- Email: [email protected]
Ingested arsenic, characteristics of well water consumption and risk of different histological types of lung cancer in northeastern Taiwan
ARTICLE IN PRESS Environmental Research 110 (2010) 455–462
Contents lists available at ScienceDirect
Environmental Research journal homepage: www.elsevier.com/locate/envres
Ingested arsenic, characteristics of well water consumption and risk of different histological types of lung cancer in northeastern Taiwan$ Chi-Ling Chen a, Hung-Yi Chiou c, Ling-I Hsu d, Yu-Mei Hsueh c, Meei-Maan Wu d, Chien-Jen Chen b,d, a
Graduate Institute of Clinical Medicine, College of Medicine, National Taiwan University, 7, Chung-Shan S. Road, 100 Taipei, Taiwan Graduate Institute of Epidemiology, College of Public Health, National Taiwan University, Taipei, Taiwan c School of Public Health, Taipei Medical University, Taipei, Taiwan d Genomics Research Center, Academia Sinica, Taipei, Taiwan b
a r t i c l e in fo
abstract
Article history: Received 31 March 2009 Received in revised form 24 July 2009 Accepted 11 August 2009 Available online 6 September 2009
Our previous study combining two arseniasis-endemic areas in Taiwan confirmed a dose–response association of lung cancer and arsenic exposure. We conducted current analysis to elucidate the dose– response relationship in lower exposure level, and to evaluate whether the association differs in different histological types. In addition, whether specific characteristics of well water consumptions increased lung cancer risk was examined in order to establish a complete risk profile for arsenic exposure. A total of 8086 residents in northeastern Taiwan were followed for 11 years and 6888 participants remained in the final analysis because 1198 residents with unknown arsenic concentration were excluded. The 178 incident lung cancers were ascertained through linkage with the national cancer registry profiles in Taiwan. All analyses were performed by Cox’s proportional hazards regression models. We found a significant dose–response trend (P ¼ 0.001) of lung cancer risk associated with increasing arsenic concentration. There was no apparent increased risk at concentrations between 10 and 100 mg/L, but concentrations between 100 and 300 mg/L showed evidence of excess risk (RR 1.54, 0.97–2.46). The relative risk was 2.25 (95% CI: 1.43, 3.55) for exposure to Z300 mg/L when compared to o10 mg/L. The significant dose–response trends and the synergistic effect of arsenic exposure and cigarette smoking can be found in squamous and small cell carcinomas, but not in adenocarcinoma. Despite lacking statistical precision, when duration is accounted for, all levels of exposure including low concentration were in the direction of increased risk of lung cancer, and these associations tended to increase with longer durations of exposure. This study provides additional evidence linking arsenic to lung cancer, and the indications that arsenic may play a more important role in certain histological type may help with further research in carcinogenic effect of inorganic arsenic on lung cancer. & 2009 Elsevier Inc. All rights reserved.
Keywords: Adenocarcinoma Arsenic Lung cancer Squamous cell carcinoma Small cell carcinoma
1. Introduction Arsenic is a widely distributed semi-metallic element that occurs naturally in various compounds in the crust of the earth. In addition to earlier ecological studies conducted in Taiwan (Chen $ Supported by Grants NSC-83-0412-B002-231, NSC-97-2314-B001-002 from the National Science Council; DOH85-HR-503PL from Department of Health, Executive Yuan; NHRI-EX97-CD9201 from the National Health Research Institutes, Taiwan. All participants provided written informed consents before the data collection procedures that had been reviewed and approved by the IRB of College of Public Health, National Taiwan University. Corresponding author at: Genomics Research Center, Adacemia Sinica, 128 Academia Road, Section 2, Nankang, Taipei 115, Taiwan. Fax: +886 2 2789 8784. E-mail address: [email protected] (C.-J. Chen).
0013-9351/$ - see front matter & 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.envres.2009.08.010
et al., 1992, 1985, 1988; Chen and Wang, 1990; Guo, 2004; Tsai et al., 1999; Wu et al., 1989), recent epidemiological studies, using cohort (Chen et al., 2004; Chiou et al., 1995; Tsuda et al., 1995) or case-control study (Chen et al., 1986; Ferreccio et al., 2000; Heck et al., 2009; Mostafa et al., 2008) designs from various countries, provided further evidence that ingestion of arsenic via drinking water increased risk of lung cancer in a dose–response manner. Our previous analysis combining cohorts from the two arseniasis-endemic areas in Taiwan (Chen et al., 2004) confirmed the elevated risk of lung cancer associated with high arsenic exposure (RR ¼ 3.29, 95% CI: 1.60, 6.78 for Z700 mg/L compared with o10 mg/L), and a significant dose–response relationship. However, there was no conclusive evidence regarding risk estimates in lower exposure group because of small case numbers (Celik et al., 2008). In addition, the proxy measurement of arsenic
ARTICLE IN PRESS 456
C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
exposure in the southwestern cohort in Taiwan in this study added uncertainty to the inference of causal relationship. More information on the hazardous health effect of ingested arsenic exposure at a lower level from epidemiological studies is warranted. Occupation studies (Axelson et al., 1978; Newman et al., 1976; Pershagen et al., 1987; Taeger et al., 2008; Wicks et al., 1981) examined arsenic exposure via inhalation, and one ecological study investigating arsenic from drinking water from Taiwan (Guo et al., 2004) suggested that there might be cell-type specificity of arsenic-related lung cancer, mainly from observing a higher proportion of either squamous cell carcinoma (Guo et al., 2004), or adenocarcinomas (Axelson et al., 1978; Newman et al., 1976; Pershagen et al., 1987; Wicks et al., 1981) in arsenic-exposed population, or both (Taeger et al., 2008). To date, one study (Heck et al., 2009) conducted in a US population with low-to-moderate arsenic exposure using individual toenail arsenic exposure information reported that higher arsenic exposure was associated with small cell and squamous cell carcinoma of the lung. However, toenail arsenic concentration is thought to be representative of only short-term exposure, mostly in the range less than 1 year prior to assessment. The residents from an arseniasis-endemic area in northeastern Taiwan were exposed to arsenic via drinking water at a relatively lower concentration than those in southwestern Taiwan (Chiou et al., 2001). Instead of sharing a couple of wells among the whole village in southwestern endemic areas, the residents in northeastern endemic areas usually have a well in their own backyard and drank from the well water for a long period of time until the implementation of tap water in the early 1990s. We conducted current analysis using only data from northeastern cohort that has better individual measurement of arsenic exposure to elucidate the dose–response relationship with incident lung cancer in lower arsenic exposure levels, and to evaluate whether the dose–response relationship differs in different histological types of lung cancer. In addition, with the supplemented information from the questionnaire, we investigated whether they drank well water for longer duration, drank from birth (latency) and still drank well water at enrollment (recency) added burden to the risk of developing lung cancer. Similarly, it is of interest to evaluate whether the relationship between arsenic concentration and lung cancer risk differs depending on the latency and recency to establishing a complete risk profile for arsenic exposure.
2. Materials and methods 2.1. Study cohort Study participants were recruited from four townships of Tung-Shan, ChuangWei, Chiao-His and Wu-Chieh in the Lanyang Basin, which was located in the northeastern coast of Taiwan. Residents here had consumed water from shallow wells (o40 m in depth) that contained inorganic arsenic since the late 1940s until tap water was implemented in the early 1990s. Detailed description of study area and enrollment procedure was described previously (Chiou et al., 1997, 2001). Briefly, a total of 8102 residents aged Z40 years from 4586 households of 18 villages in the study area participated in the baseline home interview from 1991 to 1994. Since the national identification numbers were used to link to the national cancer registry profile, 16 participants with missing or who shared the same ID numbers were excluded from the analysis, resulting in 8086 participants in the final analysis. Four well-trained local public health nurses used standardized questionnaires to conduct personal interview at study sites. Information collected via questionnaire included demographic characteristics, cigarette smoking and habitual alcohol consumption status, as well as residential and well water consumption history. Residents who agreed to participate all provided written informed consents, and the data collection procedures had been reviewed and approved by the IRB of College of Public Health, National Taiwan University.
2.2. Arsenic exposure Arsenic concentration was estimated using the water samples collected from the wells the participants used. A total of 3901 water samples were collected from individual wells of 3901 households (85.1% of 4586 household in the study) during the home interview. Because the wells of 685 households no longer existed, the arsenic exposure of 1136 residents was classified as ‘‘unknown’’. The concentration of arsenic was determined by hydride generation combined with flame atomic absorption spectrophotometery. The range of the water samples was between undetectable (o0.15 mg/L) and more than 3000 mg/L. We were unable to determine the arsenic concentration of well water samples for 62 participants, resulting in a total of 1198 participants with ‘‘unknown’’ arsenic concentration. Since these 1198 participants with ‘‘unknown arsenic concentration’’ were not associated with the risk of lung cancer, and excluding them from the main analyses did not alter either the results or the conclusions, 6888 remained in the final analysis. In addition, several measurements of arsenic exposure were evaluated, including the duration of drinking well water containing 410 mg/L of arsenic, the starting (latency) and ending age of exposure, the current status of water consumption at the enrollment (recency) and cumulative exposure status (concentration times duration). All these variables were derived from a questionnaire with detailed history of residential addresses and corresponding arsenic concentration of well water in each address that was tested. The cumulative arsenic exposure was calculated as the sum of the products of arsenic concentration (mg/L) and years of drinking water from that specific well for the period from start to quit drinking well water or recruitment if they were still drinking well water.
2.3. Identification of lung cancer cases Each subject’s national identification number was used to link with the national cancer registry profiles in Taiwan to identify newly diagnosed lung cancer cases between date of questionnaire interview and December 31, 2006. This cancer registry system was implemented in 1978 in Taiwan, and was considered a nationwide cancer registry system with updated, accurate and complete information. The different histological types of lung cancer were determined by the International Classification of Diseases for Oncology codes and categorized into ‘‘squamous cell carcinoma’’, ‘‘adenocarcinoma’’, ‘‘small cell carcinoma’’ and the remaining as ‘‘other histological types’’. After an average follow-up period of 11.5 years (SD, 3.6 years) with 79,345 person-years, a total of 178 newly diagnosed lung cancer cases were identified through data linkage with the computerized national cancer registry profile in Taiwan, resulting in a lung cancer incidence of 2.24 cases per 1000 person-years. Among them, 75(42.1%) were squamous cell carcinoma, 51 (28.7%) were adenocarcinoma, 22 (12.4%) were small cell carcinoma and the remaining 30 (16.9%) mostly consisted of ‘‘no microscopic confirmation’’ (n ¼ 12, 6.0%) and ‘‘other malignancy’’ (n ¼ 10, 5.0%). This ‘‘other malignancy’’ indicates primary lung cancer with some rare form of histological types that are not categorized.
