Lung cancer family history and exposure to occupational/domestic coal combustion contribute to variations in clinicopathologic features and gene fusion patterns in non-small cell lung cancer.

BACKGROUND: Both genetic and environmental factors contribute to the development of cancer and its mutant spectrum. Lung cancer has familial aggregation. Lung cancer caused by non-tobacco factors has unique pathological and molecular characteristics. The interaction between genetic lung cancer susceptibility and carcinogens from coal burning remains complex and understudied.
METHODS: We selected 410 non-small cell lung cancer (NSCLC) patients with a family history of lung cancer (FLC) and exposure to coal combustion between 2014 and 2017. Clinicopathologic parameters were analyzed. Reverse transcription-PCR was performed to detect ALK, ROS1, RET, and NTRK1 rearrangement.
RESULTS: Among the 410 NSCLC patients, 192 had FLC and 204 (49.8%) were exposed to occupational or domestic coal combustion. FLC patients had the same characteristics regardless of gender and coal exposure: younger age, high female ratio, adenocarcinoma, increased metastasis, later stage at diagnosis, and higher frequency of gene fusion. Sixty-seven patients (16.3%) had gene rearrangement: 51 (12.4%) harbored EML4-ALK fusions and 16 ROS1 fusions (3.9%). The highest gene fusion rate (35.1%, 33/94) occurred in patients with both FLC and high tobacco and coal exposure. ALK fusions and total gene rearrangement were closely associated with women, never smokers, younger age, FLC, and coal exposure.
CONCLUSION: FLC and exposure to coal combustion have an important impact on the clinicopathological characteristics and gene fusion mode of NSCLC, particularly in cases of higher levels of carcinogens, and genetic susceptibility has a greater impact. Our findings may help evaluate the effect of FLC and coal exposure on the pathogenesis of lung cancer.

Introduction

Lung cancer has been the most common cancer worldwide for decades.1, 2 Investigations conducted among different populations have found significant differences in the epidemiological, clinical, and molecular characteristics of lung cancer patients, especially non‐smokers, and that natural genetic susceptibility and fossil fuel use are important factors in the etiology of lung cancer.2, 3, 4, 5, 6, 7 An epidemiological survey proved that lung cancer has familial aggregation, after adjusting for tobacco smoking and other environmental factors.8, 9, 10, 11 Most studies have found an increased risk associated with a family history of lung cancer (FLC), which varies 1.3–3.5‐fold.8, 9, 10, 11, 12, 13, 14 Unlike other typical familial cancers, such as breast and colorectal cancer, genetic susceptibility to lung cancer has not been studied. Lung cancer has a range of well‐known risk factors, such as smoking, mining, shipbuilding, industrial construction, and the occupational hazards of petroleum refining.11 Globally, approximately 53% of lung cancer cases in women and 15% in men are not related to smoking.5 Statistics suggest that approximately 3 billion people on earth use coal or biomass for cooking and heating purposes.15, 16 This practice poses a long‐term risk of the development of cardiovascular and respiratory diseases, including lung cancer.3, 15, 16, 17 Evidence suggests that lung cancer caused by other non‐tobacco factors has unique clinical and histological characteristics compared to lung cancer caused by smoking.4, 5, 6, 7

Notably, Xuanwei, Yunnan Province, China, has long been the focus of large‐scale epidemiological studies, reporting some of the highest rates of lung cancer in the world, especially among people who have never smoked.7, 17, 18 This has partially been attributed to coal combustion.3, 16 The incidence of lung cancer in these rural areas also shows characteristics of family aggregation.17 The shared environment can partly explain this phenomenon, but the influence of a genetic background of susceptibility cannot be ignored.

Patients from this region could represent a good model to study the intricate mechanisms of lung cancer etiology. Although much work has been done, the dynamic and complex interactions between environmental carcinogens and human genetic background remain a mystery.4, 19 Research has suggested that non‐smokers and patients exposed to coal combustion exhibit unique driver mutation patterns and fusion genes, such as ALK and ROS1, which exhibit special signatures in NSCLC.4, 6, 7, 19 To better characterize the interaction between these two key players, this study investigated the clinical characteristics and genetic rearrangements of people with FLC and a history of exposure to coal combustion, one representing genetic causes and the other key environmental factors. Compared to fluorescence in situ hybridization (FISH) and immunohistochemistry (IHC), reverse transcription (RT)‐PCR can detect fusion genes, including variant types. It is also a fast, sensitive, high‐flux method.20, 21, 22

Methods

Patients and tissue samples

Patients with pathologically confirmed lung cancer in Ward One at the Department of Thoracic Surgery, Yunnan Cancer Hospital between February 2014 and June 2017 were included. To study the effect of FLC and exposure to occupational or household coal combustion, we focused on patients with both characteristics, based on the following criteria: (i) FLC was defined as individuals with three or more first‐degree relatives affected by lung cancer (n = 192); and (ii) patients without FLC were enrolled as a control, defined as individuals with no reported cancer diagnosed among first‐degree relatives for three generations. Eventually, 218 patients met the criteria. Patients’ exposure to occupational or domestic coal combustion was also documented. Patients with occupational coal combustion exposure, mainly residents from Xuanwei and other rural areas, include farmers who use coal to heat agricultural facilities and workers, such as furnacemen, exposed to carcinogens related to burning coal. Residents use coal domestically to cook and heat their homes. Smoking history and occupational or domestic coal exposure was obtained based on self‐report and confirmed by personal medical records.

