Document 2JEwrpXponbpkb89r8QNx0R0p

;> vrrr . "' | ' Carcinogenicity of Fibrous Glass: Pleural Response in the Rat in Relation to Mearl F. Stanton,2 Maxwell Layard,2 Andrew Tegeris, Eliza Miller,2 Margaret May,2 and Elizabeth Kent*-* ABSTRACT--Seventeen fibrous glasses of diverse type or di mensional distribution Induced different incidences of malignant mesenchymal neoplasms when implanted In the pleurae of fe male Osborne-Mendel rats for periods of mors than 1 year. Neo plastic response correlated well with the dimensional distribution of fibers. Fibers less than or equal to 1.5 it in diameter and greater than 8 it In length yielded the highest probability of pleural sarcomas, and probability trends suggested that pleural sarcoma incidence Increased with Increasincpengths of fibers with diameters of less than 1.5 it. Morphologic observations indicated that fibers less than or equal to 8 it In length were y inactivated by phagocytosis. In fibers greater than 8 it in length, the correlation of carcinogenicity with increasing length was diffi cult to explain. Since neoplastic response to a variety of types of durable fibers, particularly asbestos fibers, was similar, our ex periments reinforce the idea that the carcinogenicity of fibers depends on dimension and durability rather than physicochemi cal properties and emphasize that all respirable fibers be viewed with caution.--J Natl Cancer Inst 58: 587-603, 1977. At the 1970 Symposium on Tissue Response to Asbes tos held in Cardiff, Wales, we reported that exception ally fine fibrous glass applied to the pleura of the rat induced sarcomas that were identical to the neoplasms induced with asbestos at this site and closely resembled mesotheliomas induced in the pleura of man by asbes tos. Ancillary data suggested that the carcinogenicity of both asbestos and glass was related to the dimensional configuration of the fibers rather than to either their physicochemical properties or contaminating oils or minerals. Further support for the concept that fiber carcinogenicity depended on durability and fibrous con figuration of the proper dimension was attained in two subsequent reports in which the dimensional distribu tion of fibers of several types of asbestos, glass, and aluminum oxide was correlated with carcinogenicity {l, 2). The present report extends these findings by more precise assessments of both carcinogenicity and the di mensional distribution of fibers in the 17 different types of fibrous glass tested thus far. MATERIALS AND METHODS Rats and implantation procedures.--The rats used and the method of exposing the pleura to fibrous glass were essentially those described in (1). We have consistendy used female, specific-pathogen-free, Osborne-Mendel rats from the outbred, closed colony of the NIH Animal Production Secrion. As untreated controls, more than 4,000 old female Osborne-Mendel rats have been exam ined during the past 6 years. In our animal housing, the rats generally remained free of chronic pneumonias, and no tumors were found in the thorax that might be confused with those induced in the pleura. At weaning, the litters were mixed randomly and the rats were housed 5 per large polycarbonate cage. They were fed commercially prepared rodent pellets and water ad libi tum and treated between 12 and 20 weeks of age. As in previous experiments, a substrate pledget of coarse fibrotis glass was used as the vehicle for implant ing the test fibers. In ail experiments, the surface of flat, 45-mg pledgkts composed of autoclaved, binder-coated, coarse fibrotis glass of the type commonly used as insu lation material, was coated with a standard dose of 40 mg test fibers that had been suspended in 1.5 ml of 10% gelatin by gentle continuous agitation. The gelatin-sus pended coatings of test fiber were allowed to harden on the pledgets! at 4 C, so that the pledgets developed a rubbery consistency. Treatment consisted of a left-sided open thoracotomy under ether anesthesia and the direct placement of the pledget over the visceral pleura. Appli cation in this manner had several advantages: The pledgets could be manipulated easily so that the test fibers on the surface of the pledget could be applied uniformly to a broad surface of the visceral pleura. Additionally,1 it permitted the application of fibers that could not pass readily through a hypodermic needle, insuring that fibers were included that otherwise would have been fragmented or selectively filtered out by the injection procedures. Further, the coarse fibrous glass pledget acted as a partial barrier between, the incision and the test Material, thus retarding migration of the test fibers into the incision and surrounding tissues with the subsequent induction of tumors at extraneous sites. Initially in each experiment, 30 rats were treated. The 1-3% of rats' that died postoperatively within 3 weeks were replaced so that each group consisted of 30 long term survivors. All groups were observed daily; only moribund rats were killed. An effort to insure long term survival required the surgical removal of all mam mary tumors as soon as they became palpable. The survivors were killed during the 25th month after treat ment. Periodically during the progress of the experi ments, groups of 30 rats were treated as controls with gelatin-saturated pledgets of the coarse fibrous glass vehicle, comparable in weight (85 mg) to the test fibers plus vehicle-! The 130 treated control rats from these groups that died during the 24 months following treat ment or that were randomly killed the 25th month after treatment comprised experiment #17 on which the cal culations of significance are based. Gross necropsy find ings were recorded on all rats, and histologic sections were examined after staining with H & E from all im- Abbreviation used: H t E * hematoxylin and eosin. ` Received July 6. 1976; accepted September 15, 1976. * Laboratory jof Pathology, National Cancer Insdtute (NCI), Na tional Institutesof Health. Public Health Service, U.S. Department of Health, Education, and Welfare, Bethesda, Md. 20014. 3 Biometry Branch, NCI. * Pharmacopathics Research Laboratories. Inc., Laurel, Md. 20810. 3 We are particularly indebted to the staff of the Johns-Manville Research and Engineering Center. Denver, Colo., and the Owens- Coming Fiberglas Corporation, Toledo. Ohio, for providing and characterizing many of the glasses studied. VOL. 58, NO. 3, MARCH 1977 587 ! J NATL CANCER INST DUP 0809731 DU 009298 5 88 STANTON T AL. plant sites and all other grossly abnormal tissues. Special histologic stains were done occasionally. Tested glass fibers.