Document 3nLoe6v4dR9JzrxOq7yyv7KO

Br. ./. Cancer (1373) 28. 173 - [ PLAINTIFF'S EXHIBIT DOW-1670 ST0045337 MESOTHELIOMATA IN RATS AFTER INOCULATION WITH ASBESTOS AND OTHER MATERIALS J. C. WAGNER. G. BERRY and V. TIMBRELL From the Medical Research Council's Pneumoconiosis Unit, Llandough Hospital. Penarth. Glamorgan Received 21 March 1973. Accepted 10 April 1973 Summary.--Four experiments in which SPF Wistar rats were inoculated intrapleurally with asbestos or other materials are described. Mesotheliomata were observed in a considerable proportion of animals with all the samples of asbestos used and with a sample of brucite. A few were produced with synthetic aluminium silicate fibres and single ones with barium sulphate, glass powder and aluminium oxide. The risk of developing a mesothelioma at a given time after injection was approximately proportional to the dose. Of the UICC standard reference samples, crocidolite was the most carcinogenic and removal of the oils by benzene extraction did not alter the carcinogenicity of these samples. Chemical properties also seem unlikely to be the main factor producing mesotheliomata but the results support the hypothesis that the finer fibres are the more carcinogenic, and this is additional to the known aerodynamic advantage which the finer fibres have in penetrating to the periphery of the lung. We report here the results of 4 experi ments in which asbestos and other test materials were administered to rats by intrapleural inoculation. These experi ments were planned to obtain more infor mation on the carcinogenic effect of asbestos and other materials than could be obtained from our original 2 experi ments (Wagner and Berry, 1969). Pre liminary results of some of the present experiments were given by Wagner, Berry and Timbrell (1970), Wagner (1970, 1972). In this paper the complete results are given, with emphasis on the light they throw on the aetiology of mesotheliomata, taking into account the oils and waxes present in asbestos, other chemical charac teristics and the physical characteristics. MATERIALS AND METHODS In all 4 experiments specific pathogen-free (SPF) rats, of the Wistar strain were used. These rats had been bred at the Unit from stocks given to us by Imperial Chemical Industries. Pharmaceutical Division at Alderley Edge. Cheshire in 1964 and 1968. The following materials were used: 1. SFA chrysotile.--A super fine sample obtained from a Canadian mine, and pro duced bv water sedimentation separation from grade 7, the most fully milled commer cial product. 2. Crocidolite.--Prepared from virgin fibre from a mine in the North West Cape. Both (1) and (2) were from the same samples as used in the earlier experiments (Wagner and Berry, 1969). 3. UICC Standard reference samples.-- Samples of amosite. anthophvllite. Canadian chrysotile, Rhodesian chrysotile and croci dolite (Timbrell, Gilson and Webster. 1968) prepared following recommendations of V Union Internationale Contre le Cancer (UICC). 4. Benzene-extracted UICC Standard reference samples.--Samples of (3) which had been repeatedly extracted for 64 hours by hot benzene using a Soxhlet apparatus to remove oils and other benzene-soluble sub stances. After extraction the benzene was 174 J. C. WAGNER, G. BERRY AND V. TIMBRELL ST0045338 first allowed to evaporate naturally and There were about 12 rats per dose per dust finally the samples were warmed to 80 :C for and inoculation was during March 1965. 24 hours to remove any remaining benzene. After this treatment the samples were tested for the presence of any residual benzene by extracting test portions with cyclohexane and examining the solutions by means of ultraviolet spectrophotometry; no benzene was detected in these solutions. 5. Canadian chrysotiles.--Samples from Experiment 2.--Canadian chrysotiles The experimental materials were 7 of the 8 Canadian chrysotile samples. SFA chrysotile and saline control. The dose was 20 mg per rat. There were 16 rats for each Canadian sample. 32 for SFA chrysotile and 48 controls and moculation was during December 1966. 