You are on page 1of 7

Lung Cancer and Arsenic Concentrations in Drinking

Water in Chile
Catterina Ferreccio,1,2 Claudia Gonza´lez,2 Vivian Milosavjlevic,2 Guillermo Marshall,1
Ana Maria Sancha,3 and Allan H. Smith4

Cities in northern Chile had arsenic concentrations of 860 drinking water, as follows: 1, 1.6 (95% CI ⫽ 0.5–5.3), 3.9
␮g/liter in drinking water in the period 1958 –1970. Concen- (95% CI ⫽ 1.2–12.3), 5.2 (95% CI ⫽ 2.3–11.7), and 8.9 (95%
trations have since been reduced to 40 ␮g/liter. We investi- CI ⫽ 4.0 –19.6), for arsenic concentrations ranging from less
gated the relation between lung cancer and arsenic in drinking than 10 ␮g/liter to a 65-year average concentration of 200 –
water in northern Chile in a case-control study involving 400 ␮g/liter. There was evidence of synergy between cigarette
patients diagnosed with lung cancer between 1994 and 1996 smoking and ingestion of arsenic in drinking water; the odds
and frequency-matched hospital controls. The study identified ratio for lung cancer was 32.0 (95% CI ⫽ 7.2–198.0) among
152 lung cancer cases and 419 controls. Participants were smokers exposed to more than 200 ␮g/liter of arsenic in drink-
interviewed regarding drinking water sources, cigarette smok- ing water (lifetime average) compared with nonsmokers ex-
ing, and other variables. Logistic regression analysis revealed a posed to less than 50 ␮g/liter. This study provides strong
clear trend in lung cancer odds ratios and 95% confidence evidence that ingestion of inorganic arsenic is associated with
intervals (CIs) with increasing concentration of arsenic in human lung cancer. (Epidemiology 2000;11:673– 679)

Keywords: arsenic, lung cancer, water pollutants, smoking, synergy, case-control study, environmental epidemiology.

Humans are exposed to organic and inorganic arsenic taminated with inorganic arsenic10 and in large popula-
through environmental and occupational sources. Lung tion studies in Cordoba, Argentina,11 and northern
cancer is known to be caused by occupational exposure Chile.12,13 The purpose of the present study was to in-
to arsenic via inhalation.1 The main occupational expo- vestigate inorganic arsenic and lung cancer in northern
sures occur in workers who are engaged in smelting and Chile in a case-control study, with individual assessment
refining copper, gold, and lead ores; in producing agri- of exposure based on arsenic concentrations in water
cultural pesticides; in using arsenic as pigments and dyes; sources piped to households. It is the first large, popula-
and in manufacturing glass, semiconductors, and various tion-based lung cancer case-control study concerning
pharmaceutical substances from which there may be arsenic in drinking water. A preliminary research report
high exposure to airborne arsenic.1 The most extensive of work in progress on this study was presented at a
human exposure to inorganic arsenic, however, results scientific meeting in Brazil14 and included some of the
from naturally occurring inorganic arsenic in drinking results we present more fully in this paper.
water, long known to be a cause of skin cancer. Surpris-
ing evidence, originating from studies in Taiwan, indi- Subjects and Methods
cated that the ingestion of inorganic arsenic also in- The study area included regions I, II, and III in northern
creases mortality from cancer originating in various Chile. Details on the study area and methods are de-
internal sites, including lung.2–9 Further evidence of a scribed more fully elsewhere.14 Briefly, the population in
link between the ingestion of inorganic arsenic and region II experienced high exposure to inorganic arsenic
increased lung cancer risks was found in a small cohort in past years from natural contamination of drinking
study in Japan involving residents using well water con- water originating in the Andes mountains, whereas wa-
ter sources in regions I and III contained relatively little
arsenic (Table 1).
From the 1Facultad de Matema´ticas and 3Facultad de Ciencias Fı´sicas y Mate-
ma´ticas, Pontificia Universidad Cato´lica de Chile, Santiago, Chile; 2Grupo Para
el Desarrollo de la Investigacio´n en Salud (GREDIS, Santiago, Chile); and IDENTIFICATION OF LUNG CANCER CASES
School of Public Health, University of California, Berkeley, CA.
