Introduction

Human exposure to nanomaterials, i.e., particulate matter with a size of less than 100 nanometers in at least two dimensions, has occurred during the existence of mankind: the main sources being natural or artificial combustion processes. Typical industrial exposures of comparatively recent origin have been, e.g., fumes created through welding and flame cutting processes, exhausts from combustion engines, and others. But also the size distribution of some industrially produced pigments, catalysts, etc. have long comprised the nano range and thus created occupational exposures. Health effects occurring in exposed persons as such have been studied over years, however, typically either under the aspect of general dust exposure or with a focus on potentially substance-related toxic effects depending on the chemical composition of these dusts. Only comparatively recently, with the increasing technical ability to purposefully design and produce nanomaterials of defined chemical composition and geometrical structure, which do not occur in natural environments has the scientific attention been drawn to potential health effects that are not primarily associated with the chemistry of these materials but related to unique properties associated with the minuscule size and thus enormously large specific surface area of such particles—the subject of nanotoxicology had emerged (Oberdörster et al. 2005).

The increasing growth of nanotechnology in the workplace, and results from experimental animal studies suggesting that engineered nanomaterials may pose a health hazard to employees all underscore the need for preventive measures (Seaton 2006; Schulte et al. 2007). The term prevention comprises all measures directed at minimizing the risk associated with a specific exposure, the early detection through medical surveillance of adverse health effects resulting from such an exposure and the treatment of diseases, if they occur in spite of the aforementioned efforts. Risk reduction can be achieved through appropriate organizational and engineering controls, which are available even for this new technology (NIOSH 2007). The present paper deals with the as yet unresolved question of occupational medical surveillance of employees handling nanomaterials.

Rationale for occupational medical surveillance

There exist a variety of concepts regarding the development of occupational medical surveillance, which can also be used on a situation with exposure to nanomaterials. Basically, the decision to carry out a targeted occupational medical surveillance requires (1) knowledge about the existence or at least possibility of an exposure to a health hazard, (2) knowledge about specific health effects caused by such an exposure, (3) the availability of tests with a known sensitivity and specificity to detect such health effects in an early, preferably reversible or treatable, stage and (4) establishment to a sufficient degree of the causal relation between exposure and effect. The latter requirement is primarily aimed at avoiding unwarranted fear and anxiety in the tested individual, as well as at the prevention of unnecessary seemingly protective interventions such as restrictions in fitness for duty, which might ultimately lead to job loss and other undesirable social consequences.

Characterization of exposure

There exists no “typical” nanomaterial but a huge variety of particles in the nano size range (Aitken et al. 2006). Important differences between materials involve, for example, size, shape, chemical composition and degree of agglomeration.

Up to now, there is no generally accepted exposure metric. Feasible parameters may be (1) mass concentration, (2) particle number, (3) total particle surface area, (4) particle type including geometry, (5) chemical composition and (6) meaningful contaminants (e.g., metals from catalyst use) (Wittmaack 2007). There is a wide agreement that mass concentration is insufficient to characterize exposure to nanomaterials, and convenient, although not yet widely available, measurement techniques presently exist mainly for particle number. Furthermore, the lack of standardized and uniformly applied measurement methods limits the comparability of measurement results across different companies, sites, plants, or types of workplaces.

Biomonitoring methods, usually the “gold standard” in occupational medicine for quantifying individual exposures, have not been developed for nanomaterials in humans, and it is questionable whether they will be available in the future. Methods applied in animal experiments (e.g., radio-labeling, magnetic tagging) appear not to be an option, and serum levels of the chemical constituents of nanomaterials will usually not provide a sensitive enough outcome metric because of the minute mass concentrations involved in nano-exposures.

While thus a qualitative exposure characterization based on technical use data in a given workplace should be possible, the quantification of workplace measurement results remains a challenge. It has further been demonstrated that measured particle counts do not necessarily represent genuine influences from workplaces, but have to be separated from background exposures which depend on geography, location, weather conditions, behavioral factors, etc. (Kuhlbusch and Fissan 2006). Therefore, as a prerequisite for any study on the effects of nano exposures at the workplace on the health of employees such crude quantitative results have to be expressed as workplace-related increment of exposure.

Identification of relevant health end points

Up to now, limited knowledge exists regarding potential health effects of nanoparticles in humans (Gwinn and Vallyathan 2007). It is mostly based on animal experiments with questionable relevance of routes of administration, especially intratracheal instillation of nanoparticle-containing solutions, as well as on cell culture-based in vitro experiments. Inferences were also derived from epidemiological studies on fine dust exposures in workplaces as well as in the general environment (Oberdörster 2001).

