Nanoparticles are small-scale substances (<100 nm) with unique properties and, thus, complex exposure and health risk implications. This symposium review summarizes recent findings in exposure and toxicity of nanoparticles and their application for assessing human health risks. Characterization of airborne particles indicates that exposures will depend on particle behavior (e.g., disperse or aggregate) and that accurate, portable, and cost-effective measurement techniques are essential for understanding exposure. Under many conditions, dermal penetration of nanoparticles may be limited for consumer products such as sunscreens, although additional studies are needed on potential photooxidation products, experimental methods, and the effect of skin condition on penetration. Carbon nanotubes apparently have greater pulmonary toxicity (inflammation, granuloma) in mice than fine-scale carbon graphite, and their metal content may affect toxicity. Studies on TiO2 and quartz illustrate the complex relationship between toxicity and particle characteristics, including surface coatings, which make generalizations (e.g., smaller particles are always more toxic) incorrect for some substances. These recent toxicity and exposure data, combined with therapeutic and other related literature, are beginning to shape risk assessments that will be used to regulate the use of nanomaterials in consumer products.
Nanomaterials are a diverse class of small-scale (<100 nm) substances formed by molecular-level engineering to achieve unique mechanical, optical, electrical, and magnetic properties. Nanomaterials are expected to improve virtually all types of products (Royal Society, 2004), and commercialization of products that exploit these unique properties is increasing. However, these same properties present new challenges to understanding, predicting, and managing potential adverse health effects following exposure (Hood, 2004).
Widespread application of nanomaterials confers enormous potential for human exposure and environmental release. Technological development and applications are out-pacing research of health and environmental risks (Fig. 1). Like genetically modified organisms, the future of nanotechnology will depend on public acceptance of the risks versus benefits. Even beyond public acceptance, experience with past “miracle” materials (e.g., asbestos) advises caution in using novel substances without fully evaluating potential health risks.
As with larger-scale substances, risk assessment will be the basis of assessing and regulating nanomaterials to protect health and the environment. Many applications will likely have limited or manageable exposures and thus potential for effects; however, nanoscale compounds are not the same as their larger counterparts, and risks will be product specific (Hood, 2004; Royal Society, 2004). Although a precautionary moratorium on nanotechnology seems unwarranted, applications should have specific assessments of potential hazards, exposure, and likelihood of toxic effects. A symposium at the 2005 annual meeting of the Society of Toxicology presented recent findings in exposure and toxicity of nanoparticles that, with related research on fine-scale and ultrafine particles, will help support critical assessments of risk. These presentations summarized airborne exposures and aerosol quantification, skin penetration and potential for toxicity, inhalation toxicity of carbon nanotubes and other types of particles, and implications for risk assessment.
CHARACTERIZING AIRBORNE EXPOSURES TO NANOMATERIALS
While ingestion and skin penetration are potential exposure routes for engineered nanomaterials (Oberdörster et al., 2005), the inhalation route for airborne nanomaterials has perhaps received the most attention (Maynard and Kuempel, in press). For some materials, studies have shown that the toxicity of inhaled particles increases as particle size becomes smaller and as the overall surface area of inhaled material becomes larger (Maynard and Kuempel, in press; Oberdörster et al., 2005). The relevance of these findings to human health depends on the amount of material released into air during research or production, how its physical and chemical properties change while in the air, the likelihood of the material being inhaled, and the effectiveness of control measures to prevent inhalation. For example, handling of carbon nanotubes resulted in very low airborne concentrations of nanoparticles, consistent with the tendency of this material to aggregate into unrespirable larger masses (Maynard et al., 2004). Evaluation of the factors affecting exposure requires measurement of aerosol properties such as particle size, shape, surface area, and surface chemistry.
