Nitrated polycyclic aromatic hydrocarbons: a risk assessment for the urban citizen.

Nitrated polycyclic aromatic hydrocarbons (nitro-PAHs) are formed during incomplete combustion. Sources include emissions from vehicles (mainly diesel vehicles), heating, smoking, certain types of food processing, and incomplete combustion in general. Nitro-PAHs are direct-acting mutagens, and a number of them have been shown to be carcinogens. 2-Nitrofluorene (NF) represents a model substance for the nitro-PAHs. An attempt has been made to calculate the human cancer risk due to exposure to nitro-PAHs by two different models. In the first model, genotoxic lesions were transferred to units of Gray (gamma-irradiation), and in the second model a mega study (24,000 animals) on the carcinogenicity of one metabolite of NF was used to elucidate the risk. Gamma-irradiation of the rat liver gave rise to preneoplastic foci in a dose-dependent manner, which was statistically significant. The Gray-equivalents of chemically (NF) induced foci were calculated, and from the human nitro-PAH exposure, expressed in Sievert, a human risk estimate was calculated. In the second model, an extrapolation from laboratory animals to man was performed because tumor data on 2-acetylaminofluorene (AAF), a major metabolite of NF, were available in the literature. The tumor dose-response data on AAF was linear for tested lifetime doses. The results of both models agreed, with a risk range of 0.15-49 x 10(-6) on human cancer risk for an urban citizen.


Introduction
Incomplete combustion is a major problem in terms of pollution. Examples include emissions from energy production, industrial processes, vehicles, and smoking. The biological effects of incomplete combustion can be divided into effects on human health or on the ecosystem. Both effects can be acute or long term. The different biological responses can be related to each other because the same substance in the emissions can give rise to several reactions in the organism or the ecosystem. One such example is the group of nitrated polycyclic aromatic hydrocarbons (nitro-PAHs). For the formation ofnitro-PAHs, incomplete combusted organic material (PAH) and oxidized nitrogen (NO,) are necessary. A low pH (SO2, NOx) catalyzes the reaction. NOX is one important combustion product responsible for acidification of the environment, acute health effects (1), as well as formation of nitro-PAHs (2), which are strong genotoxic agents in mammalian systems (3)(4)(5)(6)(7). Because the formation of nitro-PAHs is catalyzed by 'ON 0 FIGURE 1. The chemical structure of 2-nitrofluorene.
NF has been studied in detail in our laboratory from an analytical point of view (36), and with regard to metabolism (37)(38)(39), lung effects (40,41), and genotoxic effects (42)(43)(44). A review of NF's prevalence and biological effects has also been published (45). The chemical structure of NF is shown in Figure 1.

Results and Discussion
For the group of nitro-PAHs, epidemiology or studies on specially selected groups are more or less impossible to perform because nitro-PAHs constitute only part of the products of incomplete combustion. It is extremely hard to define a group of people that is exposed to nitro-PAHs and a control group that is exactly the same from all aspects except that it is not exposed to nitro-PAHs. Nitro-PAHs always occur with other carcinogens (combustion products), and therefore another approach must be used to elucidate the genotoxic risk.
In our studies, NF was selected as the model compound. NF is a direct-acting mutagen in bacterial test systems and is often used as a positive control in genotoxic assays. In Table 1, a summary of some of the existing data on the genotoxicity of NF is shown.  (3,73,78) a( +) Positive effect; (-) no effect.

