Comparison of Circulating and Excreted Metals and of Autoimmunity between Two Great Plains Tribal Communities

Abstract

Metals contaminants of the environment from mine waste have been implicated as contributing agents in autoimmune disease. The current study compares metals and autoimmunity in two Tribal communities residing in the Black Hills and the Bighorn Mountains geographical regions that are scattered with extant hard rock mines. With documented drinking water contamination in both communities, in vivo levels of more than half of the measured serum and urine metals differed between the two communities and were substantially different from their national median values. Serum autoantibodies associated with systemic autoimmune disease were rare or at low-level, but antibodies to denatured (single-stranded) DNA and thyroid-specific autoantibodies were commonly elevated, especially in women. A three-tier statistical modeling process was carried out to examine individual metals exposure as predictors of autoantibody levels. For the most part only weak positive associations between individual metals and systemic autoantibodies were found, although univariate quantile regression analysis showed positive statistical associations of serum lead and antimony with anti-chromatin and anti-histone autoantibodies. Using age and gender-adjusted multivariable statistical models, metals did not predict anti-thyroglobulin or -thyroid peroxidase significantly and metals were generally negative predictors of the other autoantibodies. Overall these results suggest that elevated levels of environmental metals and metalloids in these communities may result in suppression of autoantibodies associated with systemic autoimmune disease.

1. INTRODUCTION

More than one hundred years of hard rock mining left a complex environmental legacy of more than 160,000 abandoned mines in the Western United States. The same geographic area is home to over 600,000 Native people and large segments of Native American Tribal lands [1]. The Black Hills Mountain Range and the Bighorn Mountains include metal and rare element rich geological formations explored by the extraction industry [2–5], resulting in metal leeching into the water table and surface environmental contamination [6–9]. Tribal communities enrolled in the current study are from Midwestern/Western U.S. geographical areas (in the states of South Dakota and Montana) where contamination of soil and water by mining-associated waste has been a community and health care provider concern since the 1980s [3, 9]. Sparse scientific data exists on environmental metals and metalloid exposure and their possible health effects, including immune system alterations.

Extensive immunotoxicology literature set decades ago a clear but complex path on methodological and in-depth approaches to evaluating possible hazardous chemical contaminant exposure [10]. This goal was further emphasized by the World Health Organization’s Environmental Health Criteria edition, encouraging early biomarker development for monitoring disease signatures in exposed communities [11]. This effort is especially wanting in Native People living on Tribal lands who have not been part of prior systematic, nation-wide sampling efforts by the U.S. Centers for Disease Control and Prevention’s National Center for Health Statistics such as the National Health and Nutrition Examination Survey (NHANES) of exposure evaluation of metal, metalloids and other possible chemical toxicants [12]. In addition, metal and other chemical exposure comparisons between Tribal Nations within the U.S. that might predict distinct abnormal immune system biomarkers are scarce. Toxicants associated with Tribal traditional practices, particularly fishing, land use and outdoor activities (e.g. hunting and gathering) as well as fortuitous drinking water contamination are of special interest to Tribal communities. These concerns are articulated in the USEPA Tribal Science Council report [13], but are not yet considered in federal risk estimations by the U.S. Environmental Protection Agency [13].

The current study combines measurements of serum and urine metal and metalloids with serum autoantibody from members of two Tribal communities residing in the watersheds of neighboring regions of the Black Hills and Bighorn Mountains, which are scattered with extant and/or abandoned hard rock mines [1]. Association between environmental metals exposure and autoantibodies may indicate possible immune system disruption by these environmental contaminants, analogous to the capacity of numerous ingested medications to induce autoimmunity [14–16]. As presented in other Tribal population investigations by our team, autoantibodies against histones and/or denatured DNA (dDNA or single-stranded DNA) can be induced by xenobiotics and possibly environmental factors, while chromatin and native DNA (nDNA or double-stranded DNA)-specific autoantibodies are categorized as idiopathic. Capturing both types of common autoantibodies provide a broader autoimmunity characterization [31–33]. Through accurate biomonitoring of serum and urine in communities, insight into the importance of environmental metal exposures to immune dysregulation in the form of autoantibody development could be compared.

