Skip directly to: content | left navigation | search

3.0 Comments on Topic 2: Health Effects of Asbestos and SVF Less Than 5 Micrometers in Length

This section summarizes the panelists' discussions on the role of fiber length in health effects from asbestos and SVF fibers. The meeting agenda (see Appendix D) lists the specific topics that the panelists addressed and identifies the discussion leaders for these topics. This section organizes the panelists' comments as follows: cancer effects (Section 3.1), noncancer effects (Section 3.2), mechanisms of toxicity (Section 3.3), general comments and interpretations (Section 3.4), and recommended research (Section 3.5). Section 3.6 summarizes observer comments made after the panelists completed their discussions. Some panelists submitted post-meeting comments to summarize their findings. These are included in Appendix E for the following topics: review of epidemiologic data (see Dr. Lockey's comments), review of laboratory animal studies (see Dr. McConnell's comments), and review of mechanistic studies (see Dr. Mossman's and Dr. Wallace's comments).

When evaluating health effects, panelists were asked to review findings from key studies that examined the role of fiber length on toxicity, whether in vivo or in vitro. Accordingly, this section should not be viewed as a literature review of all toxicity studies for asbestos and SVF; rather, it documents results from key studies that examined impacts of fiber length.

Although the panelists focused their initial discussions on fiber length, several panelists stressed that length is not the only factor affecting fiber toxicity. These panelists noted that toxicity is rather a complex function of the fiber dose, dimensions, and durability, as has been widely documented in the scientific literature.

3.1 Cancer Effects

This section summarizes the panelists' comments on the role of fiber length on cancer effects. The section is organized into three different types of studies: human cancer mortality studies (Section 3.1.1), studies of lung-retained fibers in humans (Section 3.1.2), and laboratory animal studies (Section 3.1.3). Within each section, comments are organized by type of fiber (asbestos or SVF) and type of cancer (lung cancer and mesothelioma).

3.1.1 Data from Cancer Mortality Studies

The panelists' comments on cancer mortality studies from occupational cohorts follow:

• Asbestos. One panelist indicated that no studies have evaluated cancer outcomes associated with fibers shorter than 5 µm, because no occupational cohort is exposed exclusively to such fibers. For insights into carcinogenicity of the short fibers, he reviewed findings reported for two occupational cohorts that were exposed predominantly (though not exclusively) to short asbestiform minerals:
  
  

  • The panelist first reviewed a study of workers at reserve mine deposits in Minnesota (Higgins et al. 1983). The workers at this site were exposed to cummingtonite-grunerite, a mineral related to amosite, and the vast majority of fibers were reportedly less than 10 µm in length. The study found no increase in overall mortality or mortality from respiratory cancers, but the panelist indicated that the average latency for the cohort was 14.7 years after initial exposure, with a maximum of 24.6 years, or a relatively short latency for development of cancer.

  • Second, this panelist reviewed studies of gold mine workers in South Dakota who also were exposed to cummingtonite-grunerite asbestiform material. He indicated that an initial study of this cohort (Gillam et al. 1976) found increased mortality from malignant respiratory disease among workers with at least 5 years of exposure. A follow-up study of this same cohort (McDonald et al. 1978), which considered workers who had worked for 21 years or longer, found no such increase, but did report increased risks of silicosis and tuberculosis. Average exposure concentrations for this site were 4.82 (±0.68) fibers per cubic centimeter, with 94% of airborne fibers being less than 5 µm in length.

  This panelist indicated that these two studies were the closest approximation he could find to occupational cohorts exposed only to fibers shorter than 5 µm, and neither showed evidence of increased cancer mortality. He advocated follow-up studies of these cohorts in view of the passage of more than 20 years since the original publications to investigate carcinogenicity of short fibers more fully.

  During this discussion, one panelist reviewed cancer mortality data published in two studies for an occupational cohort of Libby miners (McDonald et al. 1986, 2002). These studies considered 406 men who worked in the mine for at least 1 month prior to 1963; the more recent study, therefore, considers an average latency of period of more than 30 years from first exposure. This panelist noted that both studies reported elevated mortality rates for lung cancer, mesothelioma, and non-malignant respiratory disease (including asbestosis⁷ ). Of particular note, the panelist indicated that the more recent follow-up study (McDonald et al. 2002) suggests a PMR for mesothelioma of 6.7%—otherwise stated, 1 out of every 15 deaths identified in the follow-up study was from mesothelioma. He found this PMR significant because it is higher than those observed among most other cohorts studied, including crocidolite miners in South Africa and Australia.

• SVF. One panelist reviewed cancer mortality studies of occupational cohorts exposed to SVFs, including glass fibers, mineral wool, and RCFs. The panelist first noted that no studies have been conducted on occupational cohorts exposed exclusively to fibers less than 5 µm long, again because no cohorts appear to be exposed only to short fibers. However, he noted that cancer mortality at fiber glass and mineral wool production facilities has been extensively studied, both in the United States and Europe. He summarized these studies for different materials:
  
  

  • For fiber glass, the panelist noted that the studies did not find increases in respiratory cancer to be related to fiber glass exposures.

  • For rock wool and slag wool, the panelist indicated that studies of production facilities in the United States found no evidence of increased risk for respiratory cancer. At production facilities in Europe, on the other hand, initial studies have demonstrated increases in respiratory cancer, but no clear information to indicate that the increased cancer risk was specifically related to fiber exposure. A subsequent case-control study indicated no relationship between cumulative rock or slag wool exposure and lung cancer (Kjaerheim et al. 2002). The International Agency for Research on Cancer (IARC) authors did not conclude that the increased cancer was related to the exposures to rock or slag wool.

  • For RCF, which are more durable fibers than the other SVFs, the panelist indicated that no cancer mortality data have been published for occupational cohorts exposed to RCF. This panelist noted that a recent study of a relatively small cohort of plant production workers has not demonstrated increased respiratory cancer risk for RCF, nor any identified mesothelioma, but he acknowledged that the study had limited statistical power for detecting an increased risk. Results from this study have been accepted for publication in a peer-reviewed publication (Lemasters et al. 2002).

