Breathing in danger: how particulate matter pollution is putting the public at risk of lung cancer†

We are constantly exposed to chemicals and other agents in our environment that can influence our risk of tumorigenesis, but exactly how these factors contribute to cancer development is largely unknown. Fine particulate matter measuring ≤2.5 μm (PM2.5) from air pollution can accumulate in alveoli, contributing to inflammation and tissue damage. Despite prior correlative studies highlighting the mortality risk, there has been a historical reluctance to lower national standards for safe PM2.5 exposure. A recent publication further highlights the attributable risk of PM2.5 exposure with lung cancer – particularly in ‘never‐smokers’ with EGFR‐driven non‐small cell lung cancer. Importantly, it also elucidates a mechanistic link between PM2.5 exposure and tumorigenesis using in vivo models of EGFR non‐small cell lung cancer. © 2023 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

Particulate matter air pollution has long been associated with premature mortality [1]. Fine particulate matter measuring ≤2.5 μm (PM 2.5 ) can accumulate in alveoli, contributing to inflammation and tissue damage. Importantly, PM 2.5 exposure disproportionately affects lower socioeconomic populations [2]. Despite compelling evidence, the US Environment Protection Agency have only recently proposed a change [3] to reduce the National Ambient Air Quality Standards for annual PM 2.5 exposure from 12 to 9.0-10.0 μg/m 3 . The limitations of epidemiological studies (where the mechanistic basis for causality is not demonstrated) had been cited by government advisors as a key reason for their reluctance to lower this target since 2012 [4].
A recent paper by Hill, Lim, Weeden et al. [5] focused on the increasing incidence of 'never-smoker' lung cancerthe most common subtype being non-small cell lung cancer/adenocarcinoma in these patients. These tumours have a preponderance to occur in females and are associated with activating mutations of the EGFR gene. Unfortunately, despite an initial response with targeted therapies, most patients develop acquired resistance and there are limited subsequent effective treatment options.
In their paper, Hill, Lim, Weeden et al. [5] used epidemiological data from 32,957 patients with EGFRdriven lung cancers from four countries (Taiwan, UK, South Korea, and Canada). Their analysis supports previous studies correlating PM 2.5 exposure levels with lung cancer incidence [6]. Notably, one cohort of 228 Canadian females with 'never-smoker' lung cancer had a significantly higher prevalence of EGFR mutant lung cancer after just 3 years of high PM 2.5 exposure.
Building on the historical two-step model of tumorigenesis, whereby an insult to non-cancerous cells with a pre-existing mutation in an oncogene (i.e. EGFR or KRAS) can drive tumour formation, Hill, Lim, Weeden et al. [5] used ultra-deep sequence analysis of non-cancerous lung tissues from individualssome of whom were diagnosed with lung cancerrevealing the frequency of EGFR mutations to be 18%, positing the existence of cells primed for tumour promotion. The team also utilised transgenic immunocompetent mouse models to induce the EGFR L858R mutationthe second most common mutation found in 'never-smoker' patients with EGFR-driven lung adenocarcinoma. These mice were exposed to intratracheal administration of either control or two different doses of PM 2.5 (5 and 50 μg) for 3 weeks after EGFR L858R mutation induction. A striking dosedependent increase in spontaneous EGFR mutant preinvasive tumours was observed. These results were consistent in mice with EGFR L858R induction after PM 2.5 exposure, implying that the sequence of air pollution and oncogene induction may not matter. Further disproving prior dogma regarding the DNA-damaging effects of air pollution, whole genome sequencing revealed no enrichment of mutagenic signatures nor mutational burden in tumours from these mice.
Recent support for the 'initiation-promotion model' of cancer also comes from in vivo mutational profiling of tumours from mice exposed to known or suspected human carcinogens. In our study, we challenged the notion that most carcinogens exert pro-tumorigenic effects through direct mutagenesis [7]. Whole genome sequencing data from 181 mouse lung or liver tumours from chemically treated or exposure naïve mice revealed that the mean tumour mutation burden did not differ between spontaneous tumours and the majority of tumours post-carcinogen exposure. Indeed, only 3/20 (15%) carcinogens were definitively mutagenic, with exogenous mutational signatures having a strong transcriptional strand bias, supporting the likelihood that their activity was mediated through the formation of DNA adducts. Of note, however, most mutational signatures observed in chemically treated mice had endogenous origins. Using data from the Pan-Cancer Analysis of Whole Genomes (PCAWG) Consortium of 4,645 whole genome and 19,184 whole exome sequencing cases [8], Riva et al. [7] demonstrated associations between mouse and human mutational signatures of exogenous or endogenous origin. Referencing mouse orthologs of 299 human driver genes, orthogonal statistical approaches revealed consistent relationships between hotspot driver mutations in genes such as Fgfr2 and Braf with endogenous mutational signatures. These findings highlight the validity of in vivo models for human mutagen research and alongside other studies [9] suggest that most carcinogens might accelerate/promoterather than inducethe formation of endogenously arising mutant cell clones.