Nurses were recruited for the study and trained in inter-
Address correspondence to: Catterina Ferreccio, 1711 Massachusetts Avenue viewing techniques in each major city of northern Chile:
NW Apartment 628, Washington DC 20036. Arica and Iquique in region I; Antofagasta in region II,
Submitted June 21, 1999; final version accepted April 18, 2000.
and Copiapo´ in region III. They identified lung cancer
and bladder cancer cases in the public hospitals, where
Copyright © 2000 by Lippincott Williams & Wilkins, Inc. the large majority (80 –90%) of cancer patients are ad-


For each index case with lung cancer. We used hospital controls because selection of a random We selected a second control group in the same man- sample representative of the general population would ner as the cancer control group but from among patients have been prohibitively expensive. Iquique. and three public hospitals in region selection with each hospital repeated on the list the III. Vol. about 70%.674 Ferreccio et al Epidemiology November 2000. There was to obtain the same frequency distribution in our were two public hospitals in region I.169 1 1 1 1 1 1 * Population numbers were obtained from the 1992 census. skin. were referred from number of times required to create the target frequency the smaller public hospitals to the main public hospitals distribution. which includes scattered diseases. or neurologic expensive in the study area. Eligible cases were all those admitted stead of matching on hospital. the index case to their hospital.508 10 10 10 10 10 10 Putre 2. we also selected controls for a Only 20% of controls had to be interviewed in their concurrent bladder cancer study in a manner identical to homes after discharge.319 15 15 15 15 15 15 Vallenar 29. interviewed in the few major hospitals in the study area. were interviewed while chosen hospital with a cancer not known or suspected to still in the hospital. as each site has been found to be study were complicated and unusual for several reasons. 11 No. Our goal more than 1 year before the current admission.236 1 1 1 1 1 1 Copiapo´ 64. we selected two viewers went daily to the admissions department and the controls. three public hos. in whom lung as eligible controls. The smaller public selected from among patients who were admitted to the hospitals were visited approximately once a month. dence that arsenic may increase the risks of some con- trols meant that most controls could be identified and ditions within each of these disease groups. We identified the number of patients cancer was pathologically confirmed and first diagnosis admitted to each hospital in 1991 and created a fre- was either during the current hospital admission or no quency distribution of admissions by hospital.440 250 250 636 110 110 39 Calama 74. We created an ordered listing for control pitals in region II. Carmen 3.825 60 60 60 60 60 60 Huara 1. and Copiapo´. The methods we used for selection of controls for this bladder. At the same time. The majority of patients.114 1 1 1 1 1 1 Huasco 4.568 15 15 15 15 15 15 Tierra Amarilla 7. control group.828 1 1 1 1 1 1 A.240 1 1 1 1 1 1 Iquique 97. The hospital to be the source for each control pathology laboratories of these main hospitals to identify was identified as above. of Antofagasta. Selection of random admitted to the next hospital on the list with a diagnosis samples from the general population is difficult in itself other than cancer and also excluding from consideration in Chile.757 600 600 600 600 600 600 Antofagst 149. Arica. however.869 60 60 60 60 60 60 III Chan˜aral 8. The remaining 30% were visited and be related to arsenic. or prostate. and locating selected participants would be those admitted with cardiovascular.857 40 40 40 40 40 40 II Tocopilla 15.186 90 860 110 110 70 40 Mejillons 4. mitted.340 30 30 30 30 30 30 Pica 1.347 1 1 1 1 1 1 Freirina 3. Using hospital con. and who were within 4 years of age of the index case.761 10 10 10 10 10 10 Pozo Almonte 3.030 150 150 287 110 110 40 San Pedro 1. The first control was randomly patients diagnosed with lung cancer.964 250 250 636 110 110 40 Marı´a Elena 8. skin.631 15 15 15 15 15 15 Diego de Almagro 17.15 These diseases were excluded because of evi- small populations in remote areas. within a month of the admission of interviewed in their homes after discharge. related to arsenic in some studies. In this paper we report the results for the lung To avoid the problem of matching on exposure in- cancer cases only. we defined all patients to public hospitals with lung cancer in the study region admitted to any public hospital in the whole study area between November 1994 and July 1996.086 90 860 110 110 70 37 Taltal 6. We excluded as controls SELECTION OF CONTROLS patients admitted with cancer of the liver. We pooled . that described for the lung cancer cases. 6 TABLE 1. Average Arsenic Concentration in Drinking Water in the Three Study Area Regions in Northern Chile (1930 –1994) Average Arsenic Concentration (␮g/Liter) Population Years ⬎18 Years Region City or Town of Age* 1930–1957 1958–1970 1971–1977 1978–1979 1980–1987 1988–1994 I Arica 109. kidney.532 15 15 15 15 15 15 Caldera 7. The inter. Most cancer patients.

