Registration Dossier

Diss Factsheets

Administrative data

Link to relevant study record(s)

Reference
Endpoint:
basic toxicokinetics in vitro / ex vivo
Type of information:
read-across from supporting substance (structural analogue or surrogate)
Adequacy of study:
key study
Justification for type of information:
1. HYPOTHESIS FOR THE READ-ACROSS APPROACH (ENDPOINT LEVEL)
The data used for the toxicological hazard assessment of bulk silver and nanosilver are not restricted to studies referring to only these individual substances as test items. Instead, a more generic approach for the toxicological assessment of “silver” in general is adopted by also including information/data generated with other inorganic silver substances such as silver acetate.
The basic assumption for this is that the systemic toxicity of any of the inorganic silver substances is driven by the silver ion (Ag+), which is considered the primary relevant species of silver for the hazard assessment. It is noted that many silver substances have a low if not negligible solubility and are thus considered to be of low bioavailability in the human (or animal) body. Once dissolved to any relevant extent, the other moieties or ions released from silver salts, such as chloride, sulfate etc. are not considered further in the hazard assessment, since these are typically either ubiquitous ions in a physiological environment, or generally not known to cause any particular concern regarding their toxicology. On the other hand, a separate assessment is considered if a substance is shown to cause local effects, such as the soluble and corrosive silver nitrate. The approach of conducting a common hazard assessment for silver substances where possible has the overall aim of avoiding unnecessary new animal studies.
To justify this generic approach, the published literature on toxicological effects of silver was assessed, as well as collected proprietary studies from the joint members of the silver REACH registration dossiers. Furthermore, the EPMF is currently performing the following toxicological studies which will supplement the Ag mammalian toxicity database:
• An in vivo toxicokinetics study with several test substances (silver acetate, silver nitrate, metallic silver (tested as a powder), nanosilver) to strengthen the basis for read-across. The read-across approach for mammalian toxicity endpoints will be updated based on the outcome of this study.
• An EOGRTS (OECD guideline 443) with silver acetate. Testing outcomes from a study utilising silver acetate are anticipated to be fully applicable to ionic silver (Ag+) irrespective of the donor silver substance forming this ion, i.e. being applicable to freely soluble salts such as silver nitrate, sparingly soluble compounds including disilver(I) oxide, and metallic silver including nanosilver forms.
The hazard assessment is supplemented by a separate report titled “Derivation of DNELs - silver substances“, which is attached to the CSR and the technical dossier in the endpoint summary of section 7. The hazard assessment and DNEL derivation for silver substances distinguishes between the following substance categories:
I. soluble/bioavailable silver substances (examples: silver nitrate, -sulfate, -acetate). These substances are characterised by water solubilities in the range ca. 8 g/L to >2,000 g/L, which is assumed to facilitate their bioavailability, and
II. poorly soluble/poorly bioavailable silver substances (such as silver oxide, iodide, chloride, bromide), including silver metal and nanosilver. The solubilities of the substances in this group are typically in the low mg/L range or even below (some are considered “practically insoluble”) rendering them poorly accessible to dissolution processes under physiological conditions.
Further, conventional read-across grouping would usually also consider aspects of bioaccessibility (i.e., solubility in surrogate physiological media) as well as in vivo toxicokinetic data for validation purposes. The former aspect has been studied for silver substances (please refer to the endpoint summary of the toxicokinetics; IUCLID section 7.1), but with the observation that the intrinsic chloride content in these in vitro test media coupled with the poor water solubility of silver chloride obviously is a limiting factor, rendering this testing of limited use for “silver”. Conversely, the toxicokinetic database for silver substances including non-nano sized silver metal with respect to differences in bioavailability is scarce. Hence the importance and relevance of the ongoing in vivo toxicokinetics study referred to above.
For this reason, the dossier often refers to data on silver metal nanomaterials for the assessment of silver metal (for a lack of substantial data on non-nanosized silver powder forms) and beyond this for the group of poorly soluble silver substances, albeit recognising that this often constitutes a conservative read-across.

2. READ-ACROSS APPROACH JUSTIFICATION (ENDPOINT LEVEL)
ATSDR (1990) previously summarised the available toxicokinetic information on “silver” in a comprehensive review (relevant information from this review has been included in the endpoint summary of the toxicokinetics; IUCLID Section 7.1), and these findings are considered to apply to all silver substances in general, but distinguish between different bioavailabilities where relevant. The information extracted from the ATSDR review is supplemented by more recent data on (i) in vitro bioaccessibility on metallic silver (a micron-sized and nano-sized silver powder), disilver oxide and silver nitrate (unpublished but included in IUCLID Section 7.1 as Midander and Wallinder 2009), showing that the dissolved silver concentration in different artificial physiological media were very similar and independent of the silver substance tested, and (ii) published in vitro and in vivo investigations relating to the bioaccessibility/bioavailability of silver, in particular from silver nanomaterials.
In addition, an in vivo toxicokinetics study with several test substances (silver acetate, silver nitrate, metallic silver, nanosilver) is ongoing to strengthen the basis for read-across. The study design is in line with OECD guideline 417 and this study will significantly improve the existing toxicokinetics dataset for silver substances as it is the first comparative in vivo toxicokinetics study for ionic silver, metallic silver and nanosilver. The read-across approach for mammalian toxicity endpoints will be updated based on the outcome of this study.
Based on the currently available toxicokinetics information, the hazard assessment and DNEL derivation for silver substances distinguishes between the following substance categories:
I. soluble/bioavailable silver substances (examples: silver nitrate, -sulfate, -acetate). These substances are characterised by water solubilities in the range ca. 8 g/L to >2,000 g/L, which is assumed to facilitate their bioavailability, and
II. poorly soluble/poorly bioavailable silver substances (such as silver oxide, iodide, chloride, bromide), including silver metal and nanosilver. The solubilities of the substances in this group are typically in the low mg/L range or even below (some are considered “practically insoluble”) rendering them poorly accessible to dissolution processes under physiological conditions.



