Risk Factors for Lung Cancer in Never Smokers: A ... - IngentaConnect

Send Orders for Reprints to [email protected] Current Respiratory Medicine Reviews, 2016, 12, 74-117

74

Risk Factors for Lung Cancer in Never Smokers: A Recent Review Including Genetics Anne Marie Quinn*,1, William G. Newman2 and Philip S. Hasleton3 1

Department of Pathology, University Hospital of South Manchester, Wythenshawe, Manchester M23 9LT, UK 2

Manchester Centre for Genomic Medicine, St. Marys Hospital, University of Manchester, Manchester M13 9WL, UK

3

Department of Inflammation and Repair, University of Manchester, Manchester, UK Abstract: Lung cancer, a disease traditionally attributed to the effects of carcinogens from inhaled Anne Marie Quinn tobacco smoke, is now recognised in a population with either a relatively light history of smoking, or never smoking. Although second hand smoke contributes to the risk of neoplasia, the extent to which this affects different individuals varies, and probably depends on a combination of genetic, environmental, and racial factors. In addition, there are known associations of lung cancer in never-smokers with different occupational factors, such as asbestos, as well as potential risks due to naturally occurring diseases, and lifestyle factors. In recent years, targeted molecular therapies of non-small-cell lung cancer have enabled prolonged disease-free-survival times. These are based on the detection of activating mutations or translocations of genes coding for receptor tyrosine kinases, such as EGFR and ALK. They are reported usually in adenocarcinomas, and are often diagnosed in never smoking individuals or light smokers. This has supported the evolving concept of lung cancer in never-smokers as a molecularly different subtype, with its own associated clinical characteristics and histopathological features. Furthermore, genome-wide association studies have suggested that there are genetic polymorphisms, conferring an increased risk of lung cancer, which have a greater impact on the risk in never-smokers compared to smokers. This review considers the different factors that may lead to lung cancer in those considered never-smokers and, where possible, examines these in the context of relevant genetic findings.

Keywords: Never smokers, lung cancer, occupational, genetic risk factors. INTRODUCTION Smoking causes 18 forms of cancer and accounts for 30% of all cancer deaths in the United States [1]. Lung cancer (LC) is the commonest cause of cancer death worldwide, reflecting widespread tobacco use [2]. Mortality and incidence rates have historically been highest in high-income countries, particularly the United States, Australasia [3, 4] and some European countries. This pattern may be changing, since an increased risk of LC has been correlated with low socio-economic status (vide infra). The mortality and incidence are now declining, particularly in younger males and females in these affluent continents. Lung cancer has long been more common in men than women, but in many high-income countries (e.g. the US), the incidences in men and women have begun converging [5]. Over the past three decades, the incidence and mortality of lung cancer in China have rapidly increased, and now account for a third of the world lung cancer cases (600,000/ 1.8m) using recent figures. According to data from the National Central Cancer Registry (NCCR) in 2010, the crude incidence of LC in China was 46.08 per 100,000 population *Address correspondence to this author at the Department of Pathology, University Hospital of South Manchester, Wythenshawe, Manchester M23 9LT, UK; Tel: 0044 161 2912819; Fax: 0044 1161 2914809; E-mail: [email protected] 1875-6387 /16 $58.00+.00

(61.86 per 100,000 men and 29.54 per 100,000 women). In keeping with trends seen in developed countries, adenocarcinoma (ACA) has become the most predominant histological subtype of LC in China [6]. Globally the current pattern of LC occurrence reflects historic patterns of cigarette smoking; rates are low in much of Africa, as well as in Central and South America [7]. There were 1,271,724 LC deaths attributable to cigarette smoking worldwide in 2010 [8]. The World Health Organisation (W.H.O.) gives a higher figure of 1.59 million for the year 2012 [9]. The highest incidence was in the 70-74 year old age group and the lowest in the 20-24 year olds. The risks in smokers increase with the duration of smoking and number of cigarettes smoked daily, and progresssively decline following cessation, although never to the level among never-smoker(s) (NS). Lung cancer is seen in NS - at estimated rates as low as 5-10 per 100, 000 annually. There are higher rates in some populations, based on data from cohort studies [10]; the rates in smokers are up to 20-30 times higher. An estimated 86% of lung cancers in the UK are linked to tobacco smoking - 83% due to active smoking [11]. Although most lung cancers are a result of smoking, approximately 25% of LC cases worldwide are not attributable to tobacco use, accounting for over 300,000 deaths each year. These figures may appear high. In a global study by Parkin et al. it was found that 15% of LC in men and 53% © 2016 Bentham Science Publishers

Lung Cancer in Never Smokers

Current Respiratory Medicine Reviews, 2016, Vol. 12, No. 1

in women were not related to smoking [12]. In many cases, a specific cause cannot be identified. Striking differences in the epidemiological, clinical and molecular characteristics of LC arising in NS versus smokers have been identified, suggesting that they are separate entities [13-16]. As smoking decreases in the Western world, it is important to focus on the non-smoking causes of LC. Possible etiologic factors for LC among NS are given in Table 1. Table 1.

75

DEFINITION OF A NON-SMOKER A recent paper defined healthy non-smokers as subjects without a history of any airway or lung disease, breathlessness, cough, wheeze, ischemic heart disease, rheumatic disorders or a previous life-time exposure of > one pack-year of smoking [17]. This definition was taken from the paper of Johannessen et al. [18]. It is clear these subjects are not NS, as they have had direct exposure to cigarette

Occupational agents and exposure circumstances classified by the IARC Monographs programme (http://monographs.iarc.fr) as carcinogenic to humans, with the lung as a target organ (Adapted from new WHO Lung Cancer Classification). Agent, Mixture, or Circumstance

Main Industry or Use

Arsenic and arsenic compounds

Glass, metals, pesticides

Asbestos

Insulation, filters, textiles

Beryllium and beryllium compounds

Aerospace industry

Bis(chloromethyl)ether and chloromethyl methyl ether

Chemical intermediates

Cadmium and cadmium compounds

Dyes/pigments

Chromium (VI) compounds

Pigments (for textile dyes, paints, inks, plastics), metal industry, chrome plating

Coal - indoor emissions from household combustion

Fuel for heating, cooking

Coal tars

Fuel

Coal-tar pitch

Construction, electrodes

Dioxin (2,3,7,8-tetrachlorodibenzo-p-dioxin)

Chemical industry

MOPP chemotherapy (a mixture of vincristine, prednisone, nitrogen mustard, and procarbazine)

Treatment for Hodgkin lymphoma

Nickel compounds

Metallurgy, alloy, catalyst

Engine exhaust, diesel, Outdoor air pollution Particulate matter in outdoor air pollution Plutonium-239

Nuclear

Radon-222 and its decay products

Mining

Silica, crystalline

Stone cutting, mining, glass, paper

Soot

Pigments

Sulphur mustard

Chemical warfare

Talc containing asbestiform fibres

Paper, paints

Tobacco smoke, Secondhand tobacco smoking Vaping X- and gamma-radiation

Medical, nuclear Exposure Circumstances Aluminium production Coal gasification Coal-tar pitch Coke production Haematite mining (underground) with exposure to radon Iron and steel founding Painter (occupational exposure) Occupational exposures in the rubber production industry

76


smoke. In another study on chronic obstructive lung disease, a non-smoker was defined as having ˂ 5 pack-years of tobacco exposure; otherwise, the subjects were classified as smokers [19]. This figure could be over four times the previous definition. The international Burden of Obstructive Lung Disease (BOLD) study defined an ever smoker (current or former) as a person who had smoked ˃ 20 packs of cigarettes in a lifetime or ˃ 1 cigarette/day for a year [20]. No definition of a never-smoker is given in this paper. Yet another paper comes up with a further definition of a neversmoker as smoking < 100 cigarettes in a lifetime [21]. Thus providing a new definition of non- or never smoking is beyond this chapter. Such a definition would have limited utility, as many lung cancer patients will have also been exposed to passive smoke, diesel fumes or some of the other causes of ‘non-smoking’ LC discussed below. ELECTRONIC CIGARETTES (E-CIGARETTES, EVAPERS), VAPING Cigarette smoking delivers nicotine throughout the lung and is absorbed into both the systemic venous circulation, from the oropharynx and large airways, via the small airways and alveoli. The latter route of absorption generates a rapid peak in systemic arterial nicotine levels and hence rapid delivery to the brain [22]. Electronic cigarettes were invented in China in 2003 [23]. This habit strictly comes under the category of active smoking. Because non-scientific claims about e-cigarettes create confusion in public perception and people believe they are safe and less addictive, it is discussed here. The authors are grateful for several key publications, one from the National Institutes of Health (NIH) [24], and the other a policy statement from the American Association for Cancer Research and the American Society of Clinical Oncology [1]. How these devices should be regulated is unknown, as only limited peer-reviewed research exists. Regulation differs from country to country. 6,607 US adult smokers completed an online survey in March 2013. Participants viewed e-cigarette use as less likely to cause lung cancer, oral cancer, or heart disease compared to smoking regular cigarettes (all p < 0.001) [25] (Fig. 1). This may not be borne out by the facts, given time. E-Cigarettes do not combust tobacco nor contain the same number or concentrations of carcinogens and toxicants found in cigarette smoke (vide infra). E-cigarettes, a type of electronic nicotine delivery system, represent a new nicotine delivery technology. These devices deliver nicotine along with other constituents (vide infra) via an aerosol, which is inhaled. Their uptake is rapidly increasing in both adults and youths, primarily among current smokers. Introduced in the United States in 2007, e-cigarettes sales have been doubling annually and by 2013 were projected to become a nearly $2 billion industry by 2016 and $22 billion by 2022 [26, 27]. Recent limited evidence suggests among 14 year-old high Battery components

Indiacator light

Quinn et al.

school students in Los Angeles, those who had ever used ecigarettes at baseline, compared with non-users, were more likely to progress to cigarettes over the next year [28]. This preliminary data with only a small number of e-cigarette smokers therefore may conflict with Public Health England’s view that e-cigarettes have value as harm-reduction devices and tobacco-cessation interventions [26, 29]. Around 400,000 people have completely replaced smoking with electronic cigarettes [29, 30]. As nicotine is the addictive substance in tobacco cigarettes, nicotine delivery from electronic cigarettes is essential, if these products are to be effective for smoking cessation or harm reduction. There are three key elements that influence nicotine delivery from e-cigarette vapour to the human body: the nicotine content in the cartridge, which determines the amount of nicotine vaporised; the efficacy of vaporisation, which affects levels of nicotine transferred from a cartridge into aerosol; and the bioavailability of nicotine, which determines the dose and speed of absorption of nicotine from the aerosol and subsequent transfer into the blood stream and hence to nicotine receptors in the brain [29, 31]. Current ecigarettes generally deliver less nicotine per puff than conventional cigarettes [24, 31-33]. However an e-cigarette may contain up to 40 times more nicotine than a conventional cigarette and the user can compensate for the decreased nicotine per puff by employing a different frequency, depth, and intensity of puffing to obtain more nicotine [24]. E-cigarettes can harm their users by delivering toxic nicotine levels. Nicotine levels can be assessed by measuring cotinine, a nicotine metabolite and validated biomarker of nicotine uptake [34]. Two pilot studies measured saliva cotinine levels in samples from e-cigarette users and found high levels of cotinine, suggesting experienced users of e-cigarettes are able to gain as much nicotine from e-cigarettes as smokers do from cigarettes [34, 35]. Data on the effects of e-cigarettes on human physiology and health are limited in part due to their recent emergence, as well as their rapidly evolving construction and lack of standardisation [32]. In addition our understanding of ecigarettes is changing rapidly, as use increases and products evolve, making it difficult to interpret the evidence. This is because conclusions based on studies conducted a few years ago may not apply to the marketplace today [1]. A typical e-cigarette consists of a battery, a reservoir containing e-liquid, a microprocessor, an air flow sensor or activating button, and a heating element (Fig. 2 http://www. electroniccigaretteweb.com/faq/.). The e-liquid is usually a mixture of propylene glycol, glycerol, nicotine, flavourings, and other additives (which may include ethyl alcohol, stabilizers, and non-nicotine pharmacologically active compounds). The aerosols contain some toxic and carcinogenic substances, including formaldehyde, acetaldehyde, acrolein, Atomizing device

Inhaler

Liquid container

Fig. (1). Schematic of an e-cigarette. Source: http://www.electroniccigaretteweb.com/faq/.



and traces of nitrosamines. However the levels of some of these substances were 9-450 times lower than in conventional cigarette smoke and were often comparable with the amounts generated by a nicotine inhaler [35, 36]. The levels of these substances can depend on the voltage used to generate the aerosol [37]. The levels of nicotine, tobacco-specific nitrosamines (TSNAs), aldehydes, metals, volatile organic compounds, flavours, solvent carriers and tobacco alkaloids in e-cigarette refill solutions, cartridges, aerosols and environmental emissions vary considerably. The delivery of nicotine and the release of TSNAs, aldehydes brand to brand [36], and metals are not consistent across products. Phenolic compounds, polycyclic aromatic hydrocarbons and drugs have also been reported in ecigarette refill solutions, cartridges and aerosols. Varying results in particle size distributions of particular matter emissions from e-cigarettes across studies have been observed [35]. Heavy metals have also been identified in ecigarette aerosols [36, 38, 39]. Some toxic metals such as cadmium, nickel and lead are found in the vapour [36]. However, levels of these substances are much lower than those in conventional cigarettes. Regular exposure over many years is likely to present some degree of health hazard, though the magnitude of this effect is difficult to estimate. Studies are needed to assess whether the levels of toxicants in e-cigarette aerosol pose a health risk and to determine their toxicity thresholds [24]. The heating element is usually a wire or rod made from various metals (e.g., nickel, chromium or/and copper coated with silver). Rechargeable and disposable e-cigarettes are often referred to as first-generation, while tank systems and personal vaporizers are referred to as second- and thirdgeneration products, respectively [1].

77

In many devices, when a user takes a “puff,” an air flow sensor activates the flow of electricity to the heating element, which heats and aerosolizes some of the e-liquid. This aerosol is analogous to the mainstream smoke from a conventional cigarette [24, 40]. Both chromium and nickel are carcinogenic, as discussed below. Nicotine promotes LC proliferation via the alpha7-nicotinic acetylcholine receptor (alpha7-nAChR) subtype. Brown et al. envisaged continuous exposure to nicotine in squamous cell carcinoma (SqCC) of lung patients caused up-regulation of alpha7-nAChRs, which facilitated tumour growth and progression. They considered their results would be relevant to many SqCC patients exposed to nicotine via second-hand smoke (SHS), electronic cigarettes, and patches or gums to quit smoking [41]. A recent study demonstrated that cell lines (both normal epithelium and head and neck squamous carcinoma) exposed to vapour extract of e-cigarettes undergo increased rates of DNA strand breakage [42]. The second-hand mainstream aerosol exhaled by the ecigarette user may involuntarily expose non-users to the nicotine, ultrafine particles, volatile organic compounds, and other constituents released with the exhaled aerosol [38, 4345]. Substances remaining on surfaces in areas where people have used e-cigarettes may contribute to third hand exposure. For example, studies show nicotine from tobacco smoke can react with oxidizing chemicals in the air to form secondary pollutants, such as carcinogenic nitrosamines [46]. This reaction may also occur with nicotine from an e-cigarette aerosol. It is too early to know if e-cigarettes cause LC but from knowledge of some of the constituents of the vapour, it would be unwise to dismiss the possibility. Adenocarcinoma ALK HER2 BRAF PIK3CA AKT1 MAP2K1 NRAS ROS1 RET EGFR KRAS Unknown

NSCLC as one disease

Histology-Based Subtyping Others 11%

Squamous 34%

Squamous Cell Cancer

Adenoca 55%

EGFRvIII PI3KCA EGFR DDR2 FGFR1 Amp

Unknown

Fig. (2). Distribution of somatic oncogenic mutations by histology in lung adenocarcinoma and squamous cell carcinoma. With permission reproduced from “Genotyping and Genomic Profiling of Non-Small-Cell Lung Cancer” Li et al., The Journal of Clinical Oncology 31(8): 1039-49.

78


Quinn et al.

GENETIC ASSOCIATIONS

Could Lung Cancer Be a Familial Disease?

Genetic alterations of relevance to LC in NS, including known oncogenic somatic mutations, germline polymerphisms and epigenetic modifications in the form of DNA methylation patterns, are outlined in the sections that follow. For the purposes of this review mRNA expression profiles and protein expression profiles (including immunohistochemistry) are generally not included apart from specific references associated with a given alteration.

A number of publications have demonstrated an increased risk of LC in those with a first degree relative [6164]. It has been suggested that NS relatives with LC may confer a greater familial risk than smoking relatives (hazard ratio (HR) 2.48, 95% CI 1.27-4.84 versus 1.73, 95% CI 0.993.00) [62], with a possible increased risk of up to 5-fold [63]. This is supported by a recent survey of a US-based state cancer registry which compared relative risk (RR) in both distant and close relatives of LC cases (1747 smokingrelated cases and 784 non-smoking cases). Although there was a significant excess risk for relatives of both groups (p < 0.001) only the non-smoking cases demonstrated an increased RR based on comparisons of distant relatives only (p = 0.02). The fact that an elevated risk was limited to close relatives of smokers lead the authors to conclude that an environmental contribution, presumably SHS, was likely in these conditions [65]. On the other hand, contradictory results were reported from a pooled analysis of 24 casecontrol studies in the International Lung Cancer Consortium (ILCCO) [66]. After accounting for different variables including smoking status, NS were reported to have a reduced association with family history of LC (OR 1.25, 95% CI 1.03-1.52) compared to the overall analysis (OR 1.51, 95% CI 1.39-1.63) [66]. Furthermore, a large cohort study compared LC-related deaths amongst monozygotic and dizygotic twins born in the USA. The study included 15,924 male twin pairs born between 1917 and 1927 and who served in the armed forces during World War II. After 300 personyears of follow-up, it was concluded that there was no appreciable effect on LC mortality due to genetic factors [67].