  The panelist who summarized these results also specifically noted that there is no indication of a relationship between exposure to SVFs and mesothelioma. Though a small number of mesotheliomas have been reported for workers at SVF manufacturing plants, these cases have since been explained by other factors (e.g., probable prior exposure to asbestos, incorrect diagnoses).

3.1.2 Data from Human Studies of Lung-Retained Fibers (Cancer)

Additional insights on the influence of fiber length on cancer outcomes was presented for studies that analyzed the amounts and sizes of fibers retained in the human lung. In these studies, lung-retained fiber is used to characterize exposure. The panelists identified limitations associated with such studies, most notably that the measurements of lung-retained fibers (typically at autopsy) are static and do not characterize when exposure occurred or temporal variations in exposure. Moreover, because lung-retained fibers can break or partially dissolve after exposure, it is possible that the length distribution of fibers observed after death is different from the length distribution of fibers in the original exposures. The panelists provided the following comments on available studies of lung-retained fibers:

• General comments. One panelist provided general comments on fiber accumulation and human disease. First, the panelist indicated that people are exposed to fibers of varying length, with shorter fibers generally accounting for the majority of exposure (by fiber count); a similar pattern—a majority of shorter fibers—is consistently observed in the lung-retention studies. Second, because asbestos fibers with widely varying lengths are detected in lung tissue samples from all populations, this panelist concluded that the human lung, under continuing exposure conditions, is not capable of completely clearing fibers of any length to background levels—a finding that is not replicated in inhalation studies conducted in rats⁸. He demonstrated lung accumulation by displaying data from multiple studies (e.g., Sebastien et al. 1980; Case et al. 2000), which showed that all types of asbestos fibers (including long chrysotile fibers) accumulate in the lung with cumulative exposure.
  
  
• Mesothelioma. The panelists then commented on three case-control studies that examined the distribution of fiber lengths in people who died from mesothelioma (and most with matched controls). All three studies showed that risk of mesothelioma was considerably higher for individuals with larger amounts of long fibers retained in their lungs:
  
  
  • The first study (McDonald et al. 1989) examined lung tissues from 78 Canadian men and women who died of mesothelioma, as well as 78 lung tissues from age-, sex-, and hospital-matched controls. The lung samples were from pathologists' stock, without information on what parts of the lung the samples were collected from. Relative risk for developing mesothelioma was reported for different fiber types and lengths (<8 µm and >8 µm). The study found that the risk of mesothelioma was significantly related to concentrations of amphibole fibers longer than 8 µm and that fibers shorter than 8 µm accounted for none of the cancer risk.
    
    
  • The second study (Rogers et al. 1991) examined lung tissues from Australians who died of mesothelioma. Based on "the best fitting additive relative risk model," the study reported that mesothelioma risk was greatest for crocidolite asbestos fibers longer than 10 µm, followed by amosite asbestos fibers longer than 10 µm, and then by chrysotile fibers less than 10 µm. The authors suspected that the relative risk for chrysotile fibers less than 10 µm resulted from longer fibers breaking into shorter fibers.
    
    
  • The third study (Rödelsperger et al. 1999) evaluated lung tissue samples from 66 German individuals who died from mesothelioma and 66 matched controls. The study reported that "...a clear dose-response relationship up to an odds ratio of 99% has been demonstrated for the lung tissue concentration of total amphibole fibers longer than 5 µm." The study provided few details on the cancer risk associated with short fibers.
    
    
• Lung cancer. A panelist indicated that no lung-retention studies in humans have attempted to examine relationships between the length distribution of retained asbestos fibers and lung cancer. He suspected that studies have not been conducted due to the high attributable risk from smoking. This panelist noted that many studies have reported the total concentration of asbestos fibers for lung cancer in lung samples, but none of these evaluated the role of fiber length.

3.1.3 Data from Laboratory Animal Studies (Cancer)

The panelists identified several laboratory animal studies that illustrate the influence of fiber length on carcinogenicity, and made general comments about the relevance of these studies to humans. Panelists specifically referred to the following three studies when discussing how fiber length has been shown to relate to lung cancer and mesothelioma in laboratory animals:

• In the first study (Davis et al. 1986), no malignant cancers were observed in 42 rats exposed via inhalation to a short-fiber amosite mixture, while eight malignant cancers were reported in the 40 rats exposed to the long-fiber amosite mixture (30% of fibers longer than 5 µm and 5% of fibers longer than 10 µm).
  
  
• In the second study (Davis and Jones 1988), seven malignant cancers were observed among rats exposed via intraperitoneal injection to a "short" chrysotile fiber mixture, while 22 malignant cancers were observed among those exposed to the "long" fiber mixture. Cancers in the former group, however, have since been attributed to contamination of the "short" fiber samples with longer chrysotile fibers (Lippmann 1994).
  
  
• In the third study (Wagner et al. 1985), rats exposed to mixtures of erionite fibers that were mostly shorter than 5 µm did not develop mesothelioma, while every rat exposed to the longer erionite fiber mixtures developed the disease. One panelist found certain aspects of this study surprising, such as the fact that all of the rats exposed to long fibers died within 15 months, even though mesothelioma typically is not lethal in rats, and that the histopathological slides showed very intense pleural reactions. The panelists revisited this study (see Section 3.4) when discussing how chemical composition and surface properties might affect toxicity.
  

General strengths and weaknesses of laboratory animal studies The panelists provided several general comments on the utility of laboratory animal studies for understanding toxicity of asbestos and SVFs. Benefits of animal studies include the ability to (1) conduct highly controlled experiments using well-defined exposure levels and (2) evaluate health outcomes and lung-retention levels at many different time frames following exposure. Extensive lung tissue sampling and other highly invasive tests in humans, on the other hand, are only feasible at autopsy. However, panelists identified key factors that must be considered when interpreting laboratory animal studies. These factors include differences in life span, macrophage size, and airway branching patterns; relevancy of high dose and administration methods (e.g., peritoneal injection); and failure to address certain human exposure conditions (e.g., smoking). Overall, the panelists generally agreed that laboratory animal studies can provide useful insights into toxicity to humans, provided the studies are interpreted in the proper context regarding their relevancy to humans.

