Document em6KjVRr2wZr3j54J583mNGkm

r ri UNVIKONMI.N1AI. RliSI.AKCll 37, 3M-372 l!9S5l Mineral Particles, Mineral Fibers, and Lung Cancer1 Andrew Churg and Barry Wiggs Department of Pathology. University of British Coltonhia. Vancouver. British Columbia V6T IWS. Canada Received June 15. I9K4 The total fibrous and nonfibrous mineral content of the lung has been analyzed in a series of 14 men with lung cancer but no history of occupational dust exposure, and in a series of 14 control men matched for age. smoking history, and general occupational class. The lung cancer patients had an average of 525 = 369 x 10* exogenous mineral particles and 17.4 - 19.6 x 10* exogenous mineral fibers/g dry lung, while the controls had averages of 261 * 175 mineral particles and 4.7 ~ 3.2 x 10* mineral fibers/g dry lung. These differences are statistically significant for both panicles and fibers. Kaolinite. talc. mica, feldspars, and crystalline silica comprised the majoriiy of panicles of both groups. Ap proximately 909? of the panicles were smaller than 2 urn in diameter and approximately 609? smaller than I pm: the mean panicle size in the cancer group was l.l - 0.2 pm and in the control group 1.3 i 0.2 pm. In both groups, patients who had smoked more than 35 pack years had greater numbers of panicles than patients who had smoked less than 35 pack years. It is concluded that, in this study, lungs from patients with lung cancer had statistically greater numbers of mineral panicles and fibers than lungs from controls, and that smoking influences total long-term retention of panicles from all sources, t I9< Acs- demk Press. !nc. ai. >976)-1 particlcs- * cigars* sr ulate fracti ^etabolisn mate carcii Mineral; number pn castrated t most adsot recov . 198? e -l0P. of rai specif I to vs 9 hu has c ;e INTRODUCTION Except for a few specific species such as asbestos, mineral particles are noi generally considered to be carcinogens. However, a number of circumstantial observations suggest that mineral particles of a variety of types could play a role in the genesis of lung cancer in humans. Epidemiologically, some studies have found an association between lung cancer rates and concentrations of particulate air pollutants (Doll. 1978; Hitosugi. 1968; Vena, 1982). and it has been suggested that air pollution is the factor responsible for the differences in lung cancer rates between urban and rural areas (Doll. 1978). This rate differential is probably confined to cigarette smokers (Doll. 1978; Hi tosugi. 1968; Vena, 1982). It should be noted that this association is by no means universally accepted (Hammond and Garftnkel, 1980). Experimentally, it has been demonstrated that intratracheal instillation of car cinogens such as benz(a)pyrene along with an "inert" (i.e.. noncarcinogenic) mineral particle such as iron oxide greatly increases the rate of lung tumors over that found with administration of the carcinogen alone (Saffiotti et ah. 1972). Some other minerals such as titanium oxide, and carbon show similar effects, but not every mineral dust which has been tried is active in this system (Stenback et 1 Supported by a grant from the National Cancer Institute of Canada, and grant MA7820 from the Medical Research Council of Canada. 0013-9351/85 S3.00 Copyrifhi f I9K< hv Academic Press. Ins. All nfhlk of reproduction in any form reserved 364 diseases. : various diof human Autopsy Universit} couver Gt met the fc all cases mineral d dardized < no other clinical ot 10 years), category cancer or tained fre All lun 511961 0291 MINr.KAI I'AKTICU-S AND M'N(. CANCTK 365 .ung Cancer1 os oiumbut. ado been analyzed in a scries usi exposure, and in a neral occupational class, cenous mineral panicles he controls had averages fibers/g dry lung. These s. Kaolimie. talc. mica, les of both groups. Ap* neier and approximately p was I. I i 0.2 im and tad smoked more than 35 had smoked less than 35 its with lung cancer had lungs from controls, and n all sources, t Aca- nineral particles are not umber of circumstantial f types could play a role ion between lung cancer >11. 1978; Hitosugi. 1968; is the factor responsible I rural areas (Doll, 1978). mokers (Doll. 1978; Hisociation is by no means icheal instillation of car(i.e., noncarcinogenic) rate of lung^mors over : (SafTiotti et al., 1972). show similar effects, but this system (Stenback el a. and grant MA7820 from the 1976). Since cigarette smoke contains both chemical carcinogens and mineral particles, a similar effect might occur in humans, and it has heen shown that cigarette smoke particulates (Luhawy and Isaac. 1980). as well as a crude partic ulate fraction collected from the air (Warshawsky et til., 1983). slow the overall metabolism of benz(n)pyrenc and redirect it toward the formation of the proxi mate carcinogens. Mineral particles, whether derived from cigarette smoke or from the ubiquitous number present in the air. also tend to adsorb carcinogens, and it has been dem onstrated that the smallest (and hence most easily respirable) particles are the most adsorptive (Nalusch et al.. 1974). Benzo(n)pyrene and nitrosamines have been recovered from atmospheric mineral particles (Natusch el al.. 1974; Kneip (t al., 1983). Mineral particles also appear to adsorb radioactive materials, such as the 2l0Po present in cigarette smoke, and hence might deliver a carcinogenic dose of radiation to the bronchial tree (Martell. 