2. MATERIAL AND METHODS

2.1. Study populations

Cheyenne River Sioux Tribal (CRST) members were recruited from 13 reservation communities during 2014–2016. After informed consent was obtained from 225 adult CRST participants (age 18–77, mean age 41.8±13.4 (±SD) yrs, 52% female), blood and urine samples were collected at local community centers and were promptly processed at the Missouri Breaks Industries Research Inc. local office (Eagle Butte, SD). Study details can be found in previous publication as well [16]. Through the Crow Environmental Health Steering Committee’s community engagement and outreach efforts in 2014–2018, 43 participants from the Crow Reservation were enrolled (age 21–85, mean age 49.0±18.1 (±SD) yrs, 65% female). Crow participants were primarily people who had metals-contaminated home well water; they contributed samples over several months in 2018–2019 at the public health clinic at Hardin, MT (currently called the One Health Clinic, formerly known as Bighorn Valley Health Center).

2.2. Biospecimen collection and biomonitoring

Tribal community exposure to various metals and metalloids from unremediated hardrock mines (>900 gold and silver mines and possibly undisclosed uranium mines) in the Northern Great Plains of the Western U.S. was assessed by measurement of urine and serum. Five ml of blood was drawn from each participant using a red top Vacutainer (N=225 for CRST and N=43 for Crow Tribal members), and serum was separated by centrifugation after 30 minutes of clotting. Serum aliquots were stored at −80°C. Urine samples were collected at the same time, and 1.5 ml was stored at −80°C. Excreted metals concentrations were determined in 1.5 ml urine samples at the University of New Mexico College of Pharmacy Trace Metals Laboratory using inductively coupled plasma mass spectrometry (ICP-MS, Perkin Elmer, Inc., Houston, TX) using internal assay standards. Serum samples were shipped frozen to the University of Arizona Analytical Laboratory for Emerging Contaminants (Tucson, AZ) for metals testing. The lower limits of detection (LLOD) of individual metals/metalloids in serum and urine are shown in Tables S1a and S1b.

2.3. Specific autoantibody detection

Autoantibodies against denatured DNA (dDNA, single-stranded DNA), native DNA (nDNA, double-stranded DNA), histones and chromatin were quantified by in-house enzyme-linked immunosorbent assays (ELISA) as previously described [17, 18] using peroxidase-conjugated rabbit anti-human IgG (SouthernBiotech, Birmingham, AL) as the secondary, detecting antibody. These antigens were coated on microtiter plates at 1.0 μg/well; the nDNA wells were pre-coated with poly(lys,phe) (Sigma-Aldrich, St. Louis, MO) at 2.0 μg/well. IgG reactivity to thyroid-specific autoantigens was measured in parallel using thyroid peroxidase (TPO) (LSBio, Seattle, WA) coated at 0.1 μg/well and thyroglobulin (TG) (Lee BioSolutions, Maryland Heights, MO) coated at 0.4 μg/well. Positive control standard sera and negative control samples from normal blood donors were included in each assay to ensure that results from different assays could be reliably compiled for analysis. Samples exceeding the range of direct measurement of optical density (O.D.) after 1-hour incubation with the enzyme substrate were extrapolated from the O.D. values at earlier time-points as previously described [17]. Criteria for non-significant antibody activity was based on the mean +3 SD above the mean reactivity of CRST samples having <0.2 O.D. reactivity on each antigen, which was the highest level of normal sera reactivity included in each assay. The number of CRST serum samples producing <0.2 O.D. reactivity and consequently employed for calculating non-significant, background antibody activity on dDNA, nDNA, histone, chromatin, TG, and TPO was 180, 223, 221, 220, 213, and 190, respectively. Crow participants’ serum samples were analyzed with the same methodology and antibody positivities were detected the same way as results of CRST sera.

2.4. Statistical methods

Summary statistics including median as well as the first and third quartiles for continuous variables and frequency (%) for categorical variables were used to describe the demographics, metal exposures and immune outcomes in both Tribal communities. Metal concentrations were considered as continuous variables in statistical models and were log-transformed to normalize distributions when applied in models. These data as well as the 2011–2016 year databases of the National Health and Nutrition Examination Survey (NHANES) [12] representing the U.S. national toxicant distributions across participating counties were reported as the median and 95th percentile. We combined these datasets weighted in order to cover all years of sampling in the communities. The urine chemical measurements were used as raw concentrations, and values below LLOD were replaced by the LLOD value divided by √2 [19], which was demonstrated to have robust performance in comparison with other replacement techniques used historically [20]. Wilcoxon rank-sum tests were used to compare continuous variables across CRST and Crow Nation participants. Chi-squared tests were performed to compare categorical variables between the two participating Tribes. Spearman correlation coefficients were used to summarize the correlation among metals and serum autoantibodies. Correlation coefficients were evaluated for the strength of observed associations [21].

