6

Characterizing Risks to Subjects in Controlled Human Inhalation Exposure Studies

The U.S. Environmental Protection Agency (EPA) has conducted controlled human inhalation exposure (CHIE) studies with the objective of producing transient and reversible biomarker or physiologic responses to inform about biologic mechanisms of pollutant effects but do not cause clinical effects. Health risks to participants cannot be assumed to be the same for each CHIE study. Risk levels will vary according to study design (such as exposure agent, concentration, and duration) and the health status or risk profile of the individual participants. In this chapter, the committee provides guidance on methods for characterizing risk levels associated with participation in CHIE studies. It is important to note that this chapter is about the risk of clinically adverse health effects, not the likelihood of transient and reversible biomarker or physiologic responses that these CHIE studies are designed to examine.

Each planned CHIE study must be approved by EPA and the Institutional Review Board (IRB) of record. EPA has the responsibility for oversight, and the approving IRB has the additional responsibility for monitoring the progress of the study and for withdrawing its approval, if necessary. One of the main considerations in deciding if a CHIE study should go forward is to assess whether reasonably foreseeable risks to the study subjects are outweighed by the utility of the study results for informing air quality management decisions (see Chapter 2). Therefore, risk characterizations are needed to weigh study-related risks and societal benefits.

AUDIENCES FOR RISK CHARACTERIZATION ASSOCIATED WITH CHIE STUDIES

There are three distinct audiences to whom risk characterization are communicated as part of designing, reviewing, and executing CHIE studies:

1. EPA Researchers. When designing the study, researchers consider the likelihood of adverse outcomes when developing and applying criteria for including or excluding potential study subjects.
2. IRB. When determining whether to approve a CHIE study, IRB members consider the risks to the participants as well as the expected benefits of the study. They also determine if sufficient safeguards (such as protocols for medical oversight) will be in place for the study subjects.
3. Potential study subjects. When deciding whether to participate in a CHIE study, individuals consider the risks associated with participating in the study. This is a critical dimension of informed consent in this process.

It is useful to further classify risks in terms of a time frame for any adverse responses that might be observed. Would a possible adverse outcome be manifest within 1 or 2 days after the exposure (an acute effect), months or years after the exposure (a chronic effect), or both? Thus, a CHIE study will require the enumeration of potential adverse outcomes of interest, the characterization of these outcomes as proximate or long term, and a decision about how risks will be communicated to the different audiences for these calculations.

EXCLUSION CRITERIA FOR SCREENING STUDY SUBJECTS

EPA researchers develop and the IRB of record approves inclusion criteria for potential study subjects, that is, specific characteristics (such as health status and age) that are required of potential subjects. In addition, exclusion criteria are developed and reviewed for precluding the involvement of potential subjects, based on risk factors (such as preexisting diseases or sensitivity to bronchoscopy or other monitoring procedures). For example, when CHIE studies seek to involve subjects with mild asthma, risk-related criteria are needed that would allow for each individual to be screened for possible acceptance into a study. Individuals who have medical conditions that increase the likelihood of adverse effects are considered to have a higher baseline risk than healthy individuals. The various exclusion criteria presented in the protocols of each of the eight CHIE studies reviewed by the committee can be classified into these general categories:

• Disease history (such as asthma, cardiovascular disease, or other acute or chronic medical condition),
• Allergies (such as allergies to chemical vapors or gases, adhesives, or pollens),
• Pregnancy status (such as positive pregnancy test, nursing),
• Medication (such as over-the-counter [OTC] anti-inflammatory agents, OTC pain medications, and Vitamins C or E),
• Prior toxicant exposures (such as being a current smoker or occupational exposure to vapors, dusts, fumes, and gases),
• Ability to exercise (such as being unable to perform required exercise),
• Medical test results (such as uncontrolled hypertension), and
• Other (such as dialysis, hepatitis B, or fainting in response to blood).