1974; 1983). A specific class of mineral particles, namely fibers, appears to have special carcinogenic effects which relate to size and shape, especially small size and high ' length to width (aspect) ratio (Stanton et al., 1981). These effects are clearly seen in both humans and experimental animals in the genesis of mesolhelial tumors; whether fibrous shape is important for inducing other types of tumors is not clear. A fairly large body of experimental data has been accumulated concerning the acute phases of particle deposition in the human lung. A particularly interesting observation in this regard is the study of Schlesinger and Lippman (1978). in which a cast of the human bronchial tree was created and the deposition of particles examined. Schlesinger and Lippman showed that there was a good cor relation between the number of particles deposited in different bronchi and the known relative cancer incidence in the different bronchi. Little is known aboul the actual mineral content of the normal human lung, nor is there any information on regional differences in various sites, or in different diseases. Such information might shed light on the role of mineral particles in various diseases. In this study we have begun to examine the particulate content of human lungs in smokers with and without lung cancer. MATERIALS AND METHODS Autopsy lungs for this study were obtained from the autopsy services of the University of British Columbia Health Sciences Centre Hospital and the Van couver General Hospital. For the test sample. 14 male cases were selected which met the following criteria: (1) a documented history of lung cancer (confirmed in all cases by pathologic examination); (2) no history of occupational or other mineral dust exposure as determined by interviews with relatives using a stan dardized questionnaire: (3) details of smoking history available from relatives; (4) no other pulmonary disease (except chronic obstructive lung disease) by either clinical or pathological examination. These 14 cases were matched by age (within 10 years), sex. smoking history (within 10 pack years), and general occupational category (white collar, blue collar) to 14 control cases who did not have lung cancer or a history of dust exposure. Information on the control cases was ob tained from relatives using the same type of questionnaire. All lungs were Fixed in formalin. For mineralogic analysis of each case, two 366 CHURG AND W1GOS Croup TABLE I Aut and Smokinc Match Age tyears) Smoking (pack years) Comparison oi Cancer Control A.T = II hj e: 7 4.^ * 3V 41 i 21 Compariv Toial panic Total fibers samples of wet formalin fixed lung were taken from an upper lobe: one sample consisted of an approximately 4- to 5-g strip of pleura and underlying 0.5 cm of lung tissue. A second sample was taken at least 3 cm deep to the pleura. Data from the two samples were pooled for each case. An additional adjacent piece of tissue was taken and dried to constant weight to allow expression of results in terms of particles/g dry lung. Only grossly normal parenchyma was used. Mineral particles and mineral fibers were isolated using a modification of the technique which we have employed previously (Churg. 1983). In brief, this con sists of dissolving the lung in bleach, treating the remaining sediment with hydrogen peroxide, and collecting the mineral particles on a membrane filter. The particles are then transferred to coated electron microscope grids. Because the " Exogenous pai Table 2 sumn 1 fibers between broken down b; which arc clear enous origin (si erals analyzed a ) most likely repi v The mean num fibers was signi concentration of particles in lung is much higher than the concentration of as 0.025 for fibers) bestos or other mineral fibers, two different aliquots of the digested sample are prepared: a very high dilution sample for counting particles and an approximately As shown in' silica) account I 10- to 50-fold more concentrated sample for counting fibers. laneous group a For particle identification, approximately 200 sequentially encountered parti- 1 vermiculite. bio cles were counted, measured, and identified by morphology, electron diffraction, Table 4 docun and energy dispersive X-ray spectroscopy for each sample site. Only particles imately 6Vc of t with one dimension of 0.5 jim or greater were counted in this study. For fibers, approximately 50 sequentially identified fibers (defined as particles with roughly parallel sides and an aspect ratio of 3:1 or greater) were counted for each site. Major Tyh: Only fibers of length 0.5 p.m or greater were counted. Raw counts of particles and fibers were converted to particles and fibers/g dry lung using an algorithm relating number of grams of sample, and number of electron microscope grid Mineral squares counted. A total of approximately 11,000 mineral particles and 3000 min Exogenous parti. eral fibers was counted in this study. Control samples were prepared for each case by running the entire preparatory procedure but omitting tissue. Control specimens contained rare short fibers of chrysotile, and scattered pieces of talc, mica, and crystalline silica. The concen tration of these contaminants were so low that subtracting them from the observed Kaolinite Talc Mica Feldspars Silica Aluminum i values did not produce any significant change in the calculated numbers. Addi tional controls were prepared with the formalin which was used for fixation; Titanium Miscellaneous' essentially no particles were found in the formalin. Total Statistical comparisons were made using standard r tests and non-parametric tests (Mann-Whitney U Test) on the concentration data. ( Endogenous part Iron RESULTS I Apatite Table 1 shows the ages and smoking histories for the cancer and control groups. J * Values are given For each group. 