320 Occupational and Environmental Medicine 1996;53:320-327 Geographical and temporal differences in the

urinary excretion of inorganic arsenic: a Belgian Occup Environ Med: first published as 10.1136/oem.53.5.320 on 1 May 1996. Downloaded from population study

J P Buchet, J Staessen, H Roels, R Lauwerys, R Fagard

Abstract necessity to validate environmental moni- Objective-This Belgian study assessed toring programmes by directly estimating the geographical and temporal differ- the internal exposure ofthe population. ences in the exposure of the population to inorganic arsenic, a known carcinogen. (Occup Environ Med 1996;53:320-327) Methods-In the CadmiBel study (1985-9) the 24 h urinary arsenic excretion was measured, as an index ofrecent exposure, Keywords: arsenic absorption; environmental pollu- in industrialised cities (Liege: n = 664, tion; general population Charleroi: n = 291), in a rural control area (Hechtel-Eksel: n = 397), and in Arsenic is an ubiquitous element. Its toxicity rural districts in which the population mainly results from excessive exposure to inor- had possibly been exposed through the ganic derivatives, in particular those in which drinking water or the emissions of non- arsenic is in the trivalent state, such as As,03. ferrous smelters (Wezel: n = 93, The general population is exposed to inor- : n = 111, and Pelt: n = 133). In ganic arsenic through inhalation of dust parti- the PheeCad study, in 1991-5, the rural cles containing arsenic and through ingestion areas (n = 609) were re-examined of foodstuffs and tap or bottled water contain- together with an urban control area ing arsenic pumped from geological layers (Leuven: n = 152). containing arsenic, and possibly through Results-The CadmiBel results showed smoking.' The relative importance of these that after adjustment for sex, age, and sources may vary depending on the area of res- body mass index, the 24 h arsenic excre- idence and lifestyle habits. tion was on average low in Liege (91 Long term oral exposure of the general pop- nmol), Charleroi (155 nmol), Hechtel- ulation to inorganic arsenic through drinking Eksel (144 nmol), and Wezel (158 nmol), water may affect various organs such as the whereas the highest excretions were skin, the liver, the heart, the peripheral vascular found in Lommel (570 nmol) and Pelt (373 system, the central and peripheral nervous sys- http://oem.bmj.com/ nmol). During the PheeCad study, the tem, and the bone marrow. Similar effects mean 24 h arsenic excretion in the rural have been reported in subjects who have regu- areas ranged from 81 to 111 nmol. This larly consumed medications containing was lower than six years earlier and simi- arsenic, mainly Fowler's solution. However, lar to the excretion in the control town the main risk after long term ingestion of inor- (108 nmol). Longitudinal studies in 529 ganic arsenic is the induction of cancer of the in the people living rural areas confirmed skin and possibly also of various internal on October 1, 2021 by guest. Protected copyright. that their 24 h arsenic excretion had organs.3 In workers chronically exposed to decreased (P < 0-001) from 222 to 100 inorganic arsenic by inhalation mortality from nmol. As well as the drinking water, lung cancer increases.4 industry was likely to be a source of the Non-ferrous smelters have been an impor- Industrial Toxicology increased exposure in Lommel and Pelt in tant source of emission of various and Occupational inorganic Medicine Unit, 1985-9, because at that time the urinary substances (lead, cadmium, zinc, arsenic) in University ofLouvain arsenic excretion did not follow the the environment mainly in the past when regu- (UCL), 1200 Brussels regional differences in the arsenic content lations on industrial emissions were less strin- J P Buchet H Roels of the drinking water, because the fall in gent. Between 1985 and 1989 we examined a R Lauwerys the arsenic excretion over time coincided random sample of the adult population living Hypertension and with the implementation by industry of near non-ferrous smelters in Cardiovascular stricter environmental regulations, (Noorderkempen) and have found that their Rehabilitation Unit, because in individual subjects the urinary Department of lifetime integrated exposure to cadmium Molecular and arsenic excretion was inversely correlated (assessed by measuring cadmium in urine) was Cardiovascular with the distance to the nearest smelter, significantly increased compared with subjects Research, University and because an increased arsenic excre- living in areas less of Leuven (KUL), 3000 polluted by cadmium Leuven tion was only found downwind from the (CadmiBel study).5 Because non-ferrous R Fagard main smelter. The official network that smelters may release not only cadmium but J Staessen monitors the arsenic concentration in air- also arsenic we analysed urine samples for Correspondence to: borne and fall out dust did not detect the Dr J P Buchet, Clos their combined content of inorganic arsenic Chapelle-aux-Champs 30 54, high exposure in Lommel and Pelt (As1) and its methylated metabolites B-1200 Bruxelles, Belgium. between 1985 and 1989. monomethylarsonic acid (MMA) and Accepted 8 December 1995 Conclusion-This study highlights the dimethylarsinic acid (DMA). The sum of Geographical and temporal differences in the urinary excretion ofinorganic arsenic: a Belgian population study 321

