m concentrations of selected organochlorines among delivering women in Argentina . The EMASAR study †

Department of Community Medicine, Fac University of Norway, Box 6050 Langne solrunn.hansen@uit.no; jon.øyvind.odland@ Department of Biochemistry and Biom Hamilton, ON, Canada. E-mail: nieboere@m Institute of Environmental Assessment andW of Environmental Chemistry, Jordi Girona, E-mail: natalia.bravo@idaea.csic.es; joan.gr Department of Obstetrics and Gynecology, 8100, N-4068 Stavanger, Norway. E-mail: in Banco de Sangre, Cĺınica San Jorge, Onacha Argentina. E-mail: matiocevichsilvina@hotm Hospital Público Materno Infantil de Salta E-mail: marisavivianaalvarez@gmail.com Department of Research, Stavanger Unive Stavanger, Norway. E-mail: stein.tore.nilsen † Electronic supplementary information (ES Cite this: DOI: 10.1039/c7em00278e


Introduction
Most OCs ‡ were synthesized and used in the Northern Hemisphere. 1,2 Furthermore, the Arctic has become a repository for contaminants due to what is referred to as the latitude effect that involves atmospheric transportation and ocean currents. 3 As a consequence, most studies of the body burdens of OCs in humans have been conducted among Northern Hemisphere populations. 3 Comparable information for southern-latitude countries is limited. The Southern American continent constitutes a potential sink for OCs from global sources. 4 Environmental studies of air, sediments, soil, water, wildlife and biota have indeed regarded their occurrence there as a consequence a Department of Community Medicine, Faculty of Health Sciences, UiT The Arctic University of Norway, Box 6050 Langnes, NO-9037 Tromsø, Norway. E-mail: solrunn.hansen@uit.no; jon.øyvind.odland@uit.no; Tel: +47 77644836 of both global transport and past local uses. [5][6][7][8][9] DDT and HCHs have been used in agriculture, while PCBs have been linked to inputs from power plants, industrial activities, combustion and dumping. 5,6,9 Not surprisingly, these pollutants have entered the domestic food web. 4,6,10,11 Argentina ratied the Stockholm Convention that banned DDT, HCB and HCHs in agriculture uses since the late 1990s, as well as placing restrictions or bans on PCBs in 2001 (ref. 4) and more recently on the use of the insecticide dicofol. 4,9,10 OCs are persistent, lipophilic, toxic and bioaccumulate in the food chain. 3 The concentration of contaminants in maternal blood and breast milk provides important information on the exposure of the fetus and newborn, who are especially vulnerable. Endocrine disruption, reproductive effects, and impairment of immunologic development constitute worries and their adverse toxic effects may threaten human health (including trans-generational impairments). 3 Knowledge about human exposure to OCs in South America is limited. 12 The current study describes the body burden of OCs in Argentinian delivering women, and our ndings are assessed in the context of comparable studies elsewhere. Our working hypothesis is that a phenomenon similar to the Arctic latitudetransport effect is not operative in the Southern Hemisphere.