FACTORS THAT MIGHT TRIGGER AN ADVERSE OUTCOME

As indicated in Chapter 4, risks of adverse events temporally associated with a subject’s participation in a CHIE study might be affected by one or more of the following:

• Air pollutant exposures occurring independently from the CHIE study, several days prior to or during the multiday experimental protocols;
• Intended pollutant exposures during the experiments;
• Preexisting medical conditions or sensitivities of subjects to the CHIE study pollutant(s);
• Other experimental procedures during the CHIE study (such as blood sampling or bronchoscopy); and
• Chance occurrences of pathophysiologic events (such as a serious adverse cardiac or pulmonary event), although unrelated to air pollutant exposures, that might happen to subjects during the CHIE study.

Ideally, inclusion/exclusion criteria will remove participants at appreciable risk of an adverse outcome as a result of preexisting medical conditions or sensitivities of subjects to the CHIE study pollutant(s). As indicated in the previous section, EPA applies a broad set of criteria when selecting study subjects. Chance occurrences of pathophysiologic events unrelated to air pollutant exposures that might happen to subjects during the CHIE study would occur at a rate corresponding to a baseline of expected responses within the general population. The risk of adverse outcomes associated with other experimental procedures used during the CHIE study (such as bronchoscopy) typically are well characterized through extensive applications in many kinds of clinical studies, and this information could be directly communicated to the IRB and the participants as part of communicating the risks associated with the conduct of the CHIE study. If an adverse event occur during a CHIE studies in which study subjects exercise, there

might be uncertainty as to whether the exercise, the pollution exposure, or the two combined brought on the event. The goal is to have exclusion criteria that reduce the likelihood that events will occur due to exercise, though not all risk factors for events will be knowable. The situation that requires additional explicit characterization of risk corresponds to the risk of adverse response associated with the study-related exposures that occur in addition to pollutant exposures in the ambient environment. The committee focused on characterization of the risk associated with air pollutant exposures during the CHIE study and preexisting medical conditions or sensitivities of subjects to the CHIE study pollutant or pollutant mixtures.

WHAT ADVERSE OUTCOMES MIGHT BE EXPECTED AND WHEN? REASONABLY FORESEEABLE RISKS

As discussed in Chapter 2, reasonable foreseeable risks refer to the likelihood of effects for which there is some credible evidence to expect they might occur as a result of participating in a CHIE study. Credible evidence might be based on epidemiologic or toxicologic studies of the CHIE study pollutant or on other information (such as effects characterized in previous CHIE studies).

Epidemiologic studies have observed that acute effects (such as cardiovascular or respiratory response) in populations associated with ambient air pollution exposures might peak on the day of exposure, after 1 day, or over multiple days following exposure. Therefore, acute effects could be expected to occur during the period from immediately after exposure to several days later.

In contrast to acute adverse effects, observed chronic effects (such as increased incidence of lung cancer and ischemic heart disease) are associated with long-term exposures to air pollution. Such effects are considered to be associated with cumulative results that develop over longer periods. In such cases, what matters most are the effects of long-term exposure, rather than the results of a short-term exposure on any particular day. As CHIE studies typically impart a very small increase in the cumulative exposure to ambient air pollution over an individual’s lifetime, there is no credible evidence to suggest that chronic effects be considered among the reasonably foreseeable risks of those studies. However, because of associations between long-term exposure to air pollution and chronic effects, concerns have been expressed about whether CHIE study exposures pose an elevated risk of cancer and other chronic diseases (for example, see EPA, 2014a). Given those concerns, the likelihood of chronic effects needs to be included in informed-consent communications (see Chapter 7).

CHARACTERIZATION OF RISKS ASSOCIATED WITH CHIE POLLUTANT EXPOSURES

Once adverse outcomes are selected using the principle of reasonably foreseeable risks and some decision is made about whether these outcomes are likely to be acute or chronic, risk characterization is necessary. There are two possible methods for characterizing risks of acute and chronic effects:

• Quantitative approaches that obtain a risk estimate from a published epidemiologic study of ambient air exposures and adjusts the estimate proportionally according to the exposure duration of the CHIE study relative to the duration in the epidemiologic studies, and
• The use of an exposure scenario comparator (ESC) approach, which involves comparing experimental exposure concentrations and durations with ambient concentrations of similar magnitude and duration experienced by a population in everyday life at a certain location. For comparisons in which those characteristics are similar between experimental conditions and everyday life, risks are assumed to be the same. ESC approach examples are provided later in this chapter.