Clinical and pathologic data were obtained from the hospital cooperated databank (https://www.linkdoc.com), which included age, gender, histologic type, family history, smoking status, occupational exposure, and critical environmental risk factors. Tumor node metastasis (TNM) stage was reviewed according to the 8th edition of the International Association for the Study of Lung Cancer (IASLC) staging system. The majority of patients enrolled had adenocarcinoma (AD) and squamous cell carcinoma (SCC); other histology types were not included as few patients met the inclusion criteria.

Lung cancer tissues were stored in RNAlater (Sigma, St. Louis, MO, USA) immediately after surgery. A slide was cut from each sample for hematoxylin and eosin staining. Those containing > 70% cancer cells and < 10% necrosis were enrolled. The ethical committees of Yunnan Cancer Hospital approved the study. All patients signed informed consent.

RNA extraction and reverse transcription‐PCR

Total RNA was extracted using a TRIzol Reagent Kit (Invitrogen, Carlsbad, CA, USA), and then incubated with RNase‐free DNase I (Qiagen, Hilden, Germany) to remove contaminating DNA. RT was performed using a reverse transcription kit (Promega, Madison, WI, USA) to generate complementary DNA as PCR templates. RT‐PCR was performed using Sigma Taq (Sigma).

PCR products were first detected by electrophoresis and those with proper bands were sequenced by BGI Tech (Shenzhen, China; http://www.bgitechsolutions.com) for confirmation. The nucleotide sequences were verified using the BLAST program (National Center for Biotechnology Information, Bethesda, MD, USA). All positive cases were confirmed by another independent PCR reaction.

ALK rearrangement

Primers were selected based on reference to identify different EML4‐ALK variants (Table S1).22, 23, 24, 25 The thermal cycle conditions were as follows: EML4‐ALK variant 1: 95°C, 5 minutes, 40 cycles of 95°C, 30 seconds, 55°C, 30 seconds, 72°C, 1 minute; and EML4‐ALK variants 2–7: 95°C, 10 minutes, 40 cycles of 95°C, 30 seconds, 66°C, 30 seconds, 72°C, 2.5 minutes.

Other ALK rearrangements were also examined in the negative samples for EML4‐ALK, including TFG‐ALK, KLC1‐ALK, and KIF5B‐ALK. The primers and PCR conditions were designed based on previous reports (Table S1):26, 27, 28 KIF5B‐ALK: 95°C, 5 minutes, 40 cycles of 95°C, 30 seconds, 50°C, 30 seconds, 72°C, 2.5 minutes; KLC1‐ALK: 95°C, 5 minutes, 40 cycles of 95°C, 40 seconds, 55°C, 40 seconds, 72°C, 3 minutes; and TFG‐ALK: 95°C, 5 minutes, 40 cycles of 94°C, 40 seconds, 65°C, 40 seconds, 72°C, 1.5 minutes.

ROS1 rearrangement

Different ROS1 fusions were examined, including CD74‐ROS1, TPM3‐ROS1, EZR‐ROS1, SLC34A2‐ROS1, LRIG3‐ROS1, GOPC‐ROS1, and SDC4‐ROS1. The primers and PCR conditions were designed based on previous studies (Table S1).29, 30 The PCR conditions were as follows: 95°C, 5 minutes, 10 cycles of touchdown PCR (annealing from 63 to 58°C with a 0.5 decrease each cycle) and 30 cycles of 95°C, 40 seconds, 58°C, 40 seconds, 72°C, 1 minute, with a final extension of 72°C for 5 minutes.

RET rearrangement

The potential existence of CCDC6‐RET and KIF5B‐RET were also examined. The primers and PCR conditions were set according to a previous study (Table S1).31 The program for CCDC6‐RET was: 95°C, 4 minutes, 40 cycles of 95°C, 30 seconds, 60°C, 30 seconds, 72°C, 30 seconds. The program for KIF5B‐RET was: 95°C, 4 minutes, 40 cycles of 95°C, 30 seconds, 62°C, 30 seconds, 72°C, 30 seconds.

