Engineered Nanomaterials: Learning from the Past, Planning for the Future

Objective: The ongoing explosion in creation and use of engineered nanomaterials leaves stakeholders in government, industry, and labor uncertain of how best to proceed in protecting worker health. Methods: A synopsis is presented of the conference Nanomaterials and Worker Health, along with considerations of prior, analogous challenges in occupational health. Results: Progress has been made in defining and addressing the occupational threat of engineered nanomaterials, but future success demands coordinated effort. Conclusions: The conference Nanomaterials and Worker Health laid necessary groundwork for collaboration to proactively and preemptively address the occupational health effects of engineered nanomaterials.

From diverse stakeholders came diverse perspectives, some reassuring about progress made so far and some unsettling regarding the challenges ahead (Table 1). Reports from toxicologists warned about the hazards of carbon nanotubes and nano-scale metal oxides in animal models, sometimes at far lower doses than comparable particles of larger size. 5,6 We learned that the measurement of ENM exposure lacks standard methods or validated metrics, 7 and further, we learned that we do not have information about how many and where workers are potentially exposed. 8 Risk assessment models using available animal and worker data paint a concerning picture of workplace exposures that may approach toxic levels despite being at the limit of detection with current technology. 9 More heartening, however, is research already investigating several candidate biochemical markers of ENM exposure. 10 In addition, novel assays like exhaled breath condensate and noninvasive cardiovascular monitoring may join existing screening tests to identify early disease. 11 And despite the dearth of easily accessible data, epidemiologic researchers have already made progress in identifying cohorts of workers with potential exposure for future studies. 12 Coordination of efforts is facilitated by a systematic approach. 13 A useful systematic approach to such public health challenges is modeled by the cascade of occupational health prevention (Fig. 1). Each step along the cascade represents an opportunity for intervening to prevent exposure-associated illness. One feature of the cascade is that success at "higher" levels of prevention can reduce the burden "downstream." For example, substitution and elimination of toxic chemicals make personal protective equipment less important; early detection of pathologic changes by preclinical medical examination may prevent the need for rehabilitation. Another element of the cascade is surveillance, which provides feedback to "upstream" processes. Data from environmental monitoring guide engineering controls, and clinical diagnoses inform biological monitoring. Working from a framework like the cascade for occupational health prevention is valuable, because it brings together disparate efforts into a cohesive system greater than the sum of its parts.
The conference illuminated work being done at several steps along the prevention cascade, as well as opportunities for improvement. Premarket testing of ENM toxicology is ongoing, 5 although the pace of new materials is breathtaking. 4 The NIOSH nanotechnology field research team is defining the state and limitations of environmental monitoring of ENM exposure. 7 Presentations were given on the effectiveness of and progress in personal protective equipment, 14,15 biological monitoring, 6,10,11 and preclinical medical examination. 16,17 Lacking, yet, is the integration achieved by incorporating surveillance feedback along every step of the cascade. A collaborative conference, such as this one, lays good groundwork for such coordination.  design and populate exposure registries as soon as possible requires consideration of three possible disaster scenarios for the burgeoning ENM industry. As a "worst-case scenario" analogy for the risks of and for the ENM industry, consider the explosion of the Zeppelin Hindenburg in 1937. Not only was this incident a tragedy for the individuals who died in the crash, but it was likely disastrous for lighter-thanair travel, particularly hydrogen based. Although political factors and advances in airplanes certainly contributed to the decline of the industry, the dramatic footage of the burning Hindenburg ("Oh, the humanity!") signaled the end of an era of commercial dirigibles. The possibility of a similarly high-profile, "game-changing," preventable incident occurring to the nascent nanotechnology industry-or perhaps for a defined segment of it like, say, carbon nanotubes-does not far stretch the imagination, provided our imagination is primed by our prior experiences with other fibers such as asbestos.