Concentrations key results for the combined control group with those in earlier years. We conducted the main analyses using an fagasta 90% and Calama 93%. of diagnosis with that which would be predicted for a In addition. We grouped counties in the study area into five STATISTICAL METHODS exposure levels on the basis of average arsenic concen. and 59% of the households in the respective re. 11 No. the bladder cancer patients in a concurrent case-control gions. respectively. 1995. explaining the method of the study and variable (average packs of cigarettes smoked per year) in the general aim. College Station. at least once a year. were estimated on the basis selected for the lung cancer patients alone before com. with regard calculate odds ratios (ORs). TX) adjust- group selection methods used in this study. rounded. we assigned to each participant the average achieved hospital distribution of the controls with the water arsenic concentration for the county in which he target distribution. We reviewed occupational histories for copper Almost 100% of urban households are served by city smelting or refining as potential exposure to inhaled water systems. lifetime residential history. and region 3. years of school.0. trols. 1992 census. smoking. The coverage is lower in the small towns of each study with identical control selection methods. we could examine indirectly We used the lowest exposure categories as reference to the representativeness of the control group. We evaluated lifetime (1930 to the present) average trations in water supplies during the period 1958 –1970. occupation. and re. nine. noncancer controls. We also examined peak exposures on the these counties at the time of the study with that ex. (3) We compared the histories. In the 1992 census. We entered SES as a continuous variable and also stratified SES into three levels: low. separately. and house exposure in copper smelting. EXPOSURE ASSESSMENT and high. (4) We compared the actual distri. we made several validity checks. the study area in northern Chile including regions I. tration in drinking water from 1930 through 1994 for In view of the complexity of control group selection. of measurements in the 1950s. Age was treated as both a continuous and a PARTICIPANT INTERVIEWS categorical variable. region 2. Water arsenic concentrations have been compared key results for the cancer and noncancer con. commodities. and smoking. and nine The population-weighted average arsenic concentra- counties within them. and also the controls selected for 83%. Arica 92% and Iquique 94%. exposure of the source population in which cases oc- curred. when some of the highest exposures occurred. the exposure of controls should be representative of the which was from 1971 through 1977. 6 LUNG CANCER AND ARSENIC 675 these controls with those selected for the lung cancer through 1994. In this way. 64 – 80%. the population coverage by number of years worked in this occupation. 67–91%. and III have ten. II. region 2. and we have presented the findings using categorical variables The nurse interviewer read a letter of consent to all for six age strata. We first public water systems in the main cities was as follows: conducted analyses with cancer and noncancer controls region 1. or she resided for each year. analyzed separately. including cancer and 86% and Vallenar 84%. including information related to stratified analysis of smokers and nonsmokers (ever/ socioeconomic status (SES). Table 1 presents average arsenic concen- cases to increase the study power.Epidemiology November 2000. The results were similar. We calculated average bution of the controls by county of residence at the time arsenic water concentrations from 1930 to the present. 76 – 89%. Using lifetime residential bining the two control groups. amined synergy between arsenic and smoking in strati- gion 3. 1970. sex. These cities represent 88%. (1) We and III. as follows. water companies have been fied analyses involving never-smokers and ever-smokers required to carry out detailed chemical tests of the water. medium. arsenic exposure as a categorical variable with five ex- We then compared the control group distribution in posure strata. never smoked at least 100 cigarettes total in lifetime). and working in a copper smelter. II. The nurse then administered a struc- logistic regression analyses and assessed synergy in a tured questionnaire. basis of the average water concentration for each par- pected for a random population sample based on the ticipant in 1958 –1970. Since 1950. much predicted county distribution for the controls in the final higher than the population-weighted average of 212 validity check on the basis of the critical criterion that ␮g/liter for the second highest concentration period. Vol. occupation. Regions I. we calculated the average arsenic water random sample from the general population based on concentrations for the