these three arsenic species represents a reliable years, and one subject living in the Liege area, indicator of recent uptake of inorganic because her questionnaire showed that at the arsenic.6 Inhabitants of the Noorderkempen time of the examination she was taking med- and the corresponding control rural area ications containing arsenic. In this woman the Occup Environ Med: first published as 10.1136/oem.53.5.320 on 1 May 1996. Downloaded from (Hechtel-Eksel) were re-examined five years urinary arsenic excretion was 34 pmol/24 h. later (Pheecad study between 1991 and 1995). Thus, for the present analysis the CadmiBel During this second survey an additional group study group totalled 1689 subjects. of control subjects were examined in an urban area (Leuven). PheeCad study (1991-5) The 1103 CadmiBel participants living in the rural districts were invited to take part in a Materials and methods longitudinal survey, the PheeCad (Public STUDY POPULATIONS Health and Environmental Exposure to CadmiBel study (1985-9) Cadmium) study.9 This study took place A detailed protocol of the CadmiBel study has between 1991 and 1995. Eighty two subjects already been published.7 A random sample of had died. After excluding those who were 2327 men and women aged between 20 and severely ill (n = 4) or had moved (n = 7), 80 was recruited in four areas of Belgium: two 1010 people were left. Of these, 821 (8%) par- urban (Charleroi and Liege) and two rural ticipated. For the present analysis, 30 subjects (Hechtel-Eksel and Noorderkempen) districts. were excluded because not all relevant mea- Both rural areas are 10 to 15 km apart but in surements had been obtained, eight were one of them (Noorderkempen) two non-fer- excluded because their 24 hour urine sample rous smelters were active at the time of the was under or over collected,8 171 because they survey and a third one, dismantled in 1974, had potentially been exposed to heavy metals was still surrounded by industrial waste at work, and three subjects because their ques- deposits (fig 1). tionnaire showed that they had been on med- From 1985 to 1989 a total of 2327 subjects ications containing arsenic. In these subjects with a minimum age of 20 and who had lived the urinary arsenic excretion was 192, 68, and for at least eight years in the areas were stud- 39 pmol/24 h. For the present analysis the ied. The participation rate among the subjects PheeCad study group thus totaled 609 sub- contacted was 78% in the two rural areas (n = jects. In 529 (87%) of these subjects the uri- 1103) and 39% in the two urban areas (n = nary arsenic excretion had been measured 1220). Among the 2327 subjects, 214 were both at baseline (CadmiBel study) and during excluded because not all relevant measure- follow up (PheeCad study). ments could be obtained, 44 were excluded During the PheeCad study an additional because their 24 hour urine sample was judged group including 135 volunteers, mainly to be under or over collected on the basis of recruited among hospital staff, and 25 patients previously published criteria,8 343 (313 men with essential hypertension from the out and 30 women) because they had possibly patient clinic of the University Hospital in

been exposed at work, 36 because they were Leuven were also examined.10 These subjects http://oem.bmj.com/ not within the age boundaries from 20 to 79 were included in the present analysis because

Figure 1 The investigated areas in Iz.. Rural areas Belgium (insert) and (Closed in 1992) detailed description ofthe (Dismantled in 1974) rural area. on October 1, 2021 by guest. Protected copyright. .K aK Lommel

Pelt

Iv- _ Industrial sites 77 Polluted sites ± 10 km :.... Control districts

Wezel ./ Rural areas .e Brussels . ;,..:\ ~~~L-euven- Liege N -* Charleroi A Hechtel-Eksel W -E 1 km s S 30 km 322 Buchet, Staessen, Roels, Lauwerys, Fagard

they lived in an area of Belgium where no Urinary arsenic was measured after arsine smelters had been in operation, and because generation under conditions which allow the they constituted a contemporary urban control measurement of the sum of inorganic arsenic group for the PheeCad study. Timed urine and its methylated metabolites (monomethy- Occup Environ Med: first published as 10.1136/oem.53.5.320 on 1 May 1996. Downloaded from specimens for the measurement of arsenic larsonic acid (MMA) and dimethylarsinic acid were available in 152 subjects. (DMA)) without notable interference of seafood trimethylated arsenicals.'4 A possible EPIDEMIOLOGICAL METHODS but limited influence of fish consumption on The CadmiBel and PheeCad studies were the urinary excretion of DMA'5 cannot totally conducted with the use of similar epidemio- be excluded; however, the large number of logical techniques.7 All subjects were repeat- subjects examined, the low rate of fish con- edly visited at home. During each survey the sumption among Belgians, and the geographi- subjects completed the same self administered cal vicinity and similar social classes of the questionnaire inquiring into their medical his- populations under study strongly reduced the tory, current and past occupations, smoking probability that differences in urinary excre- habit, consumption of alcohol, and intake of tion of arsenic were caused by differences in medications. In the rural area social class was seafood ingestion. determined from each person's current and During the study, the laboratory partic- past occupations," according to the rules of ipated in the external quality control pro- the Office ofPopulation Censuses and Surveys gramme organised by the Centre de in London.'2 Toxicologie du Quebec, Laval University, In the two rural areas the geographical posi- Quebec, Canada. In 18 analyses the mean tion of the subjects' homes as well as of the (SD) results were 0 4% (4%) lower than the industrial settlements was determined from target values indicating that no changes in the maps drawn at a scale of 1/10000. The dis- analytical procedure occurred during the mea- tance of each house from the nearest smelter surement period. was then calculated. The creatinine concentration in serum and The CadmiBel and PheeCad participants urine and the activity of yglutamyltrans- collected a 24 hour urine sample for the mea- ferase'7 in serum were measured by automated surement of arsenic, cadmium, and creatinine enzymatic techniques (Technicon Auto- in a wide necked polyethylene container. A analyzer, Technicon Instruments, Tarrytown, venous blood sample was obtained within two New York, USA). weeks after urine collection. The serum creati- nine concentration was measured as an index STATISTICAL ANALYSIS of renal function and the serum yglutamyl- Data base management and statistical analyses transferase activity as a measure of alcohol were performed with the SAS software (The consumption. 13 SAS Institute, Cary, North Carolina). Non- normally distributed variables (arsenic in BIOCHEMICAL MEASUREMENTS urine) were logarithmically transformed and,