While EPA researchers and IRBs of record might seek quantitative estimates of risk of potential acute effects (for example, to estimate the increment of risk a CHIE study might add to the baseline risk associated with exposure to ambient air pollution), the committee focused on the ESC approach because it judges that approach to be the better alternative overall. That consideration is based upon the limited availabil-

ity of appropriate data for risk calculations and the large attendant uncertainty in the results (particularly uncertainty associated with estimating risks of short-term exposures from data on outdoor exposures for much longer periods of time). Also, because of the potential difficulty on the part of uninitiated individuals in understanding the implications of the quantitative results in the context of controlled short-term exposures, the committee judges that exposure comparators are more understandable than quantitative estimates of risk for the target audiences (researchers, IRB members, and potential study subjects).

The ESC approach also might serve the needs of the EPA researchers and IRBs. That approach would provide a useful context for considering chronic effects, which are considered to be associated with cumulated results of conditions developed over longer periods. In considering chronic effects, the incremental exposure of about 2 to 4 hours added by participation in a CHIE study to the cumulative ambient background exposure over the life of an individual would be extremely small and calculating a risk estimate would involve too much uncertainty, such that the risk estimate would have little meaning. Therefore, the committee strongly prefers the use of the ESC approach for the characterization of risks related to CHIE study participation.

USE OF THE EXPOSURE COMPARATOR APPROACH FOR CHARACTERIZING RISK

Use of comparative scenarios is most appropriate if the risk to the comparative population is likely to be higher than the risk associated with exposures in a CHIE study. Such a comparison involves the use of an ambient exposure concentration that is higher than the exposure concentration in the CHIE study and an ambient concentration duration in the comparative scenario that is at least as long as the experimental exposure duration. It is not advisable to compare a lower ambient concentration over a longer period compared to the CHIE study exposure concentration and duration. Attempting to represent an equivalent ambient exposure in this way introduces more uncertainty as to whether the risk in the comparative scenario is actually greater than that in the CHIE study.

The first step in developing a comparative exposure scenario involves identifying a population that is (or was) in a location with a higher ambient concentration for a longer period of time compared to the CHIE study experimental conditions. Insofar as possible, the comparative scenario ought to be one that the participants in the CHIE study can readily identify with and understand. However, recent improvements in U.S. air quality might make it difficult to find recent ambient concentrations that are appropriate for use in an exposure scenario. Instead, the scenario could include a population at a U.S. location in the past, or a present population in another country. Also, if a particular exposure regimen has been used numerous times in previous CHIE studies, without the occurrence of adverse effects, the accrued experience from those studies could be useful in developing a reasonable exposure comparator. We provide examples of comparative scenarios for exposure to diesel exhaust and PM_2.5.

Example: Comparative Exposure Scenario for Diesel Exhaust

Coble et al. (2010) report on a survey conducted between 1998 and 2001 of exposures to diesel-engine exhaust particles (DEPs), quantified as the airborne concentration of respirable elemental carbon, in seven underground mines in the United States in which diesel equipment was used. One of the mines (for limestone) had very little exhaust ventilation. In this mine, personal exposures to DEPs over a full shift (8 hours) for eight underground jobs (based on 97 personal samples) averaged between 313 and 488 µg/m³ (Coble et al., 2010, Table 4). By comparison, the participants in the DEPOZ study were exposed to 300 µg/m³ of DEPs for a total of 4 hours. Therefore, these miners were exposed over extended periods to higher concentrations of DEPs throughout each 8-hour shift than the participants in DEPOZ who also were exposed for only one or several days during a shorter period (4 hours). The researchers did not evaluate the health responses of the miners to the measured exposures.

