Mutagenesis Advance Access originally published online on April 6, 2006
Mutagenesis 2006 21(3):191-197; doi:10.1093/mutage/gel018
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Scoring variability of micronuclei in binucleated human lymphocytes in a case–control study
1 Department of Human Genetics, University of Ulm Germany 2 Department of Gynaecology, University of Ulm Germany
The micronucleus test in binucleated lymphocytes is a sensitive standard assay for biomonitoring, mutagenicity testing and to assess radiosensitivity of blood donors. The results vary between laboratories and scorers which led to the definition of international scoring criteria. We used these criteria in a case–control study, but nevertheless observed large differences between the seven scorers on the level of descriptive analysis. Therefore, we used the repeat measurements (267 in 98 blood donors) from this dataset (354 measurements in 185 blood donors) to analyse scoring variability in the setting of a case–control study. The variability was assessed by analysis of variance, which revealed the storage time of the blood samples, the blood donors including their disease status, and the scorers as sources of variation in the entire dataset. In addition, the coefficient of variation (CV) of the measurements was determined (overall: CV = 24.3%). After stepwise removal of biological and experimental variation by normalizations, the CV dropped to 6.8% on average, which may reflect the ‘pure counting error’. The scorer-specific CVs were between 5.5 and 9.5%. The differences between the scorers suggested by the raw data were neither related to the scorer-specific CV nor to their experience. Instead, we observed a general decline of the micronuclei frequencies towards the end of the study for all scorers. This could not be related to a change in experimental conditions or in the defined scoring criteria. An explanation could be an unintended and unrecognized change of scoring criteria. Since the change in the results did not occur in automated counting we suggest to use either reference slides in longer-lasting studies or automated counting by image analysis.
 |
Introduction |
|---|
 Top Introduction Materials and methodsResultsDiscussionReferences |
|---|
The investigation of micronuclei (MN) was proposed as an alternative and simpler approach compared with chromosome analysis of metaphase cells to assess chromosome damage in vivo (1
,2) and in vitro. The micronucleus test (MNT) in various forms has been frequently used for biomonitoring, for mutagenicity testing and also to assess the proficiency of DNA-repair. For in vitro studies, the cytokinesis-block micronucleus (CBMN) assay (3–5) has been developed. In this variant of the MNT proliferating lymphocytes appear as binucleated cells (BNCs) and provide an optimal reference for determining MN frequencies. This approach can be used for all in vitro applications of the MNT and is now the standard technique (6,7).
Based on the idea that the occurrence of some cancers may be associated with a reduced capacity of the affected person to repair DNA damage, numerous case–control studies have been carried out during recent years looking for radiosensitivity in different cancers (8), e.g. a proportion of breast cancer patients revealed a higher radiosensitivity (9,10) although the portion of radiosensitive patients varied widely in different studies using the MNT (11).
Different protocols have been used for the CBMN assay, with respect to, for example, the culture medium and duration of the culture, time and type of mutagen treatment used to induce DNA damage, and the amount of cytochalasine B (12). The overall reproducibility of the test has been reported in many studies using the coefficient of variation (CV) calculated on their overall results. However, they did not differentiate between different sources of variation like, for example, biological differences between blood donors, different blood cultures, different slides and different scorers. Although a role of the scorers for the variability of the MNT results has been noticed, only very few studies have given details on this point (12–14). To our knowledge, there are two major systematic analyses of factors contributing to the variability of the MNT (7,14). They addressed specific aspects of the problem, designed the study accordingly and used analysis of variance (ANOVA) to characterize the contribution of the factors under study. Both used small numbers of probands (one and three) and irradiation to induce a biological difference, which is rather large in consequence. This setting is quite different from, for example, biomonitoring or a case-control design, where large numbers of probands are investigated and the biological differences are expected to be small. The study by Fenech et al. (7) was carried out in the HUMN (Human Micronucleus) project and in this context a standardization of the scoring criteria was suggested (6).