The urinary cadmium7 and arsenic'4 concen- unless otherwise specified, described by the http://oem.bmj.com/ trations were measured by atomic absorption geometric mean and the extremes. spectrometry. Samples collected during both Means were compared by Student's t test surveys were analysed for arsenic within a six and proportions by a x2 statistic. Covariates of month period; urine samples were kept frozen the 24 hour urinary arsenic excretion in indi- at -210C before arsenic measurements. vidual subjects were traced by stepwise regres- Thawed urine samples were well shaken to sion. After standardization for known allow a representative sampling of precipitated confounders, the various geographical areas material when present. were compared by analysis of covariance. on October 1, 2021 by guest. Protected copyright.

Table 1 Characteristics ofthe population studies 1985-9 Results Men Women CADMIBEL STUDY (1985-9) Number 670 1019 Characteristics of the population Residence (n (%)): Table 1 summarises the characteristics of the Hechtel-Eksel 173 (25-8) 224 (22-0) 1689 CadmiBel whom Noorderkempen 97 (14-5) 240 (23 5) participants among Charleroi 116 (17-3) 175 (17 2) 308 men and 357 women were currently Lige 284 (42-4) 380 (37 3) smokers (median (range) 18 (1-75) cigarettes Clinical measurements (mean (SD)): Age (y) 47-6 (16-1) 47-0 (15-6) a day). Daily intake of alcohol (median (range) Body mass index (kg/rm) 25-5 (3 5) 25-4 (5 2) 20 (3-170) g per day) was reported by 245 Measurements on serum: men and 137 women. In the two rural areas, Creatinine (pmol/l, mean (SD)) 112 (24) 94 (19)* y-Glutamyltransferase the median (range) distance from the partici- (U/I, geometric mean (range)) 15 (2-252) 9 (2-334)* pants' houses to the nearest smelter was 3 0 Urinary measurements: (0-3-21.1) km. Volume (1/24h, mean (SD)) 1-63 (0 68) 1-68 (0 74) Arsenic (nmol/24h, geometric mean (range)) 164 (13-3 069) 133 (13-11 718)* Cadmium (nmol/24h, geometric mean (range)) 7-8 (0 4-33-9) 7-2 (0.5-70.8)*** Determinants ofthe urinary arsenic excretion Creatinine (nmol/24h, mean (SD)) 15 4 (4-0) 10-5 (2 7)* The urinary arsenic excretion was consistently Questionnaire data (n (%)): Intake of diuretics 30 (4-5) 99 (9.7)* higher in men than in women and tended to Current smokers 308 (46-0) 357 (35 0)* be curvilinearly related to age. In stepwise Alcohol intake 245 (36 6) 137 (13-4)* multiple regression (table 2), the multiple par- *P < 0 05; *** P < 0-001; sex differences. tial correlation coefficients for the linear and Geographical and temporal differences in the urinaty excretion ofinorganic arsenic: a Belgian jpopulation study 323

Table 2 Correlates ofthe urinary excretion ofarsenic (logarithmically transformed) in the squared terms of age combined were 0-07 CadmiBel study (1985-9) as identified by stepwise multiple regression (P = 0O10) in men and 0-08 (P = 0-01) in Men Women women. Overall, the urinary arsenic excretion Occup Environ Med: first published as 10.1136/oem.53.5.320 on 1 May 1996. Downloaded from Number 670 1019 went through a maximum at 30 years in men R2 0-20 0-26 and 40 years in women. The partial regression Intercept (SE): 1-67 (0-15) 1-63 (0-12) coefficients showed that, after adjustement for Partial regression coefficients (SE): all other covariates in the model including sex, Age (y) NS 0-0129 (0-0052)** from the fourth to the Age squared (y2) NS -0-000145 (0 000053)** aging seventh decade of Body mass index (kg/M2) 0-0085 (0 0042)* NS life was associated with a decrease in the uri- Residence: arsenic excretion men Hechtel-Eksel (0,1) 0-138 (0-036)*** 0-243 (0 036) nary averaging 19% in Noorderkempen (0,1) 0 541 (0-044)*** 0-601 (0 033) and 16% in women. In men, the urinary Charleroi (0,1) 0-167 (0-041)*** 0-280 (0 036) arsenic excretion was also positively correlated *P < 005; **P < 001, ***P < 0001. with body mass index (partial correlation coef- The four areas in the analysis (Hechtel-Eksel, Noorderkempen, Charleroi, and Liege) were iden- tified by three dummy variables. ficient: 0 07; P = 0-02). After adjustment for The variables which were considered, but did not enter the regression model (P > 0- 15) were: age and body mass index, the area of residence social class, log -glutamyltransferase, intake of diuretics, current smoking, and serum creatinine concentration. appeared as a strong determinant of the uri- nary arsenic output. The multiple partial cor- relation coefficients for the three dummy variables combined, which identified the four Table 3 Correlates oflog urinary arsenic excretion in the CadmiBel and PheeCad areas in the analysis (Hechtel-Eksel, Noorder- participants who resided in the rural areas with low and high exposure kempen, Liege, and Charleroi), were 0 44 < men CadmiBel study PheeCad study (P 0001) in and 050 (P < 0001) in (1985-9) (1991-5) women (table 2). The variables which did not enter Number 734 609 the model (P > 0K15) were: social class, R2 0-31 0-02 current smoking, the intake of diuretics, renal Intercept (SE) 2-00 (0-13) 2-13 (0-06) function as measured by the serum creatinine Partial regression coefficients (SE): concentration, and alcohol intake assessed by Sex (1 = men, 2 = women) -0-087 (0-029)*** -0-077 (0Q033)* the activity of serum Age (y) +0-0155 (0 0055)** NS rglutamyltransferase. Age squared (y2) -0-000178 (0 000055)*** NS The urinary excretion of arsenic adjusted for Intake of diuretics (0,1) 0- 114 (0-051)* NS the significant covariates amounted to 91, Residence: Wezel (0,1) NS -0-104 (0 046)** 155, 144, and 341 nmol/24 hours in the Liege, Lommel (0,1) 0-587 (0-039)*** NS Charleroi, Hechtel-Eksel, and Noorder- Pelt (0,1) 0-413 (0-036)*** NS kempen areas, respectively. In the Noorder- *P < 0.05; **P < 0.01; ***P < 0.001. kempen (Wezel, Pelt, and Lommel) the The four areas in the analysis (Hechtel-Eksel, Wezel, Lommel, and Pelt) were identified by three dummy variables. The covariates which were considered, but did not enter the regression excretion was significantly lower in Wezel than model (P > 0-15) were: body mass index, social class, log y-glutamyltransferase as index of in Pelt and Lommel (fig 2). (Conversion of alcohol consumption, current smoking, and renal function as measured by the serum creatinine concentration. units: arsenic, 1 jimol = 74.9 jig; cadmium, 1 nmol = 1124ng; creatinine, 1 mmol = 112-13 mg.) http://oem.bmj.com/

