Arsenic levels in drinking water and mortality of liver cancer in Taiwan

Journal of Hazardous Materials 262 (2013) 1132–1138

Contents lists available at ScienceDirect

Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat

Arsenic levels in drinking water and mortality of liver cancer in Taiwan Hung-Jung Lin a,b,c,d,1 , Tzu-I Sung e,f,1 , Chi-Yi Chen g , How-Ran Guo e,f,h,∗ a

Cancer Research Center, Chi-Mei Medical Center, Tainan, Taiwan Department of Emergency Medicine, Chi-Mei Medical Center, Tainan, Taiwan c Department of Biotechnology, Southern Taiwan University of Science and Technology, Tainan, Taiwan d Graduate Institute of Injury Prevention and Control, College of Public Health and Nutrition, Taipei Medical University, Taipei,Taiwan e Center for Occupational Health and Preventative Medicine, College of Medicine, National Cheng Kung University, Tainan, Taiwan f Department of Environmental and Occupational Health, College of Medicine, National Cheng Kung University, Tainan, Taiwan g Department of Internal Medicine, Ditmanson Medical Foundation Chia-Yi Christian Hospital, Chia-Yi, Taiwan h Department of Occupational and Environmental Medicine, National Cheng Kung University Hospital, Tainan, Taiwan b

h i g h l i g h t s  The carcinogenic effect of arsenic is well documented, but epidemiologic data on liver cancer were limited.  Arsenic levels above 0.64 mg/L in drinking water were associated with an increase in the liver cancer mortality in both genders.  The carcinogenic effect of arsenic on liver is not prominent at lower exposure levels.

a r t i c l e

i n f o

Article history: Received 18 February 2012 Received in revised form 11 August 2012 Accepted 27 December 2012 Available online 4 January 2013 Keywords: Arsenic Liver cancer Drinking water Dose–response Epidemiology Environmental health Risk assessment

a b s t r a c t The carcinogenic effect of arsenic is well documented, but epidemiologic data on liver cancer were limited. To evaluate the dose–response relationship between arsenic in drinking water and mortality of liver cancer, we conducted a study in 138 villages in the southwest coast area of Taiwan. We assessed arsenic levels in drinking water using data from a survey conducted by the government and reviewed death certificates from 1971 to 1990 to identify liver cancer cases. Using village as the unit, we conducted multivariate regression analyses and then performed post hoc analyses to validate the findings. During the 20-year period, 802 male and 301 female mortality cases of liver cancer were identified. After adjusting for age, arsenic levels above 0.64 mg/L were associated with an increase in the liver cancer mortality in both genders, but no significant effect was observed for lower exposure categories. Post hoc analyses and a review of literature supported these findings. We concluded that exposures to high arsenic levels in drinking water are associated with the occurrence of liver cancer, but such an effect is not prominent at exposure levels lower than 0.64 mg/L. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Arsenic is widely distributed in the natural environment and often transported by water [1]. Excess exposures to arsenic from drinking water have been found in Taiwan [2], China [3], Japan [4], Bangladesh [5], Thailand [6], Mexico [7], Argentina [8], Chile [9], Hungary [10], Romania [11], Italy [12], the United States [13], and many other countries [1,14–17]. In fact, the health effects of arsenic in drinking water have become a global crisis, and it was estimated that more than 80 million people

∗ Corresponding author at: Department of Environmental and Occupational Health, College of Medicine, National Cheng Kung University, 138 Sheng-Li Road, Tainan 70428, Taiwan. Tel.: +886 6 235 3535x5802; fax: +886 6 275 2484. E-mail address: [email protected] (H.-R. Guo). 1 These authors contributed equally to the work. 0304-3894/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jhazmat.2012.12.049

were at risk from arsenic in drinking water and in food crops in Bangladesh alone [5,18], with 30 million drinking water containing five times the quality guideline of 0.01 mg/L proposed by World Health Organization (WHO) [5]. Acute and chronic arsenic exposures were found to have dermal, respiratory, cardiovascular, gastrointestinal, hematological, hepatic, renal, neurological, developmental, reproductive, immunologic, genotoxic, mutagenetic, and carcinogenic effects on humans [14,16,19]. The diseases and conditions of the liver observed in previous studies included proliferation of Kupffer cells, granulomas, fibrosis, cirrhosis, and portal hypertension [20–23]. In addition, the carcinogenic effect of arsenic has been documented for more than a century [24]. Over the years, the data from epidemiology studies in the blackfoot disease (BFD) endemic area in Taiwan [25,26] have been serving as a major basis of the regulatory standards of WHO and United States Environmental Protection Agency (USEPA)

H.-J. Lin et al. / Journal of Hazardous Materials 262 (2013) 1132–1138

[25,27–29]. BFD is an obstructive peripheral vascular disease characterized by the black discoloration due to gangrenous changes of lower extremities [30,31], and it was most prevalent in four townships in the southwest coast region in Taiwan, which is generally referred to as the “BFD endemic area” [30,32]. In this area, many artesian wells that were used by residents as sources of drinking water had high levels of arsenic, and it is generally believed that BFD was caused by arsenic in drinking water because dose–response associations between the arsenic level in drinking water and prevalence of BFD were reported [30,33]. The argument was supported by the fact that skin changes caused by arsenic intoxication, including skin cancer, were also prevalent in this area [30,33]. As more and more studies showed that arsenic in drinking water might also cause internal cancers, data collected by a study in the BFD endemic area [26] were adopted as a major scientific basis of justifying the revisions of regulatory standards of arsenic due to excess risks of lung and urinary cancers [25,34]. In addition to cancers of skin, lung, and urinary bladder, liver cancers were also observed among patients of arsenic intoxication since the 1950s [35]. However, epidemiologic data are quite limited, especially those on the dose–response relationship. An ecological study in the BFD endemic area categorized the 42 villages into three exposure groups and found a positive dose–response relationship between arsenic levels and mortality of liver cancer in men (but not in women) [26], but no obvious dose–response relationship was observed when Morales et al. [34] broke down the villages further into eight exposure groups. A study in Demark enrolled 56,378 persons with arsenic levels in drinking water between 0.05 and 25.3 ␮g/L observed no dose–response relationship between arsenic exposure and liver cancer [36]. Likewise, a study in Idaho, USA observed no such dose–response relationship in the dose region from 0.1 to 950 ␮g/L [37]. To evaluate the association between arsenic levels in drinking water and occurrence of liver cancer further, we conducted a study including both the BFD endemic area and other areas in Taiwan.