The aim of the present study was to identify and evaluate factors contributing to the variability in a case–control study, carried out by several scorers during a period of 18 months. The dataset was generated in order to compare samples of sporadic breast cancer patients and controls (Varga, D, Jainta, S, Hoehne, M, et al., unpublished data) and involved seven scorers. This study comprised 446 visual single measurements in 260 blood donors and was carried out using the international scoring criteria (6,15). In order to evaluate the influence of different sources of variation, we selected the measurements from those subjects (98) from whom at least two measurements were available, giving a subset of 267 measurements with one or more parallel counts in the same blood donor by the same or different scorers.
We found that a major part of the variability can be attributed to the blood donors and technical factors (time between blood drawing and culture), whereas only a minor portion of variability is due to differences between scorers which had been identified as a major factor in the literature (7,12–14). However, we detected a general decrease of the ability of the scorers to discriminate between patients and controls over time despite a satisfying reproducibility of the counts.
|
Materials and methods |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Patients and controls
For conventional analysis we analysed a dataset of 328 measurements (counts of slides) from 120 blood donors, which had two or more measurements out of 480 visual counts in 268 donors of whom lymphocyte cultures could be set up and MN counted in a study on sporadic breast cancer and controls (Varga, D, Jainta, S, Hoehne, M, Patino-Garcia, et al., unpublished data). For multifactor ANOVA, we used a subset of 253 measurements from 91 women, excluding measurements from blood samples with a storage time >1 day. For details and the definition of subsamples see Table I. The study has been approved by the Ethics committee of the University of Ulm and from all participants informed consent had been obtained before they were included.
|
Scorers
The scorer joined the study at different time points and scorer 1 left it very early (involvement given in Figure 1). Note that three scorers (4, 5 and 6) entered the study at about trial number 50 and scorer 7 was active only in the last experiments. The seven scorers varied widely with respect to their cytogenetic experience, ranging from 3 months to 5 years and contributed between 30 and 120 counted slides. First, they all underwent training by counting 20 slides before they were included in the study. New scorers first analysed several slides together with an experienced scorer at a discussion microscope until they appeared fit to assess the cells correctly. Then they counted on their own and the counts were compared with the result of an experienced scorer. In case of major differences the slide was scored again at the discussion microscope. The counts of the trainees from the training period were not recorded. Counts obtained after the training period were equally included in the study assuming that we cannot decide which would be the true count, but cells with an equivocal assessment were still discussed together.
|
Lymphocyte cultures and slides preparation
Blood samples were stored at room temperature until starting the culture. Cell culture, irradiation and preparation of slides were performed essentially as described by Rothfuss et al. (10
). In brief, heparinized blood samples were exposed to a single dose of 2 Gy (1 min) of 137Cs gamma-rays (Gammacell 2000, Molsgaard Medical, Heorsholm, Denmark), and blood cultures were set up with chromosome medium 1 A (Gibco BRL). Forty-four hours after PHA stimulation, cytochalasine B (Sigma) was added to a final concentration of 6 µg/ml to block cytokinesis. Twenty-four hours later the cells were harvested by centrifugation, treated with 0.56% KCl solution and fixed with methanol/glacial acetic acid (5:1) mixed with an equal amount of 0.9% NaCl. Then, fixation with methanol/acetic acid (5:1) was performed three times. Air-dried slides were stained with 7% Giemsa in phosphate buffer (pH 7.0). The slides were coded and the entire analysis was carried out blindly. MN were counted in the light microscope in 500–1000 BNCs at a total magnification of x400 and the MN frequency determined per 1000 BNCs. These frequencies were used here to investigate the influence of various factors on MN count.
Scoring criteria
We used the international scoring criteria (6,12) and the scorers were first trained to score MN in BNCs of lymphocytes (20 slides).
Criteria for BNCs
- The two nuclei in the BNC should be of approximately equal size, staining pattern and staining intensity.
- The two nuclei in a BNC should have intact nuclear membranes and be situated within the same cytoplasmic boundary.