counties of residence for 1958 – the 1992 census given the age distribution of the con. 1930 –1957. Copiapo´ overall combined control group. We conducted unconditional logistic regression anal- We collected data on arsenic concentrations from 1950 yses that included the matching variables sex and age as . stratified into seven categories. SES. including measuring arsenic levels. We included smoking as a continuous study subjects. taking into account knowledge concerning trol groups separately before pooling. to tional regression analyses using StataCorp statistical investigate potential bias created by the complex control software (release 4. and the large majority of the population arsenic. We included an indicator variable for this ex- in this desert region receives water from town or city posure in the logistic regression analyses and also the supplies. (2) We compared when changes in water sources occurred. The focus in the occupational We estimated an SES score that took into account interview was to identify work with potential arsenic monthly income. ing for age. We conducted uncondi- to general population exposure to arsenic in water. We ex- region: region 1. (5) We also used this tion for region II in these years was 578 ␮g/liter. Anto.

We compared the achieved control group dis- Results tribution between hospitals with the target distribution PARTICIPATION OF LUNG CANCER CASES based on admissions in 1991 (Table 2). with one for sex and six for age age.676 Ferreccio et al Epidemiology November 2000.776 45 42 92 12 19 57 61 1.629 252 252 1.607 419 419 indicator variables. 6 TABLE 2. 151 (70%) agreed to participate and were the main hospitals of Arica and Antofagasta. Of served and expected control distributions resulted from these subjects.and sex-matching criteria. controls and compares characteristics and exposures of cancer controls. We found major During the 20 months of enrollment. strata. 11 No. and Concentration of Arsenic in Drinking Water for That Town or City in the Period 1958 –1970 Arsenic Water Population Concentration over 18 Control Control Ratio of Selected 1958–1970.137 68 44 288 38 24 106 68 0. the patient workers to recruit study subjects. Years of Numbers Numbers to Expected ␮g/Liter Age Expected Selected Controls 0–49 264.6 Iquique 10.8 50–99 104.138 84 124 717 94 98 178 222 1. One explanation for these dif- ticipation were that the patient was not at the hospital ferences may be variation in the capability of the field when we attempted to contact him or her. Selected Characteristics and Exposures among Lung Cancer Cases and Controls Controls for Control for Lung Bladder Total Cancer Noncancer Cancer Cancer Cases Controls Controls Controls Cases Cases Numbers 151 419 167 252 237 182 Percentage male 72% 61% 52% 67% 64% 57% Average age 61 64 64 64 61 68 Percentage ever smoked 80 55 56 53 55 55 Average packs of cigarettes (lifetime) 142 63 67 61 64 62 SES score 74 73 70 75 76 69 Worked in copper smelting 7 5 4 6 4 6 Average As water concentration (␮/liter) 1930–1996 171 109 103 114 106 114 1958–1970 464 280 250 301 265 300 SES ⫽ socioeconomic status. The former interviewed. those controls directly . particularly among men. the total combined control group with those of the cause of difficulties finding sufficient subjects meeting cancer and noncancer controls. TABLE 4.2 Copiapo´ 8. Assessment of Control Group Representativeness of Source Population Exposure Based on Residential Location in 1992. A series of validation checks was undertaken for the controls. Because of the discrepancy.434 66 33 0. recruitment of controls. The main differences between the ob- newly diagnosed with lung cancer in the hospital. In Antofagasta we had had moved from home and could not be located.4 Totals 622.578 55 42 175 23 26 78 68 0. controls than expected.1 Total 48. Vol.029 104 146 1. There were few refusals among cases and provided fewer controls and the latter provided more controls (less than 5%). The main reasons for nonpar.450 179 152 0. 217 subjects were discrepancies. 167 cancer controls and 252 non. we investigated various parameters relating to control group VALIDATION OF CONTROL GROUP validity.273 167 167 419 419 TABLE 3. Expected Number of Controls by Hospital in the Study Area Based on 1992 Discharges from Each Major Hospital in Northern Chile Eligible Noncancer Noncancer Eligible Cancer Cancer Total Total Selected/ Noncancer Controls Controls Cancer Controls Controls Controls Controls Expected Hospital Discharges Expected Selected Discharges Expected Selected Expected Selected Controls Arica 13. There were fewer cancer controls be. which may explain the increased patient was too sick to complete the questionnaire.9 Antfgst 16.694 70 88 1.5 400 and over 155. We included a total of 419 controls into the analysis Table 3 presents some basic information for cases and phase of the study. and the hired two field workers.3 100–399 98.

