Technical Report 122

Specific DNEL derivation for PSPs following inhalation exposure

In general, REACH requires the establishment of DNELs also for the inhalation route of exposure following a generic scheme involving, e.g. identification of the mode of action (MoA), determination of the relevant dose descriptor and point of departure (PoD) as well as applying a set of AFs.

Step 1: Gather typical dose descriptors (e.g. NOAEL, LOAEL) and/or other information on potency (dose-response / dose-effect relationship)

Step 2: Decide on mode of action (threshold or non-threshold)

Step 3: Derivation of effect levels (DNEL in case of threshold effects / DMEL in case of non-threshold effects) or the use of a qualitative approach

Step 4: Selection of the leading health effect

In the absence of relevant data from humans (e.g. epidemiological data), animal inhalation studies using rodents as preferred animal species generally represent the basis for the calculation of such DNELs (for more details see Appendix B).

In the case of low soluble particles that are regarded to be of low toxicity and do not have specific systemic toxicity, several aspects relevant for DNEL derivation have to be considered. The default extrapolation factors to be used according to REACH were established for systemic effects, rather than for local or `portal-of-entry` effects. Another difficulty is the discrimination between "critical effects”, "leading effects", and "critical DNEL”. Quantitative effect data may be derived from comparative response data for the toxic effect itself or for a "point in the chain of events” that is considered critical to the toxic response ("biomarker of effect”). In general OEL settings, the "critical effect” refers to the first "adverse” response (or measure of response) within the MoA that occurs at a dose / concentration below which neither this "critical effect” nor any other effects of health concern will appear (EU Key Documentation, Version 7, Methodology for the Derivation of Occupational Exposure Limits, 2013). The "critical DNEL” for the "leading effect” according to REACH, however, is defined as the lowest DNEL for a given exposure pattern amongst all different endpoint-specific DNELs which have to be calculated (ECHA 2012c).

The extrapolation by simple multiplication of default assessment factors from animal derived NOAELs/LOAELs (respectively NOAECs/LOAECs) as provided in the ECHA Guidance, allows only limited deviation from this approach. However, there is increasing scientific advice to incorporate biologically and mechanistically based data into such extrapolation schemes to take account of critical effects defining "overall thresholds” (Renwick & Lazarus 1998, Dorne 2010). Mechanistic understanding of the key events and processes leading to a distinct toxicological response may be the basis of a more flexible framework of "pathway” related assessment factors.

It is now scientifically accepted that all kind of PSPs (micro-sized and nano-sized) are eliciting localised pulmonary toxicity via processes that causes oxidative stress and are pro-inflammatory in nature, and which are initiating an acute, neutrophil driven inflammatory response. Acknowledging this, all particles of low solubility and no/low inherent toxicity, independent of the particle size, are following the same mode of action. Oxidative stress is evidentially implicated in this process and experimental evidence suggests a relationship between the induction of inflammation and the severity of the oxidative insult (Nel et al, 2006). It should be noted that the mechanisms leading to an oxidative and inflammatory pulmonary status is clearly threshold related as exemplified in a number of studies (Driscoll et al, 1997, Seiler et al, 2001, Gallagher et al, 2003, Rehn et al, 2003, Seiler et al, 2004, Valberg et al, 2009). In this context, a recent APSGB/HESI workshop on how to discriminate adverse effects from normal physiological lung responses concluded that increases in alveolar macrophages/neutrophils (hyperplasia) following exposure to PSP should not be interpreted as "adverse” in the absence of other lung effects like inflammation or fibrotic tissue changes (Kirsten et al, 2012). Jones and Neef (2012) concluded that independent of the duration of the study, lung burdens of approximately 0.5 mg/g lung tissue appeared to be a "transition point” between adaptive and adverse changes in the lung which is remarkably similar to "lung overload” conditions as defined by Morrow (1988). Similar views have been expressed also by Pauluhn (2010, 2011).

DNEL derivation according to REACH defaults

According to REACH guidance, a stepwise approach is proposed for establishing endpoint-specific DNELs which requires the consideration of various key aspects like mode of action, the identification of the leading health effect, impact of species differences in sensitivity and the subsequent extrapolation of animal derived data to humans (also appendix B). Critical in this process are questions concerning the appropriateness and applicability of the assessment factors predefined by the REACH default approach for DNEL derivations.

