Human beings are continuously exposed to a variety of harmful (e.g. genotoxic) and beneficial (e.g. antioxidant) chemicals. Lifestyle may render individuals susceptible to cancer because of hazardous exposures (e.g. smoking and recreational sun tanning) or insufficient intake of cancer preventive compounds (e.g. fruits and vegetables). Environmentally low-exposure situations are common, and although the risk for the individual is low, health effects on population basis can be large because of the high number of exposed individuals. There is a need for validated assays detecting genotoxic end-points in risk assessment. The alkaline single cell gel electrophoresis (comet) assay is widely used nowadays for detection of genotoxicity, and may be useful in risk assessment after sufficient validation. The aim of this review is to evaluate the strength of the comet assay at its present stage of validation in biomonitoring.
Abstract: Generation of DNA damage is considered to be an important initial event in carcinogenesis. The single cell gel electrophoresis (comet) assay is a technically simple and fast method that detects genotoxicity in virtually any mammalian cell type without requirement for cell culture. This review discusses the strength of the comet assay in biomonitoring at its present state of validation. The simple version of the alkaline comet assay detects DNA migration caused by strand breaks, alkaline labile sites, and transient repair sites. By incubation with bacterial glycosylase/endonuclease enzymes, broad classes of oxidative DNA damage, alkylations, and ultraviolet light-induced photoproducts are detected as additional DNA migration. The most widely measured enzyme sensitive sites have been those detected by formamidopyrimidine DNA glycosylase (FPG) and endonuclease III (ENDOIII). Reports from biomonitoring studies show that the basal level of DNA damage in leukocytes is influenced be a variety of lifestyle and environmental exposures, including exercise, air pollution, sunlight, and diet. Although not all types of carcinogenic exposures should be expected to damage DNA in leukocytes, the comet assay is a valuable method for detection of genotoxic exposure in humans. However, the predictive value of the comet assay is unknown because it has not been investigated in prospective cohort studies. Also, it is important that the performance of the assay is investigated in multi-laboratory validation trials. As a tool in risk assessment the comet assay can be used in characterization of hazards.
The comet assay as biomarker of exposure
Biomarkers are measurements that more or less specifically quantitate exposure, early biological effect, and susceptibility. Exposure biomarkers encompass: (i) chemicals or metabolites thereof, (ii) protein and DNA adducts, including types of DNA lesions detected by the comet assay. In relation to cancer, analysis of genotoxicity is distinguishable from other exposure biomarkers because it is a measurement of the biologically effective dose at the presumed target molecule. The comet assay is an exposure biomarker assay providing information of the biologically effective dose. Biomarkers of early biological effect e.g. chromosome aberrations and micronucleus frequency detect a stage of carcinogenesis, which is temporally later than the stage detected by exposure biomarkers. Examples of susceptibility biomarkers are gene polymorphisms or enzyme activity of components in metabolism and DNA repair.
Biomarkers must be rigorously validated before they are applicable as tools in risk assessment of diseases caused by environmental agents or other exposures. Rothman et al. (1995) have outlined the validation process of biomarkers in a systematic approach that involves four phases. In the first phase (laboratory studies) exposure-effect relationships are studied in cell cultures and animal experiments. The second step of biomarker validation is biomonitoring studies (transitional studies), designed to characterize the biomarker in normal human populations and evaluate exposure-effect relationships in applied studies of selected populations. The third step in the biomarker validation is crucial for the wide application of the biomarker because it is designed to show association between the biomarker and the disease (aetiologic studies), and it is the incidence of the disease rather than the biomarker that is the dependent variable in case-control and prospective cohort studies. The case-control studies are easier to perform, but suffer from the drawback of reverse causality (i.e. the biomarker may be an effect of disease rather than predicting development of disease). Prospective cohort studies are best suited for the assessment of whether a biomarker is predicting a future outcome of disease. As the last step of validation (public health application) biomarkers are validated as risk assessment tools.
Although many laboratory tests for genetic damage have been developed over the years, remarkably few of these have been validated as biomarkers. Validation of biomarkers is a lengthy process that takes years of laboratory work and inter-laboratory collaboration. Very few biomarkers related to research on cancer have been validated in prospective cohort studies. Probably the finest example of validation is the association between high chromosome aberration frequency and cancer incidence (Hagmar et al. 1998). Validation of biomarkers is required for genetic toxicology tests to be useful in public health assessment and for regulative purposes.
