An analysis of the spectra of genetic activity produced by known or suspected human carcinogens *
Nei l E. G a r r e t t 1, H. F r a n k S tack 1, M i c h a e l R. G r o s s i a n d M i c h a e l D. W a t e r s 2
I Northrop Services, Inc., Environmental Sciences, Research Triangle Park, NC 27709 (U.S.A.); and 2 Genetic Toxicology Division, Health Effects Research Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711 (U.S.A.)
(Received 3 February 1984) (Accepted 24 February 1984)
Summary
For 24 agents classified by the International Agency for Research on Cancer as known or suspected human carcinogens, we previously catalogued the qualitative genetic bioassay data available in the literature. In the present analysis, dose information, where available, was added to this data base: either the lowest effective dose (LED) or the highest ineffective dose (HID) was recorded for each agent and bioassay system. Bioassay systems were organized according to classes of genetic activity and subdivided by the phylogenetic level of the test organism. For each compound, the quantitative results in the test systems were represented by computer-generated bar graphs ('genetic activity spectra'). The x-axis unit values corresponded to the 100 different test systems, and the y-aKis values were the logarithmically transformed LED or HID values.
Statistical methods and pattern-recognition techniques were used to evaluate the genetic activity spectra. Spectra were compared among agents grouped according to target-organ specificity. In addition, the spectra of all possible pairs of compounds were compared to identify compounds displaying qualita- tively or quantitatively similar genetic activity. Chemically similar compounds frequently produced similar spectra of genetic activity, and it was possible to identify the most appropriate test systems for some classes of compounds. As the data base for human carcinogens is enlarged, analysis of genetic activity spectra may contribute to our understanding of the structure-activity relationships and mechanisms of action of these agents.
Currently, more than 100 test systems or bioas- says exist for evaluation of agents that produce genetic or genetic-related effects (Waters and Auletta, 1981). These test systems have evolved as a result of research directed toward detection and evaluation of chemicals and other materials that may influence human health. Because a primary concern of such programs has been the detection of agents that cause heritable mutation or cancer
* Reprint requests should be directed to MDW.
in man, a number of the bioassays use human cells, tissues or body fluids. The bioassays are conveniently divided into 5 major categories according to the type of genetic or genetic-related damage. These are (1) point/gene mutation, (2) chromosomal effects, (3) other genotoxic effects, (4) cellular transformation and (5) effects on sperm
morphology. These 5 major categories may be subdivided according to the phylogenetic level of the test organism. They may also be classified as either in vitro or in vivo systems.
rates a variety of genetic effects. Gene mutation assays include measurement of forward and re- verse specific gene mutation, sex-linked recessive lethality, and coat-color genetic changes, and host-mediated assays involving gene mutation in microorganisms. Other genotoxic effects are mea- sured as differential toxicity in repair-deficient or recombination-deficient strains, mitotic recombi- nation or gene conversion, unscheduled DNA synthesis, inhibition of DNA synthesis or DNA- strand breaks. Chromosomal effects are determined as sister-chromatid-exchange formation, micro- nuclei, chromosomal aberrations or aneuploidy, dominant lethality and heritable translocations. Cell transformation is scored as altered cell mor- phology in a number of primary and continuous mammalian cell lines. Effects on sperm include structural abnormalities in the parental and F 1 generations. Measurements of genetic or genetic- related effects using these 5 categories of bioassays provide a ' spectrum' of genetic information.
In previous articles, we reviewed the clinical and epidemiological information and catalogued the genetic bioassay data available for 24 agents classified by the International Agency for Re- search on Cancer ( IARC) as known or suspected huma n carcinogens (Waters et al., 1983a, b; IARC, 1979). This report focuses on the profiles or spec- tra of genetic activity produced by these 24 agents, the characteristics of the spectrum for each chemi- cal, and the information gained by analysis and comparison of the spectra produced by certain groups of these compounds. Spectral analysis of the data obtained from several categories of bioas- says provides information beyond that gained from a single bioassay result. Such analysis should facilitate accurate categorization and evaluation of agents that cause heritable mutation or cancer in m a n .
Methods
Qualitative aspects of the data base established for the agents described in this report were re- ported previously (Waters et al., 1983b). This qualitative data base consisted of the bioassay definition and code, test result (with and without metabolic activation), publication number, and corresponding reference. In the present study, the
published reports available as of June 1, 1981, comprising the data base were reexamined for information concerning dose. The total data base was examined to eliminate redundant data that had been published in two or more reports. This was facilitated by examining the computer files for specific bioassays and comparing reports from the same authors with equivalent bioassay results, dos- es, and so forth.
Determination of effective doses * It was not practical to record details of the
dose-response relationships for a data base con- taining over 1000 test results. Furthermore, posi- tive or negative results were often reported for a single dose. For these reasons, either the reported minimal effective dose or the maximal dose tested that produced no effect was recorded for each test agent and bioassay system. This analytical ap- proach allowed results obtained at a single dose to be compared with other results in the data base.
For negative test results, the highest dose studied was termed the highest ineffective dose (HID). If there was evidence of extreme toxicity, the next highest dose was used. A single dose tested with a negative result was considered equivalent to the HID. Similarly, for positive results, the lowest effective dose (LED) was recorded. Theoretically, the HID and LED both approach the no-ob- served-effect level as their limit (Chane t at., 1982). If probability values were reported, the dose for positive effects was extracted from tables or graphs for which the results were significant at P _< 0.05. If the probability value was not available, the reviewers made a conservative judgment of the dose required for the effect. A clear positive re- sponse was indicated when an increase in effect directly proportional to the increase in dose was observed for two or more doses; the lower of these doses was generally taken as the LED. A single dose was considered equivalent to the LED if the observed response was at least twice the sponta- neous background.
Dose information in the literature was reported
. The term 'dose' in the context of the present report does not imply consideration of exposure or treatment time and, therefore, can be considered synonymous with concentra- tion.
in various units (e.g., parts per million, percentage, mass per volume, volume per volume). Units for liquids and solids were converted to mass per volume. Gaseous compounds such as vinyl chlo- ride reported as parts per million on a volume per volume basis were converted to parts per million, mass per volume, according to the ideal gas laws. Dose units for in vitro test systems were expressed as micrograms per milliliter. A volume of 2 ml was assumed for the top agar in the microbial plate-in- corporation assays, and a 1-ml volume was used for differential toxicity assays. (No correction was made for diffusion of the compound into the bot tom agar.) Doses for in vivo bioassays were expressed as milligrams per kilogram body weight of the treated animal.
Because of the wide range of doses encountered in 100 bioassays, the logarithm of the dose was used in calculations and graphs. Because genetic activity was plotted on a log scale, differences in molecular weights of compounds did not greatly influence comparisons of their overall responses.
Representation of a genetic activity spectrum A data set consisting of a discrete number of
bioassays and quantitative values for the response as a function of dose can be easily represented in a bar graph. This was accomplished for each chemi- cal in the data base by using x-axis unit values (i = 1-100) corresponding to the 100 different test systems and y-axis values corresponding to the response function for each individual test system. The numbers of the points on the x-axis and corresponding bioassay codes and definitions are shown in Table 1. The bioassays are grouped according to classes of genetic or genetic-related activity and arranged within the classes according to phylogenetic level. Thus, a discrete array of tests and responses defines a readily interpretable profile or spectrum of genetic activity.
Spectral-line height (the magnitude of each bar) is a function of the LED or HID, which is associ- ated with characteristics of each individual test system, such as population size, cell-cycle kinetics, and metabolic competency. Thus, the detection limit of each test system is different, and across a given spectrum, responses will vary substantially. The spectra were compared and analyzed only on the basis of common test results (i.e., where the
91
same test systems were used to evaluate two or more of the chemicals under consideration). No effort was made to adjust or relate responses in one test system to those of another. Such adjust- ments are difficult unless directly comparable pro- tocols are employed.
