A Novel Strategy to Predict Carcinogenicity of Antiparasitics Based on a Combination of DNA Lesions and Bacterial Mutagenicity Tests

Genotoxicity and carcinogenicity testing of pharmaceuticals prior to commercialization is requested by regulatory agencies. The bacterial mutagenicity test was considered having the highest accuracy of carcinogenic prediction. However, some evidences suggest that it always results in false-positive responses when the bacterial mutagenicity test is used to predict carcinogenicity. Along with major changes made to the International Committee on Harmonization guidance on genotoxicity testing [S2 (R1)], the old data (especially the cytotgenetic data) may not meet current guidelines. This review provides a compendium of retrievable results of genotoxicity and animal carcinogenicity of 136 antiparasitics. Neither genotoxicity nor carcinogenicity data is available for 84 (61.8%), while 52 (38.2%) have been evaluated in at least one genotoxicity or carcinogenicity study, and only 20 (14.7%) in both genotoxicity and carcinogenicity studies. Among 33 antiparasitics with at least one old result in in vitro genotoxicity, 15 (45.5%) are in agreement with the current ICH S2 (R1) guidance for data acceptance. Compared with other genotoxicity assays, the DNA lesions can significantly increase the accuracy of prediction of carcinogenicity. Together, a combination of DNA lesion and bacterial tests is a more accurate way to predict carcinogenicity.

animals and humans (2,3). Genetic and carcinogenic damage was found to have important health implications for the induction of diseases, such as lung cancer (4), pancreatic cancer (5), bladder cancer (6), leukemia (7)(8)(9), and non-Hodgkin's lymphoma (10). Therefore, the regulatory agencies of Europe, the USA and Japan suggested that genotoxicity and carcinogenicity studies should be conducted to learn the benefit/risk ratio before commercial approval of pharmaceuticals. It was recommended by the regulatory agencies that genotoxicity testing, which was considered to be a fundamental part of the carcinogenic risk assessment, should be performed prior to commercialization. It was forbidden to use compounds with proven genotoxic properties on humans except in rare cases with adequate justifications (11). According to the present guidelines for genotoxicity testing of pharmaceuticals (12)(13)(14)(15), a standard test battery contains: (a) a test for gene mutations in bacteria, (b) an in vitro test with cytogenetic evaluation of chromosomal damage using mammalian cells or an in vitro mouse lymphoma thymidine kinase ± gene mutation assay, and (c) an in vivo test for chromosomal damage using mammalian hematopoietic cells. These assays were considered the best approach for genotoxic hazard identification and potential carcinogenic risk prediction. However, some limitations of this standard test battery in detecting genotoxicity were found. The current revised guidelines of the Veterinary International Conference on Harmonization and ICH S2 (R1) suggested that it can detect the genetic toxicity of most substances. However, for some special chemicals such as antimicrobial, it was required to supply the bacterial assay with a validated in vitro test for gene mutation in mammalian cells to detect the genetic toxicity (12,15).
How can we identify and analyze positive genotoxicity results, especially in vitro cytogenetics? Two main factors including cytotoxicity and the highest testing concentration of the tested chemicals have very important effects on the result of genotoxicity. The Organization for Economic Cooperation and Development (OECD) had changed over the years to find the most suitable toxicity required at the highest concentration. In the 1999 revision, it was recommended that at least 50% toxicity should be induced. The ICH S2B suggested that in vitro genotoxicity tests should be conducted up to a top concentration of 10 mM in 1997 (16). In fact, when the dose level exceeds 100 µM, the physiological biological reactions will be disorder and then result in positive findings in in vitro genotoxicity tests. Moreover, a study sponsored by the European Center for the Validation of Alternative Methods indicated that the high testing dose should be reduced because the false-positive results in in vitro genotoxicity occurred at concentration levels from 1 to 10 mM. Recently, the ICH updated the genotoxicity guidelines (Table 1) (11,17). It reduced the highest dose to 1 mM and supported the in vivo genotoxicity assays.
Antiparasitics were used in the market for many years, and for a large proportion of them, genotoxicity and carcinotoxicity assays were performed prior to 1980, when the bioassays were not concordant with the present guidelines. Thus, it is necessary to re-evaluate the old data (especially the cytogenetic data) under the current guidelines of ICH S2 (R1) (17).
