Mutation Research 556 (2004) 25–34

Sensitivity and variability of visual scoring in the comet assayResults of an inter-laboratory scoring exercise with the use of

silver staining

Omar Garcıaa,∗, Tania Mandinaa, Ana I. Lamadrida, Adriana Diazb, Antonia Remigioc,Yanela Gonzalezd, Janet Pilotoe, Jorge E. Gonzalezf, Aimee Alvarezg

a Centro de Protecci´on e Higiene de las Radiaciones, Calle 20, No. 4113 e/ 41 y 47, Playa, CP, 11300 La Habana, Cubab Centro de Aplicaciones Tecnologicas y Desarrollo Nuclear, Cuba

c Centro para la Producci´on de Animales de Laboratorio, Cubad Centro de Investigaciones Biom´edicas, Cuba

e Centro de Investigaci´on y Desarrollo de Medicamentos, Cubaf Centro de Productos Naturales, Cuba

g Instituto de Endocrinolog´ıa, Cuba

Received 11 March 2004; received in revised form 25 June 2004; accepted 29 June 2004

Abstract

Nineteen scorers from seven Cuban laboratories participated in this slide exercise designed to test the influence of the scorero sed. Thea 5,5 comets.T portion ofD orers outo s the dosei ec d1©

K

f eth-yote

0

n the accuracy, sensitivity and variability of the comet assay when a visual method of DNA damage evaluation is ussay was performed using human lymphocytes from a single donor exposed in vitro for 5 min at 0◦C to doses of 0, 5, 10, 20, 100 and 200�M of hydrogen peroxide. Each participant scored the same set of 14 coded slides with silver stainedhe comets were classified visually into five categories according to the appearance resulting from the relative proNA in the tail. The extent of DNA damage was expressed in arbitrary units. At zero dose the median values of 12 scf 19 were included between the values of the overall 25 and 75‰. This proportion remains practically the same a

ncreases. The lowest dose detected by this method for the majority of scorers (11) was 10�M. The coefficient of variation at thontrol dose was the highest (median value 26%), progressively declined to 20%, and starting from 25�M, values are aroun0%. The results of the exercise show the reliability of the silver staining and visual scoring for the comet method2004 Elsevier B.V. All rights reserved.

eywords:Comet assay; Silver staining; Visual scoring; Sensitivity and variability

∗ Corresponding author. Tel.: +53 7 579571/531803;ax: +53 7 579573/2030165.

E-mail address:omar@cphr.edu.cu (O. Garcıa).

1. Introduction

The comet assay is one of the most popular mods to evaluate DNA damage and repair in eukar

027-5107/$ – see front matter © 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.mrfmmm.2004.06.035

26 O. Garcıa et al. / Mutation Research 556 (2004) 25–34

cells. The widespread use of this assay in genotoxicol-ogy is associated with the test’s simplicity, low costand great sensitivity among other advantages. Guide-lines concerning technical performances of this assayhave been published recently to be used as the basis forfurther developments[1,2]. Nevertheless, informationabout the reproducibility and variability of the assayis scarce, reflecting the fact that standardisation of theassay is at an early stage.

Intercomparison exercises among laboratories per-forming the comet assay are one way to provide thesedata. This type of exercise has been performed by lab-oratories using chromosomal aberrations or micronu-clei for biological dosimetry or for others purposes,as part of their validation process[3–7]. For differ-ent technical reasons, mainly associated with samplestaining, a unique set of slides have not been used onthese exercises, even though it is evident that in thisway, more precise data about the scorer’s influence onthe final result of visual scoring methods may be ob-tained. This difficulty is also present with the cometassay, on account of the use of fluorescent dyes as themost common DNA staining agent, but it may be over-come if the comets are stained using silver nitrate. Sil-ver staining of biological samples dates from the 19thcentury. In comparison with ethidium bromide the sen-sitivity of this approach for the detection of DNA hasbeen reported to be around three times greater[8]. Arecent modification of this stain allows its use in thecomet assay ([9–12]and present work). The silver staini thisp plea nalm