NTRK1 rearrangement

The rarely reported CD74‐NTRK1 fusion was also examined. The primers and PCR conditions were set based on a previous report (Table S1).32 The PCR conditions were as follows: 95°C, 4 minutes, 10 cycles of touchdown PCR (annealing from 62 to 57°C with a 0.5 decrease each cycle) and 30 cycles of 95°C, 40 seconds, 57°C, 40 seconds, 72°C, 1 minute.

Statistical analysis

Chi‐square and Fischer's exact tests were used to analyze the association of clinicopathological parameters with FLC and occupational/domestic coal exposure. SPSS version 17.0 (SPSS, Chicago, IL, USA) was used. Statistical significance was set at P < 0.05 (two‐sided P value).

Results

Clinicopathological features of the study population

In total, 410 lung cancer patients were included. The clinicopathological characteristics are shown in Tables 1 and 2. There were 277 men and 133 women at an average age of 59 (range 28–84) years; 192 patients (male: 114, female: 78) had FLC while 218 (male: 163, female: 55) reported no FLC. The histological subtypes included 326 (79.5%) AD and 84 (20.5%) SCC. A total of 204 (49.8%; male: 177, female: 27) patients were exposed to occupational or domestic coal combustion and 231 (56.3%, all male) were smokers, among whom 173 (42.2%) were both smokers and exposed to coal combustion. To better evaluate major carcinogen exposure in the sample population, both factors were considered together. The definition of high exposure was either being a smoker or exposed to coal combustion, while low exposure was defined as a non‐smoker not exposed to coal combustion. A total of 262 (63.9%, male: 235, female: 27) subjects were classified in the high carcinogen exposure group.

Table 1

Clinical characteristics of 410 NSCLC patients grouped by FLC and exposure level

		Total FLC Case (%)			High exposure group FLC Case (%)			Low exposure group FLC Case (%)
Variables	Total	Positive	Negative	P *	Positive	Negative	P *	Positive	Negative	P *
Total number of patients	410	192 (46.8)	218 (53.2)		94 (24.4)	168 (64.1)		98 (66.2)	50 (33.8)
Gender				0.001			0			0
Male	277	114 (59.4)	163 (74.8)		75 (79.8)	160 (95.2)		39 (39.8)	3 (6.0)
Female	133	78 (40.6)	55 (25.2)		19 (20.2)	8 (4.8)		59 (60.2)	47 (94.0)
Average age: 59 years (range 28–84)				0			0			0.037
≤ 55 years	145	104 (54.2)	41 (18.8)		55 (58.5)	25 (14.9)		49 (50.0)	16 (32.0)
> 55 years	265	88 (45.8)	177 (81.2)		39 (41.5)	143 (85.1)		49 (50.0)	34 (68.0)
Histological type				0			0			0.550
Adenocarcinoma	326	180 (93.7)	146 (67.0)		84 (89.4)	96 (57.1)		96 (98.0)	50 (100)
Squamous cell carcinoma	84	12 (6.3)	72 (33.0)		10 (10.6)	72 (42.9)		2 (2.0)	0 (0)
T				0.138			0.54			0.25
T1	69	34 (17.7)	35 (16.1)		14 (14.9)	24 (14.3)		20 (20.4)	11 (22.0)
T2	178	75 (39.1)	103 (47.2)		36 (38.3)	77 (45.8)		39 (39.8)	26 (52.0)
T3	65	28 (14.6)	37 (17.0)		17 (18.1)	31 (18.5)		11 (11.2)	6 (12.0)
T4	98	55 (28.6)	43 (19.7)		27 (28.7)	36 (21.4)		28 (28.6)	7 (14.0)
N				0.033			0.21			0.18
N0	153	70 (36.5)	83 (38.1)		29 (30.9)	63 (37.5)		41 (41.8)	20 (40.0)
N1	33	10 (5.2)	23 (10.6)		6 (6.4)	18 (10.7)		4 (4.1)	5 (10.0)
N2	129	57 (29.7)	72 (33.0)		32 (34.0)	55 (32.7)		25 (25.5)	17 (34.0)
N3	95	55 (28.6)	40 (18.3)		27 (28.7)	32 (19.0)		28 (28.6)	8 (16.0)
M				0			0.007			0.004
M0	255	99 (54.4)	156 (71.6)		51 (54.3)	119 (70.8)		48 (49.0)	37 (74.0)
M1	155	93 (45.6)	62 (28.4)		43 (45.7)	49 (29.2)		50 (51.0)	13 (26.0)
Stage				0			0.037			0
I	98	45 (23.4)	53 (24.3)		16 (17.0)	42 (25.0)		29 (29.6)	11 (22.0)
II	51	17 (8.8)	34 (15.6)		8 (8.5)	25 (14.9)		9 (9.2)	9 (18.0)
III	106	37 (19.3)	69 (31.7)		27 (28.7)	52 (31.0)		10 (10.2)	17 (34.0)
IV	155	93 (48.4)	62 (28.4)		43 (45.7)	49 (29.2)		50 (51.0)	13 (26.0)

P value calculated by chi‐square or Fisher's exact test, when there is at least one cell with an expected count < 5.

