Inhaled particles and lung cancer, part B: Paradigms and risk assessment


  • Paul J.A. Borm,

    Corresponding author
    1. Particle Research, Institut für Umweltmedizinische Forschung, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany
    • Institut für umweltmedizinische Forschung, Heinrich-Heine-Universität Düsseldorf, Auf′m Hennekamp 50, D-40225 Düsseldorf, Germany
    Search for more papers by this author
    • Fax: +49-211-3389-331

  • Roel P.F. Schins,

    1. Particle Research, Institut für Umweltmedizinische Forschung, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany
    Search for more papers by this author
  • Catrin Albrecht

    1. Particle Research, Institut für Umweltmedizinische Forschung, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany
    Search for more papers by this author


Poorly soluble particles of low toxicity (PSP), such as CB, TiO2 and coal mine dust, have been demonstrated to cause lung cancer in rodents, being most pronounced in rats. Adequate epidemiologic studies do not clearly indicate increased lung cancer rates in humans exposed to such particles. This has caused controversial positions in regulatory decisions on PSP on different levels. The present review discusses the current paradigms in rodent particle carcinogenicity, i.e., (i) role of particle overload and of persistent inflammation and (ii) fibrosis as an intermediate step in particle-induced lung cancer with regard to human risk assessment. Fibrosis, which is usually considered a precursor of lung cancer in humans, was not related to lung tumors in an animal study using 6 different particles, each at 3 dosages. Lung tumors after both inhalation and intratracheal instillation of PSP are related to particle surface dose, which forwards hazard assessment at surface-based nonoverload concentrations and a standard setting using surface as an exposure metric. The scarce data available on humans do not support the overload concept but suggest a role for persistent lung inflammation. Differences in antioxidant protection between different rodent species correlate with susceptibility to PSP-induced carcinogenicity and support the need for detailed studies on antioxidant response in humans. Apart from such bridging studies, further focus is also needed on surface chemistry and modifications in relation to their adverse biologic effects. © 2004 Wiley-Liss, Inc.

Among the various existing types of particle, one should discriminate between those with inherent toxic activity, such as hard-metal dusts, welding fume and quartz dust, and those that due to their material properties have much lower acute toxicity. The latter are commonly referred to as PSP,1 also known as granular biodurable particles without known specific toxicity. PSP include DEP, CB, coal mine dust and TiO2 and are listed along with various other types of particles and fibers in Table I. Particles that have been identified and classified as human carcinogens are respirable crystalline silica (quartz, cristobalite), asbestos fibers and some hardwood dusts. Now also ambient PM (PM2.5, PM10) is suspected of carcinogenic potency,2, 3 but it is by no means clear which component or characteristics of PM are responsible for its proposed carcinogenic action. For asbestos as well as quartz the carcinogenic hazard is variable,4, 5 and this exemplifies the difficulties that regulators meet in classifying particles of a chemical entity (e.g., crystalline silica) as a lung carcinogen.

Table I. Particles Relevant for Occupational and/or Ambient Exposure, Their Potential Health Effects as Well as Some Applications and Exposure Situations
ParticleUse/exposureDurabilityCarcinogenic effect
Rat (inhalation)Human (IARC)
  1. Classification of carcinogenicity in animal studies was done as follows: +, positive in more than one animal during inhalation studies; −, negative or no inhalation studies; +/−, inadequate evidence in rats. In human studies, the IARC classification was taken as I, carcinogenic; IIA, probably carcinogenic; IIB, possibly carcinogenic; and III, not classifiable.

CB (15–500 nm)Pigments, toner, tires (20–400 nm)Insoluble, aggregates+IIB
TiO2, anatase or rutile (15–500 nm)Pigments, cosmetics, sunscreen agents (20 nm–5 μm)Insoluble, aggregates+ 
Iron oxides (FexOy)Pigments, paramagnetic diagnosticsInsoluble+/− 
Cement (CaCO3)Construction, buildingSolubleIII
DEPEngines, carsPartly soluble+ 
NiO, Ni-subsulfideExhaustInsoluble+I
TalcCosmetics, miningPartly soluble+ 
GraphiteAluminium productionInsoluble+/− 
Coal mine dustMiningPartly soluble+III
Crystalline silica (quartz, cristobalite)Quarrying constructionInsoluble+I
Amorphous silicaCleaning, paints, adsorbents, drugsReadily solubleIII
WooddustFurniture making, saw mills, art classesSoluble+/−I (some types of hardwood)
AsbestosInsulation, mining, shipyardsInsoluble+I
RockwoolInsulationLarge variation in durability+III
Refractory ceramic fibersInsulationHighly durable+IIA

Currently, the risk-assessment process for particles is complicated by several other factors. Animal carcinogenicity studies with PSP such as CB, TiO2 and coal mine dust, which were previously considered inert or “nuisance” dusts, show positive outcomes in rat models.6, 7, 8 In addition, among the PSP, uF particles or nanoparticles, usually defined as being <100 nm in diameter, have been identified as a special group. These uF particles can be an important component in inhalation exposure, especially considering their particle number and activity.9 Uf particles have often a large and active surface and are held responsible by toxicologists for a large number of acute and chronic biologic end points observed in human studies10, 11, 12 and experimental animal studies.13, 14 The risks of using nanoparticles for technologic applications has only recently become a subject of discussion.15, 16

One of the most important questions for regulating agencies is how to interpret the animal findings with regard to potential carcinogenicity and safe exposure levels for humans. In the first part of this review,17 we discussed the mechanisms considered relevant for the neoplastic actions of toxic particles as well as PSP in experimental animal studies and mainly in vitro studies. Briefly, both the intrinsic physicochemical properties of particles and the ability of particles to induce lung inflammation as well as cell and tissue proliferation have been shown to play a role in the genotoxicity, mutagenicity and carcinogenicity of particles. The first paradigm, which is based on rat inhalation studies, is that particle-induced inflammation drives both genotoxic events in airway epithelium as well as cell proliferation and tissue remodeling, which are required for mutations and progress toward neoplastic lesions. As such, particle load expressed in a metric that is most relevant to induce inflammation needs to be the main focus of concern when deriving safe exposure limits. Such a metric is of relevance with regard to uF particles as a high-surface analogue of fine particles of the same chemical composition or of particles that represent contrasting properties that should be treated as a different entity. A second paradigm is that fibrosis is an intermediate step in lung cancer in general18 and silicosis in particular,19 and on this basis, several risk-assessment committees have proposed or set exposure levels to prevent carcinogenicity for crystalline silica.20 In this second part of our review, we discuss both studies that support and those that question these paradigms, to put particle-induced lung cancer observations from both human and nonhuman studies into perspective in relation to their relevance for human risk assessment. Before doing so, some basic principles of particle toxicology are elaborated since they are a prerequisite to understanding further terminology.


