3. Power analysis of stomach, liver, blood and bone marrow to compare the number of animals required per group when determining the significance of a (a) 2-fold, (b) 2.5-fold and (c) 3-fold change in TI with 80% probability (dotted line).

Table I .
. DNA damage determined by the comet assay in various male rat tissues, 3 and 24 h after a single oral dose of 2-AAF and EMS SD, standard deviation; Sig., significance; FC, fold change above background; %I, % increase in TI above background; NS, non-significant; *P , 0.05; **P , 0.01; ***P , 0.001.

Table II .
Percentage of NDCN in male rats 3 and 24 h after a single oral dose of 2-AAF and EMS

Table III .
DNA damage determined by the comet assay in male rats 3 and 24 h after a single oral dose of MNU C. C. Smith et al.

Table IV .
Cytotoxicity results 3 and 24 h after a single oral dose of MNU

Table V .
DNA damage determined by the comet assay in male rats 7 days after a single oral dose of MNU

Table VI .
Study design from power analsyes showing the number of rats per treatment group required to show 80% confidence with different fold changes and three or five gels prepared per tissue