Prediction of 2-Year Carcinogenicity Study Results for Pharmaceutical Products: How Are We Doing
http://www.100md.com
《毒物学科学杂志》
Center for Drug Evaluation and Research, USFDA, 9201 Corporate Blvd, Rm N212, Rockville, Maryland 20850
ABSTRACT
Some have proposed that 2-year carcinogenicity studies may not be necessary if the material is a direct-acting DNA mutagen, induces liver enzymes, causes hyperplasia or toxicity in particular organs, causes cell proliferation, is cytotoxic, causes hormonal perturbations, or if one has QSAR analyses or ‘omics information. Safety pharmacology data, pharmacologic activity, metabolism data, and results of 13-week dose ranging studies (with organ weight data, clinical chemistry data, hematologic data, clinical signs and histopathologic findings) were compared with results of 2-year carcinogenicity studies reviewed by the Center for Drug Evaluation and Research (CDER)/FDA. The experience with the ICH genetic toxicology battery and alternative carcinogenicity models was also reviewed. It appears that the information available from short-term studies is not currently sufficient to accurately and reliably predict the outcome of long-term carcinogenicity studies.
Key Words: carcinogenicity; genetic toxicology; drug.
INTRODUCTION
It would be desirable to be able to reliably predict results of 2-year studies without actually conducting them. However, whether the information currently collected is sufficient to do that or if additional ancillary/supplemental studies are needed is a debated issue. To help answer this question, a retrospective look was made for all 60 systemic 2-year studies reviewed by the executive Carcinogenicity Assessment Committee (exec-CAC) of the Center for Drug Evaluation and Research (CDER)/FDA between January 2002 and December 2004. The conclusions of this committee are the final determination for drug-related effects for carcinogenic endpoints. This committee also evaluated the dose selection for the carcinogenicity studies, before the studies were conducted. Results from reviews from the original CDER/FDA primary reviewer, exec-CAC minutes, and published labeling were collated to prepare a database. Nothing was re-reviewed for this paper.
This current evaluation of correlations included genetic toxicity study results, safety pharmacology data, pharmacologic activity, metabolism data, and results of 13-week dose-ranging studies (with organ weight data, clinical chemistry data, hematologic data, clinical signs, and histopathologic findings). A separate compilation was made of the CDER/FDA experience with alternative carcinogenicity assays.
Others have suggested that the outcome of carcinogenicity studies can be predicted such that actual conduct of the studies is not necessary (Cohen, 2004; MacDonald, 2004) if one has information on whether the material is a direct-acting DNA mutagen, induces liver enzymes, causes hyperplasia or toxicity in particular organs, causes cell proliferation, is cytotoxic, causes hormonal perturbations, or if one has QSAR analyses or ‘omics information.
Most persons find the database of NCI/NTP–conducted carcinogenicity studies to be extremely useful for data analysis. The CDER/FDA database of pharmaceuticals is somewhat different from the NTP database, and the divergence has increased over the past few years. Most studies in the NTP database are in F344/N rats and B6C3F1 mice, from a source supplying the animals only to the NTP. In the CDER/FDA database, only about 13% are in F344 rats (from a non-NTP source), with 18% in Wistar rats and 68% in Sprague Dawley rats. For mice only 13% are in B6C3F1 mice, and 87% are in CD-1 Swiss mice. CDER/FDA has no control over the source or strain of animals used or the feed used. Studies may have been conducted with animals from Japan, Europe, or the United States. Some sponsors use dietary restriction, which affects the incidence of background neoplasms. In the CDER/FDA database of pharmaceuticals, most compounds that are positive in the Salmonella test and other worrisome genotoxic positive drugs may have been screened out before development. Liver and renal toxicants may be screened out by the pharmaceutical developers by the use of various biomarkers. Most liver hypertrophy for drugs appears to be due to induction of P450 enzymes. Although there was a cluster of results for 10 peroxisome proliferator-activated receptor (PPAR) agonists, three immunomodulators/ suppressants, and two nucleoside analogs submitted to CDER/FDA in the past 3 years, the rest of the pharmacologic classes were distinct. Thus, although the receptor or other agonist or antagonist or inhibitory activity was known, many of the pharmacologic consequences of the long-term administration of these drugs were not known in advance. The CDER/FDA database is gradually accumulating more results in several alternative carcinogenicity assays. Relevance of the various carcinogenicity study results in rodents to humans is extremely important when considering the risk/benefit of a pharmaceutical and is an important activity, but is not the subject of this current evaluation.