AL, alveolar lumen; AM, alveolar macrophage; BET, Brunauer-Emmet-Teller equation; BR, bronchus; CB, carbon black particles; CINC, cytokine-induced neutrophil chemoattractant; COPD, chronic obstructive pulmonary disease; CWP, coal workers' pneumoconiosis; D, diesel particles; DEP, diesel exhaust particles; ECM, extracellular matrix; HPRT, hypoxanthine phosphoribosyltransferase; ILD, interstitial lung disease; iNOS, inducible nitric oxide synthase; OR, odds ratio; P, particle deposit; PM10, particulate matter with mean diameter of 10 μm; PMF, progressive massive fibrosis; PMN, polymorphanuclear cells; PSP, poorly soluble particles of low acute toxicity; RNS, reactive nitrogen species; ROS, reactive oxygen species; SKC, squamous keratinizing cyst; TiO2, titanium dioxide; TGF, transforming growth factor; TNF, tumor necrosis factor; uF, ultrafine.


Particle deposition, particle clearance and lung overload

Particle deposition.

Apart from their specific chemical surface reactivity, a particle load or burden in the lung can induce a range of toxicologic responses that differ principally from soluble or nonparticulate toxicants. For the interpretation of particle effects, a number of D's must be taken into account, i.e., dose, deposition, dimension, durability and defense (Table II). The dose at a specific site (in the lungs) determines the potential toxicity of particles. Obviously, this deposited dose is dependent on the inhaled concentration as well as the dimensions of the particle.21 Interestingly, the deposition probability of uF particles in the respiratory tract increases steeply the smaller the particles are.22 A major fraction of the particles will be deposited onto the fragile epithelial structures of the gas-exchange region. Most dosimetry models and calculations assume a uniform deposition at the bronchial airway surfaces and, therefore, a similar target dose for all epithelial cells within the respiratory tract.23 However, analysis of particles in human lung tissue24, 25 as well as mathematical modeling26 show that rather large particle deposition occurs at the bronchial airway bifurcations independent of particle size. Moreover, major differences in particle deposition can be found between lung lobes.27 In particular, cells located in the vicinity of the dividing spur may receive local doses that are a few hundred times higher than the average dose for the total airway. However, the patch size (about 100 cells) was small and the simulated enhancement factors were subject to considerable variation. Still, this finding lends further support to the observation that neoplastic lesions by particles and fibers predominantly originate at bronchial airway bifurcations.28

Table II. Processes and Particle Parameters as the D Words That Play a Major Role in Determining the Toxic Response Upon Inhalation
D wordRelevance for particle-induced effects
Dose1. Cumulative dose for chronic effects; can be based on particle or fiber mass, fiber or particle number or particle surface dose.
 2. Bulk composition is not equal to surface.
DimensionSize (diameter, length)
DepositionDependent on dimension but also on airway properties (hot spots)
DurabilityBiopersistence dependent on defense as well as particle properties (dissolution)
DefenseMucociliary clearance, macrophage clearance, inflammatory cells. If macrophage clearance is saturated, overload occurs; dose increases exponentially with time.

Particle clearance.

The lung has extensive defense systems, such as mucociliary clearance especially in the upper airways and macrophage clearance in the lower airways and alveoli. Particle transport by macrophages from the alveolar region toward the larynx is rather slow in humans, even under normal conditions, and thus eliminates only about one-third of the deposited particles in the peripheral lung. This implies that the other two-thirds accumulate in the lower lungs without significant clearance unless the particles are biodegradable and cleared by other mechanisms.29 Therefore, the same deposition of PSP with different durability can lead to a different cumulative dose. Some particles, such as cement, will readily dissolve in the aqueous epithelial lining fluid, whereas others, such as CB, are almost insoluble and need other clearance mechanisms for their removal (Table III). If a toxic particle is not soluble or not degradable in the lung, it has high durability and there will be rapid local accumulation upon sustained exposure. For PSP, no accumulation will occur if normal clearance mechanisms are not impaired.

Table III. Schematic Illustration of Particle-Induce Pulmonary Effects in Rats
  1. Adapted from Oberdörster.9

Exposure durationInflammatory cells↑ (AM, PMN)Inflammation
Dose increaseBiochemical lavage parameters ↑ 
 Alveolar epithelial damage 
 Lung weight ↑ 
 AM clearance function↓Particle kinetics
 Particle retention↑ 
 AM aggregation 
 Interstitialization of particles (LN) 
 Cell proliferation ↑Morphology
  • Type II cells, Clara cellsFunctional changes
  • Fibroblasts 
 Collagen deposition and degradation 
 Fibrotic lociChronic diseases
 Benign and malignant tumors 

Particle overload.

During an inhalation study, at a certain time point, a lung burden is reached that exceeds the macrophage clearance capacity and results in overload effects (typically, 1–3 mg or 200–300 cm2/rat lung).23, 30Lung overload is here defined as a “consequence of exposure that results in a retained lung burden of particles that is greater than the steady-state burden predicted from the deposition rates and clearance kinetics of particles inhaled during exposure”.1 Therefore, the hallmark of particle overload is impaired macrophage clearance function, associated with pulmonary inflammation, centracinar interstitial and interstitial accumulation of particles and epithelial cell proliferation.1 Although volumetric overloading of macrophages which starts at 6% of normal AM volume31 was originally used to develop the overload concept, the surface dose is today considered a better indicator, especially in explaining the effects of uF particles.9, 13 Indeed, after large doses of PSP, they are visible in lung sections many months or years after cessation of exposure (Fig. 1).

