Mutagenicity and carcinogenicity of combustion emissions are impacted more by combustor technology than by fuel composition: A brief review

Abstract Studies during the past 50 years have characterized the carcinogenicity and mutagenicity of extractable organic material (EOM) of particulate matter (PM) in ambient air and from combustion emissions. We have summarized conclusions from these studies and present data supporting those conclusions for 50 combustion emissions, including carcinogenic potencies on mouse skin (papillomas/mouse/mg EOM), mutagenic potencies (revertants/μg EOM) in the Salmonella (Ames) mutagenicity assay, and mutagenicity emission factors (revertants/kg fuel or revertants/MJthermal) in Salmonella. Mutagenic potencies of EOM from PM in ambient air and combustion emissions span 1–2 orders of magnitude, respectively. In contrast, the revertants/m3 span >5 orders of magnitude due to variable PM concentrations in ambient air. Carcinogenic potencies of EOM from combustion emissions on mouse skin and EOM‐associated human lung cancer risk from those emissions both span ~3 orders of magnitude and are highly associated. The ubiquitous presence of polycyclic aromatic hydrocarbons (PAHs), nitroarenes, and aromatic amines results in mutagenic and carcinogenic potencies of PM that span only 1–3 orders of magnitude; most PM induces primarily G to T mutations. Mutagenicity emission factors of combustion emissions span 3–5 orders of magnitude and correlate with PAH emission factors (r > 0.9). Mutagenicity emission factors were largely a function of how material was burned (highly efficient modern combustors versus open burning) rather than what materials were burned. Combustion systems that minimize kinetic and mass‐transfer limitations and promote complete oxidation also minimize the mutagenicity of their emissions. This fundamental engineering principle can inform environmental and public health assessments of combustion emissions.


| INTRODUCTION
Polluted air, water, and soil are the main environmental causes of deaths globally, and air pollution is among the primary environmental cause of deaths and disability globally (GBD 2016Risk Factors Collaborators, 2017. The primary contributor to air pollution is combustion emissions from a wide variety of sources (Russell, 2013). Combustion is a complex progression of physical and chemical processes involving the mass transfer and exothermic chemical reactions of a fuel (reductant, usually carbonaceous, such as natural gas, oil, coal, wood, gasoline, etc.) and an oxidant (usually atmospheric oxygen in excess).
These interactions result in a highly reactive flow with fast heat and mass transfer. High temperatures sustain hundreds to thousands of elementary reactions involving stable species and free radicals. Free radical chemistry is critical for the initiation, propagation, and termination of flame reactions.
Thermodynamically, for hydrocarbon fuels, chemical equilibrium overwhelmingly favors the formation of carbon dioxide (CO 2 ) and water (H 2 O) as reaction products. Unfortunately, complete combustion (full oxidation) is almost impossible to achieve due to kinetic and mass trans- Air pollution is associated with an array of environmental (IPCC, 2021) and health effects, including prenatal and children's health (Heft-Neal et al., 2022;Johnson et al., 2021), neurological effects (Kim et al., 2020;Shaffer et al., 2021), cardiovascular disease (Lewtas, 2007), and cancer (IARC, 2016). Although people are exposed to both the particulate and gas phase of air pollution and combustion emissions, most epidemiology and experimental carcinogenesis and mutagenesis studies have linked these health effects to the particulate matter (PM), especially PM that is ≤2.5 μm in diameter (PM 2.5 ). Carbonaceous components of PM are products of incomplete combustion, and as described below, the organics bound to PM are the portion of PM most associated with the mutagenicity and carcinogenicity of combustion emissions and air pollution.
The mutagenicity and carcinogenicity of the extractable organic material (EOM) from PM, especially PM 2.5 , have been studied extensively for decades, as chronicled in many reviews (Claxton, 2014a(Claxton, , 2014bClaxton, 2015aClaxton, , 2015bClaxton, , 2015cClaxton et al., 2004;Claxton & Woodall Jr., 2007;IARC, 1986IARC, , 1989IARC, , 2004IARC, , 2010cIARC, , 2012IARC, , 2014IARC, , 2016Lewtas, 1988). These studies have identified some of the main chemical classes in the PM and the mechanisms by which PM causes a range of health effects. In addition, these studies have highlighted the sources of combustion emissions and the various ways by which an array of materials are burned, leading to potential health effects.