Although engineered nanomaterials rely on physicochemical features smaller than approximately 100 nm, inhalation exposures will potentially occur to airborne particles composed of nanomaterials (including aggregates of nanoparticles) covering a size range from a few nanometers to several micrometers in diameter. These “nanostructured particles” are potentially of concern if they can deposit in the respiratory system and have nanostructure-influenced toxicity (e.g., high surface area, high surface activity, unusual morphology, small diameters, or disaggregation into smaller particles once deposited). Classes of particles of interest, therefore, cover discrete nanometer-diameter particles, agglomerates of nanoparticles, and droplets of nanomaterial solutions, suspensions, or slurries. Particles formed from the degradation or comminution of nanomaterials may also present a potential risk if they exhibit nanostructure-dependent biological activity.
Conventional exposure monitoring methods typically rely on characterizing the mass and bulk chemistry of airborne particles, two measures that do not seem particularly appropriate to nanostructured particles. There may be ways to adapt mass-based measurement approaches to evaluating exposures to nanostructured particles, but solutions will most likely represent a degree of compromise. Measurement of aerosol number concentration (i.e., number of particles/volume) is a relatively straightforward alternative. This method may be useful for identifying sources of nanoparticle emissions (Brouwer et al., 2004); however, the relevance and specificity of number concentration measurements in isolation are questionable. Alternatively, measuring the size distribution of airborne particles provides a wealth of information on particle size, surface area, mass, and number concentration, and can be used to differentiate between background and process-specific sources (Kuhlbusch et al., 2004; Maynard and Zimmer, 2002). However, current instruments are expensive, and measurement of size distributions from the nanometer range to the micrometer range is frequently a complex process (Zimmer and Maynard, 2002).
Among methods for measuring surface area, isothermal adsorption has been used for many decades to measure the surface-area of powders; however, samples must be collected and analyzed later, and applicability for low aerosol concentrations is unclear. Although time intensive, particle analysis in the Transmission Electron Microscope (TEM) is capable of providing detailed information on projected (two-dimensional) surface area. Differential Mobility Analysis can be used to estimate surface area with the added advantage of high sensitivity at small particle diameters (Rogak et al., 1993).
Measuring the rate at which positive ions attach to airborne particles provides a near real-time measure of surface area. The rate at which ions diffuse to the surface of neutral nanometer-diameter particles is proportional to particle surface area (Keller et al., 2001). Diffusion charging-based aerosol monitors rely on this relationship to estimate airborne particle surface area. Ku and Maynard (2005) reported that a commercial diffusion charger provides good estimates of aerosol surface area for particles smaller than 100 nm in diameter over a range of morphologies (Fig. 2), with increasing underestimation for larger particles, consistent with diffusion charging theory. Whether the degree of underestimation is significant for health-related measurements depends on a number of factors and is still under investigation. An alternative diffusion charging-based device has recently been tested that showed a different response to particles than the instruments tested by Ku and Maynard. Intriguingly, the response of this device correlates closely with aerosol surface-area dose that reaches the lungs, as opposed to total surface area of the particles (Wilson, 2004).
Along with continued research on the toxicity of nanoparticles, ensuring the appropriate physicochemical characterization of materials is essential for drawing valid conclusions. Currently available instruments and methods provide some insight into relevant material properties. A range of tools are available in diverse research fields that are relevant to health-related nanoparticle characterization, and multidisciplinary collaborations are encouraged to fully utilize these resources. Long-term effective characterization of nanostructured particle exposure will require further research and development of portable and cost-effective instruments.
EVALUATION OF THE DERMAL PENETRATION OF NANOMATERIALS
Skin can be exposed to solid nanoscale particles through either intentional or nonintentional means. Intentional dermal exposure to nanoscale materials may include the application of lotions or creams containing nanoscale TiO2 or ZnO as a sunscreen component or fibrous materials coated with nanoscale substances for water or stain repellent properties. Nonintentional exposure could involve dermal contact with anthropomorphic substances generated during nanomaterial manufacture or combustion (reviewed in Oberdörster et al., 2005).
