3
Synthesis of Research Progress On Particulate Matter

INTRODUCTION

The committee’s first report set out the sources-to-health-effects framework (see Figure 1-1 in Chapter 1) that has been integral to the development of the particulate matter (PM) research portfolio and to tracking progress with the agenda (NRC 1998). The framework has proved useful for identifying needed elements of the research portfolio and for addressing integration of research findings in support of implementation of evidence-based control strategies. In this chapter, the committee reviews progress on each of the 10 original research topics (see Box 1-1 in Chapter 1), summarizing the gains in scientific knowledge for each from 1998 until the middle of 2002 with some additional updating over the next year as this report was written and particularly relevant contributions were made. The committee also considered the remaining uncertainties, and what remains to be done. In addition, the committee assessed the studies related to each topic quantitatively and qualitatively according to the five criteria listed in Chapter 2. The committee’s more extended evaluations of the progress are provided in Appendix C. The focus of the committee’s evaluation has been research funded by the U.S. Environmental Protection Agency (EPA) with additional consideration of research funded by other organizations in the United States and abroad.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress 3 Synthesis of Research Progress On Particulate Matter INTRODUCTION The committee’s first report set out the sources-to-health-effects framework (see Figure 1-1 in Chapter 1) that has been integral to the development of the particulate matter (PM) research portfolio and to tracking progress with the agenda (NRC 1998). The framework has proved useful for identifying needed elements of the research portfolio and for addressing integration of research findings in support of implementation of evidence-based control strategies. In this chapter, the committee reviews progress on each of the 10 original research topics (see Box 1-1 in Chapter 1), summarizing the gains in scientific knowledge for each from 1998 until the middle of 2002 with some additional updating over the next year as this report was written and particularly relevant contributions were made. The committee also considered the remaining uncertainties, and what remains to be done. In addition, the committee assessed the studies related to each topic quantitatively and qualitatively according to the five criteria listed in Chapter 2. The committee’s more extended evaluations of the progress are provided in Appendix C. The focus of the committee’s evaluation has been research funded by the U.S. Environmental Protection Agency (EPA) with additional consideration of research funded by other organizations in the United States and abroad.

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress RESEARCH TOPIC 1. OUTDOOR MEASURES VERSUS ACTUAL HUMAN EXPOSURES What are the quantitative relationships between concentrations of particulate matter and gaseous copollutants measured at stationary outdoor air monitoring sites and the contributions of these concentrations to actual personal exposures, especially for potentially susceptible subpopulations and individuals? Introduction Compliance with the National Ambient Air Quality Standards (NAAQS) for PM is ascertained by measuring ambient concentrations of PM at monitoring sites. With regard to the health effects of air pollution, the risks depend on personal exposure—that is, the exposures received by people in the various specific places, conceptualized as microenvironments, where they spend time. Total personal exposure represents the time-weighted average of particle concentrations in the microenvironments where people spend their time. Exposures to particles generated by outdoor sources take place not only outside but also in indoor environments where the particles penetrate. Indoor particle sources, such as cigarette smoking, insects, molds, and cooking, may thus contribute substantially to total personal exposure to particles. Research carried out in regard to this topic addresses the relationship of monitoring data for ambient air with personal exposures to PM and gaseous copollutants. Data on this relationship are needed not only for healthy people but also for those persons who are particularly susceptible to air pollution and at greatest risk for experiencing adverse effects. Such persons are referred to collectively as a “susceptible subpopulation” and are further addressed under topic 8 later in this chapter. What Has Been Learned? Research findings on topic 1 are relevant to interpreting the findings of the epidemiological studies of PM and to furthering the understanding of the relevance of monitored ambient concentrations for public health protection. Before 1997, the majority of time-series studies of morbidity and mortality associated with PM had relied on ambient air measurements taken for regulatory and tracking purposes. In using these measurement data in

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress the time-series studies, the researchers assumed that outdoor particle concentrations serve as a valid surrogate of personal exposures to ambient particles. Previous findings from monitoring studies had suggested that personal exposures differ from ambient concentrations because of particle sources in key indoor microenvironments (Dockery and Spengler 1981; Ozkaynak et al. 1993; Ozkaynak et al. 1996). In addition, most of these investigations found weak associations, often not statistically significant, between personal exposure and ambient concentrations when assessed cross-sectionally (at different locations for different people). However, these conclusions were based on a relatively small number of studies, which were originally designed to determine population exposure distributions rather than to examine the degree of association between personal exposures and ambient concentrations. For interpreting the time-series studies of air pollution and health, an understanding of the pattern of association between ambient concentrations and personal exposures over time was needed. To address this knowledge gap, the committee recommended that longitudinal panel studies of personal exposure to PM be conducted (NRC 1998). In such studies, particle and gaseous copollutant exposures of groups of individuals would be measured at successive points in time to examine the relationship between personal exposures and the corresponding ambient concentrations. Further, these studies would attempt to identify factors influencing the observed relationships. The recommended exposure assessment studies would include not only healthy individuals but also emphasize individuals susceptible to the effects of particle exposures, including persons with chronic obstruction pulmonary disease (COPD), cardiovascular disease, and asthma, as well as children and older adults. As a result of the committee’s recommendations, a large number of particle exposure studies were conducted in several cities in the United States with different climatic conditions and air pollution mixtures. Studies were also conducted in Europe and South America. Below we summarize the major findings that have emerged from either the initial or the completed analyses of the collected data. Relationship Between Personal Exposures and Ambient Concentrations Results from the recent panel studies support the hypothesis that ambient PM2.5 concentrations are significant predictors of corresponding personal exposures, over time, for the investigated cohorts (Ebelt et al.