--The 17 tested fibrous glasses are listed in table 1 in order of decreasing carcinogenicity. They are designated by symbols that indicated their source, method of preparation, and relation to each other. The seven samples with a first letter designation "M" were processed by Dr. James Leineweber and the staff of the Johns-Manville Research and Engineering Center. Their methods of preparadon, which consisted primarily of fragmentadon of aqueous suspensions by blender and subsequent separation by centrifugation and filtration, have been published in (3). The M sam ples represent three starting batches of either flameattenuated or rotarv-processed borosilicate fibrous glass. The mineral content of the batches was similar to that of the fine fibrous glasses published in (1). Since milling was not used in sample preparadon, the M sam ples were free of extraneous contamination and dif fered from each other only in dimension. The two M samples with a second letter designauon "O" (used in expts 1 and 9) were from a single starting batch of fibers with nominal diameters below 1 (z. The third letter designation of these two samples indicated the aim of final separation into either long (L) or short (S) fibers. Similarly, the M samples with a second letter designa tion "6" (used in expts 3,5, and 16) were from a starting batch of fibers of nominal diameter below 2 n with the third letter designauon indicaung that the aim of final separation was to achieve either long (L), short (S), or whole (W) fibers (i.e., nonseparated fibers). Finally, the M samples with second letter designation "8" (used in expts 8 and 15) were from a batch of fibers of nominal diameter below 5 and again long (L) and short (S) lengths were the aim in separating the two samples. The remaining four starting batches of fibrous glass were characterized in (/). The six samples with a first letter designation "K" (used in expts 2, 4, 6, 7, 10, and 12) were from a single batch of flame-attenuated borosil icate glass with a nominal diameter range of 0.06 to 3.0 ju. and labeled AAA fibrous glass. Samples designated "KW" and "KL" (used in expts 2 and 4, respectively) represent the whole untreated K fibers and a modest attempt to shear these fibers to lengths that remained long but within a range that could be more easily mea sured. We did the shearing by processing an aqueous suspension of the whole fibers in a conventional Waring blender for three periods of 10 minutes and selecting sedimentation fractions that had predominating fibers greater than 50 (i in length. Samples designated "KCP," "KUP," "K.2P," and "KFP" (used in expts 6, 7, 10, and 12) are the same K fibers processed primarily by commi nution to short lengths in a Spex ball-mill with a stain less-steel ball and chamber. This treatment contributed significantly to contamination of the fibers by metallic deposition from the chamber. The duration of milling was 30 seconds for experiments 6 and 7, 2 minutes for experiment 10, and 10 minutes for experiment 12. De pending on the duration of milling, the additional con tent of the following metals was present in the samples by spectrographic analysis: chromium, 0.1-2.3%; cop J NATL CANCER INST per, 0.001-0^025%; iron, 0.46-9.5%; magnesium, 0.01 -- 0.02%; molybdenum, 0.002-0.015%; and nickel, 0.05 1.2%. In addition, sample KCP differed from the other K samples in that it retained a thin outer binder of heatcured, urea-j-formaldehyde resin and small amounts of contaminating urea, mineral oil, and silicone. Samples \jith first letter designations "O," "P," and "Y" were from three separate batches of fibrous glass of the type and size range commonly used commercially. These fibers had diameters three to five times greacer than the M or K test fibers in the range of 5 and 25 fj.-- only rare fibers were less than 1 /x. All three samples had long fibers that required shortening for adequate appli cation and measurement. Sample "02P" (expt 11) was a silica-slag-derived mineral wool commonly used in the past as home insulation. It was composed of brittle, tapered fibers with hooklike attenuations that were highly irritating to the skin. This material contained higher levels of calcium and magnesium oxide than borosilicate glass and additionally was contaminated by metals fromja 2-minute period of milling in a stainlesssteel Spex mill. Sample "P2P" (expt 13) was a fibrous glass commpnly used in filtration processes that con tained higher levels of sodium and calcium oxides than conventional borosilicate glasses. It too was reduced to manageable lengths by comminution in the Spex ball- mill for 2 minutes with consequent contamination from this source. ^Finally, the two samples with the first letter designation! "Y" (expts 14 and 17) were from typical, modern, insulation-type fibrous glass that was coated with a phenol-formaldehyde binder. This was the same coarse-diameter fibrous glass that served as the vehicle for implantation of all test samples in this series and in those reported previously. The fibers were twinning strands of `exceptionally long length that necessitated approximations in the assessment of this dimension. Sample "YW" (expt 17), whole fibers from the pool employed ;as the vehicle, was simply increased in amount to j equal the weight of the vehicle plus test sample. Sample "Y2P" was the same glass reduced to shorter fibers by comminution in the Spex ball-mill for 2 minutes. Like the other samples with third letter desig nations of "P," this sample also contained contaminating metal from the mill. Estimation of particle size distribution.--Aliquots of eacr of the 17 samples were placed on microscopic slides anc on Formvar-coated electron microscopic grids. Wher appropriate, these aliquots were diluted with distillec water and jsonically vibrated to make more uniform!distributed suspensions. The preparations were then ai dried. Froin each slide preparation, a series of micro scopic fields thought to best represent the distributior of particles were photographed at final magnification of x 100 to x2,000 as appropriate to permit length am diameter iheasurements of each particle to the neare: 0.50 (i. In most cases, this required a series of step-wis magnifications of the same fields and the developmer of montages so that extremes in dimension could b adequately assessed. Approximately 1,000 consecutive counted plarticles within these fields yielded relative stable ratibs between a series of size ranges. At lea I | VOL. 58, NO. 3. MARCH U* DUP 0809732 DU 009299 '-'uTrinrrTif CARCINOGENICITY OF FIBROUS GLASS 589 1.000 particles were counted in each preparation except for those few with exceptionally long fibers. This tabula tion of all particles visible by light microscopy insured a sufficient sample, but it did not adequately take into account particles less than 0.50 fj. in diameter. Conse quently, the grid suspensions were viewed in an AEI EM6B electron microscope and photographed at final magnifications of x 1,555 to x 19,698, and the diameter and length of each particle in the photographic fields were measured to the nearest 0.01 /x. Again, stepwise increases in magnification and montages were useful in measuring dimensional extremes. From 200 to 1,000 consecutively counted particles were measured; the number depended on the size distribution and the stabi lization of ratios among the ranges of size" studied. By light microscopic examination of the grids, particles of dimensions previously counted at light magnifications (generally particles with diameters >0.50 /x) could be separated from the submicroscopic particles, arid a ratio between the optically visible particles and the submicro scopic particles could be determined. This ratio was used to calculate the number and dimensional distribution of submicroscopic particles presumed to be present in the fields of the 1,000 particles visible by light microscopy that had been counted previously. The dimensions of the total number of particles counted by light micros copy and those of the estimated total submicroscopic particles present in the same fields as those counted by light microscopy were then entered into an IBM 360-370 