It is unclear whether nanoparticles will penetrate the skin and have any toxicological impact. Concerns regarding dermal penetration include (1) skin or other organ cytotoxicity, (2) accumulation in skin or other organs resulting in toxicity after long-term exposure, (3) metabolism to even smaller particles with potentially increased toxicity, or (4) toxicity of photoactivated nanoparticles.
As an example of dermal contact with nanoparticles, nanoscale TiO2 and ZnO (<100 nm) are being included in sunscreens because of their ability to block ultraviolet (UV) light. TiO2 particles below approximately 200 microns do not scatter visible light but will still scatter some UVA radiation. Thus the inclusion of nanoscale TiO2 or ZnO in sunscreens has the consumer-desired goal of a clear sunscreen with UV-absorbing properties.
TiO2 can exist in several crystalline forms (e.g., anatase, rutile, amorphorous). Anatase TiO2 is a semiconductor, and absorption of a photon of light below approximately 385 nm (i.e., UVA or UVB) will result in the creation of an electron hole pair. The electron hole pair will interact with water or oxygen at the crystal surface, resulting in the generation of reactive oxygen species (ROS) including hydroxyl radicals, singlet oxygen, or superoxide (Konaka et al., 1999). ROS generated following illumination of nanoscale anatase TiO2 are bacteriocidal (Maness et al., 1999), oxidize water contaminants (Moraes et al., 2004), are genotoxic in Chinese hamster and mouse lymphoma cell lines (Nakagawa et al., 1997), and are cytotoxic to cultured HeLa cells (Cai et al., 1992), human fibroblasts (Wamer et al., 1997), Chinese hamster ovary cells (Uchino et al., 2002), and human colon carcinoma Ls-174-t cells (Zhang and Sun, 2004).
Coating the anatase TiO2 particle with inert oxides of silica, alumina, or zirconium reduces or eliminates the generation of ROS following UV irradiation (Mills and Le Hunte, 1997). As a result, the use of coated nanoscale TiO2 or ZnO in commercial products should eliminate the concern over the generation of ROS; however, Brezová et al. (2005) recently demonstrated that irradiation of commercial sunscreen products with light >300 nm resulted in the formation of ROS. What remains to be established regarding the use of TiO2 nanoparticles in commercial products is whether there are any biological consequences from the generation of ROS following UV irradiation.
A few studies have investigated skin penetration by nanoscale TiO2. A sunscreen containing 8% microfine TiO2 (10–50 nm) was applied for 2–6 weeks to the skin of human volunteers, and the penetration of TiO2 into the epidermis was evaluated using tape stripping (Tan et al., 1996). The depth of penetration was related to the number of tape strips required to remove the TiO2 from skin. TiO2 did not penetrate the skin of treated subjects (16 samples) when compared to TiO2 levels in cadaver skin (9 samples); however, removal of one high value in the cadaver group resulted in a statistically significant increase in TiO2 in the epidermis of treated subjects, providing the first evidence of skin penetration by nanoscale TiO2. The authors noted that a larger sample size would be necessary to establish this difference. Lansdown and Taylor (1997) demonstrated apparent penetration when microfine TiO2 was applied as a castor oil suspension to the backs of rabbits. Using light and electron microscopy, Schulz et al. (2002) and Pflücker et al. (2001) concluded that TiO2 nanoparticles did not penetrate the skin of human volunteers. Bennat and Müller-Goymann (2000) applied TiO2 to human skin either as an aqueous suspension or oil-in-water emulsion and evaluated skin penetration using the tape stripping method. They observed that TiO2 apparently penetrated skin when applied as an oil-in-water emulsion, and that penetration was greater when applied to hairy skin, suggesting surface penetration through hair follicles or pores. The observation that the hair follicles could be a “repository” of TiO2 was further established by Lademann et al. (1999), where TiO2 penetrated into the orifices of hair follicles. The penetration of submicron sized (0.75–6.0 μm) fluorescent spheres into hair follicles has also been reported (Toll et al., 2004).