2.4. Statistical analyses Nonparametric methods (Kruskal–Wallis or two-sample Wilcoxon rank-sum test) were used to examine the statistical significance of the median level of arsenic concentration, duration of well water consumption, as well as cumulative arsenic exposure by selected demographic characteristics. Person-years for each participant were calculated from the date of questionnaire interview to the date of cancer diagnosis, death, or December 31, 2006, whichever came first. Arsenic concentration was arbitrarily divided into o10, 10–49.9, 50–99.9, 100–299.9 and Z300 mg/L to emphasize the risk of lung cancer associated with lower exposure levels. Since there is still trace amount of arsenic in tap water, measurement error for the cumulative arsenic exposure group of zero is quite likely, we further used o100 mg/L years as reference. Additional reference group of o400 mg/L years was rationalized on the average duration of drinking well water (40 years) times the arsenic concentration allowed (10 mg/L). Relative risks (RRs) and 95% confidence intervals (CIs) were estimated by Cox’s proportional hazards regression models. The adjustment variables in the final model included age (continuous), gender (male/ female), schooling year (0/1–6/46 years), cigarette smoking status (never/ past/current) and habitual alcohol consumption (no/yes). Although there was a significant difference in the distribution of arsenic concentration between the four towns, it was not considered as a potential confounding factor because ‘‘township’’ was not associated with lung cancer. Trends across levels of categorical variables were assessed by testing the statistical significance of a single trend variable coded as the category of exposure. Since the study participants were mostly farmers, previous use of arsenic pesticides may play a role in the association between ingested inorganic arsenic via drinking water and lung cancer risk. The information of pesticide usage was collected via questionnaire, and a sensitivity analysis was performed by restricting to those who reported having never used pesticide. All the above analyses were performed by Stata Statistical Software (Version 8.2, Stata Corp., College Station, Texas).
ARTICLE IN PRESS C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
3. Results The baseline demographic characteristics of study cohort and the distribution of median arsenic concentration (mg/L), duration of well water consumption (years) and cumulative arsenic exposure (mg/L years) are shown in Table 1. The medians of arsenic concentration were similar in different age, gender, and cigarette smoking groups. Those who were older, male, were residents in Tong-Shang, received less schooling years, smoked cigarettes, were habitual alcohol drinkers and reported never being exposed to pesticide were exposed to higher cumulative arsenic exposure. Participants with ethnicity other than ‘‘Mingnan’’, which makes up the majority, appeared exposed to relatively lower arsenic levels, shorter duration and hence lower cumulative exposure. The variation in distribution of medians was most pronounced in the four townships, with residents in Tong-Shang being exposed to the highest concentration and cumulative exposure from well water, and Wu-Che had a median concentration of zero (interquartile: 0–28.0) but the longest duration of exposure. The mean arsenic concentration among wells with known arsenic concentration was 117.2 mg/L (SD ¼ 297.2 mg/L), and the mean duration of consuming well water was 42.0 years (SD ¼ 15.1 years), resulting in the mean cumulative arsenic exposure of 3523.5 mg/L years (SD ¼ 9443.5 mg/ L years) (Table 1). Among 4600 residents who were exposed to arsenic concentration Z10 mg/L, 3005 (65.3%) who responded still drank well water at the time of enrollment, and 1555 reported having stopped drinking well water before the enrollment (33.8%) (Table 2). A total of 1753 (21.7%) residents started drinking well water from birth, and about 15 percent drank well water for more than 50 years. Overall, there was about 12 percent study participants who either never drank well water or the arsenic concentration of the well water they drank were below the detection limit. Those who still drank well water containing Z10 mg/L of arsenic at enrollment or drank it from birth were associated with about a 30% increase in risk of lung cancer (RR ¼ 1.28, 95% CI: 0.90, 1.83 for still drinking; RR ¼ 1.32, 95% CI: 0.87, 1.98 for drinking from birth). The dose–response trends were all statistically significant when using three different reference groups, and the highest relative risks were around 2-fold for the highest cumulative exposure of 410,000 mg/L years (RR ¼ 1.53, 95% CI: 0.93, 2.54; RR ¼ 1.79, 95% CI: 1.12, 2.88 and RR ¼ 2.10, 95% CI: 1.34, 3.31 for 0, o100 and o400 as references, respectively). Table 3 shows the multivariate-adjusted relative risk of categories of arsenic concentration in relation to overall and different histological types of lung cancer. A significant dose– response trend (Ptrend ¼ 0.001) was found for increasing arsenic concentrations using o10 mg/L as the reference group, after adjusting for age, gender, years of education, cigarette smoking and habitual alcohol consumption status at enrollment. There was no apparent increased risk at concentrations between 10 and 100 mg/L, but concentrations between 100 and 300 mg/L showed evidence of excess risk (RR 1.54, 0.97–2.46). The relative risk was 2.25 (95% CI: 1.43, 3.55) for exposure Z300 mg/L when compared to the reference category. The significant dose–response trend of increasing arsenic concentrations and increasing risk of lung cancer was found in squamous cell carcinomas (Ptrend ¼ 0.004) and small cell carcinomas (Ptrend ¼ 0.021), but not in adenocarcinomas (Ptrend ¼ 0.664) and other histological types (Ptrend ¼ 0.197). The highest risk was found in small cell carcinomas with arsenic concentration Z300 mg/L (RR ¼ 5.30, 95% CI: 1.48, 18.9). We further investigated whether the association between arsenic concentration and lung cancer risk differed between those who reported still drinking well water at the enrollment (no/yes)
457
and the time of first exposure (from birth/not from birth) (Table 4). The dose–response trend was more prominent in those who still drank well water at enrollment (Ptrend ¼ 0.002) than those who stopped (Ptrend ¼ 0.115). The pattern of incremental lung cancer risk associated with increasing arsenic concentration was similar among those who started drinking well water from birth and those who started after birth. When compared to those who have never smoked cigarettes and consumed well water containing arsenic o10 mg/L, residents who smoked cigarettes Z25 pack-years and consumed well water with an arsenic level Z100 mg/L had a 7-fold increased risk of lung cancer (RR ¼ 6.97, 95% CI: 3.4, 14.3) (Table 5). And this increased risk was as high as 11-fold (RR ¼ 10.9, 95% CI: 3.57, 33.5) for squamous cell carcinoma combined with small cell carcinoma. Since the proportion of adenocarcinoma was known to be relatively higher in women and the prevalence of cigarette smoking was very low for women in this cohort, the analyses were repeated on men only. We found similar results with regard to the risk patterns, only that the risk estimates were higher with much wider 95 percent confidence intervals due to the smaller sample size (Table 5). All p-values for interactions between packyears of cigarette smoked (never/o25/Z25) and arsenic concentration (o10/10–99.9/Z100 mg/L) were not statistically significant. The combination of arsenic concentration and duration showed that those who drank well water containing high concentration of arsenic (Z300 mg/L) for over 50 years are at almost 10-fold increased risk of lung cancer (RR ¼ 9.71, 95% CI: 2.84, 33.2) when compared to those who drank well water below 10 mg/L for less than 30 years (Fig. 1). Among residents exposed to the same level of arsenic concentration, longer duration was associated with increased risk of lung cancer, and this trend was more prominent in the lowest and the highest arsenic concentration group.
4. Discussion In this analysis, we confirmed our earlier finding of a significant dose–response trend of increasing lung cancer risk with increasing arsenic concentration at a lower exposure level. In addition, the significant dose–response trend can be found in squamous cell and small cell carcinomas, but not in adenocarcinoma. Most studies investigating arsenic exposure via inhalation suggest a higher proportion of adenocarcinomas and a lower proportion of squamous cell carcinoma (Axelson et al., 1978; Newman et al., 1976; Pershagen et al., 1987; Wicks et al., 1981). A recent re-analysis of data from German uranium miners by Taeger et al. (2008) reported that arsenic exposure was associated with non-small cell lung cancer and the specific cell type in relation to arsenic was different mainly determined by silicosis status. However, two other studies evaluating arsenic exposure via drinking water reported similar results to the current analysis. One was an ecological study from Taiwan (Guo et al., 2004). They found that patients from arseniasis-endemic areas had higher proportions of squamous cell and small cell carcinomas, but a lower proportion of adenocarcinomas. The other was a population-based case-control study conducted in low-to-moderate arsenic exposure areas in the US (Heck et al., 2009). They reported that compared to o0.05 mg/g of toenail arsenic concentration, those with Z0.114 mg/g were associated with an about 3-fold increased risk of small cell and squamous cell carcinoma (OR ¼ 2.75, 95% CI 1.00, 7.57). We also examined the duration, recency and latency of drinking well water containing arsenic Z10 mg/L and found that drinking arsenic well water from birth and early life, still drinking
ARTICLE IN PRESS 458
C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
Table 1 Distribution of selected demographic characteristics of 6888 study participants from arseniasis-endemic area in northeastern Taiwan. Characteristics
Age at recruitment r50 50–54.9 55–59.9 Z60 P value Mean age Mean years of follow-up Gender Male Female P valuea Ethnicity Mingnan Others Unknownb P valuea Township of residence Chung-Wei Tong-Shang Chiou-Shi Wu-Che P valuec Schooling years 0 1–6 46 Unknownb P valuec Cigarette smoking status at enrollment Never Past Current Unknownb P valuec Years of cigarette smoking 0 1–35 435 Unknownb P valuec Cigarettes smoked per day 0 1–19 Z20 Unknownb P valuec Pack-years of cigarettes smoking 0 1–25 425 Unknownb P valuec Habitual alcohol consumption No Yes Unknownb P valuea Exposure to pesticide Never Ever Unknownb P valuea Mean concentration in well water (lg/L) Mean duration of consuming well water (years) Mean cumulative arsenic exposure ((lg/L) years) a b c
Number (n ¼ 6888)
Percentage or mean (SD)
Median arsenic concentration (lg/L) (25%, 75%)
Median duration (25%, 75%)
Median cumulative arsenic exposure (lg/L years) (25%, 75%)
1538 2338 1801 1211
22.3 33.9 26.2 17.6
25.6 (3.1, 28.3 (3.5, 29.6 (3.6, 25.8 (3.5, 0.7929
96.4) 93.7) 92.5) 88.5)
39 (29, 44) 40 (31, 53) 42 (33, 61) 45 (31, 67) o0.0001
677 (785, 2237) 840 (90, 2785) 943 (106, 3030) 980 (95, 3192) o0.0001
59.1 (11.0) 11.5 (3.6) 3481 3407
50.5 49.5
28.0 (3.8, 92.8) 27.6 (3.1, 93.2) 0.5993
41 (31, 52) 41 (31, 53) 0.6978
915 (109, 3020) 777 (78, 2559) 0.0005
6819 58 11
99.0 0.8 0.2
28.0 (3.5, 93.2) 1.7 (0, 48.2) 3.5 (0, 30.0) 0.0002
41 (31, 52) 52 (40, 66) 58 (44, 75) 0.0002
855 (96, 2800) 0 (0, 603) 158 (0, 1894) o0.0001
3387 1016 1394 1091
49.2 14.8 20.2 15.8
26.7 (8.5, 88.4) 93.7 (56.5, 207.4) 20.2 (1.5, 63.7) 0 (0, 28.0) 0.0001
40 (31, 43 (33, 40 (30, 43 (31, 0.0005
52) 53) 54) 55)
871 (224, 2558) 3279 (1436, 6157) 7 (63, 1926) 69 (0, 979) 0.0001
2011 4049 456 372
29.2 58.8 6.6 5.4
27.3 (4.0, 87.8) 27.8 (2.7, 93.3) 20.0 (1.4, 69.6) 46.0 (14.3, 157.6) 0.0035
41 (31, 41 (31, 40 (30, 43 (33, 0.0001
56) 52) 47) 54)
848 (92, 2900) 843 (85, 2800) 564 (17, 1659) 1233 (244, 4092) 0.0001
4065 854 1967 2
59.0 12.4 28.6 –
27.0 (3.2, 94.2) 28.6 (3.0, 92.1) 28.3, (4.0, 91.4) – 0.8213
41 (31, 52) 42 (31, 54) 40 (31, 53) – 0.1391
777 (80, 2605) 952 (116, 3143) 947 (115, 2938) – 0.0014
4065 1261 1461 101
59.0 18.3 21.2 1.5
27.0 (3.2, 27.6 (4.2, 29.5 (3.5, 26.1 (3.0, 0.5347
41 (31, 40 (31, 41 (31, 46 (35, 0.0001
777 (80, 2605) 972 (136, 2934) 943 (95, 3068) 789 (52, 3580) 0.001
4065 1110 1711 2
59.0 16.1 24.8 –
27.0 (3.2, 94.2) 32.3, (4.7, 96.8) 26.3 (3.0, 90.0) – 0.0305
41 (31, 52) 41 (31, 54) 40 (31, 52) – 0.1949
777 (80, 2605) 1051 (137, 3153) 867 (92, 2931) – 0.0001
4065 1021 1680 122
59.0 14.8 24.4 1.8
27.0 (3.2, 30.1 (4.0, 27.8 (3.6, 31.4 (3.0, 0.3829
41 (31, 41 (31, 40 (31, 46 (35, 0.2306
777 (80, 2605) 992 (135, 2940) 893 (97, 2989) 850 (65, 3668) 0.0011
5564 1314 10
80.8 19.1 0.2
26.3 (3.2, 90.3) 32.8 (5.2, 107.2) 47.2 (3.4, 77.0) 0.0021
41 (31, 53) 40 (31, 52) 44 (40, 55) 0.1499
807 (85, 2605) 1010 (147, 3515) 652 (0, 3182) o0.0001
6404 474 10
93.0 6.9 0.2
28.1 (3.6, 94.5) 20.2 (1.6, 66.3) 17.7 (0, 39.1) 0.0016
41 (31, 52) 41 (33, 54) 44 (35, 64) 0.0719
845 (92, 2797) 748 (80, 2576) 698 (0, 2237) 0.3382
117.2 (297.2) 42.0 (15.1) 3523.5 (9443.5)
P values were based on two-sample Wilcoxon rank-sum test. Those with unknown status were excluded from the statistical testing. P values were based on the Kruskal–Wallis test.