- The two nuclei within a BNC may be attached by a fine nucleoplasmic bridge which is no wider than one-third of the largest nuclear diameter.
- The two main nuclei in a BNC may touch or overlap each other especially in preparations in which the cytoplasm has been preserved. A cell with two overlapping nuclei can be scored only if the nuclear boundaries of each nucleus are distinguishable.
- The cytoplasmic boundary or membrane of a BNC should be intact and clearly distinguishable from the cytoplasmic boundary of adjacent cells.
Criteria for MN
- The diameter of MN in human lymphocytes may vary between 1/16 and 1/3 of the diameter of the main nuclei.
- MN are round or oval in shape.
- MN are non-refractile and can therefore be readily distinguished from artefacts such as staining particles.
- MN are not linked to the main nuclei but they may overlap the main nuclei.
- MN usually have the same staining intensity as the main nuclei but occasionally staining may be more intense.
Statistical analysis
Starting point for the analyses presented here were box plots based on single measurements corrected for storage time. From these distributional differences in the MN rates between the scorers in the study became obvious. Therefore, we looked for the origin of these differences using ANOVA with all the recorded information in the ANOVA models. In addition, for the same dataset the CV was determined after stepwise normalization since this CV is frequently used to characterize the data in case–control studies with the MNT.
ANOVA
Data were analysed using ANOVA [SAS statistic package (SAS Institute Inc., Cary, NC)] with the number of MN per 1000 BNC as the dependent variable. The proband's status (case/control), the storage time, the scorer (7 different persons), the interaction between scorer and status, as well as the proband herself constituted the factors in the model. They were considered to be fixed (i.e. their levels chosen in advance) with the exception of the probands who were considered to be randomly selected. The proband's age was included as a covariate. From this analysis, we obtained adjusted mean values of MN rates, estimates of the proband's variance and the residual variance (i.e. the variance not explained by model factors), and P-values for the single factors.
Modelling and model checking was done using the normal, Poisson and negative binomial probability distributions, with log and identity link functions. Goodness of fit was assessed in various subsets of the data; we found that the Poisson models were not adequate due to clear overdispersion. Negative binomial models showed about the same performance as the normal models, and we could not find clear evidence to apply a log transformation. Therefore, we report results based on the normal model without transformation of responses (using the procedure ‘MIXED’ in SAS), for ease of comparison with published studies. However, all results given have been double-checked using also the negative binomial with the log link function (using the procedure ‘GENMOD’ in SAS).
Since we initially observed a difference in the mean MN frequencies between male (mean 295; SD 63; 92 measurements) and female controls (mean 335; SD 73; 180 measurements), only women were included in the further evaluation in order not to confound status (cases/controls) with gender in the ANOVA model. From this dataset (FEMALE; see Table I), the following subsamples were formed: (i) probands with two or more MN rate measurements FEM-REP. These repeat measurements were heterogeneous: some from the same slide, from different slides from the same blood culture, some from different blood cultures from the same proband; the ‘repeat counting’ was sometimes carried out by the same scorer, sometimes by another one, in the case of multiple counts these conditions could occur also mixed. (ii) FEM-2SC: the subset of those probands with measurements from at least two different scorers. This subset was selected to evaluate differences between the different scorers. (iii) When we became aware of the fact that the results were dependent on the time within the study (represented by the trial number) we divided the samples/subsamples into the groups of EARLY—trial numbers up to 50—and LATE (trial number > 50).
Normalizations
There are differences in the MN frequencies between blood donors which in part reflect normal variation and in part systematic differences between groups of donors, e.g. cases and controls. In addition to this variation, errors introduced by blood culture, slide preparation and scoring accumulate. These errors can be dissected by stepwise normalizations and calculating the CV for each step. In addition to ANOVA we used this second approach on the same dataset because the results can, in our opinion, more easily be interpreted concerning their relevance for biological analyses. The latter procedure comprised the following steps: the CV was calculated before and after standardization of the single measurements to 1 day of storage time using the observed regression lines for the dependence of MN frequencies on storage time (days) (controls: Y = 309 + 58*X; cases: Y = 365 + 43*X;). The standardized values were used for further analyses.