8 mines (A, B, . . . H) in Canada. These were the same samples used to prepare the UICC standard reference sample of Canadian chrvsotile (Timbrell and Rendall. 1971) but were milled for our purpose more finely than the reference sample. 6. Brucite.--A specimen of brucite, which however also contained chrvsotile. This specimen was from Canadian mine H and consisted of long coarse brownish fibres above 50 cm in length. The sample was milled to respirable particle size. Experiment 3--UICC samples and Canadian chrysotiles The materials used were the 5 UICC reference samples both in the normal and oil-free forms, the 8 Canadian chrvsotile samples, brucite and barium sulphate and saline controls. The dose was 20 mg per rat. There were 24 rats for each of the Canadian samples and 32 for each of the other treat ments. Inoculation took place between November 1967 and February 1968. 7. Barium sulphate.--Used as a control. Experiment 4.-- Various dusts This was prepared in the laboratory by the addition of sulphuric acid to barium chloride solution. 8. Saline.--Sterile physiological saline was also used as a control. 9. Ceramic fibre.--A synthetic aluminium silicate, fibre. This fibre was prepared for experimental use by grinding in a ceramic ball mill and extracting the respirable fraction by settlement in air. The fibre diameters were between 0-5 and 1 /im. 10. Fibreglass.--A borosilicate. The nom inal diameters of the fibres were between 1-5 and 2-5 ^m but in fact only 30% were within this range, the range extending to 7 pm. The sample was prepared by em bedding the fibres in w'ater soluble wax. chopping in a microtome and washing away the wax. Over 60% of the fibres were longer than 20 fim. 11. Glass powder.--A borosilicate all in the respirable range (less than 8 /an pro The materials injected were ceramic fibrefibreglass, glass powder, aluminium oxide. SFA chrysotile and also the second sample of SFA chrysotile. The dose was 20 mg per rat and there were up to 36 rats per treatment (because of a shortage of animals it, was not possible to allocate 36 to all treatments and in addition inoculation fatalities could not be replaced). Inoculation took place in June and July 1969. For each experiment animals were allo cated at random to treatments. The age of the rats at inoculation was about 6 weeks for Experiments 1, and 3 and 13 weeks for Experiments 2 and 4. In Experiments 1 and 3 there were equal numbers of male and female rats, whilst in Experiment 2 there were 3 times as many females as males, and in Experiment 4 there were twice as many males as females. Methods jected area diameter). The experimental materials w ere made up 12. Aluminium oxide.--A non-fibrous mat in a suspension of physiological saline with a erial all in the respirable range (less than concentration of 50 mg/ml for Experiments 10 (um projected area diameter). 2, 3 and 4 and for Experiment l the con 13. SFA chrysotile (Second sample).--A centration was such that the required dose sample from the same mine and prepared would be present in 0-4 ml of suspension. similarly to (1). but taken several years later. The rats were anaesthetized with ether and a Experiment 1 -- Varying dose needle attached to a two-way tap was then introduced into the right axilla at the level of There were 5 doses. 0-5. 1. 2. 4 and 8 mg the second nipple. One arm of the two-way per rat, of SFA chrvsotile and crocidolite. tap was attached to a capillary manometer, MESOTHELIOMATA IN' RATS AFTER INOCULATION' WITH ASBESTOS 1 To G c e s 'io o is which gave a negative reading when the logical examination was not possible, needle reached the pleural cavity, Details of the method of inoculation were given by Wagner and Berry (1969). Following injec tion the rats a. caged in fours isolated in a special unit. They were fed on a proprietary brand of autoclaved cubes, and water ad libitum. Each rat was allowed to live until it died or appeared to be distressed and a full necropsy examination was carried out, except for a few which had been cannibalized. leaving 1112 rats included in the results. The predominant finding was that a high proportion of most asbestos treated groups developed mesotheliomata and the results will be given mainly in terms of the number of mesotheliomata and the time when they occurred. Some details of the results are given in Tables I-IV. A total of 386 mesotheliomata occurred but The results have been analysed using the the histological features of these tumours model given by Pike (1966) and shown to be valid for experiments of this type (Berry and Wagner. 