and working in copper smelting. Concentration and age.Epidemiology November 2000. Odds Ratio Estimates. 6 LUNG CANCER AND ARSENIC 677 TABLE 5. however.2 8. SES.5–5.8).6 6.1 3.4 4. Table 4 presents the results based on Controls OR Full Model 1. and socioeconomic status. the likely direction of bias is apparent. and the fifth column also includes adjustment (␮g/Liter) 0–10† 50–199 200–400 Arsenic 1930 –1994 10–29 30–49 for smoking. years of work in copper smelting. The final column shows the ratio of the selected num- Cancer Controls bers of controls to that expected. This assessment is indirect.8–10.2 9.0 95% Cl 0.4–5. the actual residential location of cases and 1. Clear trends in ORs are apparent.6–20.9 1.0–11. The overall criterion for assessing control group va- lidity in a case-control study is that the control group 1. owing to overselection of controls from Antofagasta (Table 2). Nevertheless.6 8.and Sex-Adjusted data analysis.5 3.5 1 lected controls to expected ratio of 0.5 2.4–5. however.5 4.7–9.8–17. The remaining columns of the table show that similar results are obtained if cancer controls are used alone. it is possible that some underascertainment N ⫽ 419 104 39 23 124 129 would have occurred in the same cities in which controls were underselected. The high ex- posure category is overrepresented. Number of Cases and Controls.0 1 lower water concentrations. appears Total 1. the ORs may be underesti- mated.3–11.0–19. 1.* and 95% CI for Lung Cancer by Exposure to Arsenic in Drinking Water: Average Concentration during selected for lung cancer patients.7 4. The fourth column presents ORs using the pooled controls and adjustment by sex Average Water † Referent category.6 bution for 419 randomly selected controls (column 3 of 95% CI 0.4 should provide an unbiased estimate of the exposure distribution of the general population in which the cases occurred. Vol. Although case ascertainment was Control Group thorough. The impact of this bias would be to underestimate ORs for the highest arsenic exposure.6 controls was used throughout to determine arsenic water concentrations.2–12.8 3.7 Table 4).0 4. This distribution was then compared with the actual distribution of the selected controls (column 4).6 1. Table 5 presents findings based on average drinking water concentration from 1930 (or year of birth if the N ⫽ 151 subject was born later) through 1994 using all controls Cases 9 5 8 50 79 pooled and compares results of analyses conducted with different control groups.9 5.5 5.4–21. sex. cumulative lifetime cigarette smoking. At relatively low concen- trations of 50 –99 ␮g/liter. Population num- bers in 1992 were used to estimate an expected distri- 1. The baseline exposure (N ⫽ 167 Controls) category (0 – 49 ␮g/liter) is reasonably represented (se- 1. and for age.8 95% CI 0. which would lead to overestimation of ORs.0–22.6 the only location with water concentration above 400 95% CI 0. and those selected for * Estimates compare results of analyses conducted with different control groups adjusting for age and sex only. This distribution was assessed indirectly using Noncancer (N ⫽ 252 Controls) the 1992 census. 11 No.5 3.3 ␮g/liter in the period 1958 –1970 (Table 1).9 7. The next-to-highest arsenic (N ⫽ 419 Controls) exposure category. noncancer controls are used .2 95% CI 0.6 3. Because the differences in exposure Controls) for the various control sources were small.3 2. The combined control group had been exposed to an average arsenic concentration of 109 ␮g/liter in drinking water between 1930 and 1994 and an average of 280 ␮g/liter for the peak exposure Controls Frequency Cancer (N ⫽ 237 Matched to Lung period 1958 –1970.7 1 results from major analyses using the overall combined control group.2–10. 100 –300 ␮g/liter.4 8.0 3. with underestimation of ORs in the (N ⫽ 419 Controls) highest exposure levels and overestimation of ORs at the 1.0 4.4–4.7–15.5 3. In OR Age.0–12.4–12.9 1 to be markedly underrepresented.3–3.3–17. rather than just residential location at the time of the study. 1.1 2. The extent of bias is also dependent on ascer- Controls Combined tainment of cases. we present 1.5 water arsenic concentration in the period 1958 –1970 1 that are already presented in Table 1.1 the bladder cancer cases.