For further information and data matrix see 'CSR Annex 10 - Read Across Justification Nanosilver HH_SUMMARY_200706' attached in IUCLID section 13.
Reason / purpose for cross-reference:
read-across source
Type:
other: Bioavailability based on dissolution in artificial biological fluids
Results:
The dissolved concentrations of silver in various artificial physiological media were very similar and seem independent of material type (silver metal, disilver oxide, silver nitrate).
Type:
other: Bioavailability based on dissolution in artificial biological fluids
Results:
It may be hypothesised that the complex ionic environment and the likely formation of poorly soluble silver chloride – an ubiquitous ion in physiology - leads to very similar equilibrium concentrations of dissolved silver, independent of the origin.

Particle characterization:

Ag-1

The silver metal powder, shows a broad particle size distribution that contains a fraction of smaller sized particles ( <0.1 μm in diameter). The silver particles show a fairly rough surface structure that contributes to the relatively large specific surface area, 3.1 m²/g, measured by BET analysis.

The Ag test material shows large variations in composition between different areas investigated. Areas with silver peaks (Ag 3d5/2) with a binding energy of 368.3 eV (associated with silver oxides) were observed as well as areas with the silver peak significantly shifter to higher binding energies of 369.2-370.7 eV (Ag 3d5/2). This shift is associated to metallic silver. The results imply particles with a thin surface oxide layer, or particles of varying surface coverage and thickness of the surface oxide. In addition, significant amounts of strongly oxidized carbon (carbon with single, and/or double oxygen bonding, carboxyl groups etc) of varying binding energies were detected on these particles. Their definite assignment is hazardous at this stage although it is clear that the origin is not from atmospheric contamination.

Ag-2

Ag-2, has a completely different appearance and surface morphology, just like “fine gravel” as seen from SEM images. Ag-2 particles show a broad distribution in size, including a large fraction of smaller sized particles and a large specific surface area of 7.9 m^2/g. Nano-sized silver particles demonstrate a high tendency to agglomerate, as depicted from particle size measurements in liquid medium (PBS).

The Ag-2 test material shows large variations in composition between different areas investigated. Areas with silver peaks (Ag 3d5/2) with a binding energy of 368.3 eV (most probably associated with silver oxides) were observed as well as areas with the silver peak significantly shifted to higher binding energies of 369.9-370.1 eV (Ag 3d5/2). This shifted peak is associated to metallic silver. The results imply particles with a thin surface oxide layer or particles of varying surface coverage and thickness of the surface oxide. In addition, significant amounts of strongly oxidized carbon of varying binding energies (carbon with single, and/or double oxygen bonding, carboxyl groups etc) were detected on these particles. Their definite assignment is hazardous at this stage although it is clear that the origin is not from atmospheric contamination.

Ag2O:

Silver oxide particles, Ag2O, seem to consist of larger agglomerates of finer, spherical silver oxide particles. The particles are all of similar size range and if the particles consist of agglomerated finer oxide particles, they are very stable since they do not fall apart when agitated during the size distribution analysis (this is seen from the agreement of measured particle size distribution). Silver nitrate particles are large, white salt crystal particles with a smooth surface morphology that looks like “wet ice bergs” when studied by SEM. The silver nitrate particles are large in size, which is also reflected by the small specific surface area, 0.03 m²/g, measured by BET analysis.

The Ag2O test material shows a silver peak (Ag 3d5/2) with a binding energy of 368.2 eV, associated with silver oxides. Both AgO and Ag2O show closely overlapping binding energies (367.9-368.2) why no unambiguous phase identification can be made. The oxygen component (O 1s) consists of three main peaks, located at 529.7, 531.8 and 533.6 eV, respectively. These peaks are associated with silver oxide, adsorbed hydroxyl groups, and/or bulk hydroxides, respectively Carbon (C 1s) is present on the surface as typical adventitious carbon due to atmospheric contamination.

AgNO3:

The AgNO3 test material, Figure 3, shows a silver peak (Ag 3d5/2) with a binding energy of 368.6, slightly higher compared to peaks usually associated with silver oxides (368.2 eV) and in addition, peaks of nitrogen (N 1s) and oxygen (O 1s) associated with nitrate. Carbon (C 1s) is present on the surface as typical adventitious carbon due to atmospheric contamination.

[XPS:

All test materials reveal adventitious carbon on the surface (285.0 eV) and small amount of oxidized carbon. A surface contamination layer of carbon is always observed to different extent due to the surface history, and its source is usually atmospheric and not possible to avoid.]

Bioaccessibility data- silver release:

The concentrations of silver that were released from the different particles were all over very similar and seem independent of material type (pure metal, oxide, nitrate). Since the concentrations measured after 2 hours and 24 hours of exposure were practically constant, it is concluded that the release/dissolution of silver takes place relatively fast and at these test conditions equilibrium is obtained during the time period prior to the 2 hour exposure and sampling. It may be speculated that silver dichloro complexes may have dominated the dissolution process. However no chemical speciation measurements were made to confirm this hypothesis. The released silver concentrations from all test materials were highest in phosphate buffered saline (PBS), having a neutral pH but the highest chloride content of the media investigated. Released silver concentrations were lowest in artificial gastric fluid (GST) with the most acidic pH and the lowest concentration of chlorides. The chloride concentration of the different test media is highest in PBS (5.35 g/L) > GMB (3.97 g/L) > ASW (3.05 g/L) > ALF (2.02 g/L) > GST (0.97g/L).

Table: Total concentration of released silver [μg/L] in the different test media

 

Test Material

GST  

Ag conc.

μg/L

ALF 

Ag conc.

μg/L

ASW 

Ag conc.

μg/L

GMB 

Ag conc.

μg/L

PBS 

Ag conc.

