Further evaluation of COMPACT, the molecular orbital approach for the prospective safety evaluation of chemicals

Mutation Research 412 Ž1998. 41–54

Further evaluation of COMPACT, the molecular orbital approach for the prospective safety evaluation of chemicals D.F.V. Lewis, C. Ioannides, D.V. Parke

)

School of Biological Sciences, UniÕersity of Surrey, Guildford, Surrey, GU2 5XH, UK Received 31 October 1996; revised 24 July 1997; accepted 24 July 1997

Abstract The molecular dimensions and electronic structures of the first group of 100 US NCIrNTP miscellaneous chemicals, evaluated for potential carcinogenicity by computer-optimized molecular parametric analysis for chemical toxicity ŽCOMPACT. have been re-determined. Using improved criteria for cytochrome P450 ŽCYP. substrate specificity, re-defined for CYP1 as having a COMPACT radius

(Ž D E y 9.5. q Ž ard y 7.8. 2

2

2

of - 6.5, and for CYP2E as having a collision

˚ or less and D E - 15.5, the likely substrates of CYP1 and CYP2E, which are regarded as potential diameter of 6.5A carcinogens, have been identified. In addition, log P values have been taken into account; those chemicals with log P - 0 are non-lipophilic substrates unlikely to reach the activating cytochrome enzymes, and have been regarded as non-carcinogens. The second group of 100 US NCIrNTP chemicals have also now been categorized by COMPACT into CYP1 and CYP2E substrates, and their potential carcinogenicities evaluated. Of the 203 chemicals in the 2 groups, those positive in the rodent two-species life-span carcinogenicity study Žrodent assay. were 53%, those positive in the Ames test Žmutagenicity. were 48%, and those positive in the COMPACT programme Žcarcinogenicity, mutagenicity, cytotoxicity. were 54%. Concordance between the COMPACT prediction of carcinogenicityrcytotoxicity and rodent two species life-span carcinogenicity data for the 203 chemicals is 69%, and correlation of COMPACT with Ames test data is 61%. The sensitivity of COMPACT for predicting rodent carcinogenicity is 72%, whereas the sensitivity of the Ames test for predicting carcinogenicity for the 203 chemicals was only 57%. The degree Žseverity. of rodent carcinogenicity also showed correlation with the COMPACT predictive evaluations of the chemicals. q 1998 Elsevier Science B.V. Keywords: Chemical; Safety evaluation

1. Introduction The safety evaluation of new drugs and chemicals by traditional studies in rodents and other experimental animals is beset with difficulties associated with )

Corresponding author. Molecular Toxicology Group, School of Biological Sciences, University of Surrey, Guildford, Surrey, GU2 5XH, UK. Fax: q44 Ž1483. 576-978.

species differences in metabolism and in mechanisms of toxicity w1–4x. This is particularly true in the evaluation of carcinogenic potential, which is first assessed by short-term tests such as the Ames mutagenicity assay, in the early stages of product development, to be followed several years later by life-span, two-species, rodent feeding studies. In many instances in recent years, negative mutagenicity studies have not been confirmed and subsequent

1383-5718r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII S 1 3 8 3 - 5 7 1 8 Ž 9 7 . 0 0 1 4 5 - 9

42

D.F.V. Lewis et al.r Mutation Research 412 (1998) 41–54

positive rodent carcinogenicity has resulted in disastrous financial losses due to the high costs of futile product development and safety evaluation, and loss of potential markets. Consequently, there is much interest in alternative approaches to safety evaluation, especially in predictive methods that can be used early in product development and are rapid, low-cost and relatively free from problems of species differences. One such method is computer-optimised molecular parametric analysis for chemical toxicity ŽCOMPACT., which is based on molecular structure, and structure-activity relationships in metabolism and chemical toxicity w2,5–9x. COMPACT is a method that identifies chemicals having molecular and electronic characteristics that make them suitable substrates for those cytochromes P450 which metabolically activate mutagens and carcinogens w2,7–11x. Cytochrome P4501 ŽCYP1. is the foremost family of the mixed-function oxidases that gives rise to chemical toxicity, which it does Ži. by oxygenating planar molecules that can intercalate with DNA w12x and thus function as genotoxic alkylating agents, and Žii. by genomal regulation via the cytosolic Ah receptor which results also in the initiation of the protein kinase C cascade, DNA replication and cellular proliferation w10–12x. Indeed, induction of CYP1 activity by a chemical is, in itself, a strong indicator of potential chemical carcinogenesis, because of the associated changes in epidermal growth factor receptors, leading to tumour promotion and progression w13x. Similarly, activation of CYP2E by substrates such as benzene and the small haloalkanes, results in the formation of reactive electrophiles and reaction oxygen species ŽROS. w14x, which can lead to genotoxicity and to oxidative stress culminating in tissue necrosis, mutations and malignancy w15,16x. It is known that ROS are involved in both the initiation and promotion of tumours w17,18x, and because of species differences in glutathione utilisation, ROS are likely to be a major cause of malignancy in rodents w1,2,10x. Substrates of CYP1 are characterised by their high degree of planarity, expressed as high values of areardepth2 Ž ard 2 ., and by their electronic structure w5–8,11,12x, which facilitates conformationally hindered oxygenation w9x, and they are characterised by low values of D E Ž D E s energy of the lowest unoccupied molecular orbital-energy of the highest