The associations by Tribe between each metal exposure and the six autoantibodies were examined using linear regression analyses adjusted for potential confounding variables including age and gender. These per metal linear regressions will be referred to as ‘adjusted univariate models’. Considering the skewed distribution for the metals and autoantibodies, log transformation of these variables was employed in all regression modeling. Multivariable regression models were also employed for any specific autoantibody that showed more than one significant metal association. No adjustments for the multiple comparison tests were considered due to the discovery nature of this study. The associations between metal exposure and each specific autoantibody were further resolved using quantile regression (QR) models, a statistical method that is well fitted to consider skewed distributions of outcome data [16, 22]. All analyses were conducted using R 4.0 (R Project, Vienna, Austria) and SAS v9.2 (SAS Institute Inc., Cary, NC) [23, 24].

3. RESULTS

3.1. Circulating autoantibodies in CRST and Crow Nation participants

While rigorous and definitive criteria for background antibody reactivity of disease-free individuals cannot be defined for populations such as Native American communities that may not be represented in a cross-sectional sampling of disease-free individuals from other communities, the cut-off for non-reactivity to each autoantigen was estimated from the antigen-specific background reactivity of 98% (mean+3 S.D. >mean) of the lowest reacting CRST samples on each antigen. Fig. 1 displays the gender-separated IgG antibody activity in sera of CRST and Crow Nation donors to the six tested autoantigens. Elevated antibody activity to autoantigens commonly associated with systemic autoimmune disease – histone, nDNA, and chromatin — was infrequent (2–6% of 225 CRST and 0–13% of Crow Nation individuals, Table 1) and generally very low, although two female CRST samples had 6–7-fold elevated anti-chromatin reactivity. Forty-two CRST donors, 12% of the male and 25% of the female participants, had elevated anti-dDNA antibodies, including two samples with an anti-dDNA activity 8- and 11-fold above background. Women were the most frequent among participants with thyroid-specific autoantibodies: 20 (9%) of CRST donors had elevated anti-thyroglobulin, 14 of whom were female, and the 7 elevated Crow samples were all women (16% of total Crow samples). Anti-TPO was elevated in 11 (26%) of the Crow individuals, 10 of whom were women, and in 35 (16%) of CRST donors, about half female (Table 1). The highest anti-TPO CRST and Crow donors were all women, with a mean anti-TPO antibody activity 29- and 31-fold, respectively, above the background level.

Fig. 1.

Autoantibody activity in 225 CRST and 43 Crow Nation participants. Dashed line in each panel is cutoff for upper limit of average background reactivity on each antigen. Scale is different for each panel to enable display of the full range of antibody activities.

Table 1.

Prevalence of autoantibody activity in CRST and Crow Tribal communities

Autoantibody	Bkgd* (O.D.)	Samples above Background (%)
		CRST (N=225)		Crow (N=43)
		Female (N=116)	Male (N=109)	Female (N=28)	Male (N=15)
dDNA	0.21	29 (25.0%)	13 (11.9%)	4 (14.3%)	0 (0%)
Histone	0.10	7 (6.0%)	1 (0.9%)	1 (3.6%)	2 (13.3%)
nDNA	0.08	4 (3.4%)	2 (1.8%)	0 (0%)	0 (0%)
Chromatin	0.08	2 (1.7%)	3 (2.8%)	2 (7.1%)	1 (6.7%)
Thyroglobulin	0.10	14 (12.1%)	6 (5.5%)	7 (25.0%)	0 (0%)
TPO	0.19	19 (16.4%)	16 (14.7%)	10 (35.7%)	1 (6.7%)

^*Bkgd: non-specific background reactivity as defined in Methods

By Wilcoxon rank sum statistical analysis comparing Crow Nation and CRST sera, Crow samples showed significantly elevated anti-TPO activity (p<0.018), while CRST samples tended to have higher anti-chromatin (p<0.015) and anti-dDNA (p<0.010) antibodies. Female Crow participants had higher anti-TPO, anti-TG, and anti-dDNA than Crow males (p<0.001); likewise for the CRST samples, women tended to have higher anti-TG (p<0.033), anti-TPO (p<0.007), and anti-dDNA (p<0.0003) than men (data not shown in Table 1).