Development of the comet assay and novel modifications
Important historic events of the development and novel modifications of the comet assay are outlined in table 1. It is widely acknowledged that the comet assay was described as a new method for detection of DNA damage in the late 1980s, although detection of DNA damage in gel-embedded cells had been described before that time (Møller 2005). Two different versions of the comet assay were described by Singh et al. (1988) and Olive (1989). The former version of the comet assay has been adopted by most laboratories as illustrated by the number of citations of the original publications (1650 versus 44 citations, Web of Science, October 2005).
|Year||Event or achievement|
|1978||Rydberg & Johanson introduce a method for detection of strand breaks in agarose-embedded single cells under alkaline conditions (pH≥12). The amount of single relative to double stranded DNA was measured by staining with acridine orange that emits green and red light, respectively.|
|1984||Östling & Johanson describe a modified version of gel-embedded cells with electrophoresis at pH≈9.5. When cells in this microelectrophoresis assay were γ-irradiated damaged DNA stretched toward the anode while DNA with few strand breaks remained circular.|
|1988||Singh et al. introduce electroforese at pH>13. This is often regarded as that the original comet assay publication. By October 2005 the publication has been cited 1650 times (Web of Science).|
|1989||Olive describes a different version of the comet assay with electrophoresis under neutral conditions. The images are referred to as “comets”. This publication has received surprisingly few citations (44 citations by October 2005, Web of Science)|
|1991||Gedik et al. describes the enzyme-modified version of the comet assay for detection of base damage. Nucleoids are digested with bacterial enzymes that recognize broad ranges of DNA damage and enzyme sensitive sites are obtained as additional DNA migration.|
|1996||Pfuler et al. detect DNA-DNA crosslinks by the comet assay.|
|1997||Santos et al. use in situ fluorescence hybridization for localization of breaks in specific genes.|
|2000||International guidelines published for application in genetic toxicology and biomonitoring.|
|2001||Collins et al. measure DNA repair activity in cell extracts. The incision efficiency of cell extracts from donors is measured on gel-embedded substrate DNA containing specific types of DNA lesions.|
|2005||The number of publications with comet assay end-points is high (2390 publications using “comet assay” as search term in Medline database, October 2005).|
Descriptions of comet assay protocols with discussion and recommendations of various technical variations are provided elsewhere (Olive 2002; Collins 2004). The procedure of the comet assay is shown in fig. 1. In short, single cell suspensions are embedded in agarose and lysed. Whereas blood samples are single cell suspensions by nature, tissues need to be disrupted by mechanical or enzymic treatment. I prefer mechanically tissue disruption in ice-cold buffer because this procedure diminishes the likelihood of ex vivo generation or repair of DNA lesions. The lysis treatment removes cellular and nuclear membranes and proteins. This leaves the gel-embedded DNA in the form of nucleoids. Following alkaline treatment and electrophoresis, DNA migrates toward the anode in a manner that is dependent on the number of lesions in the nucleoids. The extent of migration is visualized in a fluorescence microscope after staining of the DNA. Detection of particular sites is carried out by digestion of the nucleoids with bacterial DNA repair enzymes. In the simple form of the comet assay, DNA migrates because it contains breaks that may arise from direct action of genotoxic compounds and transient repair sites. Depending on the pH of the alkaline treatment, some DNA lesions are converted to strand breaks; these additional strand breaks are commonly referred to as alkaline labile sites. There has been controversy about what is being measured by the simple alkaline comet assay. Publications commonly describe the simple alkaline comet assay as detecting strand breaks. This is obviously imprecise because the assay measures DNA migration. However, describing the end-point of the simple alkaline comet assay as DNA migration implies that enzyme-sensitive sites and other novel applications are not related to DNA migration. Mainly due to a lack of specific expression, this publication use DNA damage as a term referring to the type of genotoxicity detected by the simple alkaline comet assay.
The comet assay has been applied in a broad range of scientific fields, including genetic toxicology, ecotoxicology, DNA repair, and apoptosis. Novel applications of the comet assay include detection of DNA-DNA crosslinks, and gene-specific DNA damage detected by application of fluorescent in situ hybridization methodology in the comet assay (Møller 2005). Also, the comet assay can be modified to measure repair activity in cellular extracts as incisions of gel-embedded DNA substrates carrying oxidative DNA lesions (Collins et al. 2001). The most widely adopted novel modification has been inclusion of enzymic digestion with specific DNA glycosylase/endonuclease enzymes that detect broad classes of DNA lesions (Collins 2004). It is possible to detect oxidized pyrimidines and purines by digestion with ENDOIII and FPG, respectively. The premutagenic 7-hydro-8-oxo-2′-deoxyguanosine lesion probably is one of the most important lesions detected by the FPG protein; this lesion is also measured by high pressure liquid chromatography coupled to electrochemical detection. Fig. 2 shows the chemical structure of some of the oxidative DNA lesions detected by ENDOIII and FPG enzymes. These enzymes have by far been the most widely used enzymes. Other enzymes have been used less frequently, namely the AlkA protein for detection of alkylation damage and T4 endonuclease V for UV-induced lesions.