Importantly, in the Hill, Lim, Weeden et al. [5] study, the pro-tumorigenic effects of exposure to PM 2.5 appeared to be immune mediated and PM 2.5 exposure after EGFR L858R induction did not increase tumour burden in immunocompromised animals ( Figure 1). However, analysis of lungs from immunocompetent mice revealed increased levels of proinflammatory myeloid cells. Immunofluorescence of interstitial macrophages (defined by CD11b + /CD68 + co-expression) from immunocompetent mice also showed increased numbers both 24 h and 7 weeks after PM 2.5 exposure, suggesting that air pollution may trigger a sustained inflammatory response.  [5] demonstrate the role of PM 2.5 from air pollution towards tumour promotion consolidating prior epidemiological observations and providing mechanistic insights. Deep sequencing of healthy lung tissue from individuals, several of whom were diagnosed with lung or other solid organ cancers, revealed the incidence of EGFR and KRAS oncogenic mutations to be 18% and 53%, respectively. PM 2.5 may drive macrophage infiltration, mediated by IL-1β and inflammation. This process may reprogram AT2 cells with predisposing oncogenic mutations towards a progenitor-like state and synergistically promote tumour initiation together with traditional risk factors, such as smoking and socioeconomic status. Administration of anti-IL-1β therapy (canakinumab) in vivo abrogates cancer formation in mice, furthering observations from the CANTOS trial [10] using canakinumab for atherosclerotic disease.
Moreover, RNA sequencing from PM 2.5 -exposed mice revealed upregulation of genes associated with macrophage infiltration, including IL-1β. IL-1β is a proinflammatory cytokine that may mediate the reprogramming of type II alveolar (AT2) cellsthe cells of origin for lung adenocarcinomatowards a progenitor state in response to injury due to air pollution. The role of IL-1β towards tumour promotion was further functionally validated through lung organoid co-culture. AT2 cells from non-particulate matter-exposed mice were co-cultured with macrophages isolated after in vivo exposure to either control or PM 2.5 , revealing that particulate matter-exposed interstitial and alveolar macrophages can vigorously promote the formation of tumorigenic EGFR mutant AT2 clones when compared to 'naïve' macrophages.
Intriguingly, treatment of PM 2.5 -exposed mice with IL-1β antibody abrogated tumour formation. These findings are consistent with an exploratory analysis from the CANTOS phase III trial designed to assess the efficacy of canakinumab (anti-IL-1β) in secondary cardiovascular prevention [10]. Lung cancer incidence and mortality were significantly lower in patients who received highdose (300 mg) canakinumab compared with placebo. Although two subsequent phase III studies examining canakinumab in both adjuvant and metastatic non-small cell lung cancer failed to demonstrate benefit [11,12], the work of Hill, Lim, Weeden et al. [5] indicates that IL-1β's role may relate to tumour initiation rather than propagation, which suggests a possible route to primary prevention.
Several questions remain despite the comprehensive nature of this work. Random forest analysis from a UK biobank cohort that found increased lung cancer incidence with PM 2.5 exposure (hazard ratio = 1.07; 1.03-1.11; p ≤ 0.001), did not detect a deleterious effect with passive smoking or degree of smoking exposure (pack-years). These findings differ compared with prior literature [13]. The 50 μg PM 2.5 dose utilised in 'highdose' in vivo experiments may also be higher than those examined in human correlative studies. Indeed, the highest quintile of PM 2.5 exposure from the abovementioned Canadian cohort was >7.27 μg/m 3 .
Further research may also be required to determine the impact of PM 2.5 exposure in other lung adenocarcinoma subtypes. EGFR exon 19 deletions represent the most common mutation found in 59-71% of EGFR mutant lung adenocarcinomas [14,15]. In comparison, EGFR L858R mutated adenocarcinomas are associated with a significantly worse prognosis [16]. These tumours also display an age-dependent increase in incidence, particularly in 'never-smoker' patients [15]. These observations highlight the biological heterogeneity of EGFR mutant lung adenocarcinomas according to mutation type. Hill, Lim, Weeden et al. [5] have utilised orthogonal preclinical models to validate tumorigenic effects of PM 2.5 exposure in the L858R subtype of EGFR mutant adenocarcinoma. The applicability of these findings across other oncogene addicted 'never-smoker' lung cancers, such as ALK-EML4 rearranged adenocarcinoma, is of considerable biological and clinical interest.
From a broader perspective, there is a need to better understand the exact constituents from air pollution together with their respective synergistic and individual roles in disease pathogenesis. Air pollution research has thus far been predominantly limited to assessing the impacts of particulate matter based on size criteria (i.e. PM 2.5 and PM 10 ) [17]. Understanding the diverse mechanisms behind air pollution-induced pathogenesis would foster improved opportunities to intervene with public health or pharmacological interventions. However, testing such interventions in clinical trials may be limited by ethical and pragmatic factors. The results of large-scale trials may be confounded by geographical discrepancies, particularly given the disproportionate health impact of air pollution on those from lower socioeconomic groups. A population-based pharmacological trial, such as one incorporating canakinumab, may be prohibitive due to the risk of drug toxicities in an otherwise healthy population. The inclusion criteria for any such study may need to be tailored towards those with specific exposures beyond safe thresholds. A more realistic undertaking may be interventions targeting clean air through various public health levers.
Nevertheless, these findings provide impetus for a widespread global rethink to reduce the impact of air pollution and environmental exposures to probable carcinogens. In their study, Hill, Lim, Weeden et al. detail a mechanism underpinning the tumour-promoting nature of PM 2.5 , utilising multiple in vivo and ex vivo assays. These are supported by large-scale genomic and epidemiological studies in populations with and without a history of cancer [5]. Governments and scientific advisory committees worldwide now have an unequivocal clear air mandate.

Author contributions statement
AB was responsible for conceptualisation and writing (original draft, review and editing). TJ was responsible for writing (review and editing) and supervision. DJA was responsible for conceptualisation, writing (review and editing) and supervision.