Correlation analysis by Spearman correlations demonstrated only fair correlations (< 0.40) between individual autoantibodies with each other. Significant correlation was observed in case of anti-denatured DNA and anti-chromatin autoantibody association (p= 0.032), likely due to overlapping molecular structure similarities of these autoantigens. Also, stronger correlations were detected among the two thyroid-specific autoantibodies (anti-TPO and anti-thyroglobulin antibodies) with a very significant p-value (p< 0.0001) pointing toward the importance of these central endorine tissue-specific autoantibodies.

3.2. Metals exposure in the study populations

Of the total of 34 metal elements and metalloids species examined in serum and/or urine in both communities, 23 analytes (10 in serum and 13 in urine) had significantly different median values between the two communities using Wilcoxon rank-sum test, 68% of all metals assayed (Table 2). The CRST urine samples median concentrations were higher in inorganic arsenic III [As (III)], arsenobetaine (AsB), barium (Ba), manganese (Mn), antimony (Sb), and uranium (U) elements while the Crow urine samples were higher in beryllium (Be), cadmium (Cd), cesium (Cs), mercury (Hg), lead (Pb), tin (Sn), and tungsten (W). Serum metal biomonitoring also revealed differences between the two communities in that Crow had significantly higher concentrations of cadmium (Cd), copper (Cu), lead (Pb), antimony (Sb), and tin (Sn). CRST serum samples had higher median concentrations of silver (Ag), aluminum (Al), beryllium (Be), manganese (Mn) and zinc (Zn).

Table 2.

Metal and metalloid biomonitoring data and comparisons of median values by Tribes

Metal	Concentration (median [Q1, Q3], μg/L)*		p value**
	Crow	CRST
sAg	0.136 (0.036, 0.285)	0.232 (0.113, 0.408)	0.0013
sAl	3.350 (2.800, 5.384)	6.012 (4.316, 10.03)	<.0001
sBa	0.833 (0.669, 1.302)	0.943 (0.478, 11.34)	0.355
sBe	0.019 (0.016, 0.022)	0.039 (0.023, 0.061)	<.0001
sCd	0.135 (0.031, 0.278)	0.013 (0.006, 0.047)	<.0001
sCu	78.10 (61.68, 86.36)	19.59 (15.79, 30.24)	<.0001
sMn	0.181 (0.147, 0.232)	0.226 (0.164, 0.337)	0.008
sMo	0.096 (0.049, 0.120)	0.102 (0.059, 0.192)	0.23
sNi	0.458 (0.386, 0.593)	0.497 (0.321, 0.705)	0.3404
sPb	0.803 (0.532, 1.096)	0.051 (0.032, 0.095)	<.0001
sSb	0.460 (0.378, 1.138)	0.176 (0.059, 0.244)	<.0001
sSn	1.507 (0.317, 2.264)	0.155 (0.077, 0.357)	<.0001
sZn	75.56 (69.97, 86.52)	89.88 (70.82, 103.62)	0.0022
uTAs	3.520 (1.730, 6.610)	3.648 (1.839, 5.735)	0.8135
uAs(III)	0.084 (0.023, 0.261)	0.247 (0.100, 0.413)	0.0021
uAs(V)	0.312 (0.138, 0.440)	0.200 (0.0015, 0.573)	0.8639
uAsB	0.0014 (0.0014, 0.177)	0.040 (0.0015, 0.151)	0.0028
uDMA	2.003 (1.088, 3.529)	2.074 (1.099, 3.525)	0.8977
uMMA	0.332 (0.179, 0.804)	0.456 (0.129, 0.824)	0.9388
uBa	0.042 (0.042, 0.042)	0.519 (0.124, 1.362)	<.0001
uBe	0.011 (0.011, 0.011)	0.0021 (0.0021, 0.0041)	<.0001
uCd	0.188 (0.077, 0.484)	0.074 (0.033, 0.175)	<.0001
uCo	0.291 (0.116, 0.556)	0.180 (0.073, 0.321)	0.052
uCs	2.629 (0.800, 4.290)	1.553 (0.602, 2.787)	0.0106
uHg	0.327 (0.187, 0.575)	0.197 (0.070, 0.418)	0.0005
uMn	0.0092 (0.0092, 0.180)	0.136 (0.072, 0.269)	0.0001
uMo	18.02 (0.57, 49.54)	9.319 (2.205, 22.43)	0.148
uPb	0.197 (0.071, 0.252)	0.074 (0.031, 0.158)	<.0001
uPt	0.0071 (0.0071, 0.0071)	0.0035 (0.0035, 0.014)	0.0732
uSb	0.016 (0.016, 0.040)	0.060 (0.031, 0.090)	<.0001
uSn	0.462 (0.166, 1.631)	0.168 (0.038, 0.460)	<.0001
uSr	49.66 (4.81, 108.23)	39.52 (15.20, 95.21)	0.7711
uTl	0.079 (0.032, 0.178)	n/a	n/a
uU	0.0014 (0.0014, 0.0014)	0.0085 (0.0021, 0.027)	<.0001
uW	0.043 (0.013, 0.122)	0.015 (0.0035, 0.049)	<.0001