Copyright © 2011 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
The Hindenburg crash is an example of a "true positive" catastrophe, one in which the adverse effect is actually associated with the exposure. In contrast, we can describe a "false positive" event that could similarly cripple a growing industry. Call these possibilities type 1 and type 2 disasters, respectively. In a type 2 disaster, a cluster of adverse health outcomes is found in association with an exposure of interest, but the cluster is random and the association is just unlucky. Epidemiology and toxicology can reveal type 2 disasters to be nothing but statistical chance, but the research process can take time that may cost an industry and society dearly.
An illustration of a type 2 disaster is the case of video display terminals (VDT). Several clusters of spontaneous abortion were reported in 1980 in suspected connection with very-and extremelylow-frequency electromagnetic fields from cathode-ray tubes in VDT. Initial types of studies of occupational exposure to VDT yielded equivocal results, but many suffered from methodological flaws, such as using cohorts designed to answer a different question or assessing exposure by self-report. Radiation toxicology studies were inconclusive. Conduct of a proper epidemiologic study, with an appropriate control cohort defined and exposure to electromagnetic fields quantified in the field, took time. Fortunately for exposed women, the best-designed studies were clearly negative, showing no difference in abortion rates between cohorts and no dose-response effect with VDT exposure. 18 When a technology is as pervasive as VDT is (or ENM will be), such "false positive" clusters of adverse events are statistically inevitable. The interval between observation of chance clusters and definitive assessment will be a time of great uncertainty for industry, workers, and society that will inevitably impede the growth or unnecessarily increase the cost of this new and important technology.
A type 3 disaster is a public concern over an exposure due to fear of adverse events even in the absence of an outbreak. Consider the struggles of the genetically modified organisms industry. Promise for modern genetically modified organism technology has at times been comparable in breathlessness-a cure for world hunger and malnutrition-to that for ENM. Exciting products have been developed, such as "golden rice" containing β-carotene and "Bt-corn" that kills pests without synthetic insecticides. The most likely biological risks of the technology are similar to those for other agricultural techniques, such as cross-pollination or adverse effects on local ecosystems. Nevertheless, widespread public concerns have arisen about direct health effects on human consumers, concerns that, some say, are out of proportion to presently observed or reasonably expected consequences. The innovative and commercial promise of this technology has been retarded by bad public relations, even in the absence of a safety failure. Given the impressive novelty of ENM, similar scares are possible for its industry and should be preemptively addressed.
The public health challenge from ENM comes not only from its potential for diverse toxicologic effects 5,6 but also from its predicted ubiquity throughout industry and society. Given enough manufacturing plants, enough users along the value chain, enough consumers of ENM, eventually a high-profile cluster of health events will occur that can be plausibly linked in time and proximity to ENM exposure. This cluster may reveal an adverse effect of ENM exposure, previously unrecognized or underappreciated, a Hindenburglike type 1 disaster that might have been prevented with more forethought and risk management. Alternatively, the cluster may be a "false positive" type 2 disaster like VDT, simply a chance grouping of unrelated health events that is statistically inevitable as the industry grows. Yet another possibility is that the cluster will happen in a movie or be a prediction of protesters, and the chilling effect of a type 3 disaster on industry may be no less severe for the outbreak being imagined. Whether the adverse effect is fact, fluke, or fiction, the human and commercial impact will grow until epidemiologists are able to precisely define the risk.
The considerable costs to establishing and maintaining exposure registries can be justified if the registries will protect workers and industry against type 1, 2, and 3 disasters. A specific example in occupational health that illustrates the value of registries is that of dioxin. In 1980, Centers for Disease Control and Prevention epidemiologists were asked by Department of Defense to study the alleged adverse effects of the herbicide "Agent Orange" used by military forces in Vietnam. To augment studies of solder exposure in the field, NIOSH proposed a study of civilian production workers exposed to a related industrial compound (2-,3-,7-,8-tetrachlorodibenzo-p-dioxin). The epidemiologists located chemical plants where exposure could be assessed, enrolled study cohorts of workers, and prospectively monitored exposure and health outcomes. Conclusions of the study were reported in 1991 19 ; the time to mount the study and obtain an answer took a full decade. If preemptive registries can shorten the response time to event clusters that will inevitably affect the ENM industry and reduce the ensuing period of uncertainty, the cost of registries will be recouped many times over.