11 patients had blue collar occupations and 3 patients had white I b Includes panicle ' NC. not calculat. collar occupations. ! 511961 0293 mineral particles and lung cancer 367 TABLE 2 orComparison Totai Mineral Partici.es ano Fibers in Lung Cancer Cases and Controls Smokinp (pack years) Mean concentration 2 SD (counts x IO*/g dry lung) 45 = 39 41 * 2\ Comparison Total particles" Total fibers Cancer 325 2 367 17 2 20 Control 261 2 175 523 P 0.01 0.025 'cr lobe: one sample * Exogenous panicles only. See lexi. inderlying 0.5 cm of to the pleura. Data Table 2 summarizes the comparison for total mineral particles and total mineral rial adjacent piece of fibers between the two groups, and Tables 3 and 4 give the particle and fiber data ression of results in broken down by mineral type. The data has been divided into particles and fibers ma was used, which are clearly of exogenous origin, and those which are most likely of endog i modification of the enous origin (see Discussion). The endogenous minerals included apatite, min i). In brief, this con- erals analyzed as pure iron, and particles analyzed as iron with phosphorus, which g sediment with hy- I most likely represent iron deposited on other minerals or on organic structures. nembrane filter. The The mean number of exogenous particles and the mean number of exogenous : grids. Because the fibers was significantly higher in the cancer group (P < 0.01 for particles; P < concentration of as- 0.025 for fibers). digested sample are As shown in Table 3, kaolinite, talc, mica, feldspars, and silica (i.e., crystalline nd an approximately silica) account for about 75 to 85% ofthe total particles. Included in the miscel laneous group are spinels, calcium compounds (? oxalate or oxide), chlorite and encountered parti- 1 vermiculite, biotite. tin. and particles which could not be identified. electron diffraction, Table 4 documents similar data for fibers. Asbestos fibers constituted approx site. Only particles imately 6% of the total for the cancers and approximately 17% of the total for is study. For fibers, rticles with roughly unled for each site, TABLE 3 Major Types of Mineral Particle in the Lungs of Cancer and Control Cases1' counts of particles using an algorithm an microscope grid tides and 3000 min- : entire preparatory rare short- fibers of silica. The <pnceni from the observed ed numbers. Addi- used for fixation; .nd non-parametric Mineral Exogenous panicles Kaolinite Talc Mica Feldspars Silica Aluminum Titanium Miscellaneous* Total Endogenous panicles Iron Apatite Cancer cases Mean 2 SD % of total 135 2 116 114 2 161 103 2 76 58 2 54 55 2 57 26 2 25 23 2 57 II (NCy 525 2 369 25.8 21.7 19.7 11.0 10.4 4.9 4.3 2.1 225 2 185 166 2 153 Control cases Mean 2 SD % of total 56 2 61 28 2 20 61 2 87 23 2 13 42 2 34 14 2 21 8 2 13 29 (NO 261 2 175 21.4 10.7 23.4 8.8 16.1 5.4 3.0 11.1 172 2 198 152 2 130 md control groups, patients had while * Values are given as panicles x lo*/g dry lung. * Includes panicles which could not be identified. ' NC. not calculated. 511961 0294 OU'K(. AM) WKiCS TAHI.K 4 Maiok Tvi*i s <>i Minihai Fihi rs is tmi l.t'Nt.s 01 Canuk and Coni him Cams" Control casrs 95 of total Mean 2 SD 95 of total Cl Co Chrysotile Tremolite. etc.* Amosile & crocidolite Talc Kaolinile Mullite Silica Feldspars Titanium Aluminum Miscellaneous' Total Endogenous minerals Iron Apatite 0.4 2 0.4 0.7 2 0.7 0.02 * 0.02 l.l 2 1.4 4.7 2 8.1 0.7 2 1.2 3.9 2 7.6 1.5 2 3.7 0.3 2: 0.3 1.2 e 3.1 2.9 (NC)1' 17.4 2 19.6 4.0 2 8.4 0.1 2 0.1 2.2 4.0 0.1 6.3 27.0 4.0 22.4 8.6 1.7 6.8 16.7 99.8 * Values are given as numbers of fibers x KT/g dry lung. * Tremolite. actinolite. anthophyllite. ' Includes particles which could not be identified. 4 NC. not calculated. 0.2 2 0.3 0.6 2 0.7 0.01 2 0.01 0.5 2 0.5 0.7 2 0.5 0.4 2 0.6 0.3 2 0.3 0.6 2 0.4 0.5 2 0.6 0.2 r 0.2 0.7 (NC) 4.7 2 3.2 0.8 2 0.6 0.6 2 0.6 4.3 12.8 0.2 10.6 14.8 8.5 6.3 12.8 10.6 4.3 14.a 100 the controls. Talc, kaolinite. mullite. feldspars, silica, rutile (TiCM and aluminum made up the majority of the identifiable exogenous fibers. Table 5 shows the overall size distribution for the particles in the cancer and control cases. The mean sizes and size distributions are roughly similar for both cancer and control groups. Similarly, the mean fiber sizes for the control group were: length. 1.5 1.3 p.m; width. 0.3 * 0.3 p.m: and aspect ratio. 6.5 6.9; and for the cancer group; length. 1.4 1.5 p.m: width. 0.2 0.2 gm; and aspect ratio. 8.0 7.9. Table 6 compares overall numbers of particles in those who smoked more than 35 and those who smoked less than 35 pack years. In both cancer and control groups, the heavier smokers had more particles, but this difference was not sta tistically significant in either group; however, it approached significance for the groups combined (see Discussion). Cancer Control TABLE 5 Particle Size Distribution in Cancer and Control Cases Mean size - SD (pm) l.l 2 0.2 1.3 2 0.2 Geometric mean size (pm) 0.9 l.l 95 in size range -------- <1 pm 1-2 pm >2 pm 67.0 29.5 3.4 58.4 28.2 13.4 In thil particle i a compi cancer The d `normal particle! of the silica, m of the ft this cord study im and pari our anal) a varietj containii phenomt particles We ha general | that stud this diffi counting report), nonsmol Nonet types an spheric S such as mineral rable to) (2) Then pollutant sotile as I dry lung.; reflect th the San I Francisci ) I 511961 0295 J If o 9* V> It t-J and Control Cam s" Control cases lean r SD *7 of lotal : = o.? 6 = 0.7 1 =: 0.01 5 = 0.5 7 r 0.5 7 = 0.3 6 r 0.4 5 2 0.6 : r 0.2 7 INC) 1.7 2 3.2 4.? 