^*Medians in bold are significantly different between the Tribes

^**Wilcoxon rank-sum test

abbreviations: n/a, not available; TAs, total arsenic concentration; AsB, arsenobetaine; DMA, dimethylarsinic acid; MMA, monomethyl arsonous acid.

3.3. Comparison of CRST and Crow metals data with that of NHANES

Many elemental metals and metalloids in urine of both Tribal communities had substantially lower or higher median values than the national median values (50^th percentiles) (Table 3), although various samples from both the CRST and Crow communities had serum and urine metals levels below the limit of detection (Tables S1a and S1b). About 20% of CRST samples had values of urinary As(V), Mn, Pt, and U and serum Zn that were higher than the 95^th percentile of the median of the NHANES samples. Serum Cu and Zn were the only serum metals regularly measured in NHANES, and only serum Zn skewed to higher values than NHANES for CRST samples.

Table 3.

Comparison of Crow and CRST metals data with NHANES^† 2011–2016 national biomonitoring data set, all participants

Metal	% above the 50th percentile of NHANES*		% above the 95th percentile of NHANES*
	Crow	CRST	Crow	CRST
sCu	4.6	1.3	2.3	0.4
sZn	36.4	62.2	4.6	20.9
uTAs	26.7	24.4	0	0.5
uAs(III)	22.2	35.3	2.2	2.3
uAs(V)	11.1	26.7	8.9	19.9
uAsB	17.8	11.3	0	0
uDMA	26.7	31.7	0	0.9
uMMA	31.1	33.9	6.7	3.2
uBa	6.7	29	2.2	6.3
uBe	0	0	0	0
uCd	53.3	24.6	6.7	1.8
uCo	40	22.3	4.4	2.2
uCs	31.1	10.7	6.7	0.9
uHg	64.4	47.1	4.4	3.6
uMn	44.4	65.6	8.9	21.9
uMo	31.1	13	6.7	0.9
uPb	17.8	10.7	0	2.2
uPt	100 **	44.2	2.2	18.8
uSb	24.4	62.5	0	5.8
uSn	55.6	25.5	2.2	1.8
uSr	33.3	25.9	6.7	2.2
uTl	26.7	n/a	8.9	n/a
uU	22.2	60.3	2.2	23.2
uW	42.2	23.2	2.2	1.3

^†NHANES Adults (20+ year old) weighted data: for uBe and uPt: NHANES 2009–2010, for all other analytes: NHANES 2011–2016

^*Values in bold are elevated (>55% of samples above the 50th percentile or 10% of samples above the 95th percentile) and are discussed in the text.

^**Only 2 participants had uPt concentrations above LOD level. Those were above the median NHANES 2011–2016 concentrations.

Abbreviations: n/a, not available; TAs, total arsenic concentration; AsB, arsenobetaine; DMA, dimethylarsinic acid; MMA, monomethyl arsonous acid

For almost all metals and arsenic concentrations in urine and serum samples the average level for Crow Nation participants was about the same as or less than the national average as measured by the Centers for Disease Control’s large, national health and nutritional (NHANES ) research study. About 9% of Crow Nation people had elevated arsenic (V) excretion in urine, and also had elevated thallium or manganese, and about 7% had elevated cesium, strontium or molybdenum exposures. Furthermore, mercury was an important metal in urine samples for which the average level in Crow participants was somewhat higher than the national NHANES median concetration.