For pharmaceuticals, whose clinical use is continuous for at least 6 months or intermittent in chronic recurrent conditions, the long-term carcinogenicity studies in rats and mice using lifetime treatment are required (19). This has remained the most frequently chosen testing strategy since proposed by regulatory authorities in 1970s. The objective of carcinogenicity studies is to discover whether a drug has the ability to cause carcinogenicity in animals and whether this tumorigenic potential poses a relevant risk to humans (19,20). To make an evaluation of carcinogenic risks to humans, the International Agency for Research on Cancer (IARC) in the 1-101 volumes of IARC monographs was published in the years from 1972 to 2011 (21). It examined 940 drugs in various groups: the carcinogenicity studies were sufficient for 107 drugs (11.4%), limited for 59 drugs (6.3%), and inadequate for 266 drugs (28.3%); and the remaining 508 drugs (54.0%) were not classifiable in terms of their carcinogenicity to humans. However, it included only 10 antiparasitics: 2 antiparasitics (Metronidazole and Dichlorvos) were classified as possibly carcinogenic to humans (Group 2B), and 8 antiparasitics (Chloroquine, Chlordimeform, Danex, Deltamethrin, Fenvalerate, Malathion, Permethrin, and Pyrimethamine) were considered non-classifiable in terms of their carcinogenicity to humans (Group 3).
Based on the above mentioned, it is meaningful to verify the extent of antiparasitics having the available results of genotoxicity and carcinogenicity studies. It is also necessary to re-evaluate in vitro genotoxicity results according to the present revised guidance. Due to the bacterial mutagenicity test alone produced misleading positive in predicting the carcinogens, we compared the combinations of bacterial mutagenicity test and other genotoxicity assays (such as cytogeneticity in vivo and in vitro, DNA lesions and mouse bone marrow micronucleus), aiming to work out a novel strategy to predict carcinogenicity.
The 136 antiparasitics that are listed in both the human andveterinary pharmacopeia were authorized by China. Forty-three and 107 antiparasitics were obtained from the human pharmacopeia and veterinary pharmacopeia, respectively. Since some parasites, including helminths, schistosome, and tapeworm,  (27), and Chinese standard guidelines (28) Mouse lymphoma assay The mouse lymphoma assay using the thymidine kinase (Tk) gene of L5178Y Tk ± −3.7.2C mouse lymphoma cell lines was found to be the closest to the in vivo environment among the different in vitro mammalian and bacterial gene-mutation testings The MLA was performed according to FDA toxicological principles for the safety assessment of food ingredients and OECD guidelines for the testing of chemicals. IV.C.1.c Mouse Lymphoma Thymidine Kinase Gene Mutation Assay (29) and Test Guideline 490: In Vitro Mammalian Cell Gene Mutation Tests Using the Thymidine Kinase Gene (30) Chromosomal aberration assay The potential of tested compound to induce structural and numerical chromosome aberrations was evaluated in Chinese  The methodology of the major carcinogenicity and genotoxicity tests were summarized in Table 2. The collected information of genotoxicity and/or carcinogenicity of antiparasitics was obtained primarily from peer-reviewed journals (e.g., Medline, Toxline, and the Registry of Toxic Effects of Chemical Substances) (22), the US National Toxicology Program, the edition of Physician's Desk Reference (23)(24)(25), the Center for Drug Evaluation and Research of the Food and Drug Administration and some relevant websites, such as http://www.updata.usa.com, http://www.osha.gov, http://www.toxnet.nlm.nih.gov, http:// www.ntp.server.niehs.nih.gov, http://www.potency.berkeley.edu, http://www.fda.gov/cder, http://www.scirus.com, and http:// www.inchem.org. For some antiparasitics, the genotoxicity and carcinogenicity data are incomplete in terms of the absence of the dose, the indication of an exogenous metabolic system in the genotoxicity assays, and the sex in carcinogenicity assays. In such cases, we presented our data in tables as obtained in these experimental conditions except for special markings. Moreover, regarding the present guidelines, the equivocal results that we found in extensive research were marked as positive in this review.