1133

2.2. Data analysis The data were analyzed by multivariate linear regression using village as the unit population. Data on men and women were analyzed separately, and the independent variables included the percentage of wells in each arsenic exposure category of each village. Because arsenic levels were divided into six categories, five variables were used in this model to estimate rate differences (RDs) between the reference exposure category (below 0.05 mg/L) and each of the five higher exposure categories (0.05–0.08 mg/L, 0.09–0.16 mg/L, 0.17–0.32 mg/L, 0.33–0.64 mg/L, and above 0.64 mg/L). This approach derives estimates of unit RDs (incremental mortality rates for a one-unit increase in independent variables) and can generate unbiased estimates from ecological data even when the dose–response relationship is not linear [42]. To adjust for effects of age, we also included percentages of residents in different age groups to the regression models as independent variables [43,44]. Likewise, three variables were used for four age groups. The original regression model can be expressed as the following: R = ˛ + ˇ1 X1 + · · · + ˇ5 X5 + 1 A1 + · · · + 3 A3

(Model 1)

2. Experimental

where for each village, R is the mortality rate of liver cancer, Xj is proportion (expressed as percentage) of wells with arsenic levels in category j, and Ap is proportion of residents in age group p. R was calculated by dividing the number of liver cancer deaths by the estimated total number of person-years observed in a given village. In this case, ˛ (intercept) is the estimated background risk, and ˇj indicates the RD associated with each 1% increase in wells in category j. Furthermore, the data were weighted with the total population in each village as the weight, and weighted regression models were applied to account for the different weights of data contributed by different villages. On the basis of results obtained through Model 1, we conducted post hoc analyses to confirm the findings. In addition to regression analyses, stratified analyses were applied. All statistical analyses were conducted using the SAS software, and all statistical tests were performed at the two-tailed significant level of 0.05.

2.1. Collection of data

3. Results

The current study includes the four hyper-endemic townships of BFD and six nearby townships. Original measurement reports from a census survey of wells conducted by the Taiwan Provincial Institute of Environmental Sanitation [32] were available for 138 villages in the study townships. Most of the measurements were made between 1974 and 1976, and a follow-up survey applying the same method to test a part of the wells found that the results were highly correlated [38,39]. According to the standard solutions used in the survey [32] and results from previous studies [40,41], we grouped arsenic levels into six categories: below 0.05 mg/L, 0.05–0.08 mg/L, 0.09–0.16 mg/L, 0.17–0.32 mg/L, 0.33–0.64 mg/L, and above 0.64 mg/L. As there were a total of 6103 wells in these 138 villages, each village had 44.2 wells on average. In Taiwan, there is one household registry office in each township, and timely reporting of deaths to those offices is mandated by law. We reviewed death certificates filed to the ten household registry offices in the study townships between January 1, 1971 and December 31, 1990 to identify cases of liver cancer. In addition, we obtained data on residents in each village from the household registry offices and divided the residents into four age groups: 0–29, 30–49, 50–69, and >70 years old. The study villages had a total population of around 235,000. As there were a total of 6103 wells in the study villages, the estimated average number of people using a particular well was 38.5.

3.1. Associations between arsenic level in drinking water and mortality of liver cancer Over the 20-year study period, 802 male and 301 female liver cancer deaths were identified in the study villages. After adjusting for age, arsenic levels above 0.64 mg/L in drinking water were associated with a statistically significant increase in the mortality of liver cancer in both genders. As shown in Model 1 and Table 1, a 1% increase in the proportion of wells in this category (arsenic levels above 0.64 mg/L) was associated with an increase of 0.21 per 100,000 persons per year in the mortality of liver cancer in men, and 0.09 per 100,000 persons per year in women. No significant difference in the mortality of liver cancer was found to be associated with any of the other four arsenic levels. 3.2. Post hoc data analysis Post hoc stratified analyses were conducted according to two features of the dose–response relationship observed using Model 1: (1) no significant effect below 0.64 mg/L and (2) a significant positive effect above 0.64 mg/L. Accordingly, two post hoc multiple regression models were applied to validate the observations. The first one combined all exposure between 0.05 and 0.64 mg/L as one predictor variable—X14 , as the following: R = ˛1 + ˇ14 X14 + ˇ5 X5 + 1 A1 + · · · + 3 A3

(Model 2)

1134

H.-J. Lin et al. / Journal of Hazardous Materials 262 (2013) 1132–1138

Table 1 Estimates of rate differences (per 100,000 persons per year) from different regression models. Arsenic level (mg/L)

Model 1 0.05–0.08 0.09–0.16 0.17–0.32 0.33–0.64 >0.64 Model 2 0.05–0.64 >0.64 Model 3 >0.64 a * **

Men

Women

Rate difference

[95% CIa ]

Rate difference

[95% CI]

−0.06 −0.03 0.06 −0.11 0.21

[−0.29, 0.18] [−0.23, 0.17] [−0.18, 0.30] [−0.39, 0.18] [0.04, 0.37]*

−0.03 0.03 0.02 0.01 0.09

[−0.14, 0.08] [−0.07, 0.12] [−0.09, 0.13] [−0.13, 0.15] [0.01, 0.16]*

−0.02 0.21

[−0.13, 0.09] [0.05, 0.38]*

0.01 0.09

[−0.04, 0.06] [0.01, 0.17]*

0.22

[0.06, 0.38]**

0.09

[0.01, 0.16]*

Confidence interval. p < 0.05. p < 0.01.