To accommodate both positive and negative responses on a continuous scale, doses were trans- formed logarithmically so that effective (LED) and ineffective (HID) doses were represented by posi- tive and negative numbers, respectively. The re- sponse or logarithmic dose unit (LDUij) for a given test system i and chemical j was represented by the expressions
LDUij = - log l0(dose) , for H I D values; LDU < 0
(1)
LDUij = - log]0 (dose × 10- 5 ),
for LED values; LDU > 0
These simple relationships define a dose range of 0 to - 5 logarithmic units for ineffective doses (1-100000 # g / m l or mg /kg) and 0 to +5 loga- rithmic units for effective doses (100000-1 /~g /ml or mg/kg) . A scale illustrating the LDU values is shown in Fig. 1. Negative responses at doses less than 1 were set equal to 1. Effectively, an LED value > 100000 or an H I D value < 1 produces an L D U = 0; therefore, no quantitative information is gained from such extreme values.
Data management and analysis A computerized da ta -en t ry program was used
to store abstracted data from approximately 600 citations on disk files (1000 test results). A random file was created for each test system, and a record within the file consisted of a 67-character-array
+ Bioassags
110 i I I I I LED 1{) 100 1000 10000 100000
LED x 10 ~ 0.00001 0.0001 0.001 0.01 01 1 Loglo[LED x 10 s] 5 4 3 2 1 0
Chinese hamster cells, forward mutation lung HGPRT locus ovary HGPRT or ATPase locus lung (V79) HGPRT locus lung (V79) ATPase locus
Mouse lymphoma (L5178Y) cells forward mutation, TK locus
Arabidopsis mutation Tradescantia mutation Drosophila melanogaster sex-linked recessive lethal test Mouse spot test Mouse specific locus test, postspermatogonial stages Mouse specific locus test, all stages
Escherichia coli pol A (W3110-P3478) spot test liquid suspension test
Bacillus subtilis rec (H17-M45), spot test
TABLE 1 (continued)
Number Code ~ Definition
32 REC 33 (WPR)
Lower Eukaryotic assays
34 YEC 35 YEH
Higher Eukaryotic assays
36 (UDB) 37 U D H 38 (UDL) 39 U DS 40 (UDX) 41 UDP 42 (UDM)
43 (IDL) 44 (IDP) 45 (IDR)
46 (DBH)
DNA-repair-deficient bacteria Escherichia coil WP100 uvrA rec or other rec strains
Saccharomyces cerevisiae gene conversion homozygosis (through recombination or gene conversion)
Unscheduled DNA synthesis in mammals human bone marrow human diploid fibroblasts HeLa cells all cell types, in vitro Xeroderma pigmentosum cells rat primary hepatocytes mouse, in vivo
Inhibition of DNA synthesis HeLa cells rat primary hepatocytes rodent cells
DNA strand break, human
C H R O M O S O M A L EFFECTS A SSA YS Sister- chromatid exchange 47 SCC 48 SCF 49 SCH 50 SCM 51 SCP 52 SCV 53 SCE 54 SCL 55 SC2 56 SC3 57 SC4
Chinese hamster ovary (CHO) cells, transformed Human fibroblasts, normal HeLa cells, transformed Humans lymphoblastoid cells, transformed Chinese hamster fibroblasts, transformed Chinese hamster lung fibroblasts (V79 cells), transformed In vitro and in vivo Human lymphocytes in vitro In vitro, all animals except human In vivo, all animals except human In vivo, human cells
Neurospora crassa
Drosophila melanogaster whole sex chromsome less partial sex chromsome loss
Mammalian cytogenetics Chinese hamster Syrian golden hamster mouse human all cell types lymphocytes, human Xeroderma pigmentosum cells
Mammalian cytogenetics bone marrow studies, all animals
93
94
TABLE 1 (continued)
N umber Code a Definition
69 CYL 70 CYG
71 CYO 72 CYS 73 CYT
74 HOC 75 TRC
Micronuclei 76 (MNC)
lymphocyte or leukocyte studies, all animals spermatogonial stem cells treated, spermatocytes observed oocyte or early embryo studies spermatogonia treated, spermatogonia observed differentiating spermatogonia or spermatocytes treated, differentiating spermatocytes observed
Hordeum cytogenetics Tradescantia cytogenetics
In vitro
In vivo hamster mouse rat
77 (MNH) 78 (MNM) 79 (MNR)
Chromosomal damage in vivo Dominant lethal test
80 (DLD) Drosophila melanogaster 81 DLM mouse 82 DL R rat
88 CTB BALB/c3T3 cells 89 CTH C3H10T1 /2 cells 90 CTL Established cell lines 91 CTK A K R / M E cells 92 CTR RLV/F i sche r rat embryo cells
93 CTV Viral enhancement 94 CT7 SA-7 v i rus /SHE cells
SPERM- MORPHOLOG Y A SSA YS 95 SPA Rat 96 SPH Human 97 SPI Mouse 98 SPR Rabbit 99 SPS Sheep
100 SPF Mouse F 1 assay
Parentheses indicate codes not specified by the EPA Gene-Tox Program (Waters et al., 1983b).
code for the chemical number, dose, qualitative test results, test code, citation number, and addi- tional information, An abbreviated citation corre- sponding to each publication was stored in an
additional file. The data could b e searched or sorted for any specified segment of the character arrays by accessing both the test data file and the citation file. The LDU was calculated by searching
for the segment con ta in ing the dose and calculat- ing the a p p r o p r i a t e logar i thmic function. The L D U s for the set of 100 b ioassays were ca lcula ted and s tored on magnet ic tape in a separa te file for each chemical . When confl ic t ing results for a given
b ioassay were encountered , the subset of da t a cor- r e spond ing to the major i ty of the test results ( + or - ) was used. Only those posi t ive or negative results accompan ied by a dose value were used to calcula te the magni tude of the spectral line. An average of the logar i thmic values of the da ta sub- set was calculated. In the few cases where numbers of posi t ive and negative results were equal, the overal l result for the c o m p o u n d was cons idered posit ive, and the doses cor respond ing to the posi- tive results were used in the calculat ion.
If the genet ic act ivi ty spec t rum of each of the 24 chemicals is c o m p a r e d with that of each o ther chemical , 276 combina t ions are possible. To de- te rmine pa t te rns or s imilar i t ies in test data , the da t a were examined over the ent ire series of 100 tests for all c o m m o n tests conduc ted for each pair of chemicals. The type and number of c o m m o n tests and the n u m b e r of c o m m o n tests with con- f l ict ing results were computed . Add i t iona l ly , the dose- re la ted features of the genetic profi le for c o m m o n tests of the 24 chemicals were compared . Agreemen t between the test results can be de- t e rmined by examin ing both the pos i t ions and magni tudes of the responses. The logar i thmic val- ues of the effective doses were used to compa re the magni tudes of the responses. This compar i son was made by first ca lcula t ing the difference in the values of the logar i thmic responses,
ILDU~j. - LDUij" I
where j~ and Jb deno te the two chemicals being compared , and i the test system. A dose- re la ted funct ion ( D R F ) was then def ined to serve as a measure of agreement of the magn i tude of the lines of the genetic prof i le or spectrum. The D R F was formed from a sum of the terms der ived f rom the co r re spond ing c o m m o n genetic tests:
D R F = 1 ~ c - ILDU~j . - LDU~j.I (2) n
i
where n is the total number of c o m m o n tests, and
Fig. 2. Schematic of the genetic activity spectra showing 5 bioassays of 2 chemicals, ja (abbreviated a) andjb (abbreviated b). Bioassays 2 and 4 are common to both chemicals; however, results are positive in test 4 for chemical b and negative for chemical a. The LDU u values determined by equations (1) in Methods are shown for the test responses. The dose-related function for this example is
L D U 4 a is negative, a n d L D U 2 a ~ L D U 2 b ; t hus
ILDU4b - LDU4al > iLDU2b - LDU2~I.