Genotoxicity and Carcinogenicity of Antiparatics
For the present analyses, an antiparasitic was regarded as genotoxic when it produced positive or equivocal results in at least one of the standard battery tests, and as a rodent carcinogen when it increased tumor incidence. Table 3 covers the information available on genotoxicity and carcinogenicity findings for each tested antiparasitic. The following genotoxicity assays were used: Ames (bacterial mutagenesis), sex-linked recessive lethal, in vitro cytogenetics (chromosome aberrations), in vivo cytogenetics [chromosome aberrations, micronucleus and sister chromatid exchange (SCE)], unscheduled DNA synthesis in vitro (UDS), MLA (mouselymphoma L5178Y TK ± assay), and other types of genotoxicity studies, including DNA fragmentation, mammalian mutagenesis HGPRT, SCE in vitro, DNA strand break analysis in vitro, and the micronucleus assay in vitro. The long-term carcinogenicity test was carried out in mice, rats, and other species. Table 4 summarizes the total number of antiparasitics and the following are included: the number of antiparasitics with at least one genotoxicity or carcinogenicity test result and with data required by the present guidelines; the number of antiparasitics only tested for genotoxicity or carcinogenicity. It also presents the antiparasitics with results in in vitro data required by present guidelines; the number of antiparasitics that have at least one result in long-term carcinogenesis assays in rats or mice; and Salmonella typhimurium (rat, liver, S-9, aroclor 1254), TA100, TA1535, TA1537, TA98 Salmonella typhimurium (hamster, liver, S-9, aroclor 1254), TA100, TA1535, TA1537, TA98 Salmonella typhimurium (rat, liver, S-9, PCB), TA100, TA98     Antiparasitics with at least one genotoxicity or carcinogenicity tests results ( Table 3) 52 (38.2%) a Antiparasitics without retrievable genotoxicity or carcinogenicity data 84 (61.8%) Antiparasitics with all genotoxicity and carcinogenicity data required by present guidelines (   Table 3: 31) 1 (0.7%) Antiparasitics with at least one results in tests for bacterial mutagenicity ( Antiparasitics with results in in vitro data required by present guidelines (   3, 7, 10, 13, 15, 17, 19, 20, 24-26, 30, 32, 33, 35, 36, 45, 50) 19 (14.0%) Antiparasitics tested for carcinogenicity in rats ( Table 3: 2, 3, 7, 10, 13, 15, 17, 19, 20, 24-26, 30-33, 35, 36, 45, 50) 20 (14.7%) Antiparasitics tested for carcinogenicity in both mice and rats ( Table 3: 2, 3, 7, 10, 13, 15, 17, 19, 20, 24-26, 30, 32, 33, 35, 36, 45, 50) 19 (14.0%) Antiparasitics tested for carcinogenicity in other species (  19 and 20 antiparasitics were tested for carcinogenicity in mice and rats, respectively. Among the antiparasitics with both the genotoxicity and carcinogenicity data, 19 antiparasitics tested for carcinogenicity in both mice and rats and only 1 in hamsters. Table 5 provides the number of antiparasitics tested for each type of assay, including the genotoxicity and carcinogenicity studies. The results are indicated as positive, negative and discordant. When carcinogenicity testing is considered, 57.9% of antiparasitics were tested negative in mice, and 73.7% in rats. Five antiparasitics (nos. 7, 10, 26, 32, and 36) and three antiparasitics (nos. 10, 26, and 36) were carcinogenic in mice and rats, respectively. The percentage of concordant results in carcinogenicity assays between mice and rats is 85.7% (12 out of 14) and only 2 (nos. 7 and 32) antiparasitics have discordant results: no. 32 tested positive in mice and negative in rats, while no. 7 produced the opposite result. The occurrence of discordant results between mice and rats may be the differences in species (e.g., metabolic enzymes).
Ten antiparasitics were in IARC of 2B and 3 ground classifications of carcinogens: Chloroquine, Danex, and Permethrin do not have available carcinogenicity data; Deltamethrin, Fenvalerate, and Malathion tested negative in rodents while positive results were given by Chlordimeform and Metronidazole. Dichlorvos (DDVP) and Pyrimethamine have discordant results of carcinogenicity in mice and rats. To interpret the tumor findings in a carcinogenicity study and provide a perspective on the relevance of rodents to human, the mechanism and some investigations in tumor profile (trans-species, trans-sex, and multisite versus single species, single sex, and single site) were suggested by the guidelines (15).