One panelist synthesized the findings from these and other relevant laboratory animal studies. This panelist first noted that the rat is an adequate model for cancers in humans⁹, because the rat has been shown to develop both mesothelioma and lung cancer, though he acknowledged that these cancers are not as aggressive in the rat as in humans. He added that the laboratory animal studies have allowed researchers to observe the progression of disease for both lung cancer and mesothelioma. Regarding the administration method, this panelist indicated that the inhalation studies were more relevant to human exposures. He noted that fiber administration by intrapleural implantation and intraperitoneal injection does not represent human exposures for several reasons (e.g., extremely large doses are administered in very short time frames, alveolar macrophage and mucociliary transport clearance mechanisms are bypassed, and the fibers inserted into the pleura might not be capable of reaching these tissues following inhalation exposure).

Overall, this panelist believed that laboratory animal data using all administration routes have shown that short fibers of any type are less potent than long fibers, both for mesothelioma and cancer, but the relative potency has not been quantified.

3.2 Noncancer Effects

This section summarizes the panelists' comments on the role of fiber length on noncancer effects and is also organized according to the different types of studies: occupational studies (Section 3.2.1), studies of lung-retained fibers in humans (Section 3.2.2), and laboratory animal studies (Section 3.2.3). Each section is further organized by noncancer endpoint. Although many different endpoints were discussed (e.g., irritation, nephrosis), the majority of discussions focused on pulmonary interstitial fibrosis and pleural abnormalities (e.g., pleural plaques, pleural thickening, and calcification).

3.2.1 Data from Occupational Studies

Overall, the discussion leader for this topic area indicated, there is limited evidence of noncancer toxicity being associated with fibers less than 5 µm in length, with two exceptions. First, he indicated that very high doses to short fibers, especially those that are durable in intracellular fluids, may have the propensity to cause interstitial fibrosis. Second, he noted that exposure to short, thin durable fibers may play a role in development of pleural plaques or diffuse pleural fibrosis if the dose is high enough. The following paragraphs review the discussion that led to these summary statements

• Asbestos. One panelist noted that no epidemiologic studies have examined populations exposed only to short asbestos fibers, because actual exposures are inevitably to a broad distribution of fiber lengths. To address this issue, the panelists commented on data reported among Libby residents, particularly the prevalence of intense bilateral pleural fibrosis in community members—some of whom reportedly did not work in the local vermiculite mine or processing plant, and did not live with mine or mill workers. One panelist was particularly concerned about the role of short fibers, noting that a very large portion of fibers in the homes are too short or too thin to be counted by conventional PCM sampling methods. He added, however, that some researchers have speculated that short (<10 µm), thin (<0.4 µm), durable fibers, particularly tremolite asbestos, may preferentially deposit on the pleural surface and therefore be associated with pleural plaques. This panelist emphasized that the relevance of short, thin fibers and the risk for pleural abnormalities has only been speculated, and needs to be further investigated. The panelists also wondered if the intense pleural effects observed in the Libby cohort might be associated with the unique mineralogy of the Libby asbestiform fibers. Pleural plaques have been associated with environmental exposures in areas where tremolite fibers naturally occur (e.g., in certain regions of Greece, Cyprus, Turkey, Canada, the Czech Republic, Romania).
  
  Some discussion focused on the extent to which pleural effects are associated with occupational versus environmental exposures. Two panelists cautioned against attempting to classify exposures in this manner, because some individual exposures were difficult to assess. For instance, some residents might not have worked at the mine or the mill or lived with mine or mill workers, but could have been highly exposed through routine contacts with these individuals in other settings. These panelists recommended that the discussion focus strictly on dose, regardless of the contributions from occupational and environmental exposures.
  
  
• SVF. One panelist indicated that the available epidemiologic studies provide no indication of increased mortality from nonmalignant respiratory disease among occupational cohorts exposed to SVF. He then summarized morbidity data for several cohorts. The summary focused on SVF production workers. Although SVF "end users" (e.g., insulators, pipe fitters, heating/ventilation workers) have also been evaluated, these studies are commonly confounded by potential asbestos exposures. Overall, this panelist concluded that the available occupational studies indicate limited overall toxicity associated with SVF exposure, with the exception of RCF exposure being associated with pleural plaques. This conclusion was based on the following observations:
  
  
  • For exposures to SVFs (all types), one panelist noted that multiple studies have found that SVF exposure among current or former smokers is associated with small additional decrements in forced vital capacity (FVC) and forced expiratory volume in 1 second (FEV1). He added that similar decrements in spirometric parameters are observed among other non-specific dust exposed industrial working populations, suggesting that the effect is not specific to SVFs.
    
    This panelist also reviewed studies, albeit limited ones, of skin irritation. These have reported irritation being related to mechanical effects of large diameter fibers (~5 µm in diameter), with increased irritation observed in hot, humid climates. Eye, upper respiratory, and lower respiratory irritation has also been reported in case studies among people accidentally exposed to high fiber concentrations, and these irritation effects are generally transient.
    
    
  • For fiber glass and mineral wool, this panelist noted that the available studies (e.g., Hughes et al. 1993), though limited in number, provide no indication of chest radiographic, interstitial, or pleural changes among production workers. The panelist added that studies have suggested an increased mortality risk from nonmalignant renal disease (e.g., nephritis, nephrosis) in occupational cohorts exposed to mineral wool, but not among those exposed to fiber glass; he questioned the biological plausibility of these outcomes.
    