High exposure is defined as being a smoker or exposed to coal use, while low exposure refers non‐smokers and no coal exposure. FLC, family history of lung cancer; NSCLC, non‐small cell lung cancer.

Table 2

Clinical characteristics of gene rearrangements detected by reverse transcription‐PCR in 410 NSCLC patients

		ALK rearrangements Case (%)			ROS1 rearrangements Case (%)			Total rearrangements Case (%)
Variables	Total	Positive	Negative	P *	Positive	Negative	P *	Positive	Negative	P *
Total number of patients	410	51 (12.4)	359 (87.6)		16 (3.9)	394 (96.1)		67 (16.3)	343 (83.7)
Gender				0			0			0
Male	277	22 (7.9)	255 (92.1)		3 (1.1)	274 (98.9)		25 (9.0)	252 (91.0)
Female	133	29 (21.8)	104 (78.2)		13 (9.8)	120 (90.2)		42 (31.6)	91 (68.4)
Average age: 59 years (range 28–84)				0.002			0.001			0
≤55 years	145	28 (19.3)	117 (80.7)		12 (8.3)	133 (91.7)		40 (27.6)	105 (72.4)
> 55 years	265	23 (8.7)	242 (91.3)		4 (1.5)	261 (98.5)		27 (10.2)	238 (89.8)
Histological type				0.006			0.05			0
Adenocarcinoma	326	48 (14.7)	278 (85.3)		16 (4.9)	310 (95.1)		64 (19.6)	262 (80.4)
Squamous cell carcinoma	84	3 (3.6)	81 (96.4)		0 (0)	84 (100)		3 (3.6)	81 (96.4)
Stage				0.013			0.87			0.014
I	98	15 (15.3)	83 (84.7)		5 (5.1)	93 (94.9)		20 (25.6)	78 (74.4)
II	51	6 (11.8)	45 (88.2)		2 (3.9)	49 (96.1)		8 (15.7)	43 (84.3)
III	106	4 (3.8)	102 (96.2)		3 (2.8)	103 (97.2)		7 (6.6)	99 (93.4)
IV	155	26 (16.8)	129 (83.2)		6 (3.9)	149 (96.1)		32 (20.6)	123 (79.4)
Smoking history				0			0.002			0
Yes (Current or ex‐smoker)	231	17 (7.4)	214 (92.6)		3 (1.3)	228 (98.7)		20 (8.7)	211 (91.3)
Never	179	34 (19.0)	145 (81.0)		13 (7.3)	166 (92.7)		47 (26.3)	132 (73.7)
Occupational or domestic coal use				0.022			0.98			0.041
Yes (current or ex‐user)	204	33 (16.2)	171 (83.8)		8 (3.9)	196 (96.1)		41 (20.1)	163 (79.9)
Never	206	18 (8.7)	188 (91.3)		8 (3.9)	198 (96.1)		26 (12.6)	180 (87.4)
Overall exposure level (tobacco and coal)				0.046			0.68			0.046
High (smoker or household coal‐user)	262	39 (14.9)	223 (85.1)		11 (4.2)	251 (95.8)		50 (19.1)	212 (80.9)
Low (never smoker & never coal‐user)	148	12 (8.1)	136 (91.9)		5 (3.4)	143 (96.6)		17 (11.5)	131 (88.5)
Family history of lung cancer				0.001			0.07			0
Present	192	35 (18.2)	157 (81.8)		11 (5.7)	181 (94.3)		46 (24.0)	146 (76.0)
Absent	218	16 (7.3)	202 (92.7)		5 (2.3)	213 (97.7)		21 (9.6)	197 (90.4)
Family history in high exposure group				0			0.06			0
Present	94	26 (27.7)	68 (72.3)		7 (7.4)	87 (92.6)		33 (35.1)	61 (64.9)
Absent	168	13 (7.7)	155 (92.3)		4 (2.4)	164 (97.6)		17 (10.1)	151 (89.9)
Family history in low exposure group				0.75			0.66			0.34
Present	98	9 (9.2)	89 (90.8)		4 (4.1)	94 (95.9)		13 (13.3)	85 (86.7)
Absent	50	3 (6.0)	47 (94.0)		1 (2.0)	49 (98.0)		4 (8.0)	46 (92.0)

P value calculated by chi‐square or Fisher's exact test, when there is at least one cell with an expected count < 5.

NSCLC, non‐small cell lung cancer.