9 60–89 22 51 0.4 TABLE 7.1–0.0 0.1 200–399 23 44 0.7 0. FIGURE 1. Median arsenic exposure.5–6.8 0.3 0.6) is consistent with that of 4. or if the matched controls selected directly for the Discussion lung cancer patients are used alone (excluding the ad.6–1.2 4.2 1.8 5.3 62 61 32.22–198.3 SES medium vs low 1.7–2. Similar results were obtained after adjust- ing for age and sex. a 70% increase in rel.9 2.7 1. Table 7 presents results for nonsmokers and smokers of cigarettes.1 1.0–11. alone. Cases had higher expo- sures than controls.6 4. In the relative risk estimate among the most highly exposed full-model logistic regression analysis.7–52. Working in Copper Smelting.7 1.5 0. the markedly increased relative risks of lung can- cer found in this study relate to exposures that predom- inately occurred 20 – 40 years before cancer diagnosis. and appar.0) is much greater than that expected (13.4–14.0 –19. cases of lung cancer in people who had been drinking ative risk associated with copper smelting. water containing even higher concentrations of inor- ently slightly higher relative risks associated with higher ganic arsenic (more than 1.2–40. we found an OR (OR ⫽ 8. A clear trend is again apparent.1–2.9 1.31–39.6–7.0–7.8–9.2 39 66 18.9 700–999 64 103 0.1 1.1 0.6 90–199 13 36 0.0 7. actual level not SES. and Socioeconomic Status (SES).3 0.9–16. Figure 1 presents the time-window pattern of exposure of cases and controls. 1930 –1994.0 * Referent category. Sex.1).1 1. Number of Cases and Controls.6 –7.1–1. Vol.1 Ever/never worked in copper smelting 1.2 30–59 4 19 0.7 2. 11 No. . According to Exposure to Arsenic in Drinking Water: Average Concentration during Peak Years of Exposure.3 2.678 Ferreccio et al Epidemiology November 2000.0 400–699 11 12 0.0 1.0–10.8 7.9.7 1.5 SES high vs low 2.8 Male vs Female 0. The OR for smokers in the highest arsenic-exposure category (32. 1958 –1970 Cases Controls OR Age-/Sex-Adjusted 95% CI OR Full Model 95% CI Average Arsenic Exposure (␮g/Liter) 0–10 11 92 1 1 10–29 3 62 0.7–4. Odds Ratio Estimates for Lung Cancer. Cumulative Lifetime Cigarette Smoking.5–12.0 ⫹ 6.4 ⱖ200 17 67 8.7 1.0) and the OR for smokers in the lowest arsenic-exposure category (6.8 1. Thus. Adjusted for Age.3] for lung from a small cohort study in Japan that reported eight cancer associated with smoking.13–116.8 Ever vs never smoked 4. 6 TABLE 6. Our findings demonstrate a positive trend in relative risk of lung cancer with exposure to increasing concentration of arsenic in drinking water among nonsmokers as well as a greater-than-additive effect for these exposures combined. The been included.3 [95% confidence interval (CI) ⫽ 2. This is the first study based on a large population ex- ditional controls derived from the bladder cancer study).1–7.6–7. Interaction of Exposure to Arsenic and Smoking on Relative Risk of Lung Cancer 1930–1994 Never Smoked Ever Smoked Average Arsenic in Drinking Cases Controls Cases Controls Water (␮g/Liter) 30 189 OR 95% CI 121 230 OR 95% CI ⱕ49 2 63 1* 20 103 6. 1958 –1970.3 3.4–19.1 – 1) on the basis of the OR for nonsmokers in the highest arsenic- exposure category (8.2 50–199 11 59 5.4 0.3 0.0 2.1 3. 95% CI ⫽ 4. All controls have risks and to evaluate synergy with other exposures.1 ⫽ 8.000 ␮g/liter.0 4. Finally. posed to arsenic in drinking water conducted to docu- Table 6 presents findings based on the period of peak ment the relation between this exposure and lung cancer exposure in Antofagasta.1–12.4 2.8–9. especially in the period 1955–1975.