    
  • For the more durable RCFs, the panelist indicated that occupational exposures have been associated with pleural changes, primarily pleural plaques. The pleural plaques were observed among approximately 3% of the production workers, but were found to be correlated with duration of RCF exposure, time since initial exposure, and cumulative RCF exposure. The panelist added, however, that the available studies have not found RCF exposure to be associated with a statistically increased risk for pulmonary interstitial fibrosis.
    
    
  • During this discussion, Dr. Ralph Zumwalde (NIOSH) informed the panel that, in the late 1970s, NIOSH studied the health implications among more than 2,000 miners who were exposed to an attapulgite clay that has fiber-like characteristics (Waxweiler et al. 1988). The clay "fibers" were less than 5 µm long, with diameters of approximately 0.1 µm. Dr. Zumwalde noted that this study, which he recalled found excess lung cancer among whites, might be useful in ATSDR's overall evaluation of short fibers. The increase in lung cancer deaths, however, was not associated with latency, duration of employment, or attapulgite exposure, and there was no increase in mortality from nonmalignant respiratory disease.

Overall, the relevance of short asbestos and SVFs to noncancer disease in humans was not entirely known. For the SVFs, only the durable RCF was found to be associated with pleural plaques; exposures to RCFs were not associated with pulmonary fibrosis, and exposures to fiber glass and mineral wools had no indication of chest radiographic, interstitial, or pleural changes. For asbestos fibers, no studies have examined the effects of exposures exclusively to short fibers. Given data collected in Libby, Montana, however, some panelists questioned whether short fibers might play a role in the observed cases of pleural plaques and diffuse pleural fibrosis; but others cautioned against inferring that the risk results from exposure to short fibers, given that the Libby samples contained significant numbers of long fibers as well.

3.2.2 Data from Human Studies of Lung-Retained Fibers (Noncancer)

Two panelists reviewed publications (case-control studies, a study recently submitted for publication, and a case report) that examine the influence of fiber length retained in the lung on the grade of pulmonary interstitial fibrosis, which is reported on a scale from 0 to 12. A summary of these studies, organized by fiber type, follows:

• Findings for tremolite asbestos. One panelist indicated that a study of tissues from chrysotile asbestos miners and millers reported an inverse relationship between fibrosis grade and length of tremolite fibers retained in the lung (Churg et al. 1989). In other words, the most severe fibrosis was observed among those with smaller (on average) tremolite fibers in their lungs. Another study (Nayebzadeh et al. 2001) and a study recently submitted for publication (Case et al. 2002b) examined fibrosis grades for different length intervals of tremolite fibers: 0-5 µm, 5-10 µm, and 10-20 µm. Both studies found the highest average fibrosis grade occurred among those with the lowest tremolite fiber length interval, or for those with average tremolite fiber length less than 5 µm.
  
  
• Findings for amosite asbestos. One study (Churg et al. 1990) examined lung tissue samples from a small group (<20) of shipyard workers and insulators selected from litigation cases. This study also found an inverse relationship between fibrosis grade and length of retained asbestos fibers (amosite fibers, in this case).
  
  
• Findings for total asbestos fibers. One panelist summarized a study (Timbrell et al. 1988) that evaluated lung tissue samples at autopsy from workers exposed in different asbestos mines. Data were collected both for retained asbestos fibers and fibrosis score. The fibrosis scores were then correlated with lung-retained asbestos characterized by three different metrics: number of fibers, mass of fibers, and surface area of fibers. The correlation was best when the surface area of retained fibers was used as a dose metric. This panelist added that the surface area dose metric has correlated well with pulmonary inflammatory responses in other animal inhalation toxicity studies that examined inflammation, including fibrosis, following exposure to particulate contaminants that are poorly soluble with low chemical reactivity. The panelists referred to this study, which did not examine the role of fiber length, several times when discussing appropriate dose metrics.
  
  
• Findings for aluminum oxide fibers. One panelist reviewed data from a case report (Churg et al. 1993) on an individual with diffuse interstitial fibrosis who was occupationally exposed to aluminum oxide fibers. The lung-retained fibers in this case were predominantly 3-4 µm long and 0.01 µm in diameter. The panelist indicated that these findings raise questions about the significance of short, thin, durable fibers in the lung, though he acknowledged that conclusions should not be drawn from a single case report.

Several panelists commented on the trends among the aforementioned studies. Two panelists, for instance, noted that the trend of shorter fibers possibly being more toxic, at least in terms of interstitial fibrous, is counterintuitive. Two other panelists, on the other hand, noted that these findings suggest that, for interstitial fibrosis, the surface area of retained fibers may be more important than the fiber length, because larger amounts of short fibers would have considerably greater surface area than smaller amounts of long fibers. Finally, some panelists wondered if the apparent inverse relationship between fiber length and fibrosis score might be explained by long fibers breaking down into shorter fibers between exposure and the time that lung samples were collected.

3.2.3 Data from Laboratory Animal Studies (Noncancer)

This section reviews the panelists' discussions on noncancer outcomes from asbestos and SVF exposure identified in laboratory animal studies. Before addressing this topic, one panelist summarized how the mammalian lung responds to exposures to inert materials, whether fibrous or particulate: once an inert material deposits in the lung beyond the conductive airways, it will either dissolve or be engulfed and cleared by alveolar macrophages; if the dose exceeds the lungs' capacity to clear the material, natural defense mechanisms may act, leading to fibrosis. Section 3.3 presents more details on the mechanisms involved in these steps. Specific comments on noncancer effects in laboratory animals, organized by endpoint, follow:

• Inflammation, pulmonary interstitial fibrosis, and pleural reactions. The panelists presented several observations when summarizing findings from laboratory animal studies on noncancer effects in the lung and pleura. First, two panelists noted that many laboratory animal studies have found pulmonary interstitial fibrosis following exposures to both fibrous material and non-fibrous particles. The sequence of events leading to the fibrosis was described (see Dr. McConnell's premeeting comments in Appendix B). When doses reach high enough levels, pleural reactions (e.g., localized acellular fibrotic changes) were observed, but one panelist questioned if the dose levels needed to elicit the pleural responses are relevant to environmental exposures in humans. Another panelist noted that the animal studies suggest that the pleural effects do not occur unless fibers are present in the pleura. When discussing interstitial fibrosis outcomes, one panelist said the long fibers appear to be more fibrogenic than the short fibers, though he stressed that short fibers alone are capable of generating fibrogenic responses if the dose is sufficiently high. The panelists' premeeting comments include specific references to studies that reported relative toxicity of short and long fibers for noncancer outcomes (see Dr. Mossman's premeeting comments in Appendix B).
  