Figure 2 The geometric CadmiBel study PheeCad study mean (SE) urinary (1985-9) (1991-5) excretion of arsenic adjustedfor sex, age (linear and squared terms), and body mass index in different areas 700 Liege examined during the 570 Charleroi on October 1, 2021 by guest. Protected copyright. CadmiBel (1985-9) and PheeCad (1991-5) Hechtel-Eksel studies. 600 -- Noorderkempen 0 E Wezel

- *__ Pelt 500 - Lommel CN4 373 a) Leuven

a, 400 -- * P<0.05 341 C: P<0.001 t P=0.08 C: 300 0

o

x : +~~~~~~~~~~1 ° 200 158 155 144 coU, 144: cn 111 108 A I 99 99 100 -- 81

0 n = 664 291 397 337 111 321 83 103 97 152 324 Buchet, Staessen, Roels, Lauwerys, Fagard

Table 4 Distance from the nearest smelter as a determinant oflog uinary arsenic These analyses showed that distance was not excretion in the rural areas surveyed 1985-9 correlated with the urinary arsenic excretion in Regression Partial regression the rural control area (Hechtel-Eksel) and in Occup Environ Med: first published as 10.1136/oem.53.5.320 on 1 May 1996. Downloaded from Residence n coefficient P value coefficient* P value Wezel both before and after adjustment for the relevant covariates listed in table 3. However, Hechtel-Eksel 397 -0-0220 > 0-05 -0-0116 > 0-05 Wezel 93 0 0919 > 0-05 0-0916 > 0-05 both in Lommel and Pelt, the distance to the Lommel 111 -0-1054 < 0-01 -0-1180 < 0 01 nearest smelter remained independently corre- Pelt 133 -0-0229 < 0-01 -0-0222 < 0-02 lated with the urinary arsenic excretion after *Adjusted for relevant covariates (see table 3). similar adjustments (table 4). Distance to the nearest smelter was measured in km. PHEECAD STUDY (1991-5) Characteristics ofthe population Table 5 shows the characteristics of the several groups examined between 1991 and 1995 In the PheeCad group, the median distance from the participants' houses to the nearest smelter was 7-8 (0O3-16-2) km. Determinants of the urinary arsenic excretion At follow up, in the PheeCad survey, the rela- tion between the urinary arsenic excretion and age was weak and failed to reach the 5% threshold of significance. When men and women were pooled, the main determinant of Figure 3 The geometric mean (SE) urinary excretion of was arsenic adjustedfor sex, age (linear and squared terms), the urinary arsenic excretion residence. and body mass index in subjects examined twice during the Living in Wezel was associated with a signifi- CadmiBel (1985-9) and PheeCad (1991-5) studies. cantly lower urinary arsenic excretion than was living in Hechtel-Eksel, Lommel, and Pelt. The partial correlation coefficient associated Further analyses were performed in the with living in Wezel was 0 1 1 (P < 0 01), and CadmiBel participants, who lived in the rural with sex was 0 09 (P = 0 02). areas (table 3). Men and women were pooled Both before and after adjustement for sex, in these analyses, because the determinants of age (linear and squared terms), and body mass the urinary arsenic excretion were largely the index, the urinary arsenic excretion in the same in the two sexes (table 2). The four dis- rural areas was considerably lower (fig 2) dur- tricts in the analyses (Hechtel-Eksel, Wezel, ing the PheeCad study than six years earlier Lommel, and Pelt; fig 1) were coded by three and did not differ from the excretion found in dummy variables. Again, the geographical the urban Leuven area, where non-ferrous position was a strong determinant of the uri- smelters had never been in operation. In

nary arsenic excretion. The multiple partial Wezel, the urinary arsenic excretion tended to http://oem.bmj.com/ correlation coefficient for the dummy variables be lower than in the three other areas (fig 2). which defined Lommel and Pelt was 053 (P < 0O001). The associations of the urinary Temporal differences in the urinary arsenic arsenic excretion with distance were assessed excretion in several geographical entities (table 4). Longitudinal studies of the urinary excretion