The usefulness of a registry depends on its comprehensiveness. One possible solution is to create a registry of companies: a list describing which companies are using which materials and processes. This kind of list that could be accessed to rapidly identify and assemble cohort groups of workers, saving the initial months to years of preparation necessary for finding appropriate industry cohorts to study an event cluster, is reported. The more detailed alternative proposed by many participants of this conference is a registry of workers: a list of individuals with their contact information, personal exposure history, and record of health outcomes obtained through regular surveillance. Examples were presented from the US Air Force and French Alternative Energies and Atomic Energy Commission experiences. 20,21 Such a registry would enable epidemiologists not only to design cohorts but also to immediately begin analyzing differences in exposure-controlled morbidity and mortality. After an event cluster, a preexisting worker registry would save the years to decades needed to prospectively follow individuals exposed to the suspected hazard. The savings in human health and public relations from quickly defining a "true positive," exonerating a "false positive," or assuaging nonspecific fear would easily justify the effort to build and maintain the exposure registry.
Although the future value of exposure registries is substantial, their implementation raises several nontrivial challenges, described in detail during this conference 22-24 that must be addressed. Issues of access, usability, and standardization of the registry database are crucial and will require coordination and cooperation among stakeholders. Agreements must satisfy industry needs regarding proprietary information and legal liability, needs that may otherwise be barriers to willing participation. In addition, a common understanding must be reached on the methods and appropriateness of medical surveillance of workers in the registry; such detailed consensus building might be the topic for a near-future workshop.
The contention surrounding medical surveillance centers on the clinical value to the worker of currently available screening tests. Concerns about the hazardous nature of ENM-supported to varying degrees by toxicology studies with animal models 5,6 and epidemiologic studies of ultrafine particles in diesel exhaust 25 -focus on diseases that are fairly common: restrictive pulmonary disease; cardiovascular disease and its diverse sequelae; malignancy of unspecified type; and systemic manifestations of chronic inflammation. These nonspecific endpoints, coupled with no biological markers of ENM burden, make individual screening a troublesome proposition from a clinical standpoint. Abnormal findings on pulmonary function tests, chest radiographs, or serum C-reactive protein may not be interpretable for the individual worker. 16 This uncertainty may be due to operating characteristics of the test (such as sensitivity and specificity) or due to potentially confounding risk factors (like smoking, diabetes, and other exposures). Clinically, uninterpretable test results are particularly problematic as exposure-related pathology may impact employability or have legal repercussions. Nevertheless, even a test that is experimental, insensitive, or nonspecific can have occupational health value for comparing cohorts. Different average results in two comparable groups may suggest an engineering control flaw, resulting in higher than expected exposure, for example, or may reflect an unexpected biologic response at exposure levels previously thought safe. In addition, careful informed consent procedures can obviate the problems of test results that are not actionable at the individual level. A major precept of medical screening in nonoccupational settings is that test results must have therapeutic or prognostic value to the patient; occupational health screening in exposed workers, however, may be justified despite an indeterminate clinical value of the test. Occupational screening can detect early indicators of adverse effects and thereby lead to reduction of exposure for coworkers, a worthy goal even if the screened individual does not directly benefit from the procedure. 26 Although existing methods and knowledge are incomplete, meaningful action is still possible. Examples of current medical surveillance were described in a variety of industrial and academic contexts. [27][28][29] The concerns raised at the conference about the limitations of medical surveillance with present technology, while important to address, do not outweigh the future benefits that detailed worker registries will accrue, namely aid to epidemiologists in rapidly addressing type 1, 2, and 3 disasters. Finally, we must keep in mind that populations other than workers may experience significant exposure in the future, so that proactive monitoring now may pay great dividends as ENM proliferate in society.