systems computer, and the following calculations were made for each sample. Assuming the particles were cylindrical, we calculated the volume of each counted particle and converted this value to weight by multiplying by the density of glass. The weights of all counted particles were then summed to yield an estimate of the fraction of the standard 40 mg dose counted. From this estimate, either the number of particles of a given size in the standard 40-mg dose or the percent by weight occupied by particles of a given size could be calculated. The counted particles were then grouped into 34 dimensional ranges that seemed the most likely categories to show differences in re sponse. These arbitrary ranges permitted simple illus trations of fiber distribution both by percent of the dose by weight that was occupied by each range or the num ber of particles per ng in each range (text-figs. 1,2). Since it was probable that the count of numbers of particles was no more accurate than one order of magni tude, they were tabulated as the common log with the characteristic of the log representing the probable limit of accuracy. RESULTS Table 1 summarizes the results in terms of survival and tumor incidence. Whether or not pleural sarcomas developed, mortality was high especially during the sec ond year after treatment. A scattering of tumors in other sites accounted for some deaths, but these neo plasms were not unusual in untreated controls and no significant trend in incidence could be related to treat ment. In all groups, deaths from other than tumors resulted primarilv from.pneumonia, particularly when inflammation spread to the untreated right lung, since treatment partly compromised the total pulmonary re serve. For these reasons, the best estimate of neoplastic response depended' on time o-f surviva.l as well as the number of tumors. Table 1 tabulates|the time of death and whether or not a pleural sarcoma was present at death for each rat in the 17 experiments. From data similarly tabulated on a weekly basis, a standard actuarial life table method was used to estimate the overall probability of developing pleural sarcomas in each experiment (4, 5). This com putation. included it table I, is derived in the following manner. Let nx be the number of rats alive at the beginning of weekxj, and dz be the number of rats dying of tumor in week x. {Then the estimated probability that a rat which survives.to the beginning of week* does not die with tumor in week x is Ar=l-4,/nr. The estimated probability that a rat dies with tumor by the end of week x is j?*-l -A-/V--/V The method adjusts for rats that died early in the exper iment without pleural sarcoma, which otherwise would cause the observed proportion of tumors to be less than the tumor probability, had all rats survived until they developed pieural sarcoma or the experiment was ter minated. Since botli treated control rats and the rats .in the experimental groups were subjected to the same implantation procedure, the risk of dying from causes other than tumor was comparable for all groups. There fore. the assumption underlying actuarial analyses ap peared to be satisfied. The estimated probabilities of pleural sarcoma in the 17 experiments rariged from 85.3% to zero. However, the results suggested division of the experiments into only three response groups excluding the treated con trols (footnotes c, d, and e in table I). The high-risk group consisted of Experiments 1-5 from samples com posed of either intact fibers or a fraction of the longest fibers from the glass samples with the finest diameters. The intermediate-risk group consisted of experiments 6, 7, and 8. Two of these three experiments were with fine fibers that hai been reduced in length by milling, and the third consisted of the long-fibered fraction of large diameter (>2 /x). The eight experiments in the low-risk group (expts 9 through 16) were, for the most part, experimentsjwith either large-diameter fibers or very short fibers, j In testing the combined experiments in each of the three groups for significant differences by the methods of Cox (6) and Mantel (7), we found that, in terms of tumor probability within groups, none of the individual experiments were significandy different from any other. However, the low-risk group was significantly different from th!e intermediate-risk group (P<0.01), which in turn was {significantly different from the highrisk group (P<0.0001). The differences between the treated controls (ejxpt 17) and the intermediate-risk and VOL. 58, NO. 3, MARCH 1977 J NATL CANCER INST I 1 I i DUP 0809733 Du 009300 590 STANTON ET AL. Table 1.--Seventeen experiments with different types offibrous glass ranked by probability ofpleural sarcoma development* Sample designation Experiment No. and rank Probability pleural sarcomi =SDs Wk after treatment MOL 1 85.3 = 13.2 KL M6L KW M6W 234 High-risk group' 5 KCP KUP M8L 67 8 Intermediate- risk group11 MOS 9 K2P | io | 02P KFP P2P Y2P 11 12 13 14 Low-risk group< 73.9 71.2 69.3 64.4 21.5 19.4 14.3 8.3 8.5 =9.1 =9.6 =17.7 =8.7 =10.3_ =9.4 =5.6 8.1 6.7 5.9 5.7 5.5 p.5 =5.4 =5.7 =5.5 =5.9 Number dying in stated period after treatment/ M8S 15 4.5 =4.4 M6S 16 0 YW 17 0 6-52 53-56 57-60 61-64 65-68 63-72 73-76 77-80 81-84 85-88 89-92 93-96 97-100 101-104 353-104 0/13 0/1 0/1 0/5 0/1 1/0 1/1 2/0 1/0 1/0 0/1 1/0 1/0 1/0 1/0 1/0 1/5 1/0 1/0 0/1 1/1 1/0 3/0 2/3 0/1 0/1 3/0 2/0 2/0 3/1 11/7 10/7 11/7 9/8 20/9 18/11 16/9 0/8 0/1 0/3 1/2 1/0 1/1 1/1 1/2 1/2 1/3 7/15 0/2 0/4 0/2 0/2 0/1 0/2 0/1 0/1 0/1 1/0 0/3 0/2 1/1 0/2 1/0 0/1 0/2 1/0 0/2 1/4 0/4 2/18 2/13 1/12 5/23 3/23 2/26 0/3 0/2 0/5 0/3 0/5 0/6 0/1 0/2 0/15 0/1 0/1 0/1 0/3 0/5 0/2 0/17 0/1 0/5 0/1 0/12 0/2 0/1 0/1 0/6 0/1 0/4 0/2 0/1 0/2 0/2 0/1 0/1 0/2 0/1 0/1 0/1 0/1 1/0 0/1 0/1 0/2 1/1 1/0 1/0 0/3 0/1 0/4 0/2 0/1 0/2 0/1 0/1 0/1 0/3 0/12 0/1 0/1 0/2 1/1 1/0 0/4 0/9 0/20 1/23 1/14 1/17 0/16 0/19 0/21 0/15 0/41 2/25 2/26 1/24 1/26 1/24 1/23 1/28 0/28 0/115 All expts employed 30 long-term surviving rats except #17 (control, a -130). Standard doses of 40 mg test fibers used in all expts. ` Estimated actuarial probability of pleural sarcoma calculated on weekly death rate of rats surviving the first 52 wk after treatment. ' High-risk group: Probability of pleural sarcoma significantly higher than intermediate-risk group. Differences between expts within the group not significant. j ' Intermediate-risk group: Probability of pleural sarcoma significantly higher than treated controls and low-risk group. Differences between expts within the group not significant. I ' ' Low-risk group: Probability of pleural sarcoma significantly higher than untreated controls. Data insufficient to distinguish difference from treated controls. Differences between expts within the group not significant. ; ' Number with pleurai sarcoma/No. without pleural sarcoma. high-risk groups were both significant (P<0.01 and P<0.0001, respectively). However, the low-risk group was not significandy different from the treated controls at the 5% level (J3=0.1) because the 130 treated controls were insufficient in number to exclude the possibility that single tumors might be the result of either the coarse glass fibers of the vehicle or simple nonspecific injury to the pleura. Subsequent to the completion of this study, three additional groups of treated controls were studied. Three pleural sarcomas were