Patients with a family history of lung cancer (FLC): Younger and with a higher female ratio

Significantly more women had FLC (P < 0.01). Because men dominated the smoking population, most women were never smokers. There were more men in the high exposure than in the low exposure group (Table 1, Fig 1a). Importantly, the distribution of patient age also reveals interesting variation in different subpopulations. The age structure curve of FLC patients increased compared to patients without FLC: there were significantly more individuals aged < 55 (P < 0.01) (Table 1, Fig 1b). A similar apparent age difference was observed in subgroups (Fig 1c). Men with FLC developed the disease much earlier than those without (median age: 55 vs. 64 years), while in women the age gap was smaller (median age: with FLC 53.5 vs. without FLC 57 years). On the other hand, if first divided by overall exposure (tobacco and coal), subjects in the same exposure group were still affected by FLC: FLC patients were younger than their counterparts. The age gap was much larger in the high exposure (median age: with FLC 53 vs. without FLC 63 years) compared to the low exposure group (median age: with FLC 55.5 vs. without FLC 58.5 years).

Figure 1

Patient age, gender, histology, and stage distribution. (a) Gender ratio in different subgroups. () Female, and () Male. (b) Patient age structure curves. Both fitted curves were 5th‐degree polynomial; the function for FLC patients is: y = −4E−07x5 + 0.0001x4−0.015x3 + 0.8543x2−23x + 234.51 (R2 = 0.715); and for patients without FLC is: y = 8E−07x5−0.0002x4 + 0.0228x3−1.1361x2 + 27.063x−247.71 (R2 = 0.8616) () Positive, and () Negative. (c) Patient age distribution in subgroups. Patient histology and tumor node metastasis stage divided by (d–h) exposure level () AD, and () SCC; () Stage I, () Stage II, () Stage III, and () Stage IV; () T1, () T2, () T3, and () T4; () N0, () N1, () N2, and () N3; () M0, and () M1 and (i–m) gender () AD, and () SCC; () Stage I, () Stage II, () Stage III, and () Stage IV; () T1, () T2, () T3, and () T4; () N0, () N1, () N2, and () N3; () M0, and () M1

FLC patients: Dominated by adenocarcinoma and a higher frequency of later stage diagnosis

Clinical characteristics were first divided according to gender and total exposure level, and then the effects of FLC were further compared between total and subgroup exposure level. AD was the major histological type in all subgroups (Table 1), but there were statistically more cases of SCC in subgroups negative for FLC, whether divided by gender or exposure (Fig 1d,i). There was a higher percentage of stage IV patients in the FLC subgroup (P < 0.05) within the same gender or exposure group (Fig 1e,j). FLC patients had more T4 and N3 disease, but only the N stage in total was statistically significant (Fig 1f,g,k,l). The greatest variation was in M stage, where FLC patients in all groups experienced significantly more distant metastasis (P < 0.05) (Fig 1h,m). Overall, patients with FLC had a higher risk of lymph node and other organ metastasis. The influence of FLC was consistent in the same gender or within the same exposure group.

ALK rearrangement: Associated with younger age, women, and never smokers

Of the total 410 patients, 51 (12.4%: 22 men 7.9%, 29 women 21.8%) had ALK rearrangement (Table 2). Patients with ALK fusion were younger (P < 0.01), with 28 (28/145, 19.3%) aged < 55 and 23 (23/265, 8.7%) aged > 55. The pathological diagnoses included 48 cases with AD and three cases with SCC (P < 0.01). The positive cases mainly occurred in stages I (15/51, 29.4%) and IV (26/51, 51.0%). Importantly, 4/15 (26.7%) patients in stage I were FLC positive, while 24/26 (92.3%) in stage IV were FLC positive. ALK rearrangement was significantly associated with female gender (P < 0.01) and non‐smoking status (P < 0.01). All 51 patients had EML4‐ALK fusion: 17 were variant 1, 6 were variant 2, and 28 were variant 3a/3b (Table S2).

ALK rearrangement: Associated with FLC and coal exposure

In this series, ALK rearrangement was strongly associated with FLC (P < 0.01): 35 patients (35/192, 18.2%) were positive (Table 2). Because genetic and environmental factors both contribute to gene mutation, subjects were further compared within the same exposure group. In the high exposure group, a high association existed between ALK fusion and FLC (P < 0.01). However, no significant association was found between ALK fusion and FLC in the low exposure group. In addition, ALK rearrangement was significantly associated with occupational or domestic coal combustion (P < 0.05): 33 patients (33/204, 16.2%) were current or ex‐coal users, while 18 (18/206, 8.7%) had never been exposed to coal combustion. When smoking and coal use were combined as overall exposure (tobacco and coal), the association still existed (P < 0.05): 39 (39/262, 14.9%) were either a coal user or a smoker, and 12 (12/148, 8.1%) were never‐coal users and never‐smokers.