Since the variability between measurements is affected also by the biological differences between different blood donors (e.g. cases and controls), these differences were removed for a better assessment of counting errors. This was achieved by adjusting the mean of each donor to the arbitrarily chosen value of 100; we will call this ‘normalization’:
 |
The CVs calculated from these normalized frequencies give a measure of the variability introduced by counting error.
|
Results |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
In the present investigation we analysed the variability of the single measurements using a dataset originally generated to investigate the difference of MN frequencies between breast cancer patients and controls. That analysis of this difference was based on mean MN frequencies from multiple measurements for the blood donors and these results will be reported elsewhere (Varga, D, Jainta, S, Hoehne, M, et al., unpublished data). Here, we report on the analysis of factors influencing the variability of single counts using the same dataset. As revealed by ANOVA the major factors influencing the MN frequency were the storage time of the blood sample, blood donors and the scorers, while the age of the blood donors did not have a general effect. When the single measurements (after standardization for storage time) were broken down to the individual scorers, major differences between them became apparent (Figure 2). As can be seen (Figure 2) the median values, the distribution of measurement as well as the difference between cases and controls varied widely.
|
ANOVA on the entire dataset (FEMALE) revealed that the factors storage time (P < 0.0001), status (P = 0.0003) and scorer (P = 0.0003) exerted a significant influence on the MN rate, while the proband's age (P = 0.40) was meaningless for MN rates after irradiation (Table II). Contrary to the impression from descriptive analysis (Figure 2) with major differences between some scorers no general interaction between status and scorer could be demonstrated (P = 0.73; Table II).
|
When the dataset was restricted to blood donors who had measurements from at least two different scorers (FEM-2SC), we observed (independently from the underlying stochastic model) that cases and controls could not be distinguished based on their mean MN rate (Table III; P = 0.98). The lack of difference between cases and controls may be in part attributed to the phase of the trial (distribution of duplicate measurements in the early and late phase; see below subset of FEM-EARLY). Scorers did not exhibit substantial discrepancies in overall mean values between this dataset and FEMALE, except for S6 who is more similar to the other scorers in FEM-EARLY (P-value for difference = 0.083). The role of probands' age was not unequivocal (P = 0.036; however, P = 0.21 under the negative binomial).
|
The distribution of individual measurements (Figure 3) suggested some dependence on the trial number, i.e. whether the respective trial had been carried out in an early or in a late phase of the study. There was a marked reduction of the observed maximal MN rates at about trial No. 50. Therefore, we performed a split analysis for trial numbers below 50 (dataset FEM-EARLY) and above this threshold. In the first period, only three scorers were involved in the evaluation while in the second period six different scorers determined MN rates. The results from this dataset (FEM-EARLY) correspond well to those from the FEMALE dataset and disease status resumes relevance as a factor (P = 0.062 and 0.028 under the two models). The difference in mean values between cases and controls here is of the same order of magnitude as in the complete dataset of females, however, based on generally higher measurement values in the early phase (Tables II and IV).
|
|
Table V shows how the total variance of MN rates was reduced by sequentially considering potentially explanatory variables in the ANOVA models. It becomes obvious that blood donors and their intrinsic properties explain more than 65% of the total variance (3885/5841), and that the blood donors' variation is reduced by more than 20% (from 3885 to 3008) when disease status is included. Altogether, the CV, calculated as the residual standard deviation divided by the mean value of all measurements, is reduced from about 23 to 12% when taking out systematic variation.