Description of the study area and data collection
The EMASAR study was conducted in the Argentinian cities of Ushuaia and Salta (for map see Fig. 1 in Økland et al. 2017). 13 Ushuaia (54.80 S, 68.30 W) is the southernmost city in the world and the capital of the Tierra del Fuego in Antártida e Islas del Atlántico Sur Province. It has a population of some 60 000 and features electronics manufacturing, sheries, natural gas and oil extraction, sheep farming, and tourism. Salta (24.78 S, 65.42 S) is the capital of the northwestern-highland Salta province. The city has around 620 000 citizens, with agriculture and related industrial activities being the main economic activities (for additional details see Økland et al., 2017). 13 The eld work was conducted in the period April 2011 to March 2012. Of the total 698 delivering women enrolled, the current study component is limited to 670 subjects (199 from Ushuaia and 471 from Salta). One person from Ushuaia and 27 from Salta were excluded due to the lack of serum lipid concentrations. Through interviews, participants completed questionnaires covering personal characteristics, socioeconomic factors, obstetrical and breastfeeding history, environmental, health and lifestyle conditions, and dietary intake. Nonfasting maternal blood samples, height and weight were obtained at 36 AE 12 hours following delivery (median 1 day, range 0-3 days). Details about the study prole, population anthropometric measurements and blood sampling procedures have been provided previously. 13 EMASAR is a collaborative project between UiT The Arctic University of Norway, the Stavanger University Hospital (Norway) and the two Argentinian partners, namely Clínica San Jorge in Ushuaia (a private institution co-responsible with a public hospital for the in-hospital deliveries in the city and the surrounding region), and the Hospital Público Materno Infantil Fig. 1 Age-and parity-adjusted maternal serum OC concentrations (mg kg À1 lipid, geomean) with 95% CI error bars, stratified by residence (Ushuaia and Salta) and maternal birthplace (native or migrated) for compounds with a detection frequency > 60%. City differences (Ushuaia versus Salta) at p < 0.001 except for PCB 153 (p ¼ 0.081). The p-values for differences between native and migrated groups are provided in Table  2; for full names of the compounds, see Section 2.2.
in Salta (a public institution responsible for all in-hospital deliveries in the city and the region).
The study (#2010/7317) was approved by the Ethics Committee of the Salta Medical Association and the Ministries of Health in both provinces. As required by law, the Norwegian Regional Committee for Medical and Health Research Ethics (REC North) approved the study (#2011/706), and it was conducted in accordance with the Helsinki declaration. Informed consent was obtained for any experimentation with human subjects in the study.

Chemical analysis
Blood collected in BD Vacutainers® (BD SST II Plus Advance 10/ 8.5 ml) was sampled and centrifuged at 2000 relative centrifugal force (RCF) for 10 minutes. The maternal serum was then transferred into vials (Sarstedt CryoPure, 2.0 ml tubes) for general analyses, or into glass vials (4.0 ml, pre-rinsed with nhexane/acetone) retained for chemical analyses of the POPs. The serum samples were transferred in a frozen state to the EMASAR Biobank at the Faculty of Health Sciences, UiT The Arctic University of Norway, Tromsø, Norway.

Statistical analysis
Statistical analyses were carried out using the IBM SPSS Statistics for Windows statistical package version 24 (SPSS Inc. Chicago, IL, USA). Descriptive details are reported as means, standard deviations, median and range or percentage, and relationships between variables were explored using Spearman's rank correlation analysis. Concentrations below the LOD were replaced by LOD/2. Our choice to limit the between-group statistical analyses to compounds with a prevalence > 60% is recommended by the US CDC. 17 Despite log 10 transformation, the OC distributions remained skewed positively to the right and were not normally distributed according to the Kolmogorov-Smirnov test and Q-Q residual plots. We therefore employed non-parametric statistics (the Mann-Whitney U test) for comparisons of OC concentrations between the two study sites. For related compounds, Levene's test demonstrated equal variances between the two study sites. In a univariate general linear model (UGML) adjusted for the continuous variables age and parity (excluding stillbirths), acceptable case-wise diagnostics were possible even though abruptions of normality were   View Article Online the group size, and thus the distribution, bootstrapping with 2000 estimates was selected. Between city comparisons (Ushuaia versus Salta) and maternal birthplaces involved four groups: (i) born and living in Ushuaia (natives, n ¼ 61); (ii) living in Ushuaia, but born in another province within Argentina (migrated, n ¼ 128); (iii) living and born in Salta (natives, n ¼ 409); and (iv), living in Salta but born in another province within Argentina (n ¼ 38). In the regression model, the coefficients for the OCs were estimates of the mean difference in logtransformed concentrations between each group (city or heritage) and the reference level. The regression coefficients were back-transformed (10 b ) to reect the ratios of change in concentrations, and in the text are described as % change ¼ (10 b À 1) Â 100. The signicant levels were set at p < 0.05. Finally, the ratios within the DDT group were calculated to distinguish ongoing chronic exposure versus dietary exposure or distant past exposure. 3,18