ROS1 rearrangement: Associated with younger age, women, and non‐smokers

Among the 410 patients, 16 (3.9%: 3 men, 3/277, 1.1%; 13 women, 13/133, 9.8%) harbored ROS1 rearrangements (Table 2). A statistically significant association was found for female gender and non‐smoking. Most subjects with ROS1 fusion were also younger (P < 0.01): 12 (12/145, 8.3%) were aged < 55 years and only 4 (4/265, 1.5%) were aged > 55. All 16 cases were diagnosed as adenocarcinoma: 10 patients had CD74‐ROS1 (E6/E34) fusion and the other six had unknown type (Table S3); none had a co‐existing ALK rearrangement. Other ROS1 fusions were not identified. No statistically significant association was observed for stage, histology type, or exposure to occupational or domestic coal combustion. The association between ROS1 translocation and FLC was also not significant (P = 0.07): 11 (11/16, 68.8%) ROS1 fusion cases had FLC (Fig 2a,b).

Figure 2

Total gene fusion distribution based on family history of lung cancer (FLC) and overall exposure level. FLC positive and (a) high and (b) low exposure. FLC negative and (c) high and (d) low exposure. () ALK male, ( ) ALK female, () ROS1 male, and () ROS1 female

Other gene rearrangements

Although potential RET and NTRK1 rearrangement were analyzed in all samples, no positive case was identified. This may have been a result of sample size and selection standards. Detection of other gene rearrangements can be performed in a larger sample or in populations of a different composition.

Total gene fusion pattern: Associated with women, never smokers, younger age, FLC, and coal exposure

All detected gene rearrangements were combined for evaluation (Table 2). A total of 67 cases (67/410, 16.3%) were identified, including 64 AD and 3 SCC (P < 0.01). Most fusion cases occurred in stages I and IV (Fig 2). A statistically significant association was also found for female gender and never smokers (P < 0.01). The majority of gene fusions (40/67, 59.7%) occurred in patients aged < 55 years (P < 0.01). When evaluated by coal exposure status and overall exposure (tobacco and coal), the association was significant (P < 0.05). A total gene fusion event was statistically associated with FLC (P < 0.01): 46 patients (46/67, 68.7%) were positive for FLC, and the association was stronger in the high exposure (P < 0.01) than in the low exposure group (P < 0.05).

ALK fusion: Asian studies revealed cross‐population/subpopulation similarities and heterogeneity

Asian studies on lung cancer patients that provided details of ethnicity, gender, and smoking status with ALK fusion are summarized in Table S4.2, 20, 21, 22, 23, 25, 27, 31, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58 In total, 33 studies were included (China 14, Japan 12, Korea 7). Among 10 837 NSCLC patients, 614 (5.7%) ALK rearrangements were identified, with a major positive rate of 1.4–12%. On average, more ALK rearrangements occurred in Chinese patients (7.9%), followed by Korean (6.9%) and Japanese (3.2%) (Fig 3a,b). When considering gender ratio and smoking history, in all three nations ALK rearrangement in men occurred 20% less frequently than in women. Interestingly, when ALK fusion cases in the 33 studies were combined, never smokers (71.2%) made up the majority carrying ALK fusions. When divided by nation, the never smoker ratio harboring ALK fusions was 3 in the Chinese group and approximately 2 each in the Japanese and Korean groups (Fig 3c,d).

Figure 3

Comparison of 33 Asian studies on ALK fusion based on gender and smoking history. (a) ALK fusions over three nations. Average (b) positive rate of ALK fusion, (c) gender ratio, and (d) smoking rate. Relationships between ALK+ and (e) gender and (f) smoking rates. () China, () Japan, () Korea, and () Present study. (g) Distribution of 33 Asian studies on ALK fusion based on gender and smoking rates. () China, () Japan, and () Korea. (h) ALK fusion of present study (total and subgroups) based on gender and smoking rates. (*P < 0.05) Note: the value of the numerator based on the condition that the denominator is zero was directly used in figures. The black bubble representing the coal exposure subgroup is right behind the transparent purple bubble (high exposure and family history of lung cancer [FLC]+). () Present study‐Total, () FLC+, () FLC−, () Smoker+, () Smoker−, () Coal exposure+, () Coal exposure−, () High exposure group, () Low exposure group, () High exposure + FLC+, () High exposure + FLC−, () Low exposure + FLC+, () Low exposure + FLC−. High exposure: either being a smoker or exposed to coal combustion; low exposure: not exposed to coal combustion and also a non‐smoker.

Analyses of the relationship between gender, smoking ratio, and ALK positive rate yielded quite interesting findings. In all three countries, increases in the number of ALK positive patients were associated with increases in the male: female ratio (Fig 3e). In terms of study distribution, China and Korea were similar, while Japan had less overlapping work with the other two. The trend line was based on 31 studies, excluding one27 that did not include a gender ratio and one55 that only included stage IIIB–IV NSCLC patients. On the other hand, in studies with a lower ALK positive rate, more never smokers tended to have ALK fusion, but the frequency of smokers became higher with increasing ALK positive rates; the distribution was similar for all three countries (Fig 3f). In summary, an increase in the frequency of ALK fusion in the subject population was accompanied by rising male ratios in the positive pool. The increase appeared to be related to smoking males, suggesting a potential link between smoking and ALK fusion in men.