Driver Mutations in Non-Small Cell Lung Cancer of Never Smokers Recently, it has been observed that there is a predisposition towards selected genetic alterations with transforming oncogenic properties in lung carcinomas of individuals designated NS. This has led to the suggestion that lung cancer arising in NS may be a different disease in terms of genetic susceptibility, prevailing risk factors, driver mechanisms and maintenance [16, 47]. Supporting evidence for this hypothesis comes from the distribution of “driver” oncogenes, i.e. genes that when mutated endow transforming, neoplastic properties to cell growth, amongst clinicopathological subgroups. For example, a recent whole genome sequencing study identified similar mutation profiles in NS of Asian and European origin, which were different from alterations detected in lung cancers of smokers from both groups [48]. LC in smokers has a higher frequency of mutations, including characteristic alterations in KRAS, TP53, BRAF, JAK2, JAK3, and the mismatch repair genes, while EGFR mutations and ALK and ROS1 fusions are detected in tumours of NS [49]. Activating mutations in the tyrosine kinase domain of the epidermal growth factor receptor (EGFR) gene are usually detected in pulmonary adenocarcinoma (ACA), and are demonstrated with increasing frequency in NS of Asian origin, compared to pulmonary ACA diagnosed in American smoking populations [16]. Although EGFR mutations have been detected in pulmonary squamous cell carcinomas (SqCCs), these are rare (cited at rates of less than 5% of SqCC) [50]. It is possible that at least some of these cases represent mixed adenosquamous carcinomas of the lung (ADSC) [51]. In addition, oncogenic translocations involving the ALK [52], RET [53], and ROS1 [54], genes have all been described in tumours arising predominantly in NS with ACA (see Table 2). The oncogene described after the Kirsten rat sarcoma virus (KRAS) is more often mutated in ACA of smokers. However the mutations described in NS are often transitions (substituting purine for purine or pyrimidine for pyrimidine), rather than transversions (exchanging a pyrimidine for a purine or vice versa) detected in smokers [14, 49, 55]. In contrast the HER2 (ERBB2) oncogene mutations are usually detected in NS with ACA, (sometimes with concurrent HER2 amplifications) [56, 57] (see Table 2). It has been suggested that BRAF mutations tend to occur in ACA of current or ex-smokers [58, 59]. However, a recent review of 5599 individuals found no significant difference in the distribution of BRAF mutations by smoking status apart from V600E mutations which were reported in 24 of 28 NS (odds ratio (OR) 0.14, 95% confidence interval (CI) 0.05-0.42, p = 0.0005) [60].

A systematic review encompassing seven twin studies, 28 case-control and 17 cohort studies concluded that a significantly increased risk of LC was present in relatives of those affected. Six of the case-control studies included data on NS and which were stratified according to the number of affected relatives; results of the analysis indicated an increasing risk in those with more than one affected family member (pooled RR of LC 1.57 (95% CI 1.34-1.84) with one relative compared to 2.52 (95% CI 1.72-3.70) for individuals with two or more affected relatives) [68]. A number of candidate genes accounting for susceptibility to familial LC are emerging. These include the regulator of G-protein 17 (RGS17) identified at the chromosome 6q23-25 locus [69]. Prior to this 6q23-25 had been identified as a potential locus for inherited susceptibility to LC by a genome-wide linkage study from the Genetic Epidemiology of Lung Cancer Consortium [70]. Three related SNPs in intron 1 of RGS-17, which influences Gprotein related signalling, were highlighted by a study including six multigenerational families with five or more affected members [69]. PARK2 is a gene that codes for an E3 ubiquitin ligase, and has been associated with Parkinson’s disease. It is also thought to act as a tumour suppressor gene [71]. A germline mutation 823C>T, which results in the missense mutation Arg275Trp in the RING finger 1 domain, has been identified in eleven families with a history of LC (OR 5.24, p = 0.009 in Caucasian individuals) [72]. Another possible tumour


suppressor gene with a role in apoptosis and DNA repair is Eyes A4 (EYA4), which can present with SNPs related to an increased hereditary risk of LC (rs7743259, rs159420, rs35689029, rs1878551, and rs2677826, p < 0.05). This correlates with proximity of the gene to a LC risk-related locus on chromosome 6q. EYA4 can also display somatic alterations including deletion, hypermethylation or reduced expression in LC [73]. A case-control study of Taiwanese NS individuals demonstrated a stronger association with EGFR mutations (OR 7.5, p < 0.001) in those with a specific family history of LC, compared to other cancers [74]. Other studies have also reported that a family history of LC is more likely in those diagnosed with EGFR mutations, usually seen in ACA and often NS [75, 76]. The Thr790Met mutation of EGFR, which confers resistance to targeted EGFR tyrosine kinase inhibitor therapy, has been reported as a germline mutation associated with inherited susceptibility to LC [77]. An analysis of a family with germline Thr790Met mutations over five generations indicated a relatively increased risk of LC in NS carriers (OR 0.31) [78]. In Japan, a family with a history of LC were found to harbour a germline mutation of HER2, Gly660Asp. The mutation leads to activation of Akt and p38, suggesting an oncogenic role [79]. Potential Impact of Genetic Polymorphisms Indirect exposure to SHS is associated with an increased risk of developing LC [80], and is linked to the duration and intensity of exposure [61]. It is likely there are additional factors, including susceptibility due to genetic polymorphisms, which alter the probability of LC development for a given individual [48]. Although this applies both to current/former smokers, of whom only 10-15% are diagnosed with LC, and to NS [81], it has been proposed that genetic factors impart a greater significance where the level of exposure to carcinogens is relatively low [82]. A single nucleotide polymorphism, or SNP, refers to a change of one nucleotide into another (for example C to A, T or G) that occurs within germline DNA with a population frequency of greater than 1%, and imparts a relatively low to moderate risk for the development of a disease. SNPs can be synonymous, or “silent” as the amino acid code remains unchanged, or alter the coding sequence (non-synonymous) [83]. Multiple investigations have been conducted into the severity of increased LC risk conferred, if any, by the presence of various polymorphisms. Many of the initial reports focused on alleles of the cytochrome P450 genes involved in phase I metabolism of polycyclic aromatic hydrocarbons present in cigarette smoke [84]. SNPs of the CYP 1A1 gene, 3801T˃C (Msp1 variant) and 2455A˃G, Ile462Val have been associated with LC in NS (ORs of 3.2, 95% CI 0.8-11.8 and 4.2, 95% CI 1.1-15.7 respectively) compared to ever smokers (ORs of 0.9, 95% CI 0.5-1.8 and 1.6,95% CI 0.8-3.4) [85]. Hung et al. conducted a pooled analysis and reported an OR of 2.99 (95% CI 1.511.91) for Ile462Val in Caucasian NS, which increased to an OR of 4.85 (95% CI 2.03-11.6) in ACA (see Table 3) [86]. The glutathione S-transferases catalyse phase II reactions during the metabolism of polycyclic aromatic hydrocarbons. A meta-analysis of the glutathione S-transferase pi (GSTP1)


79

polymorphism 313A˃G, which results in the substitution Ile105Val, reported an association between LC and NS, but this was significant only in Asian populations (OR 1.52, 95% CI 1.09-2.13) [87]. The valine amino acid substitution results in a reduced level of conjugation by the enzyme [87]. It is a reasonable hypothesis that SNPs reducing the functionality of DNA repair enzymes could lead to an increased or decreased risk of neoplasia as a result of exposure to carcinogens causing damage to DNA. In keeping with this, DNA repair mechanisms that operate at a suboptimal level have been proposed as risk factors for LC [88]. Polymorphisms of the nucleotide excision repair enzymes excision repair cross-complimentary group 1 and 2 (ERCC1 and ERCC2), have been investigated, with varying results based on smoking status [84]. Both of these genes code for nucleotide excision repair enzymes involved in the removal of larger DNA adducts that can interfere with normal replication and transcription [89]. A recent metaanalysis observed that the ERCC1 variant 19007T>C (rs11615) led to an increased risk of LC (homozygous CC versus TT, OR 1.24, 95% CI 1.04-1.48), however this association disappeared on removal of a single study. Stratification of the data by smoking status indicated that the risk was increased in NS (OR 2.39, 95% CI 1.47-3.88) [90]. ERCC2 is also known as the xeroderma pigmentosum complementary group D gene (XPD), and is located on chromosome 19q13.3. Results have been conflicting regarding the role of ERCC2 SNPs Asp312Asn and Lys751Gln, however both have been associated with an increased susceptibility to LC in NS, relative to smokers (OR 1.46, 95% CI 1.09-1.95, p = 0.01 and 1.57, 95% CI 1.192.08, p = 0.002 respectively) [91]. A separate study in China also demonstrated an increased risk in NS for the Asp312Asn variant (OR 1.46, 95% CI 1.095-1.948, p = 0.01) [92]. The base excision repair enzymes such as X-ray crosscomplementing group 1 (XRCC1) are responsible for the repair of chemical alterations of nucleotide bases that could lead to mutagenic events [89]. A case control study of over 1,000 Caucasian individuals with LC detected a relatively increased risk for NS harbouring the Arg399Gln polymorphism of XRCC1 (exon 10). While the OR for heavy smokers of > 55 pack years was 0.5 (95% CI 0.3-1.0), the risk was found to increase as cigarette pack years decreased, with an OR of 2.4 observed in NS (95% CI 1.2-5.0), p = 0.22 heterozygous and p < 0.01 homozygous [93]. This variant has also been an increased risk of developing p53 mutations in LC of NS (p = 0.03) [94]. The risk of LC was decreased in Chinese NS carrying the AG/AA genotype of exon 9 variant Arg280His XRCC1 (OR 0.38, p = 0.005) [95]. Studies have also been conducted of the mismatch repair enzymes responsible for correcting inappropriate nucleotide insertions made by DNA polymerase during replication. Amongst NS of a Taiwanese study there was an increased risk for the GG genotype of -93 (A to G, rs1800734) of the promoter region of mismatch repair gene MutL homolog 1 (MLH1), (OR 1.64, 95% CI 1.10-2.44, p = 0.013) [96]. This group did not detect an association with variants in the homologue of the bacterial MutS protein MSH2 [96]. However, an earlier study from Korea indicated an effect of MSH2 polymorphisms on ACA of NS in younger individuals (p = 0.028) [97].

80


A meta-analysis of four GWAS conducted on populations of European descent identified two rare alterations (with an allele frequency of less than 0.5%) in BRCA2 and CHEK2 [98]. Both influenced the risk of developing non-small-cell lung cancer (NSCLC). The mutation in BRCA2 (rs11571833, 9976A>T OR 2.47 p = 4.74 x 1020; 95% CI 2.04-3.00) causes loss of the C-terminal domain (Lys3326*). Despite an increased risk for LC, this particular SNP has not been associated with breast or ovarian cancer risk. The variant rs17879961 (470T>C, leads to a missense mutation (Ile157Thr) in the CHEK2 kinase which can stimulate cell cycle arrest and apoptosis. Although this was found to result in a reduced risk of LC (OR 0.38, 95% CI 0.29-0.49) an increased risk was associated in a subset of NS (p = 0.05) [98]. Polymorphisms of the tumour protein p63 (TP63) gene, a transcription factor that functions as a subunit of p53, have been proposed as risk factors for LC within the NS population. This is somewhat surprising as p63 expression by immunohistochemistry is often used to confirm the diagnosis of SqCC, generally a disease of smokers. A recent publication based on NS of the Female Lung Cancer Consortium in Asia (FLCCA) reported that the TP63 variant rs4488809 (rs4600802) had a significant interaction with the indoor use of coal for heating and cooking, with presumed inhalation of related fumes (p = 0.04). The same study associated HLA Class II allele rs2395185 with a risk of nonsmoking related LC in those burning coal indoors (p = 0.02) (vide infra on occupational lung disease) [99]. Pooled data from studies within the FLCCA indicated that the use of coal, (OR 1.3) or any solid fuel (including coal, wood or biomass) (OR 1.2) was linked with a 30 % increased risk of lung cancer (OR 1.3) [99]. Previously this group had conducted a genome-wide association study (GWAS), and identified novel susceptibility loci 10q25.2, 6q22.2, and 6p21.32, with confirmation of two previously identified loci at 5p15.33 and 3q28 [100]. Multiple SNPs of the TP63 locus, identified at chromosome 3q28, have been associated with the risk of both ACA (rs10937405, rs17429138, and rs4396880), and SqCC (rs10937405 and rs439680). However, these data were not stratified by smoking status [101]. In a cohort of female NS from Korea SNPs of TP63 (rs7631358, G>A) and also colony-stimulating factor 1 receptor (CSF1R) (rs10079250, A>G (His362Arg)), were associated with increased risk of LC [102]. CSF1R is also known as macrophage colony stimulating factor receptor and is located at chromosome 5q32. Silico modelling suggested the positive surface charge resulting from the His362Arg substitution could increase the interaction of the receptor with CSF1 [102]. Other potential candidate SNPs cited specifically in NS include rs2352028 of glycipan 5 (GPC5), detected by a genome-wide association study of NS conducted in the USA. This variant of the 3q31.3 locus was associated with an OR of 1.46, and subsequently shown to reduce the level of transcription [103]. Glycipan 5 is a member of the heparan sulphate proteoglycan that probably mediates transmembrane signalling at a cellular level [104]. It may function as a tumour suppressor gene and its expression by immunohistochemistry is reduced in NSCLC compared to normal lung tissue [105]. Recently, a metaanalysis reported the telomerase reverse transcriptase (TERT) gene polymorphism rs2736100 was linked with risk of LC

Quinn et al.

(OR 1.14-1.19) [106]. Telomerase reverse transcriptase functions in the addition of DNA sequences to chromosomal ends, maintaining genomic integrity. An earlier case control study of NS in the UK examined the effects of TERT, CLPTM1L, BAT3, and CHRNA3 polymorphisms [107]. Both TERT and CLPTM1L, which codes for a transmembrane protein with anti-apoptotic effects, are identified at chromosome locus 5p15.33. Results indicated that TERT rs2736100 (OR 0.78, 95% CI 0.63-0.97, p = 0.02) and CLPTM1L rs4975616 (OR 0.69, 95% CI 0.55-0.85, p = 7.95 x 10-4) were significantly associated with LC in NS. There was no specific association for NS between 6p21.33-BAT3 (rs3117582), or nicotinic cholinergic receptors subunit variants15q25.1-CHRNA3 (rs8042374) or 15q25.1-CHRNA3 (rs12914385) [107]. DNA Methylation Patterns in Never-Smokers with Lung Cancer Cytosine-phosphate-guanosine (CpG) islands, found in over 50% of human gene promoters, are unmethylated in normal cells [108]. The hypermethylation of cytosine in CpG islands of promoter regions of genes is a mechanism with the potential to “silence” the expression of genes by inactivating transcription. This is of particular importance in the defunctioning of tumour suppressor genes, and has been implicated in the aetiology of LC [109]. Conversely the remainder of the genome is methylated, preventing genomic instability and aberrant transcription potentially inducible by hypomethylation [108]. Experiments on rodents revealed variations in the incidence of methylation profiles of the ER gene promoter in response to exposure to different carcinogens, which included a tobacco derivative and X-radiation [110]. This suggests that patterns of promoter hypermethylation could be different in LC of NS compared to LC of ever smokers. The same study reported promoter methylation at the ER locus in 4 of 11 tumours from NS (36.4%) compared to 7 of 35 tumors from smokers (20%, p < 0.001) [110]. Methylation of the promoter region of O(6)-methylguanine-DNA methyltransferase (MGMT) was reported by one group as being significantly higher in NS than smokers with ACA (66 versus 47%, respectively p = 0.02), and also linked with increased tumour stage [111]. MGMT removes adducts from guanine and functions as a DNA repair enzyme that protects cells from the carcinogenic effects of alkylating agents. However, there were contrasting results of increased frequency of methylation of MGMT (OR 3.93) in LC of 81 smokers compared to 41 NS; this trend was also identified for p16 (OR 3.28) [112]. A study of approximately 600 cases of NSCLC revealed higher rates of methylation of promoters of p16 (p = 0.007), and APC (p = 0.0002), in ever smokers compared to NS [113]. Based on these trends of increased methylation in the presence of cigarette smoke it has been suggested that the methylation status of genes such as p16 can be used as an indicator of exposure to environmental tobacco smoke in NS [114]. The promoter regions of the nuclear retinoid X receptor genes RXRA, RXRB, and RXRG are methylated with varying frequencies in NSCLC. These steroid hormone receptors are thought to influence tumour growth and prog-



ression. Methylation of RXRG was associated with a worse survival specifically seen in NS but not smokers (p = 0.003) [115]. A study of 298 tumour samples reported that hypermethylation of the promoter region of Homeobox protein A9 (HOXA9) lead to reduced progression-free survival in NS (p = 0.01) [116]. However, methylation of Homeobox protein A5 (HOXA5) was reported as being more frequent in NSCLC of smokers than NS (p = 0.01), in Stage I disease [117]. Genetic Alterations Associated Subtypes of Lung Cancer

with

Histological

A proportion of the genetic and epigenetic studies to date have focussed on current or former smokers with LC, with the result that the majority of the included histological subtypes are SqCC. This is explained by the stronger relationship of SqCC, and small cell carcinoma (SCLC), with cigarette smoking than ACA and large cell carcinoma [61]. (See Fig. 2 for an overview of the distribution of common oncogenic somatic mutations reported in ACA and SqCC.) Studies which have focussed specifically on LC in NS have identified some variants specifically linked with individual histology in this group. For example, the pooled analysis by Hung et al. comparing LC in 302 NS with 1631 controls identified an increased risk for the 2455 A˃G, Ile462Val variant of CYP 1A1 in ACA (OR 4.85, 95% CI 2.03-11.6) compared to all histological types (OR 2.99, 95% CI 1.51-5.91). The ACA represented approximately half of the total number of LC cases [86]. The risk of developing ACA in female NS in China was increased in the presence of XRCC1 Gln399Gln genotype (OR2.62, 95% CI 1.44-4.79) and XRCC1 -77 combined TC and CC genotype (OR1.85, 95% CI 1.19-2.86) [118]. ACA has been significantly associated with LC in NS when compared to SqCC [13]. Worldwide a relative increase in proportions of pulmonary ACA has been observed [119, 120]. This is attributed at least in part to an increasing number of former, rather than current smokers, and a general transition to filtered or low-tar cigarettes [61]. The majority of the oncogenic driver mutations highlighted in NS in recent years have been detected in ACA (see Table 2). These include EGFR mutations often seen in ACA with a lepidic or papillary pattern of growth [121, 122], and ALK and ROS1 translocations often diagnosed in ACA with a signet ring cell component or cribriform morphology [123]. To date, CD74-NRG translocations have been reported in invasive mucinous ACA the majority of which have been reported in NS [124-126]. OCCUPATIONAL EXPOSURES TO FIBRES, FUMES AND METALS The International Agency for Research on Cancer (IARC) devised a rating system, based on animal and human data, by which they assign an agent, mixture, or exposure circumstance to one of five categories, ranging from group 1 (agent is carcinogenic to humans) to group 4 (agent is not carcinogenic to humans). Group 1 pulmonary carcinogens include arsenic, asbestos, beryllium, bis(chloromethyl) ether,