Conflicting results for test 4 cause the value of DRF to be substantially reduced, agreeing with the observation of dispar- ity in the genetic activity spectra.
c is a constant . The cons tan t c was assigned a value of 3.0 in these calculat ions, to yield largely posi t ive values for D R F . A D R F value of 2.0 co r responds to an average dose difference of one logar i thmic unit. Fig. 2 i l lustrates the mechanics of compar ing test spect ra of two chemicals for com- mon tests and dose- re la ted features. By the same procedure , a second dose- re la ted funct ion (DRF~) was de te rmined solely from the concordan t test results.
Results
Comparison of the genetic activity according to target-organ specificity
The genetic act ivi ty spect ra were ini t ial ly ex- amined according to target organ specif ici ty for t umor induct ion - - that is, accord ing to whether
96
B C M E ~,, C M M E o , M U S T A R D G A S m
5-
I -
O -
l t.¢3 00 03 < < ( < co ( /)( / )
-5 I ,
PT/GENE MUT. OTHER EFFECTS CHROMOSOMAL EFFECTS ~ l ~
~ I ~ ~I ~
Z
TRANS SPERM ~ 1 ~ - - I A I -I- -I--I
, , , I , ,I , , , I , I I I ' ~ I I I I
10 25 40 55 70 85 100
BIOASSAY SYSTEMS
Fig. 3. Genetic activity spectra for the volatile alkylating agents, bis(chloromethyl) ether (BCME), chloromethyl methyl ether (CMME) and bis(#-chloroethyl) sulfide (mustard gas).
the agent is carcinogenic in the respiratory tract, hematolymphopoietic system, bladder or liver.
Human respiratory-tract carcinogens. Insuffi- cient genetic activity data were available to ade- quately compare the spectra (Fig. 3) of the highly volatile alkylating agents bis(chloromethyl)ether (BCME), chloromethyl methyl ether (CMME), and bis(#-chloroethyl)sulfide (mustard gas). The struc- tures and the IARC classifications for this group of alkylating agents and for the reactive sub- stituted alkenes acrylonitrile and vinyl chloride are shown in Fig. 4. Considerably more bioassay data were available for acrylonitrile and vinyl chloride, as revealed by the spectra in Fig. 5. In the com- parison of the two carcinogens, the spectral lines were concordant for the prokaryotic point mu- tation assays using Salmonella tester strains TA1535, TA1537 and TA100 (SA5, SA7 and SA0), but the test results differed for the frameshift tester strains TA1538 and TA98 (SA8, SA9) and
the urine test (BFU). The results also conflicted for the bone-marrow cytogenetics assay (CYB).
The genetic activity spectra for the inorganic metal compounds of As, Cd, Cr and Ni and for the asbestiform minerals are shown in Figs. 6 and
ACRYLONITRILE (2) BIS(CHLO ROMETHYL)ETHER (l)
H C~ N CHz CI \ / / C=C O
H ILl CH z CI
VINYL CHLORIOE (1) CHLOROMETHYL METHYL ETHER (1)
H CI CH \ / .' C C O I \ '\
H H CH: CI
ASBESTOS As Ed Cr Ni (1) (1) (2) (1) I2)
MUSTARD GAS (1)
CH, CH:Ci I
Sx CH~ CH~CI
(I) IARC Human Carcinogen (2} IARE Probable
Fig. 4. Human respiratory-tract carcinogens.
ACRYLONITRILE VINYL CHLORIDE PT/GEII~ MUT 0THIEA EFFECTS Ct4ROM~IEIMAt EFFECTS TRANS SPERM PT/GEN| MiIT OTHER EFFECTS CHROMOSOMAL EFFECTS
- 5 I . . . . . I . . . . I , , , - • - • I . I I ~ . ~ o . . . . I . , . I . . . . . . . I . I I I I I I I I I " I I I I I I I
10 25 40 55 70 85 100 10 25 40 55 70 85 100 BIOASSAY SYSTEMS BIOASSAY SYSTEMS
Fig. 5. Genetic activity spectra for acrylonitrile and vinyl chloride.
97
7. A combined spectrum for the metals in Fig. 7 shows several conflicting test results, especially among the Salmonella tests. A series of lines grouped together reflects positive results in assays for chromosomal aberrations in vitro and cell transformation.
Very few tests gave concordant results for all the metals; however, a series of test results were concordant for one or more pairs of the metal compounds. Concordant positive results were seen for the prokaryotic point mutation assay WPU; the yeast primary DNA-damage assay YEH; the in vitro chromosomal aberration assays CYU, CYZ and CYH; and the cellular transformation assays CTC and CT7. Arsenic and chromium gave nega- tive results in the prokaryotic gene mutation assay SA5.
Each metal gave positive results in one or more tests at doses < 1 ~g/ml . At these levels, arsenic caused sister-chromatid exchange in human lymphocytes (SCL), virus-enhanced cell transfor- mation (CT7), and chromosomal aberration in hu- man cells, including lymphocytes in vitro (CYZ, CYH).
Cadmium gave positive results in CHO cell cytogenetics (CYU) and cell transformation sys- tems (CTC, CT7) at concentrations of 0.1-1 ~g/ml . Nickel at these levels caused chromosomal
aberration in vitro (CYC) and cell transformation in the Syrian hamster embryo clonal assay (CTC).
The chromium compounds included both hexavalent and trivalent compounds and lead chromate. The spectral lines most frequently rep- resent hexavalent chromium. For prokaryotic gene mutation assays using the five Salmonella tester strains and E. coil WP2 uvrA, results from experi- ments with lead chromate were averaged with the values obtained for hexavalent chromium. The positive genotoxic effects in S. cerevisiae (YEH) and the negative results in E. coli pol A (REP) were obtained with lead chromate only. Both triva- lent and hexavalent chromium gave positive re- suits for prokaryotic gene mutation (ECO), un- scheduled DNA synthesis (UDS), chromosomal aberrations (CYU, CYY, CYH), and cell transfor- mation (CTB). Negative results in the Salmonella tester strain TA1535 (SA5) came from experiments using trivalent and hexavalent chromium and lead chromate.
Chromium was the most extensively studied metal in the data base. The chromium compounds were active at concentrations < 1.0 ~g /ml and in some tests at concentrations < 0.1 t tg/ml. Bioas- says at concentrations of 1.0 /~g/ml or less in- cluded unscheduled DNA synthesis in human di- ploid fibroblasts (UDH) and other cell systems
9 8
u ~ 5 - p -
z
I I I ( / )
O ~ 0 - (.9 O . J
A R S E N I C
¢n5- I,--
I i i ,~+/)
0 c3 0 _ (.9 0 ._J
P1flOEME MUT. . . . . L~..'OTHER EFFECTS I CHNOMOSOMAL EFFECTS _I_TAANS IjSPERM L F - I - 1 - - I - - I - r
N - I -
d >> i ( . 9 ~ 9
~- m
+11 a
Q.
o
- 5 , . . . . . I . . . . J . . . . . . . i . I J - 5 I I I I 1 I I I
10 25 4 0 55 70 85 100 B I O A S S A Y S Y S T E M S
C H R O M I U M PTIGSIIEMUT OTHEREFFECTS CHROMOSOMALEFFSCTS TRAILS SPERM
I r l - - I - I
b > > - N I ~ ) - >" >- ~--
: ) Q ' J ~ j
r ~ . . . o m ~ I-- ~ {~ ~ ~_ 5 -
]fault II +I lt+l II in . . . I
- 5 I , , , , , I . . . . I . . . . . . . I . I I I r 1 I I I I
10 25 4 0 55 70 8 5 1 0 0 B I O A S S A Y S Y S T E M S
F i g . 6 . G e n e t i c a c t i v i t y s p e c t r a f o r a r s e n i c , c a d m i u m , c h r o m i u m a n d n i c k e l .