94.2) 93.4) 91.0) 81.7)
94.2) 93.2) 91.2) 85.3)
52) 50) 58) 58)
52) 50) 54) 58)
ARTICLE IN PRESS C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
459
Table 2 Distribution and multivariate adjusted relative risks of selected characteristics of arsenic-containing well water consumption in relation to lung cancer in northeastern arseniasis-endemic area in Taiwan. %
P-Y
# Cases (N ¼ 178)
RR (95% CI)a
Well water containing arsenic drinking status at enrollment 2288 Arsenic o10 mg/L Still drinking 3005 Stopped 1555 Cannot be determined 40
28.3 37.2 19.2 0.5
26,568 34,009 18,320 428
48 83 43 4
1.00 1.28 (0.89–1.82) 1.31 (0.87–1.98) 5.26 (1.88–14.7)
Age when started drinking well water containing arsenic 2288 Arsenic concentration o10 mg/L From birth 1753 40–20 1015 420–30 965 430 824 Cannot be determined 43
28.3 21.7 12.6 11.9 10.2 0.5
26,568 20,901 12,184 11,169 7991 512
48 45 24 35 23 3
1.00 1.31 1.16 1.68 1.04 4.22
(0.87–1.97) (0.71–1.90) (1.08–2.59) (0.63–1.73) (1.31–13.6)
Age when stopped drinking well water containing arsenic 2288 Arsenic concentration o10 mg/L Still drinking 3005 r50 583 450–60 509 460–70 309 470 154 Cannot be determined 40
28.3 37.2 7.2 6.3 3.8 1.9 0.5
26,568 34,009 7490 6096 3471 1220 472
48 83 9 15 11 9 3
1.00 1.27 1.11 1.36 1.16 2.08 4.45
(0.89–1.82) (0.53–2.32) (0.76–2.44) (0.60–2.25) (0.98–4.41) (1.38–14.3)
Years of drinking well water containing arsenic 2288 Arsenic concentration o10 mg/L o30 919 30–o40 1242 40–o50 1030 50–o60 799 60+ 569 Duration unknown 41
28.3 11.4 15.4 12.7 9.9 7.0 0.5
26,568 10,719 14,253 12,053 9457 5792 483
48 18 38 19 33 19 3
1.00 0.99 1.45 1.02 1.83 1.12 4.40
(0.58–1.71) (0.95–2.22) (0.60–1.74) (1.17–2.86) (0.65–1.93) (1.36–14.2)
23.9 43.3 32.8
19,196 34,908 25,207 14
31 69 78 0
1.00 1.22 (0.80–1.87) 1.44 (0.95–2.20) – 0.082
1071 2583 2078 524 632
13.3 31.9 25.7 6.5 7.8
12,558 30,161 23,732 5874 7001
32 43 51 23 29
1.00 0.56 (0.36–0.89) 0.78 (0.50–1.21) 1.37 (0.80–2.34) 1.52 (0.92–2.52) 0.002
1729 1925 2078 524 632
21.4 23.8 25.7 6.5 7.8
20,033 22,685 23,732 5874 7001
43 32 51 23 29
1.00 0.65 (0.41–1.02) 0.91 (0.60–1.36) 1.60 (0.96–2.65) 1.78 (1.11–2.85) 0.002
Cumulative arsenic exposure ((lg/L) years) o400 2534 400–o1000 1120 1000–o5000 2078 5000–o10,000 524 Z10,000 632 P trend
31.3 13.9 25.7 6.5 7.8
29,554 13,164 23,732 5874 7001
55 20 51 23 29
1.00 0.83 (0.50–1.39) 1.06 (0.73–1.56) 1.87 (1.15–3.04) 2.08 (1.33–3.27) 0.001
Characteristics
Number (N ¼ 6888)
Years of drinking well water (regardless of As conc.) r30 1646 430–r50 2981 450 2260 Unknown 1 P trend Cumulative arsenic exposure ((lg/L) years) 0 o1000 1000–o5000 5000–o10,000 Z10,000 P trend Cumulative arsenic exposure ((lg/L) years) o100 100–o1000 1000–o5000 5000–o10,000 Z10,000 P trend
a
Adjusted for age (1-year increment), gender, education levels, cigarettes smoking status and habitual alcohol consumption status at enrollment.
arsenic well water at the time of enrollment or stopping at older age was associated with slightly but not statistically significant higher risk of developing lung cancer. Furthermore, the cumulative arsenic exposure was associated with an increased risk of lung cancer in a dose–response fashion. But the significantly reduced risk of lung cancer (RR ¼ 0.56, 95% CI: 0.36, 0.89) for o1000 mg/L years compared to 0 mg/L years may indicate a threshold for arsenic exposure in relation to lung cancer. However, due to the possibility of misclassification in this group (0 mg/L years) because of the detection limit (0.15 mg/L), it warranted a cautious
interpretation. One recent ecological study (Marshall et al., 2007) investigating 50 years of lung cancer mortality in Region II of Chile found that the highest lung cancer relative risks were found in men who were exposed at earlier life (as children or adolescents). Our study found that consumption of well water containing arsenic 410 mg/L at earlier age (before 30 years old) was associated with increased lung cancer risk. The combination of arsenic concentration and duration showed that when the duration is accounted for, even at o30 years, all levels of exposure including low concentration were in the direction
ARTICLE IN PRESS 460
C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
Table 3 Incidence rates and multivariate adjusted relative risks of arsenic concentration in relation to different histological types of lung cancer in northeastern arseniasis-endemic area in Taiwan. Arsenic concentration (lg/L)
o10
10–49.9
50–99.9
100–299.9
Z300
Ptrend
All lung cancer$ Cases number (n ¼ 178) Incidence (per 105) RRa (95% CI)
48 181 1.00 Referent
51 211 1.10 (0.74–1.63)
20 194 0.99 (0.59–1.68)
28 269 1.54 (0.97–2.46)
31 397 2.25 (1.43–3.55)
0.001
Squamous cell carcinoma$ Cases number (n ¼ 75) Incidence (per 105) RRa (95% CI)
23 87 1.00 Referent
12 50 0.53 (0.26–1.07)
13 126 1.32 (0.67–2.61)
13 125 1.52 (0.77–3.00)
14 179 2.13 (1.09–4.17)
0.004
Adenocarcinoma$ Cases number (n ¼ 51) Incidence (per 105) RRa (95% CI)
14 53 1.00 Referent
20 83 1.50 (0.76–2.98)
4 39 0.70 (0.23–2.13)
6 58 1.06 (0.41–2.77)
7 90 1.63 (0.65–4.05)
0.664
Small cell carcinoma Cases number (n ¼ 22) Incidence (per 105) RR (95% CI)
4 15 1.00 Referent
8 33 2.02 (0.61–6.73)
0 0 –
4 38 2.77 (0.69–11.1)
6 77 5.15 (1.44–18.4)
0.021
Other histological types$ Cases number (n ¼ 30) Incidence (per 105) RRa (95% CI)
7 26 1.00 Referent
11 45 1.70 (0.66–4.39)
3 29 1.10 (0.28–4.25)
5 48 2.03 (0.64–6.40)
4 51 2.25 (0.65–7.71)
0.197
$
a $
Adjusted for age (1-year increment), gender, education levels, cigarettes smoking status and habitual alcohol consumption status at enrollment. Arsenic levels unknown were excluded from the analysis, resulted in 6,888 residents with 178 incident lung cancer cases.
Table 4 Multivariate adjusted relative risks of lung cancer associated with arsenic concentration stratified by duration and well water drinking status at enrollment. Arsenic concentration in well water (lg/L)
All lung cancer$ o10 10–49.9 50–99.9 100–299.9 Z300 P trend a $
Well water drinking status at enrollment
Age when started drinking well water
Still drinking RR (95% CI)a
Stopped drinking RR (95% CI)a
Started from birth RR (95% CI)a
Not started from birth RR (95% CI)a
1.00 1.00 (0.65–1.53) 1.14 (0.62–2.11) 1.34 (0.73–2.43) 2.65 (1.56–4.48) 0.002
1.00 1.32 (0.72–2.40) 0.69 (0.30–1.62) 1.55 (0.82–2.93) 1.73 (0.90–3.35) 0.115
1.00 1.06 (0.60–1.87) 0.80 (0.34–1.88) 1.35 (0.68–2.67) 2.39 (1.28–4.44) 0.027
1.00 1.08 (0.70–1.68) 1.01 (0.55–1.87) 1.57 (0.90–2.73) 2.09 (1.21–3.61) 0.009
Adjusted for age (1-year increment), gender, education levels, cigarettes smoking status and habitual alcohol consumption status at enrollment. Arsenic levels unknown were excluded from the analysis, resulted in 6,888 residents with 178 incident lung cancer cases.
of increased risk of developing lung cancer. In addition, these associations tend to increase with longer durations of exposure. However, these associations at low exposure level were not statistically significant due to inadequate power. Nevertheless, these results provided us additional insights into the carcinogenesis process of arsenic on lung cancer. That is, those who are exposed to arsenic in high concentrations for a long period of time are at a much higher risk of developing lung cancer than those who are either exposed to lower concentrations or for a shorter duration. Our previous study found indications of a synergistic effect of high arsenic exposure (Z700 mg/L) and cigarette smoking, and lung cancer risk could be as high as more than 10-fold when compared with non-smokers with an arsenic exposure level o10 mg/L (Chen et al., 2004). We repeated similar analysis on the cohort from northeastern Taiwan alone and found that the increased risk reached 7-fold at a much lower arsenic exposure level (Z100 mg/L) for heavy smokers (425 pack-years). And this indication of synergistic effect was found only in squamous and small cell carcinoma, but not in adenocarcinoma.