μg/L

Ag-1 2h

36.7±6.8

131.3±1.2

215.0±1.0

282.7±6.7

349.0±7.5

Ag-1 24h

35.3±1.2

123.0±1.0

224.0±59.8

270.0±3.5

352.3±6.5

Ag2O 2h

36.0±1.0

129.3±1.2

190.0±7.8

237.3±41.3

281.7±6.4
Ag2O 24h 36.0±1.0 123.0±0.0 184.7±2.1 264.3±1.5 338.0±2.6
AgNO32h 36.0±1.0 127.0±1.0 190.0±1.0 272.0±5.2 355.7±11.6
AgNO3 24h 34.0±1.0 120.0±1.0 186.0±2.6 259.3±1.5 347.3±2.1
Ag-2 2h  42.0 ±2.6 127.7 ± 1.2   209.3 ±14.6  278.3 ±8.1  280.7 ±92.2
Ag-2 24h   42.0 ±1.0  120.3 ±0.6 184.0 ± 1.0   260.3 ±1.5  340.0 ±1.0

Results presented as release rates of silver per unit surface area and hour of exposure (μg/cm²/hour), are shown in the table below. These rates are calculated from the released silver concentration considering the solution volume, measured BET area and sample weight, and the exposure time period. Error bars indicate the standard deviation of release rates from triplicate samples. The silver metal, silver oxide and silver nitrate particles all show higher average release rates of silver per surface area after 2 hours exposure and lower release rates after 24 hours of exposure. In agreement to the concentration findings, the highest release rates are observed when the test materials are exposed in PBS. Since there was no kinetic behaviour observed from the measured silver concentrations, the decreasing release rates are purely an effect of the normalisation by time. The release rate of silver decreased according to the following sequence for all samples investigated after 2 hours of exposure: PBS > GMB > ASW > ALF >> GST

Table: Release rates of silver [μg/cm²/hour] in the different test media.

Test Material

GST

Ag rate

μg/cm²/h

ALF 

Ag rate

μg/cm²/h

ASW 

Ag rate

μg/cm²/h

GMB 

Ag rate

μg/cm²/h

PBS 

Ag rate

μg/cm²/h

Ag-1 2h

0.070±0.012

0.25±0.01

0.41±0.00

0.51±0.01

0.67±0.03

Ag-1 24h

0.006±0.000

0.02±0.00

0.03±0.01

0.04±0.00

0.06±0.00

Ag2O 2h

0.068±0.001

0.24±0.01 0.35±0.02 0.43±0.07 0.53±0.02
Ag2O 24h 0.006±0.000 0.02±0.00 0.03±0.00 0.04±0.00 0.05±0.00 
AgNO3 2h 0.064±0.003  0.22±0.00 0.35±0.01 0.49±0.02 0.61±0.02
AgNO3 24h 0.005±0.000 0.02±0.00 0.03±0.00 0.04±0.00 0.05±0.00 
Ag-2 2h 0.078 ±0.005 0.24 ±0.01 0.39 ±0.03 0.52 ±0.02 0.54 ±0.18
Ag-2 24h 0.007±0.000 0.02±0.00 0.03±0.00 0.04±0.00 0.06±0.00

Another comparison of release data is enabled by normalizing the released amount of silver by the amount of particles loaded for a given time period (μg/μg). This quotient can function as an indicative measure of the percentage of the pure metal particles that has been released (dissolved) into solution. The results indicate that less than 0.5% of the silver metal particles have been dissolved during the 24 hours in all test media. From a bioaccessibility perspective, only a small fraction of silver is hence released regardless of the nature of the test material investigated. The results are presented as the maximum percentage determined from the triplicate samples investigated to avoid underestimating the amount of dissolved silver. This conclusion could also easily be drawn based on the results of silver concentration.

Table: Amount of released silver per amount of particles loaded [μg/μg] in the test media.

 Test Material GST Ag ratio μg/μg ALF Ag ratio μg/μg  ASW Ag ratio μg/μg GMB Ag ratio μg/μg PBS Ag ratioμg/μg
 Ag-1 2h  0.0004±0.0001  0.0014±0.0000  0.0022±0.0000  0.0028±0.0000  0.0037±0.0002
 Ag-1 24h  0.0004±0.0000  0.0013±0.0000  0.0022±0.0006  0.0027±0.0001  0.0036±0.0001
 Ag2O 2h  0.0004±0.0000  0.0013±0.0000  0.0019±0.0001  0.0024±0.0004  0.0029±0.0001
 Ag2O 24h  0.0004±0.0000  0.0013±0.0000  0.0019±0.0000  0.0026±0.0000  0.0036±0.0001
 AgNO3 2h  0.0003±0.0000  0.0012±0.0000  0.0019±0.0000  0.0026±0.0001  0.0033±0.0001
AgNO3 24h   0.0003±0.0000  0.0011±0.0000  0.0018±0.0000  0.0026±0.0001  0.0034±0.0001
Ag-2 2h  0.0004 ±0.0000  0.0013 ±0.0001  0.0021 ±0.0002 0.0028 ±0.0001  0.0030 ±0.0010
Ag-2 24h  0.0004 ±0.0000  0.0012 ±0.0000  0.0019 ±0.0000  0.0027 ±0.0000  0.0036 ±0.0000
Conclusions:
Interpretation of results (migrated information): other: see conclusion and summary
The dissolved concentrations of silver in various artificial physiological media were very similar and seem independent of material type (silver metal, disilver oxide, silver nitrate).It may be hypothesised that the complex ionic environment and the likely formation of poorly soluble silver chloride leads to very similar equilibrium concentrations of dissolved silver, independent of the originating substance. Chloride ion is ubiquitous in physiological systems.
Executive summary:

The concentrations of silver that were released from the different particles were all over very similar and seem independent of material type (pure metal, oxide, nitrate). The concentrations measured after 2 hours and 24 hours of exposure were practically constant. The released silver concentrations from all test materials were highest in phosphate buffered saline (PBS), having a neutral pH but the highest chloride content of the media investigated. Released silver concentrations were lowest in artificial gastric fluid (GST) with the most acidic pH and the lowest concentration of chlorides. The silver metal, silver oxide and silver nitrate particles all show higher average release rates of silver per surface area (μg/cm²/hour) after 2 hours exposure and lower release rates after 24 hours of exposure. In agreement to the concentration findings, the highest release rates are observed when the test materials are exposed in PBS. Since there was no kinetic behaviour observed from the measured silver concentrations, the decreasing release rates are purely an effect of the normalisation by time. The release rate of silver decreased according to the following sequence for all samples investigated after 2 hours of exposure: PBS > GMB > ASW > ALF >> GST.

