Heavy metals in fish and its association with autoimmunity in the development of juvenile idiopathic arthritis: a prospective birth cohort study

Every member of the ABIS-cohort was followed to see if they developed JIA by linking data from the ABIS-cohort and The Swedish National Patient Register [6] by using a unique 10-digit PIN [5]. We identified 59 children with an ICD code of JIA (ICD 9–10 code M08–09) and had accepted to participate in ABIS (see Fig. 1). Three of children had moved out of the ABIS region, but could be found and included with the PIN. 17/59 patients were excluded due to misdiagnosis (mostly monoarthritis that later proved to be reactive arthritis), after communication with the paediatric rheumatologists at local hospitals who reviewed the medical records of all 59 patients. Of the remaining 42, had 41 completed the questionnaire after delivery, and 32 of them had completed both the questionnaire after delivery and the 1 year follow up questionnaire. 29 of 32 had also completed the detailed nutrition diary during the child’s first year. All cases of JIA and their categories were double-checked via The Swedish paediatric JIA-registry. ANA positivity was detected in 11 patients with JIA. Data on fish consumption were collected from both the detailed diary and the questionnaires. Heavy metals were analysed in all 42 JIA patients and in an additional 40 randomly selected age and sex-matched controls.

ANA positivity was detected using indirect immunofluorescence with human epithelioid cells (Hep-2 cells) as substrate [7]. ANA positivity was defined as at least two positive results (titer of ≥1:160) analysed at least 3 months apart. RF was determined by latex agglutination test.

The analysis of heavy metals was performed by ALS Scandinavia AB (Luleå, Sweden) using the ‘ultrasensitive inductively coupled plasma sector field mass spectrometry method’ (ICP-SFMS) [8] after acid digestion with HNO3, according to the standards ISO 17294-1 (https://​www.​iso.​org/​standard/​32957.​html) and 2 (https://​www.​iso.​org/​standard/​62962.​html), and the EPA Method 200.8 (https://​www.​epa.​gov/​homeland-security-research/​epa-method-2008-determination-trace-elements-waters-and-wastes). Heavy metals analysed were: Aluminium (Al), Cadmium (Cd), Mercury (Hg), Lithium (Li), Lead (Pb), Zinc (Zn), Copper (Cu), Arsenic (As), Magnesium (Mg) and Iron (Fe).

Homogeneity of variance was tested using Levene’s test and the normality for independent variables was revised both graphically and by the Shapiro–Wilk test.

Possible confounding factors included heredity for rheumatism (JIA or RA in first- and second-degree relatives), parity, mode of delivery, preterm birth, duration of (exclusive and total) breastfeeding, introduction of formula, parental age, education, smoking habits, and whether or not the parents were born outside Sweden.

Statistics were calculated using the Statistical Package for Social Science (SPSS 25.0 software; SPSS Inc., Chicago, IL, USA).

Forty-one cases of JIA with complete questionnaires after delivery were identified. The number and proportion of patients in each JIA category are shown in Table 1. The unclassified category based on the ILAR-criteria [1], either did not fulfil criteria for any category or they met categorization criteria for 2 subtypes. Family history of RA/JIA in first- and second-degree family member are presented in Table 1.

Fish consumption more than once a week during pregnancy was associated with an increased risk of JIA (aOR 4.5 (1.95–10.4); p < 0.001). The child’s consumption of fish more than once a week during the first year of life was also associated with an increased risk of JIA (aOR 5.1 (2.1–12.4) p < 0.001), as illustrated in Table 1 and Fig. 2. 87% (27/31) of children with JIA was exposed to fish more than once a week during pregnancy or during their first year, compared to 56% in the general population (aOR 5.4 (1.9–15.4) p = 0.002).

Fish consumption more than once a week during pregnancy as well as during the child’s first year of life was associated with an increased risk of ANA-positivity (aOR 2.2 (1.4–3.6); p = 0.002 and p < 0.001, respectively). 100% of children with ANA-positivity had consumed fish more than once a week during their first year. Fish consumption at the age of 2.5 and 5 years showed no association with ANA-positivity (Table 1 and Fig. 2).

Concentrations of Al, Cd and Li in cord blood were significantly higher in the ANA-group than in controls (Table 2). Chi²-test with concentrations of heavy metals dichotomized below/above the detection level (Al 5.00 μg/l; Cd 0.05 μg/l; Li 1.00 μg/l) showed significant results for Al (p = 0.040), Cd (p = 0.008) and Li (p = 0.006). 82% of the children with ANA antibodies had concentrations of Al, Cd, or Li higher than the mean, and 64% had concentrations two standard deviations above the mean. 91% (10/11) of the ANA-positive had elevated concentrations higher than the mean of one or more of all the analysed heavy metals.

Concentrations of Al, Cd, Hg and Li in cord blood were significantly higher in the JIA-group than in controls (Table 2). Chi²-test of heavy metals dichotomized below/above the detection level (Al 5.00 μg/l; Cd 0.05 μg/l; Hg 0.200 μg/l; Li 1.00 μg/l) showed significant results for Al (p < 0.001), Cd (p < 0.001), Hg (p < 0.001) and Li (p < 0.001). 71% of the children with JIA had concentrations of Al, Cd, Hg or Li higher than the mean, and 52% had concentrations two standard deviations above the mean. 95% (40/42) in the JIA-group had elevated concentrations higher than the mean of one or more of all the analysed heavy metals.

Frequency of (total) fish consumption during pregnancy correlated with concentrations of Cd (p = 0.003), Li (p = 0.015) and Hg (p = 0.011) in cord blood, but not with concentrations of Al (p = 0.261) (Table 2). Consumption of fish from the Baltic Sea correlated with concentrations of Cd (p = 0.011), and consumption of “other fish” (defined as all sorts of fish except fish from the Baltic Sea and fresh water) correlated with concentrations of Cd (p = 0.028), Li (p = 0.006) and Hg (p = 0.012). Frequency of consumption of fish from fresh water was extremely low during pregnancy, as shown in Table 1. All statistical associations remained statistically significant when potential confounders (described in Material and methods section) were included in the model.

The small patient sample is a weakness due to the nature of the study, i.e. a birth cohort of a relatively rare disease. Different categories of JIA should be studied as separate groups [42, 43], but our study sample was too small to allow subgroup analysis. Although there were dropouts, these were not found to associate with heredity or subsequent diagnosis; therefore, it seems unlikely that our findings are a result of skewed attrition due to lost to follow-up participants. Other possible limitations are unmeasured contaminants in the fish, which may be associated with heavy metals, and leading to these observations, as well as unmeasured confounders, misclassification of diagnosis, selection bias and generalizability of the data.

Our study has some notable strengths. First, the prospective study design avoids the potential recall and selection biases of retrospective studies that collect data on nutrition and lifestyle after the diagnosis. The prospective design distributes any possible measurement biases equally between the JIA cases and non-cases, and recall bias is not likely to explain our findings. The ABIS cohort has been found to be representative for the Swedish general population in aspects such as education level [44]. Second, all cases of JIA were collected through the unique 10-digit PIN with codes from the Swedish National Patient Register with more than 99% coverage of all visits from both private and public caregivers. Subsequently, all cases were confirmed. This is a significant advantage over studies that rely only on self-report. Third, we have validated all diagnoses via The Swedish paediatric JIA-register and with the local physicians/hospitals. Fourth, the access to detailed information on important early life factors allowed us to control for several potential confounding factors that may have influenced our observed associations.