Personal characteristics
The personal characteristics of all participating women (n ¼ 698) are described elsewhere 13 and include details about mothers (e.g., age, marital status, education, employment, smoking habits, diet, and vitamin intake) and newborns. As mentioned, 28  analyzed, only six compounds had a detection rate above 60% in both places, specically PCB 118, PCB 138 and PCB 153, p,p 0 -DDT, p,p 0 -DDE and b-HCH. Five additional compounds exceeded this threshold in Ushuaia, namely p,p 0 -DDD, o,p 0 -DDT, o,p 0 -DDD, HCB, and a-HCH. We selected p,p 0 -DDE, PCB 138, PCB 153 and p,p 0 -DDT which exhibited detection frequencies (>89%) for detailed evaluations, and a somewhat more inclusive group when considering maternal birthplaces (Table S2 † (Table 1) followed by PCB 153 (respectively, lower by factors of 4.3 in Ushuaia and 9.9 in Salta based on GMs). Compared to Ushuaia, serum concentrations of PCB 118, b-HCH, p,p 0 -DDT, p,p 0 -DDE and P DDT were all higher (p < 0.003) in Salta, while there were no city differences for PCB 153 and 138. Wet-weight serum OC concentrations are summarized in Table S3. † 3.2.3 Impact of residence and place of birth on age-and parity-adjusted OC concentrations. The age-and parityadjusted concentrations of the dominant OCs (ng per lipids) by area of residence are shown in Fig. 1 and are further stratied whether born in the two study areas or migration into them from other regions of Argentina (n ¼ 636). Through bootstrapping, differences in OC concentrations between the Ushuaia and Salta women were shown to be distinct (p < 0.001), with the exception of PCB 153 (p ¼ 0.081).
Several signicant statistical differences (p # 0.010) were observed when considering residence history in terms of being native born or migrants to the two regions of Argentina (Table 2). Relative to Salta natives, those of Ushuaia had 27% lower concentrations of PCB 138 and 32% of PCB 153; and migrants into Ushuaia also exhibited reduced concentrations of PCB 118 and PCB 138 (respectively, by 47% and 25%), but not so for PCB 153. In addition, Ushuaian migrants had 46% less serum PCB 118 than those in Salta. The p,p 0 -DDE and p,p 0 -DDT concentrations for Ushuaian natives and migrants were less substantive by 60% and 70% when compared to the respective groups in Salta (Table 3). Ushuaia migrants had 58% lower concentrations of b-HCH compared to Saltanean natives (post hoc analysis). Intra-city differences in OC concentrations between native and migrated women in Salta did not reach statistical signicance; nor for Ushuaians, with the exception that PCB concentrations in native citizens were on average 29% lower (Table 2; also see Table S4 †). Age had a positive inuence on OC concentrations and parity a negative one (p < 0.05). The statistical signicance for these variables was lost for Ushuaia residents when examining each community separately (e.g., Table S5 †), but persisted for all OCs in Salta (data not shown).

Ratios of DDT-isomers
The p,p 0 -DDE/p,p 0 -DDT ratio was rather similar in Ushuaia (mean 22.8, median 11.4) and Salta (mean 19.2, median 11.6) women (p > 0.05, Table S4 †). In both places, around 18% of the women had a ratio below 5.0. By comparison, the respective median ratios (0.09) of p,p 0 -DDT/p,p 0 -DDE were identical (p > 0.05); three women in Ushuaia and ve in Salta exceeded a ratio of 1.00. In addition, the mean o,p 0 -DDT/p,p 0 -DDT ratio was calculated only for samples with concentrations above the LOD (n ¼ 183) and had values of 1.09 (Ushuaia) and 0.29 (Salta) with p < 0.001.

Correlations of OC concentrations
Based on Spearman's rho tests, signicant inter-OC correlations might be designated as weak (r < 0.4) to moderate (r 0.4-0.9) for the 5 OCs compared in Salta ( Table 3). The most robust relationships occurred between PCB congeners (r ¼ 0.59-0.90). For Ushuaia, the intra-PCB relationships were again the most robust (r ¼ 0.60-0.76); others were mostly weak or absent.