Description of key information

Key value for chemical safety assessment

Bioaccumulation potential:
no bioaccumulation potential
Absorption rate - oral (%):
1
Absorption rate - dermal (%):
1
Absorption rate - inhalation (%):
1

Additional information

Introduction:

Toxicokinetic information on “silver” is required for the assessment of the relative contribution of the possible routes of entry into the human body (inhalation, skin, ingestion), and for a comparison of relative bioavailability of different silver substances. ATSDR (1990) previously summarised the available information in a comprehensive review, from which relevant paragraphs have been directly extracted below. The summarised findings are considered to apply to all silver substances in general, but distinguish between different bioavailabilities where relevant (reference: ATSDR (1990): Toxicological profile for silver. ATSDR - Agency for Toxic Substances and Disease Registry). Further references mentioned in text directlycitedfrom ATSDR can be found in the ATSDR report.

The earlier information extracted from the ATSDR review is supplemented by more recent data on (i) in-vitro bioaccessibility on metallic silver, disilver oxide and silver nitrate (unpublished), and (ii) published in-vitro and in-vivo investigations relating to the bioaccessibility/bioavailability of silver, in particular from silver nanomaterials. A highly relevant side-by-side comparison of the toxicokinetics of a soluble silver substance (silver acetate) vs. silver nanoparticles of different sizes has recently been conducted by US FDA (Boudreau, 2012), which is however not yet published in full; preliminary results are already considered in this dossier.

The water solubility of chemical substances is widely used as a first tier for screening purposes when assessing bioavailability. An overview of water solubilities for several silver substances is presented in the table below:

 

Water solubility of the substance (typically at 20 °C)

silver nitrate, AgNO3

up to ca. 2 kg/L (handbooks data range from 710 g/L to 2150 g/L)

silver acetate, AgCH3COOH

ca. 10 g/L

silver sulfate, Ag2SO4

8.2 g/L

silver carbonate, Ag2CO3

63 mg/L

silver chloride, AgCl

1.9 mg/L

silver bromide, AgBr

140 µg/L

silver iodide, AgI

30 µg/L

disilver oxide, Ag2O

considered practically insoluble (handbook data range from 1.6 mg/L at 20°C to 46 mg/L at 25°C)

silver sulfide, Ag2S

considered practically insoluble

silver (metal), Ag

considered practically insoluble (very slow dissolution depending on particle size/surface area and medium; see chapter 1.3 of the CSR)

Based on theoretical electrochemical considerations (Batchelor-McAuley et al., 2014), one can assume that under physiological (near-neutral) pH, the oxygen reduction on the surface of AgNPs leads to the formation of hydrogen peroxide, which is of relevance for the dissolution of silver nanoparticles in aqueous oxygenated systems: the relatively slow decomposition of H2O2 on silver and the dependency of the mass-transport away from the particle means that for sufficiently small particles (≤ 1μm) the dissolution process will be driven by the two-electron, two-proton reduction of oxygen. This value of the 1μm is obtained through consideration of the relative rate of the mass-transport of hydrogen peroxide from an isolated particle compared to its rate of decomposition (1.3x10-2 cm s-1). Above this size limit, the dissolution reaction will be similar to that of bulk silver, where the formation of water will become more pronounced. The presence of Ag+ complex species such as chlorides or thiols may however make dissolution more thermodynamically favourable.

Comparative in-vitro bioaccessibility of Ag, Ag2O and AgNO3(unpublished, Midander and Wallinder, 2009)

Metallic silver (two powder samples: D50=1.9 µm and D50=35 nm), disilver oxide and silver nitrate have been subject to in-vitro bioaccessibility testing in five different artificial physiological media. Phosphate-buffered saline (PBS, pH 7.4), is a standard physiological solution that mimics the ion strength of human blood serum, artificial sweat (ASW, pH 6.5), Gamble’s solution (GMB, pH 7.4) which mimics interstitial fluid within the deep lung under normal health conditions, artificial lysosomal fluid (ALF, pH 4.5), which simulates intracellular conditions in lung cells occurring in conjunction with phagocytosis and represents relatively harsh conditions and artificial gastric fluid (GST, pH 1.5). The test items were put into these solutions at a loading of 100 mg/L and incubated in the dark at 37 °C for 2 h and 24 h, respectively. Subsequently, undissolved particles were removed by filtration, and the filtrate was analysed for dissolved silver.

The dissolved concentrations of silver in all five artificial physiological media were very similar and apparently independent of the silver substance tested (i. e., metal, oxide and nitrate), despite their vastly differing water solubility. It is hypothesised that the complex ionic environment and the formation of poorly soluble silver chloride precipitates leads to very similar equilibrium concentrations of dissolved silver, independent of the originating substance. Chloride ions are ubiquitous in physiological systems, so that the formation of silver chloride particles can be assumed to be a limiting factor for systemic bioavailability of silver following exposure via the dermal, oral or inhalation route.

Other, published in-vitro bioaccessibility data

Several authors have conducted studies relating to the bioaccessibility of silver in physiological media, including investigations on the solubility/bioaccessibility of nano forms of silver (see for example the studies by Mwilu et al. (2013), Ma et al. (2012), Levard et al. (2013) and Zhang, et al. (2013), all summarised in tabular format in this CSR). As expected, the dissolution behaviour of metallic silver, including nanoforms, depends strongly on the type of material tested, and particle size distribution, aggregation or agglomeration and the presence of coatings have been shown to influence dissolution behaviour. As a general conclusion, the presence of chloride ions and formation of poorly soluble silver chloride or silver chloride complexes can reasonably be expected to limit the concentration of free silver ions in physiological media.

Maurer et al. (2014) analysed the dissolution behaviour of silver metal NPs (10 and 50nm) by tangential flow filtration, in water was well as in simulated physiological fluids such as cell culture media, artificial lysosomal fluid and artificial alveolar fluid (acc. to Stopford et al. 2003). The study aimed at separating dissolved silver ions from Ag-NPs. Whereas centrifugation-based separation techniques may lead to NP agglomeration and very small NPs are retained in the supernatant even after ultracentrifugation, in this case the recirculating tangential flow filtration system retained large particles within the continuous flow path, while allowing ions to pass through molecular filters, providing a reproducible NP separation from dissolved ionic silver. The ionic dissolution of Ag-NPs was dependent on exposure time, chemical composition and temperature of the exposure solution. Ionic dissolution was found to be greatest in alveolar fluid for both sizes of Ag-NPs.

