Nano-objects are defined as materials with at least one dimension between 1 nm and 100 nm (nanoscale), nanoparticles have all three dimensions in the nanoscale (International Organization for Standardization, 2008). According to an EU Recommendation nanomaterials consist for 50% or more of particles having a size between 1 and 100 nm. “Particle” means here a minute piece of matter with defined physical boundaries. “Aggregate” means a particle comprising of strongly
bound or fused particles and “agglomerate” is defined as a collection of weakly bound particles or aggregates where the resulting external surface area is similar to the sum of the surface areas of the individual components (European Commission, 2011). The term “particle” is used throughout this paper to cover primary particles as well as aggregates and agglomerates (Maier et al., 2006). The identifying feature of nanostructured materials is that their internal or surface structure is in the nanoscale, but their external dimensions are typically greater than
RESEARCH ARTICLE
Deposition behavior of inhaled nanostructured TiO2 in rats:
fractions of particle diameter below 100 nm (nanoscale) and the slicing bias of transmission electron microscopy
Peter Morfeld1,2, Silke Treumann3, Lan Ma-Hock3, Joachim Bruch4,5, and Robert Landsiedel3
1Institutes and Policlinic for Occupational, Social and Environmental Medicine and Prevention Research, School of Medicine and Dentistry, University of Cologne, Cologne, Germany, 2Institute for Occupational Epidemiology and Risk Assessment of Evonik Industries, Essen, Germany, 3Experimental Toxicology and Ecology, BASF SE, Ludwigshafen, Germany, 4IBE Institute for Lung Health GmbH, Marl, Germany, and 5University Duisburg-Essen, Medical Faculty, Essen, Germany
AbstractContext: In experimental studies with nanomaterials where translocation to secondary organs was observed, the particle sizes were smaller than 20 nm and were mostly produced by spark generators. Engineered nanostructured materials form microsize aggregates/agglomerates. Thus, it is unclear whether primary nanoparticles or their small aggregates/agglomerates occur in non-negligible concentrations after exposure to real-world materials in the lung.Objective: We dedicated an inhalation study with nanostructured TiO2 to the following research question: Does the particle size distribution in the lung contain a relevant subdistribution of nanoparticles?Methods: Six rats were exposed to 88 mg/m3 TiO2 over 5 days with 20% (count fraction) and <0.5% (mass fraction) of nanoscaled objects. Three animals were sacrificed after cessation of exposure (5 days), others after a recovery period of 14 days. Particle sizes were determined morphometrically by transmission electron microscopy (TEM) of ultra-thin lung slices. Since the particles visible are two-dimensional surrogates of three-dimensional structures we developed a model to estimate expected numbers of particle diameters below 100 nm due to the TEM slicing bias. Observed and expected numbers were contrasted in 2 × 2 tables by odds ratios.Results: Comparisons of observed and expected numbers did not present evidence in favor of the presence of nanoparticles in the rat lungs. In simultaneously exposed satellite animals agglomerates of nanostructured TiO2 were observed in the mediastinal lymph nodes but not in secondary organs.Conclusions: For nanostructured TiO2, the deposition of nanoscaled particles in the lung seem to play a negligible role.Keywords: Titanium dioxide, inhalation study, lung, nanoparticles, size distribution, TEM, slicing bias
Address for Correspondence: Dr Peter Morfeld, Institute for Occupational Epidemiology and Risk Assessment of Evonik Industries Rellinghauser Straße 1-11 45128 Essen, Germany. Tel.: +49 201 177-4400; Fax: +49 201 177-4403. E-mail: [email protected]
(Received 10 August 2012; revised 05 October 2012; accepted 05 October 2012)
the nanoscale range. The definition of a nanostructured material is covered by the Working Draft document ISO/WD TS 80004-4, former ISO/TS 12921 (International Organization for Standardization, 2011).
Such materials composed of nano-objects/nanopar-ticles often exhibit very different physical, chemical and biological properties than their larger scale counter-parts. Poorly soluble nanoparticles have moved into the public and scientific focus since Heinrich et al. (1995) showed increased lung tumor incidence in rats exposed to nanostructured TiO
2 and carbon black after long-term
inhalation. The concern increased sharply with reports that nanoparticles may translocate from their site of ini-tial deposition in the respiratory tract to the bloodstream (Nemmar et al., 2002, 2001) or to the brain (Oberdörster et al., 2004); however, it should be noted that follow up studies could not confirm that inhaled (99mTc)–labeled carbon nanoparticles directly penetrate into the blood-stream and identified biases in the detection of technigas labeled particles in the blood (Mills et al., 2006). It seems that only primary nanoparticles or small aggregates/agglomerates possess the ability to translocate. In those studies where translocation was observed, the particle sizes were always smaller than 20 nm and were mostly produced by spark generators. In contrast to the artifi-cially generated nano-objects, engineered nanostruc-tured materials form microsize aggregates/agglomerates in the atmosphere (Kuhlbusch et al., 2004, Kuhlbusch & Fissan, 2006). As a consequence of the discussion about translocation of nanoparticles the question is of major relevance whether primary nanoparticles and their small aggregates/agglomerates occur in non-negligible con-centrations after exposure to real-world materials in the lung (Levy et al., 2012).
Thus, we chose nanostructured TiO2 as a nonartificial
model substance and dedicated an inhalation study in rats to the following research questions: Does the particle size distribution in the lung contain a relevant subdistri-bution of nanoparticles?
Moreover, we will discuss a further controversial issue: the efficiency of intraluminar deposited particles to pen-etrate into the interstitial space and the lung associated lymph nodes; here again particle size according to some data may play a role (Oberdörster et al., 1994).
This study was performed as a complementary exami-nation of a previous publication on this 5-day inhala-tion rat study with nanostructured TiO
2 focusing on the
relationship between surface activity and toxicity (van Ravenzwaay et al., 2009). Six rats left from the main study were sacrificed either immediately after the expo-sure (first group) or after a recovery period of 2 weeks (recovery group). Only relevant technical and basic data are presented in detail again. Toxicological endpoints were already described in the paper mentioned above. To investigate the study question, particle count and size distribution were determined morphometrically by transmission electron microscopy (TEM) of ultra-thin lung slices of rats. A very high exposure concentration
(measured: 88 mg/m3, target: 100 mg/m3) was chosen to yield sufficient lung deposition for analysis and possible translocation into lymph nodes.
As already mentioned, the term particle is used throughout to cover primary particles as well as aggre-gates and agglomerates (Maier et al., 2006). Thus, our research question is of interest when discussing the pres-ence of primary nanoparticles and/or small aggregates and/or small agglomerates in the lung after inhalation exposure. In compliance with the general definition of nanoparticles we analyzed the percentage of particles smaller than 100 nm (“ultrafines”). A particular method-ological challenge is the two-dimensional presentation of larger particles in the ultra-thin TEM section planes and the analytical approach to extract informative data thereof. The ideal thickness of an ultra-thin section is about 80 nm (Hayat, 1986). Therefore, a larger particle (e.g. 300 nm) is cut into a slice at a random section plane. The profiles of this slice visible and analyzed in the TEM sections typically underestimate the true particle diam-eter: the slicing bias. This slicing bias is amplified by the systematic oversampling of large particles (Boyce et al., 2010, Cruz-Orive, 1987, Hsia et al., 2010). Other investi-gations did not control for this relevant bias (Schaudien et al., 2011, Creutzenberg et al., 2012). We developed a model based on approximating spheres to estimate expected numbers of particle diameters below 100 nm caused by the slicing bias. Observed and expected num-bers of diameters below 100 nm were contrasted in 2 × 2 tables by odds ratios to investigate whether the observed numbers were larger than the expected and, thus, indi-cated true particle diameters below 100 nm.