and the results showed that this new variable (X14 ) was not a significant predictor of the mortality of liver cancer. This model is to examine whether there is a significant effect of levels between 0.05 and 0.64 mg/L. Under the assumption of no significant effect of arsenic in drinking water below 0.64 mg/L, the second model had only one indicator for arsenic exposure (for levels above 0.64 mg/L), as the following: R = ˛2 + ˇ5 X5 + 1 A1 + . . . + 3 A3

(Model 3)

Both Models 2 and 3 generated point estimates for the increases in liver cancer mortality associated with arsenic levels above 0.64 mg/L similar to the full model (Table 1). In post hoc stratified analyses, we validated the observed dose–response relationships further by dividing the 138 villages into the following three groups: the Group A consisted of 29 villages with all the 291 wells containing less than 0.05 mg/L of arsenic, Group B consisted of 74 villages with 30.6% of the 4286 wells containing more than 0.05 mg/L of arsenic but none with arsenic levels above 0.64 mg/L, and Group C consisted of 35 villages with 12.4% of the 1526 wells containing more than 0.64 mg/L of arsenic. If conclusions drawn on the dose–response relationship from regression analyses hold, Group C should have the highest mortality among the three, and Groups A and B should have similar mortality rates. The first projection that Group C has the highest mortality rate was generally true in all but two groups—men below 29 years of age and women between 50 and 69 years old, and the difference between Group C and the group with the highest mortality were not statistically significant in either case (Table 2). The second projection was supported by the facts that Group A and Group B had similar liver cancer mortality in four of the eight gender-age groups and that all the differences between Groups A and B were not statistically significant except for women between 50 and 69 years old. In other words, results of stratified analyses confirmed the dose–response relationships projected by the regression analyses. 4. Discussion 4.1. Literature on the association between arsenic ingestion and liver cancer The occurrence of liver cancer in patients of arsenic intoxication has been documented since about half a century ago [35]. Many of the earliest case reports were on patients taking medicines containing arsenic such as Fowler’s solution [21,45–49], and

cases attributable to environmental and occupational exposures were also reported [9,20,45,50,51]. The relatively high frequency of angiosarcoma was the focus of most case reports [9,21,45–47,49,51,52]. Epidemiologic studies on such populations, however, are very limited. A follow-up on 478 patients taking Fowler’s solution found only one case of liver cancer and obtained a standardized mortality ratio (SMR) of 1.23 (95% confidence interval [CI]: 0.04–7), while a statistically significant increase in bladder cancer was observed (SMR = 3.07) [53]. Whereas the early literature has provided sufficient evidence to argue an association between arsenic ingestion and liver cancer, for hepatic angiosarcomas at least, data on the dose–response relationship were quite limited. Using “arsenic” combined with “liver cancer” and “hepatoma” to search literatures in the PubMed, we found 19 English reports of epidemiologic studies on the association between arsenic ingestion and liver cancer [22,23,34,36,37,54–67]. A study found 5 cases amongst 258 wine growers with exposures through both ingestion and inhalation, but none in 159 non-exposed persons [22]. A study on 3141 deaths in Japan compared the mortalities before and after an episode of arsenic contamination of milk powder and found an increase in liver cancers [57]. The other studies were on the atherogenic and carcinogenic effects [23,34,36,37,54–56,58–67] of arsenic in drinking water. 4.1.1. Studies outside Taiwan A study in Argentina divided the study areas into three groups and observed a non-significant dose–response relationship in both genders [64] (Table 3). However, the study did not report actual exposure levels, but obtained a “crude average” of 0.178 mg/L for the “high exposure” group using measurements ≥0.04 mg/L only (instead of using all measurements). This was an unusual and probably unreliable way to estimate the exposure level, especially when only about 20% of the population in this group might have been exposed to levels ≥0.04 mg/L. In Chile, a study on a population with an average exposure up to 0.568 mg/L observed no increase in liver cancer (SMR = 1.1, 95% CI: 0.8–1.5) [60], but another study on a population with an average exposure up to 0.870 mg/L observed an increase in liver cancer mortality between ages 0 and 19 [54] (Table 3). A study in Demark enrolled 56,378 persons with arsenic levels in drinking water between 0.05 and 25.3 ␮g/L and observed no associations between arsenic exposure and liver cancer (RR = 0.97, 95% CI: 0.72–1.29 for 1 ␮g/L increment in arsenic level) [36]. A study in Idaho, USA also observed no associations between arsenic exposure and liver cancer (RD = 0.02 per 100,000 for 1 ␮g/L increment in arsenic level between 0.1 and 950 ␮g/L) [37]. A study in Japan identified two cases among 113 persons with arsenic levels in well water above 1 mg/L (SMR = 7.17, 95% CI: 1.28–26.05), but none among 330 persons with lower exposure levels [55]. Overall, the observations in these studies support both our findings of an increased risk associated with exposures above 0.64 mg/L and no increased risk below 0.64 mg/L. 4.1.2. Studies in Taiwan The other 11 studies were conducted in Taiwan. Two studies found a significant increase in mortality of liver cancer in both genders in the BFD endemic area but provided no data on the dose–response [23,58]. Two other studies also provided evidence supporting an association between arsenic ingestion and liver cancer but did not provide data on the dose–response [65,66]. Using the duration of consuming artesian well water (containing 0.35–1.14 mg/L of arsenic) as the exposure indicator, a study observed a dose–response trend (p < 0.01) among 65 cases and 368 controls [62], which supported our finding of an increased risk associated with high-level exposures.