12.K 0.2 10.6 14.8 8.5 6.3 12.8 10.6 4.} 14.8 100 8 = 0.6 6 = 0.6 ile (TiO;) and aluminum s. licles in ihc cancer and roughly similar for both es for the control group aspect ratio. 6.5 6.9; 2 2: 0.2 p.m; and aspect who smoked more than 'Oth cancer and control difference &$ not sta lled significance for the .OL Cases 77 in size range i 1 -2 nm >2 |im 29.5 3.4 28.2 13.4 Group Cancer Control MINERAL PARTICLES AND LUNG CANCER 369 TABLE 6 Toi ai Particles vs Smokinc, History Less than 35 pack years More than 35 pack years 216 i 190 626 * 322 308 s 159 DISCUSSION In this paper we have presented data in two areas: baseline values for mineral particle content in the lungs of workers without occupational dust exposure; and a comparison of the mineral content in lungs of patients with and without lung cancer. The data on mineral content show that the upper lobes of the lungs of a ` normal" smoker contain an average of about 260 x 10h exogenous mineral particles and about 4.7 x 10h exogenous mineral fibers/g dry lung. The majority of the particles are composed of kaolinite. talc. mica, feldspars, and (crystalline) silica, with lesser amounts of rutile (TiO:). and aluminum oxide. The components of the fiber burden are similar, with the addition of about 179f of the fibers in this control population composed of asbestos. Because we were interested in this study in the effects of particles which were definitely of exogenous origin, apatite and particles analysing as iron or iron with phosphorus have been excluded from our analyses. Deposition of iron or calcium phosphate (Brody and Hill. 1982) on a variety of other organic and inorganic structures, as well as generation of ironcontaining particles as a result of hemorrhage into the lung, are well recognized phenomena, and there is no easy way to determine what proportion of such particles might be of exogenous origin. We have previously published some data on the "normal" fiber burden in the general population of San Francisco (Churg. 1983); the mean number of fibers in that study was about 1 x 10'1 (translated into fibers/g dry lung). The reason for this difference most likely relates to the fact that, in the present report, we are counting fibers down to 0.5 p.m in length (compared to 1 urn for the previous report), and to the fact that the previous report contained both smokers and 'nonsmokers and males and females (see below). Nonetheless, two general conclusions can be drawn: (I) The majority of fiber types are more or less the same in both cities, and most likely reflect the atmo spheric burden of mineral fibers in the two areas. This is certainly true of minerals such as mullite, which appear to be good markers of coal fly ash pollution. The mineral particle types found in the Vancouver population are in general compa rable to published reports (Berry el al., 1980) of minerals found in the atmosphere. (2) There are some differences which probably relate to the specific atmospheric pollutants encountered in the two cities. For example, the mean value for chrysotile asbestos in the San Francisco population was approximately 1.0 x 106 g/ dry lung, whereas in the Vancouver population it is only 0.2 x lO6. This may reflect the fact that many large outcrops of exposed serpentine rock exist within the San Francisco Bay area. As measured by lung burden, the population of San Francisco also is exposed to attapulgite and gypsum, whereas these minerals are 370 CHUKG AND WIC,C,S virtually absent in the population of Vancouver. The reason for these differences particles and the source(s) of exposure in San Francisco are not clear. postulate Analysis of the nonfibrous particle size distribution in our cases reveals that smoking about 90% of the particles are smaller than 2 pm in greatest diameter, and 60 to exposure 70% are smaller than I pm in greatest diameter. The mean particle size of roughly cinoma. i I pm which we observed in both cancer and control patients is slightly smaller Comps than that expected from studies of maximal acute deposition in human volunteers the canc< (generally in the range of 2 to 4 pm. reviewed in Lippmann et al.. 1980). This a four tin difference may reflect the size distribution of the atmospheric burden of mineral diffcrenci panicles, or may indicate that larger particles are preferentially cleared, hence can be ol the long-term particle burden in humans is not accurately predicted by acute exposure deposition experiments. An interesting fact which emerges from this study, and one which requires patients I further examination, is the suggestion of dependence of particle load on cigarette also be it smoking. In both groups, those patients smoking more than 35 pack years had of one sr greater particle burdens than those smoking less than 35 pack years. Taking the | tional cx| two groups as a whole, the increase was present for most mineral types, although ' (2) T most marked for kaolinite and talc. This difference is not significant for the sep there is n arate test and control groups, a not surprising observation given the small sample in both { size and the evident variability of particle