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress 2000; Evans et al. 2000; Rojas-Bracho et al. 2000; Sarnat et al. 2000, 2001; Williams et al. 2000a,b; Rodes et al. 2001). Stronger associations were observed as the number of repeated measures per individual increased from a few up to 15 days and were substantially higher than those determined when including all panel data (cross-sectional analysis). Most of the panel studies found little spatial variability of ambient PM2.5, thus suggesting that spatial variation is not a strong determinant of the level of correlation within the areas studied. Collectively, the results from the panel studies, performed on several hundred individuals across various cities and different seasons, showed that there were varying degrees of association between personal exposures and ambient concentrations for the measured individuals, with almost half of the associations being non significant. For those individuals and for PM2.5, the correlation coefficients were in the range of 0.4 to 0.9. Impact of Nonambient Sources Near-real-time PM2.5 indoor measurements have underlined the importance of nearby sources (such as cooking stoves) that are within the various microenvironments in which people spend their time (referred to as microenvironmental sources) (Abt et al. 2000; Howard-Reed et al. 2000; Long et al. 2000; Rea et al. 2001; Vette et al. 2001). However, exposures in specific microenvironments have not been shown to have a strong effect on the relationship of personal exposures with ambient concentrations over time. This generally weak relationship may be explained by the fact that indoor source uses and associated emissions are intermittent and, when averaged over 24-hr sampling periods, contribute to a small fraction of the total PM2.5 exposure variability for a particular person over time. This may not be true for certain exposure situations, for instance, when ambient concentrations measured at stationary outdoor air-monitoring sites are low and individuals are exposed to strong sources in specific microenvironments (such as cigarette smoking, wood burning, and motor vehicle traffic). Impact of Ambient Concentrations on Personal Exposures The fraction of ambient particles that penetrates indoors varies considerably (from approximately 0.3 to 1.0), and it increases with the home air-exchange rate (Sarnat et al. 2002). Air conditioning and patterns of home

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress activities, such as opening windows and doors, which influence home air-exchange rates, are factors determining the rate of particle penetration from outdoor into indoor air. Consequently, individuals residing in homes with higher air-exchange rates tend to be exposed to higher fractions of ambient particles, and their personal exposures tend to be more strongly associated with ambient concentrations (Sarnat et al. 2002; Liu et al. 2003; Wallace et al. 2002). In addition, fine-particle penetration from outdoor air into indoor air depends on particle size; the penetration of the accumulation mode (aerodynamic diameter, approximately 0.3 < da < 1 μm) is higher than that of the ultrafine mode and larger particles (Long et al. 2001; Vette at al. 2001). Until recently, the variability in a person in particle exposures was thought to be primarily from microenvironmental sources. However, there is now strong evidence that a great fraction of this variability is due to the varying impact of ambient sources on the indoor environments and therefore on personal exposures (Landis et al. 2001; Williams et al. 2002). Sulfate is associated mostly with outdoor particle sources and has been used to determine the contributions of outdoor and indoor sources to personal exposures (Wilson and Suh 1997; Ebelt et al. 2000; Oglesby et al. 2000; Sarnat et al. 2000; Landis et al. 2001). Sulfate is a suitable tracer for the accumulation mode; however, it may overestimate the penetration of ambient ultrafine and coarse particles indoors. As suggested by both fine-particle sulfate and mass measurements, the fraction of ambient particles to which populations are exposed may depend on climatic conditions and home characteristics among other factors and may vary from 0.2 to 0.9 (Brown et al. 2003; Rodes et al. 2001). Cohort Effect When the longitudinal exposure studies were initiated, it was hypothesized that personal exposures may differ for different groups within a population because of time-activity differences among the investigated cohorts. To date, available evidence has not indicated intergroup difference, as hypothesized. Findings in a panel study of nonsmoking hypertensive African-Americans were compared with those in a multiracial cohort of individuals with implanted cardiac defibrillators (Wallace et al, unpublished material, 2003). Time-activity patterns exhibited considerable intracohort variability; however, statistically significant intercohort differences in PM2.5 exposures were not found. In another study, personal PM2.5 exposures were measured for healthy nonsmoking senior citizens, school children, and