H.-J. Lin et al. / Journal of Hazardous Materials 262 (2013) 1132–1138

1135

Table 2 Comparison of liver cancer mortality (per 100,000 persons per year) among Group A (all [As] < 0.05 mg/L), Group B (some [As] > 0.05 mg/L, but none > 0.64 mg/L), and Group C (some [As] > 0.64 mg/L) through post hoc stratified analyses. Predictor

0–29 years old Group A Group B Group C 30–49 years old Group A Group B Group C 50–69 years old Group A Group B Group C >69 years old Group A Group B Group C * **

Men

Women

Cases

Mortality

8 17 9

2.46 1.85 2.19

33 75 52

Comparison

p

Cases

Mortality

Comparison

p

A>B C>B A>C

0.51 0.69 0.81

1 7 4

0.34 0.83 1.08

B>A C>B C>A

0.38 0.69 0.28

34.54 27.99 44.87

A>B C > B** C>A

0.31 <0.01 0.24

5 15 10

5.39 5.82 8.64

B>A C>B C>A

0.88 0.33 0.39

47 137 81

98.99 105.27 141.31

B>A C > B* C > A*

0.71 0.03 0.05

12 64 23

23.33 45.08 38.07

B > A* B>C C>A

0.03 0.48 0.16

70 173 97

711.38 698.14 971.94

A>B C > B** C > A*

0.89 <0.01 0.03

32 79 45

245.40 218.47 338.35

A>B C > B* C>A

0.57 0.02 0.15

p < 0.05. p < 0.01.

Three of the remaining seven studies [34,61,63] were reanalyses of data from a previous study [26] (Tables 3 and 4). A preliminary report of that study gathered 42 villages into three groups (<0.30 ppm [mg/L], 0.30–0.59 ppm, and ≥0.60 ppm) according to the median arsenic levels in well water and reported a significant dose–response relationship in both genders [38,39] (Tables 3 and 4). When the full report excluded residents under 20 years of age, however, the dose–response relationship in women became not statistically significant [26] (Tables 3 and 4). The first re-analysis [61] divided the study population further into four exposure and four age groups (Tables 3 and 4) and identified all residents with liver cancers listed on death certificates as cases (the previous analyses only identified those with liver cancer as the major cause of death as cases) [26,38]. As a result, a significant dose–response relationship was observed in most age groups in both genders (Tables 3 and 4). Nonetheless, when the second reanalysis [34], broke down exposure levels further into eight groups, no dose–response relationship was observed (Tables 3 and 4). The third re-analysis [63] included 18 more villages with exposure data from another survey [67] and divided exposure levels even further into 11 groups. The researchers did not present actual risk estimates, but reported variations in the dose–response relationship and concluded no evidence of excess risk below 0.1 mg/L, which is compatible to our finding. Because the population was large and case number was large even when being broken down to groups, study power did not seem to be a problem in these analyses. Discrepancies among findings were probably attributable to the fact that the median arsenic level in each village was not a good exposure indicator. If our finding that the effect was much more prominent above 0.64 mg/L holds, the proportion of wells with arsenic levels above 0.64 mg/L will be a better exposure indicator, and it is reasonable that using the median exposure as the indicator, which is less reliable, may observe different dose–response relationships when different divisions of exposure levels were applied. An ecologic analysis of 314 townships adopted the same source of exposure data [32] as in our study but used the mean arsenic level in each township as the indicator [59]. A significant positive association was observed in both genders, but when the analyses were limited to 170 southwestern townships, slightly higher incremental mortalities associated with each 0.1 mg/L increase in the arsenic level were obtained (Table 3). Because the exposure indicators were different, it is hard to compare the risk estimates with those from our study.

The remaining studies in Taiwan did not have data on arsenic level [65–67]. Overall, findings from previous studies in Taiwan are not incompatible with our observations. 4.2. Other epidemiologic evidence The literature search also identified two studies of BFD patients with data on liver cancer, and both found an excess risk of liver cancer [30,56]. Because no BFD patients were identified among those who did not used water from artesian wells [68], some researchers treated BFD patients as a population whose exposure levels were above 0.35 mg/L—the lowest arsenic level detected in artesian well water in the endemic area of BFD [69]—in the risk assessment [62]. While this measure might not be entirely accurate, because a survey showed about one fourth of the artesian wells in that area had arsenic levels below 0.3 mg/L [67], it is true that BFD patients are a group of people exposed to high levels of arsenic in drinking water. Artesian wells in the endemic area of BFD should have an average arsenic level around 0.5–0.6 mg/L and a range around 0.010–1.752 mg/L [26,39,40,68,70,71]. Therefore, these two studies support our finding of an association between exposures to high levels of arsenic and liver cancer. 4.3. The dose–response Besides a lack of effects at low-dose levels, there are other possible reasons for not observing positive associations between exposures to low levels of arsenic in drinking water and liver cancer in our and other studies. First, the excess risks associated with arsenic levels below 0.64 mg/L might be too small for the study to detect. In addition, un-controlled confounders might affect the observed dose–response curve. In particular, Taiwan is an endemic area of virus hepatitis B, which is also a risk factor of liver cancer. None of the previous studies controlled effects of hepatitis B, but its effects as a confounder in the our study is expected to be small, if any, because the prevalence of hepatitis B in the BFD endemic area is about the same as that in the rest of Taiwan [70] and not related to the arsenic level in drinking water [72]. On the other hand, some studies indicated that arsenic might be an essential nutrient with a safety level of ingestion, and therefore, a non-linear dose–response relationship between arsenic exposure and liver cancer is biologically plausible [28,73]. In addition, a study in Belgium found that a low to moderate level arsenic in drinking

1136

H.-J. Lin et al. / Journal of Hazardous Materials 262 (2013) 1132–1138

Table 3 Epidemiologic studies on arsenic levels in drinking water and liver cancers using arsenic level as the exposure indicator. Study (reference)

Wu et al. [26]

Chen and Wang [59]

Study design (study period)

Ecologic analysis (1973–1986)

Ecologic analysis (1972–1983)