burden. However, if the cancer and histories control groups are combined, the differences for total particles approach signifi (3) H cance, and are highly significant (P < 0.01) if only one patient with very few pack (for exam, years but a large mineral burden is excluded as an anomalous case. Obviously dose, an < this observation needs to be checked with more cases and with nonsmokers. well inert Two explanations can be offered for this phenomenon. First is that it reflects the cancf increased deposition from increased exposure: i.e.. the number of particles pre unknown sented to the lung is, in a crude way, proportional to the number of cigarettes The po smoked. Since the increases in concentrations of particles with increasing noted (sc smoking were not limited to kaolinite, a cigarette smoke-derived particle (Brody nism(s) is and Craighead 1982), but were seen for most of the particle types present, and terials) bt since the mean size of particles derived from cigarette smoke is only about 0.2 undefined p.m (Keith and Derrick 1960), direct deposition from cigarette smoke is unlikely tides pre: to be very important. A more likely explanation is decreased clearance of particles inhaled from all sources. Cigarette smokers in general have longer clearance times than do nonsmokers for particulates in experimental studies (Bohning et al., 1982. Cohen el al., 1979, and see Lippmann et al., 1980 for a review), and Bohning et al.. (1982) Berry. J. P particle Bohning. D have, experimentally, demonstrated a strong correlation between clearance time and amount of smoking. However, there is a large individual variation in clearance times, and our data suggest that there is probably a marked variation in total long-term retention as well. Nonetheless, it appears from this study that smoking and im; Brod). A. I cigureit Brod). A. t and cor is probably a major factor in long-term particle accumulation within the lung. This ( Churg. A. i could well be important in the genesis of lung cancers, since experimental studies using the inert mineral dust plus carcinogen systems (see Introduction) have shown that particles which are cleared more slowly produce more cancers than l8V-:ts Cohen. D.. lung .S: Doll. R. (ij Hammond. Prrvcrn 511961 0297 eason for these differences at clear. n in our cases reveals that reatest diameter, and 60 to lean particle size of roughly patients is slightly smaller isition in human volunteers ppmann el aL. 1980). This ospheric burden of mineral eferentially cleared, hence -irately predicted by acute .. and one which requires of particle load on cigarette ire than 35 pack years had i 35 pack years. Taking the ost mineral types, although not significant for the seption given the small sample lowever. if the cancer and I particles approach signiflpalient with very few pack momalous case. Obviously s and with nonsmokers, non. First is that it reflects ne number of particles pre0 the number of cigarettes particles with increasing )ke-derived particle (Brody particle types present, and te smoke is only about 0.2 cigarette smoke is unlikely >f particles inhaled from all :arance times than do non.ning ei a(.,A982. Cohen el . and Bohnmg ei a!., (1982) on between clearance time idual variation in clearance 1 marked variation in total om this study that smoking ilation within the lung. This . since experimental studies is (see Introduction) have oroduce more cancers than MINERAL PARTICLES AND LUNG CANCER 371 particles cleared rapidly (Stenback and Rowland. 1978). Analogously, one could postulate that enhanced retention of carcinogenic particles because of cigarette smoking would be extremely important in the genesis of lung tumors: asbestos exposure, which combined with smoking greatly increases the incidence of car cinoma, may be a case in point. Comparison of the cancer and control cases produces the intriguing result that the cancer cases show about twice the content of exogenous particles and about a four times the content of exogenous fibers compared to controls, and both these differences are statistically significant. A number of hypothetical explanations can be offered for this observation: (1) We have missed one or more occupational exposures in the cancer group. Although possible in the odd case (and equally possible for the control group), this theory would not explain why 10/14 cancer patients had higher total panicle numbers than their matched controls. It would also be incompatible with our failure to find individual cases with huge numbers of one specific mineral type, a finding one might expect with an occult occupa tional exposure. (2) The cancer patients have actually smoked more than the controls. Again, there is no reason to believe that to be true, and the fact that the heavier smokers in both groups had more particles than the lighter smokers suggests that the histories are generally accurate. (3) However, it is entirely possible that the cancer patients smoked differently (for example, smoked down to the butt), thereby increasing their actual cigarette dose, an effect which, extrapolating from the data of Bohning el til.. (1982). might well increase total particle load. Alternatively, one is left with the hypothesis that the cancer patients retain more particles than noncancer patients for reasons unknown, but not necessarily related to smoking. The possible roles of mineral particles in the genesis of lung cancer have been noted (see Introduction); our data suggest that whether the important mecha nism^) is adsorption and concentration of carcinogens (including radioactive ma terials) by particles, redirection of the metabolism of carcinogens, or some as yet undefined process, there is an important relationship with total number of par ticles present in lung tissue. REFERENCES Bern. J. P.. Henoc. P.. and Galle. P. (1980). Chemical and crystallographic microanalysis of fine particles in ihe urban atmosphere. Environ. Res. 21, 150-164. Bohning. D. E.. Atkins. H. L-. and Cohn. S. H. < 1982). Long-term particle clearance in man: Normal and impaired. Ann. Occ. Hyg. 26. 