EXPERIENCE WITH TRANSGENIC CARCINOGENICITY ASSAYS
At least one quarter of mouse studies in the CDER/FDA database completed in the past 3 years are in alternative 6-month carcinogenicity mouse assays (per the International Conference on Harmonization—ICH S1B, 1997), which do not always exhibit the same breadth of possible target organ toxicity or tissue carcinogenicity effects as traditional 2-year studies. According to the ICH agreement, ICH S1B, CDER/FDA will accept an alternative carcinogenicity study in place of the traditional 2-year study. The utility of some alternative assays was reviewed by an International Life Sciences effort (MacDonald et al., 2004). The drugs positive in the ICH-agreed-to genetox battery (Table 1) have often been tested in the P53+/– assay, and results have uniformly been negative. Eight of the P53+/– negative drugs also gave negative results in the unscheduled DNA synthesis assay. Clear Ames-positive drugs and in vivo micronucleus-positive drugs gave negative results. It is not clear whether the lack of correlation is due to false positives in the in vitro genetox assays, insensitivity of the P53+/– assay, or a combination of the two. When CDER/FDA first started accepting P53+/– assays in place of 2-year mouse studies, two nongenotoxic compounds were included. Then the P53+/– assay was accepted for few compounds that were equivocally genotoxic. Subsequently, CDER/FDA has only concurred with the testing in the P53+/– assay of clearly genotoxic compounds in one or more assay in the battery.
The drugs that had no drug-related neoplasms in the P53+/– assay generally had carcinogenicity findings in traditional 2-year studies in rats (Table 2) (and in 2-year studies in mice in some cases), and the findings in rats and mice were mainly secondary to pharmacologic effects of the drugs.
Most of the transgenic results
The use of alternative carcinogenicity studies has reduced the number of traditional 2-year studies conducted for pharmaceuticals in mice by one fourth. Fewer animals are used, and development time is saved. Use of an alternative assay permitted the continued development of a drug when (1) a drug was clearly genotoxic, and results of the P53+/– assay were negative; (2) when a drug was equivocally genotoxic, and results of the P53+/– assay were negative; (3) when rat had equivocal carcinogenicity results, and results of the P53+/– assay were negative. Use of the P53+/– assay has allowed approval of a priority drug which was positive in the Syrian hamster embryo (SHE) cell assay but negative in the Tg.AC assay. It has allowed approval when the 2-year carcinogenicity study was inadequate and did not have to be repeated (for a study of inadequate duration and for a study with an inadequate high dose selection). Occasionally, it is considered more appropriate to conduct the mouse carcinogenicity assays in the traditional 2-year model. Examples would be when (1) a class effect is not expressed in rats, and results of concern were seen in mice for other members of the pharmacologic class; (2) when the pharmacologic effect is on the gall bladder; (3) when the mice but not the rats make all the human metabolites.
CDER EXPERIENCE WITH GENETIC TOXICOLOGY STUDY RESULTS AND THEIR PREDICTIVE VALUE
Of 123 drugs tested in the in vivo micronucleus tests and ultimately approved by the end of 2002, 20 drugs gave negative results in the Salmonella assay but were positive in the in vivo micronucleus test (Table 4). Although the Salmonella assay and the in vivo micronucleus test measure very different endpoints, compounds directly affecting DNA would likely affect both assays. All six of the dual Salmonella-positive and micronucleus-positive compounds were drugs that inhibited DNA synthesis. Of 123 drugs tested in the in vivo micronucleus tests on products ultimately approved, by the end of 2002, 12 drugs gave positive results in the Salmonella test but were negative in the in vivo micronucleus test. Of 123 drugs tested in the in vivo micronucleus tests on products ultimately approved, by the end of 2002, 43 drugs gave positive results in one or more of the in vitro assays but were negative in the in vivo micronucleus test.
Based on the NTP database, Zeiger (1998) suggested that the genetic toxicity tests did not complement each other, and batteries of the tests were no more predictive of carcinogenicity than a Salmonella test alone. It was also noted that positive responses in the in vitro genetox assays, other than the Salmonella test, or in vivo tests do not increase the probability that the chemical is a carcinogen. However for pharmaceuticals, many neoplastic results in 2-year rodent carcinogenicity studies may be attributed to the pharmacologic (exaggerated or not) activity of the drug, which may or may not be relevant to humans. Some examples include mammary neoplasms secondary to increased prolactin as a consequence of antagonism to dopamine; thyroid follicular cell neoplasms secondary to increased thyroid stimulating hormone (TSH); enterochromaffin neoplasms secondary to hypergastronemia ; uterine neoplasms secondary to estrogen dominance in aged female rats, as a consequence of leutinizing hormone (LH) suppression; and leiomyomas of the mesovarium in rats as a consequence of beta-2-adrenergic agonists (Table 2).