81

cadmium, chromium (IV), mustard gas, nickel, radon, and silica. Asbestos Although asbestos has been banned in many countries for years, approximately 125 million people continue to work with asbestos worldwide, with its use growing in developing countries [157]. Urban renewal projects involving mining, manufacturing, and handling asbestos-containing products have been largely responsible for maintaining the asbestos market [158-161]. According to the 2012 Mineral Commodity Summary [162], five countries accounted for an estimated 99% of the world’s asbestos mining production: the Russian Federation (1 million metric tons), the People’s Republic of China (400,000 metric tons), Brazil (270,000 metric tons), Kazakhstan (210,000 metric tons), and Canada (100,000 metric tons). Annual asbestos consumption in China is 0.5 million tons and nearly 14 million tons of chrysotile have been consumed since 1960, placing an estimated 1 million workers at risk for mesothelioma and lung cancer. Engineering controls and personal protective equipment use are unenforced and a large proportion of workers exceed the government-imposed occupational exposure limit of 0.8 fibres/mL for an 8-hour time-weighted average [163]. Since the US Geological Survey (2008) asbestos mine production in metric tons has increased in the Russian Federation, China, and Brazil [164]. A US Geological Survey trend report demonstrated that asbestos consumption also increased in China, India, Kazakhstan, and the Ukraine [158, 161]. In particular, Uzbekistan had an estimated near doubling of asbestos consumption in 2007 compared with 2003 [158, 165]. There is a strong association between asbestos exposure and an increased risk of lung carcinoma. This association was first published in 1935 [166, 167] and confirmed epidemiologically in 1955 [168, 169]. All types of commercial asbestos fibres have been implicated and a dose response relationship has been established [170, 171]. Asbestos is classed as a Class 1 carcinogen by the IARC [172]. There is also an interaction between asbestos and cigarette smoking, which increases the LC risk. There are differences in fibre types with respect to potency in the production of LC, and the role of asbestos dose vs asbestosis as the underlying aetiologic factor in the development of the lung carcinoma are debated (Fig. 3A-C). Selikoff et al. first reported the synergistic effect between asbestos and cigarette smoking in the causation of LC in a cohort of insulation workers [173]. Some studies confirmed this synergistic relationship, although a few have reported additive or supra-additive (i.e., greater than additive but less than synergistic) effects [170]. Hodgson and Darnton showed amphiboles (amosite and crocidolite) were between 10 and 50 times more potent than chrysotile in the production of LC on a fibre per fibre basis [174]. The pros and cons of this discussion have been presented in detail elsewhere, and have included the fundamental disagreements between experts on the criteria for the diagnosis of asbestosis, laboratory-specific variance in fibre burden analyses and criticisms of the imprecision of

82


Table 2.

Quinn et al.

Oncogenic driver alterations in lung cancer listed by associated histology and smoking status.

Gene

Mutation

Associated Histology and Epidemiology

Associated Smoking History

Comment

EGFR

Mutations

5 [127] to 64% [128, 129] ACA

NS or light smoking history [127, 130]

Frequency of ACA mutations varies according to population examined [127], e.g. 64% Asian [128] 26% Latino [130] 10 [129] to 18% [131] Caucasian 5% African-American [127]

KRAS

Mutations

4 [128] to 42% [131] ACA

Smokers [127, 132]

Mutation rate varies with population examined, e.g. 42% Caucasian 20% Black 0% Asian [131]

HER2

Mutations

Up to 4% ACA [57, 128, 133]

NS [57]

HER2 amplification was an unfavourable prognostic factor in those with HER2 mutation [57]

BRAF

Mutations

2% ACA [58]

Smokers [58]

V600E mutation accounts for over 80% of all reported mutations [58]

MAP2K1

Mutations

< 1% ACA [134, 135]

Smokers [135]

NRAS

Mutations

G Ile462Val

CYP1A1 [86]

GSTP1 [87]

ERCC1 [90]

Glutathione Stransferase P1 phase II metabolic enzyme

Nucleotide excision repair

Ile105Val

rs3212948 G>C

0.001 Meta-analysis of 11 studies;

rs11615 T>C rs11615 T>C never smoking versus smoking

rs3212986 C>A

8,215 Four of these 11 had data on smoking history indicating increased risk inNS

1,151

Meta-analysis of 6 studies


rs3212961 A>C

rs2298881 C>A

Homozygous: OR 1.11, [0.91-1.36] Heterozygous: OR 1.03, [0.94-1.13]

6,639

0.53 Meta-analysis of 4 studies 1,770

4,653

83

84


Quinn et al.

(Table 3) contd…..

Gene

Function

Single Nucleotide Polymorphism

Odds Ratio (OR), [95% Confidence Interval]

p Value

Type of Study

No. of Cases

No. of Controls

ERCC2 [91]

Nucleotide excision repair

rs1799793 G>A

Homozygous: OR 1.20, [1.05-1.36] Recessive: OR 1.20, [1.06-1.35]

0.006


7,121

8,962

rs1799793 G>A never smoking versus smoking

Dominant: OR 1.46, [1.09-1.95]

0.01

rs1799793 G>A

Homozygous: OR 1.023, [0.824-1.270]

0.838


8,719

11,382

Heterozygous:OR1.003, [0.936-1.074]

0.942

Dominant: OR 1.013, [0.949-1.082]

0.697

Recessive:OR1.033, [0.841-1.270]

0.755

rs1799793 G>A never smoking versus smoking

Dominant: OR1.460, [1.095-1.948]

0.01

rs13181 A>C

Homozygous: OR 1.31, [1.17-1.46], Heterozygous: OR 1.11, [1.04-1.19] Recessive: OR 1.23, [1.11-1.37] Dominant: OR 1.15, [1.08-1.23]

< 0.00001


8,396

10,510 Not stated

rs13181 A>C never smoking versus smoking

Dominant: OR 1.57, [1.19-2.08]

0.002

rs25487G>A

OR 1.3, [1.0-1.8]

Asp312Asn

ERCC2 [92]

0.004

Asp312Asn

ERCC2 [91]

Lys751Gln

XRCC1 [93]

Base excision repair

0.003 < 0.00001 < 0.00001

Not stated

Case control, Caucasian

1091

1,240

Arg399Gln

XRCC1 [95]

NS versus heavy smokers (≥ 55 pack years)

OR 2.4, [1.2-5.0] versus OR 0.5, [0.3-1.0]

0.22 heterozygous, A

Dominant: OR 0.38, [0.19-0.75]

0.005

Case control, China

247

253

Homozygous (GG versus AA): OR 1.64, [1.102.44]

0.013

Case control, Taiwan

730

730

Combined haplotype (TGGT+TGCT) with no risk allele was associated with a significantly decreased risk of ACA in NS:OR 0.44, [0.23-0.85] with no significant effect on the risk of ACA in smokers: OR 0.74, 0.491.12]

0.028

Case control, Korea

432 (141 ACA)

432

Arg280His never smoking versus smoking MLH1 [96]

Mismatch repair

rs1800734 A>G −93, promoter never smoking versus smoking

MSH2 [97]

Mismatch repair

IVS10+12A>G and IVS12−6T>C



85

(Table 3) contd…..

Gene

Function

Single Nucleotide Polymorphism

Odds Ratio (OR), [95% Confidence Interval]

p Value

Type of Study

No. of Cases

No. of Controls

TP63 [99]

Subunit of p53 tumour suppressor

rs4488809 (rs4600802) T>C

OR 1.20, [1.13-1.28]

4x10-9

Pooled data from 3 case control studies within the Female Lung Cancer Consortium Asia, all female NS

1,731

1,349

rs7631358G>A

adjusted OR 1.22, [1.07-1.39]

0.002

Case control, Korea (2 stages) all female NS

181 discovery 596 validation

179 discovery 1194 Validation

5x10-6

Pooled data from 3 case control studies within the Female Lung Cancer Consortium Asia, all female NS

1,731

1,349

adjusted OR 1.22, [1.07-1.40]

0.003






TP63 [102]

HLA class II [99]

Human leukocyte antigen class II

rs2395185 G>T

OR 1.16, [1.09-1.23]

CSF1R [102]

Macrophage colony stimulating factor receptor

rs10079250A>G

CIR1 [102]

Corepressor interacting with RBPJ1

rs13009079T>C

adjusted OR 0.80, [0.680.93]

0.003


GPC5 [103]

Glypican 5

rs2352028 G>A

LC in NS OR 1·46, [1·26-1·70]

6 × 10-6

Genome-wide association study, USA all NS

His362Arg

377 377 (first stage) (first stage)

LC in NS ACA OR 1·39, [1·16-1·66]

3×10-

rs2736100G>T

OR 0.78, [0.63-0.97]

0.02

Case control, UK all NS

239

553

CLPTM1L

rs4975616 A>G

OR 0.69, [0.55-0.85]

7.95 x 10-4


239

553

BAT3

rs3117582 A>C

OR 1.06, [0.78-1.45]

0.7


239

553

rs8042374 A>G

OR 1.05, [0.82-1.34]

0.7


239

553

rs12914385 C>T

OR 0.92, [0.74-1.15]

TERT

CHRNA3

Telomerase reverse transcriptase

nicotinic acetylcholine receptor (CHRN) gene subunit

0.47 *Removal of one study examining effects among smokers caused the association to be lost (p = 0.338). Never-smoker(s), NS; Adenocarcinoma, ACA.

The risk expressed as an attributable proportion due to asbestos among NS has been estimated at approximately 3040% [183]. These authors found the association between asbestos exposure and LC risk is basically linear, but may level off at very high exposures. The RR for LC increases between 1% and 4% per fibre-year (f-y)/mL, corresponding to a doubling of risk at 25-100 f-y/mL. However, one highquality case-control study showed a doubling at 4 f-y/mL. Although the RR of lung cancer appears higher in never and ex-smokers than in current smokers those who both smoke and have been exposed to asbestos have the highest risk. Workers were exposed to crocidolite at the Wittenoom asbestos mine in Western Australia [184]. There were three cases of LC, out of 500 NS in their series. Interestingly, their criteria for definition of a NS are not given in the paper. In addition the smoking status participants who responded only to the 1979 questionnaire (the study period ran until 2000) was assumed not to have changed throughout the study. It is possible that some of these subjects resumed smoking.

A Chinese study of chrysotile asbestos workers showed a clear exposure response trend with asbestos exposure level and LC mortality in both smokers and non-smokers. The greatest risk was observed in both smoking and asbestos exposure combined (adjusted HR 17.35, 95% CI 2.38126.57). A clear exposure response gradient was seen not only in the smokers but also in the non-smokers. This implies that smoking was not the only explanation for the difference in LC mortality between the two cohorts [185]. Para-occupational exposure to asbestos is more of an issue in mesothelioma than lung cancer. There are suggestions of raised LC rates among household contacts of asbestos workers and among individuals exposed to erionite [186]. Erionite, a mineral series within the zeolite group, is classified as a Group 1 known respiratory carcinogen. This designation resulted from extremely high incidences of

86


(A)

(B)

(C)

Fig. (3). (A) Asbestosis with lower lobe predominance and a pleural plaque in the interlobar fissure. (B) A small lung cancer with a background of asbestosis (C) histology of asbestos bodies with a clear central core and a coating of iron.

Quinn et al.

mesothelioma and an increased incidence of LC discovered in three small villages from the Cappadocia region of Turkey, where the disease was linked to environmental exposures to fibrous forms of erionite. Natural deposits of erionite, including fibrous forms, have been identified in the past in the western United States. Until recently, these occurrences have generally been overlooked as a potential hazard. In recent years, concerns have emerged regarding the potential for environmental and occupational exposures to erionite in the United States, such as erionite-bearing gravels in western North Dakota mined and used to surface unpaved roads. As a result, there has been much interest in identifying locations and geologic environments across the United States where erionite occurs naturally [187]. Out of a total of 45 deaths between 2000 and 2012, 14 deaths correspond to different neoplasms of the lung. Annual age-standardised mortality rates per thousand due to LC in a rural village of central Mexico (age >20 years) were 7.09 for males, and 4.75 for females, respectively. Erionite fibres were found in exposed rocks and soils, which can easily become airborne and be carried into streets and recreational areas near schools and homes. Other carcinogenic elements and minerals were found only in trace amounts, except for quartz dust and asbestos (chrysotile) cement sheeting, which were also present in the neighbouring villages. These results indicate that environmental exposure to erionite is an important cause of the high rates of LC mortality in the Village of Tierra Blanca, supporting previous similar reports for people exposed to erionite fibres in villages in Turkey [188]. Environmental exposure to asbestos occurs naturally (i.e. not in the region of asbestos manufacturing or asbestos mines or depots (such as in South Africa) due to close proximity to naturally occurring asbestos. This mineral presents a health threat to individuals living in certain countries bordering the Mediterranean Sea, regardless of direct occupational exposure. The geological uplifting of water-submerged oceanic plates beyond sea level bring in serpentine (including chrysotiles) asbestos, providing asbestos-containing rock and soil (ophiolites) [189]. Lung cancer incidences in Hekimhan in Malatya Province, Eastern Anatolia, were nearly 1.3-fold higher than the general population of Turkey [190]. None of the subjects revealed occupational exposure to asbestos. In another paper from Turkey, the risk of LC was 2.19 times more in subjects living ‘downtown’ [191]. There are confounders in Turkey, since some areas contain radon and biomass fuel is used in some villages for heating and cooking, as well coating the walls of the house with white stucco, containing asbestos. The mechanisms by which asbestos causes LC are not fully understood [192]. These authors identified oxidative stress a key molecular mechanism. Alveolar cell apoptosis, an important early event in asbestosis is mediated by mitochondria and p53-regulated cell pathways. This may be modulated by the endoplasmic reticulum. Oxidants have long been implicated in the activity of crocidolite and amosite, the most potent types of asbestos. The high iron content of these asbestos types appeared to be critical to the genesis of reactive oxygen species (ROS), including the highly DNA-damaging hydroxyl radical (OH•). In addition, it has been demonstrated that H2O2, the superoxide radical


(O2‾), and reactive nitrogen species are released from several types of asbestos fibers in cell-free solutions or in cells, especially alveolar or peritoneal macrophages, after phagocytosis of asbestos fibres in vitro or after inhalation. These reactive species may initiate cell signaling events both externally and within cells and act in a dose-response fashion to induce cell proliferation and injury [193]. While genomic alterations are an essential feature in the causation of cancer, large chromosomal deletions by asbestos appear to be clastogenic and associated with cell death [194]. Copy number alterations of chromosomal loci 19p13, 2p16, and 9q33.1 were significantly associated with LC tumours of individuals with a history of asbestos exposure, compared to those without. In addition the pulmonary asbestos fibre count was related to allelic imbalances and copy number alterations and correlated in a dose-dependent manner (p < 0.001) [195]. This Finnish study also demonstrated chromosomal allelic imbalances in pulmonary epithelium adjacent to LC tissue in asbestosexposed individuals [195]. At the level of single nucleotide variants, polymorphisms of intron 7 of TP53 (rs12947788 and rs12951053) were detected at a higher frequency in asbestos-exposed NSCLC (p < 0.001) than in unexposed patients or those with malignant mesothelioma (p = 0.046) [196]. Attempts to define asbestos-related SNPs associated with LC risk have been inconclusive when directed to the phase II metabolic enzymes GSTM1 and glutathione S-transferase T1 (GSTT1). A pooled analysis of 5 GSTM1 and 3 GSTT1 studies failed to find significant polymorphism associations [197]. Conversely, a meta-analysis of 24 studies reported that the GSTM1 null genotype was significantly associated with asbestos-related disease [198]. A study of occupationalrelated LC including individuals with exposure to asbestos, silica and uranium did not detect any significant associations with polymorphisms of CYP1A1 (T6235C and A4889G) and CYP1B1 codon 432 [199]. A genome-wide association study (GWAS) of asbestosassociated LC risk (relying on self-reported asbestos exposure) noted the importance of antigen processing and presentation, and Fas signalling pathways (p < 0.001, false discovery rate < 0.05) [200]. The subjects, based in Texas, USA included 1154 cases with NSCLC and 1137 controls. The polymorphism with the highest significance, rs13383928 (p = 8.9x10-5) did not attain genome-wide statistical significance. Although the gene of greatest significance, C7orf54 was located on 7q32, the majority of the other genes of significance were located on 11q13 [200]. Polymorphisms of theTGFβ1 gene Leu10Pro and Arg25Pro were successfully associated with an increased risk of asbestosis but a lower risk of LC [201]. Ceramic Fibres Refractory Ceramic Fibres (RCF), also termed aluminosilicate wools, are amorphous fibres that belong to a class of materials termed synthetic vitreous fibres. These also include glass wool, rock (stone) wool, slag wool and special purpose glass fibres [202]. In 2001, an IARC Working Group placed RCF in Group 2B (possibly carcinogenic to humans), which concluded that there was sufficient evidence


87

in experimental animals but inadequate evidence in humans to establish the carcinogenicity of RCF [203]. Epidemiological studies in the US and Europe showed an association between exposure and increased numbers of lung tumours [202, 204]. Despite this finding, provision of the product stewardship programme in the US has resulted in the adoption of a voluntary recommended exposure guideline of 0.5 fibres/millilitre (f/ml). Silica Silicon is the second most abundant chemical element in the earth's crust, comprising about 27% of the average rock. The three main crystalline forms are quartz, tridymite and cristobalite, the first being so abundant that it is often used in place of the general term crystalline silica (CS). Cristobalite and tridymite are found in rocks and soil and are produced in some industrial operations when quartz or amorphous silica is heated (such as foundry processes, calcining of diatomaceous earth, brick and ceramics manufacturing and silicon carbide production). Quartz is a common component of soil and rocks and consequently workers are potentially exposed to quartz dust in many occupations and industries. Occupational exposure to respirable crystalline silica (RCS) is a serious but preventable health hazard. Prolonged exposure to RCS has long been known to cause one of the oldest known industrial diseases, silicosis, and it has been observed that there is a greater risk in workers exposed to very fine particles of CS, as found in quartz and cristobalite flours [205]. Quartz, feldspars, olivine, micas, thomsonite, jadeite, and prehnite are all silicates. The minerals tridymite, coesite, cristobalite and stishovite are also mineral forms of silica and are stable at high temperatures and pressures. Quartz is therefore a silicate made of pure silica. The silica minerals are based on a chemical unit called the silica tetrahedron. This consists of a silicon atom surrounded by four oxygen atoms. It is reasonable to discuss silica in this section, since pure silicon is very rare in nature [206]. There is so much available oxygen that pure native silicon is almost never found naturally. Respirable silica refers to particles with a diameter less than 10 µm; these smaller particles are more likely to reach the lungs [207]. Silica in the environment does not present a health problem but in occupational settings, such as mining, working in stone quarries and granite production, ceramic and pottery industries, steel production, and sandblasting brick, fracking, concrete, and pottery manufacturing, as well as operations using sand products, such as foundry work and many others, there are health risks. Silica can provoke fibrotic changes (Fig. 4A, B) and cancer of the lung, especially if inhaled as freshly ground crystalline silica dust [208]. Worldwide, there are estimated to be tens of millions of workers exposed to silica [209], many of whom are exposed to high concentrations of this dust. New industries can cause exposure, so silica exposure occurs from hydraulic fracking to gain access to oil and gas deposits [207]. Fracking involves pumping large volumes of water and sand, with a high silica content, into a well at high pressure to fracture shale and other tight rock formations, allowing oil and gas to flow into the well. The National Institute for Occupational

88


(A)

Quinn et al.

that studied occupational silica exposure and LC risk. These were all published after the IARC Monograph [219], and included 28 cohort, 15 case-control and two proportionate mortality ratio (PMR) studies [220]. This paper is discussed in detail because of its importance.