Strengths of our study include the large sample size and longer follow-up periods. This cohort is quite homogeneous, with regard to socio-economic status (mostly farmers) and years when the tap water system commenced. All arsenic concentration information was based on individual well water samples that the cohort members drank most of their water from for a long period of time. In addition, we were able to further examine other components of arsenic exposure including the starting and stopping age of consumption of well water, and hence duration of exposure and cumulative exposure including concentration and duration based on a structured questionnaire interview of detailed residential and well water drinking history. The prospective follow-up design provides us with correct temporal relationship of the exposure and the disease. The possibility of selection bias is minimized since all incident lung cancer cases were identified through data linkage. One limitation of this study is that almost 15% of study participants are with unknown arsenic concentration information because their wells no longer existed when the study was conducted. There was no difference in age and gender distribution between the residents with unknown concentration and residents
ARTICLE IN PRESS C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
461
Table 5 Relative risks of lung cancer in relation to arsenic exposure and cigarette smoking in northeastern arseniasis-endemic area in Taiwan. All participants RRa (95% CI)
Arsenic concentration (lg/L)
Pack-years of smoking All lung cancer Never
o25
Z25
P for interaction Squamous and small cell carcinoma combined Never
o25
Z25
P for interaction Adenocarcinoma Never
o25
Z25
o10 10–99.9 Z100 o10 10–99.9 Z100 o10 10–99.9 Z100
1.00 1.22 (0.64–2.32) 1.32 (0.64–2.74) 2.14 (0.79–5.79) 1.52 (0.56–4.15) 5.30 (2.19–12.8) 4.08 (1.83–9.10) 4.19 (1.92–9.14) 8.17 (3.74–17.9) 0.3665
1.00 1.92 (0.37–9.91) 1.38 (0.20–9.83) 2.96 (0.61–14.3) 2.09 (0.43–10.1) 5.39 (1.16–25.1) 5.21 (1.23–22.1) 5.27 (1.26–22.0) 10.1 (2.42–42.5) 0.7205
o10 10–99.9 Z100 o10 10–99.9 Z100 o10 10–99.9 Z100
1.00 0.45 (0.11–1.90) 1.39 (0.40–4.80) 2.21 (0.45–10.9) 1.05 (0.17–6.29) 7.41 (1.91–28.8) 7.20 (2.05–25.3) 7.05 (2.03–24.4) 14.6 (4.25–50.3) 0.5387
1.00 0.77 (0.05–12.3) 1.37 (0.09–22.0) 2.51 (0.26–24.1) 1.19 (0.11–13.1) 7.22 (0.87–60.2) 7.78 (1.03–58.5) 7.55 (1.02–56.0) 15.4 (2.06–114.2) 0.7905
o10 10–99.9 Z100 o10 10–99.9 Z100 o10 10–99.9 Z100
1.00 1.45 (0.62–3.43) 1.06 (0.37–3.06) 1.83 (0.41–8.19) 1.95 (0.51–7.42) 1.54 (0.27–8.63) 0.85 (0.18–4.01) 0.60 (0.13–2.86) 2.10 (0.53–8.39) 0.4684
1.00 1.56 (0.14–17.2) 1.29 (0.08–20.6) 2.22 (0.23–21.5) 2.36 (0.27–20.4) 0.85 (0.05–13.8) 0.88 (0.09–8.66) 0.62 (0.06–6.11) 2.25 (0.26–19.7) 0.5044
P for interaction a
Men only RRa (95% CI)
Adjusted for age (1-year increment), gender, education levels and habitual alcohol consumption status at enrollment.
Duration (years)
< 30
Arsenic Concentration (µg/L)
<10
Odds Ratio* (95% Confidence Interval)
1.00
< 30
10 –49.9
< 30
50 –99.9
3.28 (0.82 –13.1)
< 30
100 –299.9
2.18 (0.44 –10.8)
< 30
> 300
1.91 (0.39 –9.49)
<10
2.21 (0.63 –7.70)
>30-50
2.45 (0.68 –8.79)
>30-50
10 –49.9
2.76 (0.83 –9.21)
>30-50
50 –99.9
1.75 (0.44 –7.02)
>30-50
100 –299.9
4.45 (1.28 –15.4)
>30-50
> 300
4.78 (1.33 –17.2)
<10
3.01 (0.92 –9.91)
>50
10 –49.9
2.66 (0.77 –9.16)
>50
50 –99.9
2.43 (0.63 –9.43)
>50
100 –299.9
3.35 (0.91 –12.4)
>50
> 300
9.61 (2.81 –32.8)
>50
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20
Odds Ratio (95% Confidence Interval)
Fig. 1. Multivariate adjusted relative risks of combination of duration and arsenic concentration in relation to lung cancer in northeastern arseniasis-endemic area in Taiwan. *Adjusted for age (1-year increment), gender, education levels, cigarette smoking status and habitual alcohol consumption status at enrollment.
with known concentration. When categorizing this group of people separately, we found that excluding them from the analysis did not alter the analysis results. Although the misclassification of
low exposure groups cannot be ruled out, we have no reason to believe that these misclassifications would be dependent on their arsenic exposure. Thus this non-differential misclassification
ARTICLE IN PRESS 462
C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
would pull the association estimates toward the null and underestimate the true relationship. However, even though the number of participants with unknown arsenic well water drinking status at enrollment and duration is small, it is associated with a significantly higher risk of lung cancer. This may be due to the fact that the mean arsenic concentration was 246.6 mg/L (SD ¼ 435.9 mg/L) among these 41 participants with unknown well water consumption status at enrollment and duration, while the mean was 174.0 mg/L (SD ¼ 349.0 mg/L) for those with those information. Since most of the study participants were farmers, it may raise concern of potential misclassification when they had a chance to be exposed to pesticide containing arsenic. We restricted our analysis to residents who were not exposed to pesticides and the results did not alter. Therefore, the possibility of bias caused by the misclassification was minimal. This is the first study to investigate individual arsenic exposure that included not only concentration but also duration and other characteristics of drinking well water in relation to overall and different histological types of lung cancer. The results provided additional evidence linking arsenic to lung cancer not only by arsenic concentration alone but also by characteristics of well water consumption. And more importantly, the indication that arsenic may play a more important role in certain histological type of lung cancer may provide information for further research in carcinogenic effect of inorganic arsenic on lung cancer.
References Axelson, O., et al., 1978. Arsenic exposure and mortality: a case-referent study from a Swedish copper smelter. Br. J. Ind. Med. 35, 8–15. Celik, I., et al., 2008. Arsenic in drinking water and lung cancer: a systematic review. Environ. Res. 108, 48–55. Chen, C.J., et al., 1992. Cancer potential in liver, lung, bladder and kidney due to ingested inorganic arsenic in drinking water. Br. J. Cancer 66, 888–892. Chen, C.J., et al., 1985. Malignant neoplasms among residents of a blackfoot disease-endemic area in Taiwan: high-arsenic artesian well water and cancers. Cancer Res. 45, 5895–5899.
Chen, C.J., et al., 1986. A retrospective study on malignant neoplasms of bladder, lung and liver in blackfoot disease endemic area in Taiwan. Br. J. Cancer 53, 399–405. Chen, C.J., Wang, C.J., 1990. Ecological correlation between arsenic level in well water and age-adjusted mortality from malignant neoplasms. Cancer Res. 50, 5470–5474. Chen, C.J., et al., 1988. Atherogenicity and carcinogenicity of high-arsenic artesian well water. Multiple risk factors and related malignant neoplasms of blackfoot disease. Arteriosclerosis 8, 452–460. Chen, C.L., et al., 2004. Ingested arsenic, cigarette smoking, and lung cancer risk: a follow-up study in arseniasis-endemic areas in Taiwan. JAMA 292, 2984–2990. Chiou, H.Y., et al., 2001. Incidence of transitional cell carcinoma and arsenic in drinking water: a follow-up study of 8102 residents in an arseniasis-endemic area in northeastern Taiwan. Am. J. Epidemiol. 153, 411–418. Chiou, H.Y., et al., 1995. Incidence of internal cancers and ingested inorganic arsenic: a seven-year follow-up study in Taiwan. Cancer Res. 55, 1296–1300. Chiou, H.Y., et al., 1997. Dose–response relationship between prevalence of cerebrovascular disease and ingested inorganic arsenic. Stroke 28, 1717–1723. Ferreccio, C., et al., 2000. Lung cancer and arsenic concentrations in drinking water in Chile. Epidemiology 11, 673–679. Guo, H.R., 2004. Arsenic level in drinking water and mortality of lung cancer (Taiwan). Cancer Causes Control 15, 171–177. Guo, H.R., et al., 2004. Cell type specificity of lung cancer associated with arsenic ingestion. Cancer Epidemiol. Biomarkers Prev. 13, 638–643. Heck, J.E., et al., 2009. Lung cancer in a US population with low to moderate arsenic exposure. Environ. Health Perspect. Marshall, G., et al., 2007. Fifty-year study of lung and bladder cancer mortality in Chile related to arsenic in drinking water. J. Natl. Cancer Inst. 99, 920–928. Mostafa, M.G., et al., 2008. Lung cancer and exposure to arsenic in rural Bangladesh. Occup. Environ. Med. 65, 765–768. Newman, J.A., et al., 1976. Histologic types of bronchogenic carcinoma among members of copper-mining and smelting communities. Ann. NY Acad. Sci. 271, 260–268. Pershagen, G., et al., 1987. Histological types of lung cancer among smelter workers exposed to arsenic. Br. J. Ind. Med. 44, 454–458. Taeger, D., et al., 2008. Major histopathological patterns of lung cancer related to arsenic exposure in German uranium miners. Int. Arch. Occup. Environ. Health. Tsai, S.M., et al., 1999. Mortality for certain diseases in areas with high levels of arsenic in drinking water. Arch. Environ. Health 54, 186–193. Tsuda, T., et al., 1995. Ingested arsenic and internal cancer: a historical cohort study followed for 33 years. Am. J. Epidemiol. 141, 198–209. Wicks, M.J., et al., 1981. Arsenic exposure in a copper smelter as related to histological type of lung cancer. Am. J. Ind. Med. 2, 25–31. Wu, M.M., et al., 1989. Dose–response relation between arsenic concentration in well water and mortality from cancers and vascular diseases. Am. J. Epidemiol. 130, 1123–1132.