Example: ESC Approach for PM_2.5

Ambient concentrations in the United States provide relevant data for exposure comparator scenarios. Table 6-1 presents the highest PM_2.5 1-hour concentrations recorded during 2014-2015 from EPA’s AirData website¹ and the corresponding 2-hour and 4-hour average concentrations from monitors designated for making National Ambient Air Quality Standards (NAAQS) compliance decisions. During that time there were 14 episodes recorded in the United States in which the 2-hour average PM_2.5 concentration exceeded 300 µg/m³, and two episodes in which the 2-hour average PM_2.5 concentrations exceeded 600 µg/m³. There were 9 episodes recorded in which the 4-hour average PM_2.5 concentration exceeded 300 µg/m³.

The U.S. Department of State conducts hour-by-hour PM air monitoring at the U.S. Embassy in Beijing, China, and at four U.S. consulates in China (U.S. Department of State, 2016).² At the embassy in Beijing, 1,920 episodes were recorded between 2008 and 2015 in which air concentrations of PM_2.5 remained greater than 300 µg/m³ for 2 consecutive hours, with the peak 2-hour average concentration reaching as high as 983 µg/m³. During the same period there were 32 episodes in which air concentrations of PM_2.5 remained greater than 600 µg/m³ for 2 consecutive hours.

In the KINGCON CHIE study, PM_2.5 exposure durations were 2 hours and the concentration range was 38 to 579 µg/m³ (see Table 4-1 in Chapter 4). A person, who happened to remain outside during one of the 1,920 episodes in Beijing near to where the ambient measurements were taken, would have been exposed to a higher concentration of PM_2.5 than were participants in KINGCON. The KINGCON application for IRB approval called for terminating exposure if the PM concentration exceeded 600 µg/m³ for 6 minutes, although confirmation could possibly take another 10 minutes. However, it is clear from the above that, even if that occurred, there were episodes in Beijing that lasted longer and with higher concentrations. Also we see from Table 6-1 that occasionally there have been exposure episodes in the United States that lasted longer and were at higher concentrations.

The OMEGACON CHIE study called for exposing participants for 2 hours to Chapel Hill airborne PM up to a maximum concentration of 600 µg/m³. Using the same terminating procedure as in KINGCON when the exposure concentration exceeded 600 µg/m³, actual achieved concentrations were as high as 470 µg/m³.

Thus, the highest possible 2-hour exposure concentrations of PM_2.5 in the OMEGACON and KINGCON CHIE studies were exceeded numerous times recently in Beijing. The PM_2.5 concentrations in the OMEGACON and KINGCON CHIE studies also were exceeded occasionally at various locations in the United States.

Evaluation of Exposure Comparisons Used in EPA CHIE Studies to Put Risks in Perspective

Each of the CHIE studies reviewed involved exposure to O₃, ambient PM_2.5, DEPs, or their mixtures. In addition, ENDZONE involved exposure to NO₂. In each IRB application, the planned exposures to these substances were placed in perspective by comparing them to exposures to the same substances in other situations. Quotations from the IRB applications regarding these comparisons are presented in Table 6-2. Many of those comparisons also are presented in the consent forms for potential study subjects.

__________________

TABLE 6-1 Largest Concentrations in 1-Hour PM_2.5 samples in 2014 and 2015 in EPA Air Data with Corresponding 2- and 4-Hour Average Concentrations

TABLE 6-2 Exposure Comparisons Used in EPA CHIE Studies

Each of the studies that involved O₃ exposure (DEPOZ, ENDZONE, SOZIAL, and GEMINOZ) exposed the volunteers to 300 ppb for 2 hours. Except for DEPOZ, which provided no basis for exposing volunteers to 300 ppb O₃, each of these studies compared the cumulative exposure from the planned 2-hour O₃ exposure (300 ppb over 2 hours or 600 ppb-hours) to the cumulative exposure resulting from 8 hours of exposure to O₃ allowed by the 8-hour NAAQS of 75 ppb (also 600 ppb-hours). (ENDZONE and SOZIAL incorrectly referred to this exposure as simply 600 ppb.) But a given cumulative exposure over 2 hours might possibly be more damaging than the same cumulative exposure spread out over 8 hours (see discussion below). Therefore, this comparison is of questionable relevance.