Re-evaluation of In Vitro
Genotoxicity Results Table 6 presents the incidence of misleading positive effects in in vitro cytogenicity when using the reduction in a top dose of 1 mM. Of 33 antiparasitics with at least one result in in vitro tests for SCE, chromosomal aberrations, or micronucleus in animal or human cells, 25 (75.8%) antiparasitics had at least one retrievable dose in in vitro cytogenicity assays, while 8 (24.2%) antiparasitics had no available dose. Under the current in vitro genotoxicity testing guidelines for dose limits, 10 (nos. 10, 14, 15, 16, 20, 21, 22, 32, 36, and 47) antiparasitics were identified as genotoxins at dose levels more than 1 mM. The re-evaluation results indicated the misleading positive response in the previous reports. Fifteen (nos. 1, 2, 3, 13, 17, 18, 19, 24, 25, 26, 28, 30, 42, 50, and 52) antiparasitics had in vitro genotoxicity results consistent with ICH S2 (R1). Table 7 provides the correlation among the different types of genotoxicity assays of antiparasitics, the numbers and percentages of antiparasitics that tested concordant and discordant between    The highly consistent correlation between bacterial mutagenicity and gene mutation in mammalian cells indicated that the same genetic end point tests might have the high consistency. The discordance (nos. 32 and 45) may be due to the xenobiotic metabolism in the liver and other organs between the bacteria and animals. With the comparison between in vitro cytogenetics and in vivo cytogenetics, 2 (nos. 18 and 49) antiparasitics gave positive responses in in vitro cytogenetics while no. 13 gave negative. These results were inconsistent with that in in vivo cytogenetics. With regard to the discordant results between DNA lesions and in vitro cytogenetics of the three (nos. 3, 19 and 33) antiparasitics, two (nos. 19 and 33) antiparasitics tested negative and no. 3 yield positive in DNA lesions, respectively. These results were opposite to that in in vitro cytogenetics.

A Novel Strategy for Predicting Carcinogenicity Based on the Genotoxicity Assays
Antiparasitics with both genotoxicity and carcinogenicity data are reported in Table 8 to analyze the correlation between the results of the various types of genotoxicity and carcinogenicity. The results are marked positive or negative or inconclusive. It is obvious that the concordant and discordant results occurred in all the 15 pairs of assays considered. When carcinogenicity in mice or rats was considered, the percentage of discordant results ranged from 71.4% between in vivo cytogenetics and carcinogenicity in both mice and rats to 10.0% between bacterial mutagenicity and carcinogenicity in both mice and rats. The rank order of the consistency between genotoxicity and carcinogenicity was bacterial mutagenicity > DNA lesions > in vitro cytogenetics > gene mutation in mammalian cells > in vivo cytogenetics. Table 9 showed 2 types and 10 combinations of gene-tox assays based on bacterial mutagenicity to indicate the predictivity for rodent carcinogenicity. The sequence of the predictivity was (Ames-DNA lesions) = (Ames-DNA lesions-in vitro) = (Ames-DNA    In these comparisons, the antiparasitics gave only positive results or only negative or inconclusive results in genotoxicity assay and tested positive in at least one sex of mice or rats or gave negative or inconclusive results in both species in carcinogenicity assays. The following indicated the number and corresponding percentages, as well as the numbers of drugs of Table 3. lesions-gene mutation in mammalian cells) = (Ames-In vivo-DNA) > (Ames-in vitro) = (Ames-in vivo) > (Ames-gene mutation in mammalian cells) > (Ames-in vivo-in vitro) > (Ames-gene mutation in mammalian cells-in vivo) = (Ames-gene mutation in mammalian cells-in vitro). Table 10 presents the number and the percentage of antiparasitics that were classified as non-genotoxic non-carcinogens, genotoxic non-carcinogens, non-genotoxic carcinogens, and genotoxic carcinogens according to the genotoxicity assays considered. An antiparasitic was regarded as genotoxic when a positive response was given in at least one genotoxicity assay, and carcinogenic when it was tested positive in at least one rodent sex. Of the 20 antiparasitics with retrievable results of both genotoxicity and carcinogenicity, Malathion, Diazinon, Deltamethrin, Fenvalerate, Coumaphos, Tiabendazole, Albendazole, Cypermethrin, Amitraz and Praziquantel might be classified as genotoxic non-carcinogens; Fenthion, Lindane, Chlordimeform, Fipronil, Dichlorvos, Metronidazole, Pyrimethamine, and Imidacloprid can be classified as genotoxic carcinogens; Mefloquine was considered a nongenotoxic non-carcinogen, while the non-genotoxic carcinogens only contained Atovaquone, which tested negative in bacterial mutagenicity, in vitro and in vivo cytogenetic assays, but was found to induce liver tumors in mice in a long-term carcinogenesis assay (75, 122,123).