Table 5 Characteristics ofthe PheeCad and Leuven populations on October 1, 2021 by guest. Protected copyright. PheeCad (1991-5) Leuven (1992-3) Men Women Men Women Number 221 388 82 70 Residence (n (%)): Hechtel-Eksel 133 (60 2) 188 (48 5) - - Noorderkempen 88 (39-8) 200 (51-5) - Leuven - - 82 (100 0) 70 (100 0) Clinical measurements (mean (SD)): Age (y) 50-1 (14 9) 51-0 (14-4) 33-6 (10-1) 32-5 (10-4) Body mass index (kg/m2) 26-1 (3-6) 26-4 (5-7) 24-6 (3-5) 21 8 (3 4) Measurements on serum: Creatinine (pmol/l, mean (SD)) 109 (19) 90 (21)* NA NA y-glutamyltransferase (U/1, geometric mean (range)) 17 (5-123) 12 (3-233)* NA NA Urinary measurements: Volume (1V24h, mean (SD)) 1-49 (0-57) 1-59 (0-66) 1-46 (0 68) 1-64 (0-63) Arsenic (nmol/24h, geometric mean (range)) 112 (13-1211) 92 (1-17 551)* 132 (32-998) 79 (3-1066)*** Cadmium (nmol/24h, geometric mean (range)) 7-5 (14-60-0) 7-5 (67-46-5) NA NA Creatinine (nmol/24h, mean (SD)) 15-4 (4 2) 9 9 (2.2)* 15-9 (2-6) 10-9 (1.7)* Questionnaire data (n (%)): Intake of diuretics 12 (5-4) 59 (9 7)* 0 (0-0) 0 (0-0) Current smokers 79 (35-7) 112 (28-9) NA NA Alcohol intake 73 (33-0) 28 (7.2)* NA NA *P < 0-05; ***P < 0-001; sex differences. NA = not analysed. Geographical and temporal differences in the urinary excretion ofinorganic arsenic: a Belgian population study 325

of arsenic were available in 529 PheeCad par- especially near the old zinc smelter in the ticipants. In all these subjects the urinary Lommel area, which was dismantled in

arsenic excretion decreased (P < 0-001) from 1974.22 Occup Environ Med: first published as 10.1136/oem.53.5.320 on 1 May 1996. Downloaded from 222 (range 13-11718) nmol/24 h to 100 (2) During the CadmiBel study there was a (1-17882) nmol/24 h. Moreover, in each significant inverse association between urinary community there was a significant (P < 0.001) arsenic excretion and the distance from the drop in the urinary output of arsenic (fig 3). nearest smelter. (3) Relative to the three smelters, Lommel and Pelt are located along the direction of the Discussion prevailing winds, which originate from the Inorganic arsenic can easily enter the organism south west 40% of the time and the north west by inhalation or ingestion. Once absorbed it is 20% of the time. rapidly methylated to monomethyl and mainly (4) Arsenic may have been dispersed into dimethylarsenic derivatives which are excreted the environment from secondary sources. The in urine as MMA and DMA. More than half use of waste from plants producing zinc to of the inhaled or ingested dose is eliminated in build roads and heighten terrains in the past urine within a few days; the methylated may explain the moderately increased urinary metabolites constitute up to 80% of the excretion of arsenic even in areas which are excreted arsenic,18 but a fraction of the more remote from the smelters. Many of these absorbed inorganic arsenic is always excreted roads have been covered during recent years. unchanged.'9 For practical purposes, the sum (5) The increased urinary arsenic excretion of Ask, MMA, and DMA in urine can be con- in people living near non-ferrous metal plants sidered as an indicator of recent exposure are also in keeping with a previous study that (either by inhalation or by ingestion) to inor- showed that the arsenic absorption by school ganic arsenic.20 In subjects non-occupationally age children living around a lead smelter (situ- exposed to arsenic and living in areas not envi- ated near Antwerp) was higher than in their ronmentally contaminated by inorganic counterparts residing in a rural or urban con- arsenic the amount excreted in urine (As, + trol area. Contamination of the hands and MMA + DMA) is usually below 200 nmol/24 h possibly inhaled air were found to contribute (15 jig/day).6 to exposure.26 Hand to mouth activity has also During the CadmiBel study (1985-9), the been identified as a primary source of arsenic mean daily urinary excretion of arsenic absorption in children from a community sur- adjusted for sex, age, and body mass index, rounding a copper smelter in the United ranged from 91 to 155 nmol/24 h (6-8 to 11 .6 States.27 It is possible that this mode of trans- ,ug/24 h) in Liege, Charleroi, and Hechtel- fer (direct ingestion of arsenic contaminated Eksel. This agrees well with the result of a dust) also played a part in contamination of duplicate meal study (n = 122) that we had adults particularly in middle aged men who in previously carried out in three areas of rural areas are often engaged in physically Belgium (Brussels, Charleroi, and Liege); the demanding labour in the open air, such as gar- mean oral daily intake of arsenic amounted to dening and agriculture. http://oem.bmj.com/ 160 nmol/24 h (12 gg/24 h).2' (6) From baseline (1985-9) to follow up In the CadmiBel study the mean urinary (1991-5), the urinary excretion of arsenic in excretion of arsenic was significantly lower in Lommel and Pelt dropped almost fivefold. Liege than in Hechtel-Eksel and in Charleroi. These changes coincided with major improve- This may be due to the fact that the rural con- ments in the local industrial processes. Indeed, trol area (Hechtel-Eksel) is only 10 to 15 km before 1991 the zinc ore concentrates (con-