Complementary to the above mentioned examina-tions, translocation was evaluated by chemical analysis of Ti in different organs and tissues in satellite animals (van Ravenzwaay et al., 2009) exposed simultaneously with the animals evaluated in the current study.
Methods
Test substanceWe used nanostructured titanium dioxide (TiO
2) P 25
(uncoated), CAS number 13463-67-7, batch number P1S-126 4 (FK), purity 99.5%, i.e. a solid white powder, stored dry at room temperature and supplied by Evonik Industries, Essen, Germany (see Figure 1).
Study designThis study was a complementary examination of an inves-tigation focusing on the toxicity of materials with different particle surface activity (van Ravenzwaay et al., 2009). The animals of the current study were exposed simultaneously with those of the published study. Biological effects caused by exposing rats to this high concentrations of nanostruc-tured TiO
2 were examined by state of the art methods and
reported by van Ravenzwaay et al. (2009). Therefore, only materials and methods relevant to this paper are described in detail in this section.
Inha
latio
n To
xico
logy
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
BA
SF A
G o
n 09
/10/
14Fo
r per
sona
l use
onl
y.
Nanoparticles in the lung and TEM slicing bias 941
Two groups of three male Wistar rats were exposed nose-only to nanostructured TiO
2 for 6 h/day on five
consecutive days. Concurrent control group animals were exposed to conditioned air. Three animals and their concurrent controls were sacrificed immediately after the 5-days exposure (first group), the others after a post-exposure recovery period of 14 days (recovery group). A 14-day post-exposure period was installed to examine whether the effects were progressive (as the case for quartz) or tended to be reversible. Details were published by van Ravenzwaay et al. (2009). The animals exposed to TiO
2 were examined by electron microscopy
for morphometrical measurement of particles in the lung. The control animals served as negative controls, so were sacrificed and fixed in the same way as the exposed animals.
The target concentration was 100 mg/m3. This extremely high concentration was chosen to produce substantial lung burdens, and thus to increase the chance of developing a model to study the distribution of the particles and to detect any translocation.
Generation and characterization of the test atmosphereA liquid aerosol of nanostructured TiO
2 was produced
at a target concentration of 100 mg/m3 by atomizing an aqueous suspension containing 0.5% of the test material with a two-component nebulizer (Schlick Mod. 970), fed at a constant rate by a membrane pump (KNF Neuberger AG, Freiburg, Germany). The nanostructured TiO
2 was
dispersed in highly deionized water without any disper-sant. The aerosol was generated using compressed air in a mixing stage, mixed with conditioned dilution air and passed via a cyclone into a head-nose inhalation system. The generated aerosols dried on their way to the expo-sure chamber due to low humidity of the compressed air and the large surface area of the droplets. No tempera-ture increase was necessary.
The mass concentration of the test substances in the inhalation atmosphere was determined by gravimetrical measurement after sampling on a glass fiber filter (type MN 85/90 BF, d = 4.7 cm) using a 7 mm in line probe (Millipore, Schwalbach, Germany) at 3 L/min (1.25 m/s) for 30 min. The nonvolatile aerosol concentration in mg/m3 was cal-culated. To monitor the constancy of the aerosol, an on-line scattered light photometer was used throughout the study. The mass median aerodynamic diameter (MMAD) of the particles was determined with an eight-stage cascade impactor (Sierra-Andersen) using gravimetrical measure-ments. Moreover, the atmospheres were characterized by a light-scattering spectrometer PCS 2000 (PALAS Particle Technology, Karlsruhe, Germany) and a Scanning Mobility Particle Sizer (SMPS, 3022A/3071A, TSI, USA). All the mea-surements were carried out simultaneously at the same level of the exposure chamber.
The devices provided also the possibility to display the volume distribution and the mass concentration of nanoparticles. These values were calculated conserva-tively on the basis of the count distribution, assuming the particles were spherical with a diameter of 100 nm, were solid and possess the physical density of the mate-rial (4.46 g/cm3). The mass fraction of nanoparticles was determined by dividing the conservatively estimated mass of nanoparticles by the gravimetrically determined particle mass.
Details about the generation and characterization of the test atmosphere were published elsewhere (Ma-Hock et al., 2007).
Preparation for electron microscopyAfter exposure or recovery period, the three exposed ani-mals per group and time point, as well as three concurrent control animals, were subjected to deep anesthesia with Isoflo® (Essex GmbH, Munich, Germany). After open-ing the thorax the animals were sacrificed and fixed by whole-body perfusion using caccodylate buffer as a rins-ing solution, followed by perfusion of 5% buffered glutar-dialdehyde (GAH) as fixation solution. From each of the 5 lung lobes, 10 samples (cubes of 2–3 mm edge length) of not sub-pleural tissue were taken for TEM. These samples were systematically distributed over the lung paren-chyma, thus representing the lung parenchyma of the particular lung lobe under investigation. From the mirror part of the cut surface of each lung lobe an additional sec-tion (5 mm thickness) was taken and stored in 5% GAH. For TEM analysis, the tissue samples of the lungs were refixed with 2% buffered osmium tetraoxide. Five of the 10 cubes of each lung lobe were embedded in Epon mix-ture (Polysciences Europe GmbH, Eppelheim, Germany), which were further processed to semi-thin and ultra-thin sections. Semi-thin sections (500-nm thick) were stained with azure-methylen blue-basic-fuchsin (Ambf). Based on the evaluation of the semi-thin sections, five ultra-thin sections of 40 to 70 nm thickness per cube were prepared. They were stained with uranyl acetate and lead citrate. In one of these ultra-thin sections of one of the blocks, three
Figure 1. The morphology of the test material (TiO2). Primary particle
neighboring squares with 90-μm edge length (deter-mined by the wire mesh of the grid serving as specimen holder for the ultra-thin sections) were evaluated by TEM. The number of counted TiO
2 particles profiles was
documented separately for each block, each lung lobe and each animal. The total examined area for each lung lobe was ~0.0243 mm2. The evaluated squares consisted of intact tissue and were free of bronchi and blood vessels. Thus, these study areas of equal size should represent the alveolar compartments of the lung parenchyma exam-ined in each lung lobe of each animal.
To investigate whether the findings were sufficiently representative the lungs of all animals of the first group (necropsy immediately after the last exposure) were examined additionally in a validation substudy as fol-lows: from all five lung lobes two blocks different from those in the first examination were chosen and cut ultra-thin. On the ultra-thin sections one square in the center was chosen as described above and the observed particles in the section were measured following the pro-cedures applied in the first evaluation, and by the same technical assistants.
Mediastinal lymph nodes were fixed in a similar way and examined qualitatively for morphological changes as well as for presence of Ti particles.