Contents lists available at ScienceDirect
Environmental Research journal homepage: www.elsevier.com/locate/envres
Ingested arsenic, characteristics of well water consumption and risk of different histological types of lung cancer in northeastern Taiwan$ Chi-Ling Chen a, Hung-Yi Chiou c, Ling-I Hsu d, Yu-Mei Hsueh c, Meei-Maan Wu d, Chien-Jen Chen b,d, a
Graduate Institute of Clinical Medicine, College of Medicine, National Taiwan University, 7, Chung-Shan S. Road, 100 Taipei, Taiwan Graduate Institute of Epidemiology, College of Public Health, National Taiwan University, Taipei, Taiwan c School of Public Health, Taipei Medical University, Taipei, Taiwan d Genomics Research Center, Academia Sinica, Taipei, Taiwan b
a r t i c l e in fo
abstract
Article history: Received 31 March 2009 Received in revised form 24 July 2009 Accepted 11 August 2009 Available online 6 September 2009
Our previous study combining two arseniasis-endemic areas in Taiwan confirmed a dose–response association of lung cancer and arsenic exposure. We conducted current analysis to elucidate the dose– response relationship in lower exposure level, and to evaluate whether the association differs in different histological types. In addition, whether specific characteristics of well water consumptions increased lung cancer risk was examined in order to establish a complete risk profile for arsenic exposure. A total of 8086 residents in northeastern Taiwan were followed for 11 years and 6888 participants remained in the final analysis because 1198 residents with unknown arsenic concentration were excluded. The 178 incident lung cancers were ascertained through linkage with the national cancer registry profiles in Taiwan. All analyses were performed by Cox’s proportional hazards regression models. We found a significant dose–response trend (P ¼ 0.001) of lung cancer risk associated with increasing arsenic concentration. There was no apparent increased risk at concentrations between 10 and 100 mg/L, but concentrations between 100 and 300 mg/L showed evidence of excess risk (RR 1.54, 0.97–2.46). The relative risk was 2.25 (95% CI: 1.43, 3.55) for exposure to Z300 mg/L when compared to o10 mg/L. The significant dose–response trends and the synergistic effect of arsenic exposure and cigarette smoking can be found in squamous and small cell carcinomas, but not in adenocarcinoma. Despite lacking statistical precision, when duration is accounted for, all levels of exposure including low concentration were in the direction of increased risk of lung cancer, and these associations tended to increase with longer durations of exposure. This study provides additional evidence linking arsenic to lung cancer, and the indications that arsenic may play a more important role in certain histological type may help with further research in carcinogenic effect of inorganic arsenic on lung cancer. & 2009 Elsevier Inc. All rights reserved.
Keywords: Adenocarcinoma Arsenic Lung cancer Squamous cell carcinoma Small cell carcinoma
1. Introduction Arsenic is a widely distributed semi-metallic element that occurs naturally in various compounds in the crust of the earth. In addition to earlier ecological studies conducted in Taiwan (Chen $ Supported by Grants NSC-83-0412-B002-231, NSC-97-2314-B001-002 from the National Science Council; DOH85-HR-503PL from Department of Health, Executive Yuan; NHRI-EX97-CD9201 from the National Health Research Institutes, Taiwan. All participants provided written informed consents before the data collection procedures that had been reviewed and approved by the IRB of College of Public Health, National Taiwan University. Corresponding author at: Genomics Research Center, Adacemia Sinica, 128 Academia Road, Section 2, Nankang, Taipei 115, Taiwan. Fax: +886 2 2789 8784. E-mail address: [email protected] (C.-J. Chen).
0013-9351/$ - see front matter & 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.envres.2009.08.010
et al., 1992, 1985, 1988; Chen and Wang, 1990; Guo, 2004; Tsai et al., 1999; Wu et al., 1989), recent epidemiological studies, using cohort (Chen et al., 2004; Chiou et al., 1995; Tsuda et al., 1995) or case-control study (Chen et al., 1986; Ferreccio et al., 2000; Heck et al., 2009; Mostafa et al., 2008) designs from various countries, provided further evidence that ingestion of arsenic via drinking water increased risk of lung cancer in a dose–response manner. Our previous analysis combining cohorts from the two arseniasis-endemic areas in Taiwan (Chen et al., 2004) confirmed the elevated risk of lung cancer associated with high arsenic exposure (RR ¼ 3.29, 95% CI: 1.60, 6.78 for Z700 mg/L compared with o10 mg/L), and a significant dose–response relationship. However, there was no conclusive evidence regarding risk estimates in lower exposure group because of small case numbers (Celik et al., 2008). In addition, the proxy measurement of arsenic
ARTICLE IN PRESS 456
C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
exposure in the southwestern cohort in Taiwan in this study added uncertainty to the inference of causal relationship. More information on the hazardous health effect of ingested arsenic exposure at a lower level from epidemiological studies is warranted. Occupation studies (Axelson et al., 1978; Newman et al., 1976; Pershagen et al., 1987; Taeger et al., 2008; Wicks et al., 1981) examined arsenic exposure via inhalation, and one ecological study investigating arsenic from drinking water from Taiwan (Guo et al., 2004) suggested that there might be cell-type specificity of arsenic-related lung cancer, mainly from observing a higher proportion of either squamous cell carcinoma (Guo et al., 2004), or adenocarcinomas (Axelson et al., 1978; Newman et al., 1976; Pershagen et al., 1987; Wicks et al., 1981) in arsenic-exposed population, or both (Taeger et al., 2008). To date, one study (Heck et al., 2009) conducted in a US population with low-to-moderate arsenic exposure using individual toenail arsenic exposure information reported that higher arsenic exposure was associated with small cell and squamous cell carcinoma of the lung. However, toenail arsenic concentration is thought to be representative of only short-term exposure, mostly in the range less than 1 year prior to assessment. The residents from an arseniasis-endemic area in northeastern Taiwan were exposed to arsenic via drinking water at a relatively lower concentration than those in southwestern Taiwan (Chiou et al., 2001). Instead of sharing a couple of wells among the whole village in southwestern endemic areas, the residents in northeastern endemic areas usually have a well in their own backyard and drank from the well water for a long period of time until the implementation of tap water in the early 1990s. We conducted current analysis using only data from northeastern cohort that has better individual measurement of arsenic exposure to elucidate the dose–response relationship with incident lung cancer in lower arsenic exposure levels, and to evaluate whether the dose–response relationship differs in different histological types of lung cancer. In addition, with the supplemented information from the questionnaire, we investigated whether they drank well water for longer duration, drank from birth (latency) and still drank well water at enrollment (recency) added burden to the risk of developing lung cancer. Similarly, it is of interest to evaluate whether the relationship between arsenic concentration and lung cancer risk differs depending on the latency and recency to establishing a complete risk profile for arsenic exposure.
2. Materials and methods 2.1. Study cohort Study participants were recruited from four townships of Tung-Shan, ChuangWei, Chiao-His and Wu-Chieh in the Lanyang Basin, which was located in the northeastern coast of Taiwan. Residents here had consumed water from shallow wells (o40 m in depth) that contained inorganic arsenic since the late 1940s until tap water was implemented in the early 1990s. Detailed description of study area and enrollment procedure was described previously (Chiou et al., 1997, 2001). Briefly, a total of 8102 residents aged Z40 years from 4586 households of 18 villages in the study area participated in the baseline home interview from 1991 to 1994. Since the national identification numbers were used to link to the national cancer registry profile, 16 participants with missing or who shared the same ID numbers were excluded from the analysis, resulting in 8086 participants in the final analysis. Four well-trained local public health nurses used standardized questionnaires to conduct personal interview at study sites. Information collected via questionnaire included demographic characteristics, cigarette smoking and habitual alcohol consumption status, as well as residential and well water consumption history. Residents who agreed to participate all provided written informed consents, and the data collection procedures had been reviewed and approved by the IRB of College of Public Health, National Taiwan University.
2.2. Arsenic exposure Arsenic concentration was estimated using the water samples collected from the wells the participants used. A total of 3901 water samples were collected from individual wells of 3901 households (85.1% of 4586 household in the study) during the home interview. Because the wells of 685 households no longer existed, the arsenic exposure of 1136 residents was classified as ‘‘unknown’’. The concentration of arsenic was determined by hydride generation combined with flame atomic absorption spectrophotometery. The range of the water samples was between undetectable (o0.15 mg/L) and more than 3000 mg/L. We were unable to determine the arsenic concentration of well water samples for 62 participants, resulting in a total of 1198 participants with ‘‘unknown’’ arsenic concentration. Since these 1198 participants with ‘‘unknown arsenic concentration’’ were not associated with the risk of lung cancer, and excluding them from the main analyses did not alter either the results or the conclusions, 6888 remained in the final analysis. In addition, several measurements of arsenic exposure were evaluated, including the duration of drinking well water containing 410 mg/L of arsenic, the starting (latency) and ending age of exposure, the current status of water consumption at the enrollment (recency) and cumulative exposure status (concentration times duration). All these variables were derived from a questionnaire with detailed history of residential addresses and corresponding arsenic concentration of well water in each address that was tested. The cumulative arsenic exposure was calculated as the sum of the products of arsenic concentration (mg/L) and years of drinking water from that specific well for the period from start to quit drinking well water or recruitment if they were still drinking well water.
2.3. Identification of lung cancer cases Each subject’s national identification number was used to link with the national cancer registry profiles in Taiwan to identify newly diagnosed lung cancer cases between date of questionnaire interview and December 31, 2006. This cancer registry system was implemented in 1978 in Taiwan, and was considered a nationwide cancer registry system with updated, accurate and complete information. The different histological types of lung cancer were determined by the International Classification of Diseases for Oncology codes and categorized into ‘‘squamous cell carcinoma’’, ‘‘adenocarcinoma’’, ‘‘small cell carcinoma’’ and the remaining as ‘‘other histological types’’. After an average follow-up period of 11.5 years (SD, 3.6 years) with 79,345 person-years, a total of 178 newly diagnosed lung cancer cases were identified through data linkage with the computerized national cancer registry profile in Taiwan, resulting in a lung cancer incidence of 2.24 cases per 1000 person-years. Among them, 75(42.1%) were squamous cell carcinoma, 51 (28.7%) were adenocarcinoma, 22 (12.4%) were small cell carcinoma and the remaining 30 (16.9%) mostly consisted of ‘‘no microscopic confirmation’’ (n ¼ 12, 6.0%) and ‘‘other malignancy’’ (n ¼ 10, 5.0%). This ‘‘other malignancy’’ indicates primary lung cancer with some rare form of histological types that are not categorized.