sureD hni-c be-t as-s sb e ise otalD il,e ratoryf ex-c ity,a usingd iatede

To perform this intercomparison exercise, the DNAdamage was evaluated using a visual classification ofcomets into categories according to their appearanceand the extent of DNA damage was expressed in arbi-trary units (AU) according to the system proposed byCollins et al.[13]. AU are correlated with the percent-age of DNA in the tail[13]. The percentage of DNA inthe tail is probably the most useful descriptor for com-parison of results within or between laboratories[14],since it gives an immediate, unambiguous and objec-tive indication of the appearance of comets[2], and isrelated to the DNA break frequency over a wide rangeof levels of DNA damage[2,15].

Considering the crucial role of the scorer in this sys-tem, the exercise was designed in order: (1) to establishthe accuracy of DNA damage measurement for eachscorer, when the same set of slides is analysed usingthe same criteria, (2) to test the influence of the scoreron the detection limit of the methods, and (3) to eval-uate the consistency of each scorer in relation to thelevel of DNA damage in the sample analyzed.

2. Material and methods

2.1. Study design

The idea of carrying out an intercomparison exer-cise on the comet assay was advertised by sending anemail to all Cuban laboratories working, or potentiallyi tions iques m-a g ins meta tingl dedi sen-t dyp thec

toryp ingles de-t witha n ofD rieso s in

s performed on dried slides, is permanent, androperty adds the possibility of analysing the samt any time after its preparation using a conventioicroscope.The great diversity of parameters used to mea

NA damage with the comet assay is another tecal problem that makes the comparison of resultsween laboratories difficult. The basic steps of theay are well defined[1,2], although small differenceetween laboratories may exist, but DNA damagxpressed variously as tail length, tail moment, tNA migration, percentage of DNA in the comet tatc. These parameters are established in each labo

or DNA staining using different fluorescent dyes,ited by different light sources, with different intensnd the comet images are captured and analysedifferent image analysis software and the assocquipment.

nterested in the comet assay. Preliminary informapecified that each laboratory should analyse a unet of coded slides, with different levels of DNA dage induced by hydrogen peroxide. Short trainincoring was offered for those interesting in the cossay, but with limited experience, in the co-ordina

aboratory. Finally, seven laboratories were inclun the exercise, with a total of 19 scorers. Repreatives from four laboratories included in this stuarticipated in the scoring training organised byo-ordinating laboratory (seeTable 1).

To start the exercise, the co-ordinating laborarepared, a package containing: (a) a box with a set of coded slides with silver stained comets, (b)ailed instructions on scoring procedure togetherdescription of the methods used for the inductioNA damage, (c) a set of photographs of five categof comets as guidance for classification of comet

O. Garcıa et al. / Mutation Research 556 (2004) 25–34 27

this exercise, (d) a standard Excel template for scor-ing in AU, (e) a schedule for forwarding the packagefrom one laboratory to the next. All the written docu-ments and photographs were also given electronically.The package circulated among laboratories accordingto the schedule and each laboratory returned their re-sults by email to the co-ordinator.

2.2. Main characteristics of laboratories andscorers

The codes of scorers and laboratories, previous ex-perience with the comet assay, and participation to thescoring training organised by the co-ordinating labo-ratory are reported inTable 1. Scorers with previousexperience in the comet assay were defined as thosewho had carried out the comet assay before this studyusing any methods for comet measurement.

2.3. Blood sampling and DNA damage induction

A blood sample was collected by a finger prick ofa single 45 years old male non-smoker. Lymphocyteswere isolated and exposed for 5 min on ice to dosesof 0, 5, 10, 25, 50, 100 and 200�M of hydrogen per-

Table 1Main characteristics of scorers participating to the inter-laboratory scoring comparison exercise

Laboratory code Total participants Scorer code Training at the co-ordinating laboratory Previous experience with Comet assay

1

2

3

4

5

67

oxide according to Collins et al.[13]. Two slides perdose and two gels per slide (i.e. four gels per dose) wereprepared as follows: after hydrogen peroxide treatmentlymphocytes were embedded in 140�l of 1% low melt-ing point agarose in phosphate buffered saline (PBS)at 37◦C. Cells were then transferred as two roughlyequal drops to microscope slides (frosted at one end),precoated with 1% normal agarose and dried. Each dropwas covered with an 18 mm× 18 mm coverslip and leftin the refrigerator for 5 min, after which the coverslipwas removed.