Control animals were sacrificed and fixed in the same way as those exposed to TiO
2, they were evaluated in a
qualitative way.
Morphometric measurement of particle size in the lungsIn order to determine the distribution of particle diam-eters and particularly the percentage of nanoscale-TiO
2
(<100 nm) among the total number of particles observed, a morphometric measurement in all exposed animals was performed by TEM in each lung lobe (magnification ×8000). The measurements were performed interactively using the image analyzer SIS-Analysis (Soft Imaging Systems GmbH, Münster, Germany).
From each of the particles within these three fields, we measured and recorded the largest length “large diam-eter” and largest width “small diameter,” which were per-pendicular to each other (see Figure 2).
By counting the number of TiO2 particles from equal
sizes of study areas of the alveolar compartment the resulting data estimated the number density of deposited and retained particles in the lung tissue (although biased (Boyce et al., 2010, Cruz-Orive, 1987, Hsia et al., 2010)). The measured TiO
2 particle profiles and their large and
small diameter were registered separately for each block, each lung lobe and each animal.
Estimation and analysis of the particle diameter fraction with d < 100 nm adjusting for slicing bias due to the two-dimensional representation (cutting) by TEMThe observed particle size distribution was deter-mined as described above. However, particles are
three-dimensional objects. Size assessment by electron microscopy images only two dimensions (cutting). Due to this slicing bias the true size of the three-dimensional particle is usually larger than the observed size in the ultra-thin section. To assess how many particles profiles with an observed size smaller than 100 nm were due to the slicing bias, a model was developed to calculate the expected numbers given the observed particle size distri-bution >100 nm. To approximate the situation, particles are assumed to be spherical but with smaller spherical items attached on the surface. The latter simulate the roughness of the particle surface (Figure 3).
Whereas the true diameter is D the observed diameter d̂ is almost always smaller due to the slicing bias. In the following we will estimate the probability of d̂ d< given a true diameter D; d denotes an interesting upper limit of particle diameters, and usually corresponds to 100 nm in this study. Applying the Pythagorean Theorem we get
xd D2
2 2
2 2+ ⎛
⎝⎜⎞⎠⎟
= ⎛⎝⎜
⎞⎠⎟
Therefore,
x D d y D x= − =1
22 2 and – (1)
Within the particle with diameter D, the range for a sec-tion parallel to d that corresponds to d̂ d< is y. The term z describes the diameter of the small spheres attached to the surface and operates as a roughness parameter. If this roughness z is < d a cut through the outer region (small surface spheres) will always return that d dˆ< . Thus, assuming that the cutting plane through the particle is chosen at random the probability for an observed section with d̂ d< is
P d dy z
D z( )ˆ < =
++2
(2)
Figure 2. Schematic description of the morphometrical measurement of particle sizes in the lung tissue. First, the largest length was identified and measured (= large diameter). Second, the largest length perpendicular to the large diameter was identified and measured (= small diameter).
Inha
latio
n To
xico
logy
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
BA
SF A
G o
n 09
/10/
14Fo
r per
sona
l use
onl
y.
Nanoparticles in the lung and TEM slicing bias 943
We can insert y as given by (1) so that P d d( )ˆ < can be calculated from D, d, and z.
If only one sphere with diameter D is used to approximate the truncated distribution of the particle sizes ≥d, e.g., choosing D as the average of the observed diameter ≥d, only a very crude model will be obtained. To produce a far better approximation of the distribu-tion we applied ten spheres. The observed distribution of particle diameter ≥d was broken down in deciles and the average observed diameter was calculated for each of these deciles. The ten spheres were given these average diameters whereas the roughness parameter z was kept constant. However, these observed average diameter values are too small due to the slicing bias. We corrected the averages in a crude way by calculating the degree of underestimation at a unit sphere. Taking into account that the area of the unit circle is π we corrected the observed average diameter by applying the crude factor 4/π, i.e., with 1 ≤ i ≤ n = 10
D Mi i=4
π
where Mi is mean value of the observed diameters in
the respective decile of the condition distribution given diameters ≥d.
Thus, using n = 10 spheres for approximation the prob-ability P for observed diameters <d is
Pn
P d dii
n
= <=∑ 1
1
( )ˆ (3)
P d di( )ˆ < is calculated by formula (2) when applied to the ith of the n = 10 spheres. In consequence, the overall
probability P to observe a diameter <d due to the slicing bias can be calculated from (3) after we determined the mean observed particle diameter M
i, 1 ≤ i ≤ n = 10 in the
deciles of the distribution with diameters ≥d, and when d and z are given. In the following we use P as an estimate of the conditional probability that TiO
2 particles with true
diameters ≥d are observed with a measured diameter <d.Next, we apply P to calculate expected numbers of
particles with diameters <d due to the slicing bias. To do so, we assume that the null hypothesis is true, i.e. no true particles with diameters <d exist. If N denotes the num-ber of all observed particles (with true diameters ≥d due to the null hypothesis) and N d≥ the number of particles with observed diameters ≥d it follows
expected 1 1 N
1
= == ≥
PN P P P
P P N d
[ /( )]( )
[ /( )]
− −−
This term “expected” estimates the number of par-ticles with diameters <d we expect to see due to the slic-ing bias under the null hypothesis that no true particles with diameters <d exist. Thus, we estimated “expected” by multiplying the odds P/(1–P) and the number of observed particles with diameters ≥d.
We calculated prevalence odds ratios (Rothman et al., 2008) of observed vs. expected numbers of particles with diameters <d and 0.95-confidence intervals for the odds ratio (Cornfield’s method, see Kahn, 1983). The null hypothesis was tested by Fisher’s exact test in 2 × 2 tables (Rosenbaum, 1995). We like to emphasize that these odds ratios are the statistics of interest in this analysis. Thus, this analysis does not aim at estimating the number density of particles with diameters <d. The odds ratios measure relative differences between the observed and expected particle numbers with diameter <d and are well suited to test the null hypothesis of no true particles with diameter <d.
In this study, we are interested in particles smaller than 100 nm (“ultrafines”), therefore we set d = 100 nm. The parameter z describes the roughness of the particle surface. Considering a primary particles size of ~20 nm (Maier et al., 2006), we calculated the expected numbers assuming conservatively z = 10 nm or z = 20 nm. Thus, the assumption z d< mentioned above is fulfilled. We applied this model to both groups of animals (first and recovery) and evaluated the large and the small diameter separately. The same methods were used to analyze the validation substudy.
All calculations were done with Stata 10 (StataCorp, 2008). A significance level of 5% was chosen.
Results
Characterization of the atmospheresThe major characteristics of the aerosols investigated are summarized in Table 1. The measured test substance concentrations of TiO
2 aerosols were 88 ± 6.4 (mean ±
SD, in mg/m3). On-line scattered light photometer data
Figure 3. Schematic description of the model applied to assess the expected number of particles with diameters <d = 100 nm due to the slicing bias. The true diameter of the spherical particle is D. Small spheres with diameter z are attached to the surface to simulate the roughness of the surface (z = roughness parameter). If a sampling section parallel to d cuts through the region of y and z a sample diameter will be observed <d.