  Reviewing specific studies, one panelist indicated that the intensity of noncancer responses in laboratory animals varies from one fiber type to the next. He noted, for example, that hamsters exposed to amosite asbestos had an increased incidence of pleural fibrosis, while hamsters exposed to comparable amounts of chrysotile asbestos did not; pulmonary fibrosis was evident, however, in both groups of hamsters. The panelist suspected that the different outcomes resulted from either the amosite fibers being more durable (less soluble) in the lung or the amosite fibers being more likely to translocate to the pleura.
  
  One issue that generated significant discussion was the extent to which interstitial fibrosis progresses in laboratory animals and the relevance of disease progression to humans. One panelist noted that, in every animal study he has conducted and reviewed to date, interstitial fibrosis is progressive only when asbestos exposure is ongoing. After asbestos exposure ceases, he noted, no overt signs of progressive fibrosis are apparent, although this has not been quantified in a definitive way. The inflammatory responses, microgranulomas, and bronchiolization also tend to decrease. This panelist added that fibrosis does not appear to progress and macrophage response tends to decrease when the exposure ceases, even though long asbestos fibers remain in the animals' lungs. He interpreted this trend as suggesting that short asbestos fibers in the original dose might play a role in stimulating an initial inflammatory response in the rats. Another panelist suggested that the lack of fibrosis progression, even in the presence of long fibers, might suggest that the retained fibers have been rendered inert (in comparison to the freshly inhaled fibers), possibly by being coated with biological fluids. In other words, he wondered if the freshly inhaled fibers are more likely to elicit cellular responses than fibers that have been in the lung for an extended period of time.
  
  Though not questioning the comments on fibrosis progression in animals, two panelists emphasized that the trends discussed above are not observed in humans. Citing their experiences evaluating shipyard workers and chrysotile miners, these panelists noted that fibrosis and pleural changes have progressed in humans, even after asbestos exposures ceased. Reasons why fibrosis might progress differently in rats and humans were not discussed.
  
  Commenting further on disease progression, one panelist indicated that certain noncancer effects and lung cancer appear to have consistent patterns in all animal studies he has reviewed, including studies of asbestos exposure and studies of exposure to non-fibrous particulate. Specifically, this panelist said he had not found any study in which a rodent had lung cancer, but did not have interstitial fibrosis; and he had never seen rodents with interstitial fibrosis in the absence of inflammation. These observations led the panelist to infer that lung cancer would not be expected to develop at doses that do not induce fibrosis or inflammation. He stressed that this inference is based solely on observations from previous animal studies and does not in any way suggest that fibrosis is a precursor to lung cancer-an issue that came up during the observer comments (see Section 3.6). He also added that this relative sensitivity of noncancer endpoints may not necessarily be observed in humans.
  
  
• Irritation. One panelist noted that laboratory animal studies have not studied the extent to which asbestos and SVF irritate the skin and eye. He added that histopathological studies of the nasal cavity, pharynx, larynx, trachea, and conductive airways have not identified evidence of irritation, though he acknowledged that the histopathological techniques might not have detected certain responses (e.g., increased mucus production). He cautioned that these results do not necessarily suggest that humans will not experience fiber-induced irritation in the nasal cavity, larynx, and upper respiratory tract, because the rat studies did not consider populations with impaired mucociliary clearance, as might be observed in smokers. Finally, this panelist indicated that ingestion studies in rats and hamsters have shown no evidence of irritation in the gastrointestinal tract.

3.3 Mechanisms of Toxicity

This section reviews the panelists' comments on mechanisms of toxicity, primarily as presented by the two designated discussion leaders, Dr. Mossman and Dr. Wallace. After identifying several general advantages and disadvantages of in vitro studies, the discussion leaders reviewed current theories on mechanisms of toxicity for a wide range of fibers and analogous non-fibrous particles. This section reviews key points from those presentations. Emphasis is placed on what has been established or hypothesized regarding the relative toxicities of short and long fibers. For more detailed information on mechanisms of toxicity, refer to Dr. Mossman's and Dr. Wallace's post-meeting comments in Appendix E.

• General comments on the utility of in vitro studies. To initiate discussions, one panelist listed several strengths and limitations associated with in vitro toxicity studies. First, she indicated that in vitro studies, when compared to laboratory animal studies, offer a far more controlled setting for examining mechanisms of toxicity and dose-response behavior for specific cell types. She acknowledged, however, that interpreting trends among studies using widely varying doses and multiple cell types can be complicated. Moreover, the in vitro studies are all limited in duration, typically lasting a few days, due to the limited life spans of isolated cells in the in vitro environment. Consequently, the in vitro studies cannot characterize dissolution, macrophage clearance, and other processes that occur over longer time scales. Finally, this panelist noted that doses to in vitro samples cannot readily be extrapolated to human inhalation exposures.
  
  
• Role of reactive oxygen species (ROS). One panelist reviewed a widely accepted theory of how generation of ROS might explain asbestos-related toxicity. She indicated that alveolar macrophages, as they attempt to digest foreign fibers and particles, produce an "oxidative burst" and release ROS. (Other cell types that contact asbestos fibers also release ROS.) These ROS can initiate sequences of events that have been shown in vitro to lead to outcomes such as genotoxicity, cytotoxicity, and cell proliferation.
  