occupied molecular orbital, or D E s E ŽLUMO. y E ŽHOMO... The term Ž ard 2 % D E . is known as the COMPACT ratio, which is used as a criterion in the identification of CYP1 substrates. The other criterion is a low value of D E Ž D E s- 14.0.; however, it is more precise to examine the position of a chemical on a plot of ard 2 against D E. Criteria for potential CYP1 substrates were, therefore, originally taken as: ard 2 s) 5.0, D E s- 14.0, and COMPACT ratio Žard 2 % D E . s) 0.15 w19x. Substrates of CYP2E are characterised by a low molecular collision diameter; criteria for CYP2E substrates were, therefore, ˚ originally taken as: collision diameters- 6.5 A, ard 2 s 2.0 to 6.0, D E s) 10.0, COMPACT ratio s- 0.35 w19x. Evaluation of the COMPACT molecular orbital approach to the safety evaluation of chemicals was undertaken with the first 100 structurally diverse organic chemicals of the US NCIrNTP database w19x. Of the 100 compounds, 97 were positive in the rodent life-span carcinogenicity studies, when the criterion of carcinogenesis was taken as the appearance of malignancy in one or more of the 4 segments wnamely, rat ŽM., rat ŽF., mouse ŽM., mouse ŽF.x; 91 were positive in COMPACT ŽCYP1 plus CYP2E.; but only 59r97 were positive in the Salmonella mutagenicity ŽAmes. test. The degree of concordance between COMPACT prediction of toxicity and the rodent two species life-span carcinogenicity, for these first 100 compounds, was 92%, and between COMPACT and the Ames test was 64%. Subsequent to this COMPACT evaluation of the first 100 NCIrNTP chemicals, the criteria for CYP1 specificity were redefined to accord with detailed analysis of more complete molecular and electronic parameters of characteristic substrates of this cytochrome P450, and are now defined in terms of the COMPACT radius ŽCR.. In the discrimination of CYP1 substrates, low values of D E may give erroneously high values of the COMPACT ratio, leading to a high number of false positives. This is because the boundary that discriminates CYP1 substrates from those of other cytochromes P450 is a curve which has its origin at values of D E near to 8.0 w13x. In order to construct an equation which can give rise to a reasonably close approximation, and one which can produce a single COMPACT value, the most useful function is that of an arc of a circle of radius ‘CR’,

D.F.V. Lewis et al.r Mutation Research 412 (1998) 41–54

which defines the characteristics of CYP1. The lower the value of ‘CR’, the more positive the result, i.e. COMPACT is positive for CYP1 when CR - 6.5. The COMPACT radius ŽCR. values are readily calculated for known D E and areardepth2 data using the equation transform:

(

2

2

CR s Ž D E y 9.5 . q Ž ard 2 y 7.8 . . Hence, COMPACT is positive for CYP1 if CR - 6.5. The criteria for identifying potentially carcinogenic CYP2E substrates have also been modified, for it was found that if those chemicals with collision ˚ or less are arranged in order of diameters of 6.5 A their D E values, only those with D E - 15.5 were carcinogenic in the rodent assay. Hence, the criteria adopted for identifying potentially carcinogenic CYP2E substrates were a collision diameter of - 6.5 ˚ and D E of - 15.5 w21x. The original 100 miscelA laneous chemicals w19x have, therefore, been reevaluated using the new criteria, and the second 100 chemicals have similarly been evaluated by the revised COMPACT procedure designed to more accurately reflect requirements for CYP1 and CYP2E specificity. The partition coefficient Ž P . is a quantitative expression of the lipophilicity of a chemical; the greater the lipophilicity, the greater is the value of P. The major biological interest in P values is the potential of drugs and chemicals to penetrate lipid membranes. No matter what electronic, conformational or stereochemical features of the chemical determine its pharmacological or toxicological activity, the partition coefficent has a major role in the delivery of the chemical to its site of action w20x. Hence, although molecular planarity and frontier orbital energies model the ultimate interaction of the chemical with the activating CYP1 and CYP2E, and its oxidative metabolism to an ultimate carcinogen, the lipophilicity of the chemical will determine whether or not the molecule will be able to penetrate the lipid endoplasmic reticulum and reach the cytochromes P450 therein. It was therefore concluded that although COMPACT models a chemical’s ability to be metabolically activated to carcinogens, this parametric analysis needs to be combined with a quantitative expression of lipophilicity Žlog P . to

