Generic versus substance specific approaches
Based on regulatory provisions, like REACH, a high level of protection of humans from exposure to harmful chemicals which might compromise their health must be ensured. For this it has to be demonstrated that hazards and resulting risks are properly controlled to exposures which are considered to be safe. Within REACH, such ‘human exposure limit values' (Derived No Effect Levels; DNELs) are generally based on dose descriptors like NOAELs or benchmark doses for critical effects observed in animal studies which have to be modified to human exposure situations by applying so called assessment factors to account for species differences and uncertainties.
With respect to ‘poorly soluble particles of low toxicity' (PSP), it should be noted that per definition, they are devoid of any known specific toxicity and thus localised ‘lung overload' driven pulmonary toxicity is considered the only relevant or ‘critical' effect for risk assessment purposes. This also means that for risk extrapolation merely local effects and only the inhalation route of exposure have to be considered. From this it follows, that using the described conceptual 'adverse outcome pathway' for PSP, the avoidance of lung inflammatory responses may be used as criteria for setting a safe exposure limit.
Based hereupon, there are various approaches to establish safe ‘human exposure limit values' for PSP that can be envisaged. In this respect, beside the derivation of `substance specific no effect levels' like the ‘DNELs' under REACH, also the use of `binding no effect levels` and/or ‘indicative no effect levels' like ‘general dust limit values' may be appropriate.
Indicative / Binding OELs
EU Indicative OEL values (IOELVs) are only set for substances for which an effect threshold can be identified and such values are claimed to be purely health based. Binding OEL values (BOELVs) are set according to more pragmatic principles including a toxicological evaluation as well as issues of feasibility. As the name implies, the IOELVs are not mandatory and the member states may implement them at higher, equal, or lower numerical values in their national legislations. In contrast, the BOELVs must be implemented at the same or a lower level (i.e., providing the same or a higher safety margin).
According to the `Scientific Committee on Occupational Exposure Limits`(SCOEL), uncertainty is handled by the application of uncertainty factors, henceforth called assessment factors (AFs) in concordance with the nomenclature in the REACH guidance document (ECHA, 2008). The SCOEL guidance gives no numerical recommendations for AFs but lists a number of aspects of uncertainty that might need consideration.
Aspects which should be taken into account are inter alia nature and severity of critical effect (local or systemic), nature of PoD, dose-response considerations and known species differences (SCOEL 2009).
General dust limit values
In Europe, general dust limit values are set for so called ‘inert dusts', ‘nuisance dust' or ‘particulates not otherwise regulated' (PNOR) either by regulatory bodies including EU member states or independent scientific bodies like the German MAK Commission (MAK values).
The German workplace regulation specifies an occupational exposure limit (OEL) for so called `granular bio-persistent dusts`(GBS) of 4 mg/m3 for the inhalable fraction and 1.5 mg/m3 for the respirable fraction. However, the MAK Commission recently has lowered the MAK value of the respirable fraction (for dusts of density 1) from 1.5 mg/m3 to 0.3 mg/m3 (MAK 2012a).
For workplace exposure, official occupational exposure limits (OELs) for PSP (or GBS) may already exist. Under certain circumstances OELs and/or the underlying information used for setting the OELs can be used to derive a respective DNEL. With regard to setting DNELs specifically for dusts, the REACH technical guidance document states that "for exposure to dust, it should be considered whether a derived DNEL for inhalation may have to be lowered. However, when using existing OELs for the inhalable and/or respirable fraction of dusts, the following should be considered according to ECHA:
• For non-soluble inert dusts and if the derived DNEL for inhalation is above these dust limits, the general dust limits should apply for exposure scenarios with exposure to dust.
• For significantly soluble dusts, if the derived DNEL for inhalation is above (these dust limits), the general dust limit might apply. Where it is not to be used, the rationale for any deviation from the general dust limits should be justified.
In principle, this approach is supported, with some minor specification, by the Task Force members. For PSP, as defined in this report, general dust limits may be used as long-term DNEL for workers, unless there is some substance specific toxicity data suggesting different level(s) of toxicity. For PSPs, where substance-specific inhalation data point to potential health effects at exposure levels below those general OELs, DNELs should be derived on the basis of those substance specific data but taking into account the wealth of data suggesting that the rat is the most sensitive species to effects of PSP as a result of lung overload (Chapter 5).
Generic volumetric overload Limit Value according to Pauluhn
Published evidence suggests that repeated exposure inhalation studies on rats represent the most sensitive bioassays in regard to poorly soluble particles of low inherent toxicity and that the prevention of any overload-like condition will also prevent adverse effects to occur from secondary inflammatory responses or long-term sequelae. Considering this particular high sensitivity of rats to inhaled PSP, the use of additional assessment factors to account for inter- or intra species differences is not considered to be necessary for the establishment of long-term chronic inhalation DNEL based on rat inhalation data. This conclusion matches the deliberations of expert groups convened to address PM-related chronic toxicity and is supported by an ILSI workshop which concluded that no default uncertainty factor is required to account for quantitative ifferences in deposition, air flow patterns, clearance rates and protective mechanisms between humans and animals (ILSI 2000).