3.4. Statistical associations among autoantibodies and metals in the CRST Tribe

Spearman correlations between metal exposure and autoimmunity for both Tribes are shown in Figure 2. With CRST samples, positive, albeit weak, correlations (r=0.2–0.4) were detected between serum Sb and levels of anti-histone, anti-chromatin and anti-TPO autoantibodies [21]. CRST members tended to have substantial negative correlation (r =−0.4 to −0.5) between anti-histone and/or anti-chromatin with serum Ba, Cd, Mo, Pb, and Zn. Metals in urine samples also showed negative correlations between As(III), As(V), and Hg with anti-histone and with anti-chromatin autoantibodies.

Fig. 2.

Correlation analyses between autoantibody activity and metal levels in serum (s) and urine (u) from CRST and Crow Nation participants. Scale on the right of each figure explains the color-coded correlation coefficients.

Adjusted univariate regression and multivariable analyses were carried out for all six autoantibody specificities individually (Tables S2a–f). Increased serum concentrations of Pb as well as urinary As (III), arsenobetaine, and MMA yielded negative estimates predicting anti-dDNA antibody in adjusted univariate associations (Table S2a); by multivariable analysis a statistically significant negative estimate remained only for serum Pb concentration. Anti-histone autoantibody levels were significantly positively correlated with serum Mn, serum Sb, and urinary U concentrations, which stayed in the final multivariable model but were not significant predictors of anti-histone (Table S2b). In contrast, sBa, sCd, sCu, sMo, sPb, sSn, sZn, uAs(V), and uHg were significant negative predictors of anti-histone antibodies, some with very low p-values, although by multivariable analysis p-values <0.05 remained only for sBa. Serum Cu was the only significant metal predictor for anti-nDNA (Table S2c), and this association was negative. As with anti-histone antibodies, anti-chromatin antibodies were significantly negatively predicted by sBa, sCd, sCu, sPb, sZn, uTAs, uAs(V), uDMA, and uHg by univariate analysis; of note sPb and uHg also remained significant by multivariable analysis. (Table S2d). In contrast, both adjusted univariate linear regression and univariate QR modeling of serum Sb revealed significant positive estimates of anti-chromatin antibodies that were sustained throughout nearly the entire distribution of immune response (Figure 3). Anti-TG (Table S2e) was not significantly predicted by any metals. In adjusted univariate models, anti-TPO antibodies were negatively predicted by sBa and uAs(V) and positively by sSb among CRST people. However, none of these associations transpired in multivariate age and gender-adjusted models (Table S2f).

Fig. 3.

Univariate quantile regression modeling of log anti-chromatin autoantibody (x-axis) in association with log serum antimony levels among CRST participants. Serum Sb was significantly positively associated with anti-chromatin activity in the middle section of the outcome distribution ordered as increasing antibody values.

3.5. Statistical associations among autoantibodies and metal biomonitoring in the Crow Tribe

From Fig. 2 it can be seen that fair to moderate correlations (r=0.2–0.6) of metals with several autoantibodies among Crow samples were largely negative associations, especially of serum Cu, Mo, Sb, and Sn with antibodies to dDNA, nDNA, histone, and/or chromatin. Strikingly, the overall pattern of associations was different between Crow and CRST. Multivariable modeling of Crow Nation autoantibodies showed that urinary Cs was a significant negative predictor of anti-dDNA (p=0.033, Table S3a), as was serum Sb a strong negative predictor of anti-histone antibodies (p= 0.0012, Table S3b) and anti-nDNA antibodies (p=0.021, Table S3c). Positive, albeit weak correlations (r=0.2–0.4) between serum Ag or the urinary speciated arsenic forms As (III), DMA and MMA and individual autoantibodies (anti-nDNA, anti-histone, and anti-chromatin autoantibodies) can also be observed in Fig. 2. Univariate QR modeling using serum Pb showed a strikingly similar pattern between two biochemically related autoantibody targets, histone and chromatin – as shown in Fig. 4, strongly positive β estimates were observed for both autoantibodies that manifest in the lower middle section of the outcome distribution of immune responses. By multivariable analysis serum Sb was a significant negative predictor of anti-chromatin autoantibody with a remarkably strong p-value (p=0.0011) and estimate (−0.91) (Table S3d). Serum Zn also stayed in this multivariable model but had no significant effect. The thyroid-specific autoantibodies in Crow Nation samples were each positively predicted by a single metal/metalloid analyte such as anti-TG by uAsB (p=0.038, Table S3e) and anti-TPO by sAg (p=0.033, Table S3f). However, none of these metals had significant predictive effects when multivariable, age and gender-adjusted statistical models were applied.