2.4. Comet assay

The comet assay was performed according to themethod of Singh et al.[16], as modified by Collins etal. [13] except for enzyme digestion and comet stain-ing. Basic steps of the assay were performed at 4◦C asfollow: (a) lysis: 1 h, in lysis buffer (2.5 M NaCl, 0.1 MEDTA, 10 mM Tris, 1% (v/v) Triton X-100, pH 10), (b)alkaline unwinding: 40 min in electrophoresis solution(0.3 M NaOH, 1 mM EDTA, pH 14), (c) electrophore-sis: 30 min, 300 mA, 30 V, 1 V/cm, (d) neutralization:5 min; three times in neutralising buffer (0.4 M TrispH 7.5).

3 1 Yes2 Yes18 Yes

4 3 Yes4 Yes5 Yes6 Yes

2 7 No15 No

5 8 No12 No13 No14 No16 No

3 9 Yes10 Yes11 Yes

1 17 No1 19 Yes

YesYesYes

NoNoNoNo

YesYes

YesYesYesYesYes

NoNoNo

YesNo

28 O. Garcıa et al. / Mutation Research 556 (2004) 25–34

2.5. DNA staining

After electrophoresis and neutralisation a modifiedversion of silver staining protocols published by Cerdaet al.[9] and Nadin et al.[12] was applied. Slides were(a) washed twice with deionized water, (b) placed at37◦C to dry gels (usually 1.5–2 h), (c) fixed for 10 minin fix solution (15% trichloroacetic acid, 5% zinc sul-phate heptahydrate, 5% glycerol), (d) washed twicewith deionized water, (e) dried overnight at room tem-perature, (f) re-hydrated for 5 min in deionized water,(g) placed back-to-back in a horizontal staining jar, (h)stained for 35 min in dark conditions with shaker using100 ml of freshly prepared stain solution composed by34 ml of vigorously mixed stock solution B (0.1% am-monium nitrate, 0.1% silver nitrate, 0.25% tungstosili-cic acid, 0.15% formaldehyde, v/v) and 66 ml of stocksolution A (5% sodium carbonate), (i) washed 2–3times with deionized water, (j) immersed 5 min in astop solution (acetic acid 1%), and (k) air-dried. Fi-nally each gel was protected with a 21 mm× 21 mmcoverslip to avoid damage during the intercomparisonexercise.

2.6. DNA damage evaluation

2.6.1. Number of cells scoredThe recommended number of cell to be scored per

gel was 100 (i.e. 200 per slide and 400 per dose). Nev-ertheless, scorers 12, 13, 14, 16 and 17 analysed only5 ustedt ells.

2tion

wi lvers erei ro-c lab-o lab-o datas r ofc er ofc om0 o bec r ofo sifi-

cation (0–4), and then summing the values obtained ineach gel.

2.7. Statistical analysis

The accuracy of each scorer was evaluated at eachdose by comparing the scorer median value with theoverall median per dose. The median was chosen, forthe measurement of central tendency, to reduce theinfluence of high and low values frequently reportedin this type of exercise. At each dose the 25 and 75‰from all the results were calculated and the medianvalues for each scorer were analysed in the context ofthe overall results. All individual median values fallingbetween the overall 25 and 75‰ were consideredsatisfactory[7].