1969). Fuller details are given in in the appendix, and it need only be noted that this method of analysis allows a constant c to be estimated for each of the treatment groups and that this constant, which we will refer to as the " carcinogenicity factor ". serves as a single index summarizing the mesothelioma experience of each group. It combines the will not be described here as there is nothing to add to the features described for the original experiments (Wagner and Berry. 1969). Also, in the presentation and analysis of the results, no account has been taken of the sex of the rats. The original experiments show males and females equally likely to develop a meso thelioma. and the present experiments information on the proportion of animals confirm this. which developed a mesothelioma with the times after inoculation at which the meso- Experiment 1 theliomata occurred. Also, since the method There is a relationship between the of estimation eliminates mortality due to other causes, chance variations in natural mortality between different treatment groups do not affect the treatment comparisons, nor do systematic differences in natural mortality between different experiments, such as that resulting from the animals in Experiments 2 and 4 being older than those in Experiments 1 and 3. affect comparisons between experi number of mesotheliomata and the dose for both SFA chrysotile and crocidolite. This implies that the carcinogenicity is related to dose (d) and we considered this relationship in the form of the carcino genicity factor being proportional to a power of dose. i.e. c = 6rf,) where b and p are constants. The power p was estimated ments. separately for each dust, giving 0-73 for Where significance levels are quoted they chrysotile and 0-96 for crocidolite. Be are usually based on the cAi-squared approxi cause of the small number of mesothe mation to a likelihood-ratio test. liomata. however, these estimates are not very precise and the approximate 950 RESULTS limits are 0-3-1-3 and 0-2-1-9 for chry There were 13 rats for which histo sotile and crocidolite respectively. There SKA ohry-^nle SFA chrysotile SFA chrvsotile SFA chrysotile SFA chrysotile Crocidolite Crocidolite Crocidolite Crocidolite Crocidolite ** 5 mg l mg 2 mg 4 mg 8 mg 9 o mg l mg 2 mg 4 mg 8 mg Table I.--Experiment 1 Results Number of rats Number with with histology a mesothelioma Survival time (days) of first mesothelioma 1:2 1 512 11 3 61") l '2 5 425 1 '2 4 479 1 2 8 496 1 l l 992 12 0 -- 12 .3 562 13 917 l! 5 799 Mean survival (days) 784 729 664 762 692 899 76n 777 819 689 17 b Canadian chrysotile Canadian chrysotile Canadian chrysotile Canadian chrysotile Canadian chrysotile Canadian chrysotile Canadian chrysotile SFA chrysotile Control A B C D E F H C. WAGN'ER. G. BERRY AND V. TIMBRELL Table II.--Experiment 2 Results Number of rats with histology 16 16 16 16 16 16 16 32 48 Number with a mesothelioma 8 10 5 10 7 10 4 22 0 Survival time (days) of first mesothelioma 416 416 488 461 405 421 384 376 -- Mean survival (d 642 594 702 624 373 619 602 553 728 0 {iC ^lQ 01S Table III.--Experiment 3 Results Material ClCC samples Amosite Amosite benzene-extracted Anthophyllite Anthophylhte benzene extracted Chrysotile (Canadian) Chrysotile (Canadian) benzene-extracted Chrvsotile (Rhodesian) Chrysotile (Rhodesian) benzene-extracted Croctdolite Crocidohte benzene-extracted Canadian chrysotile A Canadian chrysotile B Canadian chrysotile C Canadian chrysotile D Canadian chrysotile E Canadian chrysotile F Canadian chrysotile G Canadian chrysotile H Brucite Barium sulphate Saline control Number of rats with histology 32 32 32 32 32 32 31 32 32 30 24 22 24 23 24 24 23 23 32 30 32 Number u ith a mesothelioma Survival time (days) of first mesothelioma 12 377 11 590 6 498 14 533 10 541 9 632 7 502 5 659 19 586 19 468 14 488 9 437 9 460 12 534 9 489 13 484 16 576 14 429 18 502 1 436 0-- Mean survival (days) 716 718 761 728 747 753 693 686 682 657 712 636 717 669 660 675 659 663 680 783 818 Ceramic fibre Fibreglass Glass powder Aluminium oxide SFA chrysotile SFA chrysotile (2nd sample) Table IV.