|
In addition to the ANOVA we carried out a stepwise normalization of the data and calculated a scorer-specific CV in the usual way (CV = SD/Mean). The only obvious experimental variation was due to the time between drawing the blood sample and the time at which the culture was set up (storage time). This was the factor which contributed most to the variation. Hence, MN frequencies were adjusted to day 1 using the regression line between storage time and MN frequencies. However, the standardization for storage time decreased the CV of the scorers only marginally, as can be seen comparing the CV of the raw data (mean 24.3%) with the corrected data for storage time (mean 23.0%) in repeat measurements (Table VI). In contrast, when the measurements were normalized for individual blood donors the CV of the scorers was reduced considerably (Table VI, third column). Restricting this calculation to repeat measurements from the same blood cultures reduces the CV futher (Table VI, fourth column). These CVs come close to the ‘pure’ counting errors and these values were rather low and uniform (mean CV = 6.8%).
|
At a first glance, differences between the scorers were obvious and seemed to correlate well with their experience (Figures 1 and 2). Since these differences could however not be attributed to counting precision (scorer-specific CV, Table VI) we searched for other factors explaining the differences and found a remarkable dependence of the trial number. Starting from trial number 50 the high MN frequencies disappeared (Figure 3) independently from the individual scorer. Scorers who started later had less experience and a smaller difference between cases and controls. However, the trial number related change in the results affected all scorers equally (data not shown) and, therefore, was not related to their experience. There was no change of experimental conditions (culture, medium, incubator, lot of PHA, FCS, radiation source, etc.) that could be traced back to the time of the decline in the counts. Furthermore, this decline occurred only in visual counting whereas the counts by image analysis remained on a constant level [data not shown; the automated measurements in parallel to some of the visual counts have been reported (16) together with automated counts in a later sample, which demonstrate the unchanged levels of MN rates in patients and controls].
|
Discussion |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
A considerable number of studies have been conducted to investigate a higher radiosensitivity observed in sporadic breast cancer patients compared to controls. They all agree that a proportion of the patients are indeed more sensitive using the 90th percentile or mean + 2 SD to define the normal range. The percentage of sensitive patients was highly variable and it is obvious that this variability depends predominantly on the overall variability of the MN frequencies and thus on the MN counts. We carried out a similar study and used these data to analyse the origin of the variability in more detail.
There are three studies specifically designed to investigate sources of variation in MN counts or differences between laboratories (7,13,14). They differ from our approach in following three aspects: (i) the number of probands is low (3 and 1, respectively) and the number of repeat or parallel measurements per proband is high. This approach puts weight on the variation between scorers and this was identified as a major source of variation accordingly. (ii) The studies were carried out at a single point in time, a fact, which constitutes a difference to almost all case–control studies and which precludes detection of any difference due to unintended changes of laboratory conditions or counting criteria. (iii) The three studies used irradiation in order to test how well a biological difference can be detected. This difference is, however, rather large at the doses applied, and much larger than, for example, the difference between controls and cancer cases. All effects seen in those studies appeared also in our analysis. The most important factors were as follows: storage time of the blood samples, blood donors, status and scorers whereby status corresponds to irradiation in the other studies.
The most important observation in our study was that the effect of the scorers was rather modest compared to the biological differences between the probands. Whether or not the results from a single proband change over time and whether these have an extent similar to the differences between probands (17,18) could not be assessed in our study. However, we detected also a complete and rather sharp shift in the MN rates from experiment No. 50 onwards. High MN frequencies as well as the difference between cases and controls disappeared from the results. Such a change is highly suggestive for a switch in experimental conditions, which could be excluded for all known variables and procedures. Another explanation would be an inadvertent switch in scoring criteria. This change was, however, seen not only for those scorers who entered the study about this time but also for those two scorers who worked already during the first part of the study. This explanation seems unlikely since one has to assume that both experienced scorers changed together and introduced the new scorers into the published criteria (6) at the same time. However, in favour of the second explanation is the observation that the change in the results affected only visual counting but not the results obtained by image analysis. These automated counts are always much lower than the results from visual counting but they retained the differences between cases and controls (18). Taken together, these observations may indicate that there exist subtle differences in scoring which are not covered by the present scoring criteria, but nevertheless have considerable influence on the results.
