Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Dec 1.
Published in final edited form as: Neurotoxicology. 2024 Oct 11;105:196–207. doi: 10.1016/j.neuro.2024.10.005

Effects of mixed metal exposures on MRI diffusion features in the medial temporal lobe

Eun-Young Lee 1, Juhee Kim 1, Janina Manzieri Prado-Rico 2, Guangwei Du 2, Mechelle M Lewis 2,3, Lan Kong 4, Jeff D Yanosky 4, Paul Eslinger 2,4,6, Byoung-Gwon Kim 5, Young-Seoub Hong 5, Richard B Mailman 2,3, Xuemei Huang 2,3,6,7,8,9
PMCID: PMC11701722  NIHMSID: NIHMS2032514  PMID: 39395642

Abstract

Background:

Environmental exposure to metal mixtures is common and may be associated with increased risk for neurodegenerative disorders including Alzheimer’s disease. This study examined associations of mixed metal exposures with medial temporal lobe (MTL) MRI structural metrics and neuropsychological performance.

Methods:

Metal exposure history, whole blood metal, MRI R1 (1/T1) and R2* (1/T2*) metrics (estimates of brain Mn and Fe, respectively), and neuropsychological tests were obtained from subjects with/without a history of mixed metal exposure from welding fumes (42 exposed subjects; 31 controls). MTL structures (hippocampus, entorhinal and parahippocampal cortices) were assessed by morphologic (volume or cortical thickness) and diffusion tensor imaging [mean (MD), axial (AxD), radial diffusivity (RD), and fractional anisotropy (FA)] metrics. In exposed subjects, effects of mixed metal exposure on MTL structural and neuropsychological metrics were examined.

Results:

Compared to controls, exposed subjects displayed higher MD, AxD, and RD throughout all MTL ROIs (p’s<0.001) with no morphological differences. They also had poorer performance in processing/psychomotor speed, executive, and visuospatial domains (p’s<0.046). Long-term mixed metal exposure history indirectly predicted lower processing speed performance via lower parahippocampal FA (p’s<0.023). Higher entorhinal R1 and whole blood Mn and Cu levels predicted higher entorhinal diffusivity (p’s<0.043) and lower Delayed Story Recall performance (p=0.007).

Discussion:

Mixed metal exposure predicted certain MTL structural and neuropsychological features that are similar to those detected in Alzheimer’s disease at-risk populations. These data warrant follow-up as they may illuminate a potential path for environmental exposure to brain changes associated with Alzheimer’s disease-related health outcomes.

Keywords: mixed metal exposure, Alzheimer’s disease, medial temporal lobe, cortical thickness, diffusion tensor imaging, learning and memory

Introduction

There is a growing concern about health consequences of environmental exposure to toxic metals found in ambient air pollution (e.g., vehicle emissions), contaminated land, and occupation-related activities (Crous-Bou et al. 2020; Delgado-Saborit et al. 2021; Lee et al. 2018; Swartjes 2015). Even essential metals (e.g., Cu, Fe, Mn) can be neurotoxic in lower or higher dosage than its equilibrium status (Teleanu et al. 2018; Zhou et al. 2018), whereas non-essential metals (e.g. Cd, Hg, Pb) can be neurotoxic at low exposure levels (Sanders et al. 2009). These metals can cross the blood-brain barrier, accumulate in specific brain regions, affect cellular functions, and may contribute to neurodegenerative processes (Banerjee et al. 2014; Liu et al. 2010; Miu and Benga 2006).

Alzheimer’s disease (AD) is the most common age-related neurodegenerative disorder, comprising about 50–70% of dementia cases (Brookmeyer et al. 2011). It is characterized pathologically by accumulation of β-amyloid plaques and tau tangles in the brain, with the most prominent neuronal damage noted in the hippocampus (Li et al. 2019). Although the major early disease-related behavioral deficits entail learning/memory problems, the disease also affects executive, attentional, and language functions (Bäckman et al. 2005). The disease-modifying FDA-approved drugs (e.g., lecanemab) only have modest effects on the progression of AD (Bateman et al. 2022), underscoring the importance of identifying modifiable causes.