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress individuals with COPD in two cities (Brown et al. 2003). No significant differences among the groups were found in the relationship between personal exposures and ambient concentrations. Exposures to Gaseous Copollutants In several cohort studies, simultaneous 24-hr personal PM2.5, ozone (O3), sulfur dioxide (SO2) and nitrogen dioxide (NO2) exposures and corresponding ambient concentrations were measured using a personal multipollutant sampler. The findings of these studies suggested that PM2.5 personal exposures and ambient concentrations were correlated over time, and personal exposures to O3, SO2, and NO2 were not correlated with their respective ambient concentrations (Sarnat et al. 2001). In contrast, PM2.5 personal exposures were also correlated with O3 and NO2 ambient concentrations. Similar findings in other locations would imply that using ambient gaseous concentrations in multipollutant health-effects models along with PM2.5 might not be appropriate, because the ambient gaseous and PM2.5 concentrations are serving as surrogates for PM2.5 exposures (Williams et al. 2000c; Vette et al. 2002). No results are available, however, for carbon monoxide (CO) because short-term or continuous personal exposure measures have not been made for this pollutant gas. What Remains To Be Done? Substantial progress has been made in answering the research questions related to topic 1. The committee was able to identify a large number of studies conducted in various locations, such as Baltimore, Boston, and Research Triangle Park, NC, that had been initiated following the first report. For most, the field work is now complete, and results are being published in the peer-reviewed literature. Advances have been made in personal monitoring, and data can be feasibly collected, not only from healthy adults but also from children and persons with chronic heart and lung diseases. The monitoring studies provide the important and generally consistent finding that ambient particle concentrations are a key determinant of the longitudinal variation in personal exposure to particles for those groups studied to date. This finding is critical for interpreting the time-series analyses as well as other epidemiological studies of particles and health. Although substantial data have been collected, they are not sufficient to develop a national perspective on the relationship between ambient PM

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress concentrations and personal exposure, because data are lacking for fully representative persons and locations. Also, there is still little information about the exposures of susceptible individuals to particles and other air pollutants. Further studies are needed, particularly on those persons at the highest risk for illness or death. RESEARCH TOPIC 2. EXPOSURES OF SUSCEPTIBLE SUBPOPULATIONS TO TOXIC PARTICULATE MATTER COMPONENTS What are the exposures to biologically important constituents and specific characteristics of particulate matter that cause responses in potentially susceptible subpopulations and the general population? Introduction Research topic 2 extends research topic 1, shifting the emphasis on exposures to specific types of particles that have been found to be associated with greater risk for health effects. In the committee’s portfolio, research related to topic 2 would be implemented only after the work under topic 5 sufficiently advanced understanding of particle characteristics that determine their toxicity, as discussed below. What Has Been Learned? Before 1997, very little information existed on particle exposures and chemical composition and size characteristics of the particles. Therefore, the database on exposures to particles in relation to the characteristics of the particles, particularly those considered to convey toxicity, needed to be expanded. The committee highlighted the need to characterize the physical and chemical properties of exposure particles for both the general public and susceptible subpopulations. Specifically, population-based field studies would provide information on the distribution and intensity of exposure for defined components and specific size fractions. In addition, longitudinal studies would investigate the relationship between personal exposures and ambient concentrations for specific components and particle size fractions. Toward that end, the committee suggested that state-of-the-art personal exposure measurement methods be developed and implemented. Subse-

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress quently, comprehensive and cost-effective field studies would be designed to determine population exposures, building on the results from the longitudinal panel studies (topic 1). To date, the research conducted on exposures to the toxic components of PM (such as metals and organics) in a limited number of exposure studies focused on methods development and applications of speciation techniques. Those efforts will be useful in the initial chemical characterizations of exposure particles and in the design of future exposure studies. However, the techniques can only be fully implemented in exposure studies after ongoing and future toxicological studies identify components of biological relevance. Specific progress is detailed below. Personal sampling devices have been developed and field tested. These methods make it possible to obtain information on personal exposures to different particle fractions and their components. More specifically, new methods have been developed for PM10 and PM2.5, ionic species, elements, elemental and organic carbon, and organic compounds (Demokritou et al. 2002). In addition, new personal sampling devices allow for the simultaneous collection of gaseous copollutants, PM2.5 and PM10, and particle composition (Chang et al. 1999; Demokritou et al. 2001). The development of new sampling and analysis protocols in conjunction with the use of more sensitive analytical techniques makes it possible to improve measurement precision and accuracy. One of these advances is a decrease in the flow rates of air into sampling devices, making smaller personal sampling devices possible. Real-time personal exposure measurements of fine mass and ultrafine particles have been conducted and have demonstrated the importance of nearby (microenvironmental) sources in determining total personal exposures (Fischer et al. 2000). These measurements will be critical to efforts in identifying sources that contribute to personal exposures and link exposures to specific activities or events. Furthermore, state-of-the-art exposure health effects studies have conducted simultaneous real-time personal exposure and biological monitoring (Liao et al. 1999; Howard-Reed at al. 2000). This was done to link magnitude and duration of exposures to biologically relevant events. Specifically, studies have examined the relationships between real-time fine particles and adverse cardiac functions. A limited number of studies have conducted measurements of personal exposures to various particulate constituents, including sulfate, nitrate, ammonium, elemental and organic carbon, and elements (Ebelt et al. 2000; Sarnat et al 2000; Williams et al. 2000a,b). Such studies enable the investigation of relationships between personal exposures to specific particle constituents and the corresponding ambient concentrations.