2.4. Statistical analyses Nonparametric methods (Kruskal–Wallis or two-sample Wilcoxon rank-sum test) were used to examine the statistical significance of the median level of arsenic concentration, duration of well water consumption, as well as cumulative arsenic exposure by selected demographic characteristics. Person-years for each participant were calculated from the date of questionnaire interview to the date of cancer diagnosis, death, or December 31, 2006, whichever came first. Arsenic concentration was arbitrarily divided into o10, 10–49.9, 50–99.9, 100–299.9 and Z300 mg/L to emphasize the risk of lung cancer associated with lower exposure levels. Since there is still trace amount of arsenic in tap water, measurement error for the cumulative arsenic exposure group of zero is quite likely, we further used o100 mg/L years as reference. Additional reference group of o400 mg/L years was rationalized on the average duration of drinking well water (40 years) times the arsenic concentration allowed (10 mg/L). Relative risks (RRs) and 95% confidence intervals (CIs) were estimated by Cox’s proportional hazards regression models. The adjustment variables in the final model included age (continuous), gender (male/ female), schooling year (0/1–6/46 years), cigarette smoking status (never/ past/current) and habitual alcohol consumption (no/yes). Although there was a significant difference in the distribution of arsenic concentration between the four towns, it was not considered as a potential confounding factor because ‘‘township’’ was not associated with lung cancer. Trends across levels of categorical variables were assessed by testing the statistical significance of a single trend variable coded as the category of exposure. Since the study participants were mostly farmers, previous use of arsenic pesticides may play a role in the association between ingested inorganic arsenic via drinking water and lung cancer risk. The information of pesticide usage was collected via questionnaire, and a sensitivity analysis was performed by restricting to those who reported having never used pesticide. All the above analyses were performed by Stata Statistical Software (Version 8.2, Stata Corp., College Station, Texas).
ARTICLE IN PRESS C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
3. Results The baseline demographic characteristics of study cohort and the distribution of median arsenic concentration (mg/L), duration of well water consumption (years) and cumulative arsenic exposure (mg/L years) are shown in Table 1. The medians of arsenic concentration were similar in different age, gender, and cigarette smoking groups. Those who were older, male, were residents in Tong-Shang, received less schooling years, smoked cigarettes, were habitual alcohol drinkers and reported never being exposed to pesticide were exposed to higher cumulative arsenic exposure. Participants with ethnicity other than ‘‘Mingnan’’, which makes up the majority, appeared exposed to relatively lower arsenic levels, shorter duration and hence lower cumulative exposure. The variation in distribution of medians was most pronounced in the four townships, with residents in Tong-Shang being exposed to the highest concentration and cumulative exposure from well water, and Wu-Che had a median concentration of zero (interquartile: 0–28.0) but the longest duration of exposure. The mean arsenic concentration among wells with known arsenic concentration was 117.2 mg/L (SD ¼ 297.2 mg/L), and the mean duration of consuming well water was 42.0 years (SD ¼ 15.1 years), resulting in the mean cumulative arsenic exposure of 3523.5 mg/L years (SD ¼ 9443.5 mg/ L years) (Table 1). Among 4600 residents who were exposed to arsenic concentration Z10 mg/L, 3005 (65.3%) who responded still drank well water at the time of enrollment, and 1555 reported having stopped drinking well water before the enrollment (33.8%) (Table 2). A total of 1753 (21.7%) residents started drinking well water from birth, and about 15 percent drank well water for more than 50 years. Overall, there was about 12 percent study participants who either never drank well water or the arsenic concentration of the well water they drank were below the detection limit. Those who still drank well water containing Z10 mg/L of arsenic at enrollment or drank it from birth were associated with about a 30% increase in risk of lung cancer (RR ¼ 1.28, 95% CI: 0.90, 1.83 for still drinking; RR ¼ 1.32, 95% CI: 0.87, 1.98 for drinking from birth). The dose–response trends were all statistically significant when using three different reference groups, and the highest relative risks were around 2-fold for the highest cumulative exposure of 410,000 mg/L years (RR ¼ 1.53, 95% CI: 0.93, 2.54; RR ¼ 1.79, 95% CI: 1.12, 2.88 and RR ¼ 2.10, 95% CI: 1.34, 3.31 for 0, o100 and o400 as references, respectively). Table 3 shows the multivariate-adjusted relative risk of categories of arsenic concentration in relation to overall and different histological types of lung cancer. A significant dose– response trend (Ptrend ¼ 0.001) was found for increasing arsenic concentrations using o10 mg/L as the reference group, after adjusting for age, gender, years of education, cigarette smoking and habitual alcohol consumption status at enrollment. There was no apparent increased risk at concentrations between 10 and 100 mg/L, but concentrations between 100 and 300 mg/L showed evidence of excess risk (RR 1.54, 0.97–2.46). The relative risk was 2.25 (95% CI: 1.43, 3.55) for exposure Z300 mg/L when compared to the reference category. The significant dose–response trend of increasing arsenic concentrations and increasing risk of lung cancer was found in squamous cell carcinomas (Ptrend ¼ 0.004) and small cell carcinomas (Ptrend ¼ 0.021), but not in adenocarcinomas (Ptrend ¼ 0.664) and other histological types (Ptrend ¼ 0.197). The highest risk was found in small cell carcinomas with arsenic concentration Z300 mg/L (RR ¼ 5.30, 95% CI: 1.48, 18.9). We further investigated whether the association between arsenic concentration and lung cancer risk differed between those who reported still drinking well water at the enrollment (no/yes)
457
and the time of first exposure (from birth/not from birth) (Table 4). The dose–response trend was more prominent in those who still drank well water at enrollment (Ptrend ¼ 0.002) than those who stopped (Ptrend ¼ 0.115). The pattern of incremental lung cancer risk associated with increasing arsenic concentration was similar among those who started drinking well water from birth and those who started after birth. When compared to those who have never smoked cigarettes and consumed well water containing arsenic o10 mg/L, residents who smoked cigarettes Z25 pack-years and consumed well water with an arsenic level Z100 mg/L had a 7-fold increased risk of lung cancer (RR ¼ 6.97, 95% CI: 3.4, 14.3) (Table 5). And this increased risk was as high as 11-fold (RR ¼ 10.9, 95% CI: 3.57, 33.5) for squamous cell carcinoma combined with small cell carcinoma. Since the proportion of adenocarcinoma was known to be relatively higher in women and the prevalence of cigarette smoking was very low for women in this cohort, the analyses were repeated on men only. We found similar results with regard to the risk patterns, only that the risk estimates were higher with much wider 95 percent confidence intervals due to the smaller sample size (Table 5). All p-values for interactions between packyears of cigarette smoked (never/o25/Z25) and arsenic concentration (o10/10–99.9/Z100 mg/L) were not statistically significant. The combination of arsenic concentration and duration showed that those who drank well water containing high concentration of arsenic (Z300 mg/L) for over 50 years are at almost 10-fold increased risk of lung cancer (RR ¼ 9.71, 95% CI: 2.84, 33.2) when compared to those who drank well water below 10 mg/L for less than 30 years (Fig. 1). Among residents exposed to the same level of arsenic concentration, longer duration was associated with increased risk of lung cancer, and this trend was more prominent in the lowest and the highest arsenic concentration group.
4. Discussion In this analysis, we confirmed our earlier finding of a significant dose–response trend of increasing lung cancer risk with increasing arsenic concentration at a lower exposure level. In addition, the significant dose–response trend can be found in squamous cell and small cell carcinomas, but not in adenocarcinoma. Most studies investigating arsenic exposure via inhalation suggest a higher proportion of adenocarcinomas and a lower proportion of squamous cell carcinoma (Axelson et al., 1978; Newman et al., 1976; Pershagen et al., 1987; Wicks et al., 1981). A recent re-analysis of data from German uranium miners by Taeger et al. (2008) reported that arsenic exposure was associated with non-small cell lung cancer and the specific cell type in relation to arsenic was different mainly determined by silicosis status. However, two other studies evaluating arsenic exposure via drinking water reported similar results to the current analysis. One was an ecological study from Taiwan (Guo et al., 2004). They found that patients from arseniasis-endemic areas had higher proportions of squamous cell and small cell carcinomas, but a lower proportion of adenocarcinomas. The other was a population-based case-control study conducted in low-to-moderate arsenic exposure areas in the US (Heck et al., 2009). They reported that compared to o0.05 mg/g of toenail arsenic concentration, those with Z0.114 mg/g were associated with an about 3-fold increased risk of small cell and squamous cell carcinoma (OR ¼ 2.75, 95% CI 1.00, 7.57). We also examined the duration, recency and latency of drinking well water containing arsenic Z10 mg/L and found that drinking arsenic well water from birth and early life, still drinking
ARTICLE IN PRESS 458
C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
Table 1 Distribution of selected demographic characteristics of 6888 study participants from arseniasis-endemic area in northeastern Taiwan. Characteristics
Age at recruitment r50 50–54.9 55–59.9 Z60 P value Mean age Mean years of follow-up Gender Male Female P valuea Ethnicity Mingnan Others Unknownb P valuea Township of residence Chung-Wei Tong-Shang Chiou-Shi Wu-Che P valuec Schooling years 0 1–6 46 Unknownb P valuec Cigarette smoking status at enrollment Never Past Current Unknownb P valuec Years of cigarette smoking 0 1–35 435 Unknownb P valuec Cigarettes smoked per day 0 1–19 Z20 Unknownb P valuec Pack-years of cigarettes smoking 0 1–25 425 Unknownb P valuec Habitual alcohol consumption No Yes Unknownb P valuea Exposure to pesticide Never Ever Unknownb P valuea Mean concentration in well water (lg/L) Mean duration of consuming well water (years) Mean cumulative arsenic exposure ((lg/L) years) a b c
Number (n ¼ 6888)
Percentage or mean (SD)
Median arsenic concentration (lg/L) (25%, 75%)
Median duration (25%, 75%)
Median cumulative arsenic exposure (lg/L years) (25%, 75%)
1538 2338 1801 1211
22.3 33.9 26.2 17.6
25.6 (3.1, 28.3 (3.5, 29.6 (3.6, 25.8 (3.5, 0.7929
96.4) 93.7) 92.5) 88.5)
39 (29, 44) 40 (31, 53) 42 (33, 61) 45 (31, 67) o0.0001
677 (785, 2237) 840 (90, 2785) 943 (106, 3030) 980 (95, 3192) o0.0001
59.1 (11.0) 11.5 (3.6) 3481 3407
50.5 49.5
28.0 (3.8, 92.8) 27.6 (3.1, 93.2) 0.5993
41 (31, 52) 41 (31, 53) 0.6978
915 (109, 3020) 777 (78, 2559) 0.0005
6819 58 11
99.0 0.8 0.2
28.0 (3.5, 93.2) 1.7 (0, 48.2) 3.5 (0, 30.0) 0.0002
41 (31, 52) 52 (40, 66) 58 (44, 75) 0.0002
855 (96, 2800) 0 (0, 603) 158 (0, 1894) o0.0001
3387 1016 1394 1091
49.2 14.8 20.2 15.8
26.7 (8.5, 88.4) 93.7 (56.5, 207.4) 20.2 (1.5, 63.7) 0 (0, 28.0) 0.0001
40 (31, 43 (33, 40 (30, 43 (31, 0.0005
52) 53) 54) 55)
871 (224, 2558) 3279 (1436, 6157) 7 (63, 1926) 69 (0, 979) 0.0001
2011 4049 456 372
29.2 58.8 6.6 5.4
27.3 (4.0, 87.8) 27.8 (2.7, 93.3) 20.0 (1.4, 69.6) 46.0 (14.3, 157.6) 0.0035
41 (31, 41 (31, 40 (30, 43 (33, 0.0001
56) 52) 47) 54)
848 (92, 2900) 843 (85, 2800) 564 (17, 1659) 1233 (244, 4092) 0.0001
4065 854 1967 2
59.0 12.4 28.6 –
27.0 (3.2, 94.2) 28.6 (3.0, 92.1) 28.3, (4.0, 91.4) – 0.8213
41 (31, 52) 42 (31, 54) 40 (31, 53) – 0.1391
777 (80, 2605) 952 (116, 3143) 947 (115, 2938) – 0.0014
4065 1261 1461 101
59.0 18.3 21.2 1.5
27.0 (3.2, 27.6 (4.2, 29.5 (3.5, 26.1 (3.0, 0.5347
41 (31, 40 (31, 41 (31, 46 (35, 0.0001
777 (80, 2605) 972 (136, 2934) 943 (95, 3068) 789 (52, 3580) 0.001
4065 1110 1711 2
59.0 16.1 24.8 –
27.0 (3.2, 94.2) 32.3, (4.7, 96.8) 26.3 (3.0, 90.0) – 0.0305
41 (31, 52) 41 (31, 54) 40 (31, 52) – 0.1949
777 (80, 2605) 1051 (137, 3153) 867 (92, 2931) – 0.0001
4065 1021 1680 122
59.0 14.8 24.4 1.8
27.0 (3.2, 30.1 (4.0, 27.8 (3.6, 31.4 (3.0, 0.3829
41 (31, 41 (31, 40 (31, 46 (35, 0.2306
777 (80, 2605) 992 (135, 2940) 893 (97, 2989) 850 (65, 3668) 0.0011
5564 1314 10
80.8 19.1 0.2
26.3 (3.2, 90.3) 32.8 (5.2, 107.2) 47.2 (3.4, 77.0) 0.0021
41 (31, 53) 40 (31, 52) 44 (40, 55) 0.1499
807 (85, 2605) 1010 (147, 3515) 652 (0, 3182) o0.0001
6404 474 10
93.0 6.9 0.2
28.1 (3.6, 94.5) 20.2 (1.6, 66.3) 17.7 (0, 39.1) 0.0016
41 (31, 52) 41 (33, 54) 44 (35, 64) 0.0719
845 (92, 2797) 748 (80, 2576) 698 (0, 2237) 0.3382
117.2 (297.2) 42.0 (15.1) 3523.5 (9443.5)
P values were based on two-sample Wilcoxon rank-sum test. Those with unknown status were excluded from the statistical testing. P values were based on the Kruskal–Wallis test.
