Increase, Chromosomal aberrations leads to Increase, lung cancer

Biological Plausibility

The biological rationale linking CAs with lung cancer is strongly supported. There are many epidemiological studies that provide evidence of a link between increasing CAs and cancer incidence. Several published reports spanning over 22,000 study subjects across multiple European countries have examined the association between the presence of CAs in cultured blood lymphocytes and the incidence of cancer. In every cohort examined, the presence of CAs was predictive of cancer risk (Bonassi et al. 2000; Hagmar et al. 2004; Norppa et al. 2006; Boffetta et al. 2007; Bonassi et al. 2008). Although CSAs and CTAs both had predictive value, CSAs were considered to be slightly more indicative of cancer risk (Norppa et al. 2006).  Similarly, studies examining chromosomes in lymphocytes from lung cancer patients found significant increases in CTAs, CSAs, and overall CAs relative to lymphocytes from healthy controls. Furthermore, the CAs were shown to be significant predictors of lung cancer risk (Vodenkova et al. 2015). Analysis of MN and NPB levels within binucleated cells also found that these CAs were significantly increased in lung cancer patients relative to healthy controls (Lloyd et al. 2013; El-zein et al. 2014; El-zein et al. 2017), with very similar results for geographically-separated test and validation cohorts (El-zein et al. 2014).

Exposure to radiation has also been epidemiologically linked to the relationship between CAs and cancer. Studies of radon-exposed uranium miners have revealed evidence of an association between exposure to radon gas and an increased incidence of lung cancer (Roscoe et al. 1989; Tirmarchel et al. 1993; Smerhovsky et al. 2001; Smerhovsky et al. 2002; Vacquier et al. 2008; Walsh et al. 2010). Analysis of CAs in the blood lymphocytes of miners from the Czech Republic found that miners with higher levels of CAs had a significantly elevated risk of cancer (Smerhovsky et al. 2001; Smerhovsky et al. 2002). The results from these studies were likely not due to smoking status of the miners, as a study examining a cohort of 516 white, never-smoker American uranium miners found that the mortality rates from lung cancer were higher in the miners than in the general non-smoking population (Roscoe et al. 1989).

Beyond epidemiology, there are also many genetic and molecular studies that provide strong evidence of a relationship between CAs and cancer. A subset of these studies have investigated copy number variants (CNVs). Examination of CNVs and known cancer genes in a large population revealed that CNVs often overlap with cancer genes and thus have the potential to amplify carcinogenesis (Shlien and Malkin 2009; Ohshima et al. 2017)Moreover, using only CNV genetic information from a database, Zhang et al (2016) were able to categorize 3,480 samples into their respective cancer type based solely on the CNVs of the samples. This was accomplished by developing a panel of 19 discriminating genes that could predict cancer type with a high level of accuracy using only the CNV number. Interestingly, many of these discriminating genes have known associations with cancer or processes known to be important in cancer development (Zhang et al. 2016). Furthermore, cancer-prone individuals tend to have more CNV instability, which has been attributed to inherently less efficient DNA repair mechanisms (Shlien and Malkin 2009). In their 2013 review, Liu et al provided lists of cancer-related genes typically amplified by CNVs (ERBB2, EGFR, MYC, PIK3CA, IGF1R, FGFR1/2, KRAS, CDK4, CCDN1, MDM2, MET, and CDK6) and deleted by CNVs (RB1, PTEN, CDKN2A/B, ARID1A, MAPSK4, NF1, SMAD4, BRCA1/2, MSH2/6, DCC, and CDH1). There is also evidence associating CNVs to lung cancer specifically. Analysis of primary NSCLC samples revealed 27 chromosomal regions where CNVs were present in at least one third of the samples (Wrage et al. 2009). Furthermore, medically-relevant CNVs were found in 60% of lung cancer patients, encompassing genes such as TP53, BAP1, STK11, BRCA2, CDKN2A, and RB1 (Mukherjee et al. 2016). 