Types of scoring systems for comets
There are several types of scoring systems in the comet assay, including continuous (e.g. percent DNA in the tail (%T) and tail length) and categorical (visual scoring in arbitrary units) measurements, as well as various descriptions of the distribution of the images. It has been very popular to express DNA damage in terms of the tail moment, which is the tail length multiplied with %T. In theory, it should be possible to make direct comparisons of the same comet assay end-point between laboratories, but this has been much more difficult than expected. Recently, the tail length has been critised severely because it is only linear with respect to dose in a narrow interval and it is difficult to picture the appearance of the comets by the size of the value of the tail moment (Collins 2004). I believe that the end-points obtained as visual score and %T are superior at the moment because they have linear dose-response relationships with known strand breaking agents over a wide dose range and they can be compared between laboratories. In addition, there is a reasonably good linearity between visual score and %T (fig. 3).
Table 2 shows the results of a literature survey where the %T and visual score are obtained from control groups in biomonitoring studies (Møller 2006). For comparison with %T, the visual score was recalculated from the original publications to give a range of 0–100 arbitrary units. There is a very good concordance between the two ways of expressing the level of DNA migration; the mean (S.D.) is 9.5 (5.8) and 10.6 (8.5) for %T and visual score, respectively. This means that, as a rule of thumb, the migration of DNA expressed as %T or visual score in leukocytes from healthy human beings is 10% or 10 arbitrary units (assuming range 0–100 for visual score), respectively (Møller 2006).
|No of studies (subjects)||62 (3083)||63 (1988)||121 (4943C)|
|Mean (SD)||9.5 (5.8)||10.6 (8.5)||10.0 (7.2)|
|Median (25–75% quartiles)||8.8 (5.1–14.2)||7.8 (3.9–17.7)||8.6 (4.4–14.5)|
Basal level and dynamic range of DNA damage detected by the comet assay
Reporting comet assay results as visual score or %T has little value to people that are not familiar with the method. It is much more informative to express the level of genotoxicity in e.g. number of lesions per cell. Any comet assay end-points can be expressed as the number of lesions per cell by calibration with X-rays. The relationship between the yield of strand breaks in eukaryotic cell DNA per Gy has been investigated by the alkaline sucrose sedimentation technique (Kohn et al. 1976; Ahnström & Erixon 1981). The calibration requires a linear dose-response relationship between the X-ray dose and the comet assay end-point as e.g. shown in fig. 4. Although the conditions of X-ray irradiations and key analytical procedures (e.g. alkaline conditions) are not identical between the methods, it is possible to estimate the number of lesions per comet assay score.
We have found a linear dose-response relationship of DNA damage detected by the comet assay in X-ray irradiated cells over a range of 0–10 Gy (fig. 4), whereas others have reported linear dose-response relationships over a range of 0–8 Gy (Collins et al. 1996; Pouget et al. 1999). This corresponds to a dynamic range of 9280–11600 lesions/diploid cell detected by the simple comet assay (assuming that one Gy generates 1160 strand breaks per diploid cell). The basal level of DNA damage in primary human (diploid) lymphocytes has been estimated to contain 1000–1100 lesions detected by the simple alkaline comet assay, 1200–1900 ENDO sites, and 600–2000 FPG sites (Møller 2006). The dynamic range of enzyme sensitive sites depends on the concomitant level of DNA damage in the experimental setting because enzyme sensitive sites are obtained by subtraction of samples treated with and without enzyme.
The detection limit of the comet assay has been reported to be 5–10 cGy using X-ray or γ-radiation (Tice 1995); this corresponds to about 60 additional lesions in diploid cells or 1–2 extra lesions per chromosome. Using optimized assay conditions, it is possible to detect doses of ionizing radiation as low as 0.6 cGy (Malyapa et al., 1998), which correspond to less than 10 lesions per diploid genome. However, it should be recognized that the ultimate goal is not necessarily the lowest possible limit of detection, but rather assay conditions that enable the widest dose-response range in conjunction with a low limit of detection. This applies to animal experimental models where the range of doses can be wide, and biomonitoring studies with concomitant determination of DNA damage and enzyme sensitive sites.