Braydich-Stolle et al. (2014) characterised the impact of artificial interstitial, alveolar, and lysosomal fluid on the physical properties of Ag NPs and their rate of ionic release. The test items studied were (i) hydrocarbon-coated (Ag-HC) (25nm), and (ii) polysaccharide-coated (Ag-PS) (25nm). After dispersion in artificial fluids, the Ag-HC and Ag-PS NPs exhibited significant alterations in morphology, aggregation patterns, and particle reactivity: water-dispersed Ag-HC and Ag-PS had a primary size of 27 and 236.3 nm, respectively, with predominantly spherical morphology. In lysosomal fluid, extensive aggregation was observed for both coatings with a concurrent loss of spherical morphology (Ag-HC), with primary and aggregate particle sizes of 120 and 290 nm, respectively. For Ag-PS, the corresponding primary and aggregate diameters were 50 and 160nm. In alveolar fluid, both Ag-HC and Ag-PS also showed significant aggregation: individual and aggregate particles sizes were 68 and 204 nm for Ag-HC, and 53 and 341 for Ag-PS, respectively.

Loza et al. (2014) studied the formation of Ag-Cl NPs by adding silver nitrate to different simulated biological media at concentrations between 0.01g/L-0.1g/L (stated as being the range of the cytotoxicity of silver), with stirring at room temperature for 7d under sterile conditions without exclusion of light. The precipitated silver particulates were isolated by ultracentrifugation, re-dispersed in pure water, again subjected to ultracentrifugation and then analysed by X-ray powder diffraction, SEM and energy-dispersive X-ray spectroscopy. The initially present silver ions are bound as silver chloride due to the presence of chloride. Only in the absence of chloride, glucose was able to reduce Ag+ to Ag0. The authors concluded that the predominant silver species in biological media is dispersed nanoscopic silver chloride, surrounded by a protein corona which prevents the growth of the crystals and leads to colloidal stabilisation.

In another paper, Loza et al. (2014) describes the dissolution/precipitation behaviour of silver ions in physiological media further: whereas the equilibrium concentration of free ionic silver is determined by the solubility product of AgCl (1.7x10-10mol2/L2), ionic silver can also form complexes with organic compounds and thereby be removed from the equilibrium. However, the concentration of ionic silver is typically low in biological experiments (between 1 and 100 ppm) thereby possibly explaining why precipitation is typically not observed, since the resulting particles are very small. Conversely, AgNPs release silver ions when oxygen or hydrogen peroxide are present. In the presence of chloride ions, precipitation of silver chloride nanoparticles occurs.

Walczak, A. P. et al. (2013) performed an in vitro study on silver metal NPs (60nm) in simulated human digestion fluids (saliva, gastric and intestinal medium, with and without addition of proteins). After gastric digestion in the presence of proteins, the number of particles dropped significantly, however they increased to original values after the intestinal digestion. SEM-EDX revealed that the reduction in the number of particles was caused by their clustering; these clusters being composed of AgNPs and chlorine. During intestinal digestion, these clusters disintegrated back into single 60 nm AgNPs. In comparative experiments, intestinal digestion of AgNO3 solution in the presence of proteins also resulted in the formation of silver nanoparticles (20–30 nm) composed of silver, sulphur and chlorine.

Specific investigations on the chemical transformation of nanosilver in biological environments

Liu et al. (2012) studied the dissolution and chemical transformation of nanosilver in biological environments and draw the following conclusions from their studies: Silver nanoparticles undergo a set of biochemical transformations, incl. accelerated oxidative dissolution in gastric acid, thiol binding & exchange, photoreduction of thiol- or protein-bound silver to secondary zerovalent silver nanoparticles. Also, silver nanoparticles undergo rapid reactions between silver surfaces and reduced selenium species. Selenide is observed to rapidly exchange with sulfide in preformed Ag2S solid phases. The combined results allow proposing a conceptual model for silver nanoparticle transformation pathways in the human body: argyrial silver deposits are secondary particles formed by partial dissolution in the gastrointestinal tract followed by ion uptake, systemic circulation as organo-Ag complexes, and immobilization as zerovalent silver nanoparticles by photoreduction in light-affected skin regions. The secondary silver particles then undergo detoxifying transformations to sulfides and partly further to selenides or Se/S mixed phases through exchange reactions. The formation of secondary nano-sized particles in biological environments implies that silver nanoparticles are not only a product of industrial nanotechnology but can also also been present in the human body following exposure to more traditional chemical forms of silver. The research presented above supports the hypothesis that nanosilver particles are not absorbed to any relevant extent and/or further distributed within the body as intact particles, but rather in ionic form after dissolution. Depending on the chemical environment, secondary particles may be formed via transformation into poorly soluble forms such as metallic silver, chloride, sulfide or -selenide.

In a study comparing tissue distribution after oral dosing with Ag-NPs and AgNO3, the authors concluded that tissue levels were generally much higher in oral dosing with AgNO3, and also concluded that (i) these tissue levels were highly correlated with the dissolved (ionic) fraction of the AgNP suspension, and (ii) silver nanoparticles were detected not only in AgNP-treated rats, but also in those dosed with AgNO3, obviously demonstrating formation of these from ionic silver in vivo (van der Zande et al., 2012).

The dissolution of silver metal NPs (60nm, 5% w/w suspension in ethylene glycol) and binding of released Ag+ ions in cellulo was studied in primary murine macrophages, exposed to a total dose of 5 μg/mL AgNPs either as single exposure (acute mode) for 6h or 24h, or as 1.25 μg/mL AgNPs per day for 4 days (chronic mode). Dissolution rates were dependant on the exposure scenario: chronic exposure lead to a higher Ag+ release than acute exposure; Ag-S bond lengths were 2.41 ± 0.03A and 2.38 ±0.01A in acute and chronic exposure respectively, compatible with diagonal AgS2 coordination. Glutathione was identified as the most likely putative ligand for Ag+. Internalised AgNPs in macrophages were characterised by TEM as electron-dense deposits in the cytoplasm

Below, available key information on absorption, distribution, metabolism and excretion of “silver” as relevant for human health risk assessment of silver is presented.