Several general conclusions have emerged from these studies, and we have consolidated them in this review. To illustrate these conclusions, we present data on the extractable organics from PM of 50 combustion emissions, including the association between carcinogenic potencies in humans as determined by epidemiology and experimental carcinogenic potencies in rodents, mutagenic potencies in Salmonella, and mutagenicity emission factors in Salmonella for PM from ambient air and a wide array of combustion emissions. These data demonstrate several fundamental features of PM-associated mutagenicity and carcinogenicity that can inform assessments of environmental and health effects of air pollution and combustion emissions.

| CARCINOGENIC POTENCIES OF ORGANICS FROM COMBUSTION EMISSIONS AND ASSOCIATION WITH HUMAN LUNG CANCER RISK
The first report of the experimental carcinogenicity (in mice) of organics from polluted air was by Leiter et al. (1942). Since then, numerous studies have confirmed this observation in experimental animals, as reviewed by Claxton and Woodall (2007) and IARC (2016).
Based on a range of data, outdoor air pollution and PM in outdoor air pollution are Group 1 (known) human lung carcinogens (IARC, 2016).
A variety of combustion emissions that contribute to varying extents to the carcinogenicity of polluted air have also been evaluated as known human carcinogens (IARC, 2012), including diesel exhaust (IARC, 1989(IARC, , 2014, cigarette smoke (IARC, 1986;IARC, 2004), and household combustion of coal (IARC, 2010c). Emissions from the burning of biomass (primarily wood) have been evaluated as a Group 2A (probable) human carcinogen (IARC, 2010c). Figure 1 summarizes the carcinogenic potency on mouse skin of the EOM (primarily from PM 2.5 ) from a variety of combustion emissions and urban air. In this assay, the organics were painted onto the skin of mice, and after typically 6 months, the number of lesions (papillomas) on the skin were counted; thus, the carcinogenic potency was expressed as the number of papillomas/mouse/mg EOM. In nearly all of these studies, histopathological analyses showed that the papillomas were carcinomas, that is, malignant tumors, rather than benign tumors.
Coal tar is technically not a combustion emission but a liquid pyrolysis byproduct, produced by thermally induced devolatilization and partial distillation of coal in the absence of oxygen to produce solid coke. However, we have included it because it is among the complex mixtures for which there are carcinogenicity data in this assay. The carcinogenic potencies of these organic extracts span~3 orders of magnitude, with one of the most important human lung carcinogens, cigarette smoke, being the weakest among these extracts.
The Group 1 human carcinogen benzo[a]pyrene (B[a]P) (IARC, 2010a) was included as a positive control in most of these studies because polycyclic aromatic hydrocarbons (PAHs) such as B[a]P were shown to be a primary class of carcinogen in these extracts. The carcinogenic potency of this single compound extends the potency range an additional order of magnitude beyond that of the organic extracts.
In contrast to the~3 orders of magnitude range of carcinogenic potencies seen for the complex mixtures associated with combustionrelated emissions, the carcinogenic potencies of 1298 compounds in mice and rats span >10-million-fold (>7 orders of magnitude) (Gold & Zeiger, 1997). The basis for this difference is explored later in the section on PAHs; however, likely explanations are that the mutagenicity of combustion emissions and PM is largely due to the presence of just a few chemical classes, especially PAHs, aromatic amines, and nitro-PAHs. Thus, having a rather similar mix of compounds likely results in a rather similar range of mutagenic potencies.