(B)

Fig. (5). Silica dust clouds from delivery trucks loading into sand movers. Photo credit: NIOSH.

Fig. (4). (A) Silicosis - close up image of a macroscopic section of lung with white fibrous tissue which is becoming confluent in areas. (B) Histology of silicosis with a hyaline fibrous whorl amidst lung tissue.

Safety and Health (NIOSH) obtained 116 silica samples at 11 different fracking sites and found the silica concentrations exceeded the current OSHA (Occupational Safety and Health Administration) standard in 47% of the samples, while 79% had concentrations greater than the new standard proposed by OSHA (available at osha.gov/dts/hazardalerts/hydra ulic_frac_hazard_alert.html) (Fig. 5). A study in Turkey found high exposure levels and silicosis among nearly 75% workers who sandblasted denim [210]. In 1997, the IARC evaluated the relation between silica exposure and human cancer [211]. This was based on both animal and epidemiological data and classified crystalline silica, inhaled in the form of quartz orcristobalite, from occupational sources as carcinogenic to humans (Group 1) [211]. However, the IARC Working Group noted that carcinogenicity was not found in all industrial circumstances studied and a long debate followed the publication of the Monograph [212-216]. NIOSH and the National Toxicology Program also concluded that crystalline silica is a human carcinogen [217, 218]. In 2006 Pelucchi et al. published a review of epidemiological studies; published between 1996 to 2005

The pooled RR of LC was calculated using random effects models from all cohort studies considering occupational exposure to silica was 1.34. The RRs were 1.69 in cohort studies of silicotics only, 1.25 in studies where the silicosis status was undefined and 1.19 among non-silicotic subjects. The pooled RR was 1.41 for all case-control studies. The RRs were 3.27 in case-control studies of silicotics only, 1.41 in studies where silicosis status was undefined and 0.97 among non-silicotic subjects. The RR was 1.24 for PMR studies. The pooled RR from studies of silicotics only (seven cohort, one case-control) was 1.74 (95% CI 1.37-2.22) suggesting a close relation between silicosis and LC, supporting old theories on the role of pulmonary fibrosis induced by silica [220]. Pelucchi et al. concluded the association with LC was consistent for silicotics, but the data were limited for nonsilicotic subjects and not easily explained for workers with undefined silicosis status. This left open the issue of doserisk relation and pathogenic mechanisms and supported the conclusion that the carcinogenic role of silica per se in the absence of silicosis was still unclear. In another, older study the most comprehensive exposure metric included period- and age-specific estimates of exposure and an estimate of occupational stability, but also simpler metrics gave significantly elevated estimates of the risk ratio between 1.36 and 1.50 for LC for occupations with the highest estimated cumulative silica exposure (≥10 mg/m3-years), allowing a lag time of 20 years [221]. New studies have shown excess lung mortality occurs in silica-exposed workers who do not have silicosis and who do not smoke [209]. Reports of the relation between LC and silica exposure or silicosis, or both, have since been published, including some meta-analyses, excluding Pelucchi et al. [207, 220, 222-224] and one pooled exposure-response analysis [225]. The effect of silica exposure on LC is weak and variable in workers who do not have silicosis using this data, except for the Steenland paper [207]. The large number


of studies looking at the association of silica and LC is a product of the fact that occupational silica exposure is widespread and of public health importance, and because the risk of LC found in epidemiologic studies is low compared with other pulmonary carcinogens, such as arsenic and asbestos, requiring many studies to detect it [207]. A large Chinese study by Liu et al. [226] addressed the question of whether silicosis was a necessary precursor of LC. After excluding individuals with radiographic evidence of silicosis from the analysis (representing 427/546 LC deaths), the RRs were 1.12, 1.41, 1.58, and 1.70, respectively, by quartile of cumulative exposure demonstrating silicosis was not a requirement for LC. In addition their study showed the RR for exposure to silica is similar in smokers and non-smokers. Nonetheless, because smoking is such a strong risk factor for LC, the risks for silica exposure and smoking together are high. For example, looking at these data with the NS with low exposure as the referent group for the other three categories, the RRs for highly exposed NS, ever-smokers with low exposure, and highly exposed ever-smokers were 1.60, 3.43, and 5.07, respectively. An increase in LC risk was noted after a 15-year delay with the logarithm of cumulative exposure, and little heterogeneity was present across different industries. The low exposure-response slope of silica, as compared with other known carcinogens, might have partly accounted for the difficulty in detection of its carcinogenic effect in workers without the disorder in previous studies. Citing this important study [227], the Working Group for IARC Monographs reaffirmed crystalline silica dust as a human carcinogen in March, 2009 [211]. Pelucchi et al. showed a moderately increased LC risk with silica exposure. The problem is whether the lower RR for exposed, non-silicotics or not defined as such is related to dose (i.e.: the higher RR in silicotics is related to higher exposure). This does not exclude that lower doses are not able to induce silicosis but can still induce cancer, as in the case of asbestos, or it is the expression of the uncertainty of the data [220]. In a pooled data set of 10 cohort studies, Steenland et al. examined 65,980 subjects, two thirds of them miners [225]. There were 1,072 deaths from LC. The authors found an increasing trend in risk of LC with cumulative silica exposure: RR 1.0 (95% CI 0.85-1.3); RR 1.3 (95% CI 1.11.7); RR 1.5 (95% CI 1.2-1.9); and RR 1.6 (95% CI 1.3-2.1) compared with the lowest quintile. This dose-risk relation supports a causal relationship between silica and LC. These authors confirm this view in a recent paper [207]. Studies providing the most convincing evidence of carcinogenicity indicate increased risks of LC are restricted to those groups with the highest cumulative exposure, suggesting the existence of a threshold. However, exposureresponse relationships in the cohorts have not been consistent because exposure measures differ between studies. This makes it challenging to carry out a meta-analysis [228]. In addition the role of co-exposures cannot be excluded. Some co-exposures are deducible on the basis of common knowledge (e.g. foundry workers in the study from Starzynski et al. [229]), some are stated by the authors (e.g.:


89

potential exposure to radon in the study from Brown et al. [230]), some are controlled for by excluding co-exposed workers (PAH and asbestos in the study from Chan et al. [231]). It is difficult to assess to what extent these points may affect the overall evaluation. It is a limitation of a metaanalysis to take the risks from single studies at their face value (i.e. to disregard the possible effect of several factors reported in detail thereafter). Such factors may affect risk estimates, and can be identified by only a full examination of the data reported (see Pelucchi in tables S1 to S3) (online) [220]. It is likely that the aetiology of silica-induced lung carcinoma is similar to that of asbestos, as this is a silicate. The chronic inflammation and release of oxidants by silica is also thought to cause genotoxic damage to the lung epithelium, thereby increasing the risk of LC. Neutrophils and macrophages release several growth factors as well as free radicals that may contribute to the pathogenesis of silicosis and LC. A Chinese case-control study of 111 silicaexposed miners and 183 silicosis patients concluded that the severity of the silicosis was not influenced by SNPs of the inflammatory cytokine genes TNFα, TGFβ, or IL-10 [232]. However, similar to mechanisms suggested for arsenic carcinogenicity (vide infra), a distinctive somatic mutation pattern for LC related to silica has been proposed. A comparison of LC arising in workers with silicosis with nonsilica related LCs revealed differences in the p53 mutational spectrum. Codon 12 alterations of the KRAS gene, classically associated with LC, were absent from tumours with an associated history of silica exposure [233]. Fumes These will include some metals, which are inhaled as fumes, such as during welding. Benzidine and Beta-Naphthylamine After adjustment for smoking, exposure to bis(chloromethyl) ether, and age at first exposure, a marginally significant HR was observed for the long duration of exposure (DOE) group to benzidine and betanaphthylamine in causing LC (adjusted HR = 3.02, 95 % CI 0.84-10.93, p = 0.091), compared to the short DOE group [234]. Octachlorodipropyl Ether (s-2) Mosquito Coils Children and their parents at home are often protected from mosquitoes by insecticides. The annual worldwide consumption of the four major types of residential insecticide products (aerosols, mosquito coils, liquid vaporizers, and vaporizing mats) is in the billions of units. Mosquito coils are burned indoors and outdoors in East Asia and to a limited extent in other parts of the world, including the United States. Coils consist of an insecticide/repellent, organic fillers capable of smouldering, a binder, and additives such as synergists, dyes, and fungicide. China uses millions of coils and Indonesians buy an estimated seven billion coils annually. Coils contain pyrethroid insecticides, particularly d-allethrin, and may contain octachlorodipropyl ether (S-2, S-421) as a synergist or active ingredient. These coils probably expose children and adults to some level of bis (chloromethyl) ether

90


Quinn et al.

(BCME). BCME is formed from formaldehyde and hydrogen chloride, combustion products formed from the slow smouldering (about 8 hr/coil) of the mosquito coils. BCME is an extremely potent lung carcinogen discussed below. In a small study, coils purchased in Indonesia and in the United States contained highly variable amounts of S-2. Some coils containing S-2 were not labelled, making it impossible for consumers to make an informed decision about coil contents. Mosquito coils containing S-2 are unregistered, and their use is illegal in the United States [235].

Diesel Exhaust Emissions and Atmospheric Pollution

The status of BCME as a potent human lung carcinogen is established [236-238]. Some chemical workers in 1963 were exposed to chloromethyl ethers (CMEs) and there was an epidemic of LC. Approximate estimates of exposure were made. Standardized mortality ratios (SMRs) were calculated for LC, based on Philadelphia city rates. Over 30 years of observation, 25/67 deaths were due to LC, with a doseresponse relationship. SMRs were elevated only among 59 moderately and heavily exposed workers, peaked at 23.1 in the first decade, and then declined to 7.4 and 7.9 in later decades. The mean latency period from onset of exposure to death was 21 to 25 years and was inversely related to cumulative exposure. Three of 12 heavily exposed cases occurred in non-smokers. Small cell lung carcinoma accounted for 80% of the moderately and heavily exposed cases [239].

Air pollution is a major contributor to cardiopulmonary ill health. According to results from the Global Burden of Disease Study, outdoor particulate matter air pollution contributed to more than 3.2 million premature deaths and 76 million disability-adjusted life-years in 2010, ranking among the most important health risks world-wide [244]. The WHO estimated smoking caused 5.1 million deaths and 71% of LC cases worldwide in 2004, whereas air pollution caused 1.2 million deaths and 8% of LC worldwide in the same year [245].

Chlorinated Solvents Two large studies with a total of 4,942 cases were studied with appropriate controls [240, 241]. In the first study [240], after adjustment for exposure to asbestos, there was a positive, statistically significant association with lung cancer for men and women exposed to a combination of perchloroethylene (PCE), trichloroethylene and dichloromethane. Further adjustment for socioeconomic status slightly decreased this association. No statistically significant associations were found for other solvent combinations. The second study showed an increased risk of LC associated with occupational exposure to PCE (OR (any exposure) 2.5, 95% CI 1.2-5.6; OR (substantial exposure) 2.4, 95% CI 0.8-7.7) and to carbon tetrachloride (OR (any exposure) 1.2, 95% CI 0.8-2.1; OR (substantial exposure) 2.5, 95% CI 1.1-5.7) [241]. Again no other chlorinated solvents showed both statistically significant associations and dose-response relationships. ORs appeared to be higher among non-smokers. Both studies concluded there were suggestive indications that exposure to PCE and carbon tetrachloride may increase the risk of LC. The first study thought this association was more prominent among women, who seem to have a higher prevalence of exposure than men [240]. Organophosphates A recent review by the IARC classified Malathion and Diazinon as “probably carcinogenic to humans”, Group 2A. Malathion is used in insect control, agriculture and for public health purposes and has been associated with Hodgkin lymphoma and increased risk of prostate cancer [242].

Increased combustion of fossil fuels in the last century is responsible for the progressive change in the atmospheric composition. Air pollutants, such as carbon monoxide (CO), sulphur dioxide (SO2), nitrogen oxides (NOx), volatile organic compounds, ozone (O3), heavy metals, and respirable particulate matter, differ in their chemical compositions, reaction properties, emission, time of disintegration and ability to diffuse in long or short distances [243].

A large study from nine European countries of 312,944 cohort members contributed 4,013,131 person-years at risk [246]. This study came from a wide range of European regions, and reduced the possibility of sampling and publication bias. It also benefited from a high follow-up rate and adjustment of potential confounders, including a set of smoking variables [247]. During follow-up (mean 12.8 years), 2095 incident LC cases were diagnosed. The metaanalyses showed a statistically significant association between risk for LC and PM (particulate matter) 10 (HR 1.22, 95% CI 1.03-1.45 per 10 μg/m3). For PM2.5 the HR was 1.18(95% CI 0.96-1.46) per 5 μg/m3. The same increments of PM10 (particulate matter with an aerodynamic diameter < 10 μM) and PM2.5 (particulate matter with a diameter < 2.5 μM) were associated with HRs for lung ACA of 1.51 (95% CI 1.10-2.08) and 1.55 (95% CI 1.05-2.29), respectively. SqCCs were not significantly associated with particulate matter air pollution. The associations were stronger in participants who lived at their enrolment address throughout follow-up. Interestingly Danish and Austrian cohorts contributed more than half the lung cancer cases. Despite this fact, the cohort areas represented a wide range of air pollution concentrations, with three to 12 times higher means air pollution levels in some southern European areas than in some northern European areas. Recently the USA Environmental Protection Agency surveyed over 550,000 individuals, comparing air pollution exposure and mortality rates. An increase in the PM2.5 of 10 μg/m3 was associated with a 7.52 percent increase in mortality over a year, and a 2.14 percent increase in death rate over a two-day period [248]. An increase in road traffic of 4,000 vehicle-km per day within100 m of the residence was associated with an HR for LC of 1.09 (95% CI 0.99-1.21). The results showed no association between LC and NOx concentration (HR 1.01, 95% CI0.95-1.07 per 20 μg/m3) and traffic intensity on the nearest street (HR 1.00, 95% CI 0.97-1.04 per 5000 vehicles per day). The authors believed particulate matter air pollution contributes to LC incidence in Europe [246]. The


authors suggested air pollution with nitrates and toxic agents formed from NOx, such as nitrosamines, might be more important for risk for LC than polycyclic aromatic hydrocarbons (PAHs) in the air. The concentration of PAHs in the air has decreased substantially in many cities in developed countries throughout the past three to four decades [249]. This study is similar to the Harvard Six Cities study [250], estimate in a US cohort (351 cases) of 1.37 (1.071.75) per 10 μg/m3 and that from a Canadian study (HR 1.29, 0.95-1.76; 2390 cases) [251], but higher than results in the Netherlands [252], Japan [253], China [254] and Italy [255].


91

exposure was up to six times higher when burning smoky coal than smokeless coal. PAH measurements from unventilated fire-pits were up to five times that of ventilated stoves. Exposure also varied between different room sizes and season of measurement. PAH exposure is probably modulated by a variety of factors, including fuel type, coal source, and stove design [260]. In rural Xuanwei County, China in 1995, indoor concentrations of airborne particles and benzo [a]pyrene were compared in homes during smoky coal burning in stoves with chimneys and in unvented stoves or fire pits. A long-term reduction in LC incidence was noted after stove improvement [261].