  This panelist highlighted two key observations regarding ROS. First, in vitro studies have shown that alveolar macrophages generate more ROS when attempting to digest longer fibers, while shorter fibers can be engulfed completely by macrophages (and other cell types) with no visible damage to the cells. Second, she noted that ROS can form highly reactive hydroxyl radicals via a reaction that is facilitated by the presence of iron. Therefore, long, iron-containing fibers, like several amphibole asbestos fibers, are capable of generating an intense "oxidative burst," which might explain their greater potency, when compared to fibers that do not contain iron. Finally, this panelist noted that researchers can prevent pulmonary fibrosis in animals by administering free radical scavengers or other substances that interfere with ROS formation and reactions—a finding that argues strongly for ROS having a causative role in inducing asbestos-related fibrosis.
  
  Overall, this panelist noted that many aspects of the ROS theory help explain how fiber length and, to a lesser extent, mineral content relate to toxicity and why shorter fibers are substantially less toxic than longer ones. She presented results from several in vitro studies (e.g., Ohyama et al. 2001) that confirm that longer fibers generate a greater "oxidative burst" when they are not ingested by alveolar macrophages.
  
  
• Effects on cell signaling events. The panelist then described current research that has characterized effects of asbestos- and fiber-related cell signaling events. She explained that these events originate when fibers interact with cell surfaces, after which the cells activate transcription factors that mediate various outcomes which can be measured in vitro, such as cell proliferation, cell transformation, and cell death. She noted that cell proliferation is an important step in development in both malignant and nonmalignant disease.

  The panelist then described studies examining how selected signaling pathways are affected by asbestos and glass fibers of different lengths. She summarized studies that demonstrated activation of transcription factors and cell proliferation. First, the panelist reviewed a study (Ye et al. 1999) in which mouse macrophage cell lines were challenged with two formulations of fiber glass mixtures, one with average length of 6.5 µm, the other 16.7 µm. These challenges caused production of tumor necrosis factor-alpha (TNF-a), a cytokine involved in inflammation and fibrosis, which in turn caused activation of nuclear factor-kB. Gene promoter activation induced by the short fibers was found to be between one-third and one-half what was observed for the long fibers.

  Second, the panelist reviewed in vitro studies that examined specific aspects of cell proliferation. Although cell proliferation relates to cancer outcomes, she emphasized that development of mesothelioma and lung cancer is a multi-stage process with a long latency period, which cannot be captured in the short time frame of an in vitro study. The panelist identified many studies (see Appendix E) demonstrating that longer fibers are more apt to cause cell proliferation than are short fibers, whether from tracheal explant studies (e.g., Sesko and Mossman 1989) or intratracheal models in rats (Adamson and Bowden 1990).

• Studies of asbestos genotoxicity. The panelist indicated that researchers have been studying the genotoxicity of asbestos, both in vivo and in vitro, for more than 20 years. These studies examined a wide range of endpoints (e.g., cell transformation, chromosomal aberrations, gene mutation) in various matrices. The panelist focused, however, on a series of studies conducted to examine the role of fiber length on cell transformation and cytogenetic effects (Hesterberg and Barrett 1984, 1985; Hesterberg et al. 1986). These studies demonstrated that long, thin fibers are most potent for both types of effects, and the shortest fibers examined (less than 1.7 µm long) had no indication of tumorigenic potential. These findings, she noted, indicate that longer fibers are again more toxic, with some suggestion that fibers below a certain length threshold may not be carcinogenic at all.
  
  
• Observations regarding mechanisms of toxicity from non-fibrous particulates. One panelist addressed mechanisms of action for non-fibrous particulates having compositions similar to those in asbestos and SVFs. First, he indicated that non-fibrous crystalline silica is strongly pathogenic for fibrotic lung disease, while two polymorphs of crystalline silica-quartz and cristobalite-have recently been classified as carcinogenic (IARC 1997). In contrast, amorphous silica (more akin to SVFs) has not been shown to cause lung cancer or mesothelioma in rodents (IARC 1987). Exposure to the crystalline silica polymorphs can directly damage cells, resulting in intracellular generation of reactive oxygen species and a cascade of events (e.g., synthesis and release of cytokines, cell proliferation, secretion of collagen into the extracellular space) similar to the those evoked by asbestos fiber. In vitro studies have shown that silanols (hydroxyl groups on the crystalline surface) are associated with the initial damage to cells: loss of surface silanols caused the crystalline silica to exhibit less damaging activity, and subsequent formation of silanols restored the silica's toxicity (Pandurangi et al. 1990). These and other studies (see Appendix E) suggest that surface chemistry plays a role in silica's toxicity.
  
  This panelist noted that an important but generally ignored component for physiologically representative in vitro bioassays is that particles and fibers depositing in the lung initially contact the aqueous "hypophase" lining on the terminal airway and airsac surfaces. The hypophase layer is rich with micellar dispersion of surfactant, composed largely of lipids and lipoproteins. Of particular note, the hypophase can be simulated in vitro with dipalmitoyl phophatidyl choline (DPPC) dispersed in physiological saline. Silica particles and other materials deposited in the lung have been shown to adsorb the surfactant, which extinguishes short-term cytotoxicity—another observation indicating that the toxicity of particles in the lung is affected by surface chemistry. This panelist noted that the alveolar hypophase contains more than enough surfactant to coat and neutralize the entire surfaces of respirable particles, even in most high dust exposures (Wallace et al. 1975).
  
  Although the volume of surfactants in the alveolar hypophase is sufficient to coat respired particles completely (even in high dust exposures) and thus is theoretically capable of extinguishing the particles' toxicity, this panelist indicated that the toxicity can be restored when other cellular mechanisms remove the protective surfactant cover. Specifically, macrophages can engulf surfactant-coated particles, where they are subject to phagolysosomal enzymatic digestion which can remove the surfactant film and thus restore toxicity. Some study has shown that the surfactant film is more readily removed from crystalline silica than it is from kaolin, which suggests a mechanism by which quartz may be more toxic than kaolin; however, experimental study has not demonstrated that particle de-toxicification and re-toxicification explains the relative toxicities of these materials. Thus, surfactant coating of foreign particles deposited in the alveolar space again appears to play an important role in toxicity. The influence of surface chemistry has also been observed in quartz particles having alumino-silicate surface contamination; for such particles the surface contamination can delay for months or perhaps years the expression of fibrogenic activity.
  