Absorption

Oral absorption

ATSDR, 1990: “Based on medical case studies and experimental evidence in humans, many silver compounds, including silver salts and silver-protein colloids, are known to be absorbed by humans across mucous membranes in the mouth and nasal passages, and following ingestion. The absorption of silver acetate following ingestion of a 0.08 mg/kg/day dose of silver acetate containing radiolabelled silver (110mAg) was studied in a single female by in vivo neutron activation analysis and whole body counting: approximately 21% of the dose was retained in the body at 1 week (East et al. 1980; MacIntyre et al. 1978). ”

However, the reliability of this method is not documented and several assumptions in the publication by East et al. leading to this estimate render this a likely overestimate. In a well-documented comparative investigation assessing the bioavailability of110msilver nitrate in mice, rats, monkeys and dogs via oral, intravenous and intraperitoneal administration, only about 1% or less of an oral dose was absorbed with the exception of dogs (<10%) (Reference: Furchner et al. 1968: Comparative metabolism of radionuclides in mammals-IV. Retention of silver-110m in the mouse, rat, monkey, and dog, Health Physics 15:505-514).

A study comparing tissue distribution levels in rats after dosing with AgNPs and AgNO3showed that silver tissue levels were much higher in liver after AgNO3-dosing than with AgNPs, and also demonstrated that the tissue levels were highly correlated to the dissolved (ionic) fraction of the AgNP suspension; this suggests that mainly Ag+ passes the intestine upon oral intake (van der Zande et al., 2012).

More conclusive data may become available from a study currently being conducted in the US (Boudreau, 2012) when the results are published.

Dermal absorption

Several published sources report information on percutaneous absorption of silver, with varying degrees of reliability and relevance for risk assessment.

 Some older information exists which can however be used as supportive information. Since this data is often only available from secondary sources, dedicated endpoint records have not been prepared in the technical dossier on the studies mentioned in this paragraph. Therefore, for technical reason, these references do not appear in the automatically generated reference list in the CSR. In a well-documented study with guinea pigs less than 1% of the applied dose of silver nitrate was absorbed through the skin. However, this study has major methodical deficiencies, and is therefore considered only as supporting data (Wahlberg et al., 1965: Percutaneous toxicity of metal compounds. A comparative investigation in guinea pigs. Arch. Environ. Health 11, 201-204). A review article exists which contains references to earlier published investigations, in which an in-vivo percutaneous absorption study in guinea pigs with110Ag as tracer is described. Despite that this study also has considerable methodological shortcomings compared to current standards, the authors likewise conclude on a dermal absorption rate of <1% (Skog & Wahlberg, 1964: A comparative investigation of the percutaneous absorption of metal compounds in the guinea pig by means of the radioactive isotopes: 51Cr, 58Co, 65Zn, 110mAg, 115mCd, 203Hg. J. Invest. Derm. 43, 187-192, 1964. Cited in Hostynek, 1993: Metals and the skin. Crit. Rev. Toxicol. 171-235). Other similarly outdated data relate to either non-standard test systems or absorption through wounded or burnt skin, and is therefore not considered relevant for the assessment of percutaneous absorption through intact skin, as required for risk assessment purposes. For example, one case report study of 11 human volunteers on absorption of silver from the nasal septum after cauterisation for nose bleeds suggests a significant increase of silver blood concentrations 3 hours after administration. This study is not considered as particularly relevant for human health risk assessment because of the involved skin injury (Nguyen et al., 1999: Argyremia in septal cauterization with silver nitrate. J. Otolaryng. 28, 211-216).

Most recently, increased interest in the safety assessment of silver nanomaterials has produced a number of relevant publications on these materials:

Larese et al. (2009) compared the percutaneous absorption of polyvinylpirrolidone-coated silver nanoparticles in-vitro through intact and damaged human skin. Whereas the publication has some reporting deficiencies, the percutaneous absorption rate (percentage of the applied dose absorbed during 24h of exposure) can be calculated: the exposure concentration is given as 70 µg Ag/cm2, and the median penetration rates (over 24h) are 0.46 ng/cm2and 2.32 ng/cm2for intact and damaged skin, respectively. Thus, the percentage of the applied dose absorbed during 24h of exposure is 0.00066 % for intact skin and 0.0033 % for damaged skin.

Samberg et al. (2010) studied dermal absorption and irritation due to silver nanoparticles in-vivo in pigs. By microscopic investigations silver nanoparticles were localized only in the superficial layers (50nm particles) and on the top layer (20nm particles) of the stratum corneum and did not appear to penetrate into the deeper dermis. This may be considered as supportive of the assumption that silver does not penetrate through human skin to any relevant extent.

Brandt (2012) compared the percutaneous absorption of silver from two antimicrobial topical creams in mice in-vivo. No absorption rates are reported, but the study concludes that percutaneous absorption of silver from an antimicrobial cream containing nanoscale silver was much lower than from a cream containing (soluble) silver sulfadiazine. Analysis of inner organs and blood of mice treated with the commercial cream containing 0.1% nanoscale silver revealed extremely low percutaneous absorption rates, resulting in barely detectable silver ion concentrations with values not differing significantly from those of the untreated group.

Moiemem et al. (2010) studied the systemic absorption of silver in six patients with large scale burn wounds (median of 46.1% of the total body surface affected). Patients were treated with a commercial wound dressing containing “nanocrystalline” silver. Silver levels in blood serum were analysed over time. A significant increase of silver in blood was observed, and levels decreased following the end of treatment. None of the patients had any symptoms or signs suggesting argyria. The authors conclude that elevated silver levels in blood were similar to those reported following the use of silver sulfadiazine.

