Fish consumption habits of pregnant women in Itaituba, Tapajós River basin, Brazil: risks of mercury contamination as assessed by measuring total mercury in highly consumed piscivore fish species and in hair of pregnant women

The municipality of Itaituba (04°16’34.0” S, 55°59’01.0” W) is situated in the Tapajós River basin in the southwestern region of the state of Pará, Brazil, with a population of over 98 thousand and territory of over 62 million km² (41) (Figure 1). It is the reference centre in public health care for patients referred from other municipalities (Aveiro, Jacareacanga, Novo Progresso, Rurópolis and Trairão) and receives patients from riverbank communities, countryside, forest, and gold mining areas such as those marked in Figure 1.

Figure 1

This study included 110 pregnant women (aged 18–40; mean age 24.6±4.8 years), admitted to the obstetric clinic at the Municipal Hospital of Itaituba (state of Pará, Brazil) to give childbirth. For the purpose of this study, we designed a specific questionnaire to collect their socio-demographic data (including residence, age, education, marital status, race, and occupation), health status, weight, height, blood pressure, smoking habit, diet, alcohol use, and fruit consumption. We also collected information about their fish consumption, including the species and weekly estimates of meals (g). To help them estimate consumption per meal, we showed them a silicon (50 and 100 g) fish filet very similar to the natural fish filet.

All participants were informed about all study details (duration, objectives, method, risks, and benefits) and signed the informed consent form before they entered the study. The study followed the principles described in the Declaration of Helsinki and was approved by the Research Ethics Committee of the University of the State of Pará (UEPA), Santarém, Brazil (approval No.: CAAE 94880318.9.0000.5168 on 11/21/2018).

The participants also gave hair samples (n=110), which were taken with sterile scissors as close as possible to the scalp in the occipital region to minimise aesthetic changes. Each sample was placed in a paper envelope until analysis. Considering that hair grows 1 cm a month on average, we took only four centimetres off the tip of the hair for THg analysis to obtain an average Hg concentration in the last four months of pregnancy.

To remove impurities, hair samples were first washed with the 0.1 %, (v/v) Triton^TM X-100 solution (Sigma-Aldrich, St. Louis, MO, USA) and then with ultra-pure water. Excess water was rinsed with acetone (Sigma-Aldrich) and samples dried at room temperature. Each sample was then cut into very thin pieces with sterile scissors for better homogenisation.

Cichla spp. and P. squamosissimus are mostly piscivores, yet occasionally they also feed on crustaceans and other invertebrates (42, 43, 44, 45, 46). In general, they are sedentary, although they have a short lateral range of movement (47, 48). These two species are not protected by spawning regulations and are sold at markets throughout the year.

All samples used in this study were purchased from the same artisanal fishermen in Itaituba, by whose account the fish had been caught in the small lakes around the village of São Luiz do Tapajós in the municipality of Itaituba (Figure 1). This upper Tapajós river area receives water from small tributaries with artisanal gold mines. Purchases were always made between the 15^th and 17^th of each month from July 2018 to June 2019 to cover the dry (July–December) and the rainy (January–June) season corresponding to low and high river water levels, respectively, measured in Itaituba with a pole placed about 10 m from the bank (Figure 2). Fish were weighed and their size measured. The samples consisted of dorsolateral muscle cuts, divided by season into groups of 60 and frozen at -20 °C until analysis. The dry season group contained 30 samples of Cichla spp. and P. squamosissimus each and so did the rainy season group.

Figure 2

Aliquots (duplicates) weighing 0.005–0.007 g of finely cut hair or 0.05–0.06 g of fresh fish were placed on quartz boats and inserted into a direct mercury analyser, DMA-80 (Milestone Srl, Sorisole, Italy), as this method does not require sample pretreatment because it involves thermal degradation of the sample, catalytic conversion, amalgamation, and spectrophotometry (a mercury lamp with a wavelength equal to 253.55 nm). Oxygen (99.99 % purity) was used as the carrier gas, and the detection limit was 0.0015 ng. To ensure analytical quality we used certified reference materials NIES 13 (human hair; National Institute for Environmental Studies, Tsukuba, Japan) and BCR-463 (tuna fish flesh; European Commission Directorate - Joint Research Centre, Geel, Belgium). Respective recoveries of 94 and 97 % were routinely observed for either.

The risk was calculated for each woman by dividing hair THg concentration by either the hair reference dose (RfD) proposed by the United States Environmental Protection Agency (US EPA) of 1.0 mg/kg (49) or by the lowest observable adverse effect hair concentration (LOAEHC) of 0.3 mg/kg associated with an adverse neurodevelopmental effect (50). The US EPA hair RfD corresponds to the provisional tolerable weekly Hg intake of 0.7 μg/kg body weight (PTWI) adopted in 2001 based on neurologic developmental effects in children associated with exposure to MeHg from maternal diet in utero (51). It is more protective than other relevant RfDs (52, 53), which is why we selected it to assess the risk for the newborns of our study participants.