Preclinical studies have shown that a number of metal exposures (e.g., Cu, Mn, and Fe) can lead to Alzheimer’s-like Aβ production in brain and cognitive impairment (Guilarte 2010; Kitazawa et al. 2016). Epidemiological studies also have reported higher plasma metal levels (e.g., Al, Cd, Cu, Fe, Hg, Pb, and Se) in populations with AD and related disorders (Hare et al. 2016; Xu et al. 2018). Despite the data that associate chronic life-long metal exposure to age-related neurodegenerative diseases, it is challenging to ascertain the effects on specific subgroups. This is due to the difficulty in ascertaining exposure accurately (either distant in time or cumulative, or both), the long-latency of the diseases, and the fact that many early disease-related changes may be subtle and resemble normal aging. Moreover, some metals have shared sources of exposure, and also affect similar targets and pathways such that there can be synergistic or antagonistic effects. Most preclinical and epidemiological studies so far, however, have paid little attention to the mixture or interaction effects of co-exposed metals on neurodegenerative processes.

There are a growing number of biochemical biomarkers (e.g., Aβ and tau in blood and CSF) that can reflect Alzheimer’s-related disease process in clinical practice (Bjorkli et al. 2020). These biofluid based biomarkers, however, do not yield brain region-specific information and/or underlying mechanistic data about how some brain regions are more vulnerable to disease processes. Radioligand-based neuroimaging [e.g., β-amyloid positron emission tomography (PET)] provides brain-region specific information, but it requires radiation exposure and has limited anatomical resolution (compared to MRI, see below). More importantly, radioligand-based neuroimaging can only report on one target at a given time (Vaquero and Kinahan 2015), limiting its utility to examine multidimensional features related to neurodegeneration.

A number of brain MRI techniques have been shown to be useful non-invasive biomarkers that gauge region-specific neurodegenerative processes, including AD-related early changes (Kress et al. 2023; Shukla et al. 2023). Recent studies suggested that a variety of brain structures (e.g., fornix, cingulum, and entorhinal and parahippocampal cortices etc.) may exhibit changes early in the disease process (Echavarri et al. 2011; Lacalle-Aurioles and Iturria-Medina 2023) and better predict future AD conversion (Killiany et al. 2002; Leandrou et al. 2018). Among these brain regions, medial temporal lobe (MTL) memory areas that include the hippocampus and entorhinal and parahippocampal cortices may play a key role in AD-related cognitive decline, particularly affecting learning and memory (Eichenbaum and Lipton 2008; Sugar and Moser 2019). Previous MRI studies reported MTL morphometric (e.g., atrophy, shape) and water molecule diffusion (reflecting microstructural integrity) alterations in both early-stage AD and in at-risk populations (e.g., MCI; mild cognitive impairment) who subsequently converted to AD (Lancaster et al. 2016; Leandrou et al. 2018).

In the current study, we examined the effects of mixed metal exposure on MTL structural features and cognition in welders. Welders can serve as an epitype human population for translational research of mixed metal exposure-related neurotoxicity because welding fumes contain various metals (Subedi et al. 2019). Our overall hypothesis was that exposure to metal mixture via welding fumes is associated with MTL structural and cognitive changes like those seen in AD at-risk populations. We tested three specific hypotheses: H1) Exposed subjects (welders) will have higher medial (MD), axial (AxD), radial (RD) diffusivity, and lower fractional anisotropy (FA) values in MTL structures than controls; H2) Exposed subjects will have lower performance on processing/psychomotor speed, executive, and learning/memory tasks than controls; and H3) In exposed subjects, multiple metal exposures will predict MTL structural and cognitive metrics. Lastly, we explored potential mediation effects of mixed metal exposure on neuropsychological performance via MTL diffusion features.

Methods

Study subjects

Eighty subjects were recruited from labor unions in central Pennsylvania (USA) and the surrounding community between 2011–2013. Exposed subjects were welders who had welded a minimum of three years at any point in their lifetime. Welders performed multiple types of welding, but overall shield metal arc welding (SMAW), gas metal arc welding (GMAW), and gas tungsten arc welding (GTAW) accounted for most of the welding. Controls were matched in age- and gender and had various occupations (e.g., warehouse workers, drivers, and salesman etc.) with no history of welding were recruited from same area. All subjects denied any neurological or neuropsychiatric disorders. All subjects were male, with Movement Disorder Society Unified Parkinson’s disease Rating Scale motor exam sub-scores (UPDRS-III) < 15, Mini-Mental Status Examination (MMSE) scores >24, and Montreal Cognitive Assessment (MoCA) >19. Forty-two welders and 31 controls completed brain MRI acquisition with good quality images (five welders and two controls were excluded). A priori sample size calculation was conducted using GPower3.1. For a multivariate analysis of variance (MANOVA) design with 2 groups and 3 (e.g., ROIs for each DTI metric) to 12 (for blood metal levels) outcome variables, a minimum number of total 47 to 100 subjects depending on the number of outcome variables will give 90% power to detect group differences at a significance level of α=0.05 with an effect size of f2=0.25.