The studies that involved PM exposure (DEPOZ, KINGCON, OMEGACON, WOODSIE, and XCON) compared the planned PM exposure concentration to concentrations that could occur in different situations or environments. However, they involved exposures to PM mixtures that varied in particle size distribution and chemical composition, and only one of the studies, DEPOZ, had a reference for the comparison exposure scenario.

In DEPOZ the claims that DEP exposures of some truck drivers are generally about 100-300 µg/m³ and average 900 µg/m³ for some underground mines were not referenced. A recent assessment of DEP exposures in seven mines with diesel equipment showed much lower concentrations than 900 µg/m³ (Coble et al., 2010). The documented comparison in DEPOZ with exposures near roadways (Buzzard et al., 2009) was from a simulation study rather than from monitoring of DEP concentrations near roadways. The reported average DEP concentration of 364 µg/m³ was the highest of 10 measurements and it lasted only 9 seconds. The reported peak concentration of 860 µg/m³ was from a “typical acceleration test” in which DEP concentrations greater than 500 µg/m³ lasted only about 1 second. The committee considers it inappropriate to compare these very brief exposure situations, lasting only a few seconds, under simulated conditions, to the 2 hours of exposure to DEP planned for the DEPOZ study.

KINGCON and OMEGACON both state that exposures to PM in these studies will not exceed what would be encountered over 24 hours in a typical urban environment on a smoggy day. No reference is provided for this comparison. Because that statement might not be true in all cases, it would be much better to specify the U.S. urban centers where those concentrations were observed. Even if validated, a given cumulative exposure spread out over 24 hours might not involve the same risk as the same cumulative exposure occurring over only 2 hours.

WOODSIE compared the planned exposure concentration of 500 μg/m³ wood smoke PM to a number of exposure situations, including forest firefighters, people who use biomass fuels for cooking, people indoors in houses heated by wood burning, and people out of doors in cities where wood is used for heating. None of these comparisons were documented. Moreover, some of them seem inappropriate. Risky exposures to forest firefighters could be justified because they have accepted a high-risk societal responsibility to protect life and property. Also, they are likely to wear protective respirators at times to reduce personal exposures.

XCON compared CHIE exposures to exposures while driving along a heavily traveled highway in a city, such as Los Angeles, although this comparison was not documented. XCON also mentions an ongoing study that on some days exposed participants to concentrations as high as 1,181,000 UFP/cc with no symptoms of discomfort or clinically relevant responses, and a study in the Chapel Hill facility that exposed participants to 1 to 3 million UFP/cc with no clinically relevant responses. However, the durations of exposures in these studies were not mentioned. Also cited is Shah et al. (2008) (referred to as Frampton, 2008), which “exposed volunteers for several years to ultrafine [elemental] carbon particle concentrations of 10 million particles/cc and have not reported any clinically relevant changes.” However, “several years” apparently refers to the total time this group has been conducting these studies, as the exposures in the cited article lasted only 2 hours. Also, the Shah et al. particles had a median diameter of 27.9 ± 2.2 nm, whereas the instrument used to concentrate particles in XCON produced particles with diameters in the range 30-250 nm. Thus, the particles in the Shah et al. study were generally smaller than those in XCON and consequently could pose different levels of risk.

ENDZONE involved exposure to 500 ppb NO₂ in addition to the 300 ppb O₃ exposure. It is stated that previous human exposure studies have utilized higher concentrations of NO₂, but this statement was

not documented. It is further stated that the planned NO₂ exposures in this study were to be lower than have been measured around operating gas stoves. Two references were provided to support this statement (Goldstein et al., 1988; Leaderer et al., 1984). Leaderer et al. (1984) reported on NO₂ in houses with kerosene space heaters, not gas stoves. Furthermore, this reference was to an abstract that contained no information about NO₂ concentrations. A later publication by the same authors (Leaderer et al., 1986) apparently reported on the same study, and the highest reported average NO₂ concentration in 25 combinations, defined by location within house and number of kerosene heaters and gas stoves, was less than 50 ppb NO₂. Goldstein et al. reported on lung function versus NO₂ concentrations while cooking with a gas stove. This paper contains graphs showing roughly 200 highest average 5-minute NO₂ measurements, most of which are less than 200 ppb and only 5 appear to be greater than 500 ppb. The information provided on the ENDZONE consent form seems to imply that participants could expect to experience higher exposures to NO₂ around an operating gas stove than they will experience in the study. This is not consistent with the information provided in the two studies cited in ENDZONE.