The Mann–WhitneyU-test, using two-tailedP val-ues (P< 0.05) was used to establish the detection limitsfor each scorer (all the four AU values obtained at each

F withv d cellsa

0 cells per gel; in these cases the results were adjo give the arbitrary units as an average per 100 c

.6.2. Arbitrary unitsThe five categories used for comet classifica

ere those proposed by Collins[17], with minor mod-fications to take account of characteristic of the sitain (seeFig. 1). Several pictures of classes 1–3 wncluded for better classification during the scoring pess. The Excel template provided to participatingratories allowed the use of internal codes in eachratory, and was designed in order to register: (a)eparately for each gel of the slide, (b) the numbeomets classified as 0–4, and (c) the total numbells scored. Arbitrary units with possible values frto 400 were programmed into the Excel sheet t

alculated automatically by multiplying the numbebserved comets (from 0 to 100) by the comet clas

ig. 1. Images of silver stained comet of human lymphocytesarious degrees of DNA damage. Class 0 represents undamagend class 4 the most heavily damaged cells.

O. Garcıa et al. / Mutation Research 556 (2004) 25–34 29

dose versus corresponding values for the unexposedcells). The same test was used to determine the detec-tion of the increase in treatment (all the four AU valuesobtained at each dose versus corresponding values atthe previous dose).

The coefficient of variation (CV), calculated for du-plicate slides at each dose, was used to evaluate the con-sistency of the scorers at each dose used in the exercise.

Fr

3. Results

3.1. Accuracy of DNA damage measures

The median value per dose obtained by each scorerand its relation to the overall median, 25 and 75‰ areshown inFig. 2. At zero dose the median value of 12scorers out of 19 is included between the lines corre-

ig. 2. Median values per dose obtained by each scorer (squares inepresent the 25 and 75‰ and whiskers the range of values for each

side the box) in relation to the overall median per dose (solid line). Boxesscorer. Broken lines indicate overall 25 and 75‰.

30 O. Garcıa et al. / Mutation Research 556 (2004) 25–34

sponding to the overall 25 and 75‰. This proportionremains practically the same as the dose increase, with aminimum number of scorers (10) included between thecritical percentiles at 10�M and a maximum numberof scorers (14) at 50�M. Four scorers (6, 11, 18, 19) are

Ft

always between the critical percentiles and four scorers(7, 8, 13, 14) are out of the critical percentiles on five ormore occasions. Scorer 7 is always under the 25‰ andscorer 8 is always over the 75‰. Scorer 13 is on fiveoccasions over the 75‰. Scorer 14 obtained contrast-

ig. 3. Relationship between H2O2 dose and DNA damage expressed ihe mean values. The standard deviation in each point is not represen

n arbitrary units. Scorer values pooled by laboratories. Points representted due to line overlapping.

O. Garcıa et al. / Mutation Research 556 (2004) 25–34 31

Table 2Detection of the effect induced by the increase in H2 O2 dose

Scorer code Detection of statistically significant difference(P < 0.05) in comparison with the previous dose[H2O2] (�M)

5 10 25 50 100 200

1 − + + + + −2 − − + − + −3 + + + + + −4 − + − + + −5 − + + + + −6 − + + + + −7 − − + + + −8 − + − + − −9 − − + + − −10 + − + + + −11 − + + + + −12 − − + + − −13 − − − − − −14 − − − − − +a

15 − − + − + −16 − + − + − −17 + − − + − −18 − − + + + +19 − − − + − −Total + 3 8 12 15 11 2

(+) Indicate statistical significant difference (P< 0.05) in comparisonwith the previous dose.

a Lack of biological significance. The effect at 200�M is lowerthan at 100�M.

ing results, with median values over the 75‰ on threeoccasions (control and low doses), and median valuesunder the 25‰ on two occasions (high doses), with aremarkable underestimation of effect at 200�M dose.

Fig. 4. Intra scorer coefficient of variation (CV) for duplicate slide at the ts representthe median, 25 and 75‰ and range for data of all scorers.

The relationship between dose and effect obtainedby each scorer is shown inFig. 3.

3.2. Detection limit

The minimal detectable dose (P < 0.05) was 5�Mfor three scorers (3, 10 and 17). For the majority (11scorers, numbers 2, 4, 5, 6, 8, 11, 12, 15, 16, 18 and 19)the minimal detectable dose was 10�M. The 25�Mtreatment was the minimal detectable dose for fourscorers (1, 7, 9, 14), and only one scorer (13) was un-able to establish the difference between unexposed andexposed cells.