Tinkle et al. (2003) provide compelling evidence that nanoparticles could penetrate the skin in demonstrating epidermal and dermal penetration by fluorescent microspheres (0.5–1.0 μm) in human skin in an in vitro flexed skin model.
To assess the dermal penetration and toxicological impact of nanoparticles, future studies will need to focus on (1) characterization of the materials being tested (e.g., monodispersed or aggregated, anatase versus rutile TiO2), (2) the analytical method used to detect the nanoparticle (i.e., needs to be able to discriminate between epidermis, dermis, and hair follicles), and (3) the condition of the skin (e.g., stratum corneum) before and after testing. Dermal application and intradermal injection with CdSe quantum dots may assist in further investigating dermal penetration and systemic absorption of nanoscale materials.
PULMONARY TOXICITY OF CARBON NANOTUBES AND IMPLICATIONS FOR RISK ASSESSMENT
Nanotubes hold great promise for many applications for aerospace and other industries, including miniature electronics such as efficient wires and semiconductors; space-efficient computer memory; light-weight, exceptionally strong structures with excellent heat dissipation; miniature electromechanical devices; and highly efficient air purification systems. In addition to new technologies, space missions may encounter nanoparticles in lunar or Martian dusts. Release of nanoparticles in an enclosed environment is a great concern for crew health and for hardware engineering.
For many nanotube manufacturing and use settings, the actual amount of material reaching the lungs will likely be negligible; however, under some conditions nanotubes can be extremely small and buoyant (Maynard et al., 2004) and, thus, penetrate deep into the lung. Research sponsored by NASA investigated the potential hazard associated with three types of single-walled carbon nanotubes (SWNT) to assess the response of the lower respiratory system (Lam et al., 2004). Findings of this research relevant for conducting risk assessments are summarized below.
Properties of SWNT that Affect Toxicity
Single-walled nanotubes are graphite sheets rolled into tubes on the order of 1 nm in diameter and 1000 nm or more in length. They may be capped at either end by half-fullerene domes and achieve great strength by the sp2 bonds between carbon atoms. Some nanotubes have a strong tendency to agglomerate by van der Waals forces into tattered ropes much larger than individual fibers; whereas, others remain as a fine powder, much like carbon black. Both intentional and unavoidable impurities can affect the electrical, chemical, and mechanical properties of the particles and, thereby, their toxicity. The toxicity of the particles will also depend on whether they are persistent or cleared from the lung and whether the host can mount an effective response to sequester or dispose of the particles.
Intratracheal Instillation of SWNT into Mice
Lam et al. (2004) used intratracheal instillation to minimize the amount of material required, given the expense or limited availability for some of the SWNTs, and because of known dosage to the lungs and ease of comparison to compounds of known toxicity. Disadvantages included an unnatural delivery that could exaggerate the effects on the lungs, missing any effects on the upper airways, anesthetization of animals, and vehicle effects on the delivery and properties of the particles.
Test particles were suspended (sonicated) in heat-inactivated mouse serum to deliver doses of 0.1 or 0.5 mg/mouse in a 50 μl volume. Three types of SWNT, along with carbon black and quartz (reference dusts of known toxicity) were tested. The lungs were removed from euthanized animals and subjected to histopathology study 7 or 90 days after instillation. The types of nanotubes tested varied according to the method of preparation (electric arc discharge or vapor deposition) and the content of metals. Raw SWNTs contained 27% Fe, 1% Mo, 0.8% Ni, and 0.4% Cu; purified SWNTs contained 2% Fe; and Carbolex SWNTs contained 26% Ni, 5% Y, and 0.5% Fe.