The bacterial mutagenicity has the highest specificity but the lowest sensitivity (Table 8), while DNA lesions (in vitro and/ or in vivo) have the highest sensitivity and a lower specificity. A test with a low specificity induced a high proportion of misleading positive results. Therefore, the combination of bacterial mutagenicity and DNA lesions has high accuracy in relation to rodent cancer, which is consistent with the above analysis results. A proportion of 5.3% of antiparasitics gave positive in bacterial mutagenicity and was classified as non-carcinogens. There were 31.6% of antiparasitics that were regarded as carcinogenic while gave a negative result in bacterial mutagenicity.

DiSCUSSiON
The economic importance of parasitic infections in livestock and humans has long been recognized. Meanwhile, the most important advances in antiparasitics have come from the animal health area. Although many antiparasitics have been developed and applied to control parasitism in humans and animals, genotoxicity and carcinogenicity studies have not been conducted on a large proportion of them. Since a relationship between exposure to genotoxic compounds and carcinogenesis has been established, genotoxicity tests have been proposed for all medicinal products for human use except for some compounds (e.g., anticancer) that can interact with DNA (11). Therefore, this review was to assess the extent of antiparasitics that have been tested for genotoxic and carcinogenic activity. In addition, the ability of various types of genotoxicity assays was summarized to discriminate rodent carcinogens, which benefit to analyze the relative predictivity of carcinogenicity in rodents and humans. Furthermore, it is necessary to re-evaluate in vitro genotoxicity according the present revised guidelines.
With regard to the genotoxicity assays, compared to the positive and discordant results, the incidence of negative responses is 61.7, 61.1, 21.2, 25.8, and 15.8% for bacterial mutagenicity, gene mutation in cultured mammalian cells, in vitro cytogenetics, in vivo cytogenetics, and DNA lesions (in vitro and in vivo), respectively. It was observed that the incidence of negative responses was higher than the positive and discordant results in bacterial mutagenicity and gene mutation in cultured mammalian cells. Kasper et al. (240) reviewed the advantages and limitations of the standard genotoxicity tests in predicting the ability and the mode of action for carcinogens, which demonstrated that a totally negative response in all the standard genotoxicity assays was sufficient to prove the non-genetic toxicity of the chemicals, while the presence of a positive response in some genotoxicity assays, particularly in Ames and in vitro genotoxicity studies, did not afford support for the genetic definition of the chemicals. There have been a number of experiences in the literature regarding the high correlation among the various types of genotoxicity assays with respect to carcinogens (241,242), which suggested that a chemical that tested positive in Salmonella tended to yield positive responses in any other in vitro genotoxicity studies, for instance, chromosome aberrations (CA), SCEs, and mutations in mouse lymphoma cells (MLA) (243) A high percentage of antiparasitics tested positive in the following assays: in vitro cytogenetics, in vivo cytogenetics, and DNA lesions (in vitro and in vivo). It is worth noting that the proportion of positive responses in in vitro cytogenetics is higher than in other types of assays. The in vitro cytogenetics seems to be more sensitive to genetic substance. However, the in vitro assays always lead to a number of false-positive results in genotoxicity and the carcinogenicity in rodents (244,245). It was learned from the literature that the massive positive results only occurred at high levels of concentration. Recent surveys for in vitro cytogenetics were taken from compilations such as that of Müller et al. (246), Kirkland and Müller (247), Müller and Kasper (248), and Hilliard et al. (249). The conclusion was that the highest testing concentrations might lead to an increase in the emergence of misleading, toxicity-related positive results. In cytotoxicity and chromosome aberrations in vitro, Galloway (250) found that the positive response in genetic toxicology was caused by the cytotoxicity rather than the true drug or DNA interactions. Parry et al. (251) examined 24 carcinogens that gave positive results in in vitro genotoxicity at 1-10 mM, yet almost half of them were not mechanistically genotoxic carcinogens or had carcinogenic effects only in excessive doses. In the present review, we re-evaluate the in vitro genotoxicty according to current ICH S2 (R1) guidance. We find that the percentage of antiparasitics in agreement with the current ICH S2 (R1) guidance for in vitro genotoxicity data acceptance was 15 (45.5%). Thus, it is essential to re-evaluate in vitro genotoxicty that conducted prior to the update guideline of ICH S2 (R1) to provide a comprehensive assessment of the genotoxic effects.