Table 2shows the capacity of each scorer to iden-tify the increase in H2O2 treatment concentration. Ascan be seen scorer 3 is able to identify all the increasesstarting from 5 up to 100�M, while scorers 1, 5, 6,and 11 have similar performance, except for the initialdose increase (0–5). The increase from 100 to 200�Mis generally not identifiable, supporting the observationabout the occurrence of assay saturation in certain con-dition [17]. In general, the best performance of scorersto identify the increase occurs at 50�M dose treatment(15 successful), with also good detection rate in the pre-ceding and subsequent dose (12 and 11 successful).

3.3. Consistency

The median of intra-individual scorer CV for dupli-cate slide for all scorers is shown inFig. 4. For non-t 6%.C d by

same dose. The results shown as points, box, and whisker plo

reated cells the median CV for all scorers was 2V values less than or equal to 26% were obtaine

32 O. Garcıa et al. / Mutation Research 556 (2004) 25–34

scorers 1, 2, 4, 6, 8, 10, 13, 14, 18 and 19, anotherthree (numbers 5, 9 and 15) show CVs in the inter-val >26–50%, and two scorers (numbers 3, 11) showCVs in the interval >50–100%. Scorers 7 (106%), 12(104%), 16 (128%) and 17 (141%) obtained the highestCVs on non-treated cells, and scorers 4 (1%), 13 (3%),2 (6%), 18 (3%) and 8 (15%), the lowest CV values.As can be seen fromFig. 4 the median CVs for 5 and10�M treatment are less than 20%. CV values around10% were obtained starting from 25�M.

Two scorers (1 and 2) analysed the slides twice, witha 10-month interval time. Comparing the results ob-tained in these two independent determinations, CVvalues less than 10% were obtained by scorer 1 at 10,50, 100, and 200 doses and by scorer 2 at 50, 100 and200 doses. CV values in the interval >10% <25% wereobtained at dose 5 by scorer 1 and at dose 25 by the twoscorers. Values of CV higher than 25% were obtainedin 4 of the 14 estimations, by the two scorers at the con-trol dose and by scorer two at the 5 and 10�M doses.The highest individual CV for these two independentestimations was at the control dose (41%, scorer 2).

4. Discussion

Sensitivity and reproducibility are critical pointsfor any biomarker of DNA damage, and even whena method is thoroughly standardised it is necessary totest the ability of laboratories to produce reproducibler

onh mes[ s ine rolei sedf d yet,t ld in-d NAi

er-c l ands asedo heira s ofD sulti par-i us is

reached, and the percentage of DNA in tail is chosenas the preferred way to present the results, the DNAdamage reported in AU may be transformed into per-centage of DNA in tail. A good correlation betweencomet results by visual classification and the percent-age of DNA in the tail is generally found[13,14,20]soit is just necessary to generate and use the appropriateequations in each laboratory. Meanwhile, in this exer-cise the assay was performed and the results expressedin the same way, so it should be possible to test theinfluence of scorer on the quality of the final results.

4.1. Accuracy of DNA damage measures perscorer

Under the conditions used for accuracy evaluationin this exercise the majority of scorers obtained sat-isfactory results. Usually in intercomparison exercisesthe “true” relationship between treatment and effect isnot known, and the proportion of “correct” outcomes,is analysed in relation to the overall median[5,7]. Ad-ditional information about the scorer performance maybe obtained analysing all the dose response curves ob-tained by the scorers.

As can be seen fromFig. 3, the shape of the doseresponse curves obtained during the exercise is similarfor the majority of scorers including scorers 7 and 8with systematically low and high scoring respectivelyin relation to the overall median, but with a good con-sistency in the scoring process and also good results int hesec rtingf ingf 14s outo cor-e n thec use r forc r re-c e.