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress What Remains To Be Done? Although monitoring methods are being developed for the goals of topic 2, the uncertainties associated with the topic remain largely unaddressed. The committee’s sequence of research calls for more substantial advances under topic 5 before fully implementing topic 2. Exposure studies will be necessary for components of biological relevance. These investigations should examine relationships among personal exposures to particle components of biological relevance and corresponding ambient concentrations for susceptible subpopulations and the general public. Some of these studies should characterize exposure distributions for a variety of microenvironments, such as work, school, and transportation environments. The data from EPA’s Speciation Trends Network may provide a useful starting point for designing exposure studies that will give a national perspective and explore geographic differences in patterns of exposure. RESEARCH TOPIC 3. CHARACTERIZATION OF EMISSION SOURCES What are the size distribution, chemical composition, and mass emission rates of particulate matter emitted from the collection of primary-particle sources in the United States, and what are the emissions of reactive gases that lead to secondary-particle formation through atmospheric chemical reactions? Introduction A large variety of anthropogenic and natural emission sources contribute to airborne PM (see Box 3-1 and Table 3-1). Emissions from these sources need to be characterized for several purposes, including health effects research and implementation of PM standards. The development of laboratory exposures that assess the toxicity of emissions from specific sources (topic 5) requires knowledge of the characteristics of emitted and secondary particles. Emissions from critical sources need to be well-characterized for confidence to be placed in the source- and receptor-oriented air quality models (topic 4), particularly as they are used to develop emission-control strategies for achieving the PM NAAQS. In its second report (NRC 1999), the committee noted that traditional emission inventories have

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress BOX 3-1 Emissions and Emission Inventories Particulate matter is produced by a diverse array of emission sources. Some emissions emerge directly in the liquid or solid phase, as “primary” particles, such as combustion nuclei or mechanically generated dusts. Additional particulate species, such as sulfuric acid and its ammonium salts, ammonium nitrate, and a diverse array of organic compounds, are produced in the atmosphere by reactions that involve precursors emitted as gases. This “secondary” PM can condense on existing particles, form new particles through homogeneous nucleation, or be left as the residue of evaporated cloud droplets. An emissions inventory is an accounting of emissions. It is typically based on a census of source types (for example, number of automobiles and powerplants), their activity (number of kilometers traveled, British thermal units generated from burning fuel), and average emission factors (grams of emissions per kilometer, kilograms of emissions per million British thermal units). Emissions from a class of sources are expressed as a rate (kilogram per day) that are typically estimated as the product of source activity (for example, kilometer per day for motor vehicles or British thermal units per day for boilers) and an emissions factor (for example, kilogram per kilometer or kilogram per British thermal unit). All emission inventories involve uncertainties arising from the use of a single emissions factor to characterize all the individual elements in a broad class of sources. Inventories of primary-particle emissions present a number of special challenges. One is that direct comparisons with observed ambient concentrations are confounded by the possible contributions of secondary material. For example, particulate ratios of organic carbon to elemental carbon may be higher in ambient observations than in emission inventories because either (1) the inventories underestimate important sources of primary organic particles (such as oil-burning vehicles or vegetative combustion), or (2) the ambient particles contain secondary organics formed in the atmosphere from hydrocarbons (such as biogenics) emitted as gases and thus not covered by the particle inventory. Similarly, semivolatile species might be gases in the hot effluent sampled at a combustion source, but they condense as PM at ambient temperatures. Some of the most important sources of primary PM are “fugitive” in the sense that their emissions enter the atmosphere as puffs of indeterminate extent released at unpredictable times rather than as well-defined flows through a chimney or tailpipe. Fugitive particle sources, which include wild and prescribed fires, open trash burning, construction activities, agricultural tillage, and unpaved road use, are particularly hard to characterize in terms of activity concentrations and emission factors. All the difficulties are

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress heightened when interest extends beyond mass emission rates to include particle composition and size, as needed for health effects studies. Emission inventories merit scrutiny because they are critical and inherently uncertain inputs to air quality management. A crucial cross-check is provided by comparisons with observed ambient concentrations, receptor models, and inverse applications of source models. focused on representing PM mass emissions, and it recommended the adaptation of realistic source-test methods and their widespread application to measure mass emissions, chemical composition, and size distributions of PM. The committee also emphasized the characterization of the emission rates of reactive precursor gases (SO2, oxides of nitrogen [NOx], ammonia, and volatile and semivolatile organic compounds). The committee’s final recommendation was the construction of comprehensive national emissions modeling systems and resulting inventories that are size and chemically resolved. Because particle toxicological studies are ongoing, and air quality simulation models for state implementation plans (SIPs) are currently being developed and tested, the committee called attention to the need for immediately starting research to improve the characterization of PM emission sources. What Has Been Learned? Substantial improvements have been made since 1997 in estimates of on-road mobile-source emissions, particularly from heavy-duty diesel trucks and buses, though significant uncertainties remain even for this source category. A national, multisponsor effort, involving EPA, was made to implement standardized test methods (Gautam et al. 2002) and conduct an intercomparison study of mobile-source emissions testing facilities in the United States (Traver 2002). Effects on emissions of changes in fuels (for example, low sulfur diesel and compressed natural gas), after-treatment devices (for example, catalyzed particle trap), and operating conditions have also been characterized; the findings have informed recent regulatory decisions by EPA and California’s Air Resources Board (CARB)—among