Likewise, lung cancer-specific studies have been performed to identify chromosomes most often affected by CNV gains and losses. Analysis of DNA from primary human lung tumours and early-passage primary cell lines established from human tumours revealed that gains were most frequently found at chromosomes 3q, 5p, 7p, and 8q, while losses were most frequent at chromosome 3p (Balsara et al. 1997). Separation of CNV analyses into squamous cell carcinoma and lung adenocarcinoma groups demonstrated differences between the two lung cancers (Petersen et al. 1997; Bjorkqvist et al. 1998; Feder et al. 1998; Massion et al. 2002). In general, CNV changes were present in 84% of squamous cell carcinoma samples, but only in 68% of adenocarcinomas (Bjorkqvist et al. 1998). In squamous cell carcinoma, the most frequent gain was found at chromosome 3q (Bjorkqvist et al. 1998), specifically at 3q26 (Massion et al. 2002); other common CNV gains were found at chromosomes 8q, 5p and 7p (Bjorkqvist et al. 1998). Losses in 3p were also more common in squamous cell carcinoma than adenocarcinoma (Feder et al. 1998). In adenocarcinoma, the most common documented CNV was a gain at chromosome 7p (Feder et al. 1998).  Gains in squamous cell carcinoma were often found in genes GLUT2, THRB, PIK3CA and BCL6, and losses in FHIT, EG9F2 and CACNAID. Interestingly, CNV gains affecting PIK3CA were correlated with increased activity of PKB in squamous cell carcinoma (Massion et al. 2002). A review by Knuutila et al (1999) summarizes DNA copy number losses found in 73 human tumour types, with results separated by chromosome number.

Loss of heterozygosity is also a common occurrence in cancer. Preneoplastic lesions from seven NSCLC tumours were histologically categorized and genetically analyzed. Consistent with above studies that revealed CNV losses at chromosome 3p, loss of heterozygosity was also common at the 3p locus. Percentages of 3p loss of heterozygosity increased from hyperplastic lesions (76%) to dysplastic lesions (86%) to carcinoma in situ (100%). Overall, cumulative loss of heterozygosity was nearly doubled in carcinoma in situ and invasive carcinoma lesions relative to preneoplastic hyperplasia and dysplasia lesions (Hung et al. 1995).

Other studies have revealed a link between gene rearrangements and cancer. Truncated tumour suppressor genes TP53, BRCA1 and BRAC2 have been reported in prostate cancer (Mao et al. 2011). In lung cancer, the gene ALK has been observed to undergo rearrangements, often in the form of gene fusions with EML4 (Sanders and Albitar 2010; Sasaki et al. 2010).  These ALK rearrangements often result in increased activity of ALK, higher activation of PI3K-AKT pathways, and ultimately an increased risk of tumourigenesis (Sanders and Albitar 2010). Another common example of a gene fusion is the Philadelphia chromosome, which is formed by a translocation between chromosome 9 and 22 and results in the fusion of BCR and ABL genes. The resulting BCR/ABL gene fusion product was found to be the cause of chronic myelogenous leukemia (reviewed by (Trask 2002). This fusion may be induced by stressors such as ionizing radiation; exposure of human leukemic promyelocytic cells (HL-60) to 5 Gy of gamma-ray radiation resulted in homologous BCR and ABL genes in closer proximity to each other and to the centre of the nucleus (Bartova et al. 2000).

Several rearrangements have also been significantly associated with lung cancer. A balanced translocation at chromosome 19 that results in overexpression of Notch3 in lung epithelial cells has been identified in a number of NSCLC lung cancer cell lines and tumours. This is significant, as Notch3 is not normally expressed in the cells of the lungs (Dang et al. 2000). In fact, transgenic mice engineered to overexpress Notch3 in the lung epithelium died at birth. Analysis of these embryos at embryonic day 18.5 revealed tissue abnormalities in locations where Notch3 mRNA was found, which suggests that overexpression of Notch3 in the lungs may play a role in lung tumourigenesis (Dang et al. 2003). Significant associations have been found between rearrangements in chromosome Xp and higher NSCLC tumour stage, as well as rearrangements in 17p and lower NSCLC tumour stage; 3p and 6q rearrangements were linked with better NSCLC survival (Feder et al. 1998).