43

exclude those polar compounds unlikely to reach the cytochromes P450 of the endoplasmic reticulum. Based on observations that most P450 substrates have log P values ) 0, all chemicals with values of log P - 0 are considered to be too polar to penetrate the lipid endoplasmic reticulum membrane, and as they could not be activated to carcinogens and would be readily excreted, are regarded as negative for all aspects of COMPACT. Log P values have long been used by the Environmental Protection Agency, and other US government agencies, to model the bioavailability of potentially-toxic chemicals.

2. Methods Each of the 203 chemicals studied is listed in Table 1; associated data were taken from the work of Ashby and Tennant w22x, and molecular structures were analyzed for CYP1 and CYP2E specificity using the COMPACT procedure w8,19x. In that earlier work, molecular structures were constructed from first principles using the COSMIC modelling package w23x and minimised to give optimised geometries by the conjugate gradient method. Molecular shape parameters were calculated as previously described w19x. The energy minimised molecular structures were subjected to molecular orbital ŽMO. calculations by the CNDOr2 procedure w24x via the COSMIC software framework. The collision diameter, related to the van der Waals volume, is proportional to the cross-sectional area perpendicular to the main molecular plane. The structural alert modelling software, HazardExpert ŽCompuDrug Ltd., Budapest., was used for the generation of log P values. All calculations and measurements were performed on a Sigmex S6130 graphics terminal connected to a MicroVAX II computer running the COSMIC software package. In the present work, CYP1 substrates were identified from values of the COMPACT radius ŽCR.; positive for CR values of F 6.5. Potentially carcinogenic CYP2E substrates are identified from collision ˚ or less and D E values of diameter values of 6.5 A w x - 15.5 21 . The identification of potential CYP1 and CYP2E substrates by COMPACT, and the identification of Salmonella mutagenicity by the Ames

44

D.F.V. Lewis et al.r Mutation Research 412 (1998) 41–54

Table 1 Revised molecular parameters, Ames test and rodent life-span carcinogenicity assay for 203 miscellaneous chemicals

D.F.V. Lewis et al.r Mutation Research 412 (1998) 41–54

45

46

D.F.V. Lewis et al.r Mutation Research 412 (1998) 41–54

D.F.V. Lewis et al.r Mutation Research 412 (1998) 41–54

47

48

D.F.V. Lewis et al.r Mutation Research 412 (1998) 41–54

D.F.V. Lewis et al.r Mutation Research 412 (1998) 41–54

49

Key to carcinogenicity data: R s rat; M s mouse; m s male; f s female; P s positive; N s negative; E s equivocal; IS s inadequate study; CE s clear evidence; NE s no evidence; SE s some evidence; EE s equivocal evidence; NT s not tested; ) s difficult to evaluate. Level: A, carcinogenic to both R and M at one or more site; B, carcinogenic to only one species but at 2 or more sites; C, carcinogenic at a single site in both sexes of a single species; D, carcinogenic at a single site in a single sex of a single species; E, only equivocal evidence of carcinogenicity; F, non-carcinogenic; O.R.s overall result. † See Ref. w22x; metallic compounds have been omitted.

(

2 In the COMPACT assay ŽCYP1., values of COMPACT radius CR s Ž D E y 9.5 . q Ž ard 2 y 7.8 .

2

of 6.5 or less are taken as Žq.;

˚ or less, together with D E value of and values of ) 6.5 are taken as Žy.. In the COMPACT assay ŽCYP2E., a collision diameter of 6.5 A - 15.0 eV is taken as Žq.. When log P - 0, all aspects of COMPACT are taken as Žy.. The overall result in the rodent carcinogenicity assay is taken as Žq. if 1 or more of the 4 segments is P, CE, or SE; equivocal results ŽLevel E. are taken as negative.