observed in a total of 150 rats from these groups. Thus, although the 1 or 2 neoplasms in the low-risk experiments were sig nificantly different from untreated controls {see first paragraph in "Materials and Methods"), the specificity of low-level response seemed doubtful although it may have been the result of the large fibers of the vehicle. The relationship of the high-, intermediate-, and low- response groups to particular types of fibers suggested, as in previous experiments {I, 2), that tumor response was related to fiber dimension. A good indication of this relationship could be gained if one simply compared the high-, intermediate-, and low-risk probability groups as illustrated in figures 1-17. Ranked in order of tumor probability, the figures indicated that fineness of diame ter and increasing length seemed related to carcino genicity. To test this hypothesis, estimates of the distri bution of particles by size were made in each experiment (as described in "Materials and Methods"), and these data were correlated with response. Text-figures 1 and 2 represent a Jseries of dimensional grids, one grid for each experiment, with the fibers grouped into 34 cate gories of size and tabulated either as percentage of sample weight (text-fig. 1) or by the common logarithm of the number of particles per tg (text-fig. 2). The grid pattern was arranged so that, as one moved from lower left to upper right, the categories increased in diameter and length. Categories on the extreme left were essen tially nonfiorous particles, and particles of the upper left categories frequently represented aggregates of submicroscopic'particles and fibers. Categories on the right side of the grid were open-ended to accommodate the usual long fibers in many of the samples. In the submicroscopic ranges, fibers greater than 8 p. in length were often difficult to trace for their entire length and were included in the open-end categories. With ideal samples, particles in each sample should have been con fined to a jnarrow range of adjacent categories. Ob viously, this was not the case with most of the 17 sam ples, but the size distribution was sufficiently narrow in some samples to permit tentative conclusions. Compari son of the experiments at the two extremes of tumor probability (expt 1 vs. expts 16 and 17) indicated three patterns of nearly mutually exclusive fiber distribution. The high-risk experiment (expt 1) had a fiber distribu tion concentrated in the lower right categories of the grid, whereas the negative experiments (expts 16 and J NATL CANCER INST DUP 0809734 DU 009301 CARCINOGENICITY OF FIBROUS GLASS 591 oO CO 00 oo *} AA oo 00 00 o ''t A oO AA MOL-sl 85.3% >2.5 -4.0 >1.5 -2.5 >.50-1.5 >.25-.50 >.10-.25 >.05-.10 .01-.05 >8.0 <1 <1 <1 <1 <1 <1 <1 10 64 14 7 3 1 Kli - =2 -8.0 73 9% ,-4.0 -8.0 -4.0 s-1 -2.5 < 1 1 >50-1.5 1 ^1 >.j>5-.50 s' 1 < 1 >.10-.25 <1 <1 <1 >.05-.10 <1 <1 <1 .D1 -.05 < 1 <1 <1 5 22 3 38 19 28 ^1 1 1 <1 M6L-=3 >8.0 .71.2% >4.0-8.0 >2.5-4.0 >1.5-2.5 <1 >.50-1.5 <1 <1 >.25-.50 <1 <1 <1 >.10-.25 <1 <1 <1 >.05-.10 <1 <1 <1 .01-.05 <1 <1 15 13 32 32 1 <1 <1 <1 KW-S4 >8.0 69.3% >4.0- >2.5 -4.0 >1.5 -2.5 <1 <1 >.50-1.5 <1 <1 <1 >.25-.50 <1 <1 <1 >.10-.25 <1 >.05-. 10 <1 .01-.05 1 10 1 13 <1 11 1 30 1 30 <1 <1 <1 <1 M6W-45 64; 4% -4.0 -2.5 >.|50-1.5 <1 >.25-.50 <1 >.,10-.25 <1 >.p5-.10 <1 .01 -.05 <1 >8.0 <1 <1 <1 1 12 <1 <1 <1 <1 <1 <1 <1 <1 2 <1 48 10 11 27 30 2 1 <1 <1 looo COo COo COo KCP-=6 21.5% >2.5 -4.0 >1.5 -2.5 >.50-1.5 <1 >.25 -.50 <1 >.10-.25 <1 >.05-.10 <1 .01-.05 >8.0 3 <i 2 <i 1 32 <1 <1 <1 <1 <1 <1 <1 11 18 43 2 51 34 <1 <1 M0S-=-9 8.3% >2.5 -4.0 >1.5 -2.5 >.50-1.5 2 >.25-.50 1 >.10-.25 <1 >.05-.10 <1 .01-.05 <1 >8.0 11 6 i2 9i 5i 1i <1 <1 <1 45 <1 1 1 KUP--7 >8.0 19.4% >4.0- <1 >2.5 -4.0 i 4 >1.5 -2.5 2 3 >.50-1.5 <1 6 6 >.25-.50 <1 <1 <1 >.10-.25 <1 <1 <1 >.05-.10 <1 <1 <1 .01-.05 <1 18 26 11 3 6 75 <1 <1 <1 <1 K2P-*10 >8.0 8.1% >4.0-8.0 >2.5- 4.0 5 14 >1.5- 2.5 18 17 >.50-1.5 3 23 6 >.25-.50 <1 <1 <1 >.10-.25 <1 >.05-.10 <1 <1 .01-.05 8 5 1 M8l-=8 14 .3% >2.5 -4.0 >1.5 -2.5 >. 50-1.5 >.25 -.50 >. 10-.25 >.05-.10 -101 - .05 >8.0 <1 <1 <1 1 <1 <1 <1 1 4 41 2 31 1 15 <1 <1 <1 02P- = 11 >8.0 38 19 6.>% >4.0 9 18 <1 ! >2.5 -4.0 i 4 5 2 | >1.5 -2.5 i 1 2 <1 >,50-1.5 <1 <i <1 <1 <1 >.'25-.50 <1 <i <1 >Jl 0-.25 <1 <i >;o5-.io <1 ;oi-.05 KFP-=12 >8.0 5.9% >4.0-8.0 >2.5-4.0 >1.5-2.5 <1 >.50-1.5 <1 <1 >.25-.50 <1 <1 >.10-.25 <1 >.05-.10 <1 .01-.05 49 10 <1 <1 P2P - - 13 >8.0 5.7% >4.0-8.0 4 -4.0 2 2 -2.5 1 <1 >.50-1.5 <1 <1 <1 >.25-.50 <1 >.10-.25 <1 <1 >.05-. 10 <1 <1 .01-.05 49 29 2 <1 <1 . Y2P-=t14 >8.0 56 10 5,5% >4.0- 8.0 <1 30 4 >2.5 -4.0 1 2 5 >1.5 -2.5 1 <1 1 >150-1.5 <1 <1 <1 <1 >l25-.50 > 10-.25 <1 <1 <1 > 05-.10 01-.05 M8S-*15 >8.0 4.5% >4.0- 8.0 i3 >2.5 -4.0 1 7 23 >1.5 -2.5 4 14 23 >.50-1.5 1 6 6 3 >.25-.50 <1 <1 X10-.25 <1 >.05-.10 <1 <1 <1 .01-.05 .01-1 > 1-4 >> 4-8 8-64 > 64 M6S-416 >8.0 0% >4.0-8.0 >2.5 -4.0 >1.5 -2.5 1 6 >.50-1.5 5 34 26 18 >.25-.50 2 4 1 <1 >.10-.25 1 1 <1 >.05-.10 <1 <1 .01-.05 <1 > .01-1 1-4 4-8 8-64 64 Length p Y,W - = 17 0?i >4.0- >2.5- 4.0 2.5 >;.50-1.5 >|.25 -.50 >|. 10-.25 >!.05 -.10 .01 -.05 CO 1 >8.0 <1 83 <1 14 <1 3 <1 <1 <1 <1 .01-1 1-4 4-8 8-64 64 Text-ficure 1.--Fiber distribution into 34 dimensional categories by percent of test sample weight for each of the 17 expts ranked by probability of pleural sarcoma. i VOL. 58, NO. 3, MARCH 1977 J NATL CANCER INST DUP 0809735 DU Q09302 I STANTON ET AL. MOL-#1 >03 1 85.3% >4.0-8.0 i 5-4.0 i 5-2.5 1 >.50-1.5 2.23 2.53 0.23 3.08 >.25-.50 3.08 3.95 >.10-.25 2.93 3.35 4.53 >.05-.10 3.93 4.79 .01 -.05 3.46 4.65 KL- = 2 1 73.9% Hi.67 0.67 1.52 0.97 1.45 0.67 2.03 2.19 2.95 2,40 3.42 2.74 2.55 3,16 3.63 3.03 3.16 3.33 3.03 2.85 4.09 3.76 3.25 3.03 3.73 3.03 3.03 M6L-=3 >Ch8- 71.2% >4.0-8.0 "2.5-4.0 1.53 1.23 >1 5-2.5 1.84 2.76 2.13 ".50-1.5 2.81 2.05 2.59 3.70 2.90 >.25-.50 2.44 2.441 3.16 3.39 >.10-.25 2.44 3.52 3.44 3.62 >.05-.10 2.44 3.44 3.56 3.70 .01-.05 3.44 3.35 3.14 KW- =4 0.06 0.06 69.3% P.36 0.76 P.06 1.061 0.66 0.37 [1.41 1.98 2.48 1.80 1.29 12.42 2.51 1,06 0.67 1.70 2.75 1.67 2.71 1.67 2,51 2.21 M6W-*5 64.4% 1.10 1.58 1.10 1.09 1.95 1.16 1.10 2.18 2.74 2.01 2.28 3.17 3.21 3.69 2.34 1.40 2.84 3.11 3.54 1.70 3.22 3.251 3.69 3.10 3.5513.471 3.76 2.71 2.951 2.71 2.80 KCP-=6 21.5% >4.0-8.0 >2.5-4.0 1.44 >1.5-2.5 2.05 >.50-1.5 3.00 3.59 >.25-.50 3.24 3.38 >.10-.25 3.24 3.28 >.05-.10 2.50 2.90 .01-.05 2.20 2.98 >8.0 1.81 2.05 2.17 3.17 3.10 3.10 2.67 0.97 2.01 1.81 1.81 2.31 0.97 2.85 2.01 2.55 2.50 KUP - = 7 19.4% 2.11 2.73 2.88 3.93 3.36 3.54 3.36 3.50 3.17 3.32 2.31 1.26 2.52 2.71 3.54 2.84 2.61 2.63 2.04 1.74 [2.31 1.26 12.44 13.19 1.96 2.57 2.31 2.80 2.31 M8L -=8 14.3Ch 0.17 0.77 1.73 2.02 1.49 1.12 2.11 2.27 2.60 1.45 2.35 2.42 1.85 2.38 1.85 1.85 -, M0S-=9 8.3% >4.0-8.0 >2.5-4.0 2.97 >1.5-2.5 3.43 >.50-1.5 3.88 3.91 >.25-.50 4.34 4.02 >.10-.25 4.43 3.88 >.05-.10 5.90 4.19 .01-.05 6.77 4.63 >8.0 2.46 2.9^ 2.37 2.76 3.69 2.72 1.76 1.17 2.46 3.37 K2P- *10 8.1% TEST 3.09 3.84 3.56 4.12 4.53 3.60 4.28 3.02 1.81 4.11 3.91 2.89 2.41 2.66 1.81 02P = 11 1.78 0.48 6.7% 2.32 2.17 0.48 2.06 2.48 2.09 0.48 2.49 1.95 1.94 0.48 2.83 2.83 1.91 1.84 0.48 2.44 0.48 1.68 2.08 1.65 2.41 KFP-=12 5.9% >4.0-8.0 >2.5-4.0 2.89 >1.5-2.5 3.20 >.50-1.5 3.19 3.48 >.25-.50 2.96 1.22 >.10-.25 3.30 >.05-.10 3.27 .01-.05 >8.0 2.67 2.73 1.92 1.52 2.03 2.03 0.92 0.92 M8S--15 >8.0 4.5% >4.0-8.0 1.37 >2.5-4.0 2.21 2.75 2.90 >1.5-2.5 3.17 3.43 3.29 >.50-1.5 3.41 3.92 3.46 2.81 ^.25-.50 3.83 2.48 >.10-.25 3.81 .05-. 