  Finally, this panelist noted that researchers might glean greater understanding of the main site of asbestos fibrogenic activity from theories reported for crystalline silica. He explained that a series of studies (e.g., Bowden et al. 1989) suggests that fibrosis results from a sequence of events following interactions between crystalline silica and interstitial cells, rather than interactions with alveolar macrophages. Specifically, it is hypothesized that interactions with the interstitial cells control the stimulation of exacerbated collagen synthesis by pulmonary fibroblasts; whereas, interactions with macrophages are hypothesized as being responsible only for an inflammatory response (not fibrosis) evoking neutrophil influx to the alveolus. This panelist suggested that further research on the mechanisms of fibrogenic toxicity for asbestos should consider interactions with interstitial cells, rather than focusing largely on responses initiated by interactions with alveolar macrophages.
  
  
• Comparisons between fibrous minerals and crystalline silica particles. This panelist noted that the available in vitro studies do not explain comprehensively how asbestos fibers and crystalline silica particles differ in inducing fibrosis. Although they identify several endpoints that asbestos and crystalline silica have in common, the studies cannot predict why asbestosis appears as a diffuse fibrosis, while silicosis appears in localized nodules.

  However, some research provides insights on differences between how long fibers, short fibers, and particles contribute to cytotoxicity. Specifically, an in vitro study (Liu 1994) examined whether surfactant coating inhibits the cytotoxicity of asbestos. (As the previous bulleted item indicates, similar studies found that surfactant coating virtually extinguished the short-term toxicity of crystalline silica particles.) In the study, Chinese hamster lung cells were tested for micronucleus induction after being challenged with surfactant-coated chrysotile asbestos. The study considered how induction differs between long fibers (average fiber length of 101 µm) and shorter fibers (average fiber length of 11.6 µm). It found a slight, but not significant, decrement in cytotoxic endpoints for the long fibers and a considerable, statistically significant decrement for the shorter fibers. The findings suggest that surfactant coating is less effective at impairing toxicity for longer fibers.

  Although the studies on crystalline silica underscored the role of surface chemistry in eliciting toxic responses, changes in the surface composition in chrysotile asbestos were found to have no significant effect on in vitro genotoxic activity (Keane et al. 1999). Specifically, fibers that had been mildly leached to remove near-surface magnesium atoms exhibited comparable genotoxicity to fibers that were not treated with the leaching solution.

  Based on his review of these and other studies, one panelist suggested that more than one mechanism of toxicity may operate for asbestos and SVF, and the roles of the individual mechanisms might depend on fiber length. He explained that "frustrated phagocytosis" and its ensuing events clearly appear more relevant to long fibers (i.e., long fibers are much more likely to be only partially engulfed by alveolar macrophages), while a toxicity mechanism mediated by surface properties of phagocytized material (e.g., restoration of fiber toxicity in the intracellular matrix) would be more relevant to short fibers. In other words, part of the short fiber toxicity might be related to mechanisms involving surface chemistry, which were described in the previous bulleted item. This panelist added, however, that additional mechanisms could contribute to toxicity. As one example, he indicated that asbestos fibers penetrating the cell or cell nucleus may exercise modes of direct genetic or epigenetic damage. Whatever the mechanisms of direct fiber damage or stimulation of the cell surface, he noted, several components of the consequent intracellular response have been well defined. Appendix E provides additional detail on the responses that have been characterized, and the influence of fiber length on these responses.

• Recent advances in fiber preparation methods. One panelist noted that researchers at NIOSH have been developing a fiber size classifier (separator) that permits in vitro or perhaps limited in vivo experiments with sets of fibers of fairly well-defined length (Baron et al. 1994). The dielectrophoretic classifier reportedly can separate fibers from an airstream and produce about 1 mg/day of a given size interval. The panelist indicated that this preparation method recently was used to generate the following categories of fiber size intervals:

Cut	Fiber Length		Fiber Diameter
	Average (µm)	Standard Deviation (µm)	Average (µm)	Standard Deviation (µm)
1	32.7	23.5	0.75	0.50
2	16.7	10.6	0.49	0.27
3	6.5	2.7	0.44	0.22
4	4.3	1.0	0.40	0.15
5	3.0	1.0	0.35	0.14

3.4 General Comments and Interpretations

While discussing the influence of fiber length on asbestos and SVF toxicity, the panelists made several general comments and interpreted observations from the laboratory animal, human, and in vitro studies. This section summarizes these general comments and interpretations, while Section 4.1 reviews the panelists' individual summary statements provided at the end of the meeting.

• Evaluating toxicity based on the "reasonable certainty of no harm." When discussing asbestos and SVF toxicity, the panelists discussed the terminology they should use to characterize the hazard of fibers less than 5 µm long. One panelist recommended that the panelists consider whether the short fibers have a "reasonable certainty of no harm," drawing from the language promulgated in the Food Quality Protection Act. Given that dose-response data for humans or animals uniquely exposed to fibers less than 5 µm is largely not available, most panelists agreed that the terminology proposed was appropriate for their conclusions. They also noted that separate conclusions should be drawn for different endpoints, using a weight-of-evidence approach that draws from all types of data, including dosimetric, toxicologic, epidemiologic, and in vitro testing.
  
  
• Do fibers shorter than a certain length have a "reasonable certainty of no harm"? The panelists debated whether summary statements could be made regarding whether fibers of certain length intervals have a "reasonable certainty of no harm." Two panelists suggested that environmental exposures to fibers shorter than 5 µm would likely be free of carcinogenic effects. Other panelists, however, felt uncomfortable making such judgments for noncancer effects (e.g., pleural abnormalities), especially considering the evidence summarized in Section 3.2. Refer to Section 4.1 for the panelists' individual summary statements regarding the influence of fiber length.
  