Response of the Lung to Instillation of SWNTs
The highest dose of Carbolex SWNTs caused death in 5/9 mice, most likely due to Ni release from the SWNTs during sonication, which would not likely occur by aerosol inhalation. No other lethality occurred, although the high-dose Carbolex group lost weight, and the raw (but not purified) SWNT group had transient signs of acute toxicity. Mice showed a dose-dependent incidence of focal lesions in their lungs 90 days after instillation. Microscopically, these lesions contained bundles of SWNTs within macrophages surrounded by fibrous tissue forming granulomas with some necrotic cells. In some areas the granulomas were discrete, while in other areas the fibrotic tissue was widespread and contiguous. Alveolar wall thickening occurred in areas without obvious granulomas. Multifocal granulomas were also observed in the lungs of rats that received instillations of SWNT containing 5% Ni and 5% Co (Warheit et al., 2004).
Lungs instilled with carbon black showed the presence of opaque particles even after 90 days, but no tissue reaction. The Fe-containing nanotubes elicited inflammation that was still present after 90 days, whereas the inflammatory response had subsided in the lungs instilled with Ni-containing Carbolex nanotubes. Lungs instilled with quartz showed inflammation, but no granuloma formation or fibrosis.
Nanotubes were thus at least as toxic as quartz and much more toxic than carbon black, with some indication of the effect of metal content on toxicity. The redox properties of iron in SWNT were implicated in oxidative stress and cytotoxicity in cell cultures of human keratinocytes (Shvedova et al., 2003). Nickel is also a redox-active metal with potential to influence toxicity.
Implications for Human Risk Assessment
These findings indicate that, if inhaled into the lungs, nanotubes (depending on their content) are capable of eliciting an inflammatory, granulomatous, and fibrogenic response, and that the mass-based permissible exposure level (PEL) for respirable graphite dust may be inadequately protective for exposure to SWNTs. If a 30-g mouse were exposed to airborne nanotubes at a concentration of 5 mg/m3, the PEL for respirable graphite dust, and 40% of the respired nanotubes deposited in the pulmonary region, the lungs would accumulate a mass of nanotubes equivalent to the low dose within 4 working days and a mass equivalent to the high dose within 17 working days. Moreover, because SWNTs were more toxic than quartz based on histopathology, assuming similar relative toxicity in humans, a PEL below that for quartz dust (0.05 mg/m3) is suggested until further characterization of nanotube toxicity, assessment of the fraction of airborne nanotubes in the respirable size range, and determination of whether a mass-based standard is even applicable given evidence for a surface-area based standard (Maynard and Zimmer, 2002).
Additional questions of relevance for human health assessment include under what conditions are the nanotubes respirable; how quickly are inhaled nanotubes cleared from the lungs and what is their fate; what are the long-term tissue responses to persisting nanotubes; and how do shape (e.g., varying tube lengths) and metal impurities such as Ni and Fe influence toxicity?
IMPACT OF EXPOSURES TO NANOPARTICLES ON RESPIRATORY HEALTH: PARTICLE SIZE MAY NOT BE MORE IMPORTANT THAN SURFACE CHARACTERISTICS
Several toxicology studies in rats have reported that ultrafine or nanoparticles cause increased lung toxicity compared to larger particles of similar chemical composition at equivalent mass concentrations (reviewed by Oberdörster et al., 2005). Surface area and particle number metrics are thus postulated to play important roles in ultrafine particle lung toxicity.
A few studies have assessed lung toxicity in rats following aerosol exposures to ultrafine particles at very high particle concentrations. The results of 2-year inhalation studies with ultrafine TiO2 (P-25) or fine-sized TiO2 particles (average primary particle sizes ∼25 and ∼300 nm, respectively) have indicated that less than one-tenth (10 vs. 250 mg/m3) of the inhaled exposure concentrations of the ultrafine particle-types, compared to fine-sized particle-types, produced equivalent numbers of lung tumors in rats (approximately 16–30%) (Heinrich et al., 1995; Lee et al., 1985). In addition, shorter-term pulmonary toxicity studies with ultrafine and fine carbon black, as well as TiO2 particles in rats (Bermudez et al., 2004; Oberdörster et al., 1994), have demonstrated increased pulmonary inflammatory potency of the ultrafine particles when compared to fine-sized particulates of respective similar composition.