Global comparisons
Our OC concentration data are summarized and compared to those reported for other regions of the world in Fig. 2-4; the corresponding world-wide data are compiled in Tables S6a-e  and in Tables S6f & g †

Preamble
In addition to comparing our serum OC concentrations with those reported for other regions of the world, the inuences of the following predictor variables are discussed below: maternal age, parity, breastfeeding history, province lived in and of birth (specically, within the provinces of Ushuaia and Salta or migration into them from other Argentinian provinces), as well as dietary and environmental predictor variables.

4.2
The signature of the OCs in a global and regional comparison 4.2.1 DDT-group. In general, the observed maternal lipidadjusted serum concentrations of p,p 0 -DDE in Salta may be designated as moderate when compared with those reported for other world areas (Fig. 2). By comparison, concentrations for Ushuaian mothers were in the lower range and comparable to those reported for Arctic populations not consuming marine mammals, 3 although somewhat higher than the lowest observed such as for the residents of the coast of non-malaria South Africa. 19 The concentrations for Saltanean mothers were comparable to those reported for Alaska Yup'ik and for Baffin Island and Canadian Inuits living in Inuvik, 3 although higher than in Korea 20 and central Canadian cities 21and both above and below those reported for the Caribbean islands. 22 The Argentinian maternal levels were moderately lower than in Western Australia (wet weight, wwt), 23 central Europe like Spain 24 and Italy, 25 and considerably lower than in Bolivia, 26 China, 27 and Arctic women with a traditional diet of marine mammals. 3 By contrast, the observed concentrations were considerably lower than in countries with historical use of DDT and subsequent prohibition such as Mexico 28 and Peru, 17 Vietnam (wwt) 29 and Belize. 22 Malaria-endemic areas of South Africa with active DDT treatments have reported the highest concentrations in contrast to the low levels in the neighboring nonmalaria Indian Ocean region. 19 The Argentinian p,p 0 -DDT concentrations observed (Fig. 2) were lower than in countries with a previous or current history of DDT usage. 17,19,22,27,29 Comparable p,p 0 -DDT levels to those in Salta have been observed in Mexico, 28 Northern Greenland and Faroe Island 3 and the non-malaria Indian Ocean area in South Africa. 19 The Ushuaian levels were slightly higher than those found in Alaska and Iceland. 3 Finally, our observed concentrations of DDT-related compounds were higher (p,p 0 -DDE), comparable (p,p 0 -DDT) or lower (p,p 0 -DDD) than those reported for an adult population of Buenos Aires in 2006. 12 Relative to Ushuaia, the higher concentrations of p,p 0 -DDT (and thus also p,p 0 -DDE) in Salta are in agreement with the nearly ten-fold higher use of insecticides in the home and perhaps in agriculture. 13 Several decades ago this compound was used for malaria vector control in the northwestern regions. 4 The current presence of low levels p,p 0 -DDT and the relatively high concentrations of its derivative p,p 0 -DDE point to past comprehensive use and ongoing entry into the local food web. 4,6,7,9 The observed p,p 0 -DDE/p,p 0 -DDT ratio of around 20 in the two communities (Table S4 †) is consistent with historical use of the latter. 18 Nevertheless, chronic or more recent exposure to DDT cannot be excluded as one in ve women in both communities had ratios below 5. The o,p 0 -DDT/p,p 0 -DDT ratio is a useful index in regions with active use of dicofol. 30 In Ushuaia, the relatively high o,p 0 -DDT/p,p 0 -DDT ratio of 1.09 compared to 0.29 in Salta (Table S4 †) suggests some recent DDT input, 18 and is likely related to dicofol use. [30][31][32] Atmospheric release of dicofol in the province of Mendoza in western-central Argentina has also been reported, 9 and this insecticide still appears to be in use. 10 Furthermore, long-range transport of the relatively volatile Dicofol from neighboring regions and countries, or other southern continents, might also have some inuence. 