259-271. Brody. A. R.. and Craighead. J. E. 11975). Cytoplasmic inclusions in pulmonary macrophages of cigarette smokers. Lab. Invest. 32, 125-132. Brody. A. R.. and Hill. L. H. (1982). Interstitial accumulation of inhaled chrysotile asbestos fibers and consequent formation of microcalcifications. Amer. J. Pathol. 109, 107-114. Churg. A. (1983). Nonasbestos pulmonary mineral fibers in the general population. Environ. Res. 31, 189-200. Cohen. D.. Arai. S. F.. and Brain. J. D. (1979). Smoking impairs long term dust clearance from the lung. Science {Washington. D.C.) 204, 514-517. Doll. R. (1978). Atmospheric pollution and lung cancer. Environ. Health Perspect. 22, 23-31. Hammond. E. C.. and Garfinkel. L. (1980). General air pollution and cancer in the United States. Prevent. Med. 9, 206-211. 511961 0298 l 9 372 CIlUKtj AND WICKiS Hilosugi. M. (1968). Epidemiological sludy of lunp cancer with special reference in ihe efTeci of air pollution and smoking habit. Inst. Public Health Bull {JapanI 17, 237-256. Keilh. C. H.. and Derrick. J. C. (19601. Mcasurcmcnl of (he particle size distribution and concentra. lion of cigarette smoke by the conifugc. J. Colloid Sci. IS, >40-146. Kneip. T. J.. Daisey. J. M.. Solomon. J. J.. and Hershman. R. J. (198.1). Ai-Nilroso compounds: Ev. idence for their presence in airborne particles. Science (Washington. D.C.I 221, 104.'-1047. Lippman. M.. Yeates. D. B..and Albert. R. E. (1980). Deposition, retention, and clearance of inhaled particles. Bril. J. Indus!. Med. 37, .1.17-162. Lubawy. W. C.. and Isaac. R. S. (1980). Acute tobacco smoke exposure alters (he profile of meiaboliles produced from benzoUilpyrene by the isolated perfused rabbit lung. Toxicologv 18, .17-47 Martell. E. A. (1974). Radioactivity of tobacco trichromes and insoluble cigarette smoke particles. Nature iLondon) 249, 215-217. Martell. E. A. (1983). Radiation at bronchial bifurcations of smokers from indoor exposure to radon progeny. Prat . Natl. Aead. Sci. USA 80, 1285-1289. Mossman. B. T.. Eastman. A.. Landesman. J. M.. and Bresnick. E. (19831. Effects of crocidolile and chrysotile asbestos on cellular uptake and metabolism of benzoin (pyrene in hamster tracheal epithelial cells. Environ. Health Pcrspcct. 51. 331-335. Natusch. D. F. S.. Wallace. J. R.. and Evans. C. A. (19741. Toxic trace elements: Preferential con centration in respirable particles. Science (Washington. D.C.I 183 . 202-204. Saffiotti. U.. Montesano. R.. Sellakumar. A. R.. Celis. F.. and Kaufman. D. C. (19721. Respiratory tract carcinogenesis in hamsters induced by benzoUilpyrene and ferric oxide. Canter Rc\. 32, 1073-1081. Schlesinger. R. B.. and Lippmann. M. (1978). Selective particle deposition and bronchogenic carci noma. Environ. Res. 15. 424-431. Stanton. M. F.. Layard. M.. Tegeris. A.. Miller. E.. May. M.. Morgan. E.. and Smith. A. 11981). Relation of particle dimension to carcinogenicity in amphibolc asbestoses and other fibrous min erals. J. Natl. Canrer Inst. 67, 965-975. Stenback. F.. Rowland. J.. and Sellakumar. A. (1976). Carcinogenicity of benzotutpyrene and dusts in (he hamster lung (instilled intratracheally with titanium oxide, aluminum oxide, carbon, and ferric oxide). Oncology 33. 29-34. Stenback. F.. and Rowland. J. (1978). Role of particulate size in the formation of respirators tract tumors induced by benzo()pyrene. Ear. J. Cancer 14, 321-326. Vena. J. E. (1982). Air pollution as a risk factor in lung cancer. Anter. J. Epidemiol. 116. 42-56. Warshawsky. D.. Bingham. E.. and Niemeier. R. W. (1983). Influence of airborne particulate on the metabolism of benzo()pyrene in the isolated perfused lung. J. Toxicol. Environ. Health 11, 503-517. | \x IKnssil >; Reca Si Corn* in dnnli labeled i izalion < cudmiM ingeMu* lap u;ti< E-IgG. I while be of cudni was si ill cessatici Cadmiui ingested ot feciion by i has been a and partici phagocyto: 1976). imp 1981). and observed i Chronic e> type hypet etui.. 1981 the ability 1980). The mot a/.. 197411 of the effc 1974). the cvtic and i ' Present a. ester. Roche- I 511961 0299 mgust 1984 onal (B) CT less n tri lateral venxxmal vastot enhance nnstrason. poralorocipn of path- foamy hisneedMke d crystals: jarit cab. lemosjOeon n, X40. 403 Characteristics of Breast Cancer in an Incident Cancer Population he cere- caJafied There is era) catngiomaRbroma- thechointracerxilar abxus(es), ie usual appear ithde. 