--Experiment 4 Results Number of rats with histology 31 35 35 35 36 32 Number with a mesothelioma 3 0 1 1 23 21 Survival time (day's) of first mesothelioma 743 -- 516 646 325 382 Mean survival (d 736 774 751 710 568 639 are some theoretical grounds for choosing p to be an integer and therefore p was taken as unity. The values of the carcinogenicity factor adjusted to a dose of 20 mg were then 4-10 X 10~9 for chrysotile and 1-70 x 10~9 for crocidolite. Experiments 2, 3 and 4 Comparing first the effects of the separate Canadian samples (Table V), there is considerable variation between Experiments 2 and 3 and this is mainly because of the small number of animals in MESOTHELIOMATA IN RATS AFTER INOCULATION WITH ASBESTOS 177 I 'lES'iOOlS Table V.--Estimates of Carcinogenicity others in order amosite, anthophyllite, Factor (x 109) for Experiments 2, 3 and 4 Canadian chrysotile and Rhodesian chry Experiment sotile. The difference between the two samples of chrysotile is not significant. 2 3 4 Holmes et al. (1971) also carried out Canadian chrysotile A Canadian chrysotile B Canadian chrysotile C Canadian chrysotile D Canadian chrysotile E Canadian chrysotile F Canadian chrysotile C Canadian chrysotile H SFA chrysotile SFA chrysotile (2nd sample) Brucite Barium sulphate Ceramic fibre Glass powder Aluminium oxide 1 25 2 23 0-68 2 39 I - 89 l 90 -- I 10 4 72 -- -- -- -- -- 1 20 1 16 0-67 l 23 0-97 l 40 2 09 1 84 -- -- 121 0 04 -- -- -- ___ -- -- -- -- -- -- -- 2-85 2-28 -- -- 0 16 0 04 0 05 chemical analyses on the UICC reference samples. There were very large differ ences between the samples in the amount of the different metals present and it is clear that these bear no relation to the carcinogenicity. The sample of brucite proved as carcinogenic as the Canadian samples of chrysotile (Table V). Non-asbestos materials which produced the occasional mesothelioma were ceramic fibre, barium sulphate, glass powder and aluminium each group. There was overall about oxide. 30 % more carcinogenicity in Experiment The second sample of SFA chrysotile 2 than in Experiment 3 but the difference proved similar in carcinogenic effect to is not significant (P > 0-1). Sample C the original sample. has the lowest carcinogenicity in each experiment and overall is significantly the DISCUSSION least carcinogenic (P < 0-05) but apart The application of the test materials from this no differences between the by intrapleural inoculation may be criti samples were detected. These samples cized as unrealistic in comparison with have been analysed for certain metals human exposure, about which the animal (Holmes, Morgan and Sandalls, 1971; experiments are intended to provide Morgan and Crallev, 1973) and in Table relevant information. Nevertheless, this VI the results of these analyses are shown, type of experiment has an important together with the carcinogenicity factor part to play. With an inhalation experi obtained by combining the 2 experiments. ment, which provides a realistic route of The correlation coefficients between the entry of the test material, there are 2 carcinogenicity factor and the different factors involved. First, the penetration metals are --0-13 for iron, --0-58 for of dust through the airways and alveoli chromium, 0-04 for cobalt, --0-02 for will differ with different samples of dust nickel, --0-39 for scandium and --0-04 (Timbrell, 1965). The second factor is the for manganese. None of these is signi effect of the dust, given that it has reached ficant and it is reasonably clear from the pleura. In inoculation experiments examination of the chemical properties of only the second factor is relevant and sample C that the low carcinogenicity of hence these experiments are simpler to this sample is not because of a low content interpret. This makes intrapleural inocu of any of these