Although cigarette smoke, outdoor air pollution, diesel exhaust, and the emissions from the indoor combustion of coal, particularly smoky coal (a low-quality fossil fuel with high volatile organic content) (Mumford et al., 1990), are Group 1 human carcinogens (IARC, 2004(IARC, , 2010c(IARC, , 2014(IARC, , 2016, the carcinogenic potencies of the organic extracts from their PM on mouse skin are not the same. Thus, the carcinogenic potency of the complex mixture of organics from smoky coal emissions is 1000 times greater than that from cigarette smoke, with the potencies of diesel exhaust and urban air in between (Figure 1). Chemical analyses have shown that these various organic extracts consist of thousands of compounds; however, bioassay-directed fractionation and other methods have indicated that much of the carcinogenicity and mutagenicity of these organic extracts are largely due to a few primary chemical classes, F I G U R E 1 Carcinogenic potency of extractible organic material (EOM) on Sencar mouse skin from a variety of combustion emissions, coal tar, and benzo [a]pyrene. Data for smokeless coal, smoky coal, pine, and benzo[a]pyrene are from Mumford et al. (1990); the remaining data are from DeMarini and Lewtas (1995). Error measurements were not provided in the cited papers except in Mumford et al. (1990), preventing an overall analysis of significant differences among the various complex mixtures or B[a]P F I G U R E 2 Association between human lung cancer risk and carcinogenic potency on mouse skin of the EOM from PM of various combustion emissions. Human lung cancer data are from Albert et al. (1983); mouse tumor potencies for cigarette smoke, roofing tar, and coke oven are from Lewtas (1993). Mouse skin carcinogenic potency for smoky coal is from Mumford et al. (1990) and was extrapolated (dashed line) to estimate human lung cancer risk. Figure  modified and redrawn from Cupitt et al. (1994) to include smoky coal, and the solid line is a linear regression. Error measurements were not provided in the cited papers except in Mumford et al. (1990), preventing an overall analysis of significant differences among the samples such as PAHs, aromatic amines, and nitroarenes (nitro-PAHs) (Hecht & DeMarini, 2019;IARC, 1986IARC, , 2004IARC, , 2010bIARC, , 2010cIARC, 2012IARC, , 2014IARC, , 2016. Figure 2 illustrates the association between the carcinogenic potencies of some of these organic extracts on mouse skin and the unit lung cancer risk in humans exposed to such organics. Both the rodent carcinogenic potency for these extracts and the human lung cancer risk span 3 orders of magnitude. The strong association between the two endpoints suggests that these complex organic extracts exhibit similar tumor potencies in both species. As noted by Cupitt et al. (1994), such an association permits the extrapolation of the mouse skin carcinogenic potency to human lung cancer risk for samples where epidemiologic studies are not possible but where air or combustion emission samples could be collected and tested for their carcinogenic potency on mouse skin.
We show an example of this with smoky coal, whose EOM has been evaluated for its carcinogenic potency on mouse skin, but for which lung cancer relative to EOM exposure has not been determined ( Figure 2). Women exposed to indoor smoky coal emissions have some of the highest frequencies of lung cancer among never smokers in the world (Mumford et al., 1987). Using the comparative-potency data in Figure 2, an extrapolation based on the mouse papilloma data suggests that the EOM from smoky coal may induce~3000 lung cancer cases per million people exposed at 1 μg EOM/m 3 of inhaled air for 70 years.

| MUTAGENIC POTENCIES OF ORGANICS FROM AIR AND COMBUSTION EMISSIONS
In this review, mutagenic potency refers to the number of mutant colonies (revertants or rev) produced in the Salmonella (Ames) mutagenicity assay per either μg of EOM, μg of PM, or m 3 of air. The mutagenicity of the EOM from the PM of ambient air was first reported in 1977 (Pitts et al., 1977;Talcott & Wei, 1977;Tokiwa, 1977) and has since been followed by hundreds of additional studies (Claxton et al., 2004;Claxton & Woodall Jr., 2007;IARC, 2016). The 1970s and 1980s also marked the first reports of the mutagenicity of a variety of combustion emissions, including diesel exhaust (Claxton, 2015a;IARC, 1989IARC, , 2014; cigarette smoke (IARC, 1986(IARC, , 2004; and combustion of coal, wood, and other types of biomass (IARC, 2010c). Reviews of hundreds of studies have shown that people with elevated levels of genotoxicity biomarkers (chromosome aberrations, micronuclei, DNA damage measured by 32 P-postlabeling or the comet assay, DNA methylation, telomere shortening, etc.) associated with exposure to ambient air pollution, especially traffic-impacted air, are at increased risk for cancer (DeMarini, 2013;Demetriou et al., 2012). These studies highlight the role of mutagenic combustion emissions to air pollution, resulting in the induction of a variety of genotoxicity biomarkers. The plastic is polyethylene plastic sheets; values are the sum of the data from the prefilter and postfilter; used plastic was exposed to pesticides and soil. emissions; Figure 3 shows a histogram of a subset of these data. Using the data from Table 1    Carbon black has low %EOM and mutagenicity, and when mutagenicity was detected, it was associated with extractable nitropyrene contaminants (Rosenkranz et al., 1980), likely due to high temperature organic reactions with fuel-bound or atmospheric nitrogen.