50

D.F.V. Lewis et al.r Mutation Research 412 (1998) 41–54

test, were essentially unequivocal. However, the evaluation of the rodent carcinogenicity data, which is dependent on species, sex, and subjective histopathological evaluation, is sometimes equivocal. Therefore, as in the previous validation exercise w19x, the most stringent standards were applied, with a single positive result Ž1r4 segments or more. being regarded as an indication of positive carcinogenicity Žq., and a negative value Žy. being assigned only where no positive result occurred in any of the 4 Žrat male and female, mouse male and female. segments of the rodent assay.

trimethyl phosphate Žno. 49.; 11-amino-undecanoic acid Žno. 51.; bisŽ2-chloro-1-methyl-ethyl.ether Žno. 54.; dimethylhydrogen phosphite Žno. 58.; 1,2epoxybutane Žno.60.; chlorobenzilate Žno. 69.; diŽ2ethylhexyl.adipate Žno. 77.; 3-amino-4-ethoxyacetanilide Žno. 91.. These changes in COMPACT classification also resulted in changes in consonance, with concordances between COMPACT and the rodent assay decreasing from 92% w19x to 69% for the present study, and with the concordance between COMPACT and the Ames Salmonella assay changing from 64% w19x to 62% for the present study. The changes in COMPACT classification result

3. Results 3.1. Re-eÕaluation of first set of 100 chemicals In the initial evaluation w19x, all 100 chemicals had been examined for their potential as CYP1 substrates, but only 27 chemicals Žall negative for CYP1. were evaluated for potential CYP2E activity, of which 18r27 were positive. In this re-evaluation, all 100 chemicals were re-examined for both potential CYP1 and CYP2E activity. Using the new criteria for CYP1 ŽCR F 6.5. and CYP2E Žcollision diameter ˚ and D E F 15.5., 61 of the 100 chemicals F 6.5 A were positive for CYP1, while 36 of the 100 were positive for CYP2E, and 73r100 were positive for either CYP1 or CYP2E or both ŽTable 1.. In the re-evaluation, a further 27 of the 73 chemicals not previously examined for potential CYP2E substrate activity were found to be positive; none of the newly identified CYP2E substrates changed the earlier combined CYP1 and CYP2E COMPACT evaluations for potential carcinogenicity. In contrast, the re-evaluation of the 100 chemicals for potential CYP1 and CYP2E substrate activity led to re-classification of 15 of the chemicals, changing them all from positive to negative. These 15 chemicals, which were positive for CYP1 andror CYP2E by the original criteria w19x but became negative by the new ˚ or less; log P ) 0; criteria ŽCOMPACT radius, 6.5 A ˚ collision diameter, 6.5 A or less, and D E F 15.5., were as follows: allyl isovalerate Žcode no.1.; benzene Žno. 6.; 1,2-dichloroethane Žno. 19.; DEHP Žno.22.; 1,4-dioxane Žno. 23.; 4,4X-methylenebisŽ N, N-dimethyl.aniline Žno. 26.; nitrofen Žno. 35.;

Table 2 Concordance evaluations of COMPACT with ames test and rodent carcinogenicity assay Concordance of COMPACT evaluation with Ames test Test evaluations 1st 100 2nd 100 chemicals chemicals COMPACT Ames

q q

. 59r97 s 61% 16r100 . . 61r100s 61% . 0r97 45r100 . .

COMPACT Ames

y y

COMPACT Ames

q y

38r97

20r100

COMPACT Ames

y q

0r97

19r100

Total concordances

61q59 s61% 197

Concordance of COMPACT evaluation with rodent test COMPACT Ames

q q

73r100 s 73% 5r103

COMPACT Ames

y y

0r100

COMPACT Ames

q y

0r100

COMPACT Ames

y q

27r100 3r103

. . . 68r103s 66% . 63r103 . . 32r103

Total concordances

73q68 s69% 203

D.F.V. Lewis et al.r Mutation Research 412 (1998) 41–54

51

Table 3 Summary of concordances, sensitivities and predictivitiesa

from a closer adherence to modeled requirements for CYP1 and CYP2E specificity. The lack of concordance of this measure with rodent carcinogenicity could be attributed to a number of factors including: Ž1. carcinogenicity associated with induction of CYP4A, proliferation of hepatic peroxisomal enzymes and ROS production Žcompound nos. 22 and 77.; Ž2. elimination of polar chemicals with negative values of log P Žcompound nos. 23, 49, 58, 60 and 91.; and Ž3. carcinogenicity possibly attributable to generation of ROS through the activation of CYP2E Žcompound nos. 6, 19 and 54.. The remaining five compounds Žnos. 1, 26, 35, 51 and 69. with changed COMPACT classifications may be involved in similar, but as yet unknown, molecular mechanisms.