C A D M I U M
PT/GSIS MOT OTHER EFFECTS CHIqOMOSOUAL EFFECTS TRAMS SlMERM L- --I-- ~L. ~L. ~L- .J r " - I - I - I - I I
5 r r
/I .............. < < <
{J
-J r~
.J £3
. . . . i . . . . I . . . . . . . I . I I I l I I I I I
10 25 4 0 55 70 85 100 B I O A S S A Y S Y S T E M S
N I C K E L
PT/GENE MUT OTHER EFFI~CTS CHROMOItOM&L EFFECTS "roANS SPERM
: - I - I -|~ - 1 I
-=~ I ' i ' ' i l ' ' I I I . . . . . I
(.2 t.t. ¢O
~ t D ¢ 9
I . I I I I
1 0 2 5 4 0 5 5 7 0 8 5 1 0 0
B I O A S S A Y S Y S T E M S
(UDS); sister-chromatid exchange in CHO cells (SCC) and normal human fibroblasts (SCF, < 0.1 /~g/ml); and chromosomal aberrations in human cells, including lymphocytes in culture (CYZ and CYH), and in hamster and mouse cells (CYV and CYY, < 0 .1/~g/ml) . Chromium also gave positive results at doses < 0.1 F g / m l in the BALB/c 3T3 mouse (CTB) and RLV/Fischer rat embryo (CTR) cell transformation systems, and _< 1.0 /~g/ml in the SHE clonal transformation system (CTC).
Human hematolymphopoieticosystem carcinogens. The structures of the hematolymphopoietic-system carcinogens and their IARC classifications are
shown in Fig. 8. The number of test results for phenytoin was insufficient for comparison with those for the other chemicals. The 4 compounds (Fig. 9) for which sufficient test data existed were conveniently divided into two groups, one exhibit- ing a large number of negative test results and the other positive responses. Benzene and chloram- phenicol produced more negative than positive responses; their spectra were concordant for 3 negative test results, the prokaryotic and higher eukaryotic gene mutation assays SA5, SA8, and SRL. In chromosomal aberration tests, both of these compounds gave positive results in vitro
99
ASBESTOS
~5-
~o- _ J
PTIG|NT MUT OTIlER EFFECTS CIlROMOSOMAL EFFECTS |TRAILS SPERM I LL-- ~ L . . . *_ J
- I l - I - l - - t
METAL COMPOUNDS
> >- L)
if, i >- (..) )==
=a tj >. t=)
. . . . , I . , , , , , , | . | I [ I i " I I [ I I
10 25 40 55 70 85 100 B I OASSAY SYSTE MS
° 0 -
PTISEN! MVT _ i_ _ = __ a }~ -I" O T . , EFFIECTS. 1_1_. CIIIIOSI4~SOMA L ~,FECTS '_ , -T"ANS ,-SP1E aM,
2
!IF I . . . . . I,
I I
10 25
~N I >~ >- ~- C_) (_) (.)
fill -5- -5 . . . . i . . . . .
I I I
40 55 70 BIOASSAY SYSTEMS
L ) I - t j
i,,lil ' t ' ' l
85 100
Fig. 7. G e n e t i c ac t iv i ty spec t r a for the a s b e s t i f o r m minera l s a n d a c o m b i n e d s p e c t r u m for the meta l s As, Cr, C d a n d Ni.
(CYH) and conflicting results in vivo (CYB). An interesting result is the < 0.1 ~g /ml LED value obtained for benzene using lymphocyte cytogenet- ics in vivo (CYL).
Primarily positive results were obtained for melphalan and ethylene oxide. Results for the chemicals were concordant for the higher eukaryotic gene mutation assays L5T and SRL. The potencies of ethylene oxide and meiphalan were reflected in the number of spectral lines indicating a mean LED of < 1 ~g/ml . For ethyl- ene oxide, these low LEDs were found in the L5T, UDM, SC4, CYB and CYL tests. LEDs < 1 ~g /ml for melphalan were obtained in the L5T, SC2,
Fig. 8. H u m a n h e m a t o l y m p h o p o i e t i c sy s t em ca rc inogens .
P H E N Y T O I N ( 3 ) M
c~x5 .~/-~ -...f o c6.~I ,
(3) IARC Unclassified
CYC and CTH systems. Thus, both ethylene oxide and melphalan were effective at low doses in the mouse lymphoma forward-mutation system, a sis- ter-chromatid exchange system, and a mammalian cytogenetics system.
Human bladder carcinogens. This chemical group consists of 4 aromatic amines and 2 nitro- gen mustards. The structures and IARC classifica- tions of these compounds are shown in Fig. 10. Only 2 tests were common for cyclophosphamide and chlornaphazine, the Salmonella base-pair sub- stitution assay (SAO) and the sex-linked recessive lethal test using Drosophila (SRL). The genetic activity spectrum for cyclophosphamide is shown in Fig. 11. The data for cyclophosphamide con- sisted of 191 test results and were the most exten- sive for any agent in the data base; 82% of the test results were positive. The most sensitive test (LED < 1 #g /ml ) for cyclophosphamide was unsched- uled DNA synthesis in HeLa cells (UDL). The genetic activity spectrum for chlornaphazine is not shown, because this compound was tested in only three bioassays; however, it gave positive results in the virus-enhanced cell transformation assay (CT7) at 0.1 /~g/m[.
The spectrum of auramine displays positive re- sults that are concordant with the spectra of the aromatic primary amines 4-aminobiphenyl and 2- naphthylamine (Fig. 12) for the E. col pol A
100
5 - O3 I--
:D
U.I O3 0 c~ 0 -
o - - I
- 5 -
BENZENE
rT/gileIlle MUT OTNIEII EFFECTS CHROMOSOMAL leFFleCT~ ITRAII| III~FRII l - _ u - u
¢_
rr '- C,
!tit! ....................... ! ................. i ......... ! • i - • i l - • i . i . i • . i • • ]
10 25 40 55 70 85
BIOASSAY SYSTEMS
z
. ° ] I I
100
M E L P H A L A N
5 .
° O.
5 ~
I--
w
0 a 0 -
0 _ I
PTIGIElele MUT. OYNleR EFFECT| CHROMOSOMAL £ F FECTS TRANS SPERM
- ! - i - I - - I
F- ro I--
!L I]FFII I I . . . . . I . . . . J . . . . . . . I . I i
I I I I I I 1 10 25 40 55 70 85 100
BIOASSAY SYSTEMS
-5 - 5 -
C H L O R A M P H E N I C O L
PTIGENIE MUT OTHle R IEFFIICTS CMROMOSOMA L leFFleCTle _ | _ _ ! . . . . . .
I I " - I I I
< ,¢ I -
~ 0 I I rr ~ t u ~ a:
n- i¢/~
,x l .I . . . . . . I ' i ' " i " " i i i 10 25 40 55 70
BIOASSAY SYSTEMS
>.>-
II °
S Ol . I I
I I 85 100
E T H Y L E N E OXIDE
PT/GIEHIE MUT OTHER leFFleCTS CHROMOSOMAL EFFECTS TRAILS SPEFlM . - t - - -L - L - l - - I
- I l - - I - - I - - I
¢
i . . . . . I . .I . . . . I . I I i i i ' 1 ' i i i
10 25 40 55 70 85 100
BIOASSAY SYSTEMS
Fig. 9. Genetic activity spectra for the human hematolymphopoietic system carcinogens.
liquid suspension test (RET) and the S. cereoisiae homozygosity assay (YEH). However, the tertiary amine auramine gave negative results in all the Salmonella tester strains, unlike the primary amines, which gave largely positive results. The results in SA5, SA7, HMA and MNM were nega- tive for auramine and one or more of the primary amines.