metals. lation a more suitable method for the Turning now to the UICC reference investigation of questions such as whether samples (Table VII), there are no signi extraction of the oils alters the carcino ficant differences between the normal and genicity of an asbestos sample. The two benzene-extracted samples, and overall types of experiment supplement one 58 mesotheliomata occurred with the another and we will be reporting separ benzene-extracted samples and 56 with ately on 2 experiments in which rats the untreated samples. Crocidolite was were exposed to dust clouds of the UICC the most carcinogenic sample with the reference samples. 178 J. C. WAGNER. G. BERRY AND V. TIMBRELL znesnoois The varying dose experiment gave that all the types of asbestos, having verv results which indicated that the risk of different chemical compositions, produce developing a mesothelioma at a given mesotheliomata makes it unlikelv that the time after injection was proportional to carcinogenicity of asbestos could be due the dose. This form of dose relationship to chemical properties. was also found by Pike and Doll (1965) for Our experiments offer some evidence lung cancer and smoking in man whereas that the development of mesotheliomata Lee and O'Neill (1971) showed that after is associated with the presence of fine repeated applications of benzopyrene to fibrous material within the pleural cavity. the backs of mice the incidence rate of First, UICC Canadian chrysotile is a tumours was proportional to the square mixture of batches of material from 8 of the dose. Canadian mines, and the separate samples The carcinogenicity of the SFA chry- used were taken from the same batches sotile sample was similar in Experiments 1 (Timbrell and Rendall, 1971). The main and 2, after adjusting the former to a dose difference in the subsequent preparation of 20 mg, and in Experiment 4 was lower of the material was that the separate but not significantly so. In all these 3 Canadian samples were ground more experiments the carcinogenicity of the finely than the composite UIC'C sample. SFA chrysotile was significantly greater Comparing Tables V and VII, the carcino than in the earlier experiment (Wagner genicities of all the separate Canadian and Berry, 1969) when the estimate of the samples were greater than that of the carcinogenicity factor was 1-68 x 10~9. UICC' Canadian chrysotile. Also, the The crocidolite was also more carcinogenic samples of SFA chrysotile were from mine in Experiment 1 than in the earlier experi D. These were superfine samples and ment (c = 1-16 x 10-9) but not signi resulted in a very high carcinogenicity. It ficantly so. These differences could be should be noted that of the Canadian the result of a change in susceptibility of samples the one with the lowest carcino the rats or of a change in the dust during genicity (C) in both experiments was from storage. a mine in British Columbia whilst the The suggestion that natural oils and others were from 7 mines in the Quebec waxes (Harington, 1962), or contaminating area. However, sample C could not be oils from the preparation of the fibre distinguished from the other samples by (Harington and Roe, 1965; Roe, Walters its size distribution. and Harington, 1966) or from plastic A full quantitative analysis of our storage bags (C'ommins and Gibbs, 1969) experimental results will only be possible might contribute to the carcinogenicity of when techniques are available for com asbestos receives no support from our plete size characterization of the experi present experiments, which is in agree mental materials, both before injection ment with our original experiments and present in the lungs at postmortem. (Wagner and Berry, 1969) when removal Such techniques to determine the mass of the oils from the crocidolite sample and the diameter and length distributions resulted in no detectable change in of the particles are being developed. carcinogenicity. However, even with the characterization Harington and Roe (1965) also ad methods at present available, a relation vanced the possibility that the presence of ship emerges between the observed in trace metals might be relevant to the cidence of mesotheliomata and the physical carcinogenicity of asbestos. In our experi factors. ments with the Canadian samples the A further factor that must be taken carcinogenicity was not related to the into account is the tendency of chrysotile content of iron, chromium, cobalt, nickel, fibres to fragment longitudinally into fine scandium or manganese. Also, the fact fibrils in lung fluids, the degree of frag- MESOTHELIOMATA IN RATS AFTER INOCULATION' WITH ASBESTOS 179 Table VI.