10 3.85 2.45 2.14 .01-.05 .01-1 >1-4 ^4-8>8-64 >64 P2P- *13 2.00 5.7% 1.85 2.21 1.06 2.30 2.08 1.70 2.28 0.91 0.69 2.56 2.51 1.21 0.81 1.21 2.36 1.29 1.55 0.81 ' f M6S- *16 0% 1 1 2.45 | 2.85 4.43 4.83 4.38 i 3.79 5.17 4.88 3.87[ 2.15 5.60 4.73 3.581 5.45 4.28 1 5.16 1 .01-1 >1-4 "4-8 >8-64 "64 Y2P-=14 5.5% 1.94 2.58 3.42 2.09 1.96 1.08 2.24 0.60 2.12 2.16 1.74 1.641 1.60 Tiol 3.12 1.60 1.30 YW- = 17 0% 0.89 0.92 0.80 T.OO T.10 0.34 0.40 0.30 0.11 0.41 .01 -1 ''l -4 '-4-8 >8-64 '64 Length ri Tr.xT-figure 2.--Fiber distribution into 34 dimensional categories by common log of tile number of parudes per jtg in each size category for ea of the 17 expts ranked by probability of pleural sarcoma. J NATL CANCER INST VOL. 58, NO. 3, MARCH no icnon 4 flfl DU 009303 CARCINOGENICITY OF FIBROUS CLASS 593 17) had fiber concentrations in either the upper right or lower left categories of the grid. These three experi ments suggested that fibers in the submicroscopic range (i.e., diameters, 0.5 ft) and most likely those of great est length in this range were the most carcinogenic. This impression was further augmented by the predomi nance of fibers with diameters of less than or equal to 0.25 ft and lengths of greater than 8 pt in the remaining four high-risk experiments. Furthermore, in the lowrisk experiments (expts 9 through 16), the most consist ent deficiencies in fiber distribution were in these same extremes of fineness and length. Two of the three ex periments in the intermediate group (expts 6 and 8) also seemed to support the carcinogenicity of submicros copic long fibers by yielding levels of this type of fiber that fell becween those of the high-risk and low-risk groups. The failure of experiment # 7 to yield a higher probability of pleural sarcomas was the unexplained exception. Furthermore, the problem of a low, but po tentially significant, risk of pleural sarcomas could not be limited to long fibers in the submicroscopic range, since these categories of fibers could not be detected in most of the experiments in the low-risk group. Statistical regression techniques afforded a second more refined method of analysis. The logit transforma tion (5) was applied to the estimated tumor probabilities (P) according to the following formula: logit- To relate the logits to the particle size distribution, weighted linear regression methods were used. These methods find the linear function of the form found to provide good fits to the data. The first of these included fibers with diameters less than or equal to 0.25 ft and lengths greater than 8 ft. The estimated regres sion equation was log & -3.06 +0.993*. with a correlationj coefficient of 0.93, which reflects a highly significant relationship. The other category com prised fibers with diameters between 0.25 and 1.5 ft and lengths greater than 64 ft. The estimated regression equation was ! - log{ 14 +1.340*. with a correlation coefficient of 0.92, again reflecting a highly significant relationship. The graphs of these equations, together with the data points for each experi ment, are given in text-figures 3 and 4. Since these two explanatory variables were so highly correlated with the logits of tumor probability, it was apparent that they must themselves b highly correlated, and indeed their correlation was 0.95. The consequence of this high correlauon was that it was impossible to conclude, on the basis of these experiments, which group of fibers was the more important one for carcinogenesis. Similar regression analyses with the use of percent of total weight of particles as explanatory variable?, led to diameters between! 0.25 and 1.5 ft with lengths greater than 64 jx as the dimensional range which best explained tumor probability.|The regression equation was 2.30+0.095*, logit-a+B,x,H------Btx, which best fits the data, where the "u i= 1, , k, quan tify certain features of the particle size distributions (e.g., common logarithm of the number of particles per jug or percent total weight in various size categories). The weights used in this analysis were the estimated standard errors of the tumor probability estimates, which were obtained from the life table analysis (table 1). As previously noted, inspection of the data readily suggested that a concentration of particles in the rela tively thin (diameter, 0.25 ft) and long (length, >8 ft) dimensional categories was associated with higher prob abilities of tumor. However, it was possible that there were other not easily observable relationships between tumor probability and the particle size distributions. This possibility was explored by the regression tech niques described, using as explanatory variables either logarithms of numbers of particles or percent weight, in a variety of dimensional categories. No significant rela tionship between distributional factors and tumor prob ability was found, other than the evident positive corre lation between "long and thin" particle concentration and tumor probability. A summary of correlation coeffi cients is given in table 2. With the use of logarithms of numbers of fibers, two nonoverlapping size categories, used individually, were with a correlation coefficient of 0.88 (see table 2, text-fig. 5). For the range with a diameter of less than or equal to 0.25 /x and length of greater than 8 ft, the percent weight gave a relatively poor fit. The regression equa tion was | W(jrj) --1.84 +0.356*, with a correlation coefficient of 0.51 (see table 2, text-fig. 6). | . The equations described above could not be inter preted too rigidly j because the estimated panicle size distributions were subject to considerable sampling er ror, the range of dimensional variation in the glass types used was in some Respects limited, and the panicular choices of dimensional range as the explanatory varia bles were arbitrary and, to some extent, suggested by the data. Nevenheless, taken together, the regression analyses would suggest that glass fibers with diameters less than 1.5 ft and lengths greater than 8 ft were highly carcinogenic in the pleura of the rat and that fibers larger or smaller than these dimensions were less carci nogenic. Thus carcinogenic potential seemed deter mined by decreasing diameter until at or near the limits of optical visibility', but increasing length became an equally critical factdr within this range. Tissues at the site of implantation in rats dying during VOL. 58, NO. 3, MARCH 1977 J NATL CANCER INST DUP 0809737 DU 009304 594 STANTON ET AL. I Table 2.--Correlation coefficients of logit of tumor probability with two variables in different dimensional ranges ofparticle size 1 .. vehicle and cofnrol. In the acute stage, hyperplasia of che mesochelium and vasodilatation, edema, and poly Fiber diameter, p` morphonuclear leukocyte infiltration of the adjacent connective tissue occurred. Tissue necrosis was minimal >4 >1.5-4 >0.25-1.5 0-0.25 and the lung Ivas only superficially affected. With all 0-8 -0.34 -0.45 0.00 0.11 types of fibers^ these acute changes at the pleural sur -0.29 -0.33 -0.41 -0.26 face were promptly followed by mobilization and prolif >8-64 -0.27 -0.40 -0.09 -0.08 0.65 0.63 eration of vascular granulation tissue from mesenchyme which penetrated the overlying mesothelium and in vaded the scaffold of glass fibers. Within the granula >64 -0.10 0.63 0.92 -0.27 0.42 0.88 tion tissue, polymorphonuclear leukocytes were re placed by mononuclear leukocytes, fibroblasts, and es * Upper figure is correlation coefficient for number of particles/ dose; lower figure is correlation coefficient for percent by weight of particies/dose. pecially macrophages. The macrophages either en gulfed small particles or adhered closely to large fibers, often fusing to form giant cells that effectively enclosed large fiber segments. For the most part, this process obscured the initially hyperplastic mesothelium, and a thick, uniform1 coat of vascular granulation tissue infil trated the glass fibers and covered the affected lung surface. Slowly during the second year, the network of LOG NUMBER OF PARTICLES PER HICROGRAH Text-figure 3.