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress to be dependent on finding underlying mechanisms that would explain why cardiac effects could be anticipated. In response to the lack of a mechanistic underpinning in support of the epidemiological findings, the committee called for an ambitious agenda of carefully designed mechanistically based controlled exposure studies. Several categories of studies were listed, using three approaches: (1) controlled clinical sites, (2) animal toxicological studies, and (3) in vitro studies. What Has Been Learned? These approaches have provided new insights into mechanisms. A major gain in mechanistic understanding since 1997 involves an expansion in focus to cardiovascular and subtler pulmonary responses. In the past, investigations tended to focus on the respiratory tract as both the portal of entry for particles and the site where effects were manifest. It is increasingly recognized that the respiratory tract may serve as the portal of entry of particles that are related to health effects manifest in organs and tissue remote from the respiratory tract. Using existing epidemiological and experimental data, an interdisciplinary workshop suggested that mechanistic considerations should focus on alterations in the autonomic nervous system; ischemic responses in the myocardium; chemical effects on ion channel function in myocardial cells; and inflammatory responses triggering endothelial dysfunction, atherosclerosis, and thrombosis (Utell et al. 2002). In fact, recent studies in humans and animals have demonstrated alterations in the autonomic nervous system, cardiac repolarization, and endothelial responses in response to particles (Utell et al. 2002). Descriptive findings of electrocardiogram changes and vascular end points confirmed a role of ambient PM and surrogate particles on extrapulmonary organ functions. As a basic mechanism for these effects, local and systemic oxidative stress responses were identified, as was a central role of oxidative stress in response using in vitro models. Together with the shift in mechanistic focus, there were appreciable changes in the experimental systems used. For example, animal models used in recent years have changed appreciably, with more given to potentially susceptible animals defined both by age and disease conditions that more realistically reflect human disease. Chronic exposures of these animals have not been carried out mainly because of the practicality of sustaining colonies of animals for long periods. There has been an increased use of real-world particles, including CAPs and fine and ultrafine particles.

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress In contrast to earlier approaches involving exploration of mechanisms in highly focused studies, more integrative approaches are now being taken so that data from different disciplines can be integrated in a more cohesive consideration of biological plausibility. The result has been the development of hypotheses that focus on specific areas, including (1) inflammation, both pulmonary and systemic, with perhaps a key role played by reactive oxygen species (ROS); (2) alteration in immune competence; and (3) autonomic nervous system dysfunction. Although these mechanisms are often considered individually, they are undoubtedly interrelated. Reviews of each topic are presented below. Inflammation and Immunity The presence of an inflammatory response is an important issue, because inflammation may induce systemic effects, including an acutephase response with increased blood viscosity and coagulability, and possibly an increased risk for myocardial infarction in persons with coronary artery disease. In chronic respiratory diseases, such as asthma and COPD, inflammation is also a key pathophysiological feature. Chronic, repeated inflammatory changes of the airways may result in airway remodeling that leads to irreversible lung disease. Thus, inflammation may be involved in both acute and chronic effects. Recent controlled-exposure studies in humans indicate that several types of particles can induce an inflammatory response. Studies using CAPs, laboratory-generated carbonaceous ultrafine particles, and diesel particles have all provided evidence for effects on pulmonary or systemic inflammatory markers. For example, levels of cytokines, chemokines, and adhesion molecules following particle exposures in healthy humans have been altered in blood (Salvi et al. 1999; Ghio et al. 2000a; Frampton et al. 2001). These soluble molecules play an important role in blood-cell recruitment to atherosclerotic lesions and inflamed airways, suggesting that exposure to either CAPs or ultrafine particles may initiate endothelial and leukocyte activation, a key initial step in leukocyte recruitment. Similarly, studies in normal dogs exposed to CAPs from Boston’s air by inhalation showed increases in pulmonary inflammation measured by bronchoalveolar lavage and in circulating blood neutrophils related to increases in specific ambient particle components (Clarke et al. 2000). Another possible consequence of exposure is increased susceptibility to acute respiratory infection. Streptococcus pneumoniae-infected rats ex-