Although the large European paper quoted at the beginning of this section adjusted thoroughly for smoking in all cohorts, the authors could not exclude potential residual confounding, because data for smoking were obtained at enrolment, and they did not account for changes in smoking habits during follow-up [246]. Coal-Burning Fume Exposure A third of the world's population uses solid fuel derived from plant material (biomass) or coal for cooking, heating, or lighting. According to the WHO, more than 3 billion people worldwide depend on biomass fuels and coal for cooking and heating (Fig. 6) [256]. A disproportionate number of individuals depending on solid fuel smoke live in Africa and Asia [158]. Indoor air pollution (IAP) accounts for about 3.54 million deaths every year but not all of these will be from LC [257]. Coal and its by-products often contain significant amounts of radionuclides, including uranium which is the ultimate source of the radioactive gas radon. IAP has been listed as the sixth largest risk contributing to burden of disease in China [258]. Nearly all of China’s rural residents and a few urban residents use solid fuels (biomass and coal) for household cooking and/or heating [259]. Women and children living in severe poverty have the greatest exposures to household air pollution. These fuels derived from locally available fuel sources, such as crop waste, dung, wood, and leaves, are smoky. As mentioned above, coal is also burnt in an enclosed environment. These fuels are often used in an open fire or simple stove with incomplete combustion and result in a large amount of household air pollution, when smoke is poorly vented. Gases such as O3, nitrogen dioxide, CO and SO2 from coal; microbial and chemical volatile organic compounds; passive smoke; and outdoor ambient air are the most common types of air pollutants encountered indoors. Particulate matter less than 10µM in aerodynamic diameter (PM10), and particularly particulate matter less than 2.5 mm in diameter (PM2.5), can penetrate deeply into the lungs and have great potential for damaging health [256]. Respiratory tract cancers, including both nasopharyngeal cancer and LC, are strongly associated with pollution from coal burning but further data are needed about other solid fuels [257]. Exposure to PAHs from burning "smoky" (bituminous) coal has been implicated as a cause of the high lung cancer incidence in the Chinese counties of Xuanwei and Fuyuan (PAHs are considered above in the vaping section). PAH

Fig. (6). Indian woman cooking indoors with little apparent ventilation (Photo courtesy: Mrs. Jyoti Londhe, Chest Research Foundation, Pune, India).

A meta-analysis on 25 case-control studies (10,142 cases and 13,416 controls) summarised the association between household coal use and LC risk. In addition it explored the effect modification of this association by geographical location. Using random-effects models, household coal use was associated with LC risk among all studies throughout the world. This was especially in those studies carried out in mainland China and Taiwan. Stratification by regions of mainland China and Taiwan found a variation in effects across the regions, with south/south-eastern and southwestern China experiencing the highest risk. The elevated risk associated with coal use throughout Asia was also observed when stratifying studies by gender, smoking status, sample size, design (population vs hospital case-control) and publication language [262]. It has been estimated that over 70% of households in China burn solid fuels (including coal, wood and crop residues) [263]. A study of 40 cases of NSCLC from never smoking females in China identified an increased proportion of exon 18 EGFR mutations (7 of 15 or 46% of all detected EGFR mutations) and a relatively reduced rate of exon 21 alterations (13%). Interestingly all of those diagnosed with a mutation were exposed to fumes as a result of coal burning for cooking and heating needs. There were no apparent differences in the mutation profiles of those burning bituminous coal (n=21) and anthracite coal (n=12) (p > 0.05) [264].

92


Polymorphisms of genes coding for metabolic enzymes involved in the detoxification of PAHs could affect the risk of LC in affected individuals. Various studies of the glutathione S-transferases have been conducted that examine the risk of LC in association with IAP as an effect of coalburning, cooking, wood or biomass consumption. Results were synthesised in a meta-analysis of six studies contributing 912 cases and 1063 controls from areas of Asia [265]. There was no observed association for the GSTP1 105Val allele, however LC risk was increased in association with the GSTT1 null genotype (OR 1.49, 95% CI 1.17-1.89, p = 0.001). The presence of the GSTM1 null genotype increased risk to a level of borderline significance (OR 1.31, 95% CI 0.95-1.79, p = 0.1) and this increased (OR 1.64, 95% CI 1.25-2.14, p = 0.0003) when the analysis was limited to four studies originating from populations associated with the use of coal [265]. Recently, a GWAS was conducted on NS with LC (6,609 cases, 7,457 controls) from the Female Lung Cancer Consortium in Asia. Of five SNPs examined it was reported that use of coal was associated with TP63 rs4488809 (rs4600802) (p = 0.04) and HLA Class II rs2395185 (p = 0.02) [264]. Cooking (Fumes in Poorly Ventilated Areas from Oils) Part of this discussion overlaps with the previous section, as biomass is used as a cooking fuel. Biomass fuel smoke possesses most of the toxins found in tobacco smoke [266]. Chronic IAP exposure has been associated with acute respiratory infection in children [267], which is a cause of 59% of deaths among children younger than five years in developing countries [268], childhood cancer, and lung cancer in women [269]. There is evidence that IAP exposure and tuberculosis, which also has a high incidence in low- to middle-income countries, increases the likelihood of LC [256]. A paper studied the association of cooking conditions, fuel use, oil use, and risk of lung cancer in a developed urban population in a prospective cohort of women in Shanghai. A total of 71,320 never-smoking women were followed from 1996 to 2009. Four hundred and twenty-nine incident LC cases were identified. Most of the women used soya bean oil for cooking; few used vegetable oil. Poor kitchen ventilation was associated with a 49% increase in LC risk. Women who lived in a home with poor ventilation during both childhood and adulthood had a 69% elevated risk of LC, compared to women never exposed to poor ventilation during their lives. Interestingly, use of coal on its own was not significantly associated with LC risk. However coal use with poor ventilation and 20 or more years of using coal with poor ventilation was significantly associated with LC, compared to no exposure to coal or poor ventilation [261, 262]. IAP from coal use in homes is an IARC class 1 carcinogen [270]. Indoor coal burning increases particulate matter [271], PAHs, and heterocyclic aromatic compounds in the air. Cooking oil use was not significantly associated with LC [272]. Amongst NS females in China, those with the ERCC2 751A>C genotypes (Lys751Gln) carry an increased risk of LC, specifically ACA (adjusted OR 1.64, 95% CI 1.06-2.52).

Quinn et al.

The risk increased in those without exposure to cooking oil fumes (OR 1.98, 95% CI 1.18-3.32, p = 0.01) or fuel smoke (OR 2.47, 95% CI 1.46-4.18, p = 0.001). Since this variant is thought to have reduced DNA repair capacity this may suggest a role for unknown exposure to SHS [273]. A stratified analysis of Chinese female NS revealed an increased risk of LC in those exposed to cooking oil fumes (adjusted OR 1.59, 95% CI 1.13-2.23). The same analysis did not detect any association between cooking oil exposure and the telomerase reverse transcriptase (TERT) rs2736100 T>G polymorphism, although the SNP was associated with increased risk in the non-stratified population (OR 1.44, 95% CI 1.09-1.9 for TG and OR 1.85, 95% CI 1.29-2.65 for GG) [274]. Similarly, a case control study of Chinese never smoking females (260 cases and 318 controls) did not detect a significant interaction between the TP63 polymorphism rs10937405 C>T and exposure to cooking oil fumes, although there was an increased risk of ACA (OR 1.58, 95% CI 1.11-2.25, p = 0.011) associated with exposure history [275]. Radon As noted above, coal and its by-products often contain significant amounts of radionuclides, including uranium, which is the ultimate source of the radioactive gas radon. Radon is also produced from the decay of naturally occurring uranium in rocks and soil. This gas has a half- life of 3.8 days. The decay products, polonium 214 and 218, present radiological hazards, and, being denser than air, may accumulate on floors and in basements. Radon gas is ubiquitous at very low levels in outdoor air, and can accumulate indoors by entering homes through cracks in floors, walls and foundations. The odds ratios for LC increase with increasing radon exposure [276, 277]. In North America residential studies, the OR was 1.37 for high concentrations (exceeding 200 Bq/m3 [5.4pCi/L] relative to concentrations under 25 Bq/m3 [0.68pCi/L]) [276]. In Europe, radon exposure at home accounts for 9% of deaths from LC and 2% of all deaths from cancer [277]. In the U.S. this amounts to over 18,000 deaths per year due to radon [278]. Radon may also be released during fracking. Since 2008, the increase in volumes of gas well wastewater and levels of radon observed in the unconventional shale gas well flow-back fracking water has compelled the Pennsylvania Department of Environmental Protection Bureau of Radiation Protection to fully re-examine these oil and gas operations [279]. In Egypt, during oil and natural gas production in Abu Rudeis, activity of the most hazardous radionuclide 226Ra was found to be higher than the exemption levels established by International Atomic Energy Agency [280, 281]. At a molecular level, radon is associated with p53 mutations [282]. It has been suggested that radon exposure may be associated with a specific p53 ‘hotspot’ mutation profile, compared to other carcinogens, however, there is insufficient evidence to support this theory [283]. A systematic review of eight studies including data on 578 subjects concluded that the available data did not imply the existence of a signature radon-associated p53 mutation. The population assessed included miners as well as those with a general background [284].


Underground mines, particularly uranium mines, contain the highest levels of radon, with concentrations one order of magnitude higher than concentrations measured in homes. Epidemiological studies of miners, along with experimental evidence from animal studies [285], have established a causal relationship between occupational radon exposure and lung cancer development in miners [13, 286]. Chronic exposure to radon and its decay products is recognized as the second leading cause of LC, after active cigarette smoking. Alpha particle emissions from inhaled radon decay products, and not radon itself, cause LC [287]. Alpha particles, which can only penetrate a short distance into bronchial epithelium, induce more biological damage than beta or gamma radiation, and can induce DNA base mutations and chromosomal strand breaks [288]. Torres-Durran et al. reported a case-control study on LC in NS with or without exposure to residential radon in Galicia, Spain. Few studies have analysed the role of combined radon and ETS (environmental tobacco smoke) exposures in LC risk in nonsmokers. Following a published meta-analysis, the authors demonstrated that a radon exposure of ≥ 200 Bq·m-3 significantly increases LC risk in NS (OR 2.42, 95% CI 1.45-4.06), compared with individuals exposed to ≤ 100 Bq·m-3 [289, 290]. Within this population, the odds ratio increases to 2.84 for females exposed to ≥ 200 Bq·m-3 [289]. The main point of this article is the confirmation of two previous studies in NS: a European pooling study and a Swedish epidemiological study [277, 291]. In this Galician study, ETS exposure at home was a significant factor for increased LC risk in a population exposed to ≥ 200 Bq·m-3. ETS is difficult to evaluate in epidemiological studies and this point is outlined in the discussion of this article by Torres-Durran et al. [289]. The authors have chosen three conditions: 0, 1-35 or ≥ 36 years living with a smoker. Individuals exposed to radon ≥ 200 Bq·m-3 have an OR of 1.99 without ETS exposure and this rises to 2.75 if they have lived 1-35 years with a smoker. The two determinant items of this case-control epidemiological study are the high rate of radon detector return and the large numbers of years the population lived in the same house, in this area of Galicia, Spain. However, cases of never smoking males represented only 20%, and ACA was the main histological type (77.5%), as in NS with LC [288, 292, 293]. 22 case-control studies of residential radon and LC risk involving 13,380 cases and 21,102 controls were examined. This meta-analysis showed residential radon exposure was associated with a significantly increased risk for LC. Compared with the lowest category, people exposed to highest one of residential radon experienced 29% higher risk of LC. The risk of LC increased by 7% for every 100 Bq/m3 radon increment. Subgroup analysis displayed a more pronounced association in the studies conducted in Europe. Studies restricted to female or non-smokers showed weak associations between radon exposure and LC [294]. Arsenic and Inorganic Arsenic Compounds Arsenic (As) is a ubiquitous metalloid found in several forms in food and the environment, such as the soil, air and water. Exposure to arsenic through groundwater is a major


93

public health problem throughout the world, with people in South Asia (Bangladesh and West Bengal, India) the most seriously affected [295]. Chronic arsenic exposure through drinking water is a public health problem affecting millions of people worldwide, including at least 30 million in Bangladesh [296]. Although water consumption provides the majority of human exposure, millions of individuals worldwide are significantly exposed to As through grains, vegetables, meats, and fish, as well as through food processed or grown in water containing arsenic. As in food may occur in both organic and inorganic forms depending on the food [297]. The predominant form is inorganic arsenic in drinking water, which is both highly toxic, carcinogenic and rapidly bio-available. An increased incidence of disease mediated by this toxic chemical is the consequence of long-term exposure. Work-related arsenic exposure can occur by inhalation and skin contact with arsenical compounds in diverse industrial or agricultural settings, including mining or smelting metal ores, manufacturing or using pesticides (e.g., insecticides, herbicides, fungicides), producing or using wood preservatives (i.e., chromated copper arsenate), manufacturing or working with paints and pigments, manufacturing glass and ceramics, and producing or working with lead-arsenic alloys and electronics (e.g., semiconductors) [158]. As-related carcinogenesis, due to chronic exposure, has a latency period of 30 to 50 years [298]. Numerous epidemiological studies have found associations of chronic arsenic exposure with skin, bladder, lung, prostate, and liver cancers [299, 300]. There is a large body of evidence showing the human lung is a major target site of arsenic carcinogenicity. Several studies have shown arsenic accumulates in the lungs to a greater extent than in most other organs [301]. This could be related to the high concentration of sulphydryl groups found in lung tissue which are known to bind arsenic [302]. Contrary to the earlier view that methylated compounds are innocuous, the methylated metabolites are now recognized to be both toxic and carcinogenic, possibly due to genotoxicity, inhibition of antioxidative enzyme functions, or other mechanisms. As inhibits indirectly sulphydrylcontaining enzymes and interferes with cellular metabolism. The chemicals effects include cytotoxicity, genotoxicity and inhibition of enzymes concerned with antioxidant function. As modulates genes which function as transcription factors, thereby affecting target genes which play a role in LC promotion and progression [303]. There is some evidence that As increases EGFR-related signalling pathways, leading to increased levels of phosphorylated extracellular signalregulated kinase/mitogen-activated protein kinase (ERK/MAPK) and expression of cyclin D1 [304]. These effects are all related to nutritional factors, either directly or indirectly. Nutritional studies both in experimental and epidemiological studies provide convincing evidence that nutritional intervention, including chemoprevention, offers a pragmatic approach to mitigate the health effects of arsenic exposure. This is particularly the case in cancer, in resource-poor developing countries. Nutritional intervention, especially with micronutrients, many of which are antioxidants and share the same pathway

94


with As, acts as a host defence against the health effects of As contamination in developing countries [300]. A prospective analysis of 6,888 individuals in northeastern Taiwan measured well-water As concentration exposure for 11 years [305]. 178 lung cancers were found through linkage with the national cancer registry profiles in Taiwan. A significant dose-response trend (p = 0.001) of LC risk was associated with increasing arsenic concentration. Lung cancer risk was associated with As exposure of at least 300 mg/L, compared with exposure less than 10 mg/L. Significant dose-response trends and the synergistic effect of arsenic exposure and cigarette smoking were found for SqCC (p = 0.004) and SCLC (p = 0.02), but not in ACA (p = 0.67). When duration was accounted for, all levels of exposure, including low concentrations had an increased risk for LC. Exposure to As in drinking water during early childhood or in utero have pronounced pulmonary effects, greatly increasing subsequent mortality in young adults from both malignant and non-malignant lung disease [306]. It has been suggested that As exposure is more likely to lead to SqCC and SCLC than other histological subtypes, although data on this is limited by a lack of associated smoking history [307]. A comparative genomic hybridisation study of 22 SqCC from As-exposed individuals and 30 SqCC with no history of exposure revealed difference in copy number alterations. These included copy number gains at 19q13.33 and losses at 1q21.1, 7p22.3, 9q12, and 19q13.31 [308]. Exposure to As by drinking water is relatively high in the Northern area of Chile [308, 309]. Whole genome sequencing of SqCC from a NS living in Northern Chile with clinical signs of As exposure revealed an increase in the number of T>G/A>C transversions compared to 171 SqCC referenced from the Cancer Genome Atlas database (p < 5 x 10-6). In addition, a rare His179Asp of the DNA-binding domain was identified. This suggests a characteristic mutational signature pattern associated with As exposure [310]. Associated carcinogenicity may be related to SNPs in the metabolic enzymes glutathione S-transferase M1 (GSTM1) and CYP4501A1 however this is probably influenced by a history of smoking [311]. Aluminium Bauxite (aluminum ore) is mined in open pits, and electrolysis converts bauxite to alumina. The latter process is undertaken in large pots. The pot room environment may contain alumina dust, metal fumes and fumes from other organic and inorganic compounds. According to Gibbs and Labreche there is reasonable evidence for an association between aluminium in its production and LC [312], although PAHs and possibly other exposures emitted from the process are probably involved in LC aetiology. Inhalation exposure at seven UK secondary aluminium smelters was investigated to quantify the main exposures and identify their sources. The substances monitored were gases (carbon monoxide, hydrogen sulphide and nitrogen dioxide), total inhalable dust, metals, ammonia, PAHs, particulate fluoride salts and acids [313]. The most probable etiological agents are PAHs. Coal tar pitches from which PAHs originate have been classified by the IARC as having sufficient evidence for carcinogenicity in humans [314].

Quinn et al.