--Regression curve relating probability of tumor to logarithm of number of particles per jug with a diameter of S0.25 p and a length of >8 p. ftr/finon Hnn LOG RUNBEA OF PARTICLES PER RICROGRAN Text-ficure 4.--Regression curve relating probability of tumor to logarithm of number of particles per pg with 0.25 M^diameter S1.5 p and a length of >64 p. the experiments showed changes that could be assem bled into a plausible series of events related to the devel opment of the pleural sarcomas. The basic response to all Fibers was typical of that to most foreign bodies, and in particular to the long, coarse fibers employed as J NATL CANCER INST . _ VOL. 58, NO. 3. MARCH 1977 I DU 009305 CARCINOOENICITY OF FIBROUS CLASS 595 granulation tissue permeated with macrophages, epi thelioid cells, and foreign-body giant cells persisted. Nocabiv, the lesions acquired a minimum of collagen. The response to these long, coarse fibers was highly cellular and often showed degrees of disorganization that might be confused with incipient neoplasia. But the lesions never extended beyond the site of fiber deposi tion nor did they show the anaplastic homogeneity that signals the onset of neoplasia (fig. 18). Lesions that contained short fibers (expts 9-16), in addition to the long, coarse fibers of the vehicle, were essentially like those just described except that phagocy tosis seemed far more successful. Even in the acute stages, an abundance of particle-laden macrophages could be found both at the site and in local Ivmph nodes. With time, it was apparent that only a small part of the short fibers was transported from the site. Pleural le sions at the end of 2 years had not only the collapsed scaffold of phagocyte-laden, coarse fibers, but also an abundance of short fibers entrapped within phagocytes. Again there was little evidence of collagen deposition even in the midst of abundant particles of both extremes of size. On the other hand, lesions that predominantly con tained fine, long fibers (expts 1-5), in addition to the coarse fibers of the vehicle, followed the pattern of granulation tissue proliferation and phagocytosis as pre viously described, but phagocytosis seemed inadequate to accommodate the fine, long fibers. In as few as 8 weeks, collagen had begun to accumulate and replace the cellular granulation tissue that invested the fibers. Islands of granulation tissue remained, but for the most part the reactive pleural covering was converted to a patchy, acellular, collagenous matrix which contained abundant, fine, long fibers seemingly free of contact with phagocytes or other inflammatory cells (fig. 19). As in other groups, regional lymph nodes contained phag ocytes laden with the smaller fibers, but even though intracellular fibers were abundant, collagen deposition was not apparent in the nodes. The link between fine, long fibers and collagen depo sition was further supported by those experiments that contained small numbers of fine, long fibers among a wide distribution of fiber sizes (expts 6, 7, and 8). In these experiments, the extent of fibrosis was less than in experiments 1-5, but greater than in experiments 9-17. Although only rough comparisons could be made, it was clear that the degree of fibrosis matched the probability of pleural sarcomas. We had no conclusive indication of the exact site of tumor development, but a most likely site seemed to be mesenchyme trapped within the dense collagenous tissue. During the course of the 17 experiments, 89 pleural sarcomas were observed. The tumors appeared similar in structure to those observed previously in rats with pleural implants of either glass, asbestos, or aluminum oxide fibers (I, 2). The tumors could be divided into those composed almost entirely of elongated spindle cells, those in which the cells assumed bizarre, poly morphic configurations without identifiable characteris tics, and those rare neoplasms that assumed an acinar or papillarv epithelioid donfiguration (figs. 20-23). How ever, all three patterns seemed closely related, since they often merged in ^ single tumor, and no correlation between experiment and tumor pattern was apparent. The tumors frequently invaded mediastinal organs and occasionally implanted on the opposite pleura, but dis tant metastases were iobserved only rarely. One tumor was of particular note in that, although the primary tumor seemed composed entirely of spindle cells, im plants of the tumor to the pleura of the opposite lung tended to show a papillary epithelioid differentiation (fig. 23). | DISCUSSION I | . The results of these experiments raise at least two points that merit discussion, the relationship of fiber dimension to mechanisms of carcinogenesis and the rel evance of the findings to human exposure. An early paper (I) proposed pat fibers at the lower range of optical visibility (i.e.,|diameters, <1.5 p) that were ex ceptionally short (i.eJ, lengths, <5 p) might account for the carcinogenicity of several types of glass and asbestos fibers. This hypothesis was not supported by subsequent data (2) which, along with those of the present report, indicate that the fine-diameter fibers that are long are carcinogenic and probably become more carcinogenic as their length increases. The negligible carcinogenicity of short fibers is perhaps related to the histologic observa tion that virtually all coarse- and fine-diameter fibers with lengths of 8 p or less are efficiently entrapped' by phagocytes. Many of the smallest particles are transported to regional lymph nodes, but even those that are not seem completely sequestered within the cytoplasmic limits of macrophages and foreign-body giant cells at the site of implantation. Similar observations have been made with chrysotile fibers in vitro (8). If we assume that efficient phagocytosis accounts for the lack of carci nogenicity of fibers less than or equal to 8 p in length and that levels of tumor response depend on the quan tity of fibers not effectively handled by phagocytes, we still must explain why fibers of coarse diameter that are long are less carcindgenic than fibers of fine diameter that are long. Considering that all rats were exposed to equal weights of the various types of fibers, obviously there were far more fine, long fibers than coarse, long fibers per dose. It also holds that fine fibers would have more ends and more surface area as well as a different surface curvature than comparable weights of long, coarse fibers. These differences suggest several hy potheses. It is reasonable that the ends of exceptionally fine fibers could penetrate both ceil