--Carcinogenicity and Chemical Analysis of Canadian Samples-- Experiments 2 and 3 Combined Sample G D F B H E A e Carcinogenicity factor ( < I0) 2-37 1 -70 1 60 1 59 1 54 1 33 1 22 11-67 Iron (%) 19 40 29 34 32 32 48 2-0 Chromium parts/10* 380 930 730 780 480 435 515 1200 Cobalt parts; 10* 53 78 78 110 42 57 63 60 Nickel parts/10* 1400 1550 1900 2400 550 695 1150 1800 Scandium parts/10* 75 6-6 77 41 56 54 50 12 0 Manganese parts/106 420 600 450 580 530 610 540 420 ST0045343 Table VII.--Estimates of Carcinogenicity Factor (xlO9) for vice Reference Samples in Experiment 3 Amosite Anthophyllite Chrvsotiie (Canadian) Chrvsotiie (Rhodesian) Crocidolite Normal form 0 66 0 33 0 50 0-44 l -45 Benzene extracted form 0-66 0-77 0 42 0 31 l 87 mentation and hence the number of fibres and fibrils produced depending on the precise physical and physiological conditions. Amphibole types of asbestos, on the other hand, have characteristic fibre-diameter distributions which they appear to retain in lung tissue (Timbrell, Pooley and Wagner, 1970). To illustrate this relationship, the electron micrographs of some of the materials are presented in decreasing order of their carcinogenicity (Fig. 1-8). For reasons given by Timbrell (1973), we shall consider as " significant " fibres those that are less than 0-5 /an in diameter and also greater than 10 pm in length. The non-chrvsotile materials will be con sidered first. In the electron micrographs for UICC crocidolite (Fig. 2), UICC amosite (Fig. 4), UICC anthophyllite (Fig. 5), ceramic fibre (Fig. 7) and glass fibre (Fig. 8) it is evident that the number of " significant " fibres decreases with decreasing carcinogenicity of the materials. The glass fibre, for instance, contains long fibres but the majority of these are thicker than 0-5 pin. On the other hand, whereas a high proportion by weight of the brucite consists of large fibres, there are also present a number of very fine long fibrils. It is difficult to compare chrvsotiie samples (Fig. 1 and 6) with other types of material on this basis. But even so, the relative positions of the chrysotiles in the classifi cation by carcinogenicity appears to correspond with the number of ' signifi cant " fibres present. For example, al though the SFA chrvsotiie (Fig. 1) contains a high proportion by weight of non-fibrous particles, even before injection the fibres were in a highly dispersed state. The enormous number of fibres that complete fragmentation of chrvsotiie can produce will be clear from the illustration that a single fibre may fragment into 1000 fibrils. The fact that this SFA sample was the most carcinogenic of all the materials used corresponds to its highly dispersed state and its high content of significant " fibres. The above theory has been examined further using the results of Stanton and Wrench (1972). Their experiments were similar to ours and they used 17 samples, including several materials (UICC samples and glass fibres) after partial pulveriza tion. They analysed their results bv discounting submicroscopic fibrils and converting all longer fibres into micro fibres of standard size (1-25 x 3-75 /an) on the assumption that fragmentation of both glass and asbestos occurred in vivo. The numbers of microfibres were then compared with the carcinogenicity of the materials. At first sight their results seem to conflict with our findings. But, MESOTHELIOMATA IN RATS AFTER INOCULATION WITH ASBESTOS 181 182 J. C. WAGN'ER. G. BERRY AND V. TIMBRELL MESOTHELIOMATA IX RATS AFTER INOCULATION WITH ASBESTOS 183 13 184 J. C. WAGNER. G. BERRY AND V. TIMBRELL ST 0045348 when their materials were assessed in the