Dose information was available for 4-amino- biphenyl, benzidine and 2-naphthylamine in 17, 19 and 22 different bioassays, respectively. The struc- ture-activity relationships for these primary aromatic amines have been reviewed (Radomski et
y
PT/GENE MOT.
F" OTHER EFFECTS CHROMOSOMAL EFFECTS TRANS SPERM
++
:z ~z u C
...... i..
c>
I < I . . . ] , . . .
,0 2'+ ,'o ~'~ }0 ' BIOASSAY SYSTEMS
liii+ li i< c3
',~o
A U R A M I N E
5 - (/) I--
LLI U3 o (3 0- ( . 9 o - J
P T / H F H E MUT. OTHER EFFECTS CHROMOSOMAL EFFECTS TRAMS SPERM
I" -'+ "P . . . . . . . . - I - I - I I
-5
F-
ii III ! - r
I . . . . . I . . . I . . . . . . . I . I I I I i I I I I
10 25 40 55 70 85 100
BIOASSAY SYSTEMS
|ENZID INE
PTtHEHEMUT OTHEHEFFECTS C H R ~ A L E F F E C T S THHNS ~ H M
F i g . 12 . G e n e t i c a c t i v i t y s p e c t r a f o r t h e a r o m a t i c a m i n e s .
-5
~ 5
80
.1:;
101
al., 1979). In the genetic activity spectra, 12 tests showed agreement between one or more pairs of the primary amines. The results for one test, the effects on mouse sperm (SPI), were negative. Posi- tive results were obtained in prokaryotic assays (SA8, SA9, SA0 and WPR) and in the urine test (BFU). Primary D N A damage was seen in the prokaryotic system RET, and unscheduled D N A synthesis in higher eukaryotes (UDP). The primary aromatic amines also gave positive results in the mammalian cell transformation systems CTC,
F i g . 11 . G e n e t i c a c t i v i t y s p e c t r u m f o r c y c l o p h o s p h a m i d e .
4 - A M I N O B I P H E N Y L
PTIGEHC MVT OTHER EFFECTS CHHOMOIOMAL EFFECTS THAHS SPERM I1: _ i s - - L _I - _i__I
M ~ E H E ~ T . OTHER EFFECTS CHHOMOIOMAL EFFECTS TflANS SPERM
- ! I - ! I v
IIII g I +< II,lff I ............. [ ............................................... TL ............
I . . . . . I . . . l . . . . . . . I . I l I I I I I I I
10 25 40 55 70 85 100 BIOASSAY SYSTEMS
102
CTK, CTR and CT7. Benzidine and 2-naph- thylamine produced positive results in the Syrian hamster embryo (CTC) and A K R / M E (CTK) as- says at concentrations < 1 /~g/ml. In addition, 4-aminobiphenyl gave positive results at the same concentration in the E. coil gene mutation assay (WPU). As will be shown in the next section, data for the primary aromatic amines were sufficient for pattern-recognit ion techniques to be used to demonstrate similarities in the genetic activity spectra.
Human-liver carcinogens. The chemical struc- tures of the liver carcinogens are shown in Fig. 13. Because the literature search produced no reports of genetic bioassays of oxymetholone, test data were evaluated for only two of the liver carcino- gens, aflatoxin B 1 and vinyl chloride; their genetic activity spectra are shown in Figs. 14 and 5, re- spectively. Vinyl chloride produced positive results in 16 of 34 test systems, whereas aflatoxin B~ gave positive results in 39 of the 43 test systems used.
Aflatoxin B1 and vinyl chloride required meta- bolic activation in many of the test systems; these compounds are carcinogenic in the liver, where metabolic activation reactions are catalyzed. The spectral lines for aflatoxin B~ and vinyl chloride were discordant for the Salmonella tester strains TA1535, TA1537, TA1538 and TA98; the lower eukaryotic test systems NEF and YEH; and the BFU system using a prokaryotic or lower eukaryotic indicator organism.
Positive results for aflatoxin B~ and vinyl chlo- ride were concordant for the higher eukaryotic tests that measure gene mutation in vitro (V7H, V70) or in vivo (SRL, HMA). In addition, results
AFLATOXIN B1 (2) OXYMETHOLONE (2) o o c,R °~
VINYL CHLORIDE (1)
H CI \ / C - C
/ \ H H
(1) IARC Human Carcinogen
Fig. 13. Human liver carcinogens.
(2) IARC ProbabLe
~o
PT/GENE MUT OTHER EFFECTS CHROMOSOMAL EFFECTS TRANS SPERM
Fig. 14. Genetic activity spectrum for aflatoxin Bp
were concordant for eukaryotic tests indicating chromosomal damage: sister-chromatid exchange (SCL, SC3), chromosomal aberrations (CYB) and micronuclei (MNM). Thus, 8 of the higher eukaryotic test systems gave similar results for aflatoxin B 1 and vinyl chloride. Two prokaryotic tests (SAO and REP) and one lower eukaryotic test (YEC) also gave positive results for both chemicals. The high potency of aflatoxin B~ was indicated by responses at doses _< 1 /~g/ml in 19 test systems.
Other known or suspected human carcinogens. Two agents, diethylstilbestrol (DES) and epichlo- rohydrin, could not be assigned to any of the above-mentioned ta rge t -organ systems. DES causes vaginal clear cell adenocarcinoma, and epi- chlorohydrin is unclassified as to its human carcinogenicity (Waters et al., 1983b). The struc- tures and IARC classifications of these com- pounds are shown in Fig. 15, and their genetic activity spectra are shown in Fig. 16.
The spectrum for DES was characterized by a considerable number of both positive and negative responses. The compound was active at concentra- tions < 1 ~ g / m l for unscheduled D N A synthesis in HeLa cells (UDL) and in the RLV/Fischer rat embryo clonal assay (CTR). Positive results were obtained at a dose < 0.1 t~g/ml for sister-chro- matid exchange in normal human fibroblasts (SCF).
Epichlorohydrin produced mainly positive re-
103
DIETHYLSTILBESTROL (1) , ( - - -~ ~,"5 E- -~ ,
- o .
C2H 5
EPICHLOROHYDRIN (3) 0 / \
CII 2 - CHCH2C1
(1) IARC Human Carcinogen Fig. 15. Other human carcinogens.
(3) IARC Unclassified
suits. At < 1 /~g/ml, the compound caused point mutation in Salmonella (SA5) and chromosomal aberrations in human lymphocytes in vitro (CYH). It also produced positive results at < 0.1/~g/ml in the Bacillus subtilis rec assay (REW) and at _< 0.01 ~ g / m l in other systems for chromosomal aberra- tion in lymphocytes (CYL). It should be pointed out that the genetic activity of this gaseous agent depends on partitioning between the gas and aque- ous phases and the use of sealed detection cham- bers (Simmon, 1977; Bridges, 1978). Thus, the genetic activity of the agent may be considerably greater than the values reported in experiments where the gas was uncontained.
Computer analysis of the genetic-activity spectra The results presented in the previous section of
this report reveal some of the complexity and inherent difficulties in comparing and interpreting
the genetic-activity spectra of known or suspected human carcinogens. To a large extent, the problem is that some chemicals have been tested adequately in only a few bioassay systems. Some of the agents have been evaluated in a large number of bioas- says and others in relatively few systems. The more extensively studied chemicals, however, ap- pear to belong to definite subgroups based on their responses in a number of common genetic bioassays. Certain chemical structures may elicit a particular array of molecular damage, correspond- ing to a spectrum of responses in biological assay systems. With a large number of test results, many correlations in responses may be examined using the data analysis techniques described under Methods.