When different laboratories score slides from the same cultures prepared in the same way, there are differences between the results of the different laboratories and scorers (7,19). These were interpreted as being due to visual misclassification of the structures scored in the assay and thus to the interpretation of scoring criteria. From this type of mistakes one would expect mostly random counting errors, which appear in our data to have only a minor contribution to the variation of the counts.
Systematic differences between scorers were clearly present in our data and may have the effect to shift the whole range of counts more or less in parallel, which results in a higher or lower overall micronucleus frequency. The most prominent example from our data can be seen comparing scorer 1 and 6. There, the shift is so large that a major portion of the patients as determined by observer 6 falls into the range of measurements for controls as determined by scorer 1 (Figure 2). It is obvious that such differences between scorers obscure differences between groups (e.g. cases and controls) when combined in the same study.
Another finding was that we did not see major differences in reproducibility (CV) between the different scorers also in the late phase of the study. We found the CV to be very low (mean = 6.8%) after removal of biological and experimental variation. The CV was suggested as some kind of standard quality control for studies using the CBMN assay. In a large study including many laboratories (7) the authors found an interscorer variability 
25% (same sample, different slides), while the intrascorer variability was 11% (same slide, different areas). Similar observations to our results were reported by Hoegstedt (13) who found a CV of 6 and 11% in two scorers who had a systematic difference of 30%. Using factorial analysis, Brown et al. (14) identified blood donors and scorers as the major source of variation which agrees well with our data although the results cannot be directly compared.
Our study was carried out in a blinded fashion and the change in the results was detected after data acquisition. Under these circumstances the problem of drifting results could be prevented by the use of reference slides throughout the study. Another possibility may be to render the scoring criteria invariant which is intrinsic to image analysis using absolute measurements (16). This may allow to identify and define criteria which provide optimal discrimination between, for example, breast cancer patients and controls. These criteria are probably related to the definition of BNCs as suggest by the results shown by Varga et al. (16) and the ‘counting problem’ may be overcome by the use of image analysis, since it has been demonstrated that such systems provide highly reproducible results. In any case, the effects reported and discussed in the present study may obscure true differences by increasing variation but can hardly give raise to false positive results.
|
Acknowledgments |
|---|
We wish to thank Ingrid Peters and Karina Eiwen for excellent technical assistance. B.P.-G. was a fellow of the DAAD. This work was supported by the Deutsch Krebshilfe (Grant No. 70-2680-Vo2).
|
Notes |
|---|
*To whom correspondence should be addressed. Tel: +49 731 50023430; Fax: +49 731 50023438; Email: walther.vogel{at}uni-ulm.de
|
References |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
-
1. Heddle H.A. (1973) A rapid in vivo test for chromosome damage. Mutat. Res. 18:187–192.[Web of Science][Medline]
2. Schmid W. (1975) The micronucleus test. Mutat. Res. 31:9–15.[Web of Science][Medline]
3. Fenech M. and Morley A.A. (1985a) Solutions to the kinetic problem in the micronucleus assay. Cytobios 43:233–246.[Web of Science][Medline]
4. Fenech M. and Morley A.A. (1985b) Measurement of micronuclei in lymphocytes. Mutat. Res. 147:29–36.[CrossRef][Web of Science][Medline]
5. Fenech M. and Morley A.A. (1986) Cytokinesis-block micronucleus method in human lymphocytes: effect of in vivo ageing an low-dose x-irradiation. Mutat. Res. 161:193–198.[Web of Science][Medline]
6. Fenech M., Chang W.P., Kirsch-Volders M., Holland N., Bonassi S., Zeiger E. (2003) HUMN project: detailed description of the scoring criteria for the cytokinesis-block micronucleus assay using isolated human lymphocyte cultures. Mutat. Res. 534:65–75.[Web of Science][Medline]
7. Fenech M., Bonassi S., Turner J., et al. (2003) Intra- and inter-laboratory variation in the scoring of micronuclei and nucleoplasmic bridges in binucleated human lymphocytes. Results of an international slide-scoring exercise by the HUMN project. Mutat. Res. 534:45–64.[Web of Science][Medline]
8. Berwick M. and Vineis P. (2000) Markers of the DNA repair susceptibility to cancer in humans: an epidemiological review. J. Natl Cancer Inst. 92:874–897.