Inha
latio
n To
xico
logy
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
BA
SF A
G o
n 09
/10/
14Fo
r per
sona
l use
onl
y.
944 P. Morfeld et al.
Inhalation Toxicology
showed that the atmosphere was as constant as possible for this type of aerosol generation.
Cascade impactor measurements of particle sizes resulted in MMADs of 1.1 μm. The calculated mass frac-tions of particles <3 μm aerodynamic size ranged between 90.2% and 92.7%. The overall picture from the cascade impactor measurements indicated highly respirable aero-sols for rats. The mass concentration measured by the cascade impactor corresponded to 104% and 87% of the gravimetric mean of the respective day. These results dem-onstrate an unbiased sampling by the cascade impactor.
We characterized the atmospheres with a light-scattering spectrometer and SMPS, the results of these measurements are presented in Table 1. The distribution of nanostructured TiO
2 in the air measured by SMPS and
the light-scattering spectrometer is illustrated in Figure 4.For the nanostructured TiO
2 about 20% of the par-
ticles by number were smaller than 100 nm, calculating the mass of these particles, <0.5% of the particles were <100 nm in the atmosphere.
Body weights and clinical observationsDuring the whole study period, the animals exposed to TiO
2 did not show any clinical signs of toxicity. The body
weight development was comparable to their concurrent control animals. Before exposure the mean body weight (±SD) of the controls (n = 6; 3 for the first and 3 for the recovery group) was 221.4 ± 12.5 g, whereas that of the exposed animals (n = 6) was 223.9 ± 12.3 g. Other biologi-cal findings were already reported on (van Ravenzwaay et al., 2009).
Electron microscopy—qualitative findingsAnimals exposed to nanostructured TiO
large (up to 8 μm) aggregates and agglomerates. These aggregates and agglomerates were mainly located in the cytoplasm of the alveolar macrophages (Figure 5) and to a lesser degree they were found free as large aggregates entangled in laminar structures (surfac-tant) in the alveolar or bronchiolar lumen. Very rarely they were observed within the alveolar septal walls. In the recovery group the overall number of particles decreased and they were almost exclusively found in macrophages. By energy dispersive X-ray spectroscopy these particles were identified to be Ti-containing par-ticles (data not shown).
In mediastinal lymph nodes, TiO2 agglomerates were
found in cells regarded to be translocated alveolar mac-rophages (Figure 6).
Estimation and analysis of particle diameter fraction with d < 100 nmIn total, 2213 particles were observed in the evaluated region in the three animals sacrificed immediately after cessation of exposure (first group). We observed 908 particles in the three rats of the recovery group. The distributions of the observed particle sizes are given in Table 2.
The averages of the large diameters were 0.79 μm in the first group and 0.69 μm in the recovery group. Correspondingly, the mean values of the small diam-eters were 0.52 and 0.46 μm, respectively. Thus, con-cerning the observed particle size distribution, a shift toward smaller particle diameters was observed after the recovery period. All observed particle diameters were below 8 μm. Smallest observed diameters ranged between 14 and 32 nm, depending on the group and kind of diameter measured. The 1%-fractiles of the distribu-tions were always <100 nm, and also the 10%-fractile of the small diameter in the recovery group. However, the morphometrical measurements revealed that the major-ity of observed diameters are larger than 100 nm in both dimensions in all examined animals.
The truncated distributions of particle diameters greater than or equal to 100 nm were approximated by
Figure 4. Particle size distribution in the test atmosphere measured by light-scattering spectrometer and Scanning Mobility Particle Sizer (SMPS).
Table 1. Measured concentrations and particle size distributions in the test atmospheres.
Nanostructured TiO2
Target concentration (mg/m3) 100Measured concentration (mg/m3) mean ± SD
88 ± 6.4
MMAD (μm)/GSD 1.1/2.2
Count median diameterPCS 2000 (μm) 0.6
SMPS (μm) 0.20Count concentration of particles in SMPS (particle number/cm3)
88 × 104
Count concentration of particles in PCS 2000 (particle number/cm3)
5.6 × 104
Count concentration of particles <100 nm (particle number/cm3)
20.6 × 104
Count fraction of particles <100 nm 23.4%
Calculated* mass fraction measured <100 nm
0.5%
MMAD, mass median aerodynamic diameter; SMPS, Scanning Mobility Particle Sizer.*Assuming the particles were spherical and the density would be the physical density of 4.46 g/cm3.
Inha
latio
n To
xico
logy
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
BA
SF A
G o
n 09
/10/
14Fo
r per
sona
l use
onl
y.
Nanoparticles in the lung and TEM slicing bias 945
ten spheres as described in the Estimation and analysis of the particle diameter section. To give an example of this approximation by distribution deciles we show the truncated large diameter distribution as observed in the first group of animals immediately sacrificed after cessa-tion of exposure (Figure 7).
The distribution was clearly skewed to the right and, thus, could not be well approximated by a simple Gaussian distribution or, equivalently, by one sphere only. This skewness held for all observed distributions (cp. Table 2). However, Figure 7 shows that an approxi-mation based on the ten spheres followed the skewed distribution rather appropriately.
Next, we present the results of the analyses based on the model estimating the expected diameter below 100 nm due to the slicing bias (cp. section 2.6). Table 3 summarizes the findings and shows comparisons of observed versus expected particle numbers below 100 nm.
If observed numbers are larger than expected we have an indication that true particles exist with diameters <100 nm. This is equivalent to odds ratios greater than one. Otherwise, counted diameters <100 nm are likely due to the above described slicing bias and should not be understood as evidence in favor of nanoparticles. Three odds ratios calculated in the analysis of the first group
Figure 5. Transmission electron microscopy analysis of the lung. First group, sacrificed immediately after cessation of exposure. (A) Alveolar macrophage (arrow head) containing numerous aggregates/agglomerates of nanostructured TiO
2 within the cytoplasm. In
addition, there are small aggregates/agglomerates of nanostructured TiO2 free in the alveolar lumen or in the retro alveolar space (arrows);
*alveolar space. (B) Alveolar macrophage containing numerous aggregates/agglomerates of nanostructured TiO2 within the cytoplasm.
(C) Higher magnification of B; aggregates/agglomerates of nanostructured TiO2 located in membrane bound vesicles within the alveolar
macrophage (arrow).
Inha
latio
n To
xico
logy
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
BA
SF A
G o
n 09
/10/
14Fo
r per
sona
l use
onl
y.
946 P. Morfeld et al.
Inhalation Toxicology
were below one, and only one odds ratio showed a non-significant elevation (small diameter, roughness of 10 nm: OR = 1.16). In this case 265 diameters were observed below 100 nm but 228 were expected due to the slicing phenomenon.
The observed proportions of particles with diameters <100 nm were higher in the recovery group: significantly elevated odds ratios could be observed in this group when assuming a roughness parameter of 10 nm. If set-ting the parameter to a more realistic value of 20 nm, no indication of true diameters <100 nm was given in case of large diameters. For the small diameters we yielded a significantly elevated odds ratio of 1.37 in the recovery group (observed = 197, expected = 144). However, the lower confidence limit indicted an increase in the odds ratio by only 8%.