from where non-ferrous smelters were active taining about 80 mg As/kg) at the main zinc on October 1, 2021 by guest. Protected copyright. (Noorderkempen), whereas in the Charleroi smelter in Wezel, located upwind from area, steel furnaces probably emitted fly ash Lommel and Pelt, were stored and transported containing arsenic. in open air, whereas later these activities were The present study clearly showed that confined to closed hangars. In 1992, the pri- before 1991 men and women living in the mary zinc smelter in Pelt stopped its activity Noorderkempen, where non-ferrous smelters and only recycling of zinc products continued had been active since the beginning of this (Waeterschoot H, Union Miniere, Brussels, century, inhaled and ingested an amount of personal communication, 1995). Also other inorganic arsenic about twice as high as in the industries, which potentially released inor- other areas. Several arguments support the ganic arsenic into the environment, such as hypothesis that environmental contamination light bulb and glass manufacturing plants in due to emissions of arsenic by the smelters Lommel, had to comply with the tighter envi- may have contributed to the higher exposure ronmental regulations imposed in 1991. to arsenic in the Noorderkempen. The people living in Lommel and Pelt were (1) The highest urinary excretion of arsenic also informed around 1985 on how to reduce was found in the subjects living in the Lommel their environmental exposure to cadmium by and Pelt areas. The environmental pollution not consuming well water, by liming the soil of by cadmium in the Noorderkempen is well their kitchen gardens and by not growing leafy known and concomitant contamination by vegetables. However, the promotion of these other non-ferrous elements (Zn, Pb, As, etc) preventive measures was not coordinated has been found.722-25 Increased concentrations across communities and was also not sustained of arsenic (> 20 ppm) have been found in over time. Unlike cadmium, liming the soil some kitchen gardens in Lommel and Pelt, makes arsenic more mobile and available for 326 Buchet, Staessen, Roels, Lauwerys, Fagard

uptake by vegetables.28 These temporary mea- arsenic both through well water and through sures are therefore unlikely to have substan- dust particles. In the Noorderkempen area tially contributed to the decrease in the arsenic where the increased was exposure found, these Occup Environ Med: first published as 10.1136/oem.53.5.320 on 1 May 1996. Downloaded from exposure. layers are located at a depth of 100 m or more Another possible source of exposure to (De Smedt P, Flemish Water Distribution arsenic is through drinking water. According Company, Brussels, Belgium, personal com- to local water companies, tap water distributed munication, 1995). in the Noorderkempen and in Hechtel-Eksel is Men seem to be more exposed than women pumped from geological layers containing to inorganic arsenic. Smoking habit was not glauconite (iron arsenate) responsible for the related to the urinary arsenic excretion and water contamination by arsenic (Denteneer A, can therefore not account for these sex differ- Departement of the Environment and ences. Other factors related to lifestyle, such as Infrastructure, Ministry of the Flemish working in open air and dietary habits,29 or dif- Community, Brussels, Belgium, personal ferential toxicokinetics' of inorganic arsenic in communication, 1994). In Wezel inorganic men and women may underly these sex differ- arsenic had already been removed from the tap ences. These were however minor by compari- water before 1985, so that during the son with the influence of environmental CadmiBel as well as during the PheeCad stud- factors. ies the arsenic content of the water was very An official network monitoring the arsenic low, probably 13 nmol/l (1 yg/l) or less (Maes J, content in air and fall out dust was in opera- PIDPA, Antwerp, Belgium, personal commu- tion in the rural areas during the CadmiBel nication, 1995). In Hechtel-Eksel, Lommel, and PheeCad studies.'03' Measuring stations and Pelt the arsenic content of the drinking were posted around the non-ferrous smelters water was only reduced at a later stage. As a and in the residential areas of Lommel, Pelt, consequence, from 1985 to 1989, the arsenic and Hechtel-Eksel. Inorganic arsenic in air- concentration in pipe water had varied from borne dust was measured daily. These daily 520 to 707 nmol/l (39 to 53 gg/1) in Lommel measurements were then computerised to and from 107 to 307 nmol/l (8 to 23 jug/l) in obtain for each year the arithmetic mean, the Pelt. From 1991 onwards, these ranges 98th percentile, and the maximum. However embraced 13 and 67 nmol/l (1 and 5 jug/1) and the measurements of the arsenic concentration 27 and 53 nmol/l (2 and 4 jg/l), respectively in airborne dust were incomplete, because (Denteneer A, personal communication, they were not performed in Hechtel-Eksel and 1994). In spite of these regional and temporal also not in Lommel and Pelt from 1991 to differences in the arsenic content of drinking 1993. Moreover, only airborne particles with water, the higher urinary excretion of arsenic an aerodynamic diameter of 10 ,um or less in subjects living in the Noorderkempen were sampled. Larger particles can also be region before 1990 cannot possibly be inhaled and ingested with sputum coughed up explained by the consumption of tap water but their arsenic content was not monitored. alone. Indeed, Pelt and Hechtel-Eksel were Also, the measurements in fall out dust per- supplied from the same wells and the drinking formed at the stations of Hechtel-Eksel, http://oem.bmj.com/ water was mixed before it was distributed Wezel, Lommel, and Pelt did not include (Keustermans L, Flemish Water Distribution arsenic.'2 Company, , Belgium, personal com- From the data made available, it seems that munication, 1995). Yet, the urinary arsenic in the early 1980s the admissible thresholds excretion was significantly higher in Pelt (373 for airborne arsenic" (annual mean 0-02 nmol/24 h, 28 ug/24 h) than in Hechtel-Eksel ,ug/m'; P98 0-08 ug/m', and P100 022,g/m3) (144 nmol/24 h, 11 gg/24 h). These findings have been exceeded at regular intervals, con- on October 1, 2021 by guest. Protected copyright. strongly suggest the presence of another envi- firming the hypothesis of industrial pollution. ronmental source of inorganic arsenic in Pelt. More importantly, the official monitoring net- The differences in the arsenic content of the work did not detect the increased exposure of drinking water between Hechtel-Eksel and the population in Lommel and Pelt between Wezel during the CadmiBel era (1985-9) were 1985 and 1989, either as a consequence of also insufficient to lead to a measurable con- missing measurements or due to lack of repre- trast in the urinary arsenic output, whereas, sentitiveness of the measurements. The detec- after imposition of stricter environmental reg- tion limit of arsenic in airborne dust was ulations in 1991 and the reduction of the reportedly about 0 13 nmol/m3 (0.01 ug/m3).3" industrial emissions, a difference did emerge The present findings highlight the necessity between Hechtel-Eksel and Wezel (fig 2). to validate measurements in the environment, Water from private wells could also have for instance by directly estimating the internal been a possible source of arsenic exposure in exposure of the concerned populations. Such a the Noorderkempen. However, all houses recommendation would certainly apply to included in the CadmiBel study were supplied areas at risk of an increased exposure to toxic with tap water. In the Noorderkempen area substances, be it from natural or industrial only 18 subjects lived in houses which also had sources. a private well, whereas in Hechtel-Eksel this A prospective follow up study of the mor- number was 102. The geological layers con- bidity and mortality in the Noorderkempen taining glauconite surface locally in Hechtel- could also be initiated to find whether the Eksel (Ide G, Lisec Research Centre for increased exposure to inorganic arsenic in the Ecology and Forestry, , Belgium, 1990) past has led to adverse health effects. Indeed and could have constituted a natural source of on the basis of the risk assessment model for Geographical and temporal differences in the urinary excretion ofinorganic arsenic: a Belgian population study 327