Study area

42 villages, Taiwan

314 townships, Taiwan

Cases

Arsenic level (mg/L)

Resultsa

Age

Sex

Number

>20

Men

54

<0.3

47.78/100,000 pyb

Women

42 27 25 16 10

0.3–0.59 ≥0.6 <0.3 0.3–0.59 ≥0.6

67.62/100,000 py 86.73/100,000 py(*) 21.40/100,000 py 24.18/100,000 py 31.75/100,000 py(−)

Men

NRc

1 mg/L increment

RDd = 6.8/100,000 py

Women Men

NR NR

1 mg/L increment 1 mg/L increment

RD = 2.0/100,000 py RD = 7.0/100,000 py

Women

NR

1 mg/L increment

RD = 2.3/100,000 py

All

170 townships, southwest Taiwan Smith et al. [60]

Ecologic analysis (1989–1993)

8 cities, Chile

≥30

Both

48

0.568

SMRe = 1.1

Liaw et al. [54]

Ecologic analysis (1950–2000)

2 regions, Chile

0–19

Men

7

0.870

RRf = 8.9

Women Both

5 12

0.870 0.870

RR = 14.1 RR = 10.6

1 ␮g/L increment (0.00005–0.0253; median = 0.0007 mg/L) 1 ␮g/L increment (0.0001–0.950; median = 0.002 mg/L)

RR = 0.97

Baastrup et al. [36]

Ecologic analysis (1993–2003)

56,378 persons

50–64

Both

35

Han et al. [37]

Ecologic analysis (1991–2005)

All 44 counties in Idaho, USA

All

Both

477

Tsuda et al. [55]

Cohort study (1959–1992)

454 patients, Japan

All

Both

0

<0.05

SMR = 0

Both Both

0 2

0.05–0.99 ≥1

SMR = 0 SMR = 7.17*

Men

186

Low

SMR = 1.54*

Women

142 98 173 125 90

Medium High (0.178 mg/L)g Low Medium High (0.178 mg/L)g

SMR = 1.80* SMR = 1.84* SMR = 1.69* SMR = 1.87* SMR = 1.92*

Hopenhayn-Rich et al. [64]

a b c d e f g (*) (−) *

Ecologic analysis (1986–1991)

24 counties, Argentina

≥20

RD = 0.02/100,000

All studies were on mortality. Person-years. Not reported. Incremental rate difference. Standardized mortality ratio. Relative risk. A “crude average” estimated using all measurements ≥0.04 mg/L cancers of biliary passage and liver. p < 0.05 for trend test. p > 0.05 for trend test. p < 0.05 for comparison with the reference population.

water does not affect causes of mortality, suggesting non-linearity of the dose–response relationship for arsenic and cancer [74]. Whereas angiosarcoma seems to be a type of liver specific to arsenic exposure [45–49,75,76], the occurrence of this cancer is rare and thus has little effect on the dose–response relationship. On the other hand, a previous study found the carcinogenic effect of ingested arsenic on liver was not specific to hepatocellular carcinoma [66,67], and therefore the increase in liver cancer observed in the present study should be due to an over-all increase in all types of liver cancer. 4.4. Strengths and limitation of the current study Timely and accurate reporting of deaths is mandated by law in Taiwan, and therefore the mortality registry we used in this study is reliable and complete. Some studies have been conducted in Taiwan to assess the dose–response relationship between arsenic level in drinking water and mortality of liver cancer, but

the results were not consistent even when the same exposure data were categorized in different ways. In fact, even in the same study, the dose–response relationship might be different between the two genders [26,34]. Since the gender difference was only observed in one study, we believe the gender difference observed by Wu et al. in 1989 [26,34] was most likely due to the statistical approach. In comparison with previous studies in Taiwan, the current study had a much larger study population (40 of the 42 villages most frequently studied villages were included in our study) and longer study period (Table 3), and therefore the results should be more reliable. Furthermore, the current study had more exposure categories than most of the previous studies and thus should be able to describe the dose–response relationship more precisely. Because the present study used ecologic data, its results might be affected by “ecological fallacy” as well as other common limitations of ecological studies, such as misclassification of exposure, the lack of ability to address exposure duration, and effects of

H.-J. Lin et al. / Journal of Hazardous Materials 262 (2013) 1132–1138

1137

Table 4 Mortality (per 100,000 person-years) estimatesa from repetitive dose–response analyses of data from an ecologic study of mortality among residents of 42 villages in the blackfoot disease endemic area from 1973 to 1986. Sex

Study (reference)

Men

Chen et al. [61]

Chen et al. [38] Wu et al. [26] Women

Age

Arsenic level (mg/L)

20–29 30–49 50–69 70+

<0.10 0.9 37.9 93.7 89.4

20–29 30–49 50–69 70+ Chen et al. [38] Wu et al. [26] a (*) (−)

All >20

0.30–0.59 3.3 41.4 154.3 169.8 0.3–0.59 42.7 67.6

≥0.60 2.1 80.1 234.7 103.0 ≥0.60 68.8(*) 86.7(*)

0.10–0.29 0.0 12.5 69.7 66.8

0.30–0.59 1.3 27.6 55.3 72.1 0.3–0.59 18.8 24.2

≥0.60 0.0 29.2 88.7 203.3 ≥0.6 31.8(*) 31.8(−)

<0.30 32.6 47.8

All >20

Chen et al. [61]

0.10–0.29 0.0 40.8 141.3 184.3

<0.10 0.0 9.3 41.3 77.8 <0.3 14.2 21.4

Round-off to one digit below the decimal point for comparison. p < 0.05 for trend test. p > 0.05 for trend test.