Since 23 consecutive P53+/– negative studies were
UNEXPECTED 2-YEAR RESULTS
The CDER experience has been that unexpected results were obtained along with a few expected results in 2-year study results (Table 5). Table 5 contains 11 representative drugs with differing pharmacologic activity. The summary of all 60 studies is given in the text following the table. Even when drugs are in the same pharmacologic class, the individual drugs may have differing potency and differing specificity for other receptors, resulting in a difference spectrum of neoplasms within a class.
MORE DETAILED RESULTS FOR VARIOUS TISSUES
During the past three years more negative results than previously have been
More detailed CDER/FDA results for various tissues are described below. These results include all 60 of the 2-year studies reviewed between January 2002 and December 2004.
Liver
In 13-week or 6-month studies, hypertrophy or increased liver weight, generally a consequence of cytochrome P450 induction, was seen for 18 of 60 drugs. Liver enzyme increases were also seen in 5 of these 18, lipid alteration in 2 of these, and one of these had centrilobular degeneration and single cell necrosis; one had necrosis and no other liver findings reported. Drug-related neoplastic findings after 2 years were seen for only 10 of the 60 drugs and were generally seen in only one gender of one species/strain, even when liver alterations were seen for both species and genders.
Kidney
In 13-week or 6-month studies, nephritis, nephrosis, tubular degeneration, mineralization, increased organ weight, and/or basophilia were seen for 21 of 60 drugs, mainly in rats, yet only one had drug-related renal neoplasms in 2-year carcinogenicity studies in any gender/species, and the 6 of 60 other studies with drug-related renal neoplasms gave no indication of renal effects in 3-month studies.
Mammary
In 13-week or 6-month studies, lobular hyperplasia was seen in female rats for a multiple-targeted drug (a partial dopamine 2/serotonin 1A agonist/serotonin 2 antagonist), for which increased prolactin was seen in rats; in 2-year studies the rats that had had lobular hyperplasia had mammary neoplasms, and the mice, but not the rats, that had increased prolactin had increased mammary neoplasms. Mammary neoplasms were seen for a dopamine 2/3/serotonin receptor antagonist/partial 5-hydroxytryptamine 4 agonist, with no mammary effects at 13 weeks. Two PPAR compounds, a phosphodiesterase 4 inhibitor, and an estrogen/progestin combination, all of which had no mammary effects at 13 weeks, gave mammary neoplasms in female rats at the end of 2 years.
Thyroid
In 13-week or 6-month studies, thyroid follicular-cell hypertrophy was seen in rats for a phosphodiesterase 5 inhibitor, mineralocorticoid antagonist, neurokinin receptor antagonist, xanthine oxidase inhibitor, monoamine oxidase inhibitor, nucleoside analog, and an enzyme cofactor, and increased thyroxin (T4) was seen for seven products, yet only two of these resulted in thyroid neoplasms in rats in 2-year studies. The drugs that had C-cell neoplasms in the 2-year study had no precursor lesion in the 13-week studies, and one compound with no thyroid effects at 13-weeks gave thyroid follicular-cell neoplasms at the end of 2 years.
Adrenal
In 13-week or 6-month studies, adrenal hypertrophy or increased weight was seen in rats for nine drug products, yet none of these resulted in adrenal neoplasms in the same gender of rats in 2-year studies. Pharmacologic activity included a gamma-aminobutyric acid agonist, calcium channel blocker (two drugs), a mineralocorticoid antagonist, a partial dopamine D2 antagonist, a serotonin (5HT3) receptor antagonist, a phosphodiesterase 4 inhibitor, and a beta-adrenergic receptor blocker.
Urinary Bladder
None of the transitional cell neoplasms was predicted, but were seen for nine drugs, mostly dual PPAR-alpha, gamma agonists. Mineralization was not apparent in the short-term studies. However, urinary bladder transitional cell hyperplasia was subsequently seen in nonhuman primate studies of less than a year.
Lymph Node/Spleen
No drug-related lymphoma or leukemia was seen in 2-year studies; however splenic atrophy or extramedullary hematopoiesis and increased spleen weight were seen in 13-week studies for several drugs.
Lung
Two drugs resulted in alveolar/bronchiolar lung neoplasms in female mice, and one drug had such effects in male and female mice in 2-year studies. No lung effects were apparent in 13-week studies.