Using the data in Table 1, Figure 4 shows a moderate correlation  Table 1, except for cigarettes, which are 2R4F reference cigarettes (DeMarini et al., 2008) and urban air (Maselli et al., 2019). The three-stone fire and forced-draft stove are described in Mutlu et al. (2016). Error measurements were available for 100% soy biodiesel, crude oil, threestone fire, and forced-draft fire, but not for the other complex mixtures, preventing an overall analysis of significant differences among the samples 10 -2 10 -1 10 0 10 1 10 2 10 -3 10 -2 10 -1 The relatively narrow range of mutagenic potencies of EOM and PM from air or combustion emissions (1-2 orders of magnitude) contrasts with the somewhat wider range (3 orders of magnitude) of carcinogenic potencies for many of the same EOM samples (Figure 1). This is likely due to several reasons. One is because carcinogenesis is a multi-step process, with mutagenesis being only one of those steps.
Chemical components other than mutagens in these organic extracts, such as those that might alter gene expression, also play a role in the carcinogenicity of these extracts (IARC, 2004(IARC, , 2010c(IARC, , 2014(IARC, , 2016, accounting for a broader range of carcinogenic potencies relative to mutagenic potencies for these organic extracts. A second reason is the nature of the two assays; one is a bacterium, and the other is a rodent, with the two having vastly different biology, including different adsorption, distribution, metabolism, and excretion (ADME).

| PM-ASSOCIATED MUTAGENICITY EMISSION FACTORS OF COMBUSTION EMISSIONS
An emission factor is a measure of how much pollution is produced and emitted per some unit of activity. Although only a few papers have reported the mutagenicity emission factors for combustion emissions, such a metric allows comparison among diverse sources. Table 1 lists the mutagenicity emission factors for a variety of combustion emissions expressed in terms of the number of Salmonella revertants (rev) per either the mass of fuel burned (kg fuel) or, based on the calorific heating value of the fuel, the thermal energy released (megajoule thermal [MJ th ]); Figure 5 illustrates a sub-set of these data.
These two ways of expressing the mutagenicity emission factor are interchangeable (related by the measured heating values for the various fuels) and, using the data in Table 1, highly correlated (r = .96).