When gender, smoking history, and ALK positive rates were integrated into a plot for evaluation, the majority of studies gathered around the center, but a few drifted away from the main group (Fig 3g). Most studies fell into the category with a higher frequency of women and never smokers, exhibited a relatively lower positive rate, and focused on stage IIIB–IV NSCLC (with the exception of one study with a much higher positive rate, represented by the large white bubble in Fig 3g, which focused on stage IIIB–IV NSCLC).55 Studies located in the right upper quadrant included more men and never smokers but showed higher positive rates. Most Chinese, Japanese, and Korean studies focused on the same geographic regions, but a few of the Japanese studies contained little co‐existing work from other ethnic groups, possibly reflecting population variation.

Subgroups by genetic, occupational, or environmental factors could partly recreate heterogeneous distribution of ALK fusion from previous reports

Our overall results were consistent with previous work on ALK rearrangement, especially those from China (Fig 3e–g). When our study population was further divided by FLC, smoking status, coal exposure, and overall exposure, the diversity of subgroups gradually took shape.

In total, 12 (6 pairs) subgroups were analyzed: 8 corresponded with the category featuring women and never smokers, similar to the majority of the studies; and 4 with a higher male ratio. Moreover, compared to the 33 Asian studies, 6 subgroups were located near the center, while the other 6 drifted some distance away. Notably, 4 pairs were separated into different quadrants; in only 2 pairs were both subgroups located in the same category (coal user +/−; low exposure, FLC +/−) (Fig 3h). Interestingly, dividing the population into subgroups featured by genetic, occupational, or environmental factors could at least partly recreate the diversified distribution picture of ALK fusion from previous reports.

ALK fusion occurred in 18.2% (35/192) of patients with FLC, with a higher ratio of women and non‐smokers, and 7.3% (16/218) in patients without FLC. The ALK+ rate in never smokers was 19.0% versus 7.4% in smokers. In ALK+ cases, all smokers were men and 85.3% of non‐smokers were women; this subgroup exhibited the greatest difference in results. ALK+ occurred more frequently in subjects exposed to coal combustion (16.2% vs. 8.7%) but ALK+ cases in both exposed and non‐exposed groups were higher in women and never smokers. In the high exposure group 14.9% (39/262) were ALK+ compared to 8.1% (12/148) in the low exposure group; women and never smokers dominated these categories. Patients with both FLC and high exposure exhibited the highest rate of ALK rearrangement (26/94, 27.7%) in all subgroups, almost matching the results of a study on advanced NSCLC (30.9%). Interestingly, ALK fusion in patients without FLC but high exposure was characterized by more men, smokers, and a moderate positive rate (13/168, 7.7%), contrasting with the results of the majority of the 33 reports included in this analysis;2, 20, 45, 48, 54 however, this subgroup was similar as the results of one of the Japanese studies.46 Finally, regardless of FLC, ALK+ patients in the low exposure group were mainly women and non‐smokers, but the positive rate was slightly higher in FLC patients (9.2% vs. 6.0%). Overall, different population compositions contribute to variations in ALK rearrangement.

Discussion

Both genetic and environmental factors contribute to the development of cancer and its mutant spectrum.2, 6, 7, 13, 59, 60 There is abundant evidence of genetic predisposition to lung cancer.8, 11, 13, 14, 61 Our study found that regardless of gender and exposure level, FLC patients exhibited some common characteristics: younger age, a high female ratio, AD dominant, increased metastasis, later stage at diagnosis, and higher frequency of gene fusion. Many of these are indicative of increased aggressiveness. Our population had unique characteristics but our results were consistent with previous studies. Findings across previous studies showed both similarity and heterogeneity, depending on the subject source. Some studies reported that certain ethnic groups are affected more by inherited lung cancer susceptibility;13, 14 one indicated a lower association between never smokers and FLC;13 another suggested that FLC is common in never smokers with NSCLC;9 in regard to gender, some studies found that female relatives have a higher risk compared to male relatives10, 61 and similarly, 34% of Spanish women with lung cancer had FLC;12 and a study that enrolled mainly Caucasian patients found a link between FLC and EGFR, rather than ALK or KRAS mutation.9 These results indicate that susceptibility to lung cancer may be inherited in a complex pattern across populations, and each population or subpopulation has unique characteristics.