FACTORS AFFECTING CORRELATIONS/COMPARISONS BETWEEN WHAT IS KNOWN FROM 13-WEEK AND ANCILLARY STUDIES AND RESULTS OF 2-YEAR STUDIES
The high dose selected for 2-year studies of pharmaceuticals may not always be at a maximum tolerated dose (MTD); rodent plasma AUC values 25 times the human plasma AUC, saturation of absorption, or a limit dose may be used, per ICH (ICH S1C, 1995 and S1C(R), 1997). The most common basis for selection of doses for pharmaceuticals was body weight gain depression, followed by cardiac weight (for the PPAR alpha and gamma dual agonists), death, 25 times the human AUC at the recommended dose, or gastrointestinal toxicity. Other less common bases included central nervous system toxicity, saturation of exposure, limit dose, renal toxicity, liver toxicity, and hormonal perturbations. Non-MTD doses could conceivably result in negative carcinogenicity results. However, positive carcinogenicity results in rodents at greater than 25 times the human exposure, as measured by the plasma AUC of the drug and metabolites, are not generally considered problematic for a pharmaceutical that is not genotoxic.
DISCUSSION/CONCLUSIONS
It is concluded that, based on an evaluation of the CDER/FDA database, additional information and data, beyond what is already being collected, are needed to better predict or obviate the need for 2-year carcinogenicity studies.
As noted above, pharmaceuticals are not representative of all compounds. The ability to use short-term toxicity, genotoxicity, and other data to predict the outcome of long-term studies may not be as applicable to pharmaceuticals as to all compounds, since they are already a selected group of chemicals. Many of the mechanisms of carcinogenesis found with pharmaceuticals may not be amenable to early prediction at this time. Many of the mechanisms identified to date appear to be nongenotoxic and may require the prolonged treatment to be expressed. Current results available to the FDA do not support the idea that short-term studies accurately predict the neoplastic findings in long-term assays of pharmaceuticals.
Approval of new drugs is specific to the molecular entity, and formulation and carcinogenicity testing remains an important component in the assessment of the safety of drug products. General conclusions about carcinogenic potential and possibly inaccurate predictions about a particular drug product are not helpful in the approval process for a specific drug product.
NOTES
The opinions expressed are those of the author and do not necessarily reflect an official FDA opinion.
ACKNOWLEDGMENTS
Conflict of interest: none declared.
REFERENCES
Allen, D. G., Pearse, G., Haseman, J. K., and Maronpot, R. R. (2004). Prediction of rodent carcinogenesis: An evaluation of prechronic liver lesions as forecasters of liver tumors in NTP carcinogenicity studies. Toxicol. Pathol. 32, 393–401.
Cohen, S. M. (2004). Human carcinogenic risk evaluation: An alternative approach to the two-year rodent bioassay. Toxicol. Sci. 80, 225–229.
ICH S1B, International Conference on Harmonization ICH S1B. (1997). Guidance for Industry S1B Testing for Carcinogenicity of Pharmaceuticals. http://www.fda.gov/cder/guidance/index.htm.
ICH S2B, International Conference on Harmonization ICH S2B. (1997). Guidance for Industry Genotoxicity: A Standard Battery for Genotoxicity Testing of Pharmaceuticals. www.fda.gov/cder/guidance/index.htm.
ICH S1C, International Conference on Harmonization ICH S1C. (1995). Dose Selection for Carcinogenicity Studies of Pharmaceuticals. www.fda.gov/cder/guidance/index.htm.
ICH S1C(R), International Conference on Harmonization ICH S1C(R). (1997). Guidance on Dose Selection for Carcinogenicity Studies of Pharmaceuticals: Addendum on a Limit Dose and Related Notes. www.fda.gov/cder/guidance/index.htm.
MacDonald, J. S. (2004). Human carcinogenic risk evaluation, Part IV: Assessment of human risk of cancer from chemical exposure using a global weight-of-evidence approach. Toxicol. Sci. 82, 3–8.
MacDonald, J., French, J. E., Gerson, R. J., Goodman, J., Inoue, T., Jacobs, A., Kasper, P., Keller, D., Lavin, A., Long, G., et al. (2004). The utility of genetically modified mouse assays for identifying human carcinogens: A basic understanding and path forward. Toxicol. Sci. 77, 188–194.
Melnick, R. L., Kohn, M. C., Portier, and C. J. (1996). Implications for risk assessment of suggested nongenotoxic mechanisms of chemical carcinogenesis. Environ. Health Perspect. 104 (Suppl. 1), 123–134.