Smoking as a confounder has been considered and eliminated. Smoking rates have decreased and the smoking data in many studies are imperfect. In the Canadian studies [312], the associations are strong and there are clear exposure-response relationships for PAH and LC. The findings are also compatible with the findings in coke oven and other PAH-exposed populations, and PAHs are known to induce tumours. There are indications that as the level of exposure to coal tar pitch volatiles have decreased, so has the risk of LC. Beryllium (and Compounds) Beryllium's classification as a carcinogen is based on limited human data that show inconsistent associations with LC. Boffetta et al. [315] conducted a systematic review of epidemiologic studies on cancer amongst workers exposed to beryllium, including a study of seven U.S. production plants (which has been recently updated), patients with beryllium disease and several smaller studies. A small excess mortality from LC was detected in the large cohort, which was partially explained by confounding by tobacco smoking and urban residence. Other potential confounders were addressed. The excess mortality was mainly among workers employed, often for a short duration, in the early phase of the manufacturing industry. There was no relation with duration of employment or cumulative exposure, whereas average and maximum exposure were associated with LC risk. The use of lagged exposure variables resulted in associations with LC risk; however, these associations were due to confounding by year of birth and year of hire. The available evidence did not support a causal association between occupational exposure to beryllium and the risk of cancer. The National Institute for Occupational Safety and Health (NIOSH, USA) studied 9,225 workers employed at seven beryllium processing plants between 1940 and 1969. They based the study on work records obtained from the companies. They also obtained the death certificates of workers who had died. NIOSH found an increased risk of LC in workers exposed to beryllium at all plants combined. They found 280 deaths, but expected 221. The LC excess was confined to workers hired in the 1940s and 1950s. Among workers hired in the 1960s, the risk of LC was noticeably lower than expected; 18 deaths from LC with 29 expected. They examined the effect of smoking and county of residence and concluded that these factors could not completely explain the increased lung cancer risk. The study did not address the relationship of LC to the degree of exposure or to specific types of beryllium compound [316]. Rothman and Mosquin questioned the NIOSH data. The strongest association reported in the NIOSH study, a standardized rate ratio for death from LC of 3.68 for the highest vs the lowest category of time since first employment, is affected by sparse-data bias. This stems from stratifying 545 LC cases and their associated person-time into 1792 categories. For time since first employment, the measure of beryllium exposure with the strongest reported association with LC, there were no strata without zeroes in at least one of the two contrasting exposure categories. Reanalysis using fewer strata or with regression models gave substantially smaller effect estimates. Simulations confirmed that the original stratified analysis was upwardly biased.


Other metrics used in the NIOSH study found weaker associations and were less affected by sparse data bias [317]. The strongest association reported in the NIOSH study seems to be biased as a result of non-overlap of data across the numerous strata. Simulation results indicate that most of the effect reported in the NIOSH paper for time since first employment is attributable to sparse-data. It appears, despite conflicting data, the evidence at the moment does not favour beryllium as a cause for LC. Cadmium Cadmium (Cd) is a toxic, heavy industrial metal that poses serious environmental health hazards to both humans and wildlife. Cd and Cd-compounds are classified as carcinogenic to humans (Group 1) by the IARC [219]. The molecular mechanisms involved in Cd-induced carcinogenesis are only now beginning to be elucidated [318]. Acute exposure to Cd causes cough, dryness and irritation of the nose and throat, headache, dizziness, chest pain, pneumonitis, and pulmonary oedema [319]. Cd occurs naturally in the earth’s crust and in ocean water. Elemental Cd is a soft, silver-white metal, which is recovered as a by-product of zinc mining and refining. The non-smoking, general population is exposed to Cd primarily via ingestion of food and, to a lesser extent, via inhalation of ambient air, ingestion of drinking-water, contaminated soil or dust. Because tobacco leaves naturally accumulate large amounts of Cd [320], cigarettes are a significant source of Cd exposure for the smoking general population. Tobacco smokers are estimated to be exposed to 1.7 μg Cd per cigarette, and about 10% is inhaled when smoked [219]. The main route of Cd exposure in the occupational setting is via the respiratory tract, although there may be incidental ingestion of dust from contaminated hands and food [219]. Occupations in which the highest potential exposures occur include, cadmium production and refining, Ni (Nickel)-Cd battery manufacture, Cd pigment manufacture and formulation, Cd alloy production, mechanical plating, zinc smelting, brazing with a silver-Cdsilver alloy solder, and polyvinylchloride compounding. Although levels vary widely among the different industries, occupational exposures have decreased since the1970s [219]. An emissions inventory compiled by the California Air Resources Board [321], indicates 16 to 18 tons/year of Cd are emitted in California; stationary sources account for 80% or more of the Cd emissions. Stationary sources likely to emit Cd include secondary smelters, cement-manufacturing plants, Cd -electroplating facilities, plants burning oil or coal, and sewage sludge incinerators. Mobile sources that emit Cd include gasoline and diesel vehicles and particles resulting from tyre wear [322]. Estimates of the number of workers potentially exposed to Cd and its compounds have been developed by CAREX (CARcinogen EXposure) in Europe. Based on occupational exposure to known and suspected carcinogens collected during 1990-93, the CAREX database estimates that 207,350 workers were exposed to Cd and Cd compounds in the European Union [219].


95

For Cd -processing workers from 17 plants in the United Kingdom, the mortality from LC was significantly increased (SMR 1.12, 95% CI 1.00-1.24), with apparent positive trends with duration of employment and with intensity of exposure [323]. The increase in LC risk was stronger in the small proportion of workers with high Cd exposure (SMR 1.62, 95% CI 0.89-2.73). Follow-up of the UK Ni-Cd battery workers confirmed a slight increase in SMR for LC associated with duration of employment in high-exposure jobs. An increase in mortality rates from lung cancer was detected in a small cohort of individuals who worked in the Ni-Cd battery-producing industry in Sweden, and especially in those who had the longest duration of employment and latency [324]. These authors said the role of concomitant exposure to nickel needed further study. Further follow-up of this Swedish cohort showed an SMR for LC in male battery workers of 1.76 (95% CI 1.012.87), although without an association with estimated total Cd exposure [325]. Excess LC mortality was reported among workers employed in a US Cd recovery plant, which had been used as an arsenic smelter until 1925 [326]. A doseresponse relationship was demonstrated between the estimated cumulative exposure to Cd and LC risk [327]. In Belgium, Nawrot et al. studied subjects residing near three zinc smelters and also subjects from the area away from the Cd pollution for the incidence of cancer from initial examinations in 1985-89 to 2004. Using urinary Cd excretion and Cd in garden soil as exposure indicators, the HR for LC was 1.70 (95% CI 1.13-2.57) for a doubling of the 24-hour urinary Cd excretion, (OR 4.17; 95% CI 1.2114.4) for residence in the high exposure area vs the lowexposure area, and 1.57 (95% CI 1.11-2.24) for a doubling of the Cd concentration in soil. Cancer was also increased in the high-exposure group. Information on smoking was included in the adjustments. Data on urinary Cd excretion, adjusted for arsenic, suggested that arsenic exposure alone could not explain the observed increases in risk [328]. Xuanwei and Fuyuan, working in the Yunnan province in southwest China showed a very high incidence of LC. Previous studies in this area have shown a strong association of LC mortality with air pollution from bituminous coal combustion [329]. Dietary samples were collected from 60 families of LC and control groups and 14 trace elements were determined using inductively coupled-plasma mass spectroscopy. Dietary intake of the trace elements contributed 96.6% of total intake. Among the 14 elements tested, Cd and titanium were present at a significantly higher level in the food consumed by the cancer group, compared to the control group. The intake of selenium by the population living in the areas was much lower than normal, with the cancer group experiencing even more severe selenium deficiency. In both groups, the intakes of several essential elements (iron, copper, and zinc) from food and the drinking water were significantly lower than required, according to the Chinese Dietary Reference Intakes [330]. No mention is made of the smoking habits of the small population studied. Garcia-Esquinas et al. evaluated the association of longterm Cd exposure, as measured in urine, with cancer mortality in American Indians from Arizona, Oklahoma, and

96


North and South Dakota who participated in the Strong Heart Study during 1989-1991. This study was a prospective cohort study of 3,792 men and women 45-74 years of age who were followed for up to 20 years. Baseline urinary Cd (U-Cd) was measured using inductively coupled plasma mass spectrometry. Low-to-moderate Cd exposure was prospectively associated with total cancer mortality and with mortality from cancers of the lung and pancreas. The LC deaths due to tobacco smoking attributed to Cd exposure was estimated at 9.0% (95% CI 2.8-21.8) [331]. Previous studies have wrestled with the problem of contamination by arsenic and nickel. In the West, As appears to be less of a problem now but, as noted elsewhere, in the East it is a problem in its own right. Not all authors found Cd carcinogenic. Compared with the general population of England and Wales, mortality from LC among copper/ cadmium alloy workers was close to expectation (observed deaths 18, expected deaths 17-8, SMR 101, 95% CI 60-59). These authors documented increased risks of LC in vicinity workers only, and an increased risk of non-malignant diseases of the respiratory system at higher cumulative cadmium exposures. The authors concluded the findings did not support the hypothesis that exposure to cadmium oxide fume increases risks of mortality for LC [332]. The IARC Working Group noted that cases of LC could potentially be misclassified as non-malignant disease. There was some population overlap between these studies [219]. The weight of evidence still favours cadmium being classed as a carcinogen. In the report from the Joint Research Centre of the European Commission (EU-JRC) cadmium oxide is considered to be a suspected human inhalation carcinogen [333]. They concluded the weight of evidence collected in genotoxicity studies, long term animal experiments and epidemiological data leads to the conclusion that cadmium oxide has to be considered at least as a suspected carcinogen for LC. The mechanisms of Cd-induced pulmonary carcinogenesis are still incompletely defined [318]. Oxidative stress plays a central role in Cd- induced carcinogenesis because of its involvement in Cd-induced aberrant gene expression, inhibition of DNA damage repair, and apoptosis [334]. Epidemiological studies suggest that occupational and environmental exposure to Cd induces LC in rodents, as well in humans [335]. The overexpression of major metallothionein (MT) isoforms is associated with Cd-induced LC [336, 337]. Specifically, in vitro cell line studies revealed that MT-1A and MT-2A levels were significantly higher in cadmium-treated lung cells (CT-LC) compared to controls, suggesting MT may increase its expression in the lung to provide protection from Cd-induced toxicity [338]. Once CT-LC acquire multiple tumour characteristics, this activates multiple signalling networks sequentially. Oxidative stress conditions are induced by antioxidative proteins Heme Oxygenase-1 (HO-1) and hypoxia inducible factor-1A (HIF-1α) [338]. HIF-1expression also occurs through Cdinduced reactive oxygen species formation associated with transformation of bronchial epithelial cells [339]. Also, Cd activates AKT, GSK-3β, and β-catenin signalling in BEAS2B human bronchial epithelial cells in a ROS-dependent manner [340]. Increased expression of the proliferation

Quinn et al.

marker PCNA and CyclinD1 in Cd treated cells suggest that Cd has mitogenic potential [318]. Chronic Cd treatment in human bronchial epithelial cells may impart DNA damage and decrease DNA repair capacity and genomic instability in an indirect manner [318]. Cd progressively reduced the mRNA and protein expression of these DNA repair genes in mouse testicular Leydig cells [341]. The long duration exposure to higher doses of cadmium, however, results in increased cell survival and acquisition of apoptotic resistance. Gene expression analysis by real-time polymerase chain reaction (PCR) revealed increased expression of the anti-apoptotic gene Bcl-2, whereas decreased expression of pro-apoptotic gene Bax. Decreased expression of genes for maintenance of DNA methylation, DNMT1, and DNA repair, OGG1 and MYH, was also observed in cells exposed to cadmium for 24h. The random amplified polymorphic DNA (RAPD) assay revealed genomic instability in cells with chronic exposure to cadmium. The findings of this study indicate that mouse testicular Leydig cells adapt to chronic cadmium exposure by increasing cell survival through increased expression of Bcl-2, and decreased expression of Bax. Increased proliferation of cells with genomic instability may result in malignant transformation, and therefore, could be a viable mechanism for cadmium-induced cancers. Chromium Chromium (Cr) VI is classed by the IARC Monograph as a group 1 pulmonary carcinogen [342]. It affects individuals primarily by inhalation exposure in an occupational setting [343]. The increased risk of lung and nasal cancer occurs mainly in workers exposed to hexavalent chromium dust, generated during the refining of chromite ore, production of chromate pigments and chromium plating [344, 345]. Chromium occurs primarily in the trivalent state, and as such cannot enter cells and thus has a low toxicity. Hexavalent chromium, Cr (VI) can enter the cell through membrane anionic transporters. It is a strong oxidizing agent, as well as a skin and mucous membrane irritant. At the cellular level, Cr exposure may lead to cell cycle arrest, apoptosis, premature terminal growth arrest, or neoplastic transformation. Cr-induced DNA inter-strand crosslinks, the tumour suppressor gene p53, and oxidative processes are some of the major factors that may play a role in determining the cell response to Cr exposure [346]. Since the IARC assessment [347], other epidemiological studies have been published [343, 345, 348-357]. Most confirm the LC risk with Cr (VI) exposure. More recent studies of this cohort have reconstructed individual exposure histories to CrVI based on species-specific air monitoring data [343, 358]. They attempted to quantify the potential LC risk contribution of smoking. Cumulative exposure to CrVI was significantly associated with increased LC risk. Additionally, study results indicated that smoking did not have a substantial effect on CrVI LC risk results (i.e., smoking and CrVI appeared to contribute independently to cancer risk) since risk estimates were not appreciably sensitive to smoking designation (for the 41% of the cohort that could be classified) [359]. In the regression model


adjusted for area, age, cigarette smoking, other types of tobacco and number of jobs held, increased ORs for LC were found for any and even low exposure to asbestos, silica and Ni-Cr, with positive trends for intensity of exposure. The numbers of NS exposed to Ni-Chromium was 1137 with an occupational risk of 1.0. They estimated a 1.18-fold increased risk for combined exposure to Ni and Cr among low-exposed workers (e.g. metal mechanics) instead of highly exposed workers in nickel refinery industries and chromate production [360]. A German study updated a 1982 report on mortality at two German chromate-producing factories [349]. The object was to establish whether the change-over to a production process using lime-free conversion of chromite ore, eliminating the formation of calcium chromate, had resulted in a reduction in bronchial carcinoma mortality among workers exposed for the first time after the change-over (completed in 1958 in Leverkusen and 1964 in Uerdingen). The SMR was considerably lower in the post-change cohort, dropping from 76 deaths from bronchial carcinoma out of 739 exposed workers to 9/678 workers, indicating the probable success of the process modification. Living close to a site where hexavalent chromium compounds were used poses no risk of LC for the residents [356]. There is a significant positive trend (p < 0.05) between cumulative hexavalent chromium exposure and risk of LC [343, 353, 361]. A study of two population-based case-control groups were conducted in Montreal for exposure to nickel, chromium VI, and cadmium compounds. Lung cancer ORs were increased only among former or non-smokers: 2.5 (95% CI 1.3-4.7) for nickel exposure, 2.4 (95% CI 1.2-4.8) for chromium VI, and 4.7 (95% CI 1.5-14.3) for cadmium. The metals did not increase the risk among smokers. While excess risks due to these metal compounds were barely discernible among smokers, carcinogenic effects were seen among non-smokers [362]. Nickel Nickel is also classed by the IARC Monograph as a group 1 pulmonary carcinogen [342]. It is still not known with certainty which forms of nickel pose the risk [363]. Evidence comes from studies of nickel refinery and leaching, calcining (to heat a substance to a high temperature but below the melting or fusing point, causing loss of moisture, reduction or oxidation, and the decomposition of carbonates and other compounds or to convert (liquid material, especially radioactive wastes) to granular solids by drying at very high temperatures), and sintering (to cause a substance [metallic powder, for example] to form a coherent mass by heating without melting.) There appears to be little or no detectable risk in most sectors of the nickel industry at current exposure levels. The general population risk from the extremely small concentrations detectable in ambient air is negligible. Nickel is used in many industrial and consumer products, including stainless steel, magnets, coinage, and special alloys. It is also used for plating and as a green tint in glass. Nickel is usually part of an alloy metal, and its chief use is in the nickel steels and nickel cast irons. Ni (Nickel)Cd is also used in battery manufacture [208].


97

The inhalation of nickel-containing dust has been associated with an increased risk of respiratory cancer in workplaces that process and refine sulphidic nickel mattes. Mattes are used in photography and special effects filmmaking to combine two or more image elements into a single, final image. Usually, mattes are used to combine a foreground image (such as actors on a set, or a spaceship) with a background image (a scenic vista, a field of stars and planets) [364]. These workers are exposed to mixtures of sulphidic, oxidic, water-soluble, and metallic forms of nickel. Because there is great complexity in the physical and chemical properties of nickel species, it is of interest which specific nickel forms are associated with carcinogenic risk. A bioavailability model for tumour induction by nickel has been proposed, based on the results of animal inhalation bioassays conducted on four nickel-containing substances. The nickel ion bioavailability model holds a nickelcontaining substance must release nickel ions that become bioavailable at the nucleus of epithelial respiratory cells for the substance to be carcinogenic. In addition, the carcinogenic potency of the substance is proportional to the degree to which the nickel ions are bioavailable at that site. This hypothesis updates the nickel ion theory, which holds that exposure to any nickel-containing substance leads to an increased cancer risk. The bioavailability of nickel ions from nickel-containing substances depends on their respiratory toxicity, clearance, intracellular uptake, and both extracellular and intracellular dissolution. Although there are some data gaps, a weight-of-evidence evaluation indicates the nickel ion bioavailability model may explain the existing animal and in vitro data better than the nickel ion theory. Epidemiological data are not sufficiently robust for determining which model is most appropriate, but are consistent with the nickel ion bioavailability model. Information on nickel bioavailability should be incorporated into future risk assessments [365]. A large epidemiologic report on cancer mortality in 10 cohorts of occupationally exposed workers [366] showed the mortality from lung cancer was associated with exposure to high levels of oxidic nickel compounds, exposure to sulphidic nickel in combination with oxidic nickel, and exposure to water soluble nickel, alone or together with less soluble compounds. Grimsrud et al. showed a dose-related association between LC and cumulative exposure to water soluble nickel compounds [363]. In a case-control study of Norwegian nickel refinery workers, the authors examined dose-related associations between LC and cumulative exposure to four forms of nickel: water soluble, sulphidic, oxidic, and metallic. A job-exposure matrix was based on personal measurements of total nickel in air and the quantification of the four forms of nickel in dusts and aerosols. A dose-related effect was seen for water-soluble nickel (OR, 1.7; 95% confidence interval CI 1.3-2.2). The study showed a doserelated association between LC and exposure to watersoluble nickel. The less soluble forms possibly contributed to an elevated risk, but the authors were not able to identify separate effects, and no dose-dependent risk was observed. Chronic exposure to nickel compounds has historically been associated with an increased incidence of lung and

98


nasal cancers. The oncogenic processes however remain poorly understood. Increased activation of signalling pathways are often observed. Signalling molecules including EGFR, phosphatidyl inositol 3-kinase (PI3K), Akt and mammalian target of rapamycin (mTOR) probably play a part in nickel carcinogenesis and cancer progression [367]. In addition there is experimental evidence that epigenetic alteration plays an important role in nickel-induced carcinogenesis. Multiple epigenetic mechanisms have been identified to mediate nickel-induced gene silencing. Nickel ions can induce heterochromatinisation by binding to DNAhistone complexes and initiating chromatin condensation. The enzymes required for establishing or removing epigenetic marks can be targeted by nickel, leading to altered DNA methylation and histone modification landscapes [368]. NiCl2 (nickel hydrochloride) induces down-regulation of E-cadherin by reactive oxygen species generation and promoter hypermethylation. Nickel induces epithelialmesenchymal transformation in bronchial epithelial cells [369]. Nickel levels in lung tissue adjacent to LC tumour were assessed in 189 individuals. Results of the analysis indicated an increased likelihood of p53 mutations (OR 3.25) in those with a high level of nickel compared to those who had a relatively lower level of nickel. The same publication reported reduced DNA repair capacity of LC cell lines in response to increasing nickel levels [370]. NATURALLY OCCURRING DISEASES Interstitial Pulmonary Fibrosis and Collagen Diseases Idiopathic pulmonary fibrosis (IPF) is a non-neoplastic pulmonary disease characterised by the formation of pulmonary scar tissue in the absence of any known provocation. IPF is rare and affects approximately 5 million persons worldwide. The prevalence is estimated to be slightly greater in men (20.2/100,000) than in women (13.2/100,000) and may be increasing [371]. The mean age at presentation is 66 years [372]. The differential diagnosis includes other connective tissue diseases, a forme fruste of autoimmune disorders, chronic hypersensitivity pneumonitis and other environmental (sometimes occupational) exposures. Thus there is crossover with the collagen diseases affecting the lung and IPF. The aetiology of the carcinoma in IPF is unclear but repetitive stimulation from chronic inflammation, alveolar epithelial injury and dysregulated repair induced by IPF cause genetic errors, which in turn may predispose to the development of LC [373, 374]. Loss of heterozygosity of a number of chromosomal loci associated with LC has been demonstrated in sputum samples from individuals diagnosed with IPF. A study of 52 samples revealed alterations at 1p34.3, 3p21.32-p21.1, 5q32-q33.1, 9p21 and 17p13.1 which correspond to where the MYCL1, FHIT, SPARC, p16(Ink4) and TP53 genes have been mapped [375]. A quantitative assessment of the metaplastic epithelia in the honeycombed areas revealed squamous metaplasia (Fig. 7), which occurred more frequently in usual interstitial pneumonia (UIP) with lung cancer than in UIP without LC [376].