and nuclear mem brane and cause varieties of damage that were not le thal. Nonlethal alteration of the genome is perhaps the simplest and most direct event that could initiate cancer. However, if cellular; penetration is responsible for carci nogenesis, then the number of available fine fiber ends should determine :the numbers of tumors. Conse quently, equal weights of fine fibers of intermediate length should be more carcinogenic than fine fibers of long length. A careful review of the data suggests no VOL. 58, .VO. 3, MARCH 1977 | J NATL CANCER INST ' i j i 1 DUP 0809739 DU 009306 596 STANTON ET AL. evidence of chis. Alternately, it has been proposed that the amount of fiber surface area available for cell an chorage might be critical to carcinogenesis (9), or that the tightly curved surface of fine fibers might alter anchorage sufficiently to cause cancer. Here again, if carcinogenicity depended simply on the amount of sur face area of fine fibers, then equal weights of fine fibers of intermediate length should be equal in carcinogenic ity to fine fiBSrs of long length. Although there is some correlation between tumor incidence and the total sur face area of fibers greater than 8 /it in length, the data are not sufficient to prove this hypothesis. In fact, a comparison of the intermediate-risk group with the high-risk group suggests that, without apparent differ ence in total surface area, exceptionally long length enhances carcinogenicity (see text-fig. 2, table 2). How ever, the process of phagocytosis, as it affects all sizes of fibers, needs to be taken into account if one intends to properly assess the number of free ends or the amount of free surface, and this is a problem that does not lend itself well to quantitation. Although present data are inadequate, there may be a direct correlation between tumor probability and the amount of fiber surface free of phagocytic activity. Further, since an apparent corre lation exists between the amount of collagen in the lesion and the probability of pleural sarcoma, it is con ceivable that collagen acts to block phagocytic activity on fiber surfaces. Collagen deposition alone is unlikely to induce cancer, but fine, long fibers entrapped within collagen and thus freed of phagocytic activity might play a key role. Collagen deposition relates fiber carcinogenesis to the idea that pulmonary cancer in man is often associated with scarring (10) and to experimental carcinogenesis with inert solid sheets (11, 12). With solid sheets, carci nogenesis seems to depend on the development of a collagenous envelope at the site which is not unlike the pleural collagen induced by fine, long fibers. With both sheets and fibers, the collagen matrix may be uniquely dense and sufficiently ischemic to force the mutational selection of cells capable of uncontrolled proliferation (13). If collagen is a factor,'then the question of its character and etiology arises. Our consideration of col lagen or cancer mechanism is largely conjecture at the present time, but future experiments with fibers may be an advantage in the study of many questions concerning the general etiology of cancer. It is of course not necessary to understand the mecha nism to appreciate the relevance of these experiments to man. Many of the mesotheliomas of the pleura attrib uted to asbestos exposure in humans do not differ sub stantially from the sarcomas of the pleura induced in the rat with implants of the several types of asbestos that we and others have studied (1,14-16). Even though inhaled asbestos reaches the human pleura by a different route than asbestos implanted in the experimental animal, it is unlikely that proximal mechanisms of asbestos carcino genesis differ in rodents and man. It is our contention that the proximal carcinogenicity of asbestos depends primarily on its exceptionally fine, long dimensions and durability rather than on physicochemical properties (2). It follows ,then that durable fibers of other types should be equally carcinogenic and yield the same types of tumors if they attain the exceptionally fine, iong dimensions of asbestos. In the rat, this would seem to be true of the variety of fibers that we have studied thus far, including fhe glass fibers reported here. To date we have induced significant numbers of pleural sarcomas in rats with fine,| durable fibers of the following types: chrysotile, crocjdolite, amosite, tremolite, glass, attapulgite, dawsonite, aluminum oxide, silicon carbide, and potassium titanate. Appropriate nonfibrous controls suggest that foreign materials other than long, fine fibers may cause fibrosis, but they do not induce signifi cant incidencesJof pleural sarcomas. It seems reasonable then to assume that all dusts that contain fine, long, durable fibers j represent a potential hazard to man. Certainly, any :measurements of human exposure or consideration of tolerance levels for human exposure should take into account not only the variety of the fiber but also the dimensional character of the fiber. We have emphasized previously that our experiments explore mechanisms of fiber carcinogenesis at the proxi mal site of susceptible tissues rather than acting as an estimate of the pverall hazard of different fibers to man (1). Clearly, fibers must have the opportunity to reach the target tissuj: to cause cancer. Our experiments are inappropriate for evaluating many aspects of the envi ronmental hazard, since they circumvent those factors that might inhibit or enhance exposure through natural routes. Subtle structural differences in different types of fibers may influence the aerodynamic properties of fibers as well as their penetrability and clearance from the lung. These1 factors would seem to play an important role in pulmonary exposure to different types of asbes tos (17-19), and similar factors may well apply to other types of fibers.| Although several well-executed inhala tion studies with glass particles in animals have indicated no adverse effects (20-22), we believe these experiments have not entirely resolved the issue of carcinogenesis by inhalation, since the experiments were in part handi capped by insufficient quantitative measurements of fi ber size and dose. Inhalation experiments with glass fibers in narrowly limited dimensional ranges, particu larly those ranges found to be the most carcinogenic in the present study, should offer a broad base from which guidelines of safety might be determined. In the past, it has often seemed that only the painful truth of epidemiologic evidence was sufficient to arouse concern for a potential environmental hazard. It is reas suring then to learn that at present it is unlikely that anyone has been exposed to quantities of airborne glass fibers of a variety and size comparable to levels of asbes tos fibers that are considered to be even minimally haz ardous to man (23-32). These references also suggest that fibrous glass of the conventional type and dimen sion has not been a demonstrable cancer hazard to either man or experimental animal, presumably because it exceeds the dimensional range that is carcinogenic. However, unequivocal evidence regarding the response of man to inhaled panicles of glass would require the study of far more individuals for much longer times and J NATL CANCER INST j VOL. 58, NO. 3, MARCH 1977 OWfiORn rin n DU 009307 CARCINOGENICITY OF FIBROUS C1AS5 59 7 ;hose exposed to higher levels than those currently re ported. Our experiments indicate that the hazard may largely be confined to long fibers of exceptionally fine diameter. Glass fibers of this type have a limited com mercial usage and the hazard of exposure seems remote from the more likely exposure to asbestos. Nevertheless, it would seem prudent to regard all respirable fibers with caution. Hopefully, well-designed inhalation ex periments will establish safety guidelines, and reasona ble monitors and safeguards will make epidemiologic data an impossible achievement. REFERENCES (1) Stanton MF, Wrench C: Mechanisms of mesothelioma induc tion with asbestos and fibrous glass. J Natl Cancer [nst 48:797 821, 1972 (2) Stanton MF: Some etiological consideration's of fibre cardno- genesis. In Biological Effects of Asbestos (Bogovski P. Timbrell V, Cilson j, et al. eds.). Lyon, France. World Health Organiza tion, International Agency for Research on Cancer, Publication No. 8. 