4

eredu yseds ingt vi-

esults.In the comet assay, sensitivity depends firstly

ow the assay was performed. The use of enzy18,19], more intensive fluorescent dye, changelectrophoresis conditions, etc., play an important

n sensitivity[18] and the best parameters to be uor comet measurement have not been standardisehough it is recognised that such parameters shouicate in some way the quantity or percentage of D

n the tail[2].The protocol for silver staining applied in this ex

ise allows distinction between comet head and taiubsequently the generation of a system of AU bn a visual classification of comets according to tppearance and in particular the relative proportionNA in tail and head. The expression of the final re

n the comet assay is one of the problems in comson of results between laboratories. If a consens

he detection limit test. The general tendency on turves is a remarkable and systematic increase starom 5 to 10�M doses, with a possible plateau startrom the 100�M dose. Only two scorers, 13 andhow curves with different shape, and both weref the critical percentile in five occasions. These srs were classified as having previous experience iomet assay (seeTable 1); in both cases this previoxperience involved use of an eyepiece micrometeomet measurement, and in addition neither scoreeived training in visual scoring before the exercis

.2. Detection limit

The visual scoring of comets has been considseful for detecting large difference between analamples[21]. The detection limits established durhis exercise indicate that it is possible to detect by

O. Garcıa et al. / Mutation Research 556 (2004) 25–34 33

sual scoring doses as low as 10�M of H2O2 in theconditions used for lymphocyte treatment in this ex-ercise, with also considerable success in the detectionof the increase over a wide range of low doses. Us-ing visual classification of comets with a similar sys-tem of comet categorization, but without conversionto AU, Kobayashi et al.[20] report a significant in-crease in DNA damage in the human lymphoblastoidcell line TK6, treated for 1 h at 37◦C with 0.125�g/mlof N-methyl-N-nitro-N-nitrosoguanidine, and signifi-cant increases in DNA damage may be also deducedfrom the published data for H2O2 and methyl methane-sulfonate treatment at 0.425 and 5�g/ml, respectively.In the same work, analysing the percentage of DNAin tail and tail moment using computer-assisted systemthe detection limit was generally higher. In this paper, agood correlation between visual classification and per-centage of DNA on tail and tail moment is presented,similar to those reported on human lymphocytes[13].

4.3. Consistency and its relation to the level ofDNA damage

Consistency is one of the most critical aspects of thevisual scoring system[7] but is not often reported.

In this intercomparison exercises considerable vari-ations were found in some scorers’ estimation of AU,particularly at the background level. The cause of highCV values from scorer 7 was related to a high result inone of the gels in comparison with general low count-i thert pointi 100c cor-e t theb 4%,r ssedp rm-i catea t thec th-o ineda n-c 10%[

in-c 10%.T de-

pendent determinations done with a 10-month interval.In a comet assay study, using also AU for DNA dam-age quantification, cryopreserved lymphocytes of onedonor were tested in duplicate within a 2-week periodto calculate the intra assay variability. The coefficient ofvariation showed values of 21.3%, at the backgroundlevel, and 4.7 and 4.1%, after 10 and 50�M H2O2treatment at 4◦C for 1 h, respectively. These variationswere attributable to the handling and scoring process[23] and are in very good agreement with the generalresults obtained in our study.

5. Conclusions

In this exercise, the sensitive and variability of visualscoring combined with silver staining of comets wasevaluated. The scorers involved in the study used thesame criteria for visual scoring of comets and the sameset of slides. The majority of the scorers involved inthe exercise obtained satisfactory results. The detectionlimits established reflect the possibilities of detectingdoses as low as 10�M of H2O2 in the conditions usedin this exercise. The CV values are similar to CV val-ues obtained with other well-established visual scoringmethods. The results of the exercise confirm the relia-bility of visual scoring, and the potential of combiningit with silver staining in the comet assay. The methodis rapid, simple and may be used without fluorescencemicroscope and image analysis systems.

A

assii edd insa land,s ell-c RN6 andA

R

nn,Sin-

ng in the other three (data not shown). For anohree scorers (numbers 12, 16 and 17) a commons the scoring of only 50 cells per slide, instead ofells per slide as was recommended. Another two srs, 13 and 14, also analysed only 50 cells, and aackground level obtained CV values of 3 and 2espectively. The results of these scorers, as discureviously, differ from the rest of the scorers, confi

ng that scoring equally does not necessarily indicorrect scoring. For all scorers the median CV a

ontrol level was 26%. In other visual scoring meds the highest CV values (29 and 52%) were obtalso at the background level[7,22], and as the dose ireased the CV values tended to reduce to around7].