From the computer-generated genetic activity spectra for the 24 chemicals studied, 276 pairwise combinations were obtained. To identify com-
DIETHYLSTI LBESTROL
5- I--
LJJ
0 r~ 0- (.9 0 _I
f'TIGENE MUT.
D
IIiilll......i, ! ...........................
-1 I - I - I I
b n-" F-- " r
o 0~
tO O ~ : 3 ... G.n.
-5 o . . . . . I .... I .... , ,:~, I , I I I I I i I I I
10 25 40 55 70 85 1 O 0
BIOASSAY SYSTEMS Fig. 16. Genetic activity spectra for diethylstilbestrol and epichlorohydrin.
EPICHLOROHYDRIN
PT/gENE MUT OTNEPl EFFECTS CHFIOMQ$OMAL EFFE(TS
m < a . < t r ~ -1- o~ ~ ~
..... i l ....................................
TRAi$ $~RH - i - . 1 . . I ~1 1 - !
........ I L .............. I I . . . . . i . . . . i . . . . . . . i . J I
I I I lb 2; 4b s; 7o 8s 100
BIOASSAY SYSTEMS
104
p o u n d s t h a t d i s p l a y s i m i l a r g e n e t i c ac t iv i ty spec-
t ra , t he d a t a we re e x a m i n e d (1) for all c o m m o n
b i o a s s a y s c o n d u c t e d for e a c h p a i r o f c h e m i c a l s
a n d (2) fo r q u a l i t a t i v e a n d q u a n t i t a t i v e a g r e e m e n t
o f i n d i v i d u a l s p e c t r a l l ines. N o s t a t i s t i ca l ly s ignif i -
c a n t c o r r e l a t i o n was f o u n d in t he s p e c t r a for 9 of
t h e c o m p o u n d s : a s b e s t o s , b e n z e n e , B C M E , ch lo r -
a m p h e n i c o l , c h l o r n a p h a z i n e , C M M E , m u s t a r d gas,
n i c k e l c o m p o u n d s a n d p h e n y t o i n . H o w e v e r , of the
276 p o s s i b l e p a i r i n g s , o n l y 27 p a i r i n g s s h o w e d
a g r e e m e n t w i t h a n u m b e r of tes t s g r e a t e r t h a n t h a t
w h i c h c o u l d b e e x p e c t e d b y c h a n c e (See T a b l e 2).
TABLE 2
COMPARISONS OF GENETIC ACTIVITY PROFILES FOR SELECTED PAIRS OF CHEMICALS
Chemical Number of bioassays a Probability value Total Common Conflicting
a The probability (P) that the number of congruent results would occur by chance is given by the relation n
e = ~ (7)(0.5) ° i ~ n - - m
where 0.5 is the probability of the results matching for any common tests, n is the number of common tests, ra is the number of conflicting tests, and (7) are the binomial coefficients defined as n! / [ (n - i)!i!], n ~ i ~ O, The cutoff significance probability for inclusion of pairings was determined empirically.
F o r the se lec ted c o m b i n a t i o n s , t he n u m b e r o f b io -
a s s ays u s e d in c o m p a r i n g a n y two c h e m i c a l s r a n g e d
f r o m 20 to 75, w i t h a m e a n o f 45. T h e n u m b e r o f
c o m m o n b i o a s s a y s va r i ed f r o m 4 to 29, w i t h a
m e a n of 12. T h e n u m b e r o f c o n f l i c t i n g or d i s c o r d -
a n t c o m m o n tes t s w a s f r e q u e n t l y 0 o r 1, b u t the
m e a n was 2.4. In genera l , t he r a t i o o f c o n c o r d a n t
to c o n f l i c t i n g tes t r esu l t s was such t h a t the p r o b a -
b i l i ty was h igh ly s ign i f i can t t ha t a m a t c h in t he
p ro f i l e s fo r a pa i r o f c o m p o u n d s i n d i c a t e d c o m -
mon genetic activity. The significance level was often < 0.01.
A quantitative comparison of common bioas- says is given by the dose-related function (DRF). A value >__ 2 indicates that the effective or ineffec- tive doses for common bioassays agree within an order of magnitude, or one logarithmic unit. Many of the values for the D R F were approximately 2, indicating substantial quantitative agreement in the responses of the common test systems. The index of dose as a function of the concordant tests (DRFc) was investigated: this function averaged 1.9, with a standard deviation of 0.5, for all com- pounds listed in Table 2. The D R F c values for aflatoxin B 1 were consistently < 1.6, but the val- ues for the other compounds averaged 2.1 (n = 20, S.D. = 0.4). Aflatoxin B 1 was considerably more potent in genetic bioassays than the other com- pounds tested.
By selecting the significant relationships be- tween chemical pairs ( p _< 0.1), 4 groups of com- pounds were identified. In Table 3, the 4 chemical groups are shown with the chemical structures and the number of significant combinations matching each agent with another agent in the particular group. Generally, chemicals paired with only one other chemical were not included in a group. Group I was composed of the primary aromatic amines (2-naphthylamine, 4-aminobiphenyl and benz- idine) and acrylonitrile. Group II was composed of two nitrogen mustards (cyclophosphamide and melphalan), an epoxide (ethylene oxide), and a reactive substituted alkene (vinyl chloride). A link between Groups I and II, designated Group I - I I , was formed by 3 compounds (aflatoxin B 1, epi- chlorohydrin and chromium compounds) whose spectra matched the spectra of members of both Groups I and II. Group III consisted of arsenic, cadmium, auramine and diethylstilbestrol.
The combined concordant spectra for Group I, I - I I , II and III compounds are shown in Figs. 17-20. Each chemical group was active in a well- defined set of bioassays. Concordant test results tended to agree quantitatively, as well as qualita- tively. The mean D R F c value was 2.3 (S.D. = 0.3, n = 17) for compounds in Groups I, II and III. Overall, Group I compounds gave concordant positive results in 12 bioassays and concordant negative results in only 1 bioassay (DRFc = 2.5,
PT/GENE MUT. ,,~ 0THEREFFECTS CHROMOSOMALEFFECTS THANSSPERM - i - - I - _ l _ - I - - I - I - I - - I - - I - - I
==
2 i
I . . . . . 110 215
, . I . . . . . . . I . I I
4'0 s'~ 7'0 s; ,~ BIOASSAY SYSTEMS
Fig. 17. Bioassays with concordant results for two or more Group I compounds.
S.D. = 0.2, n = 4). Bioassay categories yielding concordant positive results included point muta- tion, DNA damage and cell transformation. Group II compounds gave concordant positive responses in 18 bioassays and concordant negative responses in 6 assays (DRF c = 2.0, S.D. = 0.4, n = 6). Con- cordant positive results in bioassays for chro- mosomal effects in mammalian systems were par- ticularly evident. Results for Group I - I I com- pounds displayed concordance similar to that for Group I compounds, with 24 concordant positive
PT/GENE MUT. OTHER EFFECTS CHROMOSOMAL EFFECTS THANS SPERM J_
5
s I . . . . i i . . . . I . . . . . . , / , I I
BIOASSAY SYSTEMS
Fig. 18. Bioassays wi th concordant results for two or more Group I - I ] compounds.