Concordances: COMPACT ŽCYP1 and 2E. with rodent assay s141r203s69% COMPACT ŽCYP1 and 2E. with Ames assay s120r197 a s61% Salmonella assay with rodent assay s121r197 a s62% No. of rodent carcinogens positive in test Sensitivity s No. of rodent carcinogens No. of rodent carcinogens positive in test Predictivity s No. of total positive in test No. of rodent nonycarcinogens in test Specificity s No. of total negatives in test

3.2. EÕaluation of the second set of 100 chemicals

Sensitivity: Predictivity: Specificity:

Using the new criteria for identifying potential CYP1 and CYP2E substrates, and log P values for lipophilicity, positive evaluations were as follows: Ames test: 34r100 plus 3 NT Žnot tested. Ž34%.; rodent assay: 8r103 Ž8%.; and COMPACT: 37r103 Ž36%.. For greater detail, COMPACT for CYP1 only was 37r103 Ž36%., for CYP2E only was 19r103 Ž18%.. Consonance values of the three methods of toxicityrcarcinogenesis evaluation of the second

COMPACTr Rodent assay

Ames assayr Rodent assay

78r108s 72% 78r110s 71% 63r93s68%

61r108s 57% 61r94s65% 59r104s 57%

a

Of the total 203 chemicals evaluated by COMPACT, 6 were not tested in the Ames assay.

group of 100 chemicals were, for rat vs. mouse in the rodent assay, 91r102 Ž91%.; rodent assay vs. Ames test, 62r97 Ž64%.; COMPACT vs. Ames, 61r100 Ž61%.; and COMPACT vs. the rodent assay, 68r103 Ž66%. ŽTable 2..

Table 4 Correlation of Rodent carconogenicity with COMPACT evaluation of 203 miscellaneous chemicals Levels of carcinogenicity a

No. of rodent carcinogens

CYP1 positive

CYP2E 2E positive

CYP1 and COMPACT with level

A

50

32

17

38

B

16

7

4

11

C

22

12

11

17

D

20

11

7

12

E F

0 0

7 20

3 13

7 25

a

Level A s carcinogenic to both rat and mouse at one or more site. Level B s carcinogenic to only one species but at 2 or more sites. Level C s carcinogenic at a single site in both sexes of a single species. Level D s carcinogenic at a single site in a single sex of a single species. Level E s any carcinogenicity is equivocal. Level F s non-carcinogenic.

Concordance of concordance of rodent carcinogenicity 38 50 11 16 17 22 12 20

Overall

s 76% s 69% s 77% s 60%

78r100s 72%

52

D.F.V. Lewis et al.r Mutation Research 412 (1998) 41–54

Combination of the two groups of 100 compounds gave positive evaluation totals for Ames of 48%; rodent assay, 53%, and COMPACT, 54% ŽTable 1.; and concordance values for the total 203 chemicals were, for rat vs. mouse 72%; rodent assay vs. Ames, 62%; COMPACT vs. Ames, 61%; and COMPACT vs. rodent assay, 69% ŽTable 3.. These values show anomalies between the first group of 100 compounds and the second group Ži.e. Ames positive: 61% and 34%; rodent carcinogenicity: 100% and 8%.which are a consequence of the arrangement of the chemicals in decreasing order of rodent carcinogenicity ŽLevels A to F, see Table 1.. In the first group, the number of positive results in the rodent assay Ž100%. is greater than in the Ames test Ž61%., in accordance with the known fact that many rodent carcinogens are not genotoxic; however, in contrast, the second group of chemicals contains more than four times the number of genotoxic mutagens Ž34%. than there are rodent carcinogens Ž8%., indicating that mutagenicity requires further pathology to progress to carcinogenicity w11x. The sensitivities of COMPACT ŽCYP1 plus CYP2E. and of the Ames test for carcinogenicity for the total 203 chemicals are shown in Table 3. The sensitivity of COMPACT for detecting carcinogens was 72% which contrasts with the sensitivity of 57% for the prediction of carcinogenicity from the Ames test; predictivity was 71% in contrast to 65% for the Ames test, and the specificity was 68% in contrast to 57% for the Ames test ŽTable 3.. The concordance of COMPACT with the severity of rodent carcinogenicity is 76% for level A Žcarcinogenic in both rodent species., 69% for level B Žone species at 2 or more sites., 77% for level C Žone species, at 1 site only. and 60% for level D Ža single site, in a single sex, of one species only. ŽTable 4..