Although many of the purported toxicological properties of nanomaterials are attributed to size rather than to chemical composition, it is evident even from the limited available data that the toxic response may vary widely with the type of nanomaterial used (Gojova et al. 2007). If it comes to judging the extent of the size-associated effects of these materials, and the possible contribution of workplace exposures to the increment of risk as compared to the “background” situation, we may draw back on a wealth of occupational medical experience with “conventional” nano exposures from, e.g., welding fumes, diesel exhaust, flour dust. Nevertheless, a case-by-case assessment of new engineered materials will remain necessary. One of the major challenges in addressing the enormous variability of existing nanomaterials, however, is the lack of appropriate and generally accepted high throughput models (Maynard et al. 2006).

There exist several hundred published papers from in vitro assays on the biological behavior and effects of nanomaterials. The studies were targeted, e.g., on cell viability, phagocytotic behavior, signaling pathways, markers of inflammation, increased oxidative stress and other endpoints. Specific attention has been devoted to the question of skin penetration, with ambiguous results for different particle types and skin models (Baroli et al. 2007; Cross et al. 2007; Rouse et al. 2007; Gamer et al. 2006; Vogt et al. 2006). It is widely accepted that in vitro assays may serve as high throughput screening instruments and help to elucidate underlying mechanisms and principles of pathogenetic processes in toxicology; however, they rarely provide results which can be extrapolated to intact organisms, let alone human beings, in a straightforward manner (Wörle-Knirsch et al. 2006). A recent ECETOC workshop concluded that in the ongoing process of hazard identification, in vitro screening strategies have to be developed to assess the possible reactivity, biomarkers of inflammation and cellular uptake of nanoparticles; however, this process has to be validated using in vivo techniques (Warheit et al. 2007).

Animal studies have demonstrated unspecific inflammatory reactions as a potential cause for pulmonary fibrogenesis and cardiovascular effects. Tumorigenic effects have only been observed in rats, and might be related to overload conditions (Borm et al. 2006). No dose–response relations can be derived from these experiments for the time being. As a particular potential cause for concern the ability of nanoparticles has been mentioned to travel via blood stream or the olfactory pathway to the brain and to other organs (Oberdörster et al. 2004; Garnett and Kallinteri 2006). This consideration, however, would not only apply for engineered nanoparticles but also for the ones from conventional occupational and environmental sources, and thus not address a new phenomenon. The pathophysiological significance of such particle redistribution remains to be elucidated.

Epidemiological data specifically derived from nanoparticle-exposed humans are currently not available (NIOSH 2007). The applicability of research results from environmental epidemiology on airborne pollutants to nano exposures in the workplace is doubtful for several reasons. Firstly, environmental exposures tend to be more complex than workplace exposures. While they do comprise exposure to nano-sized particles, there are other factors like SO2, NOX and even traffic-related noise exposures that constitute a highly complex exposure situation where—even in the presence of measurable effects—the causal agent, let alone a group of interacting agents, cannot not easily be identified. Secondly, in most environmental epidemiologic studies the target groups showing effects related to particulate matter were children, sickly or elderly people; thus, susceptible subgroups that are not relevant to a major extent in work environments. Thirdly, environmental studies which did not explicitly take nanoparticles into account produce similar results like the ones quoted as supporting nanoparticles-induced health effects in humans (e.g., Yeatts et al. 2007).

In the absence of specific human data, the extensive experience with welders and flame cutters which has been gained over the past seven or eight decades may come closest to what may be a model to estimate the potential health effects of engineered nanoparticles. Welding produces fumes containing high numbers of partly ionized, partly oxidized, sometimes highly reactive metallic particles in the nano-size range, along with irritant and toxic gases like, e.g., ozone, carbon monoxide and nitrogen oxides (Zober 1981; Blättler 1998; Wieners et al. 2000; Antonini 2003). In a comprehensive overview, which addressed respiratory, dermal, central nervous system (CNS), carcinogenic and reproductive effects of welding exposures, a relative paucity of findings becomes evident (Antonini 2003). Decrements in lung function have been observed in small numbers of heavily exposed welders, infrequent cases of interstitial fibrosis have been attributed to improper workplace ventilation or mixed dust exposure, and CNS and reproductive findings have mostly been absent or inconsistent or attributed to other welding-related exposures such as radiant heat. A clear relationship between lung cancer and welding fumes has never been established. Consequently, the IARC classified welding fumes and gases as “possibly carcinogenic”, with limited evidence in humans and inadequate evidence in experimental animals (IARC 1990).

While this experience derived from welding can of course not serve to dismiss health concerns related to engineered nanoparticles, it may to some extent help to put too far-reaching concerns into perspective.