108
PT/OENE MOT, OTHER EFFECTS CHROMOSOMAL EFFECTS TRANS SPERM I - -I~ -I- - I ~ - I - d
-I- - I - - I - - I - - I
i i
< < <
BIOASSAYSYSTEMS
Fig. 19. Bioassays with concordant results for two or more Group II compounds.
results and 1 concordant negative result. However, in contrast to Group I results, concordant positive chromosomal effects data were prominent. The data were fewer for Group III compounds, and fewer positive than negative bioassay results were represented; however, the concordance function, DRF~, was 2.4 (S.D. = 0.1, n = 6).
The most striking individual bioassay results concerned the Salmonella tester strains. Group I and I-II compounds producted frameshift muta- tions in strains TA1538 and TA98 (SA8, SA9) and base-pair substitution mutations in strain TA100
Fig. 20. Bioassays with concordant results for two or more Group III compounds.
(SA0), but not in TA1535 (SA5). Group II com- pounds, on the other hand, did not cause frameshift mutations (SA7, SA8, SA9), but produced base- pair substitution mutations in TA1535 and TA100 (SA5, SA0). Group III compounds produced nega- tive results in all of the Salmonella tester strains. Group I-II compounds aflatoxin B 1 and epichlo- rohydrin produced gene mutation in eukaryotes in vitro (NEF) and in vivo (SRL, HMA). Aflatoxin B~ also gave positive results for gene mutation in bacteria, yeast, fungi and mammalian cells (Fig. 14). Group II compounds caused gene mutation in vitro and in vivo. Two of the Group III com- pounds (arsenic and diethylstilbestrol) gave posi- tive results in the mouse lymphoma cell gene mu- tation system (L5T), but auramine yielded nega- tive results in this system.
Other genotoxic effects were much more promi- nent for the Group ] and I-II compounds, with positive effects displayed in prokaryotic (REW, RET and WPR), lower eukaryotic (YEC, YEH), and higher eukaryotic systems (UDP, UDH and UDS). Group II compounds consistently caused DNA damage in only two prokaryotic systems (REP, RET) and one higher eukaryotic system (DBH). Two chemicals in Group III gave positive results in the prokaryotic system REW, and another pair gave positive results in the lower eukaryotic system YEH.
Group I-II and Group II compounds produced sister-chromatid exchanges and chromosomal aberrations in vitro and in vivo. The Group I chemicals were not extensively tested for chro- mosomal effects. Group I-II compounds gave positive results in the micronucleus test in vitro. Group II compounds produced negative results in the dominant lethal test in Drosophila (DLD) and positive results in the micronucleus tests in the mouse and rat (MNM and MNR) and in the mouse heritable (reciprocal) translocation test (MHT). Results for Group II compounds were negative in the Drosophila aneuploidy assays (DAC, DAP). Group III chromosomal effects as- says yielded mixed results, positive for in vivo chromosomal aberration (CYB) but negative for micronuclei (MNM).
Cell-transformation systems gave concordant results for Group I and Group I-II compounds: positive responses were reported in the assays CTC,
CTK, CTR and CT7 for Group I compounds, and in CTC, CTB, and CT7 for Group I - I I com- pounds. Two chemicals from Group II and another two from Group III gave positive results in CTH and CT7, respectively.
Only one sperm-morphology assay (SPI) was employed consistently in the evaluation of known or suspected human carcinogens. Results in this assay were negative for compounds in Groups I and I II and positive for compounds in Group lII.
Discussion
Although considerable effort has been ex- pended in comparing the performance of various individual bioassay systems and bioassays of carcinogens versus noncarcinogens (Purchase et al., 1978; Rosenkranz and Poirier, 1979; de Serres and Ashby, 1981), other aspects of the data bases formed by testing chemicals for genetic or related effects have not been investigated. Until recently, a large-scale effort had not been made to assemble all of the available bioassay data, although with the advent of the EPA GENE-TOX program (Green and Auletta, 1980; Waters and Auletta, 1981) such an effort is under way. Analysis of the complete genetic toxicology data base is difficult because of the large number of bioassay systems, currently more than 100, and the thousands of chemicals that have been tested or are presently undergoing testing.
The rationale for the present study was to ex- amine a subset of the larger genetic toxicology data base consisting only of agents classified by the IARC as known or suspected human carcino- gens. The analysis of this data base was obviously restricted by the limited number and repetitions of tests that have been performed for some of the compounds. This is particularly evident for several lung carcinogens, BCME, CMME and mustard gas; the hematolymphopoiet ic system group, phenytoin and chloramphenicol; and the human bladder carcinogen chlornaphazine. Some of the other agents have not been sufficiently tested for the major categories of damage to use the genetic data on human agents to the maximum advantage.
Despite obvious limitations, a substantial amount of semiquantitative genetic bioassay data
109
can be derived from the published literature and assembled within a framework that facilitates both visual and computer-assisted assessment. The rapid comparative evaluation of both positive and nega- tive responses according to specified categories of genetic effects (e.g., poin t /gene mutation or chro- mosomal damage) makes it possible to pinpoint similarities and dissimilarities in the overall pro- files of genetic effects displayed by pairs or groups of related compounds. Thus, if a compound is classified within a group of compounds according to target-organ specificity, reactive chemical moiety, or other characteristics, its spectrum of genetic activity may be compared with spectra of other agents in the group. Using this technique, we have demonstrated that a series of common tests were often concordant within a group of target organ-specific chemicals. For example, 7 bioassays gave concordant positive results for chromium and one or more other metal com- pounds causing cancer of the respiratory tract. These results suggest that these 7 common bioas- says may be more effective for evaluating metal respiratory carcinogens than other tests among the 45 bioassays applied to the metal compounds.
The results of the present analysis suggest that because of the chemical specificity of induced genetic damage, it is inappropriate to use a simple arithmetic sum of positive and negative test results in the quantitation of overall genetic activity. A negative result in certain bioassays is often the correct response for a compound that produces positive results in a variety of other test systems. Serious attempts to assess the mutagenic or carcinogenic potential of a given chemical must take into account the overall response pattern, as well as individual test system data.
The groups of chemicals formed by analysis of agent performance in a number of common bioas- says (Table 3) were not based on target-organ specificity. Each group (I, I - I I , II, III) contained agents affecting two or more different target organs. The groups formed by these human carcinogens were correlated with chemical struc- ture and activity; for example, all of the primary aromatic amines were in Group I, and all of the alkylating nitrogen mustards were in Group II. Although chemicals within a group did not present identical spectra of genetic activity, components of
1 1 0
the spectra matched other chemicals within the group. By use of a computer, the data could be rapidly reviewed, and the fine structure of the spectra examined over the entire series of 100 bioassays.
The overall results of the analysis confirmed that Group I compounds, some of the most potent carcinogens, were frameshift mutagens. Three of the compounds in Group I (2-naphthylamine, ben- zidine and 4-aminobiphenyl) and a Group l - I I compound (aflatoxin B1) are among a series of chemicals that have a highly reactive chemical moiety and contain a planar ring system capable of intercalation with DNA. In Salmonella typhi- murium, these chemicals produced concordant positive results in strain TA1538 but were not consistently frameshift mutagens for TA1537, in- dicating a preference for the C G C G C G C G base sequence (Ames et al., 1973). Two of the com- pounds, benzidine and 2-naphthylamine, gave negative results in TA1537, but aflatoxin B~ and 4-aminobiphenyl gave positive results. All 4 com- pounds were mutagenic in the more sensitive TA100 strain, indicating their ability to cause either base-pair substitutions or frameshift mutations. Another Group I - I I compound, chromium, was also a frameshift mutagen, and the profile of genetic activity is correlated with that of aflatoxin B1; however, concordant test results differed by 1.5 logarithmic dose units. Although acrylonitrile (Group I) is structurally similar to vinyl chloride (Group II), and both are lung carcinogens, the genetic activity of acrylonitrile was more closely related to that of epichlorohydrin (Group I-II) . A projected epoxide of acrylonitrile (Milvy and Wolff, 1977), oxiranecarbonitrile
o / \ H 2 C - C - C = N
I H
is very similar in structure to epichlorohydrin. The Group II compounds are well-known al-
kylating agents. In addition to the nitrogen mustards, Group II contained ethylene oxide and vinyl chloride. Based on the data available (Fig. 19), Group II compounds did not cause frameshift mutation in Salmonella (SA7, SA8, SA9). They did, however, cause base-pair substitution in Salmonella. Because chlorinated ethylenes are
metabolized to chlorinated epoxides (Henschler, 1977), the vinyl chloride intermediate 2-chloroeth- ylene oxide,
would be expected to react similarly to ethylene oxide.