As described in the Preparation for electron micros-copy section and Morphometric measurement of particle size in the lungs section the evaluated tissue stem from one of a total of 10 blocks, which were systematically distributed over the lung parenchyma, thus representing the lung parenchyma of the particular lung lobe under investigation. To investigate whether our approach was sufficiently representative we repeated this analysis for the first group based on two other blocks out of the ten (validation substudy). We observed 1955 particles in this second approach showing a median of large diameters at 0.47 μm and of small diameters at 0.38 μm; averages were at 0.67 and 0.54 μm, respectively. When setting
Figure 6. Transmission electron microscopy analysis of mediastinal lymph nodes. First group, sacrificed immediately after cessation of exposure. (A) Control animal, (B) Animal exposed to nanostructured TiO
2. Aggregates/agglomerates within cells (arrows).
Figure 7. Distribution of large diameters ≥100 nm and 10 approximating spheres to estimate the expected number of particles <100 nm due to the slicing bias. A Gaussian distribution with the same mean and variance is shown additionally. Data of the first group of animals, immediately sacrificed after 5 days of exposure.
Table 2. Particle size distribution observed in the lung.
Large diameter and small diameter of particles in first group (sacrificed immediately after cessation of exposure) and recovery group (sacrificed two weeks after cessation of exposure). P1, P10, P50, P90, and P99 denote percentiles of the distribution.
Inha
latio
n To
xico
logy
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
BA
SF A
G o
n 09
/10/
14Fo
r per
sona
l use
onl
y.
Nanoparticles in the lung and TEM slicing bias 947
the roughness parameter to 20 nm the odds ratio for large diameters was found to be significantly below one (OR = 0.56, 0.95 CI: 0.46–0.68). For small diameters we got an odds ratio nearer to but still below one, although not significantly so: OR = 0.88, 0.95 CI: 0.75–1.07.
Discussion
This study has focused on the following question: does the particle size distribution in the lung contain a relevant subdistribution of nanoparticles? Thus, we examined the presence of primary particles or small aggregates/agglomerates in the lung after inhalation exposure to nanostructured TiO
2 particles.
The characterization of the test atmosphere showed a highly respirable dust aerosol with a small fraction of nano-objects/nanoparticles. This type of test atmosphere was considered to be typical for exposures to engineered nanomaterials. Using TEM, the deposited particles in the lung were examined immediately after the 5 days of inhalation exposure or after a recovery period of 14 days.
In electron microscopy, the particles visible in the ultra-thin section are two-dimensional random samples of the three-dimensional structures. Knowing that this slicing bias increases the fraction of small size particles visible in the ultra-thin section, we developed a model based on ten approximating spheres to estimate expected numbers of particle diameters below 100 nm caused by the slicing bias. We did so for both diameters, large and small, separately. Observed and expected numbers were contrasted in 2 × 2 tables (estimation of prevalence odds ratios).
An alternate approach may have been based on ellip-soids taking both diameters into account simultaneously but leads to a rather complicated numerical handling (Bach, 1964). We did not follow such a complicated pro-cedure because the irregularity of the agglomerates and
aggregates are not approximated well by either spheres or ellipsoids and there is no analytical solution at hand to estimate the number of expected diameters below a certain limit if the surface of the ellipsoids are modified to approximate rough structures or if the ellipsoids do not follow a constant ratio of the large to small diameters (Bach, 1965). Moreover, nano-objects are usually defined by having one dimension below 100 nm without consid-ering two diameters simultaneously.
Our approach to analyze estimated number densities of particles with diameters <100 nm is not affected by the oversampling of large particles in two-dimensional sections (Boyce et al., 2010, Cruz-Orive, 1987, Hsia et al., 2010). This is so because we approached the research question by a relative method (prevalence odds ratios). There was no need to estimate true particle number densities or true numbers or percentages of nanopar-ticles (and we did not do so). Note that it is reasonable to assume that estimated and observed number densities of nanoparticles suffer from the same kind and degree of observation bias. Therefore, the prevalence odds ratio comparing observed and expected number densities is free of this bias given the null hypothesis. Here, we follow Hsia et al. (2010, p. 402): “Where it is justified to conclude that relative bias is the same among different experimental groups, biased data can still allow valid between-group comparisons. Such data do not provide “true or accurate values but retain their comparative worth.” Mühlfeld et al. (2007) gave similar arguments.
Probability of primary nanoparticles or small agglomerates in the lung after inhalation exposureBoth in the lung and in the atmosphere, particles with diameters <100 nm were observed. However, it is not justi-fied to conclude the existence of nanoparticles in the lung from TEM observations of measured diameters below 100 nm. Such observations must always be compared to
Table 3. Evaluating the null hypothesis in the first group and the recovery group that all observed particles below 100 nm (large diameter or small diameter) are artificially produced by unfavorable slicing.
Roughness parameter
First group Recovery groupLarge diameter Small diameter Large diameter Small diameter
2-sided <0.0001 <0.0001 0.1260 <0.0001 0.0144 0.3593 <0.0001 0.0112Expected numbers under the null are calculated by a distribution model assuming 10 approximating spheres and a roughness parameter of 10 or 20 nm. Expected numbers are rounded down to the nearest integer. Odds ratios of observed vs. expected with 0.95-confidence intervals and exact p values (Fisher test) are reported.
Inha
latio
n To
xico
logy
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
BA
SF A
G o
n 09
/10/
14Fo
r per
sona
l use
onl
y.
948 P. Morfeld et al.
Inhalation Toxicology
expected values under the null hypothesis of no nanoparti-cles taking the characteristics of the measurement process into account. Clearly, the first group of animals showed no indication of any particle diameters <100 nm when adjust-ing for the slicing bias. In support of this finding from the first group, no substantial excesses of observed diameters <100 nm in comparison to the expected numbers were found in the recovery group when a realistic roughness parameter of 20 nm was used in the model. Therefore, the comparisons of observed to expected values did not gener-ate evidence in favor of the presence of nanoparticles in the rat lungs after taking the slicing bias of TEM into account.
Odds ratios calculated for the first group of animals were significantly lower than one if the roughness parameter was set to 20 nm, and even significantly below one in the analysis of the large diameters if the roughness parameter was set to 10 nm (see Table 3). This does not necessarily point at a deficiency of the model applied, but may indicate a measurement bias. The probability to observe and document diameters below 100 nm may be distorted downward because of the very large number of particles, in particular of large particles, observed in the lungs of the animals of the first group. The observer may overlook small particles in the presence of many large objects. Such a potential observation bias can be expected to decrease if the rats were exposed to a more realistic value far lower than the 88 mg/m3 chosen in this investigation. In support of this speculation, no note-worthy deficits of measured diameters <100 nm were observed in the recovery group.
In our validation approach we independently evalu-ated two other blocks out of ten in each lung lobe from the first group of animals. This second investigation confirmed the first: there was no indication of excess percentages of nanoscaled objects in comparison to the expected number of observations due to the slicing bias.