Figure 1.

Interstitial (a) and submucosal bronchial (b) macrophages loaded with DEP 129 weeks after intratracheal instillation of a cumulative dose of 15 mg DEP at 12 weeks of age. Arrows indicate areas that stain red with Sirius red to indicate that collagen deposition has occurred in the peribronchial as well as interstitial areas.

The overload concept has important implications for hazard assessment as well as for setting occupational standard values for particles when based on the outcomes of nonhuman studies.1, 32 Saturation of macrophage clearance is different from saturation of metabolic clearance, which is usually determined by Michaelis-Menten kinetics, where saturation directly occurs when the exposure concentration is near to the affinity constant (Km) of the enzyme, meaning that instantaneous saturation and clearance at maximal level (Vmax) occurs. At particle overload, macrophage clearance function is impaired, particle accumulation starts and inflammatory cell influx increases sharply.1, 32 The concept of particle overload specifically applies to PSP. Other more toxic particles, such as crystalline silica, synthetic fibers and toxic metal particles, affect AM-mediated clearance as well but at much lower lung burdens since they can actively damage AM. In addition, uF particles impair AM phagocytosis to a much greater extent than their fine counterparts when evaluated on an equal-mass basis.13, 33 Thus, every impairment of AM-mediated particle clearance should not be viewed as particle overload. In addition to effects on particle clearance, particles or particle components may have an effect on particle or microorganism phagocytosis, and this effect might contribute to the lowering of immune surveillance, possibly by enhancing neoplastic transformation upon challenge with other carcinogens. In Table III, the sequence of events suggested to occur in response to particle inhalation is summarized.

Although the main determinants of particle effects as depicted in Table II appear simple, most are interrelated. For instance, dimension, as in the case of fibers, can have profound effects on defense and thereby on cumulative dose. Long (>20 μm) fibers are not taken up by AM and, therefore, have a longer half-life in the lung compared to shorter fibers of the same material; consequently, they have a higher cumulative dose at similar inhaled fiber number or mass. Inhaled fibers >20 μm also show interception at bronchiolar bifurcations due to their length and may achieve high local doses by this process.28 For long fibers, which can be cleared only with difficulty by the immune system, the durability (or the biopersistence) appears to be the main determinant of the carcinogenic outcome.34 Fiber biopersistence in vivo and fiber dissolution in vitro are now used as screening methods in the development of new synthetic fibers to select out potential durable and pathogenic products.34 The present review, however, will specifically focus on particle carcinogenicity, and examples of fiber carcinogenicity will be used only if necessary.


uF particles (<100 nm) have posed new problems for researchers since they are small, hard to detect by microscopic techniques and much more inflammogenic than their fine analogues on an equal-mass basis.13, 14 Moreover, they can migrate to body compartments which are remote from their application or deposition sites. In particular, because of their low recognition and uptake by macrophages in the lung,33, 35 uF particles may have increased access to other cells in the epithelium, the interstitium as well as the vascular walls.36, 37 In addition, uF particles may be transported toward extrapulmonary organs via axonal transport, including trans-synaptic transport.38 Such a mechanism was first reported for 0.03 μm polio virus in monkeys and later described for nasally deposited colloidal 0.05 μm gold particles moving into the olfactory bulb of squirrel monkeys.39 Carbonaceous uF particles may translocate along the same pathway to the CNS, based on their presence in the olfactory bulb of rats after inhalation.38 Wherever they deposit or translocate, uF particles have properties such as a large surface area that can carry and absorb many endogenous substances (proteins, enzymes). Furthermore, these surfaces are sometimes chemically very active and therefore expected to react with numerous molecules, such as antioxidants, proteins or nucleic acids. This is schematically illustrated in Figure 2.

Figure 2.

Properties of uF particles that are suggested to be relevant for their potential biologic effects. Due to their large surface area, uF particles can absorb and bind numerous endogenous components nonspecifically. Also, due to their small size, many uF particles have a reactive surface that is considered to react with and inactivate many important mediators and constituents.

From inhalation studies, it has become clear that uF particles can induce more inflammation at considerably lower gravimetric lung burdens than their larger analogues.6, 13, 40 Actually, the retained particle surface has been used to describe inflammation in (sub)chronic inhalation studies.23, 41 Although a plausible concept as it is the surface of a particle that interacts with cellular structures, it is probably also an oversimplification for several reasons. Firstly, different uF particles at similar surface areas exhibited significant differences in inflammatory activity.42, 43, 44 Secondly, it is unclear whether uF particles have a different lung distribution between alveolar spaces, macrophages and interstitium45, 46 and what is their relevance for tumor formation. In relation to this, we found that uF TiO2 at similar chronic lung inflammation produced much more lung tumors than fine TiO2.47 Although observed at end point only, this observation suggests that uF TiO2, apart from causing persistent inflammation through impairment of AM clearance, has additional effects that contribute to lung tumor formation. Indeed, as we reviewed previously, a number of PSP cause genotoxic effects in vitro and in vivo in lung target cells.17 Whereas the in vivo effects can be ascribed to effects of ROS/RNS as released during particle-elicited inflammation, the in vitro effects are due to direct interaction with cells at concentrations not usually achieved during in vivo protocols.