Fig. 4.

Univariate quantile regression modeling of log anti-histone autoantibody (left, x-axis) and log anti-chromatin autoantibody (right, x-axis) in association with log serum lead levels among Crow participants. Pb was significantly positively associated with anti-histone and anti-chromatin activities in the lower-middle section of the outcome distributions ordered as increasing antibody values.

4. DISCUSSION

Of the 795 screened autoantibody activities associated with systemic autoimmune disease (anti-histone, -nDNA, and -chromatin) from 268 people, only a total of 25 samples (11 people in Crow and 14 CRST participants) from both Tribal communities (9.3% overall) were above the cutoff for normal serum reactivity, and most of the levels were low-positive (Table 1). In contrast, elevated levels of anti-dDNA antibodies were relatively common, especially in the female CRST community (25% prevalence) and was observed in 14.3% of Crow women only (Table 1); moderate levels of this specificity (0.2–0.5 OD) are frequently detected in normal individuals, and high levels (>0.5 OD) are commonly induced by long-term ingestion of many medications [15,18]. Thyroid-specific antibodies – both anti-thyroglobulin and anti-thyroid peroxidase – were also common, with 24% of CRST and 42% of Crow participants having one or both of these organ-specific antibodies (shown in Table 1 by genders as well). This finding raises the possibility that some of these people are developing or have ongoing autoimmune (Hashimoto’s) thyroiditis [27–29]. While Crow participants were more likely to have elevated anti-TPO and anti-thyroglobulin activity, and CRST donors tended to have higher anti-chromatin and anti-dDNA antibodies, the autoantibody median differences between the Tribes were significant only by Wilcoxon rank-sum statistical analyses (data not shown).

The analysis of metals in the circulation and in urine also revealed differences between the CRST and Crow communities. Both these Native American Nations are within the Great Plains region of the U.S., separated by only approximately 300 miles (500 km) near the Black Hills mountain range extending between South Dakota and Wyoming. Extensive geological and geochemical literature as well as water contamination studies indicate similar metal mixture exposures in this region of the Midwestern U.S. Tribal lands [30]. Nevertheless, of the concentrations of 22 metals and metalloids in urine samples examined from both communities, 13 were significantly different between Crow and CRST participants. Similarly, of the 13 metals assayed in serum, 10 were significantly different between the two tribes. Whether these substantial concentration differences reflect contamination of food, drinking water, air particulates, land-use patterns or possibly inherent metabolic differences between the tribes are unclear and will require further study.

We used anti-dDNA and anti-histone activities as biomarkers of environmentally-driven autoantibodies as previously employed [31, 32]. Using QR modeling of Crow participants, samples in bottom quartile of anti-histone antibody activity showed a positive association with increased serum Pb concentrations, similar to previous observations, where autoimmune marker reactivities were also found with low but increasing concentrations of various other metals [31, 33]. However, anti-histone and anti-chromatin activities in the middle and upper range of outcome quantiles (at the middle and upper values of autoantibodies) reversed this trend, that might suggest an overall immunosuppressive effect.

Almost one-quarter of CRST participants had relatively high concentrations of zinc in serum and of uranium in urine. Zinc is likely derived from traditional food consumption, which our fishing survey also confirmed (unpublished observations). However, uranium contamination, while documented in U.S. Geological Survey analyses of the Black Hills area, was not a primary toxicant previously identified in drinking water sources or targeted in federal environmental monitoring for the CRST. In addition, one-fifth of CRST participants had urinary As(V), Mn, and Pt above the 95^th percentile of the NHANES survey. Among Crow Nation participants with contaminants excreted above the 95^th percentile of national observations, about 9% of people had highly elevated arsenic (V) excretion in urine, and also had elevated thallium or manganese, and about 7% had elevated cesium, strontium or molybdenum exposures.