'thin the na, and handrtg hey are I if they d plexus anuioma 19:280- 3. Comia (xan-%b 7-552 lthomay Hand41 lutomas uroaurg -* I choroid Atlas of Armed Elizabeth L. Schmitt''2 Barbara Threatt1 The effectiveness ot ram mammography i* a sourca of concern to radiologists because neither the riba nor retromammary spce la included on the lima In good quality examinebona. One hundred seven incident cancers were detected in 10,034 aetf-referred women followed at the University of Mchigan Breast Cancer Detection Demonstration Project (UM-BCDOP) for 5 years. These cancers were analyzed for location on the Dm, method ot detection, size, histology, and the number of films required for detection. Mammography alone detected 52 (49%) of the cancers, whereas physical examination alone detected 15 (14%). The other 40 cancers were detectable on both examinations. Ail ot the 92 cancers detected by mammography were visible in both the madMateral and the craniocaudal views. Only 22% of these cancers were within 1 cm of the posterior edge of the film in the mediolateral view, 18.7% were within 1 cm of the posterior edge in the craniocaudal view. These incident cancers were smaller, and fewer were located within the posterior 1 cm of the breast than in a similar study using the UM-BCDOP prevalent cancer population. Mammography consistently detected cancer in the breast, regardless of tumor size, histologic type, or location within the breast Received December 19, 1983: accepted after revision March 27.1984. ' Department of Radntogy, University of Michi gan Medical School. Ann Arbor. Ml 48109. * Department ot Radiology. Wayne County Gen eral Hospital. 2345 Memman Rd.. Westland. Ml 48185. Address reprint requests to E. L. Schmitt. AM 143:403-408, August 1964 0361-803X/84/1432-0403 American Roentgen Ray Society The reliability of film mammography as an effective mammographic procedure has been persistentty challenged because neither the ribs nor the retromammary space is included on the films. State-of-the-art film mammography requires vigorous compression of the breast to delineate breast structures. Such compression precludes imaging of the ribs due to the curvature of the thorax. A population screened over a period of 48 months and followed since that time, using a mammographic technique that depicted neither the ribs nor retromammary space, was used to evaluate this issue. The 10,034 self-referred women followed at the University of Michigan Breast Cancer Detection Demonstration Project (UMBCDDP) were used for analysis. These women were screened yearly from the initiation of the project in 1974 and have been contacted annuafly since completion of screening in 1980. Complete information concerning medical history and follow up is available for each woman from 1974 through the present The prevalent cancers (79) and the interval cancers (16) were excluded from this study. This group of incident cancer screenees was an ideal group in which to evaluate the effectiveness of mammographic imaging by use of the film-screen method. The study included evaluation of the size, location, method of detection, extent of invasiveness, and number of films per examination. Subjects and Methods The University of Michigan BCDOP rescreened 10,034 self-referred women from June 1975 through December 1980. These women were initially screened between September 1974 and July 1976. Each patient was examined by mammography and physical examination. Independent interpretations of each tasting method were made, and the examination findings were correlated. The correlated results were reported to the women and their physicians. Mammography was performed using a dedicated Picker or a dedicated Siemens mam- ptffc ftmA X ofogij J4V.4o5-> 511961 0300 '984 Ajn 143. August 1984 less olos at tive >ive 3%) i by 3th;inq ted ied INCIDENT BREAST CANCER 405 ia !-5 de ers the am :ed ITS 76 Fig. 1.--Measumg technique on lateral (A) and craraocaudal (B) mammograms. jre A to srs Kir TABLE 2: Mammographic Abnormalities in the Mammographically Detected Cancers, University of Michigan TABLE 4: Method of Detection of Incident Cancers by Age, University of Michigan BCOOP (1975-1980) 9 BCDOP ter Age (years) trs 40 ') Inodent Cancan 197S-1980 Abnormeity No. (%) Prevalent Cancers 1974-1976 Schmitt 4 Thread (1J, No.(%) Worts [2], NO.(%) Method Physical only.......................... Physical and mammography . <50. no. (%) >50. no. (%) 8* (18) 19 (43) 7(11) 21 (33) Cancan |%) 15 (14) 40 (37) he Microcalcification only 34(37.0) 21 (28) 69(15) Mammography only.............. 17 (39) 35(56) 52 (49) he Mass with calcification 6 (6.5) 9(12) 162(35) Total.................................... 44 (41) 63(59) 107 (100) Pj Mass/density * Includes one woman who refused mammography. he distortion............ 52 (56.5)46(60) 231 (50) al Total 92- 76 462 to * i $ cancers wot undetected by mammography. nduAg one in a pattern who refused TABLE 5: Location of Lesion on Mammogram, University of Te Michigan BCOOP (1975-1990) ri le TABLE 3: Age Distribution of Incident Cancer Screenees, Breast Thrd NO. (%) Crarocaudai NO. (%) MecWiterh te University of Michigan BCOOP (1975-1980) rs ie Agdyvan) No. of Woman Examned % VMth Cancer NO. Of Cancers (%} Posterior................. Middle...................... Anterior................... 44 (47.8) 40 (43.5) 8 (8.7) 32 (34.8) 47 (51.1) 13 (14.1) >y IT 35-39 ............... >664 0.06 1 (0.9) Total................... 92 (100) 92 (100) 3jr 40-49 ............... 50-59 ............ 3876 3114 1.0 1.2 43(40.2) 38 (35.5) Note.--ToiH detected 107, 15 <14%) I undetected by mammography. >y'P 60-69 .............. 70+ ................. 1206 174 1.6 2.S 20(18.7) 5 (4.7) the incident cancers were within 1 cm of the posterior edge e Total............ 