Quinn et al.

Fig. (7). Squamous metaplasia in idiopathic pulmonary fibrosis.

Lung cancer probably develops from a stepwise accumulation of multiple acquired genetic alterations [377]. Studies have demonstrated tumour suppressor p53 and p21 proteins are upregulated in the terminal airways and the alveoli of patients with IPF and ILD (interstitial lung disease) related to collagen vascular disease [378, 379]. Cigarette smoking is an established risk factor for IPF and can be a confounding factor for IPF and LC [380]. In IPF there is a 14-fold increase in the risk of LC [381]. Recent figures give the incidence as ranging from 9.8 to 38% [382]. In one series approximately half of the lung carcinomas were incidental findings [383]. Contradictory results come from a mortality data analysis in the United States from1979 through to 1991. This study showed that LC occurred less frequently among decedents with pulmonary fibrosis (4.8%) than in decedents with chronic obstructive lung disease (10.1%) and asbestosis (26.6%) [384]. However these figures are pre-chest CT and are likely to underestimate the figures. In addition the figures were derived from analysis of death certificate reports compiled by the National Center for Health Statistics- an inherent area of misdiagnosis. Furthermore the earlier studies quoted above will not have used the present classification which was published in 2002 [385]. The malignancy commonly arises in the lung periphery and the lower lobes, where the fibrosis is most prominent (Fig. 8). The association between connective tissue disease (CTD) and malignancy has been an area of debate. A study described the clinical features of LC associated with several CTDs. The authors reviewed the clinical features of 153 reported cases from 1944 to 2000. There were 82 females and 71 males, with a median age of 58. Histological types of LC were as follows, ‘broncholoalveolar cell’ carcinoma (39 cases) (all these cases would probably now be classified as ACA) [386], ACA [36], SqCC [28], SCLC [27], large cell carcinoma [6], others [8], and unknown [10]. There was a relationship between smoking and development of LC in patients with rheumatoid arthritis (RA) and polymyositis/ dermatomyositis (PM/DM). The majority of patients with progressive systemic sclerosis (PSS) who developed LC were female, with underlying interstitial fibrosis, and most


tumours were of ‘broncholoalveolar cell’ or ACA cell type (see above). Patient characteristics were significantly different among the various groups of CTD associated with LC [387]. For patients with dermatomyositis and polymyositis there is a well-documented association with a wide range of cancers [388].


99

in rheumatoid disease [391]. Lung cancer risk was also increased in systemic lupus erythematosus (SLE) [395]. Patients with rheumatoid disease, SLE and Sjögren’s syndrome have an increased risk of lymphoma [394, 396398]. In Sjögren’s syndrome the increased incidence of lymphoma has a RR as high as 44 [399].

Fig. (9). Scleroderma associated with lung cancer. Fig. (8). Lung cancer associated with idiopathic pulmonary fibrosis.

Data from several epidemiologic studies revealed an overall two-fold increase in cancer in scleroderma (Fig. 9). Both the diffuse and limited forms of scleroderma were associated with an increased risk of cancer [389]. The prevalence of cancer indifferent series of patients with scleroderma varies between 2.8% and 8.7%, with LC being the commonest [388]. Patients with scleroderma who smoke are seven times more likely to develop LC than non-smokers with scleroderma (p = 0.008) [390]. Smokers with scleroderma and LC were more likely to smoke larger quantities than controls with scleroderma who smoked [390]. In this last study [390], peripheral lung tumours occurred earlier than bronchogenic tumours in the course of scleroderma. The median time to cancer after onset of scleroderma symptoms was 25 years (range 1-39 years) for the bronchogenic tumours (small and squamous cell) and five years (range 1-33 years) for the peripheral tumours (large cell, BAC and ACA) (p = 0.05). In a recent, retrospective study of a large number of patients (31,064 patient-year follow-up) from a specialist centre in Seoul, South Korea, patients with scleroderma showed a higher risk of LC (standardised incidence ratio [SIR] 4.917, 95% CI 1.97710.131) [391]. The frequent use of immunosuppressive drugs in rheumatic disease may predispose to the development of cancer independently of the influence of underlying disease [392]. The increase in RR for LC in patients with scleroderma may be explained by scleroderma-induced genetic injury, overdiagnosis bias, or drug therapy for scleroderma. In rheumatoid disease there was also an increase in LC in some studies, with a RR as high as 1.5 [388]. In a metaanalytic study of rheumatoid and LC [393], and an international multi-centre cohort study [394]. However Chang et al. did not document an increased incidence of LC

Five of ten DM patients developed cancer within one year of their dermatomyositis diagnosis, with a mean time between DM diagnosis and cancer diagnosis of 2.0±2.1 years, as compared to 7.4±4.2 years, 8.9±4.8 years and 6.6±4.5 years for rheumatoid disease, lupus and scleroderma patients, respectively. The tight temporal relationship between DM and cancer suggests a DM subset might be triggered as an immunologic response to cancer, similar to that seen in scleroderma patients with autoantibodies against RNA polymerase III [391]. Potential confounders in all these studies are that the pulmonary fibrosis may be a side effect of methotrexate and can predispose a patient to developing LC. Cyclophosphamide can induce the development of secondary cancers [400]. In addition, the prolonged use of corticosteroids and immunosuppressants can increase infection rates, which may then increase cancer risk [401]. Human Immunodeficiency Virus While the rates of infectious lung diagnoses, particularly opportunistic infections, have declined with combination antiretroviral therapy (cART), non-infectious and nonacquired immunodeficiency syndrome (AIDS) related lung diseases, such as chronic obstructive pulmonary disease, pulmonary hypertension and LC are increasing and have greater incidence rates in HIV-infected compared with HIVuninfected persons [402, 403]. Nearly, 900,000 people in the United States are living with diagnosed human immunodeficiency virus (HIV) infection and therefore increased cancer risk. An estimated 7,760 (95% CI 7330-8320) cancers occurred in 2010 among HIV-infected people, of which 3,920 cancers (95% CI 3480-4470) or 50% (95% CI 4854%) were in excess of expected. Lung cancer accounted for 440 cases in this group with a 52% excess over the expected number [404].

100


Only one of 22 studies reported a non-significantly decreased risk for LC among HIV-infected patients compared to the general population (SIR 0.7). The remaining studies all reported an increased risk for LC associated with HIV infection; 17 studies found a statistically significant association of HIV with LC. Despite LC being the most common non-AIDS-defining cancer, individual studies were usually limited by small numbers of cases and the precision of the estimates was limited [405]. A retrospective analysis of the Texas Department of Health cancer and HIV registries from 1990 to 1995 found a 6.5-fold increase in the incidence of LC in HIV-infected people [406]. In addition, Engels et al. found patients younger than 50 years had a statistically significantly increased incidence of LC [407]. There are several reasons HIV-infected persons may be at increased risk for LC. Cigarette smoking is more prevalent in the HIV population than the general public [408]. Additionally, the immune system performs an important function in ridding the body of tumourigenic cells [409]. Immune deficiency related to HIV infection may diminish the immune response to malignant foci [410-413]. The expression of HIV proteins did not enhance lung tumourogenesis caused by two different tobacco carcinogens, suggesting that incompletely restored immunity and/or inflammation, which persisted in most HIV patients despite controlled viraemia, underlay the excess risk of LC. Adjuvant therapies that restore immunity and lower inflam-mation may decrease LC mortality in HIV patients [414]. Another proposed mechanism for the increased risk of LC in this population is that persistent HIV-associated immuno-suppression results in reduced immune surveillance and subsequent increased risk for malignancy [405]. The types of LC in HIV infection are divided into AIDSdefining malignancies of the lung and non-AIDS-defining malignancies of the lung. In the former group there are two tumours; Kaposi’s sarcoma (KS) (Fig. 10) and Non-Hodgkin lymphoma (NHL). HIV/AIDS-associated KS most commonly occurs in homosexual or bisexual men infected with the human herpes virus-8 (HHV-8), also known as the KSassociated herpes virus. While disease is rarely isolated to the lungs, pulmonary involvement occurs frequently in HIVinfected patients with extensive mucocutaneous disease. Pulmonary KS is present in approximately 30% of patients but the rates of clinical diagnosis prior to autopsy are highly disparate [405]. Epstein-Barr virus (EBV) has been associated with the majority of HIV-associated NHL cases. Primary pulmonary lymphoma is an infrequent cause of AIDS-related lymphoma. It is of the large-cell type with uniform EBV expression [415]. Rarely, immune compromised HIVinfected patients may manifest lymphoma of the pleura or other body cavities (primary effusion lymphoma) which are strongly associated with KS-associated herpes virus infection, in addition to EBV [416]. In a retrospective review of AIDS patients undergoing autopsy at two San Francisco hospitals from 1982-1991 [417], 5.8% of NHL patients had isolated pulmonary localisation without other systemic manifestations. The lung was the commonest organ involved in those with extranodal site of disease (70%). Lymphomatoid granulomatosis (LYG)

Quinn et al.

has been described in HIV patients [418]. As this an EBVdriven disease, the finding is not surprising. Adenocarcinoma histology accounted for 36 to 50% of cases and small cell histology for 5 to 13% of cases [419]. Giant cell carcinoma [420], SqCC [421], and pleural leiomyosarcoma in children with HIV have also been described [422]. The last tumour had EBV detected by in situ hybridisation (ISH) and polymerase chain reaction in nearly all cases [423-425].

Fig. (10). Kaposi’s sarcoma.

Epstein-Barr Virus (EBV) We have documented above the role of EBV in HIV/AIDS. It also plays a part in post-transplant lymphoproliferative disorder [426]. Lymphomatoid granulomatosis (LYG) is a very rare EBV-driven lymphoproliferative disease. The atypical lymphoid cells directly accumulate within affected tissues and clinically present in the form of infiltrative lesions. It is usually a progressive disorder that virtually always involves the lung and characteristically presents as bilateral pulmonary nodules. LYG is an angiocentric and angiodestructive B-cell lymphoproliferative disorder. Histological examination reveals EBV-positive B cells admixed with a prominent background of mononuclear cells comprised of T cells, plasma cells, and histiocytes. Abnormal areas frequently congregate in monomorphous nests or larger aggregates. The malignant B cells usually are medium to large in size and express CD20, latent membrane protein 1. They are variably positive for CD30 and usually negative for CD15 assisting in the exclusion of Hodgkin lymphoma. It is hypothesised these patients have dysregulated immune surveillance of EBV. Grading of the lesions is based on morphologic features and the number of EBVpositive B cells by ISH [427]. The median age in one large series was 46 years (M: F 2.2:1). All patients had lung involvement. All patients had past EBV exposure by serology but a low median EBV viral load and lesions positive for EBV. Grading was performed predominantly on the lung biopsy at diagnosis, as grades 1-3. Grading is important as it dictates treatment [427, 428]. LYG is a distinct entity that can usually be differentiated from other EBV-associated B-cell lymphoproliferative disorders on the basis of the combination of clinical presentation, histology, and EBV studies.


In grade 1 lesions, EBV-positive cells are infrequently detected; whereas in grade 2 lesions, they are readily detected and usually number 5 to 20 high-power fields-by ISH. Grade 3 lesions consist of large atypical B cells, and EBV positive cells are very numerous and easily identified. Distinguishing grade 3 (high grade) from grades 1 and 2 (low grade) is an important distinction, as grade 3 lesions are approached therapeutically like diffuse large B-cell lymphoma with immunochemotherapy [428]. No EBV was detected in a microarray series of squamous and adenocarcinoma of lung [429]. There were discrepancies between microarray- and real-time quantitative PCR based strategies which highlighted the difficulty of validating molecular markers of disease. Another associated primary LC associated with EBV is lymphoepithelioma-like carcinoma of the lung (LELC). The distinct clinical features of LELC include no significant predilection for sex, minimal association with a history of smoking, strong association with EBV in Asians, and a predilection for early or locally advanced stage of the disease [430, 431]. EBV-associated smooth muscle tumours are rare neoplasms. They mainly affect immunocompromised patients, as seen in HIV and their clinical presentation is variable depending on size and organ involvement. EBV-associated smooth muscle tumours are a specific subcategory occurring in AIDS or post-transplant patients. These tumours can have incomplete smooth muscle differentiation but show nuclear EBER as a diagnostic feature [432]. They are usually indolent although somewhat unpredictable in behaviour. Metastases are rare, and were observed in 4/51 AIDSassociated (8%) [433], and 1/19 cases (5%) in a mixed AIDS and transplantation-associated series [434]. Based on a large survey of AIDS-associated cases, mitotic rate does not appear to distinguish cases with malignantoutcome [433]. Histologically, the EBV-associated smooth muscle tumours have a spectrum from a well-differentiated conventional-appearing spindle cell smooth muscle tumours to those composed of ovoid cells within complete smooth muscle differentiation. Intratumoural T lymphocytes may be present and even prominent, but this is not a uniform feature. These tumours are smooth muscle actin positive but often desmin negative. Some have also been classified as myopericytomas, reflecting differentiation intermediate to smooth muscle cell and pericyte [435]. A diagnostic test is demonstration of EBV RNA by ISH, which highlights the tumour cell nuclei. Human Papilloma Virus (HPV) This virus is well known in the causation of cervical cancer but less well in the lung. There is a full discussion of this issue by Thunnissen in Spencer’s Pathology of the Lung [436]. Juvenile-onset laryngeal papillomatosis has been known for some time to be caused by HPV, types 6 and 11. Lung involvement is rare (3%) with malignant transformation in only 0.5% of the patients [437]. In these SqCCs, HPV types 11 [438, 439], and 16 [440] were identified. In mothers of patients with juvenile-onset laryngeal papillomatosis, indicators of HPV infection in the genital tract have been identified, providing some support for viral transmission [441]. It has been assumed the juvenile


101

HPV chest infection was transmitted from the mother’s genital tract. However, a description as to whether the same type of HPV virus was present in mother and child has not been documented [441]. The route of transmission of HPV to lung tissue is still unclear. HPV normally invades healthy tissue by direct mucosal contact and the virus probably reaches the lung via the blood. The possibility of transmission from the cervix to the oral cavity and onwards to the larynx and lung is also plausible. HPV has been identified in pulmonary squamous metaplasia and SqCC [442, 443]. Haematogenous spread of the virus from cervical infections (subtypes 16 and 18) rather than inhaled virus (subtypes 6 and 11) is postulated, but additional studies are necessary. Once HPV presents to the host cells, it is believed HPV attaches to those cells expressing heparin sulphates. These act as primary receptors for HPV and internalise the virus. HPV protein E6 interferes with p53 and E7 with retinoblastoma protein [444]. HPV E6 and E7 are also involved in a concomitant increased expression of autocrine and/or paracrine IL-6 and downstream Mcl-1, a Bcl-2 family member which has an anti-apoptotic effect [445]. Several other possible mechanisms of molecular pathogenesis of HPV-induced lung cancer are reported [446]. Taiwanese NS had a significantly high prevalence of HPV16 /18, suggesting HPV infection as a possible etiological agent of LC in NS [447]. EGFR mutations are more frequent in NS, women, Asian ethnicity, and those with ACA. These clinicopathological features are evident in this Taiwan study [447]. One study in Japanese patients with lung cancer shows a significant association between high-risk type HPVs and EGFR mutations [448]. Oestrogen contributes to a large extent to the onset of HPV infection and tumour progression [449]. Oestrogen signalling plays a biological role in both the epithelium and the mesenchyme in the lung and thus oestrogen could potentially promote lung cancer [450]. Cross-talk between ER and EGFR in head and neck cancers arbitrates this action, which is further validated by the colocalised membrane ER and EGFR in the lung tumours [451]. This ER-EGFR cross-talk could occur in the lung tissue, which may favour HPV persistence and malignant transformation [452]. Maybe on the basis of linkage with juvenile-onset laryngeal papillomatosis, human papilloma virus is perhaps the most likely virus to play a role in lung carcinogenesis. High risk HPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 68, 73, and 82 are considered oncogenic and responsible for most anogenital carcinomas [453]. In patients with cervical squamous carcinoma and adenocarcinomas, the risk of smoking-related cancers was increased for a second primary LC [454, 455]. The SIR in patients with cervical squamous carcinoma for a pulmonary ACA was 1.58 and for pulmonary SqCC 5.45, while in patients with cervical adenocarcinoma, the SIR for pulmonary ACA and SqCC were 1.96 and 3.04, respectively [455]. According to this study, smoking was not a co-factor for cervical carcinoma. The histological clustering is interesting: for women with cervical squamous carcinoma, the SIR was 10.9 during the first four years and then decreased. In women with cervical AC, the increased risk for pulmonary SqCC and ACA

102


Quinn et al.

remained for up to 20 years. A few patients are reported with carcinomas of the cervix and lung having an identical HPV subtype [456-458].