population mobility. Further studies with exposure data on individuals and the ability of adjusting for effects of other confounders such as hepatitis B and C infection, liver cirrhosis, social-economic status, and alcohol consumption are necessary to confirm findings in this study. 5. Conclusions In conclusion, we found exposures to high arsenic levels in drinking water are associated with the occurrence of liver cancer, but such an effect was not prominent at low exposure levels. These findings suggest that both the Joint FAO/WHO Expert Committee on Food Additives guidelines for lung cancer (3.0 ␮g/kg bw per day; 2–7 ␮g/kg bw per day) and the new U.S. EPA arsenic standards for drinking water (0.1 mg/L) may provide sufficient protection against the carcinogenic effect of arsenic on liver. Acknowledgements This work was supported in part by Grants NSC89-2320-B006-015 and NSC100-2314-B-006-061 from the National Science Council of Taiwan, R.O.C. and a grant from the Cancer Research Center of the Chi-Mei Medical Center. The authors would also like to thank the household registry offices for providing data. References [1] P.L. Smedley, D.G. Kinniburgh, A review of the source, behaviour and distribution of arsenic in natural waters, Appl. Geochem. 17 (2002) 517–568. [2] H.-R Guo, C. -J. Chen, H.L. Greene, Arsenic in drinking water and cancers: a descriptive review of Taiwan studies, Environ. Geochem. Health s16 (1994) 129–138. [3] G. Sun, Arsenic contamination and arsenicosis in China, Toxicol. Appl. Pharmacol. 198 (2004) 268–271. [4] T. Tsuda, T. Nagira, M. Yamamoto, N. Kurumatani, N. Hotta, M. Harada, H. Aoyama, Malignant neoplasms among residents who drank well water contaminated by arsenic from a king’s yellow factory, J. UOEH 11 (Suppl.) (1989) 289–301. [5] T. Clarke, Delta blues, Nature 422 (2003) 254–256. [6] H.M. Foy, S. Tarmapai, P. Eamchan, O. Metdilogkul, Chronic arsenic poisoning from well water in a mining area in Thailand, Asia-Pac. J. Publ. Health. 6 (1992) 150–152. [7] M.E. Cebrian, A. Albores, M. Aguilar, E. Blakely, Chronic arsenic poisoning in the north of Mexico, Hum. Toxicol. 2 (1983) 121–133. [8] R.E. Biagini, G.C. Quiroga, V. Elias, Chronic hydroarsenism in ururau, Archivos Agentinos de Dematologia 24 (1974) 8–11.

[9] R. Zaldivar, Arsenic contamination of drinking water and foodstuffs causing endemic chronic poisoning, Beitr. Pathol. 151 (1974) 384–400. [10] M. Borzsonyi, A. Bereczky, P. Rudnai, M. Csanady, A. Horvath, Epidemiological studies on human subjects exposed to arsenic in drinking water in southeast Hungary, Arch. Toxicol. 66 (1992) 77–78. [11] H.V. Aposhian, E.S. Gurzau, X.C. Le, A. Gurzau, S.M. Healy, X. Lu, M. Ma, L. Yip, R.A. Zakharyan, R.M. Maiorino, R.C. Dart, M.G. Tircus, D. Gonzalez-Ramirez, D.L. Morgan, D. Avram, M.M. Aposhian, Occurrence of monomethylarsonous acid in urine of humans exposed to inorganic arsenic, Chem. Res. Toxicol. 13 (2000) 693–697. [12] G. Tamasi, R. Cini, Heavy metals in drinking waters from Mount Amiata (Tuscany, Italy). Possible risks from arsenic for public health in the Province of Siena, Sci. Total Environ. 327 (2004) 41–51. [13] G.J.R.N. Deane, J. Thom, G. Gentry, Evaluation of health implications of elevated arsenic in well water, Water Res. 6 (1972) 1133–1136. [14] B.K. Mandal, K.T. Suzuki, Arsenic round the world: a review, Talanta 58 (2002) 201–235. [15] J.C. Ng, J. Wang, A. Shraim, A global health problem caused by arsenic from natural sources, Chemosphere 52 (2003) 1353–1359. [16] M.M. Rahman, J.C. Ng, R. Naidu, Chronic exposure of arsenic via drinking water and its adverse health impacts on humans, Environ. Geochem. Health 31 (Suppl. 1) (2009) 189–200. [17] M. Sharpe, Deadly waters run deep: the global arsenic crisis, J. Environ. Monit. 5 (2003) 81N–85N. [18] M.F. Hossain, Arsenic contamination in Bangladesh—an overview, Agric. Ecosyst. Environ. 113 (2006) 1–16. [19] S. Kapaj, H. Peterson, K. Liber, P. Bhattacharya, Human health effects from chronic arsenic poisoning—a review, J. Environ. Sci. Health A: Toxicol. Hazard. Subst. Environ. Eng. 41 (2006) 2399–2428. [20] J.C. Pimentel, A.P. Menezes, Liver disease in vineyard sprayers, Gastroenterology 72 (1977) 275–283. [21] J.L. Cowlishaw, E.J. Pollard, A.E. Cowen, L.W. Powell, Liver disease associated with chronic arsenic ingestion, Aust. N. Z. J. Med. 9 (1979) 310–313. [22] H. Luchtrath, The consequences of chronic arsenic poisoning among Moselle wine growers. Pathoanatomical investigations of post-mortem examinations performed between 1960 and 1977, J. Cancer Res. Clin. Oncol. 105 (1983) 173–182. [23] S.-M. Tsai, T.-N. Wang, Y.-C. Ko, Mortality for certain diseases in areas with high levels of arsenic in drinking water, Arch. Environ. Health 54 (1999) 186–193. [24] J. Hutchinson, Salvarsan (“606”) and arsenic cancer, Br. Med. J. 1 (1911) 976–977. [25] EPA, National primary drinking water regulations; arsenic and clarifications to compliance and new source contaminants monitoring; final rule., Fed Reg, Environ. Protect. Agen. 66 (2001) 6976–7066. [26] M.-M. Wu, T.-L. Kuo, Y.-H. Hwang, C.-J. Chen, Dose–response relation between arsenic concentration in well water and mortality from cancers and vascular diseases, Am. J. Epidemiol. 130 (1989) 1123–1132. [27] EPA, Health Assessment Document for Inorganic Arsenic: Final Report, U. S. Environmental Protection Agency, Washington, DC, 1984. [28] EPA, Risk Assessment Forum. Special Report on Ingested Inorganic Arsenic: Skin Cancer; Nutritional Essentiality. EPA 625/3-87/013, U. S. Environmental Protection Agency, Washington, DC, 1988. [29] WHO, WHO Guidelines for Drinking Water Quality, Volume 1. Geneva, World Health Organization, Switzerland, 1993.