10.034 107* of the film on the mediolateral view. This contrasts with our ie * There were 105 women with 107 cancers. previous study of prevalent cancers [1], in which 12% were imaged In the juxtathoracic third of the breast in the cranio- within the posterior 1 cm of imaged breast tissue. The de caudal view (48%) than in the mediolateral view (35%) (table tected incident imaged cancers were smaller, and fewer were d 5). In the prevalent population, 59% of cancers were seen located in the posterior centimeter of the breast as screening posteriorly in the craniocaudal view and 38% of the cancers continued. None of these most posteriorly located cancers e posteriorly were seen in the mediolateral view. Only 2% of was detected by physical examination alone. None of the 511961 0302 406 SCHMITT AND THREATT AJR 143. August 1984 TABLE 6: Incident Cancer Size, University of Michigan BCOOP (1975-1980) Saa (cm) 0-0.5........................... 0.6-1.0 ................... 1.1-1.5................... 1.6-2.0................... 2.1-2.5 >2.6 ...................... Total...................... NO.(%) 35 (32.7) 17 (15.9) 30 (28.0) 15 (14.0) 5 (4.7) 5 (4.7) 107 (100) CumUttnm Total % 32.7 48.6 76.6 90.6 95.3 100.0 TABLE 7: Comparison of Cancer Six* and Method of Detection, University of Michigan BCODP (1975-1980) Physical only................. Mammography only................... Mammography and physical . . . Total......................................... Tumor Stz si 0 cm. no (%> >1.0 cm. no (%) 8 (15.4) 36 (69.2) 8 (15.4) 7(12.7) 16(29.1) 32 (58.2) 52 (49.0) 55(51.0) TABLE 8: Comparison of Tumor Size and Invasiveness, University of Michigan BCOOP (1975-1980) imaged cancers required a third view of the breast for detec tion of the cancer. The size of the cancer was determined by the Project Tumor Sbs (cm) 0.5, no.f%) O.S.no. (%) pathologist from either the surgical specimen or the histologic section. Of the cancers, 48.6% were 1 cm or smaller and Noninvasive...................... Invasive............................. 27 (77) 8 (23) 7 (10) 65 (90) 76.6% were smaller than 1.5 cm (table 6). This is an increase % of total tumors......... 33 67 in the number of small cancers as compared with those in our prevalent population, in which 63% were 1.5 cm or smaller. The number of detected small cancers increased with rescreening. Of the incident cancer screenees, 85% had negative axillary nodes. Andersson et al. [5] found, in a screening project of 12,765 women aged 50-69 years, that 55% of the cancers were 1 cm or smaller and that 92.5% of these patients had negative axillary lymph nodes. This differs with the size of cancers usually found in a clinically symptomatic population, in which 19% were under 1.5 cm [6]. Physical examination alone detected 15% of the cancers under 1.0 cm and 69% were detected by mammography only (table 7). Of the cancers 0.5 cm or less, 77% were noninvasive, whereas only 10% of those over 0.5 cm were noninvasive (table 8). Mammography alone detected 76.5% of the noninvasive cancers; physical the breast, and proper technical factors vwi allow adequate imaging of>cancers in any location within the breast on filmscreen mammograms. This is well demonstrated in a popu lation followed with annual screening, in which very small, early cancers were detected. Cancers detected by annual rescreening examinations are less likely to be invasive than in the prevalent cancer group. Mammography will detect small cancers consistently throughout the entire breast, including those in the juxtathoracic third of the breast. Combining physical examination and mammography enhances each method, so that the detected cancers are smaller, with a small percentage of positive lymph nodes, thereby increasing the probability of cure. examination alone detected 5.9% of the noninvasive cancers. Mammography consistently detected noninvasive as well as REFERENCES invasive cancers, regardless of tumor size, histologic type, or location within the breast. Prom our data, it is apparent that standard craniocaudai 1. Schmitt EL, Threatt B. Tumor location and detectability in mammographic screening. AJR 1982;139:761-765 2. Wolfe JN. Analysis of 462 breast carcinomas. AJR and mediolateral contact film mammograms property obtained 1964;121:846-853 with vigorous compression will image mammographically de 3. Shaprio S. Evidence on screening for breast cancer from a tectable cancers in a given population. Only ten "interval" randomized trial. Cancer 1977;39:2772-2782 carcinomas (8.5%) occurred in years 2-5 of the UM-BCDDP. 4. Wolfe J. Breast patterns as an index of risk for developing breast Of these 10 patients, seven did not have mammograms. Four cancer. AIR 1978;126:1130-1139 of these seven were under age 50 and, therefore, ineligible 5. Andersson I, Anfred L, Hilda# J. Linde! F, Junggvist U, Pettersson for repeat mammograms (change of age criteria by NCI [7]) H. Breast cancer screening with mammography. Radiology 1979;132:273-276 T and three refused repeat mammograms. The remaining "in 6. Fisher B. Slack NH. Brass U. cooperating investigators. Cancer terval" cancers (three) occurred, not because they were not of the breast size of neoplasm and prognoses. Cancer imaged, but because they were obscured by the dense sur 1989;24:1071-1080 rounding breast tissue (Wolfe P2 and Dy parenchymal pat 7. Beahrs OH. Report of the working group to review the BCOOP. terns) [4]. Presented at the National Cancer Institute consensus develop Vigorous compression, careful attention to positioning of ment meeting, Washington. DC. September 1977 511961 0303 1986 March 03 Dr. Belk: Attached is an article from Environmental Research. A copy was sent to Terry Hannan Re: E. Martinez. Margaret Attachment 511961 0290