4. Discussion As the 203 chemicals evaluated for potential carcinogenicity by COMPACT were originally selected by the NTP to be in order of decreasing severity of rodent carcinogenicity w22x, the first group of one hundred chemicals to be evaluated w19x were exclusively positive carcinogens Ž100%., whereas the second group of one hundred chemicals evaluated were

mostly non-carcinogenic Ž8% positive carcinogens.. This categorization of the chemicals in order of decreasing carcinogenicity resulted in a number of anomalous results, namely, that in the first group, the incidence of positive results in the rodent assay, the Ames Salmonella assay, and the COMPACT assay were 100%, 61% and 73%, respectively, whilst in the second group positive results were 8%, 34%, and 36%, respectively. This indicates that the second group may include many chemicals that are genotoxic Žmutagenic in Ames test., and are substrates for the activating cytochromes P4501A and 2E, yet are not rodent carcinogens. Such chemicals may be metabolized to reactive intermediates which can manifest many other forms of serious toxicity apart from carcinogenicity, including immunotoxicity e.g. halothane w25x, tissue necrosis, e.g. paracetamol, and chronic inflammatory diseases, such as scleroderma and asthma w26x, incapacitating and potentially fatal diseases that are medical problems the equal of cancer. The COMPACT and Ames test may also identify chemicals which although potentially toxic, may readily be detoxicated by Phase II metabolism, such as 8-hydroxyquinoline Ž no. 180 . , 4nitroanthranilic acid Žno. 194., and phenol Žno. 199.. However, for highly polar compounds, which are readily excretable e.g. ellagic acid, although COMPACT is positive there is no interaction with CYP1 and no genotoxicity w27x. In extensive studies of the prediction and extrapolation of rodent carcinogenicity, two major classes of rodent carcinogens have been identified, namely, putative genotoxic and non-genotoxic carcinogens w28x. Genotoxic carcinogens are mostly substrates of CYP1 and to a lesser extent, of CYP2E, and although many non-genotoxic carcinogens manifest their toxicity via ROS produced by CYP2E w11,19,29x e.g. ethanol, other mechanisms may be involved. Of the 108 chemicals in the present study that were found to be rodent carcinogens, 44 were negative in the Ames test and hence are probably non-genotoxic carcinogens. Of these, 23 were predicted not to be substrates of CYP1, although 6 were likely to be substrates of CYP2E. Typical of these non-genotoxic carcinogens are 1,1,2-trichloroethane Žno. 64., hexachloroethane Žno. 81., pentachloroethane Žno. 84., and tetrachloroethanes Žnos. 86, 87.. All of these are positive in the rodent assay, negative in the Ames

D.F.V. Lewis et al.r Mutation Research 412 (1998) 41–54

test, negative for CYP1A in COMPACT, but positive for CYP2E, which is known to metabolise halogenohydrocarbons with ROS generation. A second mechanism of non-genotoxic carcinogenesis involves the induction of CYP4A, proliferation of the peroxisomal enzymes, hepatotoxicity, peroxide production, cell proliferation and liver cancer w30x; typical of this class of non-genotoxic carcinogens are diŽ2-ethylhexyl.phthalate Žno. 22., diŽ2-ethylhexyl. adipate Žno. 77., and butylbenzyl phthalate Žno. 96., all of which were positive in the rodent assay but negative for CYP1 and CYP2E in COMPACT. A third mechanism involves the induction of protein kinase C via interaction with the Ah receptor, and microsomal enzyme induction, and is exemplified by 2,3,7,8-tetrachlorodibenzo-p-dioxin Žno. 43. but compounds of this type are planar and are positive in COMPACT w13x. Yet another mechanism is that of nitrilotriacetic acid Žno. 32. which complexes iron, facilitating its uptake by tissues and its subsequent generation of ROS and malignancy w31x. In a previous prospective evaluation of 40 miscellaneous chemicals by the COMPACT assay w32x, the concordance of COMPACT with rodent carcinogenicity was 72%, the sensitivity of COMPACT was 75%, and the correlation of COMPACT with potential genotoxicity was 63%, which compares with the previous concordance w19x of the Ames test with rodent carcinogenicity of 64% and a sensitivity of the Ames test of 56%. In the present study of 200 miscellaneous chemicals by COMPACT, concordance of COMPACT with the rodent assay was 69% ŽTable 3., concordance of COMPACT with the Ames test was 61%, and concordance of the Ames test with the rodent assay was 62%. The sensitivity of COMPACT is high, being 72% in the present study and 75% in the former smaller study w32x, and a high sensitivity is the foremost requirement for a shortterm carcinogenicity test, even if the predictivity is low. For it is essential not to miss any potential carcinogen in the first screening test, since subsequent investigation of any false positives is likely to reveal some of these chemicals as highly toxic noncarcinogens, exhibiting mutagenic, immunotoxic or other forms of chronic chemical toxicity, which is mediated by the CYP-mediated generation of reactive intermediates. COMPACT, in the present study, is characterised by greater sensitivity Ž72%. and