Assay and sample scoring variation
Recently there have been various interesting attempts to investigate the reliability of the comet assay in close detail. This has been done in multi-laboratory trials as wells as in single laboratory trials. Foremost is the effort by several laboratories in Europe through the European Standards Committee on Oxidative DNA Damage (ESCODD) collaboration study to establish standardized protocols for determination of oxidative DNA damage and reach agreement on the basal level of oxidative DNA damage (ESCODD 2003). In the later phases of the ESCODD project, the FPG version of the comet assay was included for determination of oxidative DNA damage in HeLa cells incubated with Ro19–8022 that generate oxidative DNA damage without strand breaks. Coded cryopreserved samples were distributed to ESCODD members who analyzed the level of FPG sensitive sites by their own comet assay procedure; this revealed that all the laboratories were able to detect differences in coded samples on a qualitative basis, but only half of the laboratories were able to detect a dose-response relationship (ESCODD 2003). In two different studies there were 10 times difference in the level of FPG sensitive sites reported by different laboratories; determined by the coefficient of variance (CV), the variation in values obtained from different laboratories were 57 and 66% (ESCODD et al. 2003 & 2005).
In another set of investigations, researchers have determined the variance in slide scoring. In one study, 19 investigators from 7 laboratories scored the same set of slides with cells exposed to H2O2; all but one of the investigators detected a dose-response relationship (Garcia et al. 2004). However, the variation in slide scoring was large (fig. 5); CVs ranged from 10% (100 μM H2O2) to 100% (control, estimated from data in figure). The higher reproducibility in scoring damaged nucleoids is expected if it is assumed that comet-like images represent events of genotoxic insults. Assuming that the distribution of single nuclei scores per sample follows the Poisson distribution, the confidence interval will decrease as the number of events increases. Using a similar approach, we investigated the variation in slide scoring among both experienced and non-experienced investigators. It was found that both experienced and non-experienced investigators were able to distinguish between three slides of X-ray irradiated cells with clear difference in the average level of migration (Møller et al. 2004a). The mean levels of DNA migration in the three slides were 24 arbitrary units (non-irradiated), 160 arbitrary units (low dose), and 290 arbitrary units (highest dose). The corresponding CVs of slide scoring were 20% (high dose), 38% (low dose), and 93% (non-irradiated). Although the variance in slide scoring was large also in this study, it appeared to depend on experience, i.e. the variation was lowest among the most experienced investigators.
As a perspective, strikingly few publications contain data on the assay variation of the comet assay; however those that report assay variation typically obtain CVs of 20–50% in cryopreserved assay control samples (Holz et al. 1995; Hellman et al. 1999; De Boeck et al. 2000; Møller et al. 2002, 2003 & 2004b; Speit et al. 2003; Møller 2005; Avogbe et al., 2005; Lee et al. 2005). The CVs obtained in control groups of biomonitoring studies is in the range of 36% (95% confidence interval 27–46%) (Møller et al. 2000). This suggests that most laboratories have similar assay variation. Considering the large differences obtained in the slide scoring exercises, an important contribution to the assay variation could be differences in slide scoring. An assessment of the contribution of intra-assay and inter-assay variation indicated that approximately two-third of the variance was attributed to inter-assay variation (Møller et al. 2004b). These data underscore the importance of validation trials and highlights the importance of developing standards that can be used to calibrate assays.
Application in biomonitoring studies
The comet assay has been increasingly popular as genotoxicity test in biomonitoring studies of occupational and environmental exposures. The exposure levels of carcinogens that human beings encounter in these studies are lower than the doses of genotoxic compounds in animal experimental model systems. Elevated levels of DNA damage have been observed in leukocytes of persons in putatively high-exposure circumstances due to occupation or treatment with antineoplastic agents, although data from occupational studies are conflicting (Møller et al. 2000). The maximal effect ratios of DNA damage in leukocytes have differed considerably following various treatments with antineoplastic alkylating agents. Moreover, it appears that a similar range of fold-induction of DNA damage is observed by exposure to antineoplastic agents and exposure to occupational and environmental agents (Møller et al. 2000). Assuming that patients are exposed to high doses of genotoxic compounds in chemotherapy, and antineoplastic agents are highly reactive, studies of patients receiving chemotherapy should yield higher levels of DNA damage than observed in studies of occupational or environmental exposures. Since this is not the case, it suggests that the comet assay results presently is not suitable for determination of dose-response relationships in biomonitoring studies, i.e. it is not possible to produce risk estimates by the value of the DNA damage detected by the comet assay. However, it is possible that the discrepancy in fold-induction between chemotherapy and occupational exposures is due to the use of suboptimal comet assay end-points that cannot be compared. The use of either %T or visual score in arbitrary units could ease the comparison between the studies. The far-reaching implication of this lack of clear-cut dose-response relationships is that the comet assay may not be a strong tool in risk assessment, unless for the purpose for hazard characterization. In this respect, it is important to remember that the comet assay is a reliable tool for assessment of exposure. E.g., we have found dose-dependent relationships between air pollution and the level of FPG sites in mononuclear blood cells of people exposed to the relatively clean air of Copenhagen, Denmark (Vinzents et al. 2005). A more pronounced effect in FPG sites was observed among people living in Cotonou, Benin, which is highly air-polluted because of high traffic intensity of old vehicles and poor gasoline (Avogbe et al. 2005).