Mode of Action and Dose Descriptors:

As discussed within this report, there is substantial evidence that poorly soluble particles of low toxicity, whether nanosized or microsized, exert toxicological relevant adverse effects via a threshold mode of action. Plausible mechanism of the observed pulmonary responses, even the tumorigenic action of high exposure concentrations of PSPs like TiO2 and carbon black in rats, can be interpreted by overloading of lung clearance capabilities accompanied by increased pulmonary inflammation and oxidative stress, cellular proliferation, secondary genotoxic events and (in rats) tumourigenesis. Hence, the derivation of DNELs based on NOAELs/NOAECs for poorly soluble (nano)particles of low toxicity is toxicologically justified.

Leading Health Effect/Point of Departure:

According to REACH, endpoint-specific DNELs for so called leading health effects have to be calculated and the endpoints carcinogenicity, mutagenicity and reproductive toxicity (CMR) are considered to be more severe than other endpoints. However, evidenced by several investigations, it is hypothesised that the reported tumourigenic results from PSP (nano)particles in rats are not a reliable basis for predicting human lung cancer risk (Valberg et al, 2009). This view is supported by experimental animal studies and epidemiology, which have failed so far to provide conclusive evidence of poorly soluble (nano)particle induced systemic toxicity (Gebel 2012). Based on the underlying mode of action, an impairment of lung clearance mechanisms with pulmonary inflammation and oxidative stress can be considered as critical (leading) health effect for which thresholds can be derived. Below this threshold neither the critical nor any other more serious effect will appear. Hence, substance specific NOAELs/LOAELs defining lung inflammation/oxidative stress as pathway defined leading health effects seem to be toxicologically justified dose descriptors.

Considering PSPs and with respect to the related lung overload phenomenon, the appropriate conversion of the respective NOAELs/ OAELs into an adequate `point of departure` (PoD) for DNEL derivation is of special importance. In fact, differences between various animal species and humans with regard to ventilation rates, respiratory volumes, airflow patterns and particle deposition and clearance characteristics are well documented. However, Pauluhn (2010, 2011) compared the larger size and higher number of human alveolar macrophages with that of rats and concluded from the resulting higher human alveolar macrophage volume that humans are six-times more resistant to attaining lung overload conditions compared to rats. Based hereupon and leveraged by earlier conclusions from an ILSI expert group (ILSI 2000) that no default uncertainty factor is required to account for quantitative differences in deposition, air flow patterns, clearance rates and protective mechanisms between humans and animals, NOAELs/LOAELs from rodent inhalation studies can be regarded as appropriate PoDs without additional conversion. However, it should be noted that epidemiology studies show that even under worst case exposure scenarios (exposures during former occupational conditions, e.g. amongst coal workers) no rat specific pathological lung overload conditions have been seen in humans.

Assessment Factor accounting for interspecies differences:

The interspecies assessment factor takes into account potential species-specific differences in the sensitivity towards adverse health effects and should cover uncertainties based on the default assumption that humans are about 10-fold more sensitive than experimental animals. It is argued that this difference in sensitivity between species mainly relates to a non-existing correlation between body weight with many physiologically functions. Based hereupon, the REACH guidance has split the interspecies factor into allometric scaling ("toxicokinetic”) and a default factor for remaining interspecies differences ("toxicodynamic”). The latter, however, is not scientifically based and seems to fit the original interspecies factor of 10.

However, several evaluations are highlighting that rats seem to be far more sensitive to pulmonary particle effects than humans or even other animal species. The estimation of interspecies differences in the retention kinetics demonstrated a 10-fold interspecies difference in lung burdens of rats exposed for 3 months and humans exposed long enough to attain steady state in lung burdens. Interestingly, the alveolar clearance rate of the human is thought to be independent from the particulate matter load for expected exposures, but the clearance rate for a rat depends on the amount of particles in the alveolar region. Hence, rats appear to be more susceptible to ‘‘overload”-related effects due to impaired macrophage-mediated alveolar clearance (Brown et al, 2005). In case of PSP induced lung effects, this contradicts the conservative default position that humans are more sensitive than animals. Taking the higher sensitivity of rats compared to humans as an "intrinsic” safety factor, it seems appropriate and justifiable to reduce the default AF for interspecies differences with regard to pulmonary responses from PSP exposure based on rat data. Comparable considerations have led already to proposals for respective reductions for interspecies variations (Vermeire 1999, Christensen et al, 2011, Pauluhn 2011). Also ECETOC (2003, 2010) suggests an interspecies assessment factor of 1 for local effects following inhalation exposure to dusts.