Zeiger, E. (1998). Identification of rodent carcinogens and noncarcinogens using genetic toxicity tests: Premises, promises, and performance. Regul. Toxicol. Pharmacol. 28, 85–95.(Abigail Jacobs)
ABSTRACT
Some have proposed that 2-year carcinogenicity studies may not be necessary if the material is a direct-acting DNA mutagen, induces liver enzymes, causes hyperplasia or toxicity in particular organs, causes cell proliferation, is cytotoxic, causes hormonal perturbations, or if one has QSAR analyses or ‘omics information. Safety pharmacology data, pharmacologic activity, metabolism data, and results of 13-week dose ranging studies (with organ weight data, clinical chemistry data, hematologic data, clinical signs and histopathologic findings) were compared with results of 2-year carcinogenicity studies reviewed by the Center for Drug Evaluation and Research (CDER)/FDA. The experience with the ICH genetic toxicology battery and alternative carcinogenicity models was also reviewed. It appears that the information available from short-term studies is not currently sufficient to accurately and reliably predict the outcome of long-term carcinogenicity studies.
Key Words: carcinogenicity; genetic toxicology; drug.
INTRODUCTION
It would be desirable to be able to reliably predict results of 2-year studies without actually conducting them. However, whether the information currently collected is sufficient to do that or if additional ancillary/supplemental studies are needed is a debated issue. To help answer this question, a retrospective look was made for all 60 systemic 2-year studies reviewed by the executive Carcinogenicity Assessment Committee (exec-CAC) of the Center for Drug Evaluation and Research (CDER)/FDA between January 2002 and December 2004. The conclusions of this committee are the final determination for drug-related effects for carcinogenic endpoints. This committee also evaluated the dose selection for the carcinogenicity studies, before the studies were conducted. Results from reviews from the original CDER/FDA primary reviewer, exec-CAC minutes, and published labeling were collated to prepare a database. Nothing was re-reviewed for this paper.
This current evaluation of correlations included genetic toxicity study results, safety pharmacology data, pharmacologic activity, metabolism data, and results of 13-week dose-ranging studies (with organ weight data, clinical chemistry data, hematologic data, clinical signs, and histopathologic findings). A separate compilation was made of the CDER/FDA experience with alternative carcinogenicity assays.
Others have suggested that the outcome of carcinogenicity studies can be predicted such that actual conduct of the studies is not necessary (Cohen, 2004; MacDonald, 2004) if one has information on whether the material is a direct-acting DNA mutagen, induces liver enzymes, causes hyperplasia or toxicity in particular organs, causes cell proliferation, is cytotoxic, causes hormonal perturbations, or if one has QSAR analyses or ‘omics information.
Most persons find the database of NCI/NTP–conducted carcinogenicity studies to be extremely useful for data analysis. The CDER/FDA database of pharmaceuticals is somewhat different from the NTP database, and the divergence has increased over the past few years. Most studies in the NTP database are in F344/N rats and B6C3F1 mice, from a source supplying the animals only to the NTP. In the CDER/FDA database, only about 13% are in F344 rats (from a non-NTP source), with 18% in Wistar rats and 68% in Sprague Dawley rats. For mice only 13% are in B6C3F1 mice, and 87% are in CD-1 Swiss mice. CDER/FDA has no control over the source or strain of animals used or the feed used. Studies may have been conducted with animals from Japan, Europe, or the United States. Some sponsors use dietary restriction, which affects the incidence of background neoplasms. In the CDER/FDA database of pharmaceuticals, most compounds that are positive in the Salmonella test and other worrisome genotoxic positive drugs may have been screened out before development. Liver and renal toxicants may be screened out by the pharmaceutical developers by the use of various biomarkers. Most liver hypertrophy for drugs appears to be due to induction of P450 enzymes. Although there was a cluster of results for 10 peroxisome proliferator-activated receptor (PPAR) agonists, three immunomodulators/ suppressants, and two nucleoside analogs submitted to CDER/FDA in the past 3 years, the rest of the pharmacologic classes were distinct. Thus, although the receptor or other agonist or antagonist or inhibitory activity was known, many of the pharmacologic consequences of the long-term administration of these drugs were not known in advance. The CDER/FDA database is gradually accumulating more results in several alternative carcinogenicity assays. Relevance of the various carcinogenicity study results in rodents to humans is extremely important when considering the risk/benefit of a pharmaceutical and is an important activity, but is not the subject of this current evaluation.