CAs that affect pathways controlling cellular growth and apoptosis may promote the development of cancer. In some cases, CAs may alter the activity of proto-oncogenes or tumour suppressor genes (Mitelman et al. 1997; Albertson et al. 2003). Proto-oncogene regulation may be modified such that its gene product is overexpressed; alternatively, the product of the proto-oncogene itself may be affected, producing an abnormally-functioning protein. CAs that affect tumour suppressor genes may inactivate its expression; deletion of the chromosome housing the tumour suppressor gene(s) or unbalanced structural rearrangements may also prevent the expression of tumour suppressor genes (Mitelman et al. 1997). If these alterations enhance cell growth and/or inhibit apoptosis, the cell may become excessively proliferative and unresponsive to external environmental signals (Albertson et al. 2003). There are several pathways that could conceivably be pushed towards malignant transformation by the formation of CAs, including signalling pathways AKT-PI3K-mTOR and RAS-REF-MEK. If a CA occurs within gene(s) related to either of these pathways such that the activity is augmented, this may contribute to the development of a tumour (Sanders and Albitar 2010).

Other factors that may also contribute to increasing the CAs in a tumour include aberrant centromeres and telomerase deficiencies. In some cases, centromeres may become abnormally large due to aberrant amplification. These large centromeres may no longer separate the chromosomes appropriately during cell division, increasing the CA burden in the resulting daughter cells. In telomerase-deficient tumour cells that are proliferating but not being monitored closely, the telomeres may become abnormally short. This becomes an issue for cells that continue dividing because the chromosomes may become damaged during cell division, resulting in chromosomal fusions and breakages. Ultimately, this would also increase the CAs in the daughter cells (Albertson et al. 2003).

Ionizing radiation may also play a role in carcinogenesis. A series of studies focussed on irradiating human papillomavirus (HPV18)-immortalized human bronchial epithelial cells and transplanting the cells into nude mice. The transplantation of these irradiated cells resulted in tumour induction, an effect that was not found when unirradiated cells were transplanted into the mice (Hei et al. 1994). From these tumours, 6 different tumour cell lines were established and analyzed for cytogenetics. All of the lines were found to have CAs, and all harboured losses in genetic information (Weaver et al. 1997). Establishment of further tumour cell lines and their subsequent genetic analysis confirmed that there were CAs, especially in the form of deletions, that were common among the different tumour cell lines (Weaver et al. 2000).

Whether the CA is spontaneous or inherited may also be an important factor in the development of cancer. Non-clonal CAs, which are acquired spontaneously, promote genetic instability and are thought to confer a growth advantage. Ionizing radiation and carcinogens are two stressors that are thought to push the cell towards production of non-clonal CAs, which dominate during the pre-crisis stage of tumour development. After the tumour cells have passed the crisis stage and become immortal, clonal CAs (which are stable, inherited and recurrent in the cell population) dominate the CA landscape of the tumour. Clonal CAs are thought to confer a survival advantage to the cells. Overall, it is suggested that the shifting of equilibrium between non-clonal and clonal CAs is key in the initiation and progression of cancers (Heng, Stevens, et al. 2006; Heng, Bremer, et al. 2006). Interestingly, non-clonal CAs are affected by genotype. In both mouse embryonic stem cells and cultured lymphocytes that were lacking ATM, the spontaneous frequency of non-clonal CAs were significantly increased relative to wild-type cells; the same pattern was also observed in p53-/- cells from a human lung cancer cell line and an ovarian carcinoma cell line (Heng, Stevens, et al. 2006).

There is moderate empirical evidence supporting the relationship between the incidence of CAs and the development of lung cancer in the presence of ionizing radiation. The evidence presented below is summarized in table 11, here (click link). Radon gas exposure in particular is linked to this process, and there are several published reviews that provide evidence for associations between radon exposure and the appearance of CAs, and radon exposure and the incidence of lung cancer (Jostes 1996; Al-Zoughool and Krewski 2009; Robertson et al. 2013). Genetic abnormalities found in lung cancer that result in genomic instability are also discussed by Massion (2003). Overall, however, there is little empirical evidence available supporting a dose and incidence concordance, some empirical evidence supporting a temporal concordance, and little empirical evidence supporting essentiality for this KER.