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress posed to PM demonstrated increased pulmonary burdens of bacteria, circulating white blood cells, extent of pneumococcal-associated lung lesions, and incidence of bacteremia (Zelikoff et al. 1999). Subsequent studies implicated the iron content in mediating these effects. These findings suggest that PM, especially the soluble iron component, affects the host immune response during pulmonary infection and helps to explain some epidemiological observations. Cardiovascular Effects There is growing clinical and epidemiological evidence that ambient air pollution can precipitate acute cardiac events, such as angina pectoris, cardiac arrhythmias, and myocardial infarction, with the majority of excess PM-related deaths attributable to cardiovascular disease. Clinical studies of young and older subjects exposed to CAPs have shown reductions in heart rate variability (HRV) and increases in blood fibrinogen levels (Devlin et al. 2000, 2003). In another study, cardiac repolarization and responses of the parasympathetic nervous system were blunted during recovery from exercise immediately after exposure to ultrafine particles (Frampton 2001; Frampton et al. 2002). Similarly, animal studies are linking exposure to PM with changes in cardiac function, including induction of arrhythmias and an increased incidence of myocardial infarction. Inhaled PM exacerbated ischemia in a model of coronary artery occlusion in conscious dogs. Exposure to CAPs significantly increased peak ST-segment elevation during a 5-minute coronary artery occlusion (Wellenius et al. 2003). Investigators have focused on systemic inflammation and alterations in vascular endothelial function to explain these cardiac phenomena. Humans exposed to ambient particles showed increased blood levels of endothelins, which affect vascular tone and endothelial function (Vincent et al. 2001a,b), and altered vascular tone assessed by an ultrasound technique (Brook et al. 2002). In summary, an impressive array of findings from in vitro, animal, and human studies have provided a much more robust understanding of the potential mechanisms responsible for particle-induced cardiovascular events. Although a definitive mechanism has not been established to explain either increases in cardiac arrhythmias or myocardial ischemia, it has become clear that particles are capable of inducing many of the intermediate steps that are linked to adverse cardiac outcomes.

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress Oxidative Stress Recent work has focused on oxidative stress as an underlying mechanism relevant to pulmonary, cardiovascular, and other systemic effects. PM generates ROS, which provide pro-inflammatory stimuli to bronchial epithelial cells and macrophages. These cellular targets release cytokines and chemokines, enhancing the response to allergens. PM might therefore act as an adjuvant that strengthens the response of the immune system to environmental allergens. Hallmarks of allergic inflammation include increased immunoglobulin E (IgE) production, eosinophilic bronchial inflammation, airway hyperresponsiveness, and increases of NO in exhaled air. Diesel exhaust particles (DEP) markedly enhanced the antibody response and lipid peroxidation in allergic animals, while pretreatment with an antioxidant minimized the response (Whitekus et al. 2002). These findings are consistent with human nasal challenge studies supporting the role of DEP as an adjuvant in an already established allergic response, as well as in an exposure to neo-allergens. More recent studies found that diesel exhaust inhalation increases inflammatory markers (such as lung neutrophils and eosinophils) in healthy volunteers, supporting the hypothesis that diesel exhaust can worsen respiratory symptoms. DEP alone might augment levels of IgE, trigger eosinophil degranulation, stimulate release of various cytokines and chemokines, and stimulate the TH2 pathway (Pandya et al. 2002). Taken together, these findings might be relevant in explaining the increased number and severity of asthma attacks related to acute or short-term increases in PM levels in an urban setting and could implicate DEP and other types of PM as factors in asthma exacerbations. ROS associated with exposure to PM might play a role in cardiovascular effects. Quinones and other compounds that produce ROS might contribute to disease-related vascular dysfunction caused by PM exposure. That possibility could become particularly relevant as understanding of the role of PM in endothelial dysfunction expands and could further explain the mechanisms underlying cardiovascular events. What Remains To Be Done? Despite progress since 1997, uncertainties still exist in the scope and significance of experimental data in explaining the epidemiological findings on risks of PM. There are important limitations in the understanding of the relevance of mechanisms observed in animal and in in vitro systems for

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress humans. That is particularly the case in extrapolations from high-dose animal exposure to low-concentration human environmental exposure. Similar problems occur in understanding the relevance of mechanistic observations from nonphysiological exposure routes, such as instillation, to the normal inhalation route of pollutant exposure. The findings from the clinical, animal, and in vitro experimental work have often not included dose-response relationships. Such dose-response studies are an important element of confirming a mechanistic basis in support of the epidemiological findings. In addition, similar physiological, cellular, and molecular responses to PM in different species help to provide a mechanistic underpinning to the epidemiological observations. To date, the mechanistic observations have been largely in the realm of physiological and cellular mechanisms. The molecular mechanistic basis for the observed health effects is yet to be explored but is a necessary approach in moving forward. This approach is likely to become increasingly important as the research community moves into the discipline of molecular epidemiology. Another major uncertainty relates to the lack of the understanding of the relationships between the mechanisms responsible for acute versus chronic health effects. As focus shifts to findings from epidemiological studies on chronic health effects, a similar shift will be required of the mechanistic studies. At present, it is unclear how the mechanistic findings from acute health effects studies will relate to the mechanisms underlying chronic health effects. Finally, much of the exploratory, hypothesis-generating research done to date has focused on identifying mechanisms. The next step is to more clearly understand mechanisms underlying exposure-response relationships, recognizing that it is likely that most mechanisms will have some element of exposure (dose) dependence. This issue is critical to understanding the relevance of the various mechanisms described in experimental systems to ambient PM concentrations typically encountered by people. RESEARCH TOPIC 10. ANALYSIS AND MEASUREMENT To what extent does the choice of statistical methods in the analysis of data from epidemiological studies influence estimates of health risks from exposures to particulate matter? Can existing methods be improved? What is the effect of measurement error and misclassification on estimates of the association between air pollution and health?