94.2) 93.4) 91.0) 81.7)
94.2) 93.2) 91.2) 85.3)
52) 50) 58) 58)
52) 50) 54) 58)
ARTICLE IN PRESS C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
459
Table 2 Distribution and multivariate adjusted relative risks of selected characteristics of arsenic-containing well water consumption in relation to lung cancer in northeastern arseniasis-endemic area in Taiwan. %
P-Y
# Cases (N ¼ 178)
RR (95% CI)a
Well water containing arsenic drinking status at enrollment 2288 Arsenic o10 mg/L Still drinking 3005 Stopped 1555 Cannot be determined 40
28.3 37.2 19.2 0.5
26,568 34,009 18,320 428
48 83 43 4
1.00 1.28 (0.89–1.82) 1.31 (0.87–1.98) 5.26 (1.88–14.7)
Age when started drinking well water containing arsenic 2288 Arsenic concentration o10 mg/L From birth 1753 40–20 1015 420–30 965 430 824 Cannot be determined 43
28.3 21.7 12.6 11.9 10.2 0.5
26,568 20,901 12,184 11,169 7991 512
48 45 24 35 23 3
1.00 1.31 1.16 1.68 1.04 4.22
(0.87–1.97) (0.71–1.90) (1.08–2.59) (0.63–1.73) (1.31–13.6)
Age when stopped drinking well water containing arsenic 2288 Arsenic concentration o10 mg/L Still drinking 3005 r50 583 450–60 509 460–70 309 470 154 Cannot be determined 40
28.3 37.2 7.2 6.3 3.8 1.9 0.5
26,568 34,009 7490 6096 3471 1220 472
48 83 9 15 11 9 3
1.00 1.27 1.11 1.36 1.16 2.08 4.45
(0.89–1.82) (0.53–2.32) (0.76–2.44) (0.60–2.25) (0.98–4.41) (1.38–14.3)
Years of drinking well water containing arsenic 2288 Arsenic concentration o10 mg/L o30 919 30–o40 1242 40–o50 1030 50–o60 799 60+ 569 Duration unknown 41
28.3 11.4 15.4 12.7 9.9 7.0 0.5
26,568 10,719 14,253 12,053 9457 5792 483
48 18 38 19 33 19 3
1.00 0.99 1.45 1.02 1.83 1.12 4.40
(0.58–1.71) (0.95–2.22) (0.60–1.74) (1.17–2.86) (0.65–1.93) (1.36–14.2)
23.9 43.3 32.8
19,196 34,908 25,207 14
31 69 78 0
1.00 1.22 (0.80–1.87) 1.44 (0.95–2.20) – 0.082
1071 2583 2078 524 632
13.3 31.9 25.7 6.5 7.8
12,558 30,161 23,732 5874 7001
32 43 51 23 29
1.00 0.56 (0.36–0.89) 0.78 (0.50–1.21) 1.37 (0.80–2.34) 1.52 (0.92–2.52) 0.002
1729 1925 2078 524 632
21.4 23.8 25.7 6.5 7.8
20,033 22,685 23,732 5874 7001
43 32 51 23 29
1.00 0.65 (0.41–1.02) 0.91 (0.60–1.36) 1.60 (0.96–2.65) 1.78 (1.11–2.85) 0.002
Cumulative arsenic exposure ((lg/L) years) o400 2534 400–o1000 1120 1000–o5000 2078 5000–o10,000 524 Z10,000 632 P trend
31.3 13.9 25.7 6.5 7.8
29,554 13,164 23,732 5874 7001
55 20 51 23 29
1.00 0.83 (0.50–1.39) 1.06 (0.73–1.56) 1.87 (1.15–3.04) 2.08 (1.33–3.27) 0.001
Characteristics
Number (N ¼ 6888)
Years of drinking well water (regardless of As conc.) r30 1646 430–r50 2981 450 2260 Unknown 1 P trend Cumulative arsenic exposure ((lg/L) years) 0 o1000 1000–o5000 5000–o10,000 Z10,000 P trend Cumulative arsenic exposure ((lg/L) years) o100 100–o1000 1000–o5000 5000–o10,000 Z10,000 P trend
a
Adjusted for age (1-year increment), gender, education levels, cigarettes smoking status and habitual alcohol consumption status at enrollment.
arsenic well water at the time of enrollment or stopping at older age was associated with slightly but not statistically significant higher risk of developing lung cancer. Furthermore, the cumulative arsenic exposure was associated with an increased risk of lung cancer in a dose–response fashion. But the significantly reduced risk of lung cancer (RR ¼ 0.56, 95% CI: 0.36, 0.89) for o1000 mg/L years compared to 0 mg/L years may indicate a threshold for arsenic exposure in relation to lung cancer. However, due to the possibility of misclassification in this group (0 mg/L years) because of the detection limit (0.15 mg/L), it warranted a cautious
interpretation. One recent ecological study (Marshall et al., 2007) investigating 50 years of lung cancer mortality in Region II of Chile found that the highest lung cancer relative risks were found in men who were exposed at earlier life (as children or adolescents). Our study found that consumption of well water containing arsenic 410 mg/L at earlier age (before 30 years old) was associated with increased lung cancer risk. The combination of arsenic concentration and duration showed that when the duration is accounted for, even at o30 years, all levels of exposure including low concentration were in the direction
ARTICLE IN PRESS 460
C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
Table 3 Incidence rates and multivariate adjusted relative risks of arsenic concentration in relation to different histological types of lung cancer in northeastern arseniasis-endemic area in Taiwan. Arsenic concentration (lg/L)
o10
10–49.9
50–99.9
100–299.9
Z300
Ptrend
All lung cancer$ Cases number (n ¼ 178) Incidence (per 105) RRa (95% CI)
48 181 1.00 Referent
51 211 1.10 (0.74–1.63)
20 194 0.99 (0.59–1.68)
28 269 1.54 (0.97–2.46)
31 397 2.25 (1.43–3.55)
0.001
Squamous cell carcinoma$ Cases number (n ¼ 75) Incidence (per 105) RRa (95% CI)
23 87 1.00 Referent
12 50 0.53 (0.26–1.07)
13 126 1.32 (0.67–2.61)
13 125 1.52 (0.77–3.00)
14 179 2.13 (1.09–4.17)
0.004
Adenocarcinoma$ Cases number (n ¼ 51) Incidence (per 105) RRa (95% CI)
14 53 1.00 Referent
20 83 1.50 (0.76–2.98)
4 39 0.70 (0.23–2.13)
6 58 1.06 (0.41–2.77)
7 90 1.63 (0.65–4.05)
0.664
Small cell carcinoma Cases number (n ¼ 22) Incidence (per 105) RR (95% CI)
4 15 1.00 Referent
8 33 2.02 (0.61–6.73)
0 0 –
4 38 2.77 (0.69–11.1)
6 77 5.15 (1.44–18.4)
0.021
Other histological types$ Cases number (n ¼ 30) Incidence (per 105) RRa (95% CI)
7 26 1.00 Referent
11 45 1.70 (0.66–4.39)
3 29 1.10 (0.28–4.25)
5 48 2.03 (0.64–6.40)
4 51 2.25 (0.65–7.71)
0.197
$
a $
Adjusted for age (1-year increment), gender, education levels, cigarettes smoking status and habitual alcohol consumption status at enrollment. Arsenic levels unknown were excluded from the analysis, resulted in 6,888 residents with 178 incident lung cancer cases.
Table 4 Multivariate adjusted relative risks of lung cancer associated with arsenic concentration stratified by duration and well water drinking status at enrollment. Arsenic concentration in well water (lg/L)
All lung cancer$ o10 10–49.9 50–99.9 100–299.9 Z300 P trend a $
Well water drinking status at enrollment
Age when started drinking well water
Still drinking RR (95% CI)a
Stopped drinking RR (95% CI)a
Started from birth RR (95% CI)a
Not started from birth RR (95% CI)a
1.00 1.00 (0.65–1.53) 1.14 (0.62–2.11) 1.34 (0.73–2.43) 2.65 (1.56–4.48) 0.002
1.00 1.32 (0.72–2.40) 0.69 (0.30–1.62) 1.55 (0.82–2.93) 1.73 (0.90–3.35) 0.115
1.00 1.06 (0.60–1.87) 0.80 (0.34–1.88) 1.35 (0.68–2.67) 2.39 (1.28–4.44) 0.027
1.00 1.08 (0.70–1.68) 1.01 (0.55–1.87) 1.57 (0.90–2.73) 2.09 (1.21–3.61) 0.009
Adjusted for age (1-year increment), gender, education levels, cigarettes smoking status and habitual alcohol consumption status at enrollment. Arsenic levels unknown were excluded from the analysis, resulted in 6,888 residents with 178 incident lung cancer cases.
of increased risk of developing lung cancer. In addition, these associations tend to increase with longer durations of exposure. However, these associations at low exposure level were not statistically significant due to inadequate power. Nevertheless, these results provided us additional insights into the carcinogenesis process of arsenic on lung cancer. That is, those who are exposed to arsenic in high concentrations for a long period of time are at a much higher risk of developing lung cancer than those who are either exposed to lower concentrations or for a shorter duration. Our previous study found indications of a synergistic effect of high arsenic exposure (Z700 mg/L) and cigarette smoking, and lung cancer risk could be as high as more than 10-fold when compared with non-smokers with an arsenic exposure level o10 mg/L (Chen et al., 2004). We repeated similar analysis on the cohort from northeastern Taiwan alone and found that the increased risk reached 7-fold at a much lower arsenic exposure level (Z100 mg/L) for heavy smokers (425 pack-years). And this indication of synergistic effect was found only in squamous and small cell carcinoma, but not in adenocarcinoma.