EPIDEMIOLOGIC RESEARCH: PREVENTIVE MEDICINE FOR A GROWING INDUSTRY
Epidemiologic research, in contrast to surveillance, is a preplanned study of limited duration that collects adequate-which often means detailed-information and samples regarding exposure to hazards, health outcomes, and other relevant variables defined in advance. The goals of analytic epidemiology are to characterize the relationship between exposure and outcome, often focusing on dose response. In contrast, exposure registries usually utilize data previously collected for administrative purposes and often have little information on confounding exposures. While registries are useful in recognizing hypothesized associations, they are of less value in assessing dose-response relationships.
Epidemiologic research is the means of confirming or refuting purported associations between hazards and health outcomes. Exposure registries of workers can provide data-the accumulation of which is the most time-intensive step in the research processneeded for epidemiologic analysis. Such research can save worker health by identifying true risk as well as reduce the commercially crippling uncertainty that accompanies suspected risk.
As described at this conference, exploratory epidemiologic studies of the ENM workforce have already begun, 12 and they should continue in earnest. All the necessary prerequisites of an epidemiologic study of ENM are present. Work in toxicology is prioritizing potential hazards. Surveys and registries are identifying vulnerable populations. Although agreement on standardized methods is yet needed, we can characterize exposure at least crudely. Through surveillance, we will assess morbidity and mortality. Perhaps, the most compelling reason to encourage and support epidemiology is that such an effort will push the development of novel methods and shape the careers of young researchers whose future expertise will be greatly needed as ENM become ubiquitous.
Several conference participants noted that we must be wary of "paralysis of analysis" or "letting the perfect be the enemy of the good." It is true that we do not yet know enough about ENM characteristics, measurement, toxicity, biological markers, and health outcomes to design precise surveillance, registries, or epidemiology.
Given the scope of the challenge, however, we will likely never have as complete knowledge as we would wish. Like the clinician with a sick patient whose illness does not fit neatly into a diagnosis, we must act on imperfect information. We must be bold and wise, and importantly, we must be unafraid to change course as needed.
One answer to analysis paralysis is adoption of the well-known Shewhart cycle of iterative process improvement as a model of how ENM stakeholders should view exposure registries and medical surveillance. In the cycle of "Plan-Do-Study-Act," we are in the first stage, planning our objectives and what processes may achieve them. The necessary next stage is to do implementation of the proposed processes; only after trying them can we study their effectiveness in meeting our objectives. Study leads to action on the causes of differences between goals and current results. The cycle then repeats with new plans informed by previous attempts. Registries and surveillance should be started according to our best knowledgebest guesses when necessary-with the understanding that they will be continually refined by future conferences and workgroups in light of emerging evidence.

CONCLUSION
Someday, society may look back on the early stages of the ENM industry and ask whether appropriate caution was taken. Regret for insufficient prudence is certainly part of the history of other widespread occupational exposures in the twentieth century, such as asbestos. Nevertheless, the participants of this conference should be able to look back in pride at the proactive steps taken in Keystone. The broad array of presentations and discussions helped to define the state of the art and lay foundation for future consensus and standardization. Stakeholders in government and industry together are accepting their ethical obligation to share the costs of this wondrous new technology with the workers who will inevitably bear the brunt of exposure.
The challenge is where to go hence. Clearly, industrial innovation will be raging forward. 30 This conference presented clear messages that we must hasten our pace along numerous fronts, including toxicology, assessment of potential occupational exposure, and detection of biological effects. Although we yet need agreement on certain crucial details, we have from this conference, a path on which to proceed with epidemiologic studies and exposure registries. The efforts of this conference and the meetings to follow it reduce the chances of widespread adverse health effects and commercially crippling uncertainty as we enter the brave new world of ENM.