Dose and Incidence Concordance

There is a lack of empirical evidence to show dose and incidence concordance between CAs and lung cancer, particularly in the field of ionizing radiation. As described above, numerous studies are available that provide strong evidence linking CAs to cancer incidence, radiation exposure to CA frequency, and radiation exposure to cancer incidence. However, there is a lack of studies that assess whether increasing doses of a stressor, such as ionizing radiation, translate into dose-dependent changes in CA frequencies and dose-dependent changes in cancer incidences. Attempts were thus made to locate studies using similar radiological and biological conditions that assessed CA frequency and/or cancer incidence in response to a stressor.

Evidence from several epidemiological studies suggests a dose/incidence concordance between the appearance of CAs and lung cancer incidence upon radiation exposure. In humans, this association has been studied in different cohorts of uranium miners that were occupationally exposed to radon in the 1900s. (It is important to note that radon exposure was reported as working level months (WLM). One WLM is calculated based on 170 hours of exposure to one working level (WL), where 1 WL refers to the equivalent of 1.3 x10⁵ MeV of alpha particle energy in 1 L of air. Damage from 1 WLM is thought to be equivalent to 0.8 - 10.0 mGy (Jostes 1996); this corresponds to 100 - 1250 WLM/Gy.) One of these uranium miner studies examined the relationship between radon exposure, CAs and lung cancer. In a cohort consisting of 225 radon-exposed miners from the Czech Republic, 1,323 cytogenetic assays were performed and 20 cases of respiratory and intrathoracic organ cancers were recorded. Over the course of their employment, mine workers were estimated to be exposed to 1.7 - 662.3 WLM, with approximately one-third of miners exposed to doses above 80 WLM. There were significant associations found between the radiation dose and both the percentage of aberrant cells and frequency of chromatid breaks. Furthermore, an increased risk of lung cancer was revealed in subjects with high frequencies of CAs. Radon exposure was also found to be a significant predictor of lung cancer incidence (Smerhovsky et al. 2002). In addition, other studies examining uranium miners from different countries found a significant association between the cumulative radon exposure and the risk of lung cancer (Tirmarchel et al. 1993; Vacquier et al. 2008; Walsh et al. 2010).

Evidence from in vivo and in vitro studies has also revealed similar associations. In a study examining mouse bronchial epithelial cells for CAs, 1 Gy of X-ray radiation was found to induce a significant increase in the percentage of binucleated cells with MN relative to unirradiated controls (Werner et al. 2017). Similarly, rats irradiated with 1 Gy of thoracic X-rays between 1 and 15 weeks of age were found to develop significantly more lung tumours than unirradiated controls (Yamada et al. 2017). Several studies using lung and non-lung cell lines have also shown that a dose-dependent increase in CAs occurs with increasing radiation doses of X-rays between 0 and 5 Gy (Yamada et al. 2002) and alpha particles between 0 and 2.23 Gy (Nagasawa et al. 1990; Deshpande et al. 1996; Yamada et al. 2002; Stevens et al. 2014). Coinciding with these results, oncogenic transformations were found in non-lung cell lines irradiated with similar radiation doses; specifically, dose-dependent increases in oncogenic transformations were evident between 0 and 2.5 Gy of both X-rays (Robertson et al. 1983) and alpha particles (Robertson et al. 1983; Miller et al. 1996). Relative to X-ray exposed cells, those exposed to alpha particles had more MN accumulated per Gy (Yamada et al. 2002) and more oncogenic transformations (Robertson et al. 1983). Likewise,  in vivo  studies demonstrated a dose-dependent increase in MN in lung fibroblasts isolated from Wistar rats exposed to 0 - 11.3 Gy of gamma-ray radiation (Brooks et al. 1995), Wistar rats exposed to 0 - 323 WLM of radon, and Syrian hamsters exposed to 0 - 278 WLM of radon (Khan et al. 1995). When the incidence of lung carcinomas was examined in Sprague-Dawley rats exposed to radon and radon progeny at exposure levels similar to uranium miners, there was a dose-dependent increase in lung carcinomas between 25 and 3000 WLM (Monchaux et al. 1994). 