In 2010 EPA promulgated, for the first time, a 1-hour NAAQS for NO₂ (100 ppb) (75 Fed. Reg. 6474 [2010]). Thus, NO₂ exposures planned for ENDZONE (500 ppb for 2 hours) exceeded this standard by a factor of 5. It is inconsistent to compare the planned exposure to O₃ to the O₃ NAAQS, while ignoring the fact that in the same study the planned exposure to NO₂ is five times higher than the 2010 NO₂ NAAQS.

As the above review indicates, most of the comparison exposure scenarios in these CHIE studies were inadequate. Many were undocumented, and those that were documented were not all appropriate.

Exposure regimes that produce the same cumulative exposure may not produce comparable effects, even if the exposure length is only a few hours. For example, exposures to variable O₃ concentrations over 6-8 hours can elicit somewhat larger decrements in forced expiratory volume for 1 second (FEV₁) than a constant exposure over the same time period even if the cumulative exposures are equal (Hazucha and Lefohn, 2007). Thus, the practice used in the IRB applications of comparing a 2-hour exposure to O₃ to an 8-hour constant exposure at the 8-hour NAAQS might not be a valid comparison even if the cumulative exposures are comparable. Similar situations might occur for exposure to PM and other air contaminants. Thus, the most valid comparisons to planned exposures in human exposure studies are comparisons to populations exposed to comparable or higher concentrations for comparable or longer times.

RECOMMENDATIONS

The committee recommends that risk-characterization objectives be addressed by using an ESC approach in which the risk associated with a CHIE study exposure is likely to be lower than the risk to the comparative population.

To illustrate that the risk associated with the participation in a CHIE study is likely lower that the risk to the comparative population, the comparative scenario involves a documented ambient exposure concentration that is higher than the exposure concentration in the CHIE study and an exposure duration in the comparative scenario that is at least as long as the experimental exposure duration.

The comparative exposure scenarios should be documented fully so that the reasonableness of the comparison can be evaluated. Comparative exposure scenarios should be based on populations in the United States, insofar as possible. When that is not feasible, scenarios involving populations with demographics and life styles as similar as possible to those in the United States should be given precedence.

In planning a CHIE study, EPA should obtain the appropriate monitoring data for exposure comparator scenarios using locations where populations are or were exposed to ambient concentrations exceeding the exposure concentration envisioned for the CHIE study. When developing such comparator scenarios, differences between the CHIE study subjects and the comparative population (for example, regarding health and susceptibility) should be considered. In addition consideration should be given to the reliability of the ambient exposure monitoring data available for the

comparison (for example, data obtained from personal monitors, fixed-site monitors, or geospatial estimates) and the potential inaccuracies in personal exposure estimates that can result from use of monitored ambient concentrations.

Such considerations in comparing study subjects with specific populations and experimental exposures with ambient pollutant concentrations also would be required in attempting to develop quantitative risk estimates. Alternatively, if a particular exposure regimen has been used numerous times in previous CHIE studies, without the occurrence of adverse effects, the accrued experience from those studies could be useful in developing a reasonable exposure comparator.

It is important to note that if a CHIE study is being proposed for which no appropriate ambient concentrations (past or present) can be found, and no previous CHIE studies without adverse effects are applicable, it might be an indication that the CHIE study requires further explicit justification or should not be conducted.

For communication with study subjects, risk should be characterized in a descriptive and comparative manner using an ESC approach. This should be useful in explaining the type of exposure that will be similar to exposure as part of the study. These should be evidence based with an explanation of how they were developed. Comparison should be accessible and familiar (see Chapter 7 for more detail). Consistent expression of these complex concepts for individuals with limited health numeracy is essential.