We compared the tumor incidence of several uF particles (CB, TiO2) and their fine analogues at different doses as well as other durable (diesel DEP) and soluble (amorphous silica) particles. Particles were instilled intratracheally in doses up to 60 mg for uF particles and up to 120 mg for fine particles. Under these conditions, a very high initial dose rate (mg/day) was achieved, which according to other studies13, 23, 28 leads to impairment of lung clearance. Comparison of lung tumors at similar gravimetric dose (30 mg) of several uF particles (Table IV) shows that the amount of lung tumors induced by 3 different insoluble uF particles is proportional to their surface area. DEP, with the lowest surface area (34 m2/g), induced 22 tumors in 46 animals, while CB (300 m2/g) induced 40 tumors in 45 animals. The high surface amorphous silica induced few lung neoplasms, which might be due to its high solubility, i.e., low durability in vivo. However, caution must be taken since, apart from their small size, uF particles for commercial applications often have different chemical surface areas, obtained by surface modification. Evidence for the relevance of the chemical surface in PSP has also come from studies with surface-modified uF TiO2. Surface modifications resulting in enhanced hydrophobic uF TiO2 have generally led to amelioration of the inflammatory response,13, 48, 49 though initial studies showed increased toxicity of surface-modified uF TiO2.50

Table IV. Tumor Incidence and Types in Rats Treated Intratracheally with Various Doses of PSP and Killed 125–129 Weeks After Treatment
TreatmentDose (mg)Rats at riskSquamous tumorsBronchioloalveolar tumors
  1. Tumors are discriminated into squamous cell tumors and those originating from alveolar type II cells, referred to as bronchoalveolar tumors. Female Wistar rats (190 g) were treated with multiple intratracheal instillations (<6 mg) to reach the final cumulative dose at weekly intervals. Between 125 and 129 weeks of age, animals were killed, the lungs embedded and 3 lung lobes cut into sections for histopathology. NKCT, nonkeratinizing (noncystic) tumor; SCC, squamous cell carcinoma; BAA, bronchioloalveolar adenoma; BAC, bronchiolo-alveolar carcinoma. Statistical analysis showed significant differences in tumor types between the different particles (χ2 test, p < 0.001).

Fine CB(95 nm, 20 m2/g)30479171615
uF CB(14 nm, 300 m2/g)1527120361
DEP(34 m2/g)7.54810000
Fine TiO2(250 nm, 9.1 m2/g)604600660
uF TiO2(30 nm, 50 m2/g)15421011377
Amorphous silica(14 nm, 200 m2/g)153703000
 Series 1 4500000
 Series 2 4600000
Total  1278115158116


The ability of PSP such as CB and TiO2 to induce chronic inflammation, fibrosis, neoplastic lesions and lung tumors in rats has been well established.1, 6 The pulmonary carcinogenicity of CB has been demonstrated in 2 chronic inhalation studies.7, 8 Lung tumors were also noted for TiO2 in chronic inhalation studies at comparable exposures.51, 52 Tumors associated with experimental exposure to PSP are generally of 2 types: those originating from alveolar type II cells, which are called bronchoalveolar tumors, and squamous or epidermoid tumors, which are associated with bronchiolarization and considered to arise from areas with squamous metaplasia.53, 54 In general, the tumors include bronchioloalveolar adenomas, nonkeratinizing and keratinizing squamous cell tumors, bronchioloalveolar carcinomas, squamous cell carcinomas and adenosquamous carcinomas.55, 56

Previous findings of chronic inhalation studies have forwarded the concept that the retained particle dose as surface area (m2/lung), but not mass, is strongly correlated to the rat lung tumor response.13, 52, 53 This surface concept was reinforced by findings from Driscoll,41 when comparing rat lung tumor response as observed in a number of particle inhalation studies with the mass and surface area of retained particles (Fig. 3a). Only the surface area showed a highly significant correlation with tumor response, and the shape of the curve was consistent with the existence of a threshold. Interestingly, a similar curve was obtained (Fig. 3b) when the tumor responses after intratracheal instillation, as listed in Table IV, were converted into positive lung tumor responses per rat and related to instilled dose as surface area. This shows that all PSP fit on a similar line, suggesting that, in agreement with inhalation and instillation studies, the instilled surface area is the driving mechanism in causing lung tumors. Typically, both inhalation and instillation curves indicate a threshold (lung cancer prevalence not different from controls) between 0.2 and 0.3 m2 particle surface/lung. For instillation studies, this threshold almost coincides with the 7.5 mg dose DEP (BET 34 m2/g) and suggests that a similar gravimetric load would cause lung cancers with PSP in inhalation studies. Interestingly, the surface dose to cause lung cancers (0.2–0.3 m2) is about 10-fold higher than the surface load needed to cause neutrophilic inflammation by PSP (200–300 cm2).23, 31 This difference may have both a kinetic (particle clearance completely impaired) and a mechanistic (presence of particles leads to direct effects on tissue remodeling) background. Interestingly, our data set also suggests that when using surface as a metric, uF and fine analogues show similar carcinogenic potency, in contrast to their inflammatory effects, as discussed previously.

Figure 3.

Association between tumor response and particle surface area for various PSP gathered from different rat studies. (a) Results based on inhalation studies of Driscoll.41 (b) Results based on a study using a set of fine and uF particles in different doses by intratracheal instillation in rats and histopathologic tumor score (Table IV) after 129 weeks. Closed and open circles represent lung tumor prevalence (>1/animal) with and without SKCs as a relevant tumor for human risk assessment, respectively. In both cases, a similar curve (straight line) is obtained, with a threshold of 0.2–0.3 m2 surface dose/rat lung. Exclusion of SKC did not affect the outcome of instillation studies.