Standard Protocol Approvals, Registrations, and Patient Consents

Written informed consent was obtained in accordance with the Declaration of Helsinki and approved by the Penn State Hershey Medical Center Internal Review Board.

Ascertainment of occupational metal exposure history and whole blood metal levels

Metal exposure history metrics were estimated by an established metal exposure questionnaire (Lee et al. 2015) and calculated as follows:

Metal exposure through welding occurring in the 90 days prior to the study visit

HrsW90 = (weeks worked)*(h/week)*(fraction of time worked related directly to welding)

E90 = estimate of the cumulative metal exposure via welding

Lifetime metal exposure including welding years

YrsW =years spent welding during the subjects’ life

ELT =estimate of cumulative metal exposure via welding fumes over the individual’s life

We obtained whole blood the morning (ca. 08:00 a.m.) of the day that cognitive tests and brain MRI were performed. The whole blood samples were analyzed by Inductively Coupled Plasma Mass Spectrometry (ICP-MS) for trace minerals including Cr, Cu, Fe, Mn, Pb, Se, Sr, and Zn. Digestion was performed by microwave methods using the Discovery SPD digestion unit (CEM, Matthews, North Carolina). After digestion, the samples were analyzed for trace minerals using the Thermo (Bremen, Germany) Element 2 SF-ICP-MS equipped with a concentric glass nebulizer and Peltier-cooled glass cyclonic spray chamber. Calibration standards were prepared to determine blood metal levels were for each single element in a linear range from 0.025 to 10 ng/g obtained from SCP Science, USA. Concentrations for bulk minerals (e.g., Ca, K, Mg, and Na) were determined by ICP-OES (Optical Emission Spectrometry) analysis on the Thermo iCAP equipped with a polypropylene cyclonic spray chamber at the University of North Carolina, Chapel Hill, NC, USA (Lee et al. 2015).

Neuropsychological tests (NPTs) and scores

Six neurobehavioral domains were examined by subtests from the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS), Wechsler Adult Intelligence Scale- Third Edition (WAIS-III); Delis-Kaplan Executive Function System (D-KEFS); and Wechsler Memory Scale- Third Edition (WMS-III): (1) processing/psychomotor speed (Symbol Search; Stroop-Word; Stroop-Color; Trail Making-A subtests); (2) executive function (Symbol-Digit Coding; Stroop-Color-Word; Trail Making-B; Phonemic Fluency subtests); (3) language (Picture Naming; Semantic Fluency); (4) learning/memory (List Learning; Delayed List Recall; Immediate Story Recall; Delayed Story Recall; Delayed Figure Recall subtests); (5) visuospatial processing (Line Orientation; Figure Copy subtests); and (6) attention/working memory (Digit Span; Spatial Span; Letter-Number Sequencing subtests). Individual norm scores were converted to z-scores [(individual norm-based scores – mean norm) / (standard deviation of the norm)] to unify different norm (T- or scaled) scores.

MRI acquisition and processing

All MR images were acquired using a Siemens 3 T scanner (Magnetom Trio, Erlangen, Germany) with an 8-channel head coil. First, high-resolution T1-weighted (T1W) images were acquired for anatomical segmentation. For T1W images, MPRAGE sequences with Repetition Time (TR)/Echo Time (TE)=1540/2.3 ms, FoV/matrix=256×256/256×256 mm, slice thickness=1 mm, slice number=176 (with no gap), and voxel spacing 1×1×1 mm were used. For diffusion tensor imaging (DTI), TR/TE=8300/82 ms, b value=1000 s/mm2, diffusion gradient directions=42 and 7 b=0 scans, FoV/matrix=256×256/128×128 mm, slice thickness=2 mm, and slice number=65 were used.

Regions of Interest:

Medial temporal regions that previously had been reported as AD-related early changes (hippocampus, and entorhinal and parahippocampal cortices; Figure 1) were selected as regions-of-interest (ROIs). They were defined for each subject using Freesurfer.(Fischl et al. 2004) Segmentation quality then was confirmed visually by a reviewer blinded to group assignment.