1973. pp 289-294 " (3) Speil S, Leineweber J: Personal experiences with malting sam ples of fibres for biological experiments. In Fibres for Biologi cal Experiments (Pelnar PV. ed). Montreal, Canada. Institute of Occupational and Environmental Health. 1974, pp-45-50 (4) Pilgrim HI, Oowo JE: Correcung for extraneous death in the evolution of morbidity or mortaiitv from tumors. Cancer Res 23:45-48. 1963 (5) Armitage P: Statistical Methods in Medical Research. New York. John Wiley & Sons. 1971, pp 376-377, 410-414 16) Cox D: Regression models and life tables, j R Stac Soc [B] 34:187 220, 1972 (7) Mantel N: Chi-square tests with one degree of freedom: exten sions of the Mantel-Haenszel procedure. J Am Stat Assoc 58:690-700. 1963 (8) Allison A: Experimental methods--cell and tissue culture: Ef fects of asbestos particles on macrophages, mesothelial cells and fibroblasts. In Biological Effects of Asbestos (Bogovski P. Timbrell V, Gilson J, etal, eds.). Lyon. France, World Health Organization, International Agency for Research on Cancer, Publicauon No. 8, 1973, pp 89-93 (9) Maroudas N. O'Neal C. Stanton M: Fibroblast anchorage in cardnogenesis by fibres. Lancet 1:807-809, 1973 (10) Raeburn C, Spencer H: Lung scar cancers. Br J Tuberc 51:237 245, 1957 (11) Oppenheimer B, Oppenheimer E, Stout A, et al: The latent period in cardnogenesis by plasucs in rats and its relation to the presarcomatous stage. Cancer 11:204-213, 1958 (12) Brand K. Buoen L. Johnson K. etal: Edological factors, stages, and the role of the foreign bodv in foreign body tumorigenesis: A review. Cancer Res 35:279-286, 1975 (13) Alexander P, Horninc E: Observadons on the Oppenheimer method of indudng tumors by subcutaneous implantauon of plastic films. In Cardnogenesis: Mechanisms of Acuon (Wol- stenholme G, O'Connor M, eds.). Boston, Litde Brown & Co., 1959, pp 12-25 (14) Wacner JC: The pathogenesis of tumors following the intra- pleural injecdon ofjasbestos and silica. In Morpnoiogv of Ex- perimenul Respiratorv Cardnogenesis (Nettesheim P. Hanna MG. Deatherage JW, eds.). Oak Ridge National Laboratory, Oak Ridge. Tennessee. Atomic Energy Commission Sympo sium Series No. 21 j 1970. pp 347-358 (15) Macner D, McDonald AD: Malignant mesothelial tumors. His tologic type and asbestos exposure. N EnglJ Med 287:570-571, 1972 I (16) Churg J, Rosen SH,|Moolten S: Histological characteristics of mesothelioma associated with asbestos. Ann NY Acad Sd 132:614-622. 1965 j (17) Timbrell V, Griffiths DM. Pooley FD: Possible biological im portance of fibre diameters of Souch African amphiboies. Na ture 232:55-56. 1971 (18) Timbrell V: Physical! factors as etiological mechanisms. In Bio logical Effects of Asjsestos (Bogovski P. Timbrell V, Gilson JC. etal, eds.). Lyon. Fjrance. World Health Organization. Inter national Agency for Research on Cancer. Publicauon No, 8. 1973. pp 295-303 | (19) Gilson JC: Asbestos cancer: Past arid future hazards. Proc R Soc Med 66:395-403. 1973 (20) Schepers G: Pulmonary histologic reactions to inhaled fiberglas- piastic dust. Am J Pathol 35:1169-1187, 1959 (21) Gross P: The effects o'f fibrous glass dust on the lungs of animals. In Symposium on Occupational Exposure to Fibrous Glass, vol 2. Rockville. Maryland. National Institute for Occupational Safety and Health. 1974, pp 31-38 (22) Kushner M: The effects of intratracheal instillation of glass fibers of varying size in guinea pigs. In Symposium on Occupa tional Exposure to Fibrous Glass, vol 2. Rockville. Maryland. National Institute for Occupational Safety and Health. 1974. pp 20-31: quoted by. Gross P: The biological categorization of inhaled fiberglass dust. Arch Environ Health 31:101--107* 1975 (23) Nadonal Emission Standards for Hazardous Air Pollutants: Fed eral Register 40(199);48292-4831l. 1975 (24) Enterune PE. DeCoufle P, Henderson V: Respiratorv cancer in relation to occupational exposures among retired asbestos workers. Br J Indust Med 30:162-166, 1973 (25) Briush Occupational kygiene Sodety Committee on Hygiene Standards: Hygiene standards for airborne amounts of asbes tos dust. Ann Occup Hyg 16:1-5, 1973 (26) BjureJ, Soderholm B. Widimsky j: Cardiopulmonary function studies in workers dealing with asbestos and glasswool. Thorax 19:22-29, 1964 j (27) Newhouse ML Asbestos in the work place and the community. Ann Occup Hyg 16:97-102, 1973 (28) Hill JW, Whitehead WS, Cameron JD. et al: Glass fibres: Absence of pulmonary hazard in production workers. Br J Indust Med 30:174-179, 1973 (29) Enterune P, Henderson V: The health of retired Fibrous glass workers. Arch Envirpn Health 30:113-116. 1975 (30) Corn M. Savsone E: Determination of total suspended particu late matter and airborne fiber concentrations at three fibrous glass manufacturingfadliues. Environ Res 8:37-52, 1974 (31) Gross P, Tuma J, deTreville R: Lungs of workers exposed to fiberglass. Arch Environ Health 23:67-76. 1971 (32) Skulte P, ed: Occupational exposure to fibrous glass. In Pro ceedings of a Symposium of the National Institute for Occupa tional Safety and Health. DHEW Publication No. (NIOSH 76 151). Washington, D.C., U.S. Govt Print Off, 1976 DUP 0809741 VOL. 58, NO. 3, MARCH 1977 J NATL CANCER INST DU 009308 msemam | i Ficures 1-17.--Silhouettes of samples of the 17 types of glass fibers used. Figures are Numbered to correspond to the experiment number anc rank for probability of inducing pleural sarcoma. Figures A are all at x 750; figures'B are all at x 3.200. J598 STANTON' ETT AL. DU 009309 DUp 0809742 STANTON T AL. DU 009310 DUP 0809743 DUP 0809744 STANTON ET AL. DU 009311 n smra 85 DUP 0809745 DU 009312 Figure 18.--Typical pleural response to noncarrinogenic glass Fibers. Note phagocytic activity around Iarge*diameter vehicle glass and minute particles of glass, from expt M6S-#I6. H & E. x 100 j Ficure 19.--Typical preneoplastic pleural response to carcinogenic glass fibers. Note numerous fine, long fibers in dense acellular matrix and patchy character of the granulomatous response, from expt KL~#2. H Sc E. x 100 | 602 STANTON ET AL. C <t DU 009313 i C DUP 0809747 i -. - -^**--*~*'^Y irin r fy~T Ficures 20-23.--Pleural sarcomas with characteristics that range from spindle to pleomorphic to tubulopapillarv epitheloid cell patterns. See also E ' "fof' ^'Ure ^ *s 1 PaP`UarT pleur*l implant from pleural sarcoma of contralateral pleura composed predominentiy of spindle cells. H Sc STANTON ET AL. 603 DU 009314