Using volumetric load as a dose metric, as shown in Figure 3, the effects for uF and fine particles fit with different curves, whereas at high volumetric dose the effects are similar, uF particles being much more effective at low volumetric burden at causing lung tumors (Fig. 4a). Although overload was originally defined as a volumetric phenomenon,30 later studies have indicated that it is the large surface area per unit of mass that renders particles more reactive with biologic tissues and that most biologic responses, including inflammation, are more closely related to surface area.23, 41, 45, 49 Several other pieces of information suggest that the “volume curve” (Fig. 4a,b) is an epiphenomenon. For example, when comparing the particles used in our study, little differences were seen in tumor types induced by fine particles vs. their uF counterparts. Furthermore, uF particles have been shown to impair macrophage phagocytosis of fine particles33 and to cause macrophage aggregation.57 Both effects would result in retardation of clearance independent of the uptake of particles and the resulting volumetric load, which was originally supposed to be the cause of macrophage impairment. Apart from the above considerations, several other mechanisms may play a role. UF particles are highly inflammogenic when compared at equal mass to their fine counterparts, and there is evidence that they can produce ROS directly from their surface.43 As discussed in part A of this review,17 this ROS formation may, besides inflammation, also play a role in genotoxicity and tissue proliferation. Indeed, in vitro studies have demonstrated genotoxicity for various uF PSP.17 In addition, uF particles interfere with histone acetylation,58 which possibly facilitates DNA damage, to induce mitochondrial damage59 and to affect influx of Ca2+ ions60 in various target cells.

Figure 4.

Link between tumor response and administered particle volume for both uF and fine particles. (a) Tumor response for fine (open symbols) and uF (closed symbols) particles and based on the same data depicted in Figure 2b, this time indicating specific subgroups of particles. (b) Same data but aggregated for fine and uF particles. Closed and open circles represent lung tumor prevalence (>1/animal) with and without SKCs as a nonrelevant tumor for human risk assessment, respectively.

Although the biologic plausibility for differential regulation of uF particles with regard to risk assessment is obvious, the meaning of our findings from instillation studies, as depicted in Figure 3b, is limited for several reasons. Route of administration, total dose and dose rates are much higher than in inhalation studies and without doubt have led to immediate impairment of clearance in rats treated with fine and uF particles. From the volumetric load (>5 μl), as applied in the instillation studies (Fig. 4a,b), it can be deduced that the fractional clearance rate in the lung has been minimal during the entire follow-up and that overload and associated events, such as inflammation and possibly direct genotoxicity, could have occurred immediately following instillation and probably for larger parts of the study interval.


A mechanism that is largely understudied by experimental approaches is the relation between lung fibrosis and lung cancer. Clinical evidence suggests increased incidence of cancer in patients with ILD, including sarcoidosis, tuberculosis and asbestosis.18 However, how ILD predisposes to subsequent lung cancer development remains unknown.61 In epidemiologic studies on the relation between particle exposure and lung cancer, fibrosis is often considered a marker of particle exposure since ILO-defined chest radiograph scores usually show a linear relationship with cumulative exposure or years of exposure.62, 63 The association between fibrosis and lung cancer has been reported in a number of epidemiologic studies on lung cancer in coal miners64 and silica-exposed workers.65 In a retrospective cohort study of 14,000 miners, Morfeld et al.66 found no link between quartz exposure and lung cancer but did find a strong association between CWP and lung cancer. Based on this outcome, parallel pathways for fibrosis and lung cancer were suggested, which may use the same genes. Although the biologic mechanisms leading to both pathologic outcomes are considered to be at least partially similar, the kinetics of both pathologies are clearly different in humans and experimental animals. Where particle-induced fibrosis at inhalation of quartz is already visible at 28 days in rats during inhalation of 5–7 mg/m3 or upon intratracheal instillation of 1–5 mg of quartz,67 a tumor response is visible only 1 or 2 years after chronic inhalation or single instillation.55, 56

Lung fibrosis is the net result (collagen deposition) of increased fibroblast proliferation and ECM remodeling, which is switched on and maintained by cytokines such as TNF-α63, 68 and TGF-β,69 as demonstrated using appropriate knockout mouse models or specific antibodies. Both TNF-α and TGF-β are highly implicated in tumor biology, TNF-α via multiple mechanisms and TGF-β via its role in liver regulation and lung cell growth.70 One can readily assume that abnormalities in TNF-α and TGF-β expression, as they affect progression of fibrosis,71, 72 are also related to abnormalities in formation or repression of lung tumor growth by particles. Previously, we evaluated the concomitant presence of lung fibrosis and neoplastic lesions in rats chronically exposed to different fine and uF particles by intratracheal instillation at different doses. Interestingly, all particles showed the ability to induce deposition of peribronchial and interstitial fibrosis, including the soluble amorphous silica. A linear dose response was obtained when relating particle mass to severity index of fibrosis (Fig. 5). The subjective severity score was confirmed by objective morphometric assignment of areas with collagen deposition.73 In contrast to lung tumors, where surface area is the best metric (Fig. 3a,b), the best relation for fibrosis is obtained with particle mass. Moreover, at similar mass dose, no differences were noted in the extent of chronic fibrosis between fine and uF particles (Table V).

Figure 5.

Association between particle-induced fibrosis and instilled particle mass. Lung fibrosis was scored on a severity index scale between 0 and 4 based on hematoxylin and eosin–stained sections. Animals were treated with different mass doses of fine and uF particles, and fibrosis was scored after death at 125–129 weeks. Curve shows the relation between extent of fibrosis as mean and SE of 46–48 animals/data point. (b) A significant correlation was obtained, with a logarithmic curve explaining 66% of all variance (r2 = 0.66). Similar curves with volumetric and surface dose only explained 36% and 27%, respectively.

Table V. Fibrosis and Collagen Deposition in Lungs of Female Wistar Rats 125–129 Weeks After Intra-Tracheal Instillation of Several PSP at Equal Gravimetric Dose (30 mg)
Particle treatmentFibrosis (SI)1Interstitial collagen (mm2/lung)2Peribronchial collagen (mm2/lung)2
  • 1

    Fibrosis was scored by severity index (SI) in hematoxylin and eosin–stained lung sections (5/animal) of paraffin-embedded lungs (data are means ± SE of 48 animals/group) in gradations: 0, no fibrosis; 1, fibrosis (>0).

  • 2

    Collagen deposition was measured in a subgroup of 10 animals/group. Lung sections were stained by Sirius red, and interstitial and peribronchial fibrosis was measured by morphometric analysis to determine particle, lumen and collagen surface. Data are means ± SEM of 10 animals/group.