Metabolism of 2-Nitrofluorene
Although NF is a chemically stable molecule, it is extensively metabolized in the organism. After oral administration of NF, the major part of the dose is excreted within 48 hr (37,39). After 4 hr, approximately 2% of the dose has been metabolized by the intestinal microflora, absorbed, and metabolized (several steps) in the liver, distributed in the circulation, filtered by the kidneys, and excreted in urine. The excretion of metabolites is accompanied by excretion of mutagenicity. Typically, direct-acting mutagenicity (-S9) dominates over mutagenicity in the prescence of S9, both in urine and feces (37,39).
The in vivo formation (37) of the potent carcinogen 2-acetylaminofluorene [AAF (46)] is indicated. After an oral dose of NF to rats, NF is reduced to 2-aminofluorene (AF) by the intestinal microflora, acetylated to AAF, and further hydroxylated in the liver, resulting in OH-AAF compounds, which can be excreted as such or in conjugated form. This metabolic route is quantitatively the most important. AAF has been a model compound for chemical carcinogenesis since the Wilson's early finding of its carcinogenic potential in 1941 (47). AAF has been used in a number of bioassays and has been characterized from many different points of view (48). AAF is not found in the environment, and occupational exposure can only occur when AAF is used in research. It is thus of concern when an environmental pollutant (NF) commonly found in diesel exhaust (9,(49)(50)(51) has the capacity to be metabolized to this potent carcinogen in vivo. Other nitro-PAHs have been shown to form acetylated metabolites, although the biological significance of these metabolites is not known (52,53).
After oral administration of NF, an alternative metabolic route results in the formation of OH-NFs. This is seen after induction of the cytochrome P450 system in vivo. It is also seen in isolated, perfused lung and liver as well as in germ-free animals (37)(38)(39). OH-AAFs are considered to be detoxification products (54) and have a low mutagenic potency (55). OH-NFs, on the other hand, are more mutagenic than NF itself (39). So far, nothing is known about the carcinogenic potential of OH-NFs, but it cannot be excluded that they are carcinogenic. They may, for example, be involved in the tumor formation seen in the forestomach after oral dosing of NF. No forestomach tumors are seen after administration of AF or AAF. OH-NFs may also play a role in the formation of subcutaneous tumors after skin application of NF (28,56). The metabolism of NF is shown in Figure 2.
Are Animal Data Relevant to Humans?
One can always argue whether data on animal metabolism are relevant to humans, but in the case of nitro-PAHs, there are a number of factors indicating that animal studies are relevant to the human situation: a) Reduction of nitro-PAHs to amino-PAHs can be performed by anaerobic fecal bacterial suspensions from humans as well as from rats (57)(58)(59). b) Human liver S9 bioactivates AF and AAF to mutagens (60). c) Human hepatoma cell lines can (or) ± iNF (it, iv) Lung Liver / Lung (iv) Liver perform nitroreduction as well as ring hydroxylation of NP (61). d) Liver microsomal metabolism ofAAF is similar in rats and humans (62). e) Human lymphocytes metabolize AAF to ringand N-OH derivatives of AAF (63). J) AAF metabolism is similar in cultures of epithelial cells from human and rat bladder (64). g) It has been shown that the carcinogen AAF given orally to humans results in the same urinary metabolites as in the rat (65).

Risk Calculations Using AAF
In the AAF model, combined data have been used from the metabolic pathways (NF metabolized to AAF in vivo) and a mega-study (66) performed on 24,000 animals to elucidate the carcinogenic potential of AAF at low lifetime dosing. The tumor data on the liver have been used for the extrapolation to man. The dose-response curve for tumor formation in the animal study after lifetime dosing was linear. In the calculations (67), the animal data have been extrapolated to man. The positive and negative comments that could be made regarding this model are found in Table  2. With an estimated human lifetime dose range for an urban citizen, as discussed by Moller et al. (73), the human cancer risk is p(x) = 1.1 -49 x 10-6. Weaknesses and strengths of the AAF model are discussed in Table 2

Risk Calculations Using y-Radiation Eqivalents
The y-model is totally different from the AAF model in that known human risk data on y-irradiation is the basis for the risk calculation. The genotoxic effects in the rat liver after exposure to NF occur as preneoplastic lesions (44). The same genotoxic lesions can be formed after exposure to -y-irradiation of the rat livers (67). The dose of y-irradiation was exact in terms of dose maxima, exposed area, standard deviation of the dose, etc., because y-irradiation apparatus for human cancer treatment was used (67). In this way the genotoxic lesions caused by NF could be converted to y-irradiation equivalents (Gy). The total human dose of nitro-PAHs could then be expressed in radiation units, which made it possible to use human risk data on carcinogenesis after y-irradiation exposure to large populations (68). With the same estimated life dose of nitro-PAHs as used in the AAF model, the human cancer risk for an urban citizen is p(x) = 0.30 -39 x 10-6. Weaknesses and strengths of the -y-model are discussed in Table 3.