In an earlier study by the same group of researchers (Vlachou, 2007) similar investigations were carried out with patients with smaller scale burn wounds (median of 12% of the total body surface affected). A significant increase of silver in blood was observed during treatment, and levels decreased to baseline levels following the end of treatment. No haematological or biochemical indicators of toxicity associated with the silver absorption were observed. The overall finding is not considered relevant for human health risk assessment of industrial chemicals, in view of the involvement of damaged/wounded skin.

George et al. (2013) studied the absorption of silver from a nanocrystalline silver wound dressing when applied for 4-6 days to intact human skin of 16 healthy patients. The analysis of absorption/penetration was conducted by microscopy (optical and SEM) and XRD analysis of skin samples, as well as by silver analysis in blood serum. Although, according to the authors silver nanoparticles may penetrate into intact human skin in vivo beyond the stratum corneum as deep as the reticular dermis, the absorbed silver species appear to precipitate in clusters across the epidermis. However, despite this silver deposition in the dermis, silver nanoparticles did not reach systemic circulation.

Trop et al. (2006) observed argyria-like symptons and increased blood silver levels in a clinical case report on treatment of burn wounds with a silver-coated wound dressing. This indicates some absorption of silver through damaged/burned skin, but the absorption rate was not quantified. Therefore, and since uptake through damaged skin is not relevant for the risk assessment under REACH, this study is not considered further.

As supportive information, reference is made to a study not directly related to silver as such: Campbell (2012) assessed the disposition of inert polystyrene nanoparticles in mammalian skin using confocal microscopy. These polystyrene nanoparticles when applied in aqueous suspension could only infiltrate the stratum disjunctum, i. e. skin layers in the final stages of desquamation. This minimal “uptake” was independent of contact time size of nanoparticles tested. Overall, these results demonstrate objectively and semi-quantitatively that (inert) nanoparticles of a wide size range cannot penetrate beyond superficial skin layers of the barrier and are unlikely to reach the viable cells of the epidermis or beyond.

Overall, the available data for soluble silver substances as well as silver nanomaterials indicate dermal absorption rates well below 1% of the applied dose. Data obtained with silver nanomaterial preparations (which can be assumed to also contain a certain amount of dissolved, ionic silver) indicate some penetration into the stratum corneum, but detailed follow-up investigations have shown that this bound material does not become systemically available, but instead is lost via desquamation.

These conclusions above are consistent with the methodology proposed in HERAG guidance for metals (HERAG fact sheet - assessment of occupational dermal exposure and dermal absorption for metals and inorganic metal compounds; EBRC Consulting GmbH / Hannover /Germany; August 2007). The following default dermal absorption factors (reflective of full-shift exposure, i. e. 8 hours) are therefore considered adequately conservative both for (soluble) silver substances as well as silver nanomaterials, i. e. 1.0% from exposure to liquid/wet media, and 0.1% from exposure to dry (dust).

Inhalation absorption

Model calculations of inhalation absorption based on laboratory dustiness tests with representative materials

Experimental investigations have been conducted on six samples of different silver compounds: silver metal (3 different sizes), disilver oxide (2 different batches) and one representative sample of silver nitrate (crystalline powder). The samples were subject to mechanical agitation in a rotating drum apparatus and the mass fraction of the material that becomes airborne was determined (“total dustiness”). In addition, the particle size distribution of the airborne dusts was determined with a cascade impactor. Then, the MPPD model was used, to estimate the fractional deposition of such dusts in three regions of the human respiratory tract: (i) extrathoracial fraction (head), (ii) tracheo-bronchial fraction (TB) and (iii) the pulmonary fraction (EBRC, 2010: Investigations on dustiness and particle size of airborne dusts of six silver compound samples. Unpublished report for the Precious Metals Consortium, EBRC Consulting GmbH, Hannover, Germany). The results are presented in tabular form below.

MPPD model results: Fractional deposition (%) in different regions of the respiratory tract. Data on physical particle size of the original test materials and on total dustiness is also given:

 

 

Silver powder Batch PMC1

Silver powder Batch PMC2

Silver powder Batch PMC3

Disilver oxide Batch PMC1

Disilver oxide Batch PMC2

Silver nitrate Batch PMC1

D50

ca. 30 µm

ca. 2 µm

ca. 35 nm

ca. 4 µm

ca. 7-18 µm

ca. 370 µm

Total dustiness

155 mg/g

91 mg/g

248 mg/g

149 mg/g

126 mg/g

0.83 mg/g

Deposition (total)

45.4 %

51.2 %

53.3 %

46.1 %

49.1 %

37 %

Deposition (head), fH

45.3 %

50.1 %

52.4 %

45.3 %

47.9 %

36.8 %

Deposition (TB), fTB

0.1 %

0.4 %

0.3 %

0.3 %

0.5 %

0.1 %

Deposition (pulmonary)

fPU

0.0 %

0.7 %

0.6 %

0.6 %

0.8 %

0.1 %

Based on the fractional deposition, an inhalation absorption factor can be calculated under the following assumption: The material deposited in the head and tracheobronchial regions would be translocated to the gastrointestinal tract without any relevant dissolution, where it would be subject to an assigned default gastrointestinal uptake at a ratio of 1% (see above). The material that is deposited in the pulmonary region may conservatively be assumed by default to be absorbed to 100%. This absorption value is chosen in the absence of relevant scientific data regarding alveolar absorption although knowing that this is a conservative choice specifically for poorly soluble substances. Formula: (fH+ fTB) * absoral+ fPU* abspul= inhalation absorption factor. Example calculation for batch PMC3: (52.4% + 0.3%) * 1 % + 0.6 % * 100% = 1.13 %

Estimated inhalation absorption factors, assuming 1% absorption in the gastrointestinal tract (considered applicable to soluble silver substances; for the absorption of metallic silver this is considered a conservative worst-case) of material initially deposited in the head or tracheobronchial region and 100% absorption of material deposited in the pulmonary region:

 

Silver powder Batch PMC1

Silver powder Batch PMC2

Silver powder Batch PMC3

Disilver oxide Batch PMC1

Disilver oxide Batch PMC2

Silver nitrate Batch PMC1

Estimated inhalation absorption factor

0.45 %

1.21 %

1.13 %

1.06 %

1.28 %

0.47 %

Given the inherently conservative assumption of 100% absorption for pulmonary deposition and 1% oral absorption for the poorly soluble substances, it is considered adequate to take a value of 1% forward for inhalation absorption of silver substances.