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress Statistical Analysis Introduction Statistical analysis of data is the basis for making inferences about the underlying relationship between health and air pollution from epidemiological data. The goal of this research topic is to develop appropriate methods to analyze collected data and to understand the potential influences of these methods on the inferences that are made. Rapid developments in computing hardware and in statistical software have fostered the development and application of increasingly sophisticated statistical methods for analysis of large and complex epidemiological databases. What Has Been Learned? At the time of the committee’s first report, several statistical models had been developed to analyze the relationship between daily health end points and daily air quality measures, which were widely used for analysis of time-series data related to morbidity and mortality. Other key issues, such as measurement error, harvesting,1 and spatial analytical methods, had not yet been addressed rigorously but were recognized as methodological concerns in interpreting the findings of time-series studies. To some extent, the statistical literature addressed these issues generically, but they had not yet been applied to the type of data collected in health and air pollution epidemiological studies. Since 1997, several new statistical methods have been introduced to analyze the temporal association between air quality measures and health. Because approaches to analysis varied widely among researchers, comparisons of findings across locations were complicated by the possibility that methodological differences in analytical methods, rather than biological differences in the effects of particles, contributed to differing levels of association across locations. The National Morbidity, Mortality and Air Pollution Study (NMMAPS) (Samet et al. 2000a,b) was a major effort designed partly to overcome that problem by applying the same methods to 1   The term harvesting refers to the question of whether air pollution leads to the death of people who are highly susceptible and near death (and die a few days earlier than they would have with no air pollution exposure) or whether air pollution leads to the death of people who are not otherwise near death.

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress data from multiple locations. The NMMAPS approach used data from multiple cities across the United State. The cities were selected solely on the basis of size, thereby avoiding bias from picking a particular city and assuring representativeness of the findings. Work by the NMMAPS investigators led to the identification of a problem in the application of the SPlus software’s GAM (generalized additive model) function as applied to air pollution time-series data (Dominici et al. 2002). That finding, along with more detailed assessment of the methods applied to time-series data throughout the 1990s, indicated other methodological issues that had a potential impact on the effect estimates and their standard errors. The magnitude of the bias varied in a complex fashion with underlying modeling assumptions and the data structure of particular locations. Further examination of those estimates (Ramsay et al. 2003a,b) indicated that the standard errors of the measures of association were systematically underestimated, resulting in the potential to increase the level of statistical significance. Given the implication of those new findings with regard to the time-series studies, EPA slowed closure of its criteria document for PM and organized a framework for reanalysis of key data sets. In November 2002, EPA convened a workshop at which several investigators presented their results after applying several methods to the same data sets. For some data sets, the results appeared to be robust across several alternative methods that were applied. In other cases, the results differed, sometimes to the point that results would be statistically significant under one method but not under another. The differences occurred not only within the widely used GAM framework but also between GAM and other approaches, such as the generalized linear model (GLM), and among assumptions used within the GLM framework (HEI 2003). To date, there is no consensus about which analytical method is “correct.” Researchers are confronted with the need to estimate relatively small associations in the presence of potential confounding by weather and seasonality. Until the implications of the alternative analytical approaches are fully understood and until there is some scientific consensus about the appropriate method to use, researchers must explore the sensitivity of results to alternative modeling approaches (Sarnat et al. 2000). A further source of sensitivity has come with the increasing use of the case-crossover design, an alternative individual-level approach to assessing exposure-response on short time frames. The optimal uses of this design are still being explored. One recent report compared the results from a case-crossover

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress analysis with the more widely used time-series methods (Fung et al. 2003). The time-series models inherently make assumptions about the appropriate time domain for air quality data to be related to health studies. Some recent studies considered continuous air quality data and suggested alternative exposure metrics (exposure lengths less than 1 day or peak exposures). Distributed lags have also been introduced to estimate the temporal relationship between exposure and response in more detail. Several studies considered the issue of mortality displacement or harvesting, using various analytical strategies. The majority of these studies found that a significant number of deaths cannot be attributed to harvesting alone (Zeger et al. 1999; Schwartz 2000; Dominici et al. 2000), and the findings of several others suggested that harvesting can be substantial (Smith et al. 1999; Murray and Nelson 2000). Significant methodological improvements have been made in other areas as well. The identification and treatment of spatial autocorrelation (an interdependence between variables in different locations) have been addressed in studies that examined patterns in health and air quality indices in several geographic areas (Krewski et al. 2000; Burnett et al. 2001). However, the effects of concurvity noted in time-series studies (Ramsey et al. 2003a) are also apparent in spatial analyses (Ramsey et al. 2003b). New methods have been applied that allow the combinations of results across several studies, for example, in several cities in which a common methodology was applied. What Remains To Be Done? Although the committee’s previous reports had found substantial progress related to this topic, recent findings on the sensitivity of time-series results to modeling approaches are an indication that further methodological research is needed. Time-series studies are likely to remain important for estimating the health effects of air pollution on populations, and a more complete understanding of the implications of modeling approaches is needed. Additionally, the issue of harvesting or mortality displacement needs further investigation, and the seeming discrepancy between the strength of associations of PM with mortality in the daily time-series studies and the cohort studies needs explanation.