  
• Why arbitrarily establish critical fiber lengths? Because humans are always exposed to fibers with a wide distribution of fiber lengths, one panelist wondered if ATSDR or environmental agencies could develop a universal algorithm that quantifies health risks associated with different types of fiber mixtures. For instance, an algorithm might include different relative toxicity factors for different fiber length intervals (e.g., 0-5 µm, 5-10 µm, 10-20 µm, and so on). Such data could then be applied to the distribution of fiber lengths measured in the environment to assess site-specific risks. This panelist acknowledged that the relative toxicity data do not appear to be available to support this approach, but he noted such a universal algorithm would be far less arbitrary than completely ruling out fibers having dimensions below a certain level. He added that such an algorithm can eventually account for other factors (e.g., biopersistence) that are also known to affect toxicity. In short, this panelist indicated that it is theoretically possible to express health risk as a function of fiber dose, dimension, and durability, though he noted that one would need additional research into dose-response and extensive inputs from biostatisticians to develop such an algorithm.
  
  
• Other influences on toxicity. While recognizing that the focus of the meeting was on how fiber lengths affect toxicity, the panelists noted that many additional factors determine the toxicity of a fiber mixture. Examples of other factors include dose, fiber composition (mineral type), physical state (amorphous or crystalline), surface area, and surface properties. The panelists cited several examples of why length alone might not adequately predict toxicity. First, the panelists noted that the cancers observed in the study of rats exposed to erionite (Wagner et al. 1985) could not be explained by fiber length alone; they suggested that the unique findings of this study might be best explained by unique surface chemistry or the mineral's relatively large internal surface area (2.5 m²gram). Second, the panelists noted that fiber durability likely explains why asbestos fibers and SVFs of the same length are not equally toxic. Due to these and other observations, a panelist noted, ATSDR might overlook other important factors that influence toxicity if it focuses exclusively on fiber length.

3.5 Research Needs

The panelists identified several research needs when discussing the influence of fiber length on health effects. In general, the panelists encouraged thorough planning of any future study, emphasized the need for having well characterized exposures, and advocated involving researchers from multiple disciplines (e.g., epidemiologists, physicians, toxicologists, mineralogists). All research needs mentioned during this session of the meeting are documented here:

• Several panelists indicated that further study should be conducted among the residents of Libby, Montana, to understand the effect of fiber length on toxicity. One suggestion was, through the cooperation of the community and consent of residents, establishing a protocol to analyze lung and pleural tissue from community members who die, regardless of the cause of death. Another suggestion was to track the progression of the observed pleural disease.
  
  
• Given the health outcomes observed in Libby, one panelist encouraged focusing future research in laboratory animals on understanding fiber dose-response behavior for the visceral and parietal pleura. Such studies could use fibers from Libby to examine how doses to the pleura and progression of toxic responses vary with fiber length and between fibers and non-fibrous particles. Another panelist added that a well-constructed study can investigate multiple toxic endpoints.
  
  
• For added insights on toxicity of short fibers, possibly those from Libby or Lower Manhattan, the panelists suggested conducting an in vitro study using several cell types (e.g., rat pleural mesothelial cells, tracheal epithelial cells) to examine multiple endpoints that can be confirmed in animal models, such as cell proliferation and cytotoxicity.
  
  
• One panelist suggested that future study of environmental exposures could focus on residential development in areas with increased levels of naturally occurring tremolite asbestos (e.g., the Sierra foothills in California), but he added that a high level of cooperation from the local community would be essential to the success of any such study.
  
  
• To assess human health effects associated with exposures to short fibers, one panelist recommended follow-up study of two cohorts of miners, one in South Dakota and one in Minnesota, who were exposed predominantly to short cummingtonite-grunerite fibers. Further study would take into account a longer latency period and might reveal insights on the role of fiber length in toxicity.
  
  
• Several panelists encouraged NIOSH to continue to re-analyze personal exposure samples collected on membrane filters in the 1960s and 1970s from textile workers in Charleston, South Carolina. This suggestion followed an observer comment that informed the panel of NIOSH's planned work on this project.

3.6 Observer Comments and Ensuing Discussions

Observers were given two opportunities to provide comments on the second day of the meeting. The panelists were not required to respond to the observer comments. However, some comments led to further discussion among the panelists, as documented here. The observer comments are summarized in the order they were presented:

Comment 1: John Hadley, representing the North American Industrial Manufacturers

Comment 2: David Bernstein, consultant in toxicology

Comment 3: Aubrey Miller, EPA

Comment 4: Mark Maddaloni, EPA Region 2

Comment 5: David Bernstein, consultant in toxicology

Comment 6: Joel Kupferman, New York Environmental Law Project

Comment 7: Ralph Zumwalde, NIOSH

Comment 8: Suresh Moolgavkar, University of Washington

Comment 9: Jay Turim, Sciences International, Inc.

⁷One panelist, when reviewing a draft of this report, indicated that death certificate data typically use a single code for all non-malignant respiratory disease. He added that asbestosis probably accounts for a minority of these deaths when compared to chronic obstructive lung disease.

⁸When reviewing a draft of this report, another panelist indicated that animal studies have found that animals are also not capable of completely clearing fibers of all lengths to background levels.

⁹The panelists noted differences in asbestos-related cancers in rats and humans. One panelist said that lung cancer in rats tends to be bronchioalveolar, and develops in the distal lung, while lung cancer in humans largely tends to occur in proximal areas of the lung. He wondered if differences in fiber deposition patterns (due to differing airway sizes and branching patterns) might explain differences in where lung cancers develop in rats and humans. Another panelist cautioned against expecting that lung cancer would develop in the same parts of the lung in rats and humans, primarily because of the confounding factor of cigarette smoking in humans.