However, the common perception that all nanoparticle-types are more toxic and inflammogenic than fine-sized particle-types of similar content is based upon a limited number of studies; in fact, a systematic comparison of only two particle-types (TiO2 particles and carbon black particles) and essentially only one type of TiO2: P25 (80% anatase, 20% rutile). A variety of physical characteristics, other than particle size, may play significant roles in modifying toxicity of nanoparticles (Yamamoto et al., 2003). These include, but are not limited to surface treatments of particles (Warheit et al., 2003b), the tendency of aerosolized particles to aggregate/disaggregate, the method of synthesis of particles (i.e., whether the particle was generated in the gas or liquid phase), particle shape, and/or surface charge.
In considering the influence of particle size and surface treatments on lung toxicity, three sets of pulmonary bioassay studies are currently being conducted and the preliminary findings are described below.
First, because fine-sized Min-U-Sil quartz particles are classified by the International Agency for Research on Cancer (IARC) as Category 1 (i.e., human carcinogen), one would expect that nanoscale quartz particles could be even more potent pulmonary toxicants than highly cytotoxic fine-sized quartz particles. Fine-sized quartz particles (Min-U-Sil, average diameter = 1.6 μm) or nanoscale quartz particles (mean particle size ∼50 nm) were intratracheally instilled into the lungs of rats at doses of 1 or 5 mg/kg. Following exposures, assessments of bronchoalveolar lavage fluid biomarkers of lung toxicity were made at 1 day, 1 week, and 1 and 3 months. Pulmonary exposures to the Min-U-Sil quartz particles produced significantly enhanced and sustained lung inflammatory responses compared to the effects produced by the nanoscale quartz particles. These studies were repeated with smaller nanoscale quartz (range = 10–20 nm) samples or fine-sized quartz sample (400 nm), with Min-U-Sil quartz particles as a positive reference quartz particle-type. Although the study has not been completed, the results thus far indicate that the order of inflammatory potency was nanoscale quartz = Min-U-Sil > fine-sized quartz particles (Fig. 3).
In a second set of studies, rats were exposed to fine-sized TiO2 particles (300 nm), TiO2 nanoscale rods (250 nm × 30 nm), or to TiO2 nanoscale dot particles (10 nm) at intratracheal instillation doses of 1 or 5 mg/kg, with lung assessments conducted at 1 day, 1 week, and 1 and 3 months postexposure. TiO2 dust is known have low toxicity and solubility. Preliminary results have thus far demonstrated no significant differences among any of the particle-exposed groups compared to vehicle controls with regard to inflammatory or cytotoxic lung responses at any postexposure time periods. Moreover, following a transient (24-h post exposure) neutrophil-associated inflammatory response, subsequent immunological responses were macrophage mediated for all three particle types. By contrast, Min-U-Sil quartz particles produced persistent inflammatory response (through 3 months postexposure) characterized by the presence of neutrophils, lymphocytes, and macrophages. Interim conclusions suggest that particle size was not a determinant of toxicity, in contrast to previous studies of fine-sized TiO2 and ultrafine P25 TiO2 (Bermudez et al., 2004; Heinrich et al., 1995; Lee et al., 1985). Thus, P25 may not be representative of all nanoscale TiO2 particle-types, and pulmonary toxicity for each fine-sized and nano-sized particle types should be evaluated on a case-by-case basis.
A third set of studies (Warheit et al., 2005) investigated the effects of various surface treatments (0–6% alumina [Al2O3] and/or 0–11% amorphous silica [SiO2]) on the toxicity of commercial TiO2 particle formulations. Pulmonary bioassay data from instillation exposures in rats to TiO2 particle-type formulations were compared to reference base TiO2 particle types, along with vehicle controls. The TiO2 particle formulations with the largest concentrations of both alumina and amorphous silica surface treatments produced mildly enhanced adverse pulmonary effects when compared to the reference base control particles. These results along with other studies (Derfus et al., 2004; Warheit et al., 2003a,b) indicate that surface treatments on particles (likely of more relevance for human exposure than reference particle-types) can modify (either up or down) particle toxicity.