The normality of data distribution was tested with the Kolmogorov-Smirnov and Shapiro-Wilk test significant differences (p<0.05) confirmed with one-way analysis of variance (ANOVA). The software used was Prism GraphPad 8 (GraphPad Software, San Diego, CA, USA).

Our participants declared consuming 21 fish species during pregnancy (Table 1), 13 of which are non-piscivorous. Piscivorous Cichla spp. and P. squamosissimus were at the top of consumed fish in that group, as reported by 44 and 67 % of participants, respectively.

The fish did not significantly differ in weight and length between the seasons (Table 2). It has been claimed that larger and heavier carnivorous fish tend to concentrate more Hg (54), but we found no significant correlation between THg concentration and fish weight or size (Figure 3), which is in line with an earlier report for Cichla spp. in Jacareacanga, which is 300 km upstream the Tapajós River from Itaituba (36), whereas contrasting correlations were reported for carnivorous species Pseudoplatystoma sp. (surubim) and Brachyplatystoma flavicans (dourada), also consumed in Itaituba (39) or herbivorous species caratinga (Geophagus spp.) and carnivorous traíra (Hoplias malabaricus) caught in Jacareacanga (36).

Figure 3

Table 3 shows fish THg concentrations for both species by season. Dry season concentrations did not significantly differ between the two species, but Cichla spp. showed a 37 % lower THg (p<0.05) in the dry season, whereas THg concentrations in P. squamosissimus did not significantly differ between seasons.

The maximum Hg concentration in fresh fish flesh recommended by the joint Food and Drug Administration (FDA) and the US EPA for consumption by pregnant or breastfeeding women and children below 12 years of age is 0.5 mg/kg (55). The following fish species should be avoided by this vulnerable population group: king mackerel, marlin, orange roughy, shark, swordfish, tilefish, and tuna bigeye (55), as they contain mean THg (recorded between 1990 and 2012) of 0.73, 0.49, 0.57, 0.98, 1.00, 1.12, 0.69 mg/kg fresh weight, respectively (56). In the dry season, 71 % of Cichla spp. and 80 % of P. squamosissimus specimens had THg above this recommended limit of 0.5 mg/kg. In the rainy season, this percentage dropped to 37 % for Cichla spp. and to 71 % for P. squamosissimus specimens.

Total mercury concentrations found in the present study are in the range of those previously reported for carnivorous fish from the Tapajós River basin (40), ranging from 0.4 to 1.51 mg/kg, and Itaituba in particular (8), ranging from 0.9 to 3.2 mg/kg. Previous studies showed that near Itaituba, THg concentrations in Cichla spp. ranged from 0.19 to 1.00 mg/kg (10, 57, 58) and in P. squamosissimus from 0.09 to 1.35 mg/kg (11, 58). Nowadays, after more than twenty years of studies at the Tapajós River basin, the THg levels in carnivorous fish are still above the safe limit of 0.5 mg/kg, and the same is true for THg concentrations in P. squamosissimus of the upper Amazon (59).

Most of our study participants lived in the urban (n=75) area of Itaituba, followed by rural residents (n=17) and other (n=10) cities close to Itaituba (Figure 4). We found no difference in hair THg levels between these three groups of residents.

Figure 4

Table 4 shows hair THg levels by age, education level, marital status, race, and occupation groups. It does not include smoking as none reported to have smoked. None of these parameters significantly influenced hair THg concentration, nor did alcohol and fruit consumption (data not shown). The same is true for body mass index (mean ± SD: 27.4±4.4, median: 26.7, interquartile range: 24.4–30.2, n=110) (Table 5), and systolic/diastolic blood pressure (Table 6).

Mean (±SD) hair THg concentration of all participants (n=110) was 1.6±1.5 mg/kg, ranging from 0.09 to 17.97 mg/kg (median: 0.95 mg/kg, interquartile range: 0.6–2.2 mg/kg, n=108), which is below previously reported means in small villages along the Tapajós River (10, 12, 14, 39, 54, 58, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73). However, it exceeds the US EPA RfD (49) of 1 mg/kg for hair. In 30 participants (27 %) it was 2–6 mg/kg, in four (3.6 %) >6 μg/g, and in one as high as 17.9 mg/ kg.

Table 7 shows hair THg concentrations by fish consumption. Only the group consuming >400 g/week had significantly higher THg than the group consuming the lowest amounts (0–100 g/week) (p<0.05). Considering the US EPA RfD, health risk was increased in 33 % of the women consuming up to 100 g of study fish a week and in 56 % of those consuming more than that (Table 7). Considering the LOAEHC, neurodevelopmental risks of MeHg to the unborn child from maternal fish consumption (49) were increased in 85 % of women consuming up to 200 g/week and in all women consuming more than that (Table 7).