arsenic developed by the Environmental Kreatininbestimmung. Clin Chim Acta 1975;62:333-9. 17 Persyn JP, van der Silk W. A new method for the determi- Protection Agency34 and the toxicokinetics of nation of gammaglutamyl transferase in serum. J Clin after exposure through the Chem Clin Biochem 1976;14:421- 7. arsenic excretion Occup Environ Med: first published as 10.1136/oem.53.5.320 on 1 May 1996. Downloaded from 18 Buchet JP, Lauwerys R, Roels H. Urinary excretion of inor- oral or pulmonary route, one may hypothesise ganic arsenic and its metabolites after repeated ingestion of that the increased exposure in Noorder- sodium metaarsenite by volunteers. Arch Occup Environ Health 1981;48:111-8. kempen during the 1980s may have been asso- 19 Buchet JP, Lauwerys R, Roels H. Comparison of the uri- ciated with an increased cancer risk. nary excretion of arsenic metabolites after a single dose of sodium arsenite, monomethylarsonate or dimethylarsi- nate in man. Int Arch Occup Environ Health 1981;48: The population surveys have been supported by the 71-9. International Lead Zinc Research Organisation (Research 20 Offergelt JA, Roels H, Buchet JP, Boeckx M, Lauwerys R. Triangle Park, North Carolina, USA), the Province of Relation between airborne arsenic trioxide and urinary (Hasselt, Belgium), the municipalities Hechtel-Eksel and excretion of inorganic arsenic and its methylated metabo- Lommel (Belgium), the National Fund for Medical Research, lites. BrJ7 Ind Med 1992;49:387-98. the Incentive Program "Health Hazards" (Science Policy 21 Buchet JP, Lauwerys R, Vandevoorde A, Pycke JM. Oral Office, Prime Minister's Services), the Ministry of the Flemish daily intake of cadmium, lead, manganese, copper, Community (Department of Labour and Social Affairs), the chromium, mercury, calcium, zinc, and arsenic in Health and Safety Directorate of the Commission of the Belgium: a duplicate meal study. Food and Chemical European Communities, Astra Pharmaceuticals, and Bristol- Toxicology 1983;21:19-24. Meyers-Squibb (all in Brussels, Belgium). 22 Werkgroup Zware Metalen in de Kempen. Inventariserend onderzoek inzake de verontreiniging met zware metalen in de 1 TSDR Agency for Toxic Substances and Disease Registry. Kempen. Genk, Belgium: LISEC Studiecentrum voor Toxicological profile for arsenic. Atlanta, GA: US Depart- Ecologie en Bosbouw, 1985. (In Dutch.) ment of Health and Human Services, 1993. 23 Vangronsveld J, Van Assche F, Clijsters H. Reclamation of a 2 Kasper ML, Schoenfield L, Strom RL, et al. Hepatic base industrial area contaminated by non-ferrous metal: angiosarcoma and bronchioloalveolar carcinoma induced in situ metal immobilization and revegetation. Environ by Fowler's solution. JAMA 1984;252:3407-8. Pollution 1995;87:51-9. 3 Chen CJ, Wang CJ. Ecological correlation between arsenic 24 Vangronsveld J, Cammaer C, Van Assche F, Clijsters H. level in well water and age-adjusted mortality from malig- Reclamation of a "desert like" site in the north east of nant neoplasms. Cancer Res 1990;50:5470-4. Belgium: evolution of the metal pollution and experi- 4 Pershagen G. Lung cancer mortality among men living ments in situ. In: Farmer JG, ed. 8th International near an arsenic-emitting smelter. Am J Epidemiol 1985; Conference on Heavy Metals in the Environment. 122:684-94. Edinburgh: CEP Consultants, 1991:58-61. 5 Lauwerys R, Bernard A, Buchet JP, et al. Does environ- 25 Vangronsveld J, Carleer R, Clijsters H. Transfer of metals mental exposure to cadmium represent a health risk? and metalloids from soil to man through vegetables culti- Conclusions from the CadmiBel study. Acta Clin BeIg vated in polluted gardens: risk assessment and methods 1991;46:219-25. for immobilization of these elements in soils. In: 6 Buchet JP, Lauwerys R, Roels H. Comparison of several Varnavas SP, ed. 6th Int Conference on Environmental methods for the determination of arsenic compounds in Contamination. Edinburgh UK, CEP Consultants. 1994: water and in urine. Int Arch Occup Environ Health 1980; 142-5. 46:11-29. 26 Buchet JP, Roels H, Lauwerys R, et al. Repeated surveil- 7 Lauwerys R, Amery A, Bernard A, Bruaux P, Buchet JP, lance of exposure to cadmium, manganese, and arsenic in Claeys F, et al. Health effects of environmental exposure school-age children living in rural, urban, and non- to cadmium: objective, design and organization of the ferrous smelter areas in Belgium. Environ Res 1980;22: CadmiBel study: a cross-sectional morbidity study car- 95-108. ried out in Belgium from 1985 to 1989. Environ Health 27 Plissar L, Lowry-Coble K, Kalman DA, Hughes JP, van Perspect 1990;87:283-9. Belle G, Covert DS, et al. Pathways of human exposure 8 Staessen J, Bulpitt CJ, Fagard R, Joossens JV, Lijnen P, to arsenic in a community surrounding a copper smelter. Amery A. Salt intake and blood pressure in the general Environ Res 1990;53:29-47. population: a controlled intervention trial in two towns. Jf 28 Campbell JA, Stark JH, Carlton-Smith CH. Effects of Hypertens 1988;6:965-73. arsenic in sludge applied to soil. In: Lekkas TD, ed. 9 Staessen JA, Buchet JP, Lauwerys RR, Lijnen P, Roels H, 5th International Conference on Heavy Metals in the Fagard R on behalf of the Working Groups. Public impli- Environment. Edinburgh: CEP Consultants, 1985;1: cations of environmental exposure to cadmium and lead: 478-80. an overview of epidemiological studies in Belgium. _J 29 Buchet JP, Lauwerys R. Fish consumption and exposure to