The details regarding the scales and technologies associated with these combustion studies were not always described in the papers; Based on data from a wide variety of combustion emissions (Table 1) This conclusion is illustrated for wood in Figure 7, which shows the rev/MJ th for wood burned six different ways, with a pellet-fueled   Champion et al. (2020); for boiler and fireplace from DeMarini and Lewtas (1995); and for forced-draft (FD) stove, naturaldraft (ND) stove, and three-stone fire (3-SF) from Mutlu et al. (2016). Error measurements were not reported for the boiler and fireplace, preventing an overall analysis of significant differences among the samples F I G U R E 8 Mutagenicity emission factors in Salmonella TA98 + S9 for polyethylene plastic burned either in a rotary kiln (DeMarini et al., 1992) or by simulated open burning (Linak et al., 1989). Error measurements were not provided in these two papers, preventing an analysis of significant differences between the samples As the transient polyethylene charge was introduced, there was a short period of time when excess oxygen from the continuous natural gas flame was sufficient to completely oxidize the vaporizing gases. This was followed by an oxygen-deficient period while evolving gases completely overwhelmed the available oxygen, and finally, as the rate of vaporization slowed, the systems became fuel-lean, and oxidation returned to near completion. During the period when oxygen dropped to zero, conditions are best described as high temperature pyrolysis, similar to processes used to produce carbon black. PM emitted during the first and third phases of this process were low because the available oxygen and temperatures were sufficient to fully oxidize the evolving gases, and although PM emissions during the second high temperature pyrolysis phase were high, they had low %EOM (see Table 1) and, thus, correspondingly low mutagenicity per MJ th . In contrast, the open burning of agricultural plastic reflected the buoyancy and natural convective forces that controlled fuel/air mixing. Excess oxygen was always available (but not necessarily where and when it was needed), temperatures were relatively low, and mixing was poor. Available oxygen was never zero, but Another way to consider this issue is to look at an example where different types of fuels were combusted by the identical technology yet produced rather similar mutagenicity emission factors. An example of this is the combustion of 100% petroleum diesel, 100% canola biodiesel, 100% waste vegetable oil biodiesel, or 100% soy biodiesel in an identical 4.3-kW diesel generator (Table 1). As described by Mutlu et al. (2015c), biodiesel fuels are composed of a relatively small set of fatty acid methyl esters (FAME) that are unique to each bio-oil. Soy-based biodiesel, for example, is composed primarily of six, C 17 and C 19 straight chain FAME species with 0, 1, 2, and 3 C═C double bonds. Petroleum diesel fuels, in contrast, are defined by a distillate fraction, and typically contain C 10 to C 22 compounds with boiling points between 130 and 400 C. Petroleum diesel fuel is a complex mixture of thousands of compounds, including normal and branched parafins, olefins, naphthenes, aromatics, and PAHs. Aromatic compounds typically comprise 20%-50% of petroleum diesel fuels. Despite the considerably different chemical compositions of these fuels, the rev/MJ th or rev/kg fuel each spanned a factor of 8Â (<1 order of magnitude). All of the 100% biodiesel fuels had mutagenicity emission factors that were less than half of that of 100% petroleum diesel (Table 1; DeMarini et al., 2019).

| CORRELATIONS BETWEEN MUTAGENICITY AND POLLUTANT EMISSION FACTORS: IMPORTANT ROLE OF PAHs
There are strong correlations (r ≥ .9) between the mutagenicity emission factors and pollutant emission factors for a variety of cookstoves burning wood (Champion et al., 2020;Mutlu et al., 2016).    (Figure 9c). However, we did not find a significant correlation between these emission factors for the emissions from the 4.5-kW diesel generator (Tables 1 and 2). This is consistent with the discussion above in that these different diesel fuels were combusted in the same diesel engine. PAH and mutagenicity values for the three biodiesel blends each vary by factors of two, and those for petroleum diesel are another factor of two larger. There is a slight trend of increasing mutagenicity with increasing PAH, but with a p value of .33, the correlation is not significant.
This behavior may be related to minor fuel effects, but our hypothesis that combustion technologies rather than fuel compositions are most important is supported by this lack of correlation. These analyses add to those discussed below involving the strain/S9 specificity of Salmonella, mutation spectra, and bioassay-directed fractionation and chemical analyses, which show that PAHs play an important role in the mutagenicity and carcinogenicity of combustion emissions. IARC, 2010b; Kucab et al., 2019). G to T base substitutions are the primary base substitutions induced by PAHs and aromatic amines (DeMarini, 2000). In contrast, the primary classes of inferred base substitutions induced in the absence of S9 by municipal waste incinerator emissions and its fractions are G to A and G to C, along with G to T (DeMarini et al., 1996). Incinerator emissions exhibit direct-acting mutagenicity (i.e., are mutagenic in the absence of S9) because they contain nitro-PAHs, which are direct-acting mutagens. Thus, these emissions were tested for mutagenicity and their mutation spectra in the absence of S9. The G to A, G to C, and G to T mutations noted above reflect the types of mutations induced by various types of nitro-PAHs, which are direct-acting mutagens in these emissions (DeMarini et al., 1996). Nitro-PAHs induce more of these classes of F I G U R E 9 Correlation between mutagenicity and PAH emission factors for (a) agricultural (polyethylene) plastic and kerosene burned in a rotary kiln (Linak et al., 1989) Tables 1 and 2 mutation than does the PAH B[a]P or the aromatic amine 4aminobiphenyl (DeMarini, 2000;DeMarini et al., 1996;Kucab et al., 2019;Levine et al., 1994).