1138

H.-J. Lin et al. / Journal of Hazardous Materials 262 (2013) 1132–1138

[30] W.-P. Tseng, Effects and dose–response relationships of skin cancer and blackfoot disease with arsenic, Environ. Health Perspect. 19 (1977) 109–119. [31] S. Yeh, Studies on endemic chronic arsenism in southwest coast of Taiwan, Rep. Inst. Pathol. Natl. Univ. Taiwan 14 (1963) 1–23. [32] M.-C. Lo, Y.-C. Hsen, B.-K. Lin, Second Report on the Investigation of Arsenic Content in Underground Water in Taiwan. Taichung, Taiwan Provincial Institute of Environmental Sanitation, Taiwan, 1977. [33] W.-P. Tseng, H.-M. Chu, S.-W. How, J.-M. Fong, C.-S. Lin, S. Yeh, Prevalence of skin cancer in an endemic area of chronic arsenicism in Taiwan, J. Natl. Cancer Inst. 40 (1968) 453–463. [34] K.H. Morales, L. Ryan, T.-L. Kuo, M.-M. Wu, C.-J. Chen, Risk of internal cancers from arsenic in drinking water, Environ. Health Perspect. 108 (2000) 655–661. [35] F. Roth, Chronic arsenic poisoning of Moselle vineyard-workers, with special reference to arsenic cancer, Z. Krebsforsch. 61 (1956) 287–319. [36] R. Baastrup, M. Sorensen, T. Balstrom, K. Frederiksen, C.L. Larsen, A. Tjonneland, K. Overvad, O. Raaschou-Nielsen, Arsenic in drinking-water and risk for cancer in Denmark, Environ. Health Perspect. 116 (2008) 231–237. [37] Y.Y. Han, J.L. Weissfeld, D.L. Davis, E.O. Talbott, Arsenic levels in ground water and cancer incidence in Idaho: an ecologic study, Int. Arch. Occup. Environ. Health 82 (2009) 843–849. [38] C.-J. Chen, T.-L. Kuo, M.-M. Wu, Arsenic and cancers, Lancet 1 (1988) 414–415. [39] T.-L. Kuo, Arsenic content of artesian well water in endemic area of chronic arsenic poisoning, Rep. Inst. Pathol. Natl. Univ. Taiwan 20 (1968) 7–13. [40] T.-L. Kuo, Investigation on the arsenic levels in well water in the blackfoot disease endemic area, Chin. J. Public Health 15 (Suppl.) (1996) 116–125. [41] H.-R. Guo, H.-S. Chiang, H. Hu, S.R. Lipsitz, R.R. Monson, Arsenic in drinking water and incidence of urinary cancers, Epidemiology 8 (1997) 545–550. [42] H.-R. Guo, S.R. Lipsitz, H. Hu, R.R. Monson, Using ecological data to estimate a regression model for individual data: the association between arsenic in drinking water and incidence of skin cancer, Environ. Res. 79 (1998) 82–93. [43] P.R. Rosenbaum, D.B. Rubin, Difficulties with regression analyses of ageadjusted rates, Biometrics 40 (1984) 437–443. [44] H.-R. Guo, Comparison of three methods for age adjustment in ecological studies: using a study on arsenic in drinking water and bladder cancer as an example, BMC Public Health 11 (2011) 820. [45] H. Falk, J.T. Herbert, L. Edmonds, C.W. Heath Jr., L.B. Thomas, H. Popper, Review of four cases of childhood hepatic angiosarcoma—elevated environmental arsenic exposure in one case, Cancer 47 (1981) 382–391. [46] J.J. Lander, R.J. Stanley, H.W. Sumner, D.C. Boswell, R.D. Aach, Angiosarcoma of the liver associated with Fowler’s solution (potassium arsenite), Gastroenterology 68 (1975) 1582–1586. [47] W. Regelson, U. Kim, J. Ospina, J.F. Holland, Hemangioendothelial sarcoma of liver from chronic arsenic intoxication by Fowler’s solution, Cancer 21 (1968) 514–522. [48] J.W. Roat, A. Wald, H. Mendelow, K.I. Pataki, Hepatic angiosarcoma associated with short-term arsenic ingestion, Am. J. Med. 73 (1982) 933–936. [49] M.L. Kasper, L. Schoenfield, R.L. Strom, A. Theologides, Hepatic angiosarcoma and bronchioloalveolar carcinoma induced by Fowler’s solution, J. Am. Med. Assoc. 252 (1984) 3407–3408. [50] Y. Liu, W. Gao, J.M. Siegfried, J.L. Weissfeld, J.D. Luketich, P. Keohavong, Promoter methylation of RASSF1A and DAPK and mutations of K-ras, p53, and EGFR in lung tumors from smokers and never-smokers, BMC Cancer 7 (2007) 74. [51] M.-H Tsai, R. -N. Chien, S. -Y. Hsieh, C. -F. Hung, T. -C. Chen, I. -S. Sheen, Primary hepatic angiosarcoma: report of a case involving environmental arsenic exposure, Chang Gung Med. J. 21 (1998) 469–474. [52] N.-C. Huang, S.-R. Wann, H.-T. Chang, S.-L. Lin, J.-S. Wang, H.-R. Guo, Arsenic, vinyl chloride, viral hepatitis, and hepatic angiosarcoma: a hospital-based study and review of literature in Taiwan, BMC Gastroenterol. 11 (2011) 142. [53] J. Cuzick, P. Sasieni, S. Evans, Ingested arsenic, keratoses, and bladder cancer, Am. J. Epidemiol. 136 (1992) 417–421. [54] J. Liaw, G. Marshall, Y. Yuan, C. Ferreccio, C. Steinmaus, A.H. Smith, Increased childhood liver cancer mortality and arsenic in drinking water in northern Chile, Cancer Epidemiol. Biomar. Prev. 17 (2008) 1982–1987.