We found a similar trend in our study; as the dosereases CV values tend to reduce to values aroundhis is also true for the results obtained in two in

cknowledgments

The editorial assistance of A. Collins and S. Bons greatly appreciated. Part of this work was performuring O. Garcia’s stay at the laboratory of A. Collt the Rowett Research Institute, Aberdeen, Scotupported by grant no.: 064546/Z/01/Z from the Wome Trust. This work was supported by Project P34 from the Cuban Agency of Nuclear Energydvanced Technology.

eferences

[1] R.R. Tice, E. Agurell, D. Anderson, B. Burlinson, A. HartmaH. Kobayashi, Y. Miyamae, E. Rojas, J.C. Ryu, Y.F. Sasaki,

34 O. Garcıa et al. / Mutation Research 556 (2004) 25–34

gle cell gel/comet assay: guidelines for in vitro and in vivo ge-netic toxicology testing, Environ. Mol. Mut. 35 (2000) 206–221.

[2] A. Hartmann, E. Agurell, C. Beevers, S. Brendler-Schwaab, B.Burlinson, P. Clay, A. Collins, A. Smith, G. Speit, V. Thybaud,R.R. Tice, Recommendations for conducting the in vivo alkalineComet assay, Mutagenesis 18 (2003) 45–51.

[3] M. Bianchi, N.O. Bianchi, J.G. Brewen, K.E. Buckton, L.Fabry, P. Fischer, P.C. Gooch, M. Kucerova, A. Leonard, R.N.Mukherjee, U. Mukherjee, S. Nakai, A.T. Natarajan, G. Obe, F.Palitti, J. Pohl-Ruling, H.G. Schwarzacher, D. Scott, T. Sharma,E. Takahashi, C. Tanzarella, P.P.W. van Buul, Evaluation ofradiation-induced chromosomal aberrations in human periph-eral blood lymphocytes in vitro. Result of an IAEA-coordinatedprogramme, Mutat. Res. 96 (1982) 233–242.

[4] D.C. Lloyd, A.A. Edwards, A. Leonard, G.L. Deknudt, L. Ver-schaeve, A.T. Natarajan, F. Darroudi, G. Obe, F. Palitti, C. Tan-zarella, E.J. Tawn, Chromosomal aberrations in human lym-phocytes induced in vitro by very low doses of X-rays, Int. J.Radiat. Biol. 61 (1992) 335–343.

[5] O.F. Garcıa, A.T. Ramalho, M. Di Giorgio, S.S. Mir, M.E.Espinoza, J. Manzano, N. Nasazzi, I. Lopez, Intercomparisonin cytogenetic dosimetry among five laboratories from LatinAmerica, Mutat. Res. 327 (1995) 33–39.

[6] C. Lindholm, H. Romm, G. Stephan, E. Schimid, J. Moquet,A. Edwars, Intercomparison of translocation and dicentric fre-quencies between laboratories in a follow-up of the radiologicalaccident in Estonia, Int. J. Radiat. Biol. 78 (2002) 883–890.

[7] M. Fenech, S. Bonassi, J. Turner, C. Lando, M. Ceppi, W.P.Chang, N. Holland, M. Kirsch-Volders, E. Zeiger, M.P. Bi-gatti, C. Bolognesi, J. Cao, G. De Luca, M. Di Giorgio,L.R. Ferguson, A. Fucic, O. Garcia Lima, V.V. Hadjidekova,P. Hrelia, A. Jaworska, G. Joksic, A.P. Krishnaja, T.K. Lee,A. Martelli, M.J. McKay, L. Migliore, E. Mirkova, W.U.Muller, Y. Odagiri, T. Orsiere, M.R. Scarfi, M.J. Silva, T. So-funi, J. Suralles, G. Trenta, I. Vorobtsova, A. Vral, A. Zi-

ofhu-ring03)

43

etfood,

[ .N.etec-

[11] P. Reinhardt-Poulin, J.R. McLean, Y. Deslauriers, W. Gorman,S. Cabat, M. Rouabhia, The use of silver stained “comets” tovisualize DNA damage and repair in normal and xerodermapigmentosum fibroblasts, Photochem. Photobiol. 71 (2000)422–425.