Our hair THg findings are comparable with those reported for women of childbearing age from the island of Vieques offshore from Puerto Rico (median: 0.66 mg/kg, in 26.8 % of these women above the US EPA RfD) (74), Sweden (median: 0.7 mg/kg, in 20 % above the US EPA RfD) (75), and Moroccan coastal communities (median: 1.18 mg/kg, in 50 % above the US EPA RfD) (76).

Exposure in newborns has already been evidenced by a Spanish study (77) which established hair THg concentrations of 1.4 and 2.0 mg/kg in Madrid and Sabadell, exceeding the US EPA RfD in 39 and 60 % of newborns, respectively. Adverse effects have been highlighted worldwide for maternal hair THg concentrations in the range and even below those measured in our study: lower cognitive scores in six month-old US children associated with maternal hair THg concentration of 0.55 mg/kg (78, 79), impairment in behavioural ability in Chinese newborns associated with maternal hair THg concentration of 1.25 mg/kg (80), and lower psychomotor development index in 30-month-old Seychelles children associated with maternal hair THg concentration of 5.7 mg/kg (81). In indigenous Suriname communities, preterm birth (<37 weeks) was associated with maternal hair THg concentration of 3.5 mg/kg (82), and Michigan women who delivered too early (<35 weeks) had hair THg concentrations between 0.55 and 2.5 mg/kg (83).

Although mean hair THg concentration found in this study (1.6±0.2 mg/kg) is below that of the Barreiras community close to Itaituba (10.38 mg/kg) (13), it is nevertheless quite higher than those recorded in women of childbearing age from Barreiras and São Luis do Tapajós (villages close to Itaituba) between 1999 and 2012 (1.07±0.03 and 0.74±0.05 mg/kg, respectively) (84) and between hair THg concentrations found in women of childbearing age living in Santarém (300,000 inhabitants, a city located at the confluence of the Tapajós and Amazon Rivers) and Oriximiná (63,000 inhabitants, a city located on the Trombetas River, 40 km upstream from its confluence with the Amazon River) of 1.1±0.2 mg/kg and 2.5±1.0 mg/kg, respectively (85).

Previous studies conducted in women from villages located in the Tapajós River basin showed higher consumption of fish than our study participants: the mean amounts of fish ingested by women during one meal varied from 80±25 g in Açaituba to 147±72 g in São Luis do Tapajós (58). Mean fish portions (fish per meal per capita) of 243±135 g (median: 200 g) have been reported for the riverside general population (including adult men and women) of the Madeira River basin (86), and an average fish consumption of 110 g/day for fishermen of the Alta Floresta region (87) and for villagers of the Balbina reservoir (88). One should also take into account socioeconomic implications. The two piscivorous fish species evaluated in this study are more expensive than other species and most of our participants are of modest means, judging by self-reported elementary school level in nearly 50 % and unemployment in nearly 82 %, (Table 4), who therefore eat other, cheaper fish. This suggests that women with higher income may be at higher risk of Hg exposure than those with low income.

Previous studies have shown that non-piscivorous fish from the Tapajós River basin have considerably lower THg concentrations than piscivorous fish: 0.18 vs 1.66 mg/kg (8), 0.03–0.30 mg/kg vs 0.40–1.51 mg/kg (40), 0.095 vs 0.297 mg/kg (36), 0.05–0.08 vs 0.37–0.84 mg/kg (39), and 0.01–0.04 vs 0.12–0.66 mg/kg (89), respectively.

Considering that fish is beneficial for humans, non-piscivorous fish should be the best choice for consumption, especially during pregnancy. The supply of non-piscivorous fish bred in aquacultures has increased around Itaituba in recent times, such as that of the species Colossoma macropomum (the local name is tambaqui), which most of our participants (n=90) have reported eating. With its THg content of 0.08±0.03 mg/kg (39), 0.03±0.01 mg/kg (89), and 0.04±0.01 mg/kg (40), it may provide a healthier alternative at lower price. In addition, Itaituba is nowadays connected with the rest of the country by roads, which makes diet choices more diversified. This, however, is still not true for small villages along the Tapajós River banks. Furthermore, the expansion of local cattle and farms has increased the supply of beef and chicken at prices lower than those of fish.

This study has some limitations. The number of participants (n=110) is low, although it allows statistical analysis. Only two fish species have been sampled (although they are the two most favoured piscivorous species by our participants), and the ideal experimental protocol would have been to sample all of the species consumed and then collect consumption data for each participant (weight consumed on a weekly or monthly basis for each species) in order to calculate dietary exposure. However, its implementation was not feasible, as people are reluctant to measure the weight of each fish in their meal, and we could only get a rough estimation of weekly servings.