Mutation spectra in tumors associated with exposure of people to these combustion emissions are consistent with the mutation spectra of organic extracts of these emissions in experimental systems.
For example, G to T base substitution is the primary class of mutation found in the TP53 and KRAS genes in lung tumors of women whose lung cancer is associated with exposure to smoky coal PAHs (DeMarini et al., 2001), and it is the primary mutation induced in Salmonella by a condensate of smoky coal emissions (Granville et al., 2003). Likewise, nitro-PAH mutations are also found in the lung tumors of nonsmokers (COSMIC, 2021;Zhang et al., 2021), whose lung cancer may be associated with exogenous exposures such as diesel exhaust where nitro-PAHs are a primary cause of mutagenicity (DeMarini et al., 2004;Mutlu et al., 2013). This is consistent with the general observation that a mutagen induces the same primary class of base substitution across species, from bacteria to humans (DeMarini, 2000).

| ROLE OF THE GAS PHASE ON THE MUTAGENICITY OF COMBUSTION EMISSIONS AND AIR POLLUTION
There are~12 studies on the mutagenicity of the gas phase of ambient air, most of which evaluated organic extracts of XAD or polyurethane foam (PUF) in the Salmonella mutagenicity assay (IARC, 2016).
In general, these extracts exhibited both direct and indirect (S9dependent) mutagenic activity, whereas organic extracts of most PM samples are generally more mutagenic in the presence of S9 rather than in the absence of S9. The contribution of the gas phase to the mutagenicity of air is less than or equal to that contributed by PM. No studies have evaluated the mutagenicity of semi-volatiles (as measured from organic extracts of XAD or PUF) from combustion emissions; however, Linak et al. (1989) evaluated some chemical characteristics of XAD extracts from the simulated open burning of agricultural plastic (polyethylene sheets) or kerosene, used to initiate combustion, and evaluated separately as a blank. As reviewed by Riedel et al. (2018) and Zavala et al. (2018), an additional 12 studies have evaluated the mutagenicity and mutation spectra of simulated gas phase atmospheres generated by the interaction of volatile organic compounds (VOCs) with simulated sunlight (UV) in smog chambers. Although the gas phase of air is mutagenic, all risk assessments for lung cancer from polluted air (Hamra et al., 2014;IARC, 2016) or combustion emissions such as diesel exhaust (IARC, 2014) are based solely on the PM fraction.

| CONCLUSIONS
Extensive studies have shown that combustion emissions are universally mutagenic and carcinogenic. The carcinogenic potencies on mouse skin of the EOM from the PM of combustion emissions are highly associated with the unit lung cancer risk in humans exposed to those emissions. These carcinogenic potencies span 3 orders of magnitude both in mouse and humans. Consistent with this is the mechanistic evidence from hundreds of studies showing that people exposed to ambient air pollution, especially air impacted with traffic exhaust, have elevated levels of a variety of genotoxicity biomarkers and an increased risk for lung cancer. Although chemically diverse, the universal presence of PAHs, and frequently of nitro-PAHs and aromatic amines, in PM account for much of the mutagenic and carcinogenic activity of PM-associated air pollution and combustion emissions. This is reflected in the mutation spectra of these classes of compounds and of the EOM from these emissions in experimental systems and in the mutations found in tumors in humans whose cancers are associated with exposure to these emissions (such as cigarette smoke or smoky coal emissions); the primary base substitution being G to T mutations, reflective of PAHs. Although only a few studies have examined the mutagenicity of the gas phase of air pollution, studies of real-world samples and of atmospheres generated by the interaction of VOCs with simulated sunlight (UV) in smog chambers show that the gas phase is also mutagenic. Collectively, these data provide some fundamental principles to guide environmental and public health assessments of air pollution and combustion emissions.