Misleading positive results were found not only in in vitro but also in in vivo genotoxic assays. Increasing experience suggested that the occurrence of a positive response in rats and mice micronucleus tests was not the consequence of intrinsic genotoxicity but drug-related disturbances in the physiology (252), such as lysosomal damage, ATP depletion or impairment of mitochondrial function and the release of DNA endonucleases. However, at the time of writing, there has still been no amendment to the guidelines requirements of in vivo genotoxicity for dose limitations and toxicity to avoid irrelevant physiological responses. Furthermore, there is no consensus as to the highest testing concentration in in vitro genotoxicity assays. The method for the detection of toxicity has greatly changed in recent years, and the limitations of dose and toxicity in genotoxicity testing in OECD and ICH should be adjusted to adapt to the new changes. The standard genotoxicity system also needs to identify the cytotoxicity and genotoxicity clearly.
There are many explanations that could account for the existence of different results in the various types of genetic tests. The differences are the following: the detection of the genetic end point; the xenobiotic metabolism between bacterial mutagenicity and mammalian cells; the effective dose between in vitro and in vivo, especially the in vivo decomposition; the relative sensitivities of various genotoxicity assays to genetic damage; the metabolic activation pathway and metabolizing enzymes among species. In vivo activity, which is designed to study the mechanisms of mutagenicity in the potential target organs of rodents, is the best method to confirm the differences in cytogenetics between in vivo and in vitro. Except for the irrelevant biological reaction at high doses, it is also accepted that the metabolic activation process and metabolites could induce genetic toxicity. Some evidence suggested that the genetic toxicity of compounds may be prototypes or metabolites. For the drugs that are theoretically nitrosatable in the presence of amine, the interaction resulted in the formation of genotoxic-carcinogenic N-nitroso compounds (253). However, the current standard of genotoxicity assays cannot distinguish whether the positive results are derived from the drugs or their metabolites directly.
In Table 7, the percentage of concordant results between bacterial mutagenicity and carcinogenicity in both mice and rats is 90.0%, which is higher than any other correlation pairs. The same conclusion was drawn by Snyder and Green (19) in a review of the genotoxicity of marketed pharmaceuticals. Data from 467 marketed drugs were collected and no combination of gene-tox assays provided a higher predictivity of rodent carcinogenesis than the bacterial mutagenicity test itself (19). In two studies conducted by Zeiger, one identified 172 chemicals that gave negative or equivocal results in 2-year rodent assays, yet 38 (22.1%) chemicals produced positive results in Salmonella (243). Another found that among 158 drugs that tested negative in carcinogenicity assays, 33 (21%) were Salmonella mutagens (254). However, a chemical that tested negative in Salmonella testing cannot be regarded as a non-carcinogenicity because the percentage of rodent carcinogens that are not mutagenic is about 50% (254). It was also reported that the predictivity for rodent carcinogenicity of bacterial mutagenicity ranged from approximately 77 to 98% (254,255). The remaining 2-23% was classified as non-carcinogen with positive result in bacterial mutagenicity, which demonstrated the flaw and insufficiency on the prediction carcinogenicity of bacterial mutagenicity.
Therefore, it requires efforts to overcome the deficiencies of bacterial mutagenicity and improve the predictivity for carcinogenicity. We try to find which genotoxicity assay(s) considered could enhance the prediction of bacterial mutagenicity to rodent carcinogenicity. Our approach has many differences and improvement compared to Snyder and Green (19), who examined only five combinations of gene-tox assays, such as Ames-in vitro cytogenetics, Ames-in vivo cytogenetics, In vitro cytogenetics-in vivo cytogenetics, MLA-in vivo cytogenetics, and MLA-in vitro cytogenetics (19). These combinations have no DNA lesions tests and no taking bacterial mutagenicity as center. A review suggested that DNA lesion alone could contribute to the prediction of carcinogenicity in mice (255). In the present article, as shown in Table 8, DNA lesion testing can significantly increase the predictivity of Ames from 90 to 100%, suggesting that the combination of DNA lesions and bacterial mutagenicity obtained higher prediction of carcinogenicity.