In further support of this dose-dependency between CAs and lung cancer, analyses of lung tissue with varying levels of tumourigenesis exhibited corresponding accumulations of CAs. In a Kras^LA2 mouse model of lung cancer, tumours collected when the mice were 6 months of age were categorized according to size. Genomic instability in the form of CNVs significantly increased with increasing tumour size; this was especially true in chromosome 6, which houses the KRAS gene (To et al. 2011). Similarly, analysis of human lung tissue categorized according to the level of damage (normal epithelium, hyperplasia, metaplasia, dysplasia, carcinoma in situ or invasive carcinoma) found that the loss of heterozygosity was increased with higher levels of tissue damage (Thibervile et al. 1995; Wistuba et al. 1999). These findings were especially pronounced in the chromosome 3p region (Wistuba et al. 1999), specifically at 3p21-22 (Thibervile et al. 1995). Interestingly, a review by Zabarovsky (2002) suggests that there may be multiple tumour suppressor genes at chromosome 3p, which are thought to play an important role in carcinogenesis (Zabarovsky et al. 2002). Loss of heterozygosity was also found commonly at 9p21-22 and 5q21 (Thibervile et al. 1995).

Temporal Concordance

There is some empirical evidence of temporal concordance between CA frequency and lung cancer incidence after exposure to ionizing radiation. With respect to the time of irradiation, CAs have been shown to occur prior to lung tumourigenesis. Results from a number of different studies found that an increased CA burden was evident within hours or days of irradiation (Nagasawa et al. 1990; Khan et al. 1995; Deshpande et al. 1996; Yamada et al. 2002; Stevens et al. 2014; Werner et al. 2017). The development of cancer, however, was a longer process. In vitro oncogenic transformations were documented weeks after irradiation (Robertson et al. 1983; Miller et al. 1996), while in vivo lung tumours were not detected until months or years after the radiation exposure (Tirmarchel et al. 1993; Yamada et al. 2017). In terms of examining increased CAs and cancer incidence directly, injection of a CA-carrying agent into mice was shown to induce cancer within 21 - 31 days of the injection (Pear et al. 1998; Kuramochi et al. 2001).

Essentiality

There are few studies available that demonstrate the essentiality of CAs for the induction of lung cancer. However, two agonist-type studies were found that supported the relationship between CAs and cancer, though they were not specific to the lung. In the first study, addition of a known pulmonary carcinogen to cultures of peripheral blood lymphocytes from both lung cancer patients and healthy controls resulted in significantly increased MN, NPBs and nuclear buds relative to the respective untreated cultures (Lloyd et al. 2013). The second study demonstrated a clear relationship between the BCR/ABL translocation and chronic myelogenous leukemia. In this study, BALB/c mice that were lethally irradiated received a bone marrow transplant containing retroviruses carrying the BCR/ABL translocation. Within 21 - 31 days of the transplant, all of the infected mice were found to have a disease equivalent to the human chronic myelogenous leukemia (Pear et al. 1998). 

A further study manipulated TSCL1 dynamics in a xenograft mouse model. The human lung cancer cell line, A549, harbours a loss of heterozygosity at chromosome 11, which results in highly reduced levels of TSCL1. Upon injection of these cells into BALB/c mice, tumours were detectable at the injection sites by 3 weeks post-injection. In an effort to correct this defect, mini genes were engineered to carry a full-length TSCL1 gene and transfected into A549 cells which were then injected into mice. Although tumours still developed, they were fewer in number and slower growing (Kuramochi et al. 2001). Thus correction of one CA may have a measurable effect on cancer progression.

Uncertainties and inconsistencies in this KER are as follows:

1. CNVs are often difficult to detect in cancer cells, even with current advances in next generation sequencing. This is due to the sheer number of CNVs that could possibly be present within one tumour; the unknown ratio of cancer cells and healthy cells within a tumour sample; the unknown ploidy of tumours; and the possible presence of multiple clones in one tumour, including possible low-number subclones that may be difficult to detect (Liu et al. 2013).  
2. In some studies, smoking does not affect the CA-cancer relationship (Bonassi et al. 2000; Bonassi et al. 2008; El-zein et al. 2014; Vodenkova et al. 2015; El-zein et al. 2017), but it does have a significant effect in other studies (Paik et al. 2012; Lloyd et al. 2013; Minina et al. 2017).
3. In a study examining MN in lung fibroblasts isolated from Wistar rats and Syrian hamsters exposed to radon, Syrian hamsters were found to have a significantly increased rate of MN per 1000 bincleated cells per Gy relative to rats. According to the literature however, Wistar rats have a higher documented sensitivity to radon-induced lung cancer than Syrian hamsters (Khan et al. 1995).