Cardiovasc Risk 1995; (in press). arsenic. State of the art. Arch Public Health 1994;52: http://oem.bmj.com/ 10 Staessen JA, Birkenhkger W, Bulpitt CJ, et al. The relation- 171-4. (In French.) ship between blood pressure and sodium and potassium 30 Instituut voor Hygiene en Epidemiologie Stations Zware excretion during the day and at night. J7 Hypertens 1993; Metalen. Evaluatie van de gehalten aan zware metalen in 11:443-7. de omgevingslucht in Belgie. Jaarrapport. Ministerie van 11 Staessen JA, Fagard R, Amery A. Life style as a determi- Volksgezondheid en Leefmilieu. Brussel: IHE, 1992. nant of blood pressure in the general population. Am _J 31 Arseengehalte in de Noorderkempen. Erembodegem, Hypertens 1994;7:685-94. Belgium: Vlaamse Milieumaatschappij, 1994. (In Dutch.) 12 Office of Population Censuses and Surveys. Classification of 32 De neerslagkruiken in Vlaanderen. Overzichtsrapport occupations. London: OPCS, 1980. 0110411986-3110311992. Erembodegem, Belgium: 13 Yamada Y, Ishizaki M, Kido T, et al. Alcohol, high blood Bestuur Meemetten en Onderzoek, Dienst Milieurap- and serum yglutamyltranspeptidase level. portering, Vlaamse Milieumaatscappij, 1994. (In Dutch.) pressure, on October 1, 2021 by guest. Protected copyright. Hypertension 1991;18:819-26. 33 Evaluatie van de gehalten aan zware metalen in de omgev- 14 Buchet JP, Lauwerys R. Evaluation of exposure to inor- ingslucht in Vlaanderen. Dertiende Jaarrapport ganic arsenic in man. In: Facchetti S, ed. Analytical tech- 0110411992-3110311993. Erembodegem, Belgium: niques for heavy metals in biological fluids. Amsterdam: Bestuur Meetnetten en Onderzoek, Dienst Milieurap- Elsevier, 1981:75-90. portering, Vlaamse Milieumaatschappij, 1994. (In 15 Buchet JP, Pauwels J, Lauwerys R. Assessment of exposure Dutch) to inorganic arsenic following ingestion of marine organ- 34 Health assessment documentfor inorganic arsenic. Final report. isms by volunteers. Environ Res 1994;66:44-51. Research Triangle Park, North Carolina: US Environ- 16 Bartels H, Bohmer M. Eine Mikromethode zur mental Protection Agency, 1984.(EPA-600/8- 83-021F.)