Strengths of our study include the large sample size and longer follow-up periods. This cohort is quite homogeneous, with regard to socio-economic status (mostly farmers) and years when the tap water system commenced. All arsenic concentration information was based on individual well water samples that the cohort members drank most of their water from for a long period of time. In addition, we were able to further examine other components of arsenic exposure including the starting and stopping age of consumption of well water, and hence duration of exposure and cumulative exposure including concentration and duration based on a structured questionnaire interview of detailed residential and well water drinking history. The prospective follow-up design provides us with correct temporal relationship of the exposure and the disease. The possibility of selection bias is minimized since all incident lung cancer cases were identified through data linkage. One limitation of this study is that almost 15% of study participants are with unknown arsenic concentration information because their wells no longer existed when the study was conducted. There was no difference in age and gender distribution between the residents with unknown concentration and residents
ARTICLE IN PRESS C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
461
Table 5 Relative risks of lung cancer in relation to arsenic exposure and cigarette smoking in northeastern arseniasis-endemic area in Taiwan. All participants RRa (95% CI)
Arsenic concentration (lg/L)
Pack-years of smoking All lung cancer Never
o25
Z25
P for interaction Squamous and small cell carcinoma combined Never
o25
Z25
P for interaction Adenocarcinoma Never
o25
Z25
o10 10–99.9 Z100 o10 10–99.9 Z100 o10 10–99.9 Z100
1.00 1.22 (0.64–2.32) 1.32 (0.64–2.74) 2.14 (0.79–5.79) 1.52 (0.56–4.15) 5.30 (2.19–12.8) 4.08 (1.83–9.10) 4.19 (1.92–9.14) 8.17 (3.74–17.9) 0.3665
1.00 1.92 (0.37–9.91) 1.38 (0.20–9.83) 2.96 (0.61–14.3) 2.09 (0.43–10.1) 5.39 (1.16–25.1) 5.21 (1.23–22.1) 5.27 (1.26–22.0) 10.1 (2.42–42.5) 0.7205
o10 10–99.9 Z100 o10 10–99.9 Z100 o10 10–99.9 Z100
1.00 0.45 (0.11–1.90) 1.39 (0.40–4.80) 2.21 (0.45–10.9) 1.05 (0.17–6.29) 7.41 (1.91–28.8) 7.20 (2.05–25.3) 7.05 (2.03–24.4) 14.6 (4.25–50.3) 0.5387
1.00 0.77 (0.05–12.3) 1.37 (0.09–22.0) 2.51 (0.26–24.1) 1.19 (0.11–13.1) 7.22 (0.87–60.2) 7.78 (1.03–58.5) 7.55 (1.02–56.0) 15.4 (2.06–114.2) 0.7905
o10 10–99.9 Z100 o10 10–99.9 Z100 o10 10–99.9 Z100
1.00 1.45 (0.62–3.43) 1.06 (0.37–3.06) 1.83 (0.41–8.19) 1.95 (0.51–7.42) 1.54 (0.27–8.63) 0.85 (0.18–4.01) 0.60 (0.13–2.86) 2.10 (0.53–8.39) 0.4684
1.00 1.56 (0.14–17.2) 1.29 (0.08–20.6) 2.22 (0.23–21.5) 2.36 (0.27–20.4) 0.85 (0.05–13.8) 0.88 (0.09–8.66) 0.62 (0.06–6.11) 2.25 (0.26–19.7) 0.5044
P for interaction a
Men only RRa (95% CI)
Adjusted for age (1-year increment), gender, education levels and habitual alcohol consumption status at enrollment.
Duration (years)
< 30
Arsenic Concentration (µg/L)
<10
Odds Ratio* (95% Confidence Interval)
1.00
< 30
10 –49.9
< 30
50 –99.9
3.28 (0.82 –13.1)
< 30
100 –299.9
2.18 (0.44 –10.8)
< 30
> 300
1.91 (0.39 –9.49)
<10
2.21 (0.63 –7.70)
>30-50
2.45 (0.68 –8.79)
>30-50
10 –49.9
2.76 (0.83 –9.21)
>30-50
50 –99.9
1.75 (0.44 –7.02)
>30-50
100 –299.9
4.45 (1.28 –15.4)
>30-50
> 300
4.78 (1.33 –17.2)
<10
3.01 (0.92 –9.91)
>50
10 –49.9
2.66 (0.77 –9.16)
>50
50 –99.9
2.43 (0.63 –9.43)
>50
100 –299.9
3.35 (0.91 –12.4)
>50
> 300
9.61 (2.81 –32.8)
>50
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20
Odds Ratio (95% Confidence Interval)
Fig. 1. Multivariate adjusted relative risks of combination of duration and arsenic concentration in relation to lung cancer in northeastern arseniasis-endemic area in Taiwan. *Adjusted for age (1-year increment), gender, education levels, cigarette smoking status and habitual alcohol consumption status at enrollment.
with known concentration. When categorizing this group of people separately, we found that excluding them from the analysis did not alter the analysis results. Although the misclassification of
low exposure groups cannot be ruled out, we have no reason to believe that these misclassifications would be dependent on their arsenic exposure. Thus this non-differential misclassification
ARTICLE IN PRESS 462
C.-L. Chen et al. / Environmental Research 110 (2010) 455–462
would pull the association estimates toward the null and underestimate the true relationship. However, even though the number of participants with unknown arsenic well water drinking status at enrollment and duration is small, it is associated with a significantly higher risk of lung cancer. This may be due to the fact that the mean arsenic concentration was 246.6 mg/L (SD ¼ 435.9 mg/L) among these 41 participants with unknown well water consumption status at enrollment and duration, while the mean was 174.0 mg/L (SD ¼ 349.0 mg/L) for those with those information. Since most of the study participants were farmers, it may raise concern of potential misclassification when they had a chance to be exposed to pesticide containing arsenic. We restricted our analysis to residents who were not exposed to pesticides and the results did not alter. Therefore, the possibility of bias caused by the misclassification was minimal. This is the first study to investigate individual arsenic exposure that included not only concentration but also duration and other characteristics of drinking well water in relation to overall and different histological types of lung cancer. The results provided additional evidence linking arsenic to lung cancer not only by arsenic concentration alone but also by characteristics of well water consumption. And more importantly, the indication that arsenic may play a more important role in certain histological type of lung cancer may provide information for further research in carcinogenic effect of inorganic arsenic on lung cancer.
References Axelson, O., et al., 1978. Arsenic exposure and mortality: a case-referent study from a Swedish copper smelter. Br. J. Ind. Med. 35, 8–15. Celik, I., et al., 2008. Arsenic in drinking water and lung cancer: a systematic review. Environ. Res. 108, 48–55. Chen, C.J., et al., 1992. Cancer potential in liver, lung, bladder and kidney due to ingested inorganic arsenic in drinking water. Br. J. Cancer 66, 888–892. Chen, C.J., et al., 1985. Malignant neoplasms among residents of a blackfoot disease-endemic area in Taiwan: high-arsenic artesian well water and cancers. Cancer Res. 45, 5895–5899.
Chen, C.J., et al., 1986. A retrospective study on malignant neoplasms of bladder, lung and liver in blackfoot disease endemic area in Taiwan. Br. J. Cancer 53, 399–405. Chen, C.J., Wang, C.J., 1990. Ecological correlation between arsenic level in well water and age-adjusted mortality from malignant neoplasms. Cancer Res. 50, 5470–5474. Chen, C.J., et al., 1988. Atherogenicity and carcinogenicity of high-arsenic artesian well water. Multiple risk factors and related malignant neoplasms of blackfoot disease. Arteriosclerosis 8, 452–460. Chen, C.L., et al., 2004. Ingested arsenic, cigarette smoking, and lung cancer risk: a follow-up study in arseniasis-endemic areas in Taiwan. JAMA 292, 2984–2990. Chiou, H.Y., et al., 2001. Incidence of transitional cell carcinoma and arsenic in drinking water: a follow-up study of 8102 residents in an arseniasis-endemic area in northeastern Taiwan. Am. J. Epidemiol. 153, 411–418. Chiou, H.Y., et al., 1995. Incidence of internal cancers and ingested inorganic arsenic: a seven-year follow-up study in Taiwan. Cancer Res. 55, 1296–1300. Chiou, H.Y., et al., 1997. Dose–response relationship between prevalence of cerebrovascular disease and ingested inorganic arsenic. Stroke 28, 1717–1723. Ferreccio, C., et al., 2000. Lung cancer and arsenic concentrations in drinking water in Chile. Epidemiology 11, 673–679. Guo, H.R., 2004. Arsenic level in drinking water and mortality of lung cancer (Taiwan). Cancer Causes Control 15, 171–177. Guo, H.R., et al., 2004. Cell type specificity of lung cancer associated with arsenic ingestion. Cancer Epidemiol. Biomarkers Prev. 13, 638–643. Heck, J.E., et al., 2009. Lung cancer in a US population with low to moderate arsenic exposure. Environ. Health Perspect. Marshall, G., et al., 2007. Fifty-year study of lung and bladder cancer mortality in Chile related to arsenic in drinking water. J. Natl. Cancer Inst. 99, 920–928. Mostafa, M.G., et al., 2008. Lung cancer and exposure to arsenic in rural Bangladesh. Occup. Environ. Med. 65, 765–768. Newman, J.A., et al., 1976. Histologic types of bronchogenic carcinoma among members of copper-mining and smelting communities. Ann. NY Acad. Sci. 271, 260–268. Pershagen, G., et al., 1987. Histological types of lung cancer among smelter workers exposed to arsenic. Br. J. Ind. Med. 44, 454–458. Taeger, D., et al., 2008. Major histopathological patterns of lung cancer related to arsenic exposure in German uranium miners. Int. Arch. Occup. Environ. Health. Tsai, S.M., et al., 1999. Mortality for certain diseases in areas with high levels of arsenic in drinking water. Arch. Environ. Health 54, 186–193. Tsuda, T., et al., 1995. Ingested arsenic and internal cancer: a historical cohort study followed for 33 years. Am. J. Epidemiol. 141, 198–209. Wicks, M.J., et al., 1981. Arsenic exposure in a copper smelter as related to histological type of lung cancer. Am. J. Ind. Med. 2, 25–31. Wu, M.M., et al., 1989. Dose–response relation between arsenic concentration in well water and mortality from cancers and vascular diseases. Am. J. Epidemiol. 130, 1123–1132.