Distribution

ATSDR, 1990: "The distribution of silver to various body tissues depends upon the route and quantity of silver administered and its chemical form. An oral dose of silver, following absorption, undergoes a first pass effect through the liver resulting in excretion into the bile, thereby reducing systemic distribution to body tissues (Furchner et al. 1968). The subsequent distribution of the remaining silver is similar to the distribution of silver absorbed following exposure by the inhalation and dermal routes and following intramuscular or intravenous injection. Silver distributes widely in the rat following ingestion of silver chloride (in the presence of sodium thiosulfate) and silver nitrate in drinking water (at 88.9 mg silver/kg/day for silver nitrate) (Olcott 1948); The amount of silver in the various tissues was not measured, although qualitative descriptions of the degree of pigmentation were made. High concentrations were observed in the tissues of the reticuloendothelial system in the liver, spleen, bone marrow, lymph nodes, skin, and kidney. Silver was also distributed to other tissues including the tongue, teeth, salivary glands, thyroid, parathyroid, heart, pancreas, gastrointestinal tract, adrenal glands, and brain. Within these tissues advanced accumulation of silver particles was found in the basement membrane of the glomeruli, the walls of blood vessels between the kidney tubules, the portal vein and other parts of the liver, the choroid plexus of the brain, the choroid layer of the eye, and in the thyroid gland (Olcott 1948; Moffat and Creasey 1972; Walker 1971) ".

Tissue distribution studies involving i. v. administration of soluble radiotracer (110m) silver nitrate indicate that liver is a likely target organ for silver: 2 hours p. a., the highest tissue concentrations were in liver (12.4% of dose), with substantially lower levels in other organs (spleen 1.2% and kidneys 1.1%, for example (Gregus & Klaassen, 1986).

Some information on tissue distribution following exposure of experimental animals to nanosilver is available from recent toxicological studies by Sung et al. (2009) and Kim et al. (2010). Tables with the corresponding data are contained in the respective study summaries in the technical dossier and only the brief summary is presented here.

In a subchronic inhalation toxicity study conducted with nanosilver (Sung et al. 2009) rats were exposed to 49, 133 and 515 µg/m³ for 6h per day, 5 days per week for 13 weeks. At the end of the study, the authors measured silver concentrations in liver, kidneys, olfactory bulb, brain and lung tissues and report the following finding: “Silver concentration in lung tissue from groups exposed to silver nanoparticles for 90 days were a statistically significant (p < 0.01) and increased with dose. There was also a clear dose-dependent increase in the silver concentration in the blood, and dose-dependent increase in the liver silver concentration for both genders. Silver concentration in the olfactory bulb was higher than in brain, and increased in a dose dependent manner in both the male and female rats (p < 0.01). Interestingly, silver concentrations in the kidneys showed a gender difference, with the female kidneys containing two to three times more silver accumulation than in male kidneys”.

In a subchronic oral toxicity study with nanosilver (Kim et al. 2010), four groups of rats received 0, 30, 125 and 500 mg/kg/day of silver via gavage for 13 weeks and silver levels in selected tissues were measured: Results for testes, liver, kidney, brain and lung are reported in the publication and the authors summarise their findings as follows: “There was a statistically significant (P < 0.01) dose dependent increase in the silver concentration of all the tissue samples from the groups exposed to silver nanoparticles in this study. In addition, a two-fold higher accumulation of silver in the kidneys of female rats when compared with the male rats occurred across all the dose groups indicating a marked gender-dependent distribution”.

Further data are anticipated to originate from toxicokinetic and toxicity studies currently being conducted in the US (Boudreau, 2012).

Metabolism

Silver is not subject to any metabolism in its true sense regardless of its original chemical speciation, with one exception which relates to the formation of argyria particles through reaction of ionic silver to sulfide/selenide particles (postulated mechanism: Liu et al. 2012; van der Zande et al., 2012 – for details, see section above).

Excretion

ATSDR, 1990: "Following oral exposure to silver acetate in humans, silver is eliminated primarily in the faeces, with only minor amounts eliminated in the urine (East et al. 1980). The rate of excretion is most rapid within the first week after a single oral exposure (East et al. 1980). Whole-body retention studies in mice and monkeys following oral dosing with radiolabelled silver nitrate indicate that silver excretion in these species follows a biexponential profile with biological half-lives of 0.1 and 1.6 days in mice and 0.3 and 3 days in monkeys. In similarly exposed rats and dogs, silver excretion followed a triexponential profile with biological half-lives of 0.1, 0.7, and 5.9 days in rats and 0.1, 7.6, and 33.8 days in dogs (Furchner et al. 1968). Data for whole body clearance of silver at two days after exposure for these four species are presented in Table 2-5 (Furchner et al. 1968). Transit time through the gut may explain some of these interspecies differences in silver excretion. Transit time is approximately 8 hours in mice and rats, and approximately 24 hours in dogs and monkeys (Furchner et al. 1968). Animals excrete from 90% to 99% of an administered oral dose of silver in the feces within 2 to 4 days of dosing (Furchner et al. 1968; Jones and Bailey 1974; Scott and Hamilton 1950). Excretion in the faeces is decreased and deposition in tissues, such as the pancreas, gastrointestinal tract, and thyroid, is increased when saturation of the elimination pathway in the liver occurs as a result of chronic or high level acute exposure to silver (see Table 2-4) (Constable et al. 1967; Olcott 1948; Scott and Hamilton 1950). "

In studies with bile-duct cannulated rats, the total average silver excretion of an i. v. -dose in the first 4 days p. a. was 72.3% of dose and almost exclusively via faeces (72.0%), whereas only 0.3% were excreted via urine. This underlines the high relevance of biliary excretion in the case of silver, being rapid and extensive in rats, with 45% of the dose appearing in bile already in the first 2 hours p. a. (Gregus and Klaassen, 1986).