33,34 4.2.2 HCB and the HCH-group. The observed concentrations of HCB and b-HCH (Table 1) were mostly in the low range compared to other jurisdictions (Fig. 3). In Ushuaia, the concentrations of both compounds were comparable to those for the women residents of Korea 20 and HCB concentrations were comparable to those reported in Northern Norway. 35 Serum HCB concentrations for mothers in Italy, 25 Spain 24 and remote Arctic regions 3 were up to two-fold higher, and up to sixfold in China. 27 Before the use of pure lindane (g-HCH) technical mixtures were used, which encompassed isomeric compositions of 60-70% a-HCH, 5-12% b-HCH and 10-15% g-HCH. 36 Generally speaking, data on human maternal body burdens of a-HCH are scarce and are low or below the detection limit for mothers in China, 27 Australia, 23 Canada and Mexico. 28 Our Argentinian levels (Table 1), particularly those in Ushuaia, are comparable in magnitude to those reported in South Africa (2 mg kg À1 lipid) 37 and China (5.2 mg kg À1 lipid). 38 As depicted in Fig. 3, the observed b-HCH concentrations were higher than those from Canada, Greenland and Alaska, 3 comparable to those from Mexico, 28 Korea 20 and South Africa, 39 but lower than those reported for the Faroe Island 3 and Spain. 24 The high concentrations observed in China are consistent with its extensive organochlorine pesticide production and use. 27 The generally low b-HCH concentrations among Arctic populations contrast those of other OCs, and perhaps reect decades of worldwide restrictions and prohibitions.
The observed HCB, a-HCH and b-HCH concentrations likely reect regional diversities in past emissions. HCB is still released into the Argentinian environment as a byproduct of pesticide use, various industrial activities, open burning processes, waste disposal and landlls. 9 The 3-fold higher a-HCH levels in Ushuaia, and its lack of a relationship with HCB, PCBs, p,p 0 -DDT and p,p 0 -DDE (Table 3) may reect source-specic pathways. It is known that a-HCH is retained less in nature and humans than b-HCH. 36 Their comparable concentrations in Ushuaia mothers remain unexplained, as is the observation in China that a-HCH levels exceeded those of b-HCH. 38 Atmospheric long-range transportation of this semi-volatile HCH to colder climates may be relevant. Furthermore, the relatively higher exposure to the waterinsoluble b-HCH in Salta suggests greater historical local usage of HCH pesticides compared with more central and southern remote areas. The positive correlations of b-HCH at both sites with both PCBs and p,p 0 -DDE (see Table 3) suggest common sources of exposure, while the scarcity of g-HCH is consistent with its lower bioaccumulation potential and more rapid degradation in nature. 36 By contrast, a study in South Africa showed almost non-detectable maternal concentrations of HCB, a-HCH and b-HCH and a complete predominance of g-HCHthis was explained by on-going use of lindane. 37 4.2.3 PCB-congeners. PCB congeners 153 and 138 were the dominant PCBs, and this is like the footprint observed elsewhere. Furthermore, the observed PCB 118/PCB 180 ratio of 2-3 contrasts those observed elsewhere, such as in the Arctic (though with some Russian observations exempted), 3 but is in line with previous studies within Argentina. 6 Presumably this ratio reects local use of technical mixtures with congener-specic composition. 6,11 Despite prohibitions, Argentinian electric power-generation equipment, aged industrial plants, stockpiles and the other sources already mentioned continue to be sources of PCBs and continue to impact the environment and thus the food chain. 9,11 However from a global perspective, the Argentinian production and use of PCBs have been low. It is estimated that only 3% of the total global historical use of PCB occurred in the Southern Hemisphere, with Argentina contributing 0.1%. 40 This is consistent with the data for PCB 153 in Fig. 4.