EXPERIENCE WITH TRANSGENIC CARCINOGENICITY ASSAYS
At least one quarter of mouse studies in the CDER/FDA database completed in the past 3 years are in alternative 6-month carcinogenicity mouse assays (per the International Conference on Harmonization—ICH S1B, 1997), which do not always exhibit the same breadth of possible target organ toxicity or tissue carcinogenicity effects as traditional 2-year studies. According to the ICH agreement, ICH S1B, CDER/FDA will accept an alternative carcinogenicity study in place of the traditional 2-year study. The utility of some alternative assays was reviewed by an International Life Sciences effort (MacDonald et al., 2004). The drugs positive in the ICH-agreed-to genetox battery (Table 1) have often been tested in the P53+/– assay, and results have uniformly been negative. Eight of the P53+/– negative drugs also gave negative results in the unscheduled DNA synthesis assay. Clear Ames-positive drugs and in vivo micronucleus-positive drugs gave negative results. It is not clear whether the lack of correlation is due to false positives in the in vitro genetox assays, insensitivity of the P53+/– assay, or a combination of the two. When CDER/FDA first started accepting P53+/– assays in place of 2-year mouse studies, two nongenotoxic compounds were included. Then the P53+/– assay was accepted for few compounds that were equivocally genotoxic. Subsequently, CDER/FDA has only concurred with the testing in the P53+/– assay of clearly genotoxic compounds in one or more assay in the battery.
The drugs that had no drug-related neoplasms in the P53+/– assay generally had carcinogenicity findings in traditional 2-year studies in rats (Table 2) (and in 2-year studies in mice in some cases), and the findings in rats and mice were mainly secondary to pharmacologic effects of the drugs.
Most of the transgenic results
The use of alternative carcinogenicity studies has reduced the number of traditional 2-year studies conducted for pharmaceuticals in mice by one fourth. Fewer animals are used, and development time is saved. Use of an alternative assay permitted the continued development of a drug when (1) a drug was clearly genotoxic, and results of the P53+/– assay were negative; (2) when a drug was equivocally genotoxic, and results of the P53+/– assay were negative; (3) when rat had equivocal carcinogenicity results, and results of the P53+/– assay were negative. Use of the P53+/– assay has allowed approval of a priority drug which was positive in the Syrian hamster embryo (SHE) cell assay but negative in the Tg.AC assay. It has allowed approval when the 2-year carcinogenicity study was inadequate and did not have to be repeated (for a study of inadequate duration and for a study with an inadequate high dose selection). Occasionally, it is considered more appropriate to conduct the mouse carcinogenicity assays in the traditional 2-year model. Examples would be when (1) a class effect is not expressed in rats, and results of concern were seen in mice for other members of the pharmacologic class; (2) when the pharmacologic effect is on the gall bladder; (3) when the mice but not the rats make all the human metabolites.
CDER EXPERIENCE WITH GENETIC TOXICOLOGY STUDY RESULTS AND THEIR PREDICTIVE VALUE
Of 123 drugs tested in the in vivo micronucleus tests and ultimately approved by the end of 2002, 20 drugs gave negative results in the Salmonella assay but were positive in the in vivo micronucleus test (Table 4). Although the Salmonella assay and the in vivo micronucleus test measure very different endpoints, compounds directly affecting DNA would likely affect both assays. All six of the dual Salmonella-positive and micronucleus-positive compounds were drugs that inhibited DNA synthesis. Of 123 drugs tested in the in vivo micronucleus tests on products ultimately approved, by the end of 2002, 12 drugs gave positive results in the Salmonella test but were negative in the in vivo micronucleus test. Of 123 drugs tested in the in vivo micronucleus tests on products ultimately approved, by the end of 2002, 43 drugs gave positive results in one or more of the in vitro assays but were negative in the in vivo micronucleus test.
Based on the NTP database, Zeiger (1998) suggested that the genetic toxicity tests did not complement each other, and batteries of the tests were no more predictive of carcinogenicity than a Salmonella test alone. It was also noted that positive responses in the in vitro genetox assays, other than the Salmonella test, or in vivo tests do not increase the probability that the chemical is a carcinogen. However for pharmaceuticals, many neoplastic results in 2-year rodent carcinogenicity studies may be attributed to the pharmacologic (exaggerated or not) activity of the drug, which may or may not be relevant to humans. Some examples include mammary neoplasms secondary to increased prolactin as a consequence of antagonism to dopamine; thyroid follicular cell neoplasms secondary to increased thyroid stimulating hormone (TSH); enterochromaffin neoplasms secondary to hypergastronemia ; uterine neoplasms secondary to estrogen dominance in aged female rats, as a consequence of leutinizing hormone (LH) suppression; and leiomyomas of the mesovarium in rats as a consequence of beta-2-adrenergic agonists (Table 2).