As PSP driven pulmonary responses are local (point-of-entry) effects and are not depending on metabolic rate or systemic absorption, allometric scaling does not apply. Additionally, a species independent generally low distribution rate of low soluble nano- or microsized particles within the organism is to be expected and thus no significant differences in the toxicokinetics have to be assumed. Additionally, the systemic toxicity of such particles and thus also the toxicodynamic differences, are regarded to be low. In this respect, ILSI (2000) proposed an uncertainty factor of 1 for both, neoplastic and non-neoplastic endpoints as sufficient to account for toxicokinetic and toxicodynamic parameters. Since the factor for remaining uncertainties (default 2.5) is mainly introduced to cover differences in the toxicodynamics, a factor of 1 for remaining species differences is also justified.

Based on the established higher sensitivity of rats compared to humans, an assessment factor of 1 for remaining species differences is also supported by the members of this ECETOC Task Force.

Assessment Factor accounting for intraspecies differences:

Based on REACH, an intraspecies assessment factor should account for the heterogeneity in the sensitivity of the human population. Variations in the individual responses may be due to genetic polymorphisms affecting metabolism, differences in toxikinetics and toxicodynamics, sex, general health status, age but also various other factors. Although for concentration dependent local irritative port-of-entry effects an interspecies and intraspecies AF of 1 may be considered sufficient, information on intraspecies variations in response to particle induced pulmonary inflammatory responses is relatively scarce. Based hereupon, ECHA has not refined the default intraspecies factors for local effects but is proposing the same as for the extrapolation of systemic effects (5 for workers and 10 for the general population). This is not in line with the recommendations from the ILSI workshop (ILSI 2000), where uncertainty factors of 1 to account for toxicokinetic and toxicodynamic parameters were considered sufficient for both, neoplastic and non-neoplastic endpoints following chronic particulate exposures. Pauluhn (2010) recently proposed an approach for (nano)particulate induced lung overload effects using a mechanistic model in which intraspecies adjustments were also not applied because such local port-of-entry effects exhibits thresholds and are not dependent on metabolism. A comparable approach was followed within the NEDO project from the Japanese National Institute AIST as adverse effects related to pulmonary inflammation are local and only dependent on the surface related alveolar deposition of particles (Nakanishi et al, 2011). ECETOC (2003, 2010) evaluated the intraspecies variability within the human population based on the data sets from Hattis et al (1987, 1999, 2002), Hattis and Silver (1994) and Renwick and Lazarus (1998) and concluded that intraspecies factors of 3 for workers and 5 for the general population seems to be sufficient for irritative local effects from (bio)soluble substances where effects are mainly driven by cytotoxicity. However, for poorly soluble particles of low toxicity the general dust limit should apply (ECETOC 2010).

A synopsis of the available data indicates that the potency of PSP to induce inflammation-related pulmonary responses due to lung overload seems to be solely related to the biokinetics rather than on PSP inherent properties. As insoluble particles of low toxicity generally lack any significant systemic bioavailablity and lung overload related findings are considered localised `portal-of entry` effects independent of any local metabolism, default factors used to extrapolate systemic toxic effects are exaggerated and scientifically not plausible. The ECETOC Task Force members of this Report therefore support the conclusion that for concentration dependent local effects, like inflammation driven portal-of-entry effects under lung overload conditions, an intraspecies AF of 1 is considered sufficient.

Assessment factor for exposure duration extrapolation:

According to REACH Guidance, default assessment factors for the duration of the exposure have to be used. A factor of 3 for extrapolation from subacute to subchronic extrapolation and a factor of 2 for subchronic to chronic exposure should be used. This in turn means, that the same default AF should be used for both systemic effects and for local `port-of-entry` responses in the respiratory tract following inhalation exposure. The view of using the same default AF concurrently was justified by a statistical analysis of technical reports from the US National Toxicology Program (NTP) on subacute, subchronic and chronic inhalation studies with locally acting substances which revealed decreases in effect concentrations by factors of 3.2 (subacute to subchronic), 2.7 (subchronic to chronic) and 6.6 (subacute to chronic) (Kalberlah et al, 2002).

However, a re-analysis of the NTP data sets carried out by ECETOC (2010) was not able to confirm these findings. Additionally, as most of the analysed data sets refer to substances not falling under the definition of PSP, the appropriateness of these default AFs seem to be scientifically not plausible and are considered to be excessively conservative. However, the Task Force members of this report were nevertheless not able to identify sufficient specific examples to analyse the influence of study duration on the NOAELs of PSPs to derive scientifically defendable AFs and recommends that the AFs should be further investigated and reconsidered as new knowledge becomes available.