Based on the available literature and these preliminary findings, a number of additional factors are likely to influence the pulmonary toxicity of nanoparticles, in addition to particle size and surface area. These include but are not limited to the following:
(1) Particle number and size distribution;
(2) Dose of particles to target tissue;
(3) Surface treatments on particles, particularly for engineered nanoparticulates;
(4) The degree to which engineered nanoparticles aggregate/agglomerate—which strongly influences particle deposition characteristics in the lung;
(5) Surface charges on particles;
(6) Particle shape and/or electrostatic attraction potential;
(7) Method of particle synthesis—i.e., whether formed by gas phase (fumed) or liquid phase (colloidal/precipitated) synthesis and post-synthetic modifications, which likely influence the aggregation behavior of particles.
FRAMEWORK FOR RISK ASSESSMENT
Recent studies on nanoscale substances, together with considerable related research on ultrafine particles (UFP) from air pollution (e.g., Englert, 2004; Oberdörster et al., 2005), metal fume (e.g., Kuschner et al., 1997a,b), and mineral fibers (e.g., Cugell and Kamp, 2004) provide an initial basis for evaluating the primary issues in a risk assessment framework of nanomaterials (Fig. 4).
Identification of hazards depends on the diverse characteristics of these particles discussed above. Hazards of novel structures will be less predictable than smaller-scale versions of previously studied substances (e.g., metal oxides such as TiO2 or ZnO). Encapsulation of the material either by surface coatings or within a matrix affects the reactivity and biological mobility of nanoparticles (Warheit et al., 2003a,b), but the durability of such encapsulation needs consideration.
Exposure assessment includes the entire life cycle of nanomaterials from synthesis to disposal. Exposure is likely highest for workers, although product specific evaluations should consider consumer uses, wear, disposal, and potential for environmental release and fate for products containing nanomaterials (Fig. 5). As noted above, several nanomaterials (e.g., carbon nanotubes [Maynard et al., 2004], metal oxides [Bennat and Muller-Goymann, 2000]) tend to aggregate, which may reduce their ability to penetrate membranes, reach deep airways, or disperse. However, aggregation by fullerenes in water has been associated with increased solubility and antibacterial properties (Fortner et al., 2005). Determining whether nanoparticle exposure would occur for specific products is a critical initial step in assessing potential risks.
Inhalation and dermal routes have been the primary focus for health effects research of nanoparticles, as described above. Research on medical device (e.g., Yamamoto et al., 2003), diagnostic, and therapeutic (e.g., Lockman et al., 2004) applications is also contributing to understanding the toxicokinetics and toxicodynamics of nanomaterials. Such studies indicate that skin (see above) and the blood–brain barrier do not appear to be readily penetrated, and penetration depends on specific conditions. As in the lung, phagocytic cells (macrophages, monocytes) of the reticuloendothelial system (RES) rapidly uptake nanoparticles after intravenous injection. Various nanoparticle coatings have thus been evaluated in drug delivery and imaging research to evade the RES and keep particles in circulation to reach targeted sites (e.g., tumors) (Jain et al., 2005).
Differences among compounds and studies are apparent in systemic absorption of particles from the lung. Nemmar et al. (2002) but not Brown et al. (2002) reported that inhaled nanocarbon particles in humans readily pass into systemic circulation, where they are a concern for cardiovascular toxicity. Iridium nanoparticles administered to rats by endotracheal tube were cleared by the airways to the gastrointestinal tract and eliminated, with less than 1% translocating to other organs (Kreyling et al., 2002). For systemically absorbed particles, the liver may be a primary translocation site with spleen, bone marrow, heart, kidney, bladder, and brain as secondary sites (Nemmar et al., 2001, 2002; Oberdorster et al., 2005; Takenaka et al., 2004). Uptake and translocation via olfactory and trigeminal nerve axons to the brain of rodents, as reported for airborne UFP (Oberdörster et al., 2005) and manganese, but not iron (Rao et al., 2003), could be a more direct route to the brain than across the blood–brain barrier (Oberdörster et al., 2005).