Recently, a volume based generic concentration of 0.54l PM resp/m3 was proposed as a defensible OEL based on both, generic theoretical considerations as well as empirical evidence, which allow easily the calculation of respective mass concentrations by multiplication of the volume concentration with the particle agglomerate density (Pauluhn 2010, 2011). However, a proper definition of the relevant type of density is still due (also Section 2.2.1). In addition, Pauluhn also concluded that repeated inhalation studies on rats should be performed using an experimental window of a cumulative volume load of respirable particles in the `no-adverse-effect-range` not exceeding 10 μl/lung to avoid that retention half-times of 1 year would be surpassed. According to Pauluhn (2011), inhalation studies exceeding such a threshold volume may lead to meaningless findings which are difficult to extrapolate to any real-life exposure scenario.
Human equivalen concentration (HEC)
In times of growing interest in computational toxicology, the calculation of a human equivalent concentration (HEC) for inhalable dusts by dosimetric models like MPPD is inviting (Jun Ho Ji 2012, Hartwig 2011, Pauluhn 2011). The general hypothesis behind this concept is that equivalent doses lead to equivalent effects in different species. Hence, a HEC calculated from a NOAEL could in principle be considered as a conservative OEL.
However, the necessary adjustments for deriving such a reference concentration needs careful consideration as these adjustments may influence what is considered the most sensitive organ and/or critical effect, i.e. the most sensitive endpoint may vary for different durations or routes of exposure resulting in different HECs from the same external inhaled concentration. In this respect, the adjustment for systemic effects requires the integration of a more complex set of dose-response considerations compared to adjustments for local portal-of-entry effects between animals and humans. For local non neoplastic effects, there are strong indications that equivalent exposures generally result in similar effects independent whether experimental animals or humans are considered. In these cases comparable and less complex biological feed-back mechanisms between species do allow a more reliable derivation of reference concentrations. Such generalised procedures in deriving HECs do also exist for substances eliciting their responses locally, e.g. particles effects in the respiratory tract. By applying proper dosimetric adjustment factors, respecting species-specific physiology and anatomy, to duration-corrected exposure concentrations (e.g. daily average), such human equivalent concentrations are already used to derive risk based OELs by various regulatory bodies, e.g. within the German Committee on Hazardous Substances Committee (AGS).
Although the MPPD model already considers dosimetric principles like deposition and clearance there are some points that may limit the value of a HEC calculated by the MPPD model (Kalberlah and Schuhmacher-Wolz 2011):
• Different lung tumour types and tumour locations in animals and humans indicate that different target tissues should be considered for a proper dose transformation.
• The actual regional deposition fraction may deviate significantly from the mean values taken for the different lung compartments and used for HEC calculation.
• In humans deposition and retention may be significantly influenced as they are normally not exposed to the identical particle distribution as is in the animal experiment.
• The conservative differences in the clearance mechanisms, expressed by an elimination half life time of 700 days may be already compensated by alternative mechanisms of "detoxification”, e.g. translocation into the interstitium.
Whereas the first problems are of general nature and cannot be considered easily in calculations the latter one can be adapted by using different clearance rates than the defaults. Proposals for alternative clearance rates have already been made in literature (Pauluhn 2011, Hartwig, 2011) and were considered by the German MAK commission in their recently derived MAK value of 0.3 mg/m3 for (unit density) respirable granular biopersistent dusts (`GBS`) based on the HEC concept (MAK 2012a).
Experimental case-by case approach
REACH requires endpoint specific information to identify the so called 'leading health effect' which then has to be used for derivation of the final DNEL (ECHA 2012c). Generally this information is coming from sub chronic (90-day) experimental animal studies. According to the REACH framework, every endpoint specific DNEL has to be calculated, one for each identified adverse health effect and relevant exposure route.
The OEL hazard assessment basically aims at identifying the `critical effect`, i.e., the first adverse effect that appears as dose (or exposure level) increases. The underlying assumption is that if exposure is kept below the critical effect level, neither the critical effect nor other more serious effects will appear. In contrast, in the derivation of DNELs according to the REACH framework, several endpoint-specific DNELs have to be calculated, one for each identified adverse health effect. The lowest of the endpoint-specific DNELs for each relevant exposure route is then chosen as the final DNEL and the corresponding effect being called the `leading effect`. In the case of PSPs, which by definition are devoid of any specific systemic toxicity, only the inhalation route of exposure is relevant to generate meaningful data for assessing related risks. Based hereupon and taking into account `adverse outcome pathway` principles, screening studies with focus on the 'critical health effect', namely lung inflammatory responses, may be useful to generate the necessary data for extrapolation purposes and for ensuring animal welfare principles. A conceptual outline of a respective study design allowing for model specific adjustments for testing is given in Appendix A of this document.