Since 23 consecutive P53+/– negative studies were
UNEXPECTED 2-YEAR RESULTS
The CDER experience has been that unexpected results were obtained along with a few expected results in 2-year study results (Table 5). Table 5 contains 11 representative drugs with differing pharmacologic activity. The summary of all 60 studies is given in the text following the table. Even when drugs are in the same pharmacologic class, the individual drugs may have differing potency and differing specificity for other receptors, resulting in a difference spectrum of neoplasms within a class.
MORE DETAILED RESULTS FOR VARIOUS TISSUES
During the past three years more negative results than previously have been
More detailed CDER/FDA results for various tissues are described below. These results include all 60 of the 2-year studies reviewed between January 2002 and December 2004.
Liver
In 13-week or 6-month studies, hypertrophy or increased liver weight, generally a consequence of cytochrome P450 induction, was seen for 18 of 60 drugs. Liver enzyme increases were also seen in 5 of these 18, lipid alteration in 2 of these, and one of these had centrilobular degeneration and single cell necrosis; one had necrosis and no other liver findings reported. Drug-related neoplastic findings after 2 years were seen for only 10 of the 60 drugs and were generally seen in only one gender of one species/strain, even when liver alterations were seen for both species and genders.
Kidney
In 13-week or 6-month studies, nephritis, nephrosis, tubular degeneration, mineralization, increased organ weight, and/or basophilia were seen for 21 of 60 drugs, mainly in rats, yet only one had drug-related renal neoplasms in 2-year carcinogenicity studies in any gender/species, and the 6 of 60 other studies with drug-related renal neoplasms gave no indication of renal effects in 3-month studies.
Mammary
In 13-week or 6-month studies, lobular hyperplasia was seen in female rats for a multiple-targeted drug (a partial dopamine 2/serotonin 1A agonist/serotonin 2 antagonist), for which increased prolactin was seen in rats; in 2-year studies the rats that had had lobular hyperplasia had mammary neoplasms, and the mice, but not the rats, that had increased prolactin had increased mammary neoplasms. Mammary neoplasms were seen for a dopamine 2/3/serotonin receptor antagonist/partial 5-hydroxytryptamine 4 agonist, with no mammary effects at 13 weeks. Two PPAR compounds, a phosphodiesterase 4 inhibitor, and an estrogen/progestin combination, all of which had no mammary effects at 13 weeks, gave mammary neoplasms in female rats at the end of 2 years.
Thyroid
In 13-week or 6-month studies, thyroid follicular-cell hypertrophy was seen in rats for a phosphodiesterase 5 inhibitor, mineralocorticoid antagonist, neurokinin receptor antagonist, xanthine oxidase inhibitor, monoamine oxidase inhibitor, nucleoside analog, and an enzyme cofactor, and increased thyroxin (T4) was seen for seven products, yet only two of these resulted in thyroid neoplasms in rats in 2-year studies. The drugs that had C-cell neoplasms in the 2-year study had no precursor lesion in the 13-week studies, and one compound with no thyroid effects at 13-weeks gave thyroid follicular-cell neoplasms at the end of 2 years.
Adrenal
In 13-week or 6-month studies, adrenal hypertrophy or increased weight was seen in rats for nine drug products, yet none of these resulted in adrenal neoplasms in the same gender of rats in 2-year studies. Pharmacologic activity included a gamma-aminobutyric acid agonist, calcium channel blocker (two drugs), a mineralocorticoid antagonist, a partial dopamine D2 antagonist, a serotonin (5HT3) receptor antagonist, a phosphodiesterase 4 inhibitor, and a beta-adrenergic receptor blocker.
Urinary Bladder
None of the transitional cell neoplasms was predicted, but were seen for nine drugs, mostly dual PPAR-alpha, gamma agonists. Mineralization was not apparent in the short-term studies. However, urinary bladder transitional cell hyperplasia was subsequently seen in nonhuman primate studies of less than a year.
Lymph Node/Spleen
No drug-related lymphoma or leukemia was seen in 2-year studies; however splenic atrophy or extramedullary hematopoiesis and increased spleen weight were seen in 13-week studies for several drugs.
Lung
Two drugs resulted in alveolar/bronchiolar lung neoplasms in female mice, and one drug had such effects in male and female mice in 2-year studies. No lung effects were apparent in 13-week studies.