53

specificity Ž68%. than the Ames test Žsensitivitys 57%; specificitys 57%. and should be a more effective screening test for potential chemical carcinogens than is the currently widely-used Ames test. In addition, COMPACT may have a role in chemical safety evaluation that extends beyond the identification of potential carcinogens, and has considerable future potential in the rapid screening of chemicals for many forms of serious chronic toxicity, including carcinogenicity. Compounds with COMPACT ratios or collision diameters near the limits of criteria may be weak inducers of CYP1 or CYP2E respectively, andror lack specificity, inducing one or more CYP proteins. In such cases, limited enzyme induction studies ŽENACT. can determine both CYP specificity and potency of induction for a given chemical w33x, and enable a more informed prediction of potential carcinogenicityrtoxicity to be made. ENACT studies also yield information concerning the potential of chemicals for cell proliferation and inflammation, phenomena which are now considered high risk factors for human cancer w34x.

References w1x D.V. Parke, C. Ioannides, Role of cytochromes P-450 in mouse liver tumor production, in: D.E. Stevenson, R.M. McClain, J.A. Popp, T.A. Slaga, J.M. Ward, H.C. Pitot, ŽEds... Mouse Liver Carcinogenesis, Mechanisms and Species Comparisons, Alan R. Liss, New York, 1990, pp. 215–230. w2x D.V. Parke, C. Ioannides, D.F.V. Lewis, The role of the cytochromes P-450 in the detoxication and activation of drugs and other chemicals, Can. J. Physiol. Pharmacol. 69 Ž1991. 537–549. w3x G.B. Gori, Are animal tests relevant in cancer risk assessment? A persistent issue becomes uncomfortable, Regulat. Toxicol. Pharmacol. 13 Ž1991. 225–227. w4x G.B. Gori, Adjudicating cancer causation: Scientific, legal and political conflicts, Regulat. Toxicol. Pharmacol. 13 Ž1991. 309–325. w5x C. Ioannides, D.F.V. Lewis, D.V. Parke, Computer modelling in predicting carcinogenicity, Eur. J. Cancer Prevent. 3 Ž1993. 275–282. w6x D.F.V. Lewis, C. Ioannides, D.V. Parke, Molecular dimensions of the substrate binding site of cytochrome P-448, Biochem. Pharmacol. 35 Ž1986. 2179–2185. w7x D.F.V. Lewis, C. Ioannides, D.V. Parke, Structural requirements for substrates of cytochromes P-450 and P-448, Chem.-Biol. Interact. 64 Ž1987. 39–60. w8x D.F.V. Lewis, C. Ioannides, D.V. Parke, Prediction of chem-