[12] S.B. Nadin, L.M. Vargas-Roig, D.R. Ciocca, A silver stainingmethod for single-cell gel assay, J. Histochem. Cytochem. 49(2001) 1183–1186.

[13] A.R. Collins, A.-G. Ma, S.J. Duthie, The kinetics of repair of ox-idative DNA damage (strand breaks and oxidised pyrimidines)in human cells, Mutat. Res. (DNA Repair) 336 (1995) 69–77.

[14] A.R. Collins, M. Dusinska, M. Franklin, M. Somorovska, H.Petrovska, S. Duthie, L. Fillion, M. Panayiotidis, K. Raslova, N.Vaughan, Comet assay in human biomonitoring studies: relia-bility, validation and applications, Environ. Mol. Mut. 30 (1997)139–146.

[15] P.L. Olive, J.P. Banat, R.E. Durand, Heterogeneity in radiation-induced DNA damage and repair in tumor and normal cellsusing the “comet” assay, Radiat. Res. 122 (1990) 86–94.

[16] N.P. Singh, M.T. McCoy, R.R. Tice, E.L. Schneider, A sim-ple technique for quantitation of low levels of DNA damage inindividual cells, Exp. Cell Res. 175 (1988) 184–191.

[17] A.R. Collins, The Comet Assay, Principles, Applications, andlimitations, in Methods in Molecular Biology, in: V.V. Didenko(Ed.), In situ Detection of DNA Damage; Methods and Pro-tocols, vol. 203, Humana Press Inc., Totowa, NJ, 2002, pp.163–177.

[18] N.P. Singh, Microgels for estimation of DNA strand breaks,DNA protein crosslinks and apoptosis, Mutat. Res. 455 (2000)111–127.

[19] A.R. Collins, S.J. Duthie, V.L. Dobson, Direct enzymic detec-tion of endogenous oxidative base damage in human lympho-cytes DNA, Carcinogenesis 14 (1993) 1733–1735.

[20] H. Kobayashi, C. Sugiyama, Y. Morikawa, M. Hayashi, T. So-funi, A comparison between manual microscopic analysis and

ore-

[ pli-Biol.

[ bil-tion-

[ Re-aksumaniron.

jno, Intra- and inter-laboratory variation in the scoringmicronuclei and nucleoplasmic bridges in binucleatedman lymphocytes. Results of an international slide-scoexercise by the HUMN project, Mutat. Res. 534 (2045–64.

[8] C.R. Merril, Silver staining of proteins and DNA, Nature 3(1990) 779–780.

[9] H. Cerda, H. Delincee, H. Haine, H. Rupp, The DNA “comassay” as a rapid screening technique to control irradiatedMutat. Res. 375 (1997) 167–181.

10] N. Kizilian, R.C. Wilkins, P. Reinhardt, C. Ferrarotto, J.RMcLean, J.P. McNamee, Silver-stained comet assay for dtion of apoptosis, Biotechniques 27 (1999) 926–930.

computerized image analysis in the single cell gel electrophsis assay, MMS Commun. 3 (1995) 103–115.

21] P.L. Olive, DNA damage and repair in individual cells: apcations of the comet assay in radiobiology, Int. J. Radiat.75 (1999) 395–405.

22] K.L. Radack, S.M. Pinney, G.K. Livingston, Sources of variaity in the human lymphocyte micronucleus assay: a populabased study, Environ. Mol. Mut. 26 (1995) 26–36.

23] O. Holz, R. Jorres, A. Kastner, T. Krause, H. Magnussen,producibility of basal and induced DNA single-strand bredetected by the single-cell gel electrophoresis assay in hperipheral mononuclear leukocytes, Int. Arch. Occup. EnvHealth 67 (1995) 305–310.