FACTORS AFFECTING CORRELATIONS/COMPARISONS BETWEEN WHAT IS KNOWN FROM 13-WEEK AND ANCILLARY STUDIES AND RESULTS OF 2-YEAR STUDIES
The high dose selected for 2-year studies of pharmaceuticals may not always be at a maximum tolerated dose (MTD); rodent plasma AUC values 25 times the human plasma AUC, saturation of absorption, or a limit dose may be used, per ICH (ICH S1C, 1995 and S1C(R), 1997). The most common basis for selection of doses for pharmaceuticals was body weight gain depression, followed by cardiac weight (for the PPAR alpha and gamma dual agonists), death, 25 times the human AUC at the recommended dose, or gastrointestinal toxicity. Other less common bases included central nervous system toxicity, saturation of exposure, limit dose, renal toxicity, liver toxicity, and hormonal perturbations. Non-MTD doses could conceivably result in negative carcinogenicity results. However, positive carcinogenicity results in rodents at greater than 25 times the human exposure, as measured by the plasma AUC of the drug and metabolites, are not generally considered problematic for a pharmaceutical that is not genotoxic.
DISCUSSION/CONCLUSIONS
It is concluded that, based on an evaluation of the CDER/FDA database, additional information and data, beyond what is already being collected, are needed to better predict or obviate the need for 2-year carcinogenicity studies.
As noted above, pharmaceuticals are not representative of all compounds. The ability to use short-term toxicity, genotoxicity, and other data to predict the outcome of long-term studies may not be as applicable to pharmaceuticals as to all compounds, since they are already a selected group of chemicals. Many of the mechanisms of carcinogenesis found with pharmaceuticals may not be amenable to early prediction at this time. Many of the mechanisms identified to date appear to be nongenotoxic and may require the prolonged treatment to be expressed. Current results available to the FDA do not support the idea that short-term studies accurately predict the neoplastic findings in long-term assays of pharmaceuticals.
Approval of new drugs is specific to the molecular entity, and formulation and carcinogenicity testing remains an important component in the assessment of the safety of drug products. General conclusions about carcinogenic potential and possibly inaccurate predictions about a particular drug product are not helpful in the approval process for a specific drug product.
NOTES
The opinions expressed are those of the author and do not necessarily reflect an official FDA opinion.
ACKNOWLEDGMENTS
Conflict of interest: none declared.
REFERENCES
Allen, D. G., Pearse, G., Haseman, J. K., and Maronpot, R. R. (2004). Prediction of rodent carcinogenesis: An evaluation of prechronic liver lesions as forecasters of liver tumors in NTP carcinogenicity studies. Toxicol. Pathol. 32, 393–401.
Cohen, S. M. (2004). Human carcinogenic risk evaluation: An alternative approach to the two-year rodent bioassay. Toxicol. Sci. 80, 225–229.
ICH S1B, International Conference on Harmonization ICH S1B. (1997). Guidance for Industry S1B Testing for Carcinogenicity of Pharmaceuticals. http://www.fda.gov/cder/guidance/index.htm.
ICH S2B, International Conference on Harmonization ICH S2B. (1997). Guidance for Industry Genotoxicity: A Standard Battery for Genotoxicity Testing of Pharmaceuticals. www.fda.gov/cder/guidance/index.htm.
ICH S1C, International Conference on Harmonization ICH S1C. (1995). Dose Selection for Carcinogenicity Studies of Pharmaceuticals. www.fda.gov/cder/guidance/index.htm.
ICH S1C(R), International Conference on Harmonization ICH S1C(R). (1997). Guidance on Dose Selection for Carcinogenicity Studies of Pharmaceuticals: Addendum on a Limit Dose and Related Notes. www.fda.gov/cder/guidance/index.htm.
MacDonald, J. S. (2004). Human carcinogenic risk evaluation, Part IV: Assessment of human risk of cancer from chemical exposure using a global weight-of-evidence approach. Toxicol. Sci. 82, 3–8.
MacDonald, J., French, J. E., Gerson, R. J., Goodman, J., Inoue, T., Jacobs, A., Kasper, P., Keller, D., Lavin, A., Long, G., et al. (2004). The utility of genetically modified mouse assays for identifying human carcinogens: A basic understanding and path forward. Toxicol. Sci. 77, 188–194.
Melnick, R. L., Kohn, M. C., Portier, and C. J. (1996). Implications for risk assessment of suggested nongenotoxic mechanisms of chemical carcinogenesis. Environ. Health Perspect. 104 (Suppl. 1), 123–134.
Zeiger, E. (1998). Identification of rodent carcinogens and noncarcinogens using genetic toxicity tests: Premises, promises, and performance. Regul. Toxicol. Pharmacol. 28, 85–95.(Abigail Jacobs)