9. Scott D., Barber J.B., Levine E.L., Burrill W., Robert S.A. (1998) Radiation-induced micronucleus induction in lymphocytes identifies a high frequency of radiosensitive cases among breast cancer patients: a test for predisposition? Br. J. Cancer 77:614–620.[Web of Science][Medline]
10. Rothfuss A., Schuetz P., Bochum S., Volm T., Eberhardt E., Kreienberg R., Vogel W., Speit G. (2000) Induced micronucleus frequencies in peripheral lymphocytes as a screening test for carriers of a BRCA1 mutation in breast cancer families. Cancer Res. 60:390–394.
11. Baeyens A., Van Den Broecke R., Makar A., Thierens H., De Ridder L, Vral A. (2005) Chromosomal radiosensitivity in breast cancer patients: influence of age of onset of the disease. Oncol. Rep. 13:347–353.[Web of Science][Medline]
12. Bonassi S., Fenech M., Lando C., et al. (2001) HUman MicroNucleus project: international database comparison for results with the cytokinesis-block micronucleus assay in human lymphocytes: I. Effect of laboratory protocol, scoring criteria, and host factors on the frequency of micronuclei. Environ. Mol. Mutagen. 37:31–45.[CrossRef][Web of Science][Medline]
13. Hoegstedt B. (1984) Micronuclei in lymphocytes with preserved cytoplasm: a method for assessment of cytogenetic damage in man. Mutat. Res. 130:63–72.[CrossRef][Web of Science][Medline]
14. Brown J.K., Williams A., Withers H.R., Ow K.T., Gramacho C., Grey R., Amies C. (1997) Sources of variability in the determination of micronuclei in irradiated peripheral blood lymphocytes. Mutat. Res. 389:123–128.[Medline]
15. Fenech M. (2000) The in vitro micronucleus technique. Mutat. Res. 455:81–95.[Web of Science][Medline]
18. Varga D., Johannes T., Jainta S., Schuster S., Schwarz-Boeger U., Kiechle M., Patino-Garcia B., Vogel W. (2004) An automated scoring procedure for the micronucleus test by image analysis. Mutagenesis 19:391–397.
16. Vral A., Thierens H., Baeyens A., De Ridder L. (2002) The micronucleus and G2-phase assays for human blood lymphocytes as biomarkers of individual sensitivity to ionizing radiation: limitations imposed by intraindividual variability. Radiat. Res. 157:472–477.[CrossRef][Web of Science][Medline]
17. Vral A, Thierens H., Baeyens A., De Ridder L. (2004) Chromosomal aberrations and in vitro radiosensitivity: intra-individual versus inter-individual variability. Toxicol. Lett. 149:345–352.[Medline]
19. Fenech M. (1998) Important variables that influence base-line micronucleus frequency in cytokinesis-blocked lymphocytes—a biomarker for DNA damage in human populations. Mutat. Res. 404:155–165.[Web of Science][Medline]
Received on December 8, 2005; revised on February 28, 2006; accepted on March 13, 2006.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  J. M. Battershill, K. Burnett, and S. Bull Factors affecting the incidence of genotoxicity biomarkers in peripheral blood lymphocytes: impact on design of biomonitoring studies Mutagenesis, November 1, 2008; 23(6): 423 - 437. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Varga, J. Hoegel, C. Maier, S. Jainta, M. Hoehne, B. Patino-Garcia, I. Michel, U. Schwarz-Boeger, M. Kiechle, R. Kreienberg, et al. On the difference of micronucleus frequencies in peripheral blood lymphocytes between breast cancer patients and controls Mutagenesis, September 1, 2006; 21(5): 313 - 320. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||