Amorphous silica1.21 (0.13)3,063 (243)3,556 (262)
DEP1.40 (0.13)6,282 (356)1,089 (281)
Fine CB1.38 (0.15)6,766 (651)6,791 (830)
uF CB1.40 (0.11)6,366 (413)6,141 (370)
Fine TiO21.44 (0.14)NDND
uF TiO21.21 (0.15)6,775 (330)7,585 (425)

To relate the presence of lung fibrosis to lung neoplastic changes, we correlated the presence of lung fibrosis (severity index > 0) with the presence of lung tumors on the single-animal level. The resulting ORs, which describe the association between fibrosis and lung cancer for each particle type, are shown in Table VI. This calculation shows, for DEP and fine TiO2, a clear positive association between fibrosis and (multiple) tumors. In contrast, for uF TiO2, no positive association was seen between fibrosis and lung tumors. Although both types of CB (high and low surface) showed a slight positive association, the ORs (fine, 1.18; uF, 2.60) did not reach statistical significance. A change in cut-off criteria for the presence of lung fibrosis (severity index > 1 or 2) did not change these conclusions, but the ORs merely decreased and the confidence intervals became wider due to a smaller number of cases.

Table VI. Relation Between Tumor Incidence and Fibrosis (Severity Index >0) Calculated as ORs and Confidence Intervals from Animal Experiments with Lifetime Follow-up After Intratracheal Instillation of Several Low-Toxicity Fine and uF Particles
Particle treatmentAnimals at riskFibrosis and tumorsFibrosis and multiple tumors
  1. Tumor incidence was determined in rats killed 125–129 weeks after treatment and treated intratracheally with various high doses (range 7.5–120 mg) of poorly soluble, nontoxic particles. Tumors were discriminated into squamous cell and those originating from alveolar type II cells, called bronchioloalveolar tumors. Groups of different doses were pooled to obtain statistical power (confidence only given when lower limit is >1, which indicates statistical significance). Fibrosis was scored on the same hematoxylin and eosin–stained sections as the tumors, and the OR was derived from the concomitant presence of fibrosis, defined as severity index ≥1 and the presence of one or more tumors/animal. NS, not significant.

Fine CB1421.18 (NS)1.92 (NS)
uF CB1232.60 (NS)1.22 (NS)
Fine TiO29410.6 (1.31–85.4)1.35 (1.19–351)
uF TiO21341.17 (NS)0.89 (NS)
DEP1403.40 (1.10–10.5)3.03 (NS)

Therefore, in contrast to human clinical data,18 with regard to particles in the rat overload model, there appeared to be no general relation between the presence of fibrosis and lung tumors. However, a drawback of our data is that we were able to look at both outcomes only at a single chronic end point. It is recommended that future studies measure the onset and extent of collagen deposition at earlier time points by noninvasive methods (3–6 months) and correlate this to later development of neoplastic changes in the lungs of the same animal.


Based on the available data, our current knowledge (reviews1, 17, 41, 74) of particle-induced lung tumors in experimental animals can be summarized by stating that all inhaled particles, fibrous and nonfibrous, are likely to induce lung tumors in rats, provided that these particles are (i) inhaled chronically or instilled intratracheally at sufficiently high dose, (ii) respirable to the rat and (iii) highly durable. As we have discussed in this review, the retained lung burden leading to lung tumors can differ for different particles and, apart from dose, greatly depends on particle properties such as surface area and chemistry, cytotoxicity and size/dimensions (Table II). The gravimetric dose needed for the onset of particle overload and risk for subsequent neoplastic events is 1 mg/g lung tissue or 200–300 cm2 surface burden of PSP.23, 28 The surface dose where lung tumors start to develop after both inhalation and instillation of PSP lies between 0.2 and 0.3 m2 per lung, which conforms to 7–8 mg of PSP of average surface area (20–40 m2/g). Although some of the particles listed in Table I have been characterized as confirmed human carcinogens (group I by IARC), PSP such as coal mine dust,66 pigmentary TiO275 (Boffetta et al., unpublished) and CB were not associated with an increase in lung cancer in exposed workers. Thus, hazard assessment using rat studies raises the question of whether particles that induce tumors in this bioassay should be labeled as possible or even probable human carcinogens. In the following paragraphs, this aspect will be further outlined.

Is the rat paradigm valid for other animal species?

It is now generally accepted that the continued presence of nontoxic particle material in the lungs, upon impairment of AM clearance, leads to a chronic inflammatory response, fibrosis and tumorigenesis in the rat.17 As discussed earlier in this review, the overall pattern is one of chronic inflammation, which occurs upon saturation of lung clearance by overloading of macrophages. At this point, particle accumulation starts and inflammatory cell influx increases sharply. The influx of neutrophils and associated DNA damage and proliferation17, 76 are responsible for the mutagenicity,77 and the lung tumors after chronic particle exposure to PSP are due to their mutagenesis. This concept is illustrated in Table III. However, several studies have generated data that deviate from this paradigm.

Firstly, a number of studies question the validity of the inflammation paradigm. In a rat study, mutations in lung epithelial cells were determined after exposure to crystalline and amorphous silica, at exposure levels titrated to result in similar levels of neutrophilic inflammation.78 No increased mutagenicity, as determined by HPRT assay, was found in the rats exposed to amorphous silica compared to controls. However, crystalline silica (lower dose) exposure did induce clear HPRT-based mutagenicity, indicating a role for particle-specific properties.78 Similarly, depletion of circulating neutrophils in rats by injection of antineutrophil serum before short-term inhalation of quartz particles (3 days, 100 mg/m3) did not affect acute lung damage by quartz.79 It remains to be determined whether in a similar model the chronic lung damage and mutagenesis can be attenuated. Interestingly, studies using a similar model (infusion with anti-CINC antibodies) noted a reduction of oxidative DNA damage after exposure to hyperoxia, in support of a role for neutrophilic inflammation in DNA damage in the lung.80 However, it remains to be investigated whether the findings in these models using toxic quartz or hyperoxia can be reproduced with PSP.