Overall, however, the current cross-sectional sampling of metals/metalloids in serum and urine should minimize concerns in the CRST and Crow communities of possible excessive metals exposures that could be implicated in population toxicity, illness, or disability. However, immune system effects of long-term metal exposures need to be further considered in risk assessment of Tribal community health.

One purpose of this study was to determine if metal biomonitoring information was quantitatively related to and thereby implicated in autoantibody development, possibly preceding clinical symptoms and development of autoimmune disease. Using univariate linear regression analyses, only weak positive associations were found between individual urinary or serum metals and autoantibodies, and the concerned metals were different between Crow and CRST participants. However, QR statistical modeling revealed enhancement of anti-chromatin and anti-histone autoantibodies by lead and antimony in each community (Pb in Crow, antimony in CRST partciopants), paralleling phenomena seen with drug-induced autoantibodies [15, 34], at the lower metals exposure range.

5. CONCLUSIONS

It is unlikely that environmental metals can be implicated as independent agents in the induction or amplification of autoantibodies in these communities, especially since substantially elevated levels of autoantibodies were rarely detected in these populations with the exception of thyroid-specific autoantibodies. In fact, the strongest correlations between metal biomonitoring results and autoantibodies were in a negative direction, and these negative associations were observed with all the examined autoantibodies except for the anti-thyroid antibodies.

Curiously, the metals implicated in negative association with autoantibodies were different in CRST and Crow communities. Whether this observation reflects differences in metal metabolism, chemical form, toxicity, or other inherent differences between the two communities that might underlie possible global, immunosuppressive tendencies of systemic autoimmunity is unclear. That anti-thyroglobulin and anti-thyroid peroxidase autoantibodies were prevalent and not negatively correlated with serum metals suggest that spontaneous induction of these autoantibodies is resistant to immune-suppression by in vivo metals. Further population, animal, and/or mechanistic immunotoxicological studies may help explain these findings.

There are several limitations to this study. While we included in each antibody assay normal asymptomatic individuals as representative sera to estimate the cutoff for negligible reactivity to each antigen, the true cutoffs would ideally be established using several dozen sera from Tribal representatives living in a relatively uniform and unexposed environment. In addition, the small sample size of the Crow community biospecimens prevented a more robust analytical workup of this data. The current study could not obtain full information of the participants’ medication use, confirmed clinical diagnoses of autoimmune diseases and/or other chronic diseases. However, future studies could focus on family history of autoimmune disease, occupations, and socioeconomic factors of the participants that could have confounded the observed associations with metals described in the current analysis. Furthermore, only a single time-point of serum and urine collection were analyzed, reflecting short-term metal exposures, with spot urine analysis particularly vulnerable to variable metal exposure and therefore not necessarily representative of the steady-state exposure level. In addition, solid phase ELISA methods may not accurately reflect in vivo autoantibodies in solution [36–37], and we employed a relatively limited repertoire of possible autoantigens. Finally, we were unable to obtain the identical panel of metal analyses from both urine and serum samples due to their assay in different laboratories, precluding comparisons between these biospecimens in the same individual.

In conclusion, while environmental metals in these Tribal communities were generally significant predictors of an immunosuppressive character for systemic autoimmune biomarkers, as supported by some immunotoxicological investigations in the literature [38–41]. Our quantile regression modeling suggested an immune stimulatory capacity of select metals at low to medium exposure levels. Age and female gender were also significant predictors of the systemic autoantibodies, while tissue-specific (thyroid) autoantibodies displayed no dependency on environmental metals. Overall, examination of the complex relationship between autoimmunity and environmental metals in communities living close to mine waste has the potential to provide insight into immune perturbations associated with autoimmune disease.

HIGHLIGHTS.

• This study compares metal exposures and autoimmunity in two proximate Tribal communities residing in the Black Hills and in the Bighorn Mountains of the Nothern Great Plains, a geographical area that is scattered with extant hard rock mines.

• Antibodies to denatured (single-stranded) DNA and thyroid-specific autoantibodies were commonly elevated, especially in female Tribal participants.

• The negative relationship between systemic autoantibodies and environmental metals or metalloids may suggest an immune-suppression phenomenon that does not affect thyroid-specific autoantibodies.