REFERENCES manner used for our own samples, the order obtained by classification of the materials according to the number of " significant " fibres was in good agree ment with the reported order of carcino genicity. The smaller number of mesotheliomata observed from pulverized materials did not correlate with the estimated total number of particles that this pulverization would produce, but did correlate with the estimated number of Berry. G. A Wagner. J. C. (1969) The Application of a Mathematical Model Describing the Times of Occurrence of Mesotheliomas in Rata following Inoculation with Asbestos. Br.J. Cancer, 23 58 Commins, B. T. A Gibbs. G. W. (1909) Contamina ting Organic Material in Asbestos. Br. J. Cancer 23, 358. Harington, J. S. (1962) Occurrence of Oils Con taining 3 : 4-Benzpyrene and Related Sub stances in Asbestos. S'ature, Lond., 193, 43. Harington, J. S. A Roe, F. J. C. (1965) Studies of Carcinogenesis of Asbestos Fibres and their Natural Oils. Ann. N.Y. Acad. Set., 132. 439. Holmes, A., Morgan. A. A Sandalls, F. J. (1971) Determination of Iron, Chromium, Cobalt, Nickel " significant " fibres. Stanton and Wrench (1972) also pro duced mesotheliomata with very fine fibreglass. Mesotheliomata were not pro duced by our sample of fibreglass but they were by synthetic aluminium silicate fibre, which was finer than our glass fibre (Fig. 7 and 8). Again, the apparent con tradiction is explicable in terms of phy sical characterizations. The 2 samples of fibreglass can only be compared using light microscope size data; in our sample 55% of the fibres had diameters exceeding 2-5 fim compared with less than 10% of and Scandium in Asbestos by Neutron Activation Analysis. Am. industr. Hyg. Ass. J., 32, 281. Lee, P. N. A O'Neill, J. A. (1971) The Effect of Time and Dose Applied on Tumour Incidence Rate in Benzopyrene Skin Painting Experiments. Br. J. Cancer, 25, 759. Morgan, A. A Cralley, L. J. (1973) Chemical Characteristics of Asbestos and Associated Trace Elements. In Proc. Conf. Biological Effects of Asbestos. Lyon, 2-5 October, 1972. In print. Pike, M. C. (1966) A Method of Analysis of a Certain Class of Experiments in Carcinogenesis. Bio metrics, 22, 142. Pike, M. C. & Doll, R. (1965) Age at. Onset of Lung Cancer: Significance in Relation to Effect of Smoking. Lancet, i, 665. Roe, F. J. C., Walters, M. A. <fe Harington, J. S. (1966) Tumour Initiation by Natural and Con taminating Asbestos Oils. Int. J. Cancer, 1, 49L. Stanton and Wrench's sample. Although a direct association between physical factors and the development of mesotheliomata has not been demon strated, these characteristics appear to be the relevant properties. If the finer fibres are the more carcinogenic when applied to the pleura then, since the finer fibres are also able to penetrate to the pleura more easily after inhalation, these 2 factors would combine together to give the finer fibres more relative importance than even the aerodynamic differences would suggest. Stivton, M. F. <fe Wrench, C. (1972) Mechanisms of Mesothelioma Induction with Asbestos and Fibrous Glass. J. natn. Cancer Inst., 48. 797. Timbrell, V. (1965) The Inhalation of Fibrous Dusts. Ann. X.Y. Acad. Set., 132, 255. Timbrell, V. (1973) Physical Factors as Aetiological Mechanisms. In Proc. Conf. Biological Effects of Asbestos. Lyon. 2-5 October, 1972. In print. Timbrell, V., Gilson, J. C. <fc Webster. I. (1968) UICC Standard Reference Samples of Asbestos. Int. J. Cancer, 3. 406. Timbrell, V., Pooley, F. & Wagner, J. C. (1970) Characteristics of Respirable Asbestos Fibres. In Pneumoconiosis. Proc. Intemat. Conf. Johan nesburg, 1969. Ed. H. A. Shapiro. Cape Town: Oxford University Press, p. 120. Timbrell. V. <fc Rendall, R. E. G. (1971) Prepara tion of the UICC Standard Reference Samples of Asbestos. Powder Technol., 5, 279. Clearly, the experiments described in this paper have involved a large amount of daily effort over a number of years and we are grateful to all our colleagues who have contributed to this. We are also grateful to Dr B. T. Commins of the Wagner, J. C. (1970) The Pathogenesis of Tumours following the Intrapleural Injection of Asbestos