As noted above, the relevant inhalation dosimetry in risk assessments of nanoparticles may be surface area or particle number rather than mass per volume or per body weight, although the complexity of other properties preclude generalizations to all nanoparticles (described above; Yamamoto et al., 2003).
Despite this complexity, some patterns are emerging for the more studied substances. The primary mechanism of action by inhalation or dermal routes appears to be free radical generation and oxidative stress associated with surface reactivity (Oberdorster et al., 2005). Oxidative stress associated with TiO2 nanoparticles, for example, results in early inflammatory responses such as an increase in polymorphonuclear cells, impaired macrophage phagocytosis, and/or fibroproliferative changes in rodents (Bermudez et al., 2004).
An inverse relationship has been observed between lung clearance rate and nano P25 TiO2 toxicity in rodent species (i.e., order of increasing clearance and decreasing lung toxicity in a 90-day inhalation study: rats > mice > hamsters; Bermudez et al., 2004). Lung clearance of particles in humans is greater than in rats. Moreover, lung cancer in rats exposed to high levels of inert particles is thought to occur by a lung-overload mechanism that is not relevant for humans (Borm et al., 2004). On the other hand, translocation to other organs may be a greater concern for humans. Such species differences are a concern in extrapolating results from animal models to humans (Borm et al., 2004).
Intratracheal inhalation studies in rats indicate that ultrafine or nano TiO2 is less inflammatory than quartz (Rehn et al., 2003), nickel, or cobalt (Zhang et al., 1998), or carbon black (Renwick et al., 2004). Studies in humans exposed to ZnO metal fume (including nanoscale particles) have shown a relatively low order of toxicity associated with signs of pulmonary inflammation and metal fume fever that does not progress to pulmonary disease (ACGIH, 2001; Kuschner et al., 1997a). By contrast, greater toxicity is expected for more reactive nanoparticles such as fullerenes, carbon nanotubes, or perhaps UFP.
Although most toxicological studies with nanomaterials have been in vitro, or short-term in vivo studies involving unnatural delivery (e.g., intratracheal instillation) in limited species and types of nanoparticles, the National Toxicology Program is planning short and long-term studies, including oral, dermal, and inhalation exposures for some nanoparticles (http://ntp-server.niehs.nih.gov/files/nanoscale05.pdf). Nanomaterial research and risk assessments will ultimately need to address multiple potential health effects including cardiovascular, carcinogenicity, reproductive/developmental, immunological, and neurological.
Screening assessments of exposures to the more studied metal oxides could be conducted by developing toxicity benchmarks using the weight of evidence from studies of (1) nanoscale metal oxides in the toxicological and pharmacological literature, (2) fine-scale forms corrected for the proportionally greater surface area of nano-scale particles, (3) more toxic particles such as UFP, and (4) the toxicology and epidemiology of metal fume. Uncertainties in such assessments will have to be considered given data limitations; however, collectively, the available studies are beginning to reveal important features necessary for initial risk assessments of specific nanoparticles.
*Exponent, Bellevue, Washington 98007; †Woodrow Wilson International Center for Scholars, Washington, DC 20004–3027; ‡National Center for Toxicological Research and National Toxicology Program Center for Phototoxicology, U.S. Food & Drug Administration, Jefferson, Arkansas 72079; §National Aeronautics and Space Administration, Houston, Texas 77058; ¶DuPont Haskell Laboratory, Newark, Delaware 19714; ∥Environ International, Houston, Texas 77002