Age, sex, and many environmental exposures have been reported to affect the level of DNA damage detected by the comet assay (Møller et al. 2000). Environmental exposures include antioxidants, exercise, sunlight, air pollution; these exposures affect the level of DNA damage in applied studies, but they are virtually never important determinants in cross-sectional studies. Several studies have described seasonal variation of the level of DNA damage by the comet assay (Møller et al. 2002; Sørensen et al. 2003; Tsilimigaki et al. 2003). It has been argued that the seasonal variation is due to sunlight exposure, but the contribution of other seasonal changes in environmental exposures cannot be ruled out. Seasonal variation is an issue that needs to be investigated in further detail; especially the mechanism of the effect remains to be elucidated in more detail. It should be pointed out that the seasonal variation is mainly observed for DNA damage detected by the simple alkaline comet assay, whereas similar effects are less investigated for enzymic sensitive sites or non-excisting. Presently, the reports of single exposures and the effect of age and sex should be considered as potential confounding factors, but it is very important that future studies assess the magnitude of effects of these factors. An assessment of the contribution of single exposures or interaction between exposures is only possible in large investigations. Alternatively, re-analysis of published data in joined databases may serve as platform for elucidation for the effect of lifestyle factors. As exemplified by determination of DNA damage (table 2), it is possible to include data from 4943 patients in a joint database.
There are publications of many intervention studies investigating the effect of antioxidants and antioxidant-rich food products by the comet assay. It is not possible to perform a formal meta-analysis because of differences in study designs and treatment. However, a critical survey of the reports of antioxidant supplementation trials indicates that protective effects were more convincing in short-term studies (i.e. studies lasting less that 24 hr) than in the long-term studies. Unfortunately, many of the long-term studies have sequential study design that cannot control for period effects. We have devised a small scoring system for evaluation of the study design in which studies could obtain scores from zero (weak design) to three (strong design): points were given for studies with a placebo group, parallel design, and inclusion of sampling after the intervention period. This assessment showed that studies reporting oxidative DNA damage-lowering effect had poorer design than studies showing null effect, and this could not be explained by differences in the statistical power to detect difference between two groups (Møller & Loft 2002). In addition, DNA damage detected by the simple comet assay usually showed no effect of antioxidant supplementation, whereas there was a tendency that protective effects in terms of ENDOIII and FPG sensitive sites are observed following antioxidant supplementation in male subjects (Møller & Loft 2004). These antioxidant intervention studies have been conducted in healthy subjects. The conclusions cannot easily be extrapolated to the whole population because of differences in the antioxidant status, and the aging process may render subjects more susceptible to oxidative stress. It is possible that the effect of antioxidants is observed mainly in oxidatively stressed subjects.