Of course, this relative adjustment was not validated by a comparison with “true” percentages of nanoparti-cles. Note that even approaches like the disector method (Boyce et al., 2010, Cruz-Orive, 1987, Hsia et al., 2010) do not give an unbiased estimate that may be used as a gold standard. In addition we argued that our relative method does not suffer from any obvious bias. Furthermore, our model assumes an internal surface structure similar to the primary particles of TiO
2 and thus, is obviously more
realistic than an evaluation based on observed percent-ages only that simply ignores the slicing bias of TEM. The results of the current study convincingly showed that naive analyses based on observed numbers only could be severly biased (Schaudien et al., 2011; Creutzenberg et al., 2012).
Evidently, agglomeration of the aerosolized and inhaled TiO
2 particles compromises a 20% fraction of par-
ticles <100 nm whereas only agglomerates >100 nm were determined in the lungs and lymph nodes. Agglomeration occurs when particles with a higher zeta potential immerse into an ionic solution. In the lung the particles are deposited in the lung lining fluid, which is composed
of phospholipids, surfactant proteins at the outer surface and in the hypophase of low weight proteins (albumin) together with a complex ionic salt mixture. According to the in vitro data of some laboratories agglomeration is feasible under these conditions (e.g. Bihari et al., 2008).
After spraying fluorescein isothiocyanate (FITC) labeled nanoparticle amorphous silica, dispersed in pure H
2O into lungs agglomerates of the amorphous silica at
the inner alveolar surface outside of the cells were pres-ent already after 30 min; three days later the labelled and agglomerated particles were exclusively located in alveolar macrophages (Martin Wiemann; IBE gGmbH, Münster, Germany pers. communication, 2012). It is con-ceivable that in our inhalation study similar conditions are effective upon the deposition of the partly nanosized TiO
2 at the inner alveolar surface outside of the cells were
present already after 30 min. In summary, for nanostruc-tured TiO
2 material, exposure to nanoscaled particles in
the lung does not seem to play a significant role. This is in line with observations and calculations by Maier et al. (2006) that aggregates and agglomerates consisting of nanostructured TiO
2 do not seem to desintegrate into
smaller structures when exposed to fluids similar to the lung surfactant. A recent review on carbon black lends support to the conclusion of no desintegration in the lung (Levy et al., 2012). However, by the methods applied we cannot rule out that nanoscaled objects may have pen-etrated into the mediastinal lymph nodes rather quickly and before our TEM observation started. Thus, a discus-sion of the translocation probability of the particles after inhalation exposure appears to be indicated.
Probability of translocation of manufactured nanostructured TiO
2 after inhalation exposure
It is generally known that nanoscaled materials can display increased activity in comparison to the bulk materials. Translocation into vasculature was shown for denaturated human serum albumin nanocolloidal particles (Nemmar et al., 2001), carbon nanoparticles (Nemmar et al., 2002) and nanosized TiO
2 generated by
a spark generator (Geiser et al., 2005). Later studies dem-onstrated that the ability to translocate into secondary organs across membranes like the air-blood barrier is related to the type of material and particle size (Kreyling et al., 2009), and that this fraction of particles translocat-ing into the vasculature and secondary organs is minimal (Kreyling et al., 2009, Mills et al., 2006, Möller et al., 2008). Less than 0.1% of the total deposition of carbon particles (25 and 80 nm) was found in secondary organs 24 h after the exposure (Kreyling et al., 2009).
In contrast to those research materials, manufactured nanostructured materials are often present as agglomer-ates due to van der Waals forces (Donaldson et al., 2006, Frogley & Wagner, 2002, Hussain et al., 2006, Pokropivnyi, 2002). Dispersing manufactured nanostructured materi-als results in agglomerates up to several micrometers in the test atmosphere. Depending on the material property, primary particles and small aggregates/agglomerates
Inha
latio
n To
xico
logy
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
BA
SF A
G o
n 09
/10/
14Fo
r per
sona
l use
onl
y.
Nanoparticles in the lung and TEM slicing bias 949
amount to only a minimal fraction of less than one mass percent of the total dust (Ma-Hock et al., 2007). This fact was confirmed in the current study. Thus, if a very small particle size is the triggering parameter for translocation, then the translocation ability of many manufactured, aggregated and agglomerated nanostructured materials should be extremely low.
In the present study, we observed large agglomer-ates (with a median large diameter of 0.6 μm and a median small diameter of 0.37 μm) rather than singular primary particles in the lung (first group, Table 3). Less than 1% of the observed particles were smaller than 0.058 or 0.035 μm, respectively. And these observed particle fractions were clearly overstated due to the slic-ing bias (Probability of translocation of manufactured nanostructured TiO
2 after inhalation exposure section).
Accordingly, TiO2 was only detected in the lung and the
mediastinal lymph nodes in simultaneously exposed satellite animals (Table 4) as previously reported (van Ravenzwaay et al., 2009). The TiO
2 content was below
the detection limit of 0.5 μg per organ in liver, kidney, spleen, and basal brain with olfactory bulb, correspond-ing to 0.025% of the total deposition. In accordance to the large agglomerate size, this finding indicates that the translocation rate of manufactured TiO
2 into sec-
ondary organs was negligibly low.The mediastinal lymph nodes were not considered
as a secondary organ because they are lung-draining lymph nodes. However, the following questions appear to be of importance. How to assess this fraction of parti-cles that translocates from the lung to the lung-draining lymph nodes? Which mechanisms trigger the penetra-tion to the remote areas of the lung, in particular the interstitial space and the lung lymph nodes, often also denoted together as “translocation to lymph nodes.” Does particle or particle aggregation size matters in this respect?
An elaborated TEM study (Geiser et al., 2008) showed that after an exposure of one hour to singlet nanoparti-cles, generated by a spark generator, these nanoparticles
deposited at the intra-alveolar surface, escaped the mac-rophage uptake; and 80% of the deposited nanoparticles still remained at the intraluminar site after 24 h (Geiser et al., 2005).
In a 12 weeks inhalation experiment two types of TiO2
(a 20 nm ultrafine and a 250 nm fine TiO2, concentrations
at about 22 mg/m3) were compared with regard to their inflammogenic potential and their impact on clearance behavior. On aerolization both types of TiO
2 formed
aggregates with a MMAD of ~74 nm. The ultrafine sample exerted much more inflammation accompanied by a significantly larger fraction migrating to the interstitial space and the lung lymph nodes. These data indicated that surface area of the retained particles appears to be a better dose determinant for inflammation than the volu-metric load of the alveolar macrophages (Oberdörster et al., 1994), a concept which has been confirmed by other studies (Sager & Castranova, 2009, Sager et al., 2008). Whether these surface-associated effects stem from smaller aggregates decayed from the deposited aerolized aggregates is not clear. However, observations and calculations (Maier et al., 2006) do not support such an interpretation. Furthermore, Pauluhn argued in favor of a volumetric dose approach instead of using surface-related dose metrics, based on theoretical considerations and empirical data (Pauluhn, 2010). In the present inves-tigation the morphometric evaluation of particle aggre-gation records a static situation, freezing the dynamic process (prevalence study). Thus, a very small fraction decaying into ultrafine TiO
2 particles—if it exists—was
not detectable by our methods.A quite common mechanism for the lymphatic trans-
location is the sequestration of toxic non-ultrafine par-ticles from the intra-alveolar space into the remote areas of the lung (interstitium and lymphnodes), as described by Klosterkötter for fine particles such as quartz DQ12 (“lymphotropism of quartz”) (Klosterkötter & Einbrodt, 1965). There is evidence that toxicity and ensuing inflammatory response governs the process of seques-tration. When blocking the cytotoxicity of quartz with PNO the penetration of particles into the lung lymph nodes was inhibited (Klosterkötter & Gono, 1969). The accompanying study (van Ravenzwaay et al., 2009) shows a higher penetration into the remote areas of the lung by quartz DQ12 as compared to pigmentary TiO
2
and nanoscaled TiO2. In this study, pigmentary TiO
2
was clearly more active than nanoscaled TiO2 due to a
dose dependent higher degree of inflammation. Similar results were obtained by Pauluhn when comparing two 10 and 40 nm AlOOH exposures for adverse lung responses with on aerolization MMADs of 0.6 and 1.7 μm, respectively in three concentrations (0.4, 3, and 28 mg/m3) (Pauluhn, 2010). Only the high dose expo-sure triggered inflammation (PMN in the BAL) and retardation of the clearance. Penetration of particles into lung lymph nodes “appears to be highly dependent on the accumulated dose and inflammatory response.” Taken together, these studies support the conclusion
Table 4. TiO2-levels as measured by inductively coupled plasma
atomic emission spectrometry across test animals (ICP-AES, see van Ravenzwaay et al. (2009)).