There is evidence of a response-response relationship between radiation exposure and CAs in cells of the lung, and between radiation exposure and the risk of lung cancer in radon-exposed miners. In two different studies using lung fibroblasts isolated from irradiated rodents, there was a positive, linear, dose-dependent relationship found between the radiation dose and the number of MN (Brooks et al. 1995; Khan et al. 1995). A number of in vitro studies also confirmed the presence of a positive, linear dose-dependent relationship between the number of radiation-induced CAs and the radiation dose (Nagasawa et al. 1990; Yamada et al. 2002; Stevens et al. 2014). In studies examining mortality from lung cancer in radon-exposed uranium miners from France and Germany, there was a positive linear relationship between the radon exposure and risk of lung cancer mortality (Tirmarchel et al. 1993; Walsh et al. 2010). This relationship was found to be exponentially modified by the age at median exposure, the time since median exposure, and the radon exposure rate (Walsh et al. 2010). Furthermore, oncogenic transformations in C3H10T1/2 cells irradiated with alpha particles were found to increase in a positive, linear dose-dependent fashion (Miller et al. 1996).

There is evidence suggesting that time-related predictions can be made for CA incidence and the development of lung cancer after exposure to ionizing radiation. CAs have been demonstrated to occur within hours of irradiation and persist for days afterwards.  In mouse bronchial epithelial cells, 1 Gy of X-ray radiation induced a significant increase in the percentage of binucleated cells with MN by 24 hours post-irradiation. These levels remained significantly elevated at 48 hours and 72 hours post-irradiation, though there was a time-dependent decrease in the percentage of cells with CAs. By 7 days post-irradiation, these levels were no longer significantly different from controls (Werner et al. 2017). In a similar study, lung fibroblasts were isolated and cultured from Wistar rats, Syrian hamsters and Chinese hamsters after exposure to 323, 278 and 496 WLM of radon, respectively, at 0.2, 15, and 30 days post-exposure. In all species, MN levels were highest at 0.2 days post-irradiation, and decreased over 30 days. The MN levels in the irradiated fibroblasts, however, remained significantly elevated at all time points relative to unirradiated control cells (Khan et al. 1995). Other in vitro studies have shown the presence of CAs within 13 - 82 hours post-irradiation (Nagasawa et al. 1990; Deshpande et al. 1996; Yamada et al. 2002; Stevens et al. 2014). It was noted in one study that the number or sister chromatid exchanges per cell were significantly higher than non-irradiated control cells at 72 hr post-irradiation, but these levels did not change appreciably at 74, 76, 78 or 82 hours post-irradiation (Deshpande et al. 1996).

In comparison to the time between radiation exposure and CA detection, there is a much longer gap between radiation exposure and the incidence of lung cancer. Oncogenic transformations in fibroblasts irradiated with alpha particles or X-rays were present 4 - 8 weeks after radiation exposure (Robertson et al. 1983; Miller et al. 1996). In vivo irradiation of 1 week-, 5 week- and 15 week-old rats by 1 Gy of thoracic X-rays was found to induce lung tumours months to years after the radiation treatment, with the highest risk for lung tumours found in rats that died between 600 and 900 days of age (Yamada et al. 2017). Similarly, French uranium miners exposed to radon and radon progeny for a minimum of two years were diagnosed at least 10 years after the first radon exposure (Tirmarchel et al. 1993).

Furthermore, direct injection of a CA into mice has also been shown to result in cancer several weeks after the CA administration. Injection of tumourigenic A549 cells that harbour a loss of heterozygosity at chromosome 11 resulted in tumour growth 3 weeks after injection (Kuramochi et al. 2001). Similarly, administration of the BCR/ABL translocation resulted in the mouse equivalent of chronic myelogenous leukemia by 21 - 31 days post-injection (Pear et al. 1998).

Not identified.