54

w9x

w10x

w11x

w12x

w13x

w14x

w15x

w16x w17x

w18x

w19x

w20x w21x

w22x

D.F.V. Lewis et al.r Mutation Research 412 (1998) 41–54 ical carcinogenicity from molecular and electronic structure, Toxicol. Lett. 45 Ž1989. 1–13. D.F.V. Lewis, C. Ioannides, D.V. Parke, Molecular orbital studies of oxygen activation and mechanisms of cytochromes P-450-mediated oxidative metabolism of xenobiotics, Chem.-Biol. Interact. 66 Ž1989. 263–280. D.V. Parke, C. Ioannides, D.F.V. Lewis, Metabolic activation of carcinogens and toxic chemicals, Human Toxicol. 7 Ž1988. 397–404. D.V. Parke, The cytochromes P450 and mechanisms of chemical carcinogenesis, Environ. Health Perspect. 102 Ž1994. 852–853. C. Ioannides, D.V. Parke, The cytochrome P4501 gene family of microsomal haemoproteins and their role in the metabolic activation of chemicals, Drug Metab. Rev. 22 Ž1990. 1–86. C. Ioannides, D.V. Parke, Induction of cytochrome P4501 as an indicator of potential chemical carcinogenesis, Drug Metab. Rev. 23 Ž1993. 485–502. M. Ingelman-Sundberg, A.L. Hagbjork, ¨ On the significance of cytochrome P450-dependent hydroxyl radical-mediated oxygenation mechanism, Xenobiotica 12 Ž1982. 673–686. B. Halliwell, J.M.C. Gutteridge, Role of free radicals and catalytic metal ions in human diseases: An overview, Meth. Enzymol. 186 Ž1990. 1–85. H. Sies, Role of reactive oxygen species in biological processes, Klin. Woch. 69 Ž1991. 965–968. B.D. Goldstein, G. Witz., J. Zimmerman, C. Gee, Free radicals and reactive oxygen species in tumour promotion, in: R.A. Greenwald, G. Cohen ŽEds... Oxy Radicals and their Scavenger Systems: Cellular and Medical Aspects, vol. 2, Elsevier, Amsterdam, 1983, pp. 321–325. G. Witz, Active oxygen species as factors in multistage carcinogenesis, Proc. Soc. Exp. Biol. Med. 198 Ž1991. 675– 682. D.F.V. Lewis, C. Ioannides, D.V. Parke, Validation of a novel molecular orbital approach ŽCOMPACT. for the prospective safety evaluation of chemicals, by comparison with rodent carcinogenicity and Salmonella mutagenicity data evaluated by the US NCIrNTP, Mutation Res. 291 Ž1993. 61–77. A. Albert, Selective Toxicity, 7th ed., Chapman and Hall, London, 1985, pp. 65–84. D.F.V. Lewis, M.G. Bird, D.V. Parke, Molecular modelling of CYP2E1 enzymes from rat, mouse and man: An explanation for species differences in butadiene metabolism and potential carcinogenicity, and rationalization of CYP2E substrate specificity, Toxicology 118 Ž1997. 93–113. J. Ashby, R.W. Tennant, Chemical structure, Salmonella

w23x

w24x

w25x

w26x

w27x

w28x

w29x

w30x

w31x

w32x

w33x

w34x

mutagenicity and extent of carcinogenicity as indicators of genotoxic carcinogenesis among 222 chemicals tested in rodents by the US NCIrNTP, Mutation Res. 204 Ž1988. 17–115. J.G. Vinter, A. Davis, H.R. Saunders, Strategic approaches to drug design. I. An integrated software framework for molecular modelling, J. Comput. Aided Mol. Design 1 Ž1987. 31–51. J.A. Pople, D.P. Santry, G.A. Segal, Approximate self-consistent MO theory: I. Invariant procedures, J. Chem. Phys. 43 Ž1965. S129–135. A.L. Parke, C. Ioannides, D.F.V. Lewis, D.V. Parke, Molecular pathology of drug-disease interactions in chronic autoimmune inflammatory diseases, Inflammopharmacology 1 Ž1991. 3–37. A. Parke, D.V. Parke, The pathogenesis of inflammatory disease: Surgical shock and multiple system organ failure, Inflammopharmacology 3 Ž1995. 1079–1099. A.D. Ayrton, D.F.V. Lewis, R. Walker, C. Ioannides, Antimutagenicity of ellagic acid towards the food mutagen IQ: Investigation into the possible mechanisms of action, Fd. Chem. Toxicol. 30 Ž1992. 289–295. J. Ashby, D. Paton, The influence of chemical structure on the extents and sites of carcinogenesis for 522 rodent carcinogens and 55 different exposures, Mutation Res. 286 Ž1983. 3–74. D.F.V. Lewis, C. Ioannides, D.V. Parke, A retrospective evaluation of COMPACT predictions of the outcome of NTP rodent carcinogenicity testing, Envir. Health Perspect. 103 Ž1995. 178–184. D.F.V. Lewis, B.G. Lake, The interaction of some peroxisome proliferators with the mouse liver peroxisome proliferator-activated receptor Žppar.: A molecular modelling and quantitative structure-activity relationship study, Xenobiotica 23 Ž1993. 79–96. N.E. Preece, D.E. Hall, J.A. Howarth, L.J. King, D.V. Parke, Effects of acute and sub-chronic administration of iron nitrilotriacetate in the rat, Toxicology 59 Ž1989. 37–58. D.F.V. Lewis, C. Ioannides, D.V. Parke, A prospective toxicity evaluation ŽCOMPACT. on 40 chemicals currently being tested by the National Toxicology Program, Mutagenesis 5 Ž1990. 433–435. C. Ioannides, D.F.V. Lewis, D.V. Parke DV, The use of computers in the safety evaluation of drugs and other chemicals, Eur. J. Drug Met. Pharmacokin. 3 Ž1994. 225–233. B.N. Ames, Mutagenesis and carcinogenesis: Endogenous and exogenous factors, Environ. Mol. Mutagenesis 14 ŽSuppl. 16. Ž1989. 66–77.