Human Dose of Nitro-PAHs
The basis for the risk calculations is the human lifetime dose of nitro-PAHs. Based on literature data, an estimated lifetime dose range has been calculated, which is discussed in detail in M6ller et al. (67). The combined risk range is 0.15-49 x 10-6. In the dose calculations, food has been considered to be zero in terms of nitro-PAH dose, although this is not the case; food can even be the predominate source of nitro-PAHs (16,17). Cities like Berlin and Beijing represent the upper risk range, and cities like Tokyo and Kawasaki represent the lower risk range. High traffic intensity, a lot of diesel vehicles, and coal as source of energy lead to doses that pose a high risk. Some comments on the dose are shown in Table 4. In Tables 5 and 6, NF analyses in urban air and particle extracts are shown. In Table 7 the factors that could increase or decrease the calculated risk for nitro-PAHs are discussed. Although no definite risk limit has been set, a risk of 1 x 10-6 is in general is considered to be the limit for an unacceptable risk. That risk limit could be exeeded by a factor ofup to 50 times in the case of nitro-PAHs. Table 3. Weaknesses and strengths of the y model.

Weaknesses
Are foci generated from chemicals and y-irradiation in the same manner? Recent human risk data on y-irradiation indicate that the risk is underestimated.

Strengths
Well-known risk data on humans after exposure to-y-irradiation exists. Conversion of genotoxic lesions to radiation-equivalents. The y-model is an in vivo model. Are the literature data on nitro-PAH levels representative? Sampling temperatures are not given (might influence distribution gas/particle phase). Limited information available on the food levels of nitro-PAHs. Could be a major source of nitro-PAHs. Can plants (for food consumption) along roads absorb and metabolize nitro-PAHs? What is the relationship between oral and inhaled dose? Are the potent direct-acting mutagens, the OH-NFs, also carcinogens? Strengths A large number of analyses on urban air. 2-Nitrofluorene seems to be a good model for nitro-PAHs in terms of dose. An equal distribution of nitro-PAHs in the urban environment due to a very large number of sources. Abbreviations: nitro-PAHs, nitrated polycyclic aromatic hydrocarbons; OH-NF, 'Big city, many vehicles, typical western-type city, although former East Berlin has coal as a fuel for heating, which might influence the air quality in former West Berlin where the analyses have been performed.   (10). Certain combustions may generate more nitro-PAHs than vehicles [airplanes and ships (51,74)]. Cigarette smoke might be an unknown source for nitro-PAHs. Water-phase reactions (PAH + nitrite) could be an additional source for nitro-PAHs (20,21). Cooking could be a predominate source of nitro-PAHs if the Japanese data on chicken are generally applicable (17). Nitro-PAHs can act as tumor promoters (44) and/or affect tumor progression. Exposure to nitro-PAHs could lead to an increased risk of miscarriages and malformed offspring because metabolites of 2-nitrofluorene (NF) can cause malformation in laboratory animals (82). Certain other substances can function as co-carcinogens with nitro-PAHs: one example is benzo(a)pyrene (BaP) (83). Compared to other nitro-PAHs, NF, which was the base for calculations, has a low bacterial mutagenicity. Contamination of the environment of nitro-PAHs might be an addition to the lifetime dose (15). If particles are very small (<1.5 ,um), they contain more nitro-PAHs (84).
Certain risk behaviors (being a child and breathing closer to the sources of emissions for instance) are not included. Certain risk situations are not included (indoor exposure of diesel exhaust in garages, mines, ferries, etc.). Certain occupational risk behavior is not included such as cutaneous exposure of soot (workshops, chimney sweeps, etc.). Metabolism of other nitro-PAHs might be more important than NF in terms of genotoxicity. Metabolism without intestinal microflora (oxidative pathway) leads to the formation of more genotoxic metabolites than 2-acetylaminofluorene (AAF) (37)(38)(39). Local sources in the countryside could give rise to relatively high levels of nitro-PAHs (diesel tractors, for instance). The risk calculations performed on the victims from Hiroshima and Nagasaki, which is the basis for the rad equivalent calculation, can be underestimated (81). If nitro-PAHs reacts with cell components other than DNA, it is possible that other toxic reactions might occur (i.e., reactions with proteins).
Decrease risk Measured levels of NF in Berlin might not be representative of cities in general (it has been assumed in the calculation that urban levels can be as low as 10% of the Berlin concentration, which is one reason for the width of the risk range). If NF is not a good model for nitro-PAHs in general in terms of biological effects. The increased use of catalytic cleaning devices, although this is not the case for diesel vehicles for the moment (which is so far the main known source for nitro-PAHs). If animal models are not relevant to human risk. Metabolism of other nitro-PAHs is less important than NF in terms of genotoxic risk. If metabolism of NF via the oxidative pathway leads to less genotoxic metabolites than AAF. If -y-generated foci are different from chemically induced foci. If antagonism between NF and other xenobiotics reduce the promotive response.