Secondly, a number of studies have documented large contrasts in the response of different animal species to PSP. Two large studies have compared lung tumor prevalence in hamsters, mice and rats in response to chronic inhalation of particles. In a chronic inhalation study with filtered and unfiltered diesel exhaust (19 hr/day, 5 days/week), hamsters showed no lung tumors and mice showed an increase only in bronchioloalveolar carcinomas upon exposure to either filtered or unfiltered exhaust; rats showed a clear carcinogenic response to unfiltered diesel exhaust.7, 8 A large inhalation study was performed with hamsters, mice and rats, using both fine and uF TiO2.81 Animals were exposed to 10, 50 or 250 mg/m3 fine TiO2 or to 0.5, 2 or 10 mg/m3 uF TiO2 for 13 weeks and studied at the end of exposure as well as after recovery periods of 1, 3, 6 and 12 months, respectively. Lung burden was much lower in hamsters, especially at concentrations that caused overload in rats and mice (50 mg/m3 for fine TiO2, 10 mg/m3 for uF TiO2). Despite similar lung burdens in both rodent species, the inflammatory and histologic responses were less severe in mice than in rats and appeared to diminish with time. The picture that emerges from the above studies is that there is a marked difference between rats and other rodents in either clearance or tissue response to PSP.

As a potential explanation for these differences, the antioxidant defense capacities in rats and hamsters exposed to quartz were compared.82 Animals were instilled with different doses of quartz and uF CB, and production of ROS by BAL cells as well as expression of catalase and metallothionein in whole lung were measured. Hamsters linked lower oxidant production to larger upregulation of antioxidants compared to rats, which was consistent with a lower tissue injury and proliferation in hamsters compared to rats. However, these findings should be interpreted cautiously since earlier it was shown that enhanced mRNA expression of antioxidant enzymes by particles and fibers does not always lead to enhanced enzyme content and protecting activity.83 Nevertheless, increased understanding of species differences in antioxidant capacity may provide useful information for hazard comparison as well as appropriate risk assessment. Due to both their basal levels and their ability to upregulate or modify their levels or activities, antioxidants are likely determinants of responsiveness to particle effects. Apparently, hamsters have a far more robust antioxidant defense than rats. At present, we are just beginning to understand how antioxidant response can be a crucial effect modifier between particle exposure and susceptibility to tumor formation. However, increasing evidence is emerging that the rat is a particularly sensitive species to oxidative injury, whereas the hamster has been suggested to be more like humans in its antioxidant defense and oxidant generation.82 Also, investigators in infection models now consider the hamster more like humans because the induction of iNOS activity in the hamster was absent in Leishmania infection.84

The rat paradigm: valid for humans?

Little is known about the validity of the mechanism described in the previous paragraph and illustrated in Table III for humans. Questions relevant to this issue follow.

  • 1Is exposure to PSP associated with lung cancer in humans?
  • 2Are there similarities or differences between human and rodent lung cancers?
  • 3Is there evidence for overload in humans?
  • 4Do inflammation, hyperplasia and fibrosis play a role in human lung cancer?

A summary of the answers to these questions is given in Table VII, while more detail is given in the text below. Exposure to a number of particle types is associated with lung cancer in humans, but most of the particles in Table I are not PSP but highly toxic (quartz, asbestos) or highly complex, such as PM, where the role of the particle component remains to be proven. Current epidemiologic studies on workers exposed to fine TiO2 or CB do not reveal increased lung cancer incidence, though this may be related to the lower exposures in current industrial practice than typically occurred in the decades where silica and asbestos studies were performed.

Table VII. Comparison of Human and Rat Lung Responses Playing a Role in Lung Tumor Formation by PSP
EvidenceInadequate evidence in occupational epidemiologic studiesNumerous inhalation and instillation studies
Tumor typeNo SKC, rarely adenomasNo small or large large cell anaplastic tumors
Particle distribution and overloadNo “rodent-type overload” in ex-coal minersStarts at 200–300 cm2/rat lung or 1 mg/g lung
Inflammation (% PMN)Acute inflammation with PM (low-dose), chronic inflammation in occupationally exposed workersInflammation driven by particle dimensions, toxicity and surface dose
Antioxidant responseAntioxidant enzymes affected in particle-induced lung diseases, no data available to compare to rodentsUpregulation but lower than in hamsters and mice

Therefore, the first question could be answered “no” since there is no clear-cut epidemiologic evidence that PSP cause lung cancer in humans. We also know that the pathogenesis of lung cancer in humans and rodents is a multistep process that involves sequential accumulation of molecular alterations in tumor-suppressor genes (e.g., p53), oncogenes (e.g., myc, K-ras, H-ras, HER2/neu) cell cycle control and growth factor regulation. Many of these genes have been found altered in both human and rat lung tumors,85 though distinct differences in, e.g., mutation pattern and frequencies are seen between the soluble carcinogen benzo(a)pyrene and fibrous particles, suggesting different pathways in carcinogenesis.86 There are several differences between tumor cell types in rodent and human lung cancers. The most prevalent type of tumor in animal studies is the bronchioloalveolar adenoma, followed by keratinizing squamous cell tumor, adenosquamous carcinoma and squamous cell carcinoma (Table IV). The majority of these induced lung cancers, including adenomas and squamous cell carcinomas, are also seen in humans. However, small cell anaplastic tumors common in humans are unknown in rats. Reversibly, SKCs, which are very common in particle-induced lung tumors in rats (Table IV), are unknown in humans. To evaluate whether this is relevant to risk assessment for humans, we excluded SKC from the dose–response analysis shown in Figures 3b and 4b. This analysis showed that the relation between surface dose (m2/lung) and lung tumor presence remains strong, with the exception of uF CB. For some reason, instillation of this particle caused a relatively large percentage of SKC compared to other treatments. When SKCs are excluded, the relation between volumetric dose and lung tumors changes considerably (Fig. 4b), a more diffuse pattern appears and the clear distinction between fine and uF particles (Fig. 4a) disappears.