Availability of suitable medical tests

The potentially relevant health end points ascribed to engineered nanoparticles (cardiovascular, pulmonary, inflammatory) are unspecific, have a high prevalence in the general population, and share multiple non-occupational risk factors. The very few existing studies with experimental short-term human exposure to nanomaterials have failed to demonstrate reproducible health outcomes suitable for occupational medical surveillance (Beckett et al. 2005; Scharrer et al. 2007). Nevertheless, on the basis of findings in experimental animals a variety of potential effect parameters have been proposed for medical surveillance of nano-exposed employees, such as heart rate variability, blood-clotting parameters, pro-inflammatory cytokines, etc. None of these parameters are specific, most are not validated as individual health risk indicators, and some require sophisticated equipment not routinely available. Other tests like, e.g., ECG, chest X-ray and pulmonary function are well-established diagnostic tools for the assessment of symptomatic persons or high-risk populations with a known or expected typical pattern of findings. No such pattern is known with regard to nanoparticles; thus, no individualized interpretation of results would be possible.

Arguments pro and contra “nano-specific” medical surveillance

While our considerations so far dealt with the feasibility of medical surveillance, also the appropriateness of conducting medical surveillance can be judged according to established criteria referenced, e.g., by Borak et al. (2006): (1) burden of suffering, (2) accuracy and reliability of the examination method, (3) effectiveness of early detection, (4) harms of screening and (5) benefits outweighing harms.

The first criterion requires that the health condition to be averted must be sufficiently common in the surveyed population and pose a substantial threat to health to justify routine screening. While this is certainly true in the general population for some of the health end points discussed in the context of nanotoxicology, the utter lack of knowledge regarding the population attributable risk, if existing, associated with occupational nano exposures precludes any conclusions on this point.

The second criterion refers to the availability of tests with established and sufficient sensitivity, specificity and positive predictive value for the outcome of interest. As we have already pointed out such data is currently lacking.

The third criterion implies that screening is justified only if finding the condition in an earlier stage leads to a better health outcome than if screening was not performed. This may be the case for some of the conditions discussed in the given context, e.g., early diagnosis of a cardiovascular disease may elicit a surgical intervention to prevent a myocardial infarction from occurring. This is, however, a general medical consideration that applies with or without exposure to nanomaterials. In the context of occupational medical surveillance this criterion further implies workplace interventions can be conceived that can mitigate or even remove a work-related causal factor and thus improve the health prospect of the affected individual.

This points directly to the limitations imposed by the fourth criterion, potential harms caused by the screening (Schulte and Salamanca-Buentello 2007). These harms include adverse effects of the tests themselves, e.g., chest radiographs are associated with a certain, albeit small, risk of inducing a malignancy through ionizing radiation (Konietzko et al. 2001). This known small risk has to be put into perspective against the unknown and unpredictable benefit of searching for results that no one can even describe for the time being. These harms also include the stigmatizing and distressing effects of false-positive tests followed by unnecessary further invasive procedures and eventually treatment.

The fifth and last criterion, whether benefits outweigh harms, can obviously not be judged on the basis of existing knowledge.

Taken together, we conclude that there is currently no basis for targeted “nano-specific” occupational medical surveillance.

Current concepts for the safe handling of nanomaterials

The substance-related health risk at a given workplace is commonly defined as hazard (specific for a given substance) times magnitude of exposure. It is obvious that in the absence of sufficient hazard information the outcome of this equation cannot simply be assumed to be zero. Against this background, BASF currently puts specific weight on exposure control in workplaces. At workplaces in research and production where nanoparticles are handled, risk assessments have been conducted through monitoring of exposures using state-of-the-art methods. For the safe handling of nanoparticles BASF has established specific working procedures, which are summarized in a respective guideline (BASF 2004). Particle emissions at the workplace are primarily avoided by manufacturing in closed systems or by using effective extraction systems. Appropriate personal protective equipment (PPE) has been defined for such operations where an exposure potential cannot be excluded.

As a consequence of the above considerations, workers handling nanomaterials are not subjected to any unusual medical surveillance. Instead, routine practices related, e.g., to respirator use or shift work are employed in accordance with current legal requirements, best practice and BASF policy.

Recommendations and conclusions

While there is no evidence-based foundation for targeted “nano-specific” occupational medical screening, general medical screening with methods aimed at some of the health outcomes under discussion can be performed, especially if they are of high relevance in the general population. Such screening should be devised weighing the risk to benefit ratio for the tests in consideration, keeping in mind the risks associated with untargeted medical surveillance. Where such a program is newly adopted, it should be communicated to the stakeholders that all examination outcomes whatsoever will have no consequences regarding employability, liability, etc.

The results of such general medical examinations can provide a basis for future epidemiologic studies. Given the high incidence of the suspected end points in the general population, it will take a long time and a large number of “truly” exposed persons to achieve enough statistical power to detect workplace-related effects. Nevertheless, gathering of health information in such cohorts might lead to the detection of rare but specific health outcomes, if ever they exist. Therefore, the establishment of exposure registries to enable the conduct of large-scale multi-centric prospective epidemiologic studies is recommended.