Factors inuencing OC concentrations
Generally speaking, we conclude that the exposures in Salta for the dominant OCs exceeded those in Ushuaia, with PCB 153 exempted. Taken together, the relatively robust inter-correlation patterns for PCBs, p,p 0 -DDE and b-HCH in both communities (Table 3) suggest more or less a common exposure pathway. However, in Ushuaia they are somewhat more limited for a-HCH, p,p 0 -DDD and o,p 0 -residues and suggest additional sources. The impact of inter-country migration was not strong, but nevertheless measurable. Ushuaian migrants had considerably lower serum PCB 118, p,p 0 -DDE and p,p 0 -DDT than those who had moved into Salta. Interestingly, PCB-153 appears to have contributed considerably to the overall population body burden of PCBs in Ushuaia. Geographic differences in OC levels as observed likely also reect history, socioeconomic, culture, physiological factors, lifestyle and dietary differences. 22,24,[41][42][43] Age as a positive predictor of serum OC concentrations and parity and previous breastfeeding as negative ones are well understood. 35,44 Because of the relatively long half-lives (in years) of OCs (typically > 7 years), 45 body burdens increase over time. OC storage in lipid tissues and loss via breastfeeding explain the observed negative impact of parity.
In Argentina, dietary foodstuffs are typically produced within the nation, but with regional differences. Extensive importation of foods into Ushuaia suggests an additional contrasting factor. OCs are oen associated with animal lipid-rich aquatic and terrestrial food web sources, including dairy products and eggs, or even fruit, vegetables and grains. 35,41,46 As indicated by others 6 and conrmed in our recent publication, 13 sh contributed only 10-15% of the Argentinian diet. Argentinian studies have also suggested that freshwater species contain OCs due to the pollution of rivers. 6,11 Consumption of fatty meats and freshwater sh has been identi-ed as a dietary predictor variable of OCs in Argentinian women. 47 Such sources would likely affect the inland residents of Salta more.
Salta province in the north has a subtropical climate, while Tierra del Fuego is at a lower latitude and has a cooler climate. Its prevailing winds are from the south-west, and thus a latitude effect equivalent to that operative in the Arctic region seems unlikely. However, Ushuaia's expanding industry, economy, population and tourism constitute environmental challenges that include heavily polluted waterfronts and limited wastedisposal systems. 8

Strengths and limitations
The present study of OCs in two distinct provinces of Argentina uses the methodological and analytical approaches employed in AMAP studies and our laboratory participated successfully in the most recent AMAP Analytical Ring Test. 15 Nearly all invited women participated, and the sample sizes for the two regions are comparable to those for other population monitoring studies. We consider the representativeness of our sample set as acceptable. Our ndings would facilitate the planning of other surveys of other regions of Argentina, as its generalizability might best be limited to the investigated regions.
Low detection frequencies for some compounds might have introduced some bias into the statistical analyses. A few cases of extreme values of DDTs were also found, but sensitivity analyses supported their inclusion. We acknowledge that our comparisons of our data with those of multiple studies are subject to uncertainties related to varying methodological and analytical variances. However, most of the publications referred to employed the AMAP study approach, and the respective laboratories have participated in the mentioned Inter-laboratory Ring test. 3 Our use of lipid-adjusted concentrations in our comparisons might have helped to minimize bias such as that associated with different pregnancy sampling periods. 18,44

Conclusions
To the best of our knowledge, the EMASAR study is the rst conducted in Argentina on OC concentrations in the sera of women related to pregnancy. Overall, our results indicate that the serum OC concentrations in Argentinian delivering women were in the lower range. Even so, the geographical distributions of the DDTs, HCB, HCHs and PCBs for the Argentinian southernmost and northwest mothers are inferred to reect differences in domestic sources, regional diversity in historical and current uses, prior and current industrial emissions, and potential contributions of disposal sites, stockpiles and longrange transfer. Due to the wide range of pesticide use, some individuals are deemed to be at higher risk. Compared to other countries, the natives of Ushuaia in the Tierra del Fuego province are considered to be subject to lower levels of POPs than in many other countries. By comparison, the exposure of the Salta group is considered to be moderately elevated. The relatively high PCB 118/PCB 180 ratio observed for both Argentinian communities contrasts that seen elsewhere and likely reects the use of technical mixtures with congener-specic composition.

Funding
We acknowledge The Norwegian Department of Foreign Affairs and the Arctic Monitoring and Assessment Programme (AMAP) (2011/706-13) for generously funding the study.

Conflicts of interest
There are no conicts to declare.