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress Measurement Error Introduction Measurement error is inherent in most studies of environmental factors and disease, potentially affecting exposures of interest, confounding and modifying factors, and outcomes. In epidemiological studies, the individual’s exposure to pollutants of concern cannot be known for all relevant time averages. The difference between the actual exposure and the measured exposure is known as measurement error. Generally, three components are in this measure: errors due to instrument error; errors due to the unrepresentativeness of an air quality monitor; errors due to differences between the monitored pollution measures and the average actual exposure. There is substantial statistical and epidemiological literature on measurement error, but the committee identified a number of issues specific to assessing exposures to PM and the health consequences of these exposures. A particular concern is the use of central-site monitoring data as an indicator of personal exposure in the time-series studies. What Has Been Learned? Zeger et al. (2000) developed a framework for measurement error in the context of air pollution epidemiological studies. They showed that under a wide range of circumstances, measurement error might result in underestimates of the association between air pollution variables and risk for adverse health effects. In recent years, more data have become available to examine the statistical properties of measurement error. They include data on the statistical distributions of the differences between personal exposures to a variety of pollutants and ambient measures for the same pollutants (Sarnat et al. 2001). Other studies (Ito et al. 2001) have tried to characterize the geographic variability in pollution measures. These data enable some validation of the statistical assumptions made in the developed frameworks and will provide input data for models that will estimate the impacts of measurement error. Because data will be available for several pollutants, it will be important to address this problem in a multipollutant context. These analyses might be limited by the lack of reliable data for short-term personal exposures to CO. Mallick et al. (2002) explored the use of methods for adjusting for exposure measurement error in the Cox regression model used to describe mortality associated with long-term exposure to PM air pollution.

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress There are many issues that influence an understanding of the health risks associated with particles. Research is continuing on these issues; however some of them are associated with considerable uncertainty. Some are addressed elsewhere in this section (for example, model selection, measurement error) and in this document (for example, differential toxicity of different particulate components, impacts of simultaneous exposure to copollutants); however, there have been no general framework and methodology to consider the quantitative impact of the totality of these uncertainties. Such a framework with corresponding methodology could not only prove useful in identifying the most critical uncertainties but could also be used to set priorities. What Remains To Be Done? The limited application of the framework developed for measurement error suggests that measurement error per se will not negate the positive associations found between air pollution and health effects. More precise estimates of the magnitudes and statistical distributions of measurement error need to be incorporated into multipollutant models to provide more reliable quantitative estimates of the impact of measurement error and of the relative importance of the various pollutants on health impacts. Greater consideration of this issue will give more credence to risk assessments used to support regulatory decisions. Frameworks have been developed which consider many of the components which influence an understanding of the PM-health relationship, and sensitivity analyses have been undertaken for some of these components. A recent NRC report (Estimating the Public Health Benefits of Proposed Air Pollution Regulations [NRC 2002]) addressed this issue and recommended some possible approaches, including Monte Carlo analysis and decision analytic tools. SUMMARY To date, the greatest measurable gains have been made on the topics with a narrower scope, such as exposure assessment and dosimetry. Substantial new evidence on exposure to particles has been reported, and there is now an enhanced understanding of the determinants of personal exposures to particles in ambient air (topic 1). Substantial progress has been made in assessing PM exposures of healthy individuals as well as suscepti-

OCR for page 64
Research Priorities for Airborne Particulate Matter: IV - Continuing Research Progress ble subpopulations. Equipment and protocols for this purpose were available before the committee’s first report, and the new funding made available for this topic led to a clear advance in the available evidence. Although monitoring methods are being developed to assess exposures of susceptible subpopulations to hazardous PM components (topic 2), more substantial advances are needed in assessing the components themselves (topic 5) before fully implementing topic 2. Topic 6, dosimetry of particles, is of narrow scope, and an understanding of particle dosimetry in the lung had already been well-established. Dosimetry models have been enhanced in the past few years, although not yet sufficiently developed for those with chronic heart and lung disease. Research methods have been further elaborated, and insights have been gained into the statistical modeling of data on air pollution and health (topic 10). Substantial methodological research has yielded new analytical strategies and an enhanced understanding of several issues, including measurement error and possibly mortality displacement. Methods have been described for combining large amounts of data to detect the effects of air pollution with greater sensitivity. In addition, new methodological issues in time-series analyses have been identified and solutions proposed. Regarding the combined effects of PM and gaseous copollutants (topic 7), epidemiological and toxicological research has provided little indication that PM effects vary with levels of other major pollutants in ambient air; however, much research on topic 7 is needed. New knowledge about PM health effects in susceptible subpopulations (topic 8) has been developed in the past 5 years. Despite such advances in knowledge, substantial uncertainties still need to be addressed concerning those subpopulations. Finally, a critical information gap, which is related to the characteristics of particles determining risks to health and the sources of more hazardous particles, remains largely unaddressed. An understanding of health risks in relation to particle characteristics lies largely in the domains of topics 5 and 9, and information on their sources and concentrations is the focus of topics 3 and 4. Progress on topic 5 has been slow, despite its central place in moving forward on the committee’s agenda. In the final chapters of this report, we offer recommendations on how to move forward more quickly on this topic.