C - - C
Indeed, chloroethylene oxide was shown to be the most effective of the vinyl chloride metabolites in inducing gene mutation and gene conversion in yeast (Loprieno et al., 1977). Chloroacetaldehyde, also a metabolite of vinyl chloride, has been sug- gested as the primary mutagenic metabolite of cyclophosphamide (Whitehouse et al., 1974); how- ever, cyclophosphamide, in contrast to chloro- acetaldehyde, shows no preference for the TA100 over the TA1535 tester strain (McCann et al., 1975).
The genetic activity spectrum of cyclophospha- mide matched equally well with compounds from Groups I - I I and II. Because of its lack of concor- dance with Group I compounds, cyclophospha- mide was considered a member of Group II. The Group I - I I compounds correlated well with com- pounds from both Groups I and 1I; apparently, Group I - I I is a subset of the major groups (I and II). It is noteworthy that chlornaphazine contains both the naphthalene ring found in 2-naphthyla- mine (Group I) and the bis-(2-chloroethyl) reactive group of cyclophosphamide (Group II). With fur- ther genetic bioassay testing of chlornaphazine, it may be predicted that the compound will fit in Group I - I I or one of the major groups (I or 1I).
The Group III compounds did not cause base- pair substitution or frameshift mutation in Salmonella and may act by a mechanism other than gene mutation. Auramine and diethylstil- bestrol have somewhat similar overall chemical structures and displayed similar genetic activity. Auramine may he an example of an intercalating agent that produces frameshift mutation in sys- tems other than Salmonella (Rosenkranz and Poirier, 1979).
It is widely recognized that chemicals produce an array of potentially damaging effects on genetic material. As the amount of bioassay data has increased, it has become possible to test the hy- pothesis that similar chemical compounds or those possessing specific reactive groups will exhibit a similar spectrum of genetic activity. We have found by an analysis of quantitative genetic bioassay data on 24 known or suspected human carcinogens that (1) chemically similar compounds produce similar spectra of genetic activity, (2) certain groups of test systems are more appropriate than others for some classes of chemical compounds, and (3) certain groups of compounds can be recognized by a well-defined series of concordant test results. As the data base concerning human carcinogens is enlarged, interrelationships among chemicals and their effects derived from analysis of the spectrum of genetic activity may supply useful information concerning the ultimate structural damage to the genome.
Acknowledgments
The authors wish to acknowledge the construc- tive comments of Drs. L. Claxton, D. De Marini, D. Jansen, M. Moore, B. Most, S. Nesnow, D. Neubert, A. Stead, J. Wahrendorf and E. Zeiger.
References
Ames, B.N., W.E. Durston, E. Yamaski and F.D. Lee (1973) Carcinogens are mutagens: A simple test system combining liver homogenates for activation and bacteria for detection, Proc. Natl. Acad. Sci. (U.S.A.), 70, 2281-2285.
Bridges, B.A. (1978) On the detection of volatile liquid muta- gens with bacteria: Experiments with dichlorvos and epi- chlorhydrin, Mutat ion Res., 54, 367-371.
Chan, P.K., G~P. O 'Hara and A.W. Hayes (1982) Principles and methods for acute and subchronic toxicity, in: A.W. Hayes (Ed.), Principles and Methods of Toxicology, Raven, New York.
de Serres, F.J., and J. Ashby (Eds.) (1981) Evaluation of Short-Term Tests for Carcinogens, Elsevier/North-Hol- land, New York.
Green, S., and A. Auletta (1980) Editorial introduction to the reports of 'The Gene-Tox Program', An evaluation of bio- assays in genetic toxicology, Mutation Res., 76, 165-168.
Henschler, D. (1977) Metabolism and mutagenicity of
halogenated olefins - - a comparison of structure and activ- ity, Environ. Hlth. Perspect., 21, 61-64.
IARC (1979) Chemicals and Industrial Processes Associated with Cancer in Humans, Suppl. 1 to IARC Monographs, Vol. 1-20, International Agency for Research on Cancer, Lyon.
Loprieno, N., R. Barale, S. Baroncelli, H. Bartsch, G. Bronzetti, A. Cammellini, C. Corsi, D. Frezza, R. Nieri, C. Leporini, D. Rosellini and A.M. Rossi (1977) Induction of gene mutations and gene conversions by vinyl chloride metabo- lites in yeast, Cancer Res., 36, 253-257.
McCann, J., V. Simmon, D. Streitwieser and B.N. Ames (1975) Mutagenicity of chloroacetaldehyde, a possible metabolic product of 1,2-dichloroethane (ethylene dichloride), chloro- ethanol (ethylene chlorohydrin), vinyl chloride, and cyclophosphamide, Proc. Natl. Acad. Sci. (U.S.A.), 72, 3190-3193.
Milvy, P., and M. Wolff (1977) Mutagenic studies with acrylonitrile, Mutation Res., 48, 271-278.
Purchase, I.F.H., E. Longstaff, J. Ashby, J.A. Styles, D. Ander- son, P.A. Lefevre and F.R. Westwood (1978) An evaluation of 6 short-term tests for detecting organic chemical carcinogens, Br. J. Cancer, 37, 873-903.
Radomski, J.L., W.L. Hearn and T. Radomski (1979) Struc- ture-activity relationships amongst the primary aromatic amines in the induction of bladder cancer, in: W.B. Deich- mann (Ed.), Toxicology and Occupational Medicine, Else- vier /North-Holland, New York.
Rosenkranz, H.S., and L.A. Poirier (1979) Evaluation of the mutagenicity and DNA-modifying activity of carcinogens and noncarcinogens in microbial systems, J. Natl. Cancer Inst., 62, 873-892.
Simmon, V.F. (1977) Structural correlations of carcinogenic and mutagenic alkyl halides, in: I.M. Asher and C. Zervos (Eds.), Structural Correlates of Carcinogenesis and Muta- genesis, FDA Office of Science, U.S.A., pp. 163-171.
Waters, M.D., and A. Auletta (1981) The Gene-Tox program: Genetic activity evaluation, J. Chem. Inf. Comput. Sci., 21, 35-38.
Waters, M.D., N.E. Garrett, C.M. Covone-de Serres, B.E. Howard and H.F. Stack (1983a) A review and evaluation of human genetic bioassay data for some known or suspected human carcinogens, in: A. Castellani (Ed.), The Use of Human Cells for the Evaluation of Risk from Physical and Chemical Agents, Plenum, New York, pp. 425-465.
Waters, M.D., N.E. Garrett, C.M. Covone-de Serres, B.E. Howard and H.F. Stack (1983b) Genetic toxicology of some known or suspected human carcinogens, in: F.J. de Serres and A. Hollaender (Eds.), Chemical Mutagens: Principles and Methods for Their Detection, Vol. 8, Plenum, New York, pp. 261-341.
Whitehouse, M.W., F.W. Beck, M.M. Droge et al. (1974) Lymphocyte deactivation by (potential immunosupressant) alkylating metabolites of cyclophosphamide, Agents Ac- tions, 4, 117-124.