Figure 1.

Figure 1.

Open in a new tab

Example of automatically segmented regions of interest [hippocampus, entorhinal and parahippocampal cortices] on T1-weighted MPRAGE images.

Morphologic Metrics:

Hippocampal segmentation for volume estimation and cortical parcellation for thickness calculation were performed with the Freesurfer image analysis suite (http://surfer.nmr.mgh.harvard.edu/). Processing included motion correction, removal of non-brain tissue using a hybrid watershed/surface deformation procedure, and automated Talairach transformation. Within temporal lobe, deep gray matter volumetric structures including the hippocampus then were segmented and cortical gray matter structures including entorhinal and parahippocampal cortices were parcellated (Fischl et al. 2004).

Pallidal index (PI):

The PI was derived from the ratio of GP T1W intensity to FWM (Frontal White Matter) intensity [PI=(GP/FWM) × 100].

MRI R1 values:

To calculate R1 values (1/T1), ROIs were co-registered onto the whole brain T1 time images generated by the scanner using an affine registration implemented in 3D Slicer (www.slicer.org; Rueckert et al. 1999).

MRI R2*values:

The magnitude images of multigradient-echo images were used to generate R2* maps utilizing a voxel-wise linear least-squares fit to a monoexponential function with free baseline using in-house MATLAB (The MathWorks, Inc., Natick, MA) tools. The ROIs were co-registered onto the R2* maps using an affine registration implemented in 3D Slicer R1 and R2* values from the globus pallidus also were calculated since brain Mn accumulation was mostly reported in this region.

Diffusion Metrics:

DTI quality control and tensor reconstruction were performed using DTIPrep (Oguz et al. 2014). DTI maps then were estimated via weighted least squares. The diffusion maps were co-registered onto T1W images using ANTS, and the transformation matrix was applied inversely to bring the hippocampal region to the DTI maps. Four DTI indices [fractional anisotropy (FA), AxD (axial diffusivity), RD (radial diffusivity), and MD (mean diffusivity)] were calculated out of three diffusivity eigenvalues (λ1, λ2, λ3). FA is a weighted average of pairwise differences of the three eigenvalues and may represent the degree of diffusion anisotropy. AxD is the largest eigenvalue (λ1) and RD is an average of the remaining two eigenvalues, both of which may indicate the orientation of the diffusion. MD is an average of the three eigenvalues, providing the overall diffusion magnitude.

Statistical analysis

Group comparisons of demographic data were conducted using one-way analysis of variance (ANOVA). For group comparisons of metal exposure measurements (exposure history, whole blood metal levels, and MRI PI, R1, and R2* values), multivariate analysis of variance (MANOVA) was used to control intercorrelations among dependent variables. Whole blood metal levels were log-transformed to attenuate the impact of potential extreme values. Group comparisons of MRI and neuropsychological metrics were conducted using multivariate analysis of covariance (MANCOVA) with adjustment for age and education level. When comparing hippocampal volume, total intracranial volume (TIV) also was used as a covariate.

Multiple statistical analyses were conducted to examine the relationship between multiple metal exposure and health outcome (brain structural and neuropsychological metrics). First, Pearson partial correlation analyses were used to examine associations among exposure measures with adjustments of age as a covariate. For the associations of individual blood metal levels with other exposure measures, other blood metal levels additionally were adjusted. Second, multiple linear regression analyses were conducted. For these analyses, metal exposure measurements (exposure history, blood metal levels, and MRI PI, R1 and R2* values) were treated as predictors with adjustment for age and education and neuropsychological and MRI metrics were treated as outcome variables. MRI metrics also were used as predictors for neuropsychological scores as outcome variables. Third, we explored metal mixture effects using Bayesian kernel machine regression (BKMR) analyses (Bobb et al. 2015) to provide flexible modeling of non-linear associations between multiple metal predictors and health outcome variables (MRI/cognition). These analyses allow testing the overall metal mixture and its pairwise interaction effects in addition to single exposure variable effects. Lastly, we also explored mediation effects using causal mediation analyses (Valeri and Vanderweele 2013) to test whether neuropsychological metrics were linked to metal exposure measurements either directly, or indirectly via MTL structural metrics, with age and education level included as covariates (Supplementary eFigure S1). To test mediation effects of blood metal on neuropsychological metrics, other blood metals