Importantly, FLC tend to have greater impact on lung cancer when subjects were exposed to a high level of carcinogens; studies also report that the lung cancer risk is further amplified by smoking.14 This was reflected in our results (Fig 1c): the median age of FLC patients was 10 years younger (53 vs. 63 years) in the high exposure group. On the other hand, if all factors were equal and the study population was divided by pollution level, certain somatic mutation may occur less frequently in an unpolluted environment. Our results showed a smaller age gap in the low exposure group (median age: with FLC 55.5 vs. without FLC 58.5 years). Similarly, regarding ALK fusion and total gene rearrangement, the association with FLC was stronger in the high than the low exposure group (Table 2). Our findings indicate that even within the same exposure group, patients are still affected by FLC, but a susceptible genetic background makes them more vulnerable in a polluted environment.

The ALK fusion gene is recognized as an important oncogenic driver gene in NSCLC (main type EML4‐ALK).19, 24, 25, 50 Incidence of ALK fusion in NSCLC is approximately 0.99–12%, with no significant differences between Asian and Western populations.19, 20, 49 ROS1 rearrangement occurs in a small subset (0.5%–2%) of NSCLC patients and is associated with slight or never smokers and adenocarcinoma histology. CD74‐ROS1 is the major type (~40%), followed by EZR‐ROS1 and unknown type (both ~15%).62 Our results regarding ALK and ROS1 fusion were similar to previous reports with the exception of elevated ROS1 fusion frequency (3.9%).29, 30, 49, 62 A meta‐analysis including 1178 ALK rearranged cases from 20 541 NSCLC patients indicated that age, gender, smoking status, histology, tumor stage, and ethnicity may be a source of between‐study heterogeneity.19 According to the reference studies from three nations, there is obvious across population/subpopulation similarity and heterogeneity. Interestingly, similar to the reference reports, our result of an ALK fusion increase with a rising male ratio in the positive pool was associated with smoking (Fig 3e,f), suggesting a potential relationship between smoking and ALK fusion in men. However, as many previously published reports from Asian countries found that ALK rearrangement is associated with female gender and non‐smokers, this hypothesis requires further investigation.22, 36, 37, 50, 54

One parameter only represents one piece of the puzzle, and for many investigators it can be difficult to collect complete patient data, such as: disease history, personal genetic makeup, and every occupational or environmental risk factor. Therefore, it can be difficult to determine what kinds of missing data may cause differences between study results. Although it is a well‐known assumption that variation between studies is mainly caused by different population compositions,2, 19 the division of our study sample into subgroups by genetic, occupational, or environmental factors enabled a picture of diversity in ALK fusion to take shape. This not only provided support for the assumption, but also suggested that different kinds of parameters should be carefully considered when conducing cancer research.

When all identified gene rearrangement events were combined for consideration (Fig 2), inherited susceptibility and exposure to environmental carcinogens appeared to be significant factors causing gene rearrangement. The combination pushed the gene fusion rate to 35.1% (33/94) in the subgroup with both characteristics. The high frequency of gene fusion events and the association with women, never smokers, and younger age may be unique characteristics of the subject population. Other research on lung cancer conducted in this region also found special mutation signatures, such as higher K‐RAS mutation rates (15%–29%), compared to other Asian populations (2–7%), including smokers and never smokers.59, 63, 64, 65 Coal combustion releases a cocktail of carcinogens including polycyclic aromatic hydrocarbons and fine particulate matter (PM2.5), many of which are defined by the International Agency for Research on Cancer and the World Health Organization as class I carcinogens.3, 16, 66 Certain carcinogens may have a specific effect on the human genome. For example, p53 gene mutations in tumors from female never smokers in Xuanwei exhibited characteristics induced by polycyclic aromatic hydrocarbons, different from those in lung cancer attributed to smoking.6 All of these findings support the hypothesis that patients exposed to coal combustion could display a unique mutational spectrum in lung cancer tumors.5, 7, 63, 64

Our study found that FLC and exposure to coal combustion influence clinicopathologic features and gene fusion patterns in NSCLC. Our results reveal the potential uniqueness of the subject population, particularly that inherited susceptibility exerts a greater impact on the age of onset and frequency of gene fusion when subjects are exposed to high levels of carcinogens. Finally, our study showed that dividing a population by genetic, occupational, or environmental factors could partly recreate the diversified distribution picture of gene fusion presented over different reports. Our findings may help to evaluate the effect of FLC and coal exposure on the pathogenesis of lung cancer and highlight the significance of integrating various parameters into clinical and theoretical research.

Table S1. Primer sets for detection of gene rearrangements.

Table S2. Clinicopathological characteristics of 51 non‐small cell lung cancer (NSCLC) patients with ALK rearrangements.

Table S3. Clinicopathological characteristics of 16 non‐small cell lung cancer (NSCLC) patients with ROS1 rearrangement.

Table S4. Investigation of 33 ALK rearrangement studies.

Click here for additional data file.