The answers to questions 3 (overload in humans) and 4 (inflammation and cancer in humans) are derived from the best-studied group with regard to occupational exposure to PSP, i.e., coal miners. Interestingly, coal miners show all the nonmalignant pulmonary outcomes of particle exposure, including lung function decrease, emphysema and pneumoconiosis.85 Clinical studies in coal miners have shown that they have a number of characteristics (Table III), including activated macrophages that release more ROS and inflammatory mediators, which is typically related to the stage or progression of disease (review87). However, some end- points, such as PMF, which are related an imbalance in cytokine release88 are not seen in rodents.86 Also, increased oxidative DNA damage in peripheral blood was detected in coal miners,89 though coal mine dust is not considered to cause lung cancer in humans90 and the observed effects are considered to be merely a marker of systemic oxidative stress. In coal miners also, the issue of particle overload has been studied using pharmacokinetic modeling of lung burdens in relation to historic exposure. Although high particulate lung burdens (up to 40 mg/g lung) have been measured, 2 studies concluded that clearance in coal miners differs from that in rats and that there is no evidence for a “rodent-type overload” in the human lung.91, 92 Unfortunately, no data are available on the surface area of these coal dusts, and it might be that this high-mass dose translates into a low dose when expressed as surface area.

Few studies are available on the inflammatory response in the human lung during particle exposure, though studies with PM showed a clear acute inflammatory response after bronchosegmental instillation of 80 μg of PM10 extracts.93 In addition, nasal instillation of PM10 suspensions and extracts has shown considerable acute inflammation within the upper respiratory tract (review94). Inflammation in the lung during or after particle exposure was again demonstrated in coal miners by showing that PMN counts in bronchoalveolar lavage of 20 ex-coal miners were related to the amount of quartz that remained in the lungs.95 Since cumulative coal dust exposure did not significantly add to this relationship, this effect is not ascribed to PSP but to toxic quartz. In addition, the relationship was very much driven by 2 individuals with high (>0.5 g) quartz lung burden.95 Even less is known about protective and compensating responses in humans, such as induction of antioxidant enzymes, including superoxide dismutases, glutathione peroxidase and catalase, as well as the role of nonenzymatic antioxidants, such as glutathione, ascorbic acid and vitamin E, which can prevent or downregulate oxidative stress. Not much is known about pulmonary antioxidant responses in humans during inhalation of particles, but chronic changes in antioxidant status during particle-related lung diseases such as CWP96, 97 and COPD98 as well as acute changes after smoking99 or episodes of farmer's lung100 illustrate that such responses occur and can be used for therapeutic intervention. We and others showed that antioxidants in blood and macrophages of coal miners are affected by exposure and during disease.101 We also found a relation between individual progression of respiratory disease and some antioxidant factors.102 In summary, as illustrated in Table VII, a number of the steps demonstrated in rats likely also occur in humans exposed to particles but there are also differences in tumor types as well as in the occurrence of particle overload and antioxidant depletion.


We described, in part A of this review, the current state of knowledge on the mechanisms of particle-induced lung carcinogenesis and, in this part, its implications for risk assessment of particle-induced tumors. The major focus here was on the capacity and properties of PSP to cause lung tumors in rats and whether the mechanisms and pathways are equally relevant for humans. The central hypothesis based on rat studies is that insoluble particles cause toxicity-driven (toxic particles) or surface-driven (PSP) pulmonary inflammation, which leads to genotoxic, proliferative and tissue remodeling progress toward fibrosis and neoplastic lesions. In the first part of this review,17 we discussed the major importance of ROS/RNS in these processes. The evidence for a secondary, inflammation-driven process in rats is abundant but still not conclusive since real intervention studies with PSP are lacking. Dose–response curves between lung tumors and instilled or retained dose metrics show the best correlation for particle surface area, indicating that this parameter should be the basis for future risk assessment, setting of standard values and compliance measurements. However, fibrosis, which is often regarded as a pathologic substrate associated with neoplastic sequels, is correlated to mass. Furthermore, fibrosis was not consistently associated with lung tumors in rats. Extrapolation to humans remains difficult because large data gaps are present, such as on antioxidant responses and the significance of inflammation-driven genotoxicity and proliferation in the human lung. Nevertheless, we consider current legislation103 calling for CB and TiO2 to be labeled as carcinogens based on positive rodent studies to be premature. In addition, a separate consideration of uF particles with regard to their carcinogenicity is not supported by the dose–responses curves of fine and uF particles using surface as a dose metric. However, the relation between lung tumors and volumetric dosing supports differential handling of uF particles due to their potent carcinogenicity.

Apart from surface and volume as dose metrics, uF particles have chemical properties and dynamics different from fine analogues, and other end points may be more relevant.9, 10, 11, 12 Apart from particle surface area, uF particle surface chemistry is important both from commercial and biologic points of view. The quartz issue has demonstrated how minor changes in surface components of fine particles have tremendous effects on their biologic properties.5, 76 Therefore, apart from studies on the validity of the rat model for humans, one should consider the importance of surface chemistry in particle-induced biologic responses.104 Bridging studies include (i) experimental approaches to modify in vivo inflammation with subsequent long-term follow-up of particle effects and (ii) those using antioxidant response profiling to bridge differences between rodents and humans in particle-induced neoplasms.


In this study we used biologic material and data from an animal study that was originally designed and performed by Drs. F. Pott and M. Roller (Medical Institute for Environmental Hygiene, Düsseldorf, Germany; TSCHG AZ §8 23.05-230-3-74/94). We acknowledge the work of Dr. W. Mahlke in the preparation of the light microscopy photographs (Fig. 1) and morphometric evaluations of fibrotic lesions as well as Ms. V. Suri, Ms. K. Bol, Ms. Y. Steinfartz and Ms. B. Adolf for support in the conduction of the study, fixation, cutting and staining of all sections. Finally, we thank Dr. M. Herbst for advice during the completion of the study and Dr. W. Drommer for the conduction of and advice on the histopathologic evaluations.