LIFESTYLES

A systematic search of PubMed for articles of HPV infection in human subjects with NSCLC up to September 2012 was done [452]. Smoking status was not fully reported in all studies, so this information was supplemented wherever possible. Differences in the distribution of patients with and without HPV infection were tested by the Chisquared test. The authors identified 46 eligible articles, including 23 from Asian countries (n=2337 NSCLC cases), 19 from European countries (n=1553) and four from North and South America (n=160). The HPV prevalence was 28.1% (95% CI 26.6-30.3%), 8.4% (95% CI 7.1-9.9%) and 21.3% (95% CI 15.2-28.4%), respectively. 11/23 studies from East Asia (N=1110) and 4/19 from Europe (n=569) provided information on smoking status. The number of NS was 392 patients (33.9%) in East Asia and 54 patients (14.8%) in Europe. The HPV prevalence in East Asian countries was similar between never and ever smokers (33.9% vs 39.2%, p = 0.080). The HPV prevalence rate was 33.9% in East Asia from eleven studies including 68.7% in Taiwan, 60.0% in Korea, 23.8% in the central part of China and 12.4% in Japan. In European countries the incidence was lower, between 0.1 and 10%.

In the USA, the number of male deaths due to lung cancer fell after 1990, and for women after 2002, a decrease attributed to a reduction in tobacco use [461, 462]. Regardless of smoking habits, environmental tobacco smoke, or second-hand smoke (SHS), remains a potential source of aerosolised carcinogens. In the USA, cigarettes are the commonest source of SHS. Other sources include pipes and cigars [463]. Cigarettes contain approximately 600 ingredients, however when burned the resulting smoke includes over 7000 chemicals [464]. It is known that at least 69 of these agents found in smoke cause cancer [5, 465, 466]. These include acetaldehyde, aromatic amines, arsenic, benzene, benzo [α]pyrene, beryllium, 1,3-butadiene, cadmium, chromium, cumene, ethylene oxide, formaldehyde, nickel, polonium-210, PAHs, tobacco-specific nitrosamines, vinyl chloride [5, 464-466].

When the analysis was limited to HPV types 16 and 18 which have higher oncogenic risk, a significantly higher prevalence was observed in Asia (23.1%, 95% CI 21.525.2%, n=2307) than in Europe (4.4%, 95% CI 3.5-5.2%, n=1434, p < 0.001) or America (15.6%, 95% CI 10.3-22.1%, n=160, p = 0.003). Apart from the world-wide regional difference, LC associated with HPV infection was not evenly distributed within Japan. It was particularly high in Okinawa (43.9%, 95% CI 37.7-50.2%, n=255), south of mainland Japan, but it was notably low in Tokyo (0.3%, 95% CI: 0.71.6%, n=341). Based on the literature confirming the presence of HPV in LC in NS, the virus plays a role in carcinogensis in the disease. Based on the published data, PCR was chosen as the primary HPV detection method in the big Japanese analysis [452]. Different primer systems targeting relatively conserved nucleotide sequences have been developed with the aim of detecting a wide spectrum of HPV types. The sets used most frequently are GP5+/6+, MY09/11, PGMY and SPF10 [459]. The second senses methods were used mainly in the studies quoted in their paper. Another potential limitation of this paper was the limited assessment of HPV subtypes. In the majority of studies in their analysis, the high-risk subtypes of HPV were focused on HPV 16 and 18. However, other high-risk HPV types were also reported: HPV 31 and 33 were detected in 11/14 studies and 8/21 studies, respectively. The higher prevalence of HPV 33 infection was reported in Korean lung cancer patients (31.3%, n=112) compared to other Asian and Western countries (3.1%, n=511) [460]. Since only limited data are available currently with regard to the prevalence of minor HPV infections, future studies assessing more comprehensive HPV infections are warranted.

Second-Hand Smoke

It has been estimated that a nonsmoker who lives with a smoker can have an increased risk of LC development, of up to 20-30% [463]. The length of exposure and number of cigarettes smoked by a spouse are proportionate to the acquired risk [467]. This was deduced as a result of a metaanalysis of 37 studies by Hackshaw et al. in 1997, and has since been widely cited. A follow-up examination of the same 37 studies used concluded that the increase in risk was 24% as reported (RR of 1.24, 95% CI 1.13-1.36, p < 0.001), but cautioned that there was a trend towards publication bias, with a reduction in risk from 24% to 15% (95% CI 3-28%, p = 0.014) on exclusion of 40% of the studies [468]. SHS increases the likelihood of LC for all those exposed, however this risk is higher for individuals of a younger age group [469]. Specifically, the risk of LC can double upon household exposure, during childhood and adolescence, of the equivalent of 25 smoker years [470]. The intake of SHS can also affect overall survival rates of early (resected Stage I or II) NSCLC and appears to be linked to duration of exposure [471]. An examination of histological subtypes associated with SHS exposure amongst NS reported that exposure predisposes to a higher risk of SCLC (OR 3.09, 95% CI 1.62-5.89), followed by large cell carcinoma (OR 1.48, 95% CI 0.89-2.45), SqCC (1.41, 95% CI 0.99-1.99) and ACA (OR 1.26, 95% CI 1.10-1.44) [472]. This was based on pooled data from 18 case-control studies in the ILCCO. Obesity A number of comparisons of body mass index (BMI) and LC risk have concluded there is an appreciable inverselyrelated risk present, i.e., as the BMI increases, the calculated HRs fall. This trend has been observed in current or former smokers, but not in NS [473, 474]. Although confounding effects by smoking have been suggested, the data suggest that other explanations merit consideration [475, 476]. Interestingly, a report from Finland concluded that although an elevated BMI in females was linked with decreased


overall cancer risk in smokers, the risk was increased in NS [477]. A retrospective review of 857 patients with LC noted the BMI was higher in male former/NS (28.8 kg/m2) than current/ever smokers (26.8 kg/m2) (p = 0.0005), suggesting non-smoking status can confer an association with obesity [478]. This tendency was also noted in a cohort of NS where BMI and waist circumference were assessed [479]. Physical activity is unrelated to LC risk in NS, and RR may be confounded by the effects of cigarette smoke amongst ever smokers [480]. Obesity is a component of metabolic syndrome, which has been linked with the aetiology of cancers, arising outside of the lung. However its role in LC is unclear [481, 482]. The presence of metabolic syndrome does not appear to have a significant impact on the prognosis of early stage NSCLC [483]. Diet Dietary factors, including vitamin intake and alcohol use, may play a role in the development of LC, although this is disputed [484]. NS tend to have a healthier diet than smokers, for example more fruits, fibre and Vitamin A [485], which could lead to confounding results. A comparison of NS with a regular intake of low-fat foods, fruits and vegetables, against those with a varied diet, indicated a reduction in risk for those with the healthier pattern of eating. This finding retained significance after adjustment for other factors, including SHS [486]. A mixed population by smoking status experienced a reduction in LC risk upon increased consumption of apples and pears, with a decrease amongst smokers who ingested root vegetables [487]. Cruciferous vegetables (plants with long narrow seed pods, e.g. cabbage, turnip, broccoli) may protect against the development of LC, in NS females (HR 0.59, 95% CI 0.40.87, p = 0.05) [488] and also in smokers [489]. Specific effects on overall risk have been recorded for Chinese cabbage (OR 0.53, 95% CI 0.34-0.86, p = 0.004), chives (OR 0.54, 95% CI 0.35-0.85, p = 0.006) carrots (OR 0.51, 95% CI 0.33-0.8, p = 0.006) and celery (OR 0.4, 95% CI 0.26-0.63, p < 0.0001) have been recorded on populations from northeast China [490]. Carrots were also highlighted as a dietary protective agent (RR 0.7, 95% CI 0.38-1.27, p = 0.05) by a study of NS from Stockholm [491]. This study also noted risk reduction on consumption of non-citrus fruits (RR 0.58, 95% CI 0.28-1.22), and a trend for reduced risk on increasing intake of beta-carotene, p = 0.07 [491]. It is possible that beta-carotene can negatively affect the risk of cancer whilst interacting with the effects of cigarette smoke; increased intake has been linked with higher cancer risk in smokers (HR 2.14, 95% CI 1.16-3.97), while reducing the risk in NS (HR 0.44, 95% CI 0.18-1.07) [492]. Consumption of red meat (OR 1.8, 95% CI 1.5-2.2) or processed meat (OR 1.7, 95% CI 1.4-2.1) can lead to an elevated risk of LC (p < 0.001 for both), with risk increasing in NS (OR 2.4, 95% CI 1.4-4.0 and 2.5, 95% CI 1.5-4.2 respectively, p = 0.001 for both) [493]. These findings, originating from a case-control study in Italy (1,903 cases and 2,073 controls), contrast with the protective effect of meat recorded by a group in Singapore, where over 70% of the meat ingested was white meat (chicken or fish) and the entire study population were NS [494]. Fish consumption


103

was inversely related to LC risk [494]. This data may relate to the reduction in smokers’ risk associated with increased vegetable fat consumption [495] and increased LC mortality linked with intake of saturated fat [496]. Both the high intake of alcohol (heavy drinking defined by ≥ 60 g/day, OR 1.44, 95% CI 1.01-2.07)) and abstinence (OR 1.42, 95% CI 1.032.01) have been linked with increased risk in smokers but not NS, implying that the influence of alcohol is modulated by smoking [497]. Beta-cryptoxanthin is a precursor of vitamin A and is classed as a carotenoid. Raising the intake of foods high in this phytochemical may reduce LC risk [498]. Evaluation of data from a cohort of postmenopausal women revealed reduced risk associated with vitamin D for NS (HR 0.37, 95% CI 0.18-0.77, p = 0.01), although there was no association with the total population [499]. Riboflavin (vitamin B2), necessary for specific oxidative-reductive enzyme reactions, can reduce LC risk in current smokers (HR 0.53, 95% CI 0.29-0.94) but not former smokers or NS [500]. Isoflavones, which are derived from soybeans, have been linked with a decreased risk of LC in NS. Over eleven years of follow-up, multivariate HR revealed an inverse association for Japanese NS men (HR 0.43, 95% CI: 0.210.90, p = 0.024) comparing highest with the lowest quartile of isoflavone intake. A non-significant trend was reported for NS women (HR 0.67, 95% CI: 0.41-1.10, p = 0.135) [501]. Marijuana Marijuana or cannabis is derived from the Cannabis sativa plant, and is usually smoked as a hand-rolled cigarette or “joint” [502]. The issue of whether smoking of cannabis (marijuana) can lead to LC is under current debate. Marijuana smoke includes many of the toxins present in cigarette smoke [503], and is linked with histological anomalies of the bronchial mucosa, oxidative stress, abnormal function of alveolar macrophages, and exposure to elevated levels of tar [504]. Under conditions of “heavy” use of cannabis (defined by at least fifty occasions of consumption), the risk of LC in males has been associated with a HR of 2.12 (95 % CI 1.08-4.14) over a 40-year follow-up period [503]. However, pooled analyses of data have failed to demonstrate an overall significant association [504-506]. Further studies allowing for confounding factors, such as tobacco smoke, are warranted. Bidi Bidi is a hand-rolled presentation of tobacco, available in India. The tobacco is wrapped in dried Tendu leaf, and is imported by other Asian countries [507, 508]. Unlike Western countries where there has been a transition to filtered cigarettes, and a relative increase in ACA, the patterns of inhalation and toxin intensity associated with bidi use have resulted in a persistent distribution of SqCC [509]. In keeping with a potentially increased risk of exposure to carcinogens, smoking of bidi has an associated greater risk of LC (OR 8.3,95% CI 3.7-7.3 versus OR 5.2, 95% CI 3.77.3) [510] than that related to cigarettes [510-512]. The incidence of LC is proportional to the amount of bidi smoked and the duration of the smoking habit (p < 0.001) [513]. A stratified analysis of a North Indian population indicated an

104


additive effect of CYP1A1 polymorphisms and bidi smoking for SqCC and SCLC [514]. Social Class Socioeconomic status (SES) has been associated with elevated likelihood of LC [515-520] and retains statistical significance after adjustment for smoking (OR 1.38) [521]. Despite these findings, there is some doubt regarding the true effect of SES as the extensive modelling of data incurporating the effects of smoking argues that SES has no valid impact on risk [522]. A prospective screening trial in the UK highlighted an inverse relationship between LC risk and SES, which was accompanied by an increased likelihood to respond to the questionnaire with increasing SES [523]. This may reflect the concept of LC as a disease that is preventable with increasing SES and education [524]. Treatment and survival outcomes may also be affected by economic status assessed by income [525, 526]. Hormonal Factors The ILCCO conducted a pooled analysis of six case control studies and found a reduced risk for LC associated with both oral contraceptive (OR 0.81, 95% CI 0.68-0.97), and hormone replacement therapy (HRT) use (OR 0.77, 95% CI 0.66-0.9) [527]. This was consistent with results of a meta-analysis of 25 studies which detected a reduced risk due to use of HRT (OR 0.91, 95% CI 0.83-0.99, p = 0.033), also seen in NS subgroups (OR 0.86, 95% CI 0.75-0.99, p = 0.042) [528]. However, females who had undergone artificial menopause could have an increased risk (OR 1.51, 95% CI 1.17-1.94, p = 0.001) [528]. Results of the EAGLE casecontrol study expand these findings, reporting reduced risks with factors associated with oestrogen production, including later age at menopause [529]. The effects of HRT could be explained by the expression of ERα and ERβ within lung tumour tissue [530]. Amongst post-menopausal women, HRT led to a decreased risk of NSCLC with cytoplasmic ERα expression (43%, 95% CI, 0.36-0.90), nuclear ERβ expression, (56%, 95% CI 0.26-0.75) or both (58%, 95% CI 0.24-0.74) [531]. ER receptor negative NSCLC was not associated with HRT [531]. The expression of ERα in ACA has been significantly associated with NS and the presence of an EGFR mutation [532]. CONCLUSION Clinical, geographic and molecular data now suggest that LC in NS could be categorised as a different disease from LC arising in current or former smokers. The risk of LC in NS varies according to the potential carcinogen being considered. There are specific agents, such as arsenic, linked with an elevated risk of LC, which may have been accidentally included within the confines of a selected geographic population, defined by the related exposure. Other exposures, such as inhaled aluminium or air pollution, have been defined as carcinogenic but not conclusively proven to cause LC. Hazards such as asbestos have been established as a causative agent of LC, with proven risk dependent on exposure and interactions with the effects of cigarette smoke. In such cases there may be associated genetic polymorphisms and/or characteristic somatic mutat-

Quinn et al.

ions that confer oncogenic properties and proliferative advantages. Hereditary features of genes influencing smoking behaviour and nicotine addiction (such as the variants of the nicotinic cholinergic receptor subunit 15q25.1-CHRNA3) may also play a role, and can potentially act as confounding factors [533]. The potential carcinogenic roles of viruses will be affected by additional variants, such as the state of the immune system and the virus isoforms considered. It is almost certain that the likelihood of developing LC, associated with a given agent, is dependent on a combination of factors including intensity and duration of exposure, genetic predisposition, race, gender and additional environmental factors. Furthermore, in many cases, neoplasia will arise as a result of more than one causative factor. The difficulties related to extracting the significance of each agent are compounded by the fact that exposure may not always be adequately reported or quantified. Under-reporting of exposure will act as a confounding factor in the estimation of relative risk. This is especially true of SHS, and also applies to the risk related to diseases, dietary or lifestyle choices. Ironically, the elucidation of molecular pathways relevant to NS will most likely require the application of a genetic tobacco-related signature to establish the presence or absence of environmental smoke. This may be possible with the advent of molecular signatures based upon mutational load, numbers of nucleotide transversions relative to transitions [534], and gene expression profiles [535]. Alternatively, it may be possible to screen for SHS exposure by measuring levels of cotinine (which results from nicotine metabolism) in body fluids [218]. Associations with neoplasia will be clarified with further study. Genetic alterations and individual molecular pathways can potentially be confirmed, and may be of particular importance with regard to molecules that are not involved in cigarette smoke metabolism [536]. As the spectrum of individuals diagnosed with LC expands beyond the cigarette smoker, studies with the statistical power to illuminate the precise aetiologies and genetic mechanisms in NS will be invaluable. ABBREVIATIONS AIDS

= Acquired immunodeficiency syndrome

ACA

= Adenocarcinoma

ADSC

= Adenosquamous carcinoma

alpha7-nAChR = Alpha7-nicotinic acetylcholine receptor As

= Arsenic

BCME

= Bis(chloromethyl) ether

BMI

= Body mass index

BOLD

= Burden of Obstructive Lung Disease

Cd

= Cadmium

CT-LC

= Cadmium treated lung cells

CO

= Carbon monoxide

CAREX

= Carcinogen exposure

CME

= Chloromethyl ether

Cr

= Chromium



OSHA

= Occupational Safety and Health Administration

O3

= Ozone

PCE

= Perchloroethylene

PI3K

= Phosphatidyl inositol 3-kinase

PM2.5 μM

= particulate matter with a diameter