Need for inter-laboratory validation studies
There is progress on agreement of experimental procedures in the comet assay, yet there is need for validation and standardization of the comet assay in inter-laboratory collaborations (ESCODD et al. 2005). The framework used by the multi-laboratory micronucleus (HUMN) project is an interesting set-up because the assays share the same problems and advantages such as ease of slide scoring and the feasibility of scoring large number of nuclei per sample. This worldwide collaborative study was recently started with the objectives to compare baseline values of micronuclei frequency in lymphocytes and exofoliated cells from different laboratories, establish standard protocols for the micronucleus assay, and initiate a prospective cohort study with micronucleus assay data from each laboratory (Fenech et al. 1999). There is a certain degree of subjectivity in scoring of micronuclei that may explain the large variation in the baseline levels between different laboratories over the world. In order to overcome this problem, members of the HUMN project initiated an inter-laboratory slide-scoring exercise based on detailed instructions for scoring cells and included a set of reference illustrations for different types of micronuclei. This effort indicated that all laboratories were able to detect a dose-response relationship of coded samples of cells exposed to ionizing radiation, yet with large variation in micronucleus frequency between laboratories (Fenech et al. 2003). As for the comet assay, it is only detection of FPG sensitive sites that has been investigated in a multi-laboratory validation trial, whereas many laboratories have done their own validation of other endpoints by determination of dose-response relationships of genotoxic agents and optimization of experimental procedures to the particular need of the research. Important problems needed to be solved can be summarized as follows: (i) differences in baseline levels of DNA lesions, protocols, and scoring methods, (ii) conflicting reports of the effect of age, sex, season, and smoking.
Validation status of the comet assay as a biomarker in biomonitoring
Animal experimental models are pivotal in the first phase of the validation process (laboratory studies) because the large rodent carcinogen databases provide unique opportunities to compare the performance of the biomarker with other analytical methods of genotoxicity. There has been published many studies of exposure to genotoxic compounds, but an outstanding Japanese study that investigated the generation of DNA damage by exposure to 208 rodent carcinogens and non-carcinogens in eight organs of mice and rats indicates that the comet assay is a good screen-test for in vivo genotoxicity (Sasaki et al. 2000). By far, detection of DNA damage has been the preferred end-point investigated by comet assay in animal experimental studies, although the action mechanism will not be elucidated by detection of DNA damage only. This is unfortunate because the novel applications encompassing enzymic detection of DNA damage, gene-specific lesions, and cross-links facilitate a much more detailed understanding of the mechanism of genotoxicity. There is no study like the Japanese for these novel applications of the comet assay. Relatively few laboratories have adopted the enzyme-modified version of the comet assay for detection of oxidative DNA damage, yet it has been validated in the ECSODD study where approximately half of the laboratories detected a dose-response effect. Also, validations reported by individual research groups have indicated that the enzymic detection of oxidative DNA damage is reliable on dose-response basis after exposure to ionizing radiation and Ro19-8022 (Collins et al. 1996; Pouget et al. 1999; Risom et al. 2003).
The second step of biomarker evaluation has shown that genotoxic effects can be measured by the comet assay in leukocytes and various non-lymphatic tissues of human. In addition, there is extensive knowledge of factors affecting the basal level of DNA damage in healthy human beings. The potential of the comet assay to detect DNA damage in leukocytes has been investigated in cancer patients receiving therapy with antineoplastic alkylating agents. The aggregated data from biomonitoring studies indicate that the comet assay is a useful assay for estimation of exposure. Considering that the comet assay is a reliable technique for determination of DNA damage in various animal organs, analysis of DNA damage in other cells than leukocytes is infrequent in biomonitoring studies, most likely reflecting the difficulty in obtaining human biopsy material. The number of investigations of DNA damage in non-lymphatic tissues of human beings is expanding slowly, and may form a new direction of comet assay applications in the near future.
The third step in the biomarker validation (aetiologic studies) is the application in case-control studies where the incidence of the disease rather than the biomarker is the measured parameter. The association between the effect of a biomarker and disease can be expressed as the odds ratio by this approach. Reports from case-control studies containing comet assay endpoints indicate progress in this step of the validation process. At the moment, though, the majority of the case-control studies are better characterized as applied studies. There are some case-control studies that have shown increased odds ratio for adenocarcinoma of the esophagus (Olliver et al. 2005), bladder cancer (Schabath et al. 2004), breast cancer (Smith et al. 2003), and thyroid cancer (Sigurdson et al. 2005) among patients with high level of DNA damage. However, it should be stressed that biomarker-based case-control studies are likely to be associated with reverse causality, and the results from prospective cohort studies may turn out to be less exciting. Most likely, the prospective cohort studies with biobank material do not have cryopreserved leukocytes in medium suitable for analysis of DNA damage by the comet assay. It is essential that samples are collected and stored in a way that is appropriate for later analysis by the assay. In case a multi-laboratory project is initiated, it will be possible to prospectively collect data from biomonitoring studies into a large database as is being done in the HUMN project. Probably this means that information from prospective cohort studies involving the comet assay is not an issue in the near future. It will take a number of years before the comet assay will reach the final step in the validation process (public health application), assuming that it performs well in prospective cohort studies.
The study was supported by a grant from Else og Mogens Wedell-Wedellsborgs Fond.