because arsenic concentrations in water supplies vary by city and geographic location throughout the study area. 11 No. Arsenic ingestion and internal cancers: a review. Chuang YC. arsenic exposure and lung cancer. Lung cancer and arsenic exposure in drinking water: a case-control study in considered (Table 6). Chen CJ. day. Lyon: International Agency for Research chosen from the highly exposed city of Antofagasta than on Cancer. Cancer Res 1985. Horng SF. This stability in control group expo.8 for men (95% CI ⫽ gested arsenic and smoking in causing lung cancer is 3. Holtzman D. would have effectively ergistic action between the two exposures. Impacto en Salud atribuible a exposicio´n a Arse´nico: un estudio Ecolo´gico. Arsenic in Drinking Water. and just 12. The only previous study with any dose-re. Acknowledgments gasta hospital patient were selected as a control for a We thank Adriana Tapia and the following institutions for their collaboration with the study: Servicio de Salud de Arica. Scand J Work Environ Health 1993. Marshall G. Bates MN. Hertz-Piccioto I.11 the over. Milosavjlevic V. Dose-response relation between shows little difference in effect estimates according to arsenic concentration in well water and mortality from cancers and vascular average exposure. Kuo TL. Am J Epidemiol 1989. Chiang MH.4 –31. Washington DC: National Academy Press. Sancha AM. Aoyama H.7–3. Hence. You SL. Am J Epidemiol 1995. When coupled with lung cancer findings related to gistic effect of the two exposures. and Servicio de Salud de Copiapo´. Wu HY. Argentina. Haque R. Ingested arsenic and internal cancer: a historical cohort exposure was the cohort study in Japan. Smith AH. The findings. overall population standardized mortality ratios for lung Published evidence concerning synergy between in- cancer in region II of Chile of 3. sup. 17. Nevertheless. Int of subjects was small at all dose categories. the control selection area in Taiwan.15 On the basis of small arsenic exposure in Taiwan2–9 and Argentina. For example. Ferreccio C. Kurumatani N. which has 15. 6 LUNG CANCER AND ARSENIC 679 specified) with 0. analyses using various control groups shown in Table 3 6.3.66:888 – 892. A retrospective study on malignant neoplasms of bladder. A meta-analysis of studies of inhalation of in- all of Chile. Chen CW. Smith AH. Biggs ML. seven-year follow-up study in Taiwan. Pu YS. Wu MM. Vol. Alexeeff G. Gonza´lez C. 4. bladder and kidney due to ingested inorganic arsenic in drinking water. the average number of cigarettes risk estimates reported here are also consistent with the smoked per day was only 13. 8:452– 460. Am J Epidemiol 1992.5– 4. synergy between arsenic ingestion and smok- all evidence is sufficient to conclude that ingestion of ing was suggested in the Japanese cohort study. diseases. Hwang YH. Gonza´lez C. Wu MM.27:561–569. Mino Y. Kuo TL. and the contrast with exposure among cases. thing.51 expected cases (standardized mor.1: 414 – 415. Goycolea M. IARC Monographs on the however. 1997. Br J Cancer 1986. Clear trends in dose response are apparent when and lung cancer mortality in a region of northern Chile due to arsenic in concentrations are averaged over 1930 –1994 (Table 5) drinking water. numbers. especially the J Epidemiol 1998. Servicio de Salud de Iquique.7)13 using limited. Smith AH. matched on exposure. standards.147:660 – 669. Wu HY. support a syn- hospital. Babazono A. Chen CJ. Lin JS..130:1123–1132. The use of hospital controls with matching by limits are broad. Cancer Res 1995.1 for women (95% CI ⫽ 2. 7. both tables is consistent with supralinearity. National Research Council. 2. 