There are three types of DNA lesions: (a) the formation of DNA adducts; (b) DNA repair synthesis (UDS); and (c) the induction of DNA strand breaks and cross-links. An analysis of correlations between the induction of DNA lesions and carcinogenic activity was conducted in 2010 (256). It noted that the carcinogenic activity of some drugs can be correctly predicted by DNA lesion assays, yet neglected in the standard 3-test battery. Thus, DNA lesion assays were considered the best supplement for the standard 3-test battery. The occurrence of the highest predictivity in a combination of bacterial mutagenicity and DNA lesions in our review suggested a close relationship between genotoxicity and carcinogenic activity. The bacterial mutagenicity test was often used to measure the ability of a drug to cause mutations rather than a definitive test of the carcinogens. The in vivo DNA lesion tests can detect the chemicals that reach the appropriate target with an effective dose to convert into a permanent mutation by reacting with DNA. In a few cases, the mutation escaped monitoring to survive and subsequently, carcinogenicity was generated through a loss of restriction of cell division. The in vivo DNA lesions can identify this "survived mutation. " Thus, the combination of bacterial mutagenicity and DNA lesions showed a higher and more accurate predictivity of carcinogenicity.
The correlation between the results of genotoxicity and carcinogenicity assays of antiparasitics was indicated in Table 9. Among the antiparasitics that were classified as genotoxic carcinogens, 69.2% tested positive in in vitro and/or in vivo DNA lesions exhibiting a greater sensitivity to carcinogens than any other types of genotoxicity assays. Eight out of 19 (42.1%) antiparasitics gave negative results in bacterial mutagenicity and were identified as non-carcinogens. Sensitivity and specificity are commonly used to describe the capability of in vitro genotoxicity assays (257). Sensitivity is defined as the percentage of genotoxic carcinogens that produced positive results in the considered test, and specificity is regarded as the ratio of non-carcinogens that gave negative responses. The ability of a battery of three in vitro genotoxicity tests to discriminate between rodent carcinogens and non-carcinogens was made by Kirkland et al. to increase the specificity of a valid test (258). The conclusion was that the "profile" of the genotoxicity results, such as the concentration, the level of toxicity and magnitude of response, provided a body of evidence to predict the carcinogenic results (259).
The rodent bioassays were useful and relevant for predicting risks of human cancers (260). The epigenetic changes with a loss of restriction of cell division (261) and the DNA oxidative stress damage were likely to produce cancer. Trosko and Upham found that the changes in gene expression caused by cell communication systems play a key role in the imbalance of cell proliferation, differentiation, and apoptosis, eventually promoting the tumor process (262). A large number of rodent tumor findings were found not relevant for humans (262) recently. It is worth noting that traditional carcinogenicity studies are largely not predictive of human cancer risk, therefore the well-suited approaches were proposed, for instance, the genetically modified animal models (15), and in vitro carcinogenicity screening assays based on gene expression profiling (16,263). From the perspective of prospects, a more useful and accurate method to predict the carcinogenicity in humans is very urgent.
Herein, 136 antiparasitics were collected from both human and veterinary pharmacopeia. Due to the design of toxicity and the highest concentration in in vitro genetic toxicity tests have changed enormously in current guidelines, the reliably of old data were evaluated and as low as 45.5%. For a larger proportion of antiparasitics, whose genotoxicity and/or carcinogenicity results were not retrievable, the retesting based on revised guidelines should be done to make a safety assessment of human health. The combination of DNA lesions and bacterial mutagenicity is more accurate for predicting carcinogenicity than bacterial mutagenicity alone or together with any other genotoxicity testing. Development of this method for predicting carcinogens should be applied to reduce the misleading hazard alerts of the new and effective drugs.

AUTHOR CONTRiBUTiONS
ZY conceived the idea. XW analyzed and discussed data. QL analyzed and discussed data and wrote the article. ZL performed and revised the experiments. AI and FZ revised the article. All the authors discussed the results and contributed to the final manuscript.