[55] T. Tsuda, A. Babazono, E. Yamamoto, N. Kurumatani, Y. Mino, T. Ogawa, Y. Kishi, H. Aoyama, Ingested arsenic and internal cancer: a historical cohort study followed for 33 years, Am. J. Epidemiol. 141 (1995) 198–209. [56] C.-J. Chen, M.-M. Wu, S.-S. Lee, J.-D. Wang, S.-H. Cheng, H.-Y. Wu, Atherogenicity and carcinogenicity of high-arsenic artesian well water. Multiple risk factors and related malignant neoplasms of blackfoot disease, Arteriosclerosis 8 (1988) 452–460. [57] T. Yorifuji, T. Tsuda, H. Doi, P. Grandjean, Cancer excess after arsenic exposure from contaminated milk powder, Environ. Health Prev. Med. 16 (2011) 164–170. [58] C.-J. Chen, Y.-C. Chuang, T.-M. Lin, H.-Y. Wu, Malignant neoplasms among residents of a blackfoot disease-endemic area in Taiwan: high-arsenic artesian well water and cancers, Cancer Res. 45 (1985) 5895–5899. [59] C.-J. Chen, C.-J. Wang, Ecological correlation between arsenic level in well water and age-adjusted mortality from malignant neoplasms, Cancer Res. 50 (1990) 5470–5474. [60] A.H. Smith, M. Goycolea, R. Haque, M.L. Biggs, Marked increase in bladder and lung cancer mortality in a region of Northern Chile due to arsenic in drinking water, Am. J. Epidemiol. 147 (1998) 660–669. [61] C.-J. Chen, C.-W. Chen, M.-M. Wu, T.-L. Kuo, Cancer potential in liver, lung, bladder and kidney due to ingested inorganic arsenic in drinking water, Br. J. Cancer 66 (1992) 888–892. [62] C.-J Chen, Y.-C. Chuang, S.-L. You, T.-M. Lin, H.-Y. Wu, A retrospective study on malignant neoplasms of bladder, lung and liver in blackfoot disease endemic area in Taiwan, Br. J. Cancer 53 (1986) 399–405. [63] K.G. Brown, C.-J. Chen, Significance of exposure assessment to analysis of cancer risk from inorganic arsenic in drinking water in Taiwan, Risk Anal. 15 (1995) 475–484. [64] C. Hopenhayn-Rich, M.L. Biggs, A.H. Smith, Lung and kidney cancer mortality associated with arsenic in drinking water in Cordoba, Argentina, Int. J. Epidemiol. 27 (1998) 561–569. [65] H.-F. Chiu, S.-C. Ho, L.-Y. Wang, T.-N. Wu, C.-Y. Yang, Does arsenic exposure increase the risk for liver cancer? J. Toxicol. Environ. Health A 67 (2004) 1491–1500. [66] H.-R. Guo, The lack of a specific association between arsenic in drinking water and hepatocellular carcinoma, J. Hepatol. 39 (2003) 383–388. [67] S.-N. Lu, N.-H. Chow, W.-C. Wu, T.-T. Chang, W.-S. Huang, S.-C. Chen, C.-H. Lin, B.I. Carr, Characteristics of hepatocellular carcinoma in a high arsenicism area in Taiwan: a case–control study, J. Occup. Environ. Med. 46 (2004) 437–441. [68] K.-P. Chen, H.-Y. Wu, Epidemiologic studies on blackfoot disease 2: a study of source of drinking water in relation to the disease, J. Formos. Med. Assoc. 61 (1962) 611–618. [69] K.-P. Chen, H.-Y. Wu, T.-C. Wu, Epidemiologic studies on blackfoot disease in Taiwan 3: physicochemical characteristics of drinking water in endemic blackfoot disease area, Momoris Coll. Med. Natl. Taiwan Univ. 8 (1962) 115–129. [70] S. Yeh, S.-W. How, A pathological study on the blackfoot disease in Taiwan, Rep. Inst. Pathol. Natl. Univ. Taiwan 14 (1963) 25–73. [71] R.Q. Blackwell, T.H. Yang, I. Ai, Preliminary report on arsenic levels in water and food from the endemic blackfoot area, J. Formos. Med. Assoc. 60 (1961) 1139–1140. [72] S.-F. Tsai, Ecological correlation study on major risk factors of various malignant neoplasms in Taiwan, 1972–1983. Master Thesis. Taipei, Taiwan: National Taiwan University (1987). [73] E.O. Uthus, Evidence for arsenic essentiality, Environ. Geochem. Health 14 (1992) 55–58. [74] J.P. Buchet, D. Lison, Mortality by cancer in groups of the Belgian population with a moderately increased intake of arsenic, Int. Arch. Occup. Environ. Health 71 (1998) 125–130. [75] S.-Y. Ho, C.-C. Tsai, Y.-C. Tsai, H.-R. Guo, Hepatic angiosarcoma presenting as hepatic rupture in a patient with long-term ingestion of arsenic, J. Formos. Med. Assoc. 103 (2004) 374–379. [76] J. Liu, B. Zheng, H.V. Aposhian, Y. Zhou, M.L. Chen, A. Zhang, M.P. Waalkes, Chronic arsenic poisoning from burning high-arsenic-containing coal in Guizhou, China, Environ. Health Perspect. 110 (2002) 119–122.