13.3:23–31.135:462– 476. Tsuda T. Biggs ML. Hertz-Piccioto I. as is usually done. Arteriosclerosis 1988. lung. confidence selection.19: ently low OR associated with smoking alone is due the 217–226. International Agency for Research on Cancer.53:399 – 405.7. Smith AH.1) and 3.55:1296 –1300. 10. Br J gested arsenic and lung cancer risk is important when Cancer 1992. Sancha AM. problem does not affect causal inference in that. Lung and kidney cancer mor- included three dose-exposure categories. that the control selection criteria were not Evaluation of the Carcinogenic Risk of Chemicals to Humans. .10 Even in inorganic arsenic increases the risk of lung cancer. organic arsenic and cigarette smoking supports a syner- tion. if any. vol. Am J Epidemiol 1998. Lipsett M. Wu MM. Smith AH. 8. the present study graph FONDEF Proyecto 2-24. It is clear. lung and liver in blackfoot disease endemic highest exposures (Table 4). Ogawa T. The shape of the dose-response relation between in. Wang JD. Wu HY.10 This study only study followed for 33 years. 23. the direction of this bias is clear sian well water and cancers. if an Antofa. Universidad de Chile. Cheng SH. Chen CJ. Arsenic and cancers (Letter).45:5895–5899. Hsueh YM. Cad Saude Publica 1998. CH. and also when the peak exposure period 1958 –1970 is 14. Lancet 1988. Chen CJ. excluding region II. Epidemiology 1992.141:198 –209.16. Santiago. 1. Some fully adhered to and that relatively more controls were Metals and Metallic Compounds. Lee SS.0). as the referent popula.13 In our case-control study. Atherogenicity evidence of increased lung cancer risks associated with and carcinogenicity of high arsenic artesian well water: multiple risk factors arsenic in drinking water. Ferreccio C. two lowest (no cases reported for ⬍50 ␮g/liter. Wu MM.Epidemiology November 2000. and the number tality associated with arsenic in drinking water in Cordoba. In addition. The dose-response information in northern Chile. Milosavjlevic V. however. Synergism fact that the majority of smokers in a survey of two major between occupational arsenic exposure and smoking in the induction of lung cities in region II smoked fewer than ten cigarettes per cancer. hospital. Incidence of Internal cancers and ingested inorganic arsenic: a ports the validity of the study findings. Cancer potential in liver. previously been explored in the context of lung cancer 16.14(suppl 3):193–198. Mono- one case for 50 –990 ␮g/liter). the present study with more than ten times the number The main weakness of the study concerns control of lung cancer cases as in the Tsuda study. Marked increase in bladder data. Chen CJ. correcting for this bias would only add to the 5.17 We note that the appar. Chile: Facultad Ciencias Fı´sicas y is the first to provide potentially useful dose-response Matema´ticas. even among patients bidity ratio ⫽ 15. Liaw KF. Observations on the dose-response curve for risks from inhaled arsenic. Yamamoto E. Lin TM. sponse information based on knowledge of individual Kishi Y.10 The relative with lung cancer. Chen CJ. the likelihood is that both patients would have been drinking from the same water supply and would have had similar exposure to the very high arsenic levels References in Antofagasta water supplies in past years. Chiou HY. considering population cancer risks and drinking water 9. evaluation of and related malignant neoplasms of blackfoot disease. Marshall G. Chen CJ. Hopenhayn-Rich C. 1999. Huang sure. Malignant neoplasms among from the lower-exposure cities of Arica and Iquique residents of a blackfoot disease-endemic area in Taiwan: high-arsenic arte- (Table 2). 1980. 95% CI ⫽ 7. Hopenhayn-Rich C. Thus. Lin TM. in that it would result in underestimation of risks for the 3. Kuo TL. 11.39 –141. Chuang YC. Servicio lung cancer patient also admitted to an Antofagasta de Salud de Antofagasta.