and Silica. In Morphology of Experimental Respiratory Carcinogenesis. Proc. Conf. Gatlinburg 13-16 May 1970. Ed. P. Xetteaheim. M. G. Hanna Jr. and J. W. Deatherage Jr. U.S. Atomic Energy Commission Symposium Series No. 21. p. 347. Wagner, J. C. (1972) The Significance of Asbestos in Tissue. In Recent Results in Cancer Research, MRC Air Pollution Unit who prepared the benzene-extracted samples and the barium sulphate sample. Vol. 39. Current Problems in the Epidemiology of Cancer and Lymphomas. Ed. E. Grundmann and H. Tuhnius. Berlin: Springer-Verlag. p. 37. Wagner, J. C. & Berry, G. (1969) Mesotheliomas MESOTHELIOMATA IX RATS AFTER INOCULATION WITH ASBESTOS 185 ST00453h9 in Rats following Inoculation with Asbestos. Br. J. Cancer, 23. 567. Wagser, J. C., Berry. G. & Timbreil, V. (1970) Mesotheliomas in Rats Following the Intra-pleural Inoculation of Asbestos. In Pneumoconiosis. Proc. Internal. Con/. Johannesburg, 1969. Ed. H. A. Shapiro. Cape Town: Oxford University Press, p. 216. APPENDIX The results have been analysed using the model given by Pike (1966) and shown to be valid for experiments of this type (Berry and Wagner 1969). The agespecific death rate of animals dying with a mesothelioma t days after injection is taken as ck(t-w)k~l where c, k and w are constants. These 3 constants could be estimated separately for each treatment of each experiment but correlations between the estimates make them imprecise. In the original experiments, with a total of 417 mesotheliomata, it was shown that k could be taken as 3 for all treatments but that w. the lapse period before any meso theliomata occurred, varied with treat ment: in particular the lapse period for amosite was found to be about 200 days longer than that for chrysotile and cFoddolite but there was an isolated mesothelioma occurring with amosite after only 398 days. For the experiments being reported here, although the estimates of the lapse period vary widely over the different treatments, this wide range could be due to the imprecision of the estimates, and there is no strong evidence that it is invalid to use a common value. Also, there is no evidence of the lapse period being dependent on dose. Therefore, in the analysis the best estimates of k and w based on all our evidence were used; these are w = 270 and k = 3-25. Even with all the data these estimates are not very precise; for example the pairs (300, 2-9) and (220, 3-8) would be acceptable, as also would the estimates (250,3 0) which we used in our preliminary reports. However, these alternatives all lead to similar conclusions. The constant c was estimated for each of the treatment groups and this constant, which we will refer to as the " carcino genicity factor ", serves as a single index summarizing the mesothelioma experience of each treatment group. It combines the information on the proportion of animals which developed a mesothelioma with the times after inoculation at which the meso theliomata occurred. Also, the method of estimation eliminates mortality due to other causes, so that neither chance nor systematic variations in natural mortality between different groups will affect com parisons between these groups. As an example of the elimination of natural mortality, the proportion of rats developing mesotheliomata after injection with SFA chrysotile in Experiments 2 and 4 (69% and 64%) were similar to the pro portion (65%) in SPF rats in our original experiment (Wagner and Berry, 1969). However, the rats in Experiments 2 and 4 were injected at 13 weeks of age opposed to 6 weeks in our original experiment, and also the natural mortality was less in our original experiment, even after allowing for this age difference. Hence the carci nogenicity of the chrysotile was least in our original experiment and this is re flected in the values of the carcinogenicity factor which were 1-7 x 10-9 in the original experiment, 4-7 x 10-9 in Experi ment 2 and 2-9 x 10-9 in Experiment 4 (Table V). The non-significant difference between the values for Experiments 2 and 4 was revealed after eliminating the chance lower natural mortality in the group used in the latter experiment.