Test group AnimalTiO
2 in the
lung (μg)
TiO2 in the
lymph nodes (μg)
First group: sacrificed immediately after exposure
1 2219 2.52 1785 2.2
3 2068 1.7Mean 2024 2.1
Recovery group: sacrificed 14 days after exposure
4 1551 8.35 1668 11.7
6 1418 5.7Mean 1546 8.6
The animals were exposed simultaneously with the animals evaluated in the current study. No TiO
2 detected in secondary
organs.Method detection limit: 0.5 μg TiO
2 per g tissue.
Inha
latio
n To
xico
logy
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
BA
SF A
G o
n 09
/10/
14Fo
r per
sona
l use
onl
y.
950 P. Morfeld et al.
Inhalation Toxicology
that toxicity-related inflammation governs the trans-location into the remote areas of the lung parenchyma independently of particle size.
Besides particle size the surface chemistry may also play a role for translocation (Tian et al., 2009). The reported studies demonstrated the potential transloca-tion ability of certain nanomaterials and this toxicologi-cal aspect should not be dismissed in the course of an occupational safety evaluation. Therefore, with respect to translocation, manufactured nanostructured substances should be toxicologically examined in such a way that the investigated material is as similar as possible to the dust that may be released at production sites or during later stages of the product life cycle.
In summary, lymphatic translocation does not prove a decay of particles into nano-objects. The arguments and opinions expressed in Oberdörster et al. (1994) and Takenaka et al. (1986) are a non-sequitur.
A general limitation of this studySix animals are obviously not enough to generate a robust finding. The exposure concentration chosen (88 mg/m3) was substantially higher than exposures at the working place or those of downstream users. However, this high concentration elevated the sensitivity of our study to detect small particles and translocation. Thus, we believe that this investigation gives at least some insight into the occurrence and transfer of nanoparticles after very high exposures with nanostructured TiO
2, and it demonstrates
the complications research is confronted with if a valid argument shall be developed about the presence of nanoparticles in the lung.
Conclusion
For nanostructured TiO2 material, the presence of
nanoscaled particles in the lung does not seem to play a significant role. However, by the methods applied we cannot rule out that nanoscaled objects may have pene-trated into the mediastinal lymph nodes. But no TiO
2 was
detected in secondary organs (below detection limit).
Acknowledgment
We acknowledge Mrs Inta Koegel and Mrs Vanessa Hebestreit for their excellent technical assistance.
Declaration of interest
BASF SE and Evonik Industries produce TiO2 or produce
products containing TiO2.
References
Bach G. 1964. Über die Bestimmung der Anzahl dreiachsiger Ellipsoide aus der Anzahl ihrer Schnittellipsen in zufälligen Schnittebenen. ZAMP 15:205–209.
Bach G. 1965. Zufallsschnitte durch ein Haufwerk von Rotationsellip-soiden mit konstantem Achsenverhältnis. ZAMP 16:224–232.
Bihari P, Vippola M, Schultes S, Praetner M, Khandoga AG, Reichel CA, Coester C, Tuomi T, Rehberg M, Krombach F. 2008. Optimized dispersion of nanoparticles for biological in vitro and in vivo studies. Part Fibre Toxicol 5:14. http://www.ncbi.nlm.nih.gov/pubmed/18990217. Last accessed on 6 November 2012.
Boyce RW, Dorph-Petersen KA, Lyck L, Gundersen HJ. 2010. Design-based stereology: introduction to basic concepts and practical approaches for estimation of cell number. Toxicol Pathol 38:1011–1025.
Creutzenberg O, Bellmann B, Korolewitz R, Koch W, Mangelsdorf I, Tillmann T, Schaudien D. 2012. Change in agglomeration status and toxicokinetic fate of various nanoparticles in vivo following lung exposure in rats. Inhal Toxicol 24:821–830.
Cruz-Orive LM. 1987. Particle number can be estimated using a disector of unknown thickness: the selector. J Microsc 145:121–142.
Donaldson K, Aitken R, Tran L, Stone V, Duffin R, Forrest G, Alexander A. 2006. Carbon nanotubes: a review of their properties in relation to pulmonary toxicology and workplace safety. Toxicol Sci 92:5–22.
European Commission. 2011. Commission recommendations of 18 October 2011 on the definition of nanomateral. J European Union 38–40. http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:2011:275:FULL:EN:PDF. Last accessed on 6 November 2012.
Geiser M, Casaulta M, Kupferschmid B, Schulz H, Semmler-Behnke M, Kreyling W. 2008. The role of macrophages in the clearance of inhaled ultrafine titanium dioxide particles. Am J Respir Cell Mol Biol 38:371–376.
Geiser M, Rothen-Rutishauser B, Kapp N, Schürch S, Kreyling W, Schulz H, Semmler M, Im Hof V, Heyder J, Gehr P. 2005. Ultrafine particles cross cellular membranes by nonphagocytic mechanisms in lungs and in cultured cells. Environ Health Perspect 113:1555–1560.
Hayat MA. 1986. Basic techniques for transmission electron microscopy. Orlando: Academic Press.
Heinrich U, Fuhst R, Rittinghausen S, Creutzenberg O, Bellmann B, Koch W, Levsen K. 1995. Chronic inhalation exposure of Wistar rats and two different strains of mice to diesel engine exhaust, carbon black, and titanium dioxide. Inhal Toxicol 7:533–556.
Hsia CC, Hyde DM, Ochs M, Weibel ER; ATS/ERS Joint Task Force on Quantitative Assessment of Lung Structure. 2010. An official research policy statement of the American Thoracic Society/European Respiratory Society: standards for quantitative assessment of lung structure. Am J Respir Crit Care Med 181:394–418.
Hussain F, Hojjati M, Okamoto M, Gorga RE. 2006. Review article: polymer-matrix nanocomposites, processing, manufacturing, and application: an overview. J Compos Mater 40:1511–1575.
International Organization for Standardization. 2008. ISO/TS 27687. Nanotechnologies - Terminology and definitions for nano-objects - Nanoparticle, nanofibre and nanoplate. http://www.iso.org/iso/home/store/catalogue_tc/catalogue_detail.htm?csnumber=44278&commid=381983. Last accessed 6 November 2012.
International Organization for Standardization. 2011. ISO/TS 80004-4, Nanotechnologies -- Vocabulary -- Part 4: Nanostructured materials 2011. http://www.iso.org/iso/home/store/catalogue_tc/catalogue_detail.htm?csnumber=52195. Last accessed 6 November 2012.
Kahn HA. 1983. An introduction to epidemiologic methods. New York: Oxford University Press, Inc.
Klosterkötter W, Einbrodt HJ. 1965. [Quantitative animal experiments on the transport of dust from the lungs to the regional lymph nodes]. Arch Hyg Bakteriol 149:367–384.
Klosterkötter W, Gono F. 1969. Über den Einfluss von Polyvinylpyridin N-oxid-Aerosol auf die Rention, Penetration und Elimination von Quarz. Minister für Wirtschaf, Mittelstand und Verkehr des Landes NRW. In: Silikosebericht NRW, ed. Ergebnisse von Untersuchungen auf dem Gebiet der Staub- und Silikosebekämpfung im Steinkohlenbergbau. Bd. 7. Detmold: Bösmann, 159–164.
Inha
latio
n To
xico
logy
Dow
nloa
ded
from
info
rmah
ealth
care
.com
by
BA
SF A
G o
n 09
/10/
14Fo
r per
sona
l use
onl
y.
Nanoparticles in the lung and TEM slicing bias 951
Kreyling WG, Semmler-Behnke M, Seitz J, Scymczak W, Wenk A, Mayer P, Takenaka S, Oberdörster G. 2009. Size dependence of the translocation of inhaled iridium and carbon nanoparticle aggregates from the lung of rats to the blood and secondary target organs. Inhal Toxicol 21 Suppl 1:55–60.
Kuhlbusch TA, Neumann S, Fissan H. 2004. Number size distribution, mass concentration, and particle composition of PM1, PM2.5, and PM10 in bag filling areas of carbon black production. J Occup Environ Hyg 1:660–671.
Kuhlbusch TA, Fissan H. 2006. Particle characteristics in the reactor and pelletizing areas of carbon black production. J Occup Environ Hyg 3:558–567.
Levy L, Chaudhuri IS, Krueger N, McCunney RJ. 2012. Does carbon black disaggregate in lung fluid? A critical assessment. Chem Res Toxicol 25:159–164.
Ma-Hock L, Gamer AO, Landsiedel R, Leibold E, Frechen T, Sens B, Linsenbuehler M, van Ravenzwaay B. 2007. Generation and characterization of test atmospheres with nanomaterials. Inhal Toxicol 19:833–848.
Maier M, Hannebauer B, Holldorff H, Albers P. 2006. Does lung surfactant promote disaggregation of nanostructured titanium dioxide? J Occup Environ Med 48:1314–1320.
Mills NL, Amin N, Robinson SD, Anand A, Davies J, Patel D, de la Fuente JM, Cassee FR, Boon NA, Macnee W, Millar AM, Donaldson K, Newby DE. 2006. Do inhaled carbon nanoparticles translocate directly into the circulation in humans? Am J Respir Crit Care Med 173:426–431.
Möller W, Felten K, Sommerer K, Scheuch G, Meyer G, Meyer P, Häussinger K, Kreyling WG. 2008. Deposition, retention, and translocation of ultrafine particles from the central airways and lung periphery. Am J Respir Crit Care Med 177:426–432.
Mühlfeld C, Geiser M, Kapp N, Gehr P, Rothen-Rutishauser B. 2007. Re-evaluation of pulmonary titanium dioxide nanoparticle distribution using the “relative deposition index”: Evidence for clearance through microvasculature. Part Fibre Toxicol 4:7.
Nemmar A, Hoet PH, Vanquickenborne B, Dinsdale D, Thomeer M, Hoylaerts MF, Vanbilloen H, Mortelmans L, Nemery B. 2002. Passage of inhaled particles into the blood circulation in humans. Circulation 105:411–414.
Nemmar A, Vanbilloen H, Hoylaerts MF, Hoet PH, Verbruggen A, Nemery B. 2001. Passage of intratracheally instilled ultrafine particles from the lung into the systemic circulation in hamster. Am J Respir Crit Care Med 164:1665–1668.
Oberdörster G, Ferin J, Lehnert BE. 1994. Correlation between particle size, in vivo particle persistence, and lung injury. Environ Health Perspect 102 Suppl 5:173–179.
Oberdörster G, Sharp Z, Atudorei V, Elder A, Gelein R, Kreyling W, Cox C. 2004. Translocation of inhaled ultrafine particles to the brain. Inhal Toxicol 16:437–445.
Pauluhn J. 2010. Multi-walled carbon nanotubes (Baytubes): approach for derivation of occupational exposure limit. Regul Toxicol Pharmacol 57:78–89.
Pokropivnyi VV. 2002. Non-carbon nanotubes (review). Part 3. Properties and applications. Powder Metall Met Ceram 41: 123–135.
Rosenbaum PR. 1995. Observational studies. New York: Springer-Verlag.Rothman KJ, Greenland S, Lash TL. 2008. Modern epidemiology. 3rd
ed., Philadelphia, PA: Lippincott Williams & Wilkins.Sager TM, Castranova V. 2009. Surface area of particle administered
versus mass in determining the pulmonary toxicity of ultrafine and fine carbon black: comparison to ultrafine titanium dioxide. Part Fibre Toxicol 6:15.
Sager TM, Kommineni C, Castranova V. 2008. Pulmonary response to intratracheal instillation of ultrafine versus fine titanium dioxide: role of particle surface area. Part Fibre Toxicol 5:17.
Schaudien D, Knebel JW, Mangesldorf I, Voss J-U, Koch W, Creutzenberg O. 2011. Dispersion and retention of dusts consisting of ultrafine primary particles in lungs. Research Project F 2133. Dortmund/Berlin/Dresden: Federal Institute for Occupational Safety and Health (BAuA): 142. http://www.baua.de/en/Publications/Expert-Papers/F2133.html. Last accessed 6 November 2012.
StataCorp. 2008. Stata statistical software: release 10. College Station, TX: StataCorp LP. http://stata.com. Last accessed 6 November 2012.
Takenaka S, Dornhöfer-Takenaka H, Muhle H, 1986. Alveolar distribution of fly ash and of titanium dioxide after long term inhalation by Wistar rats. J Aerosol Sci 17:361–364.
Tian F, Razansky D, Estrada GG, Semmler-Behnke M, Beyerle A, Kreyling W, Ntziachristos V, Stoeger T. 2009. Surface modification and size dependence in particle translocation during early embryonic development. Inhal Toxicol 21 Suppl 1:92–96.
van Ravenzwaay B, Landsiedel R, Fabian E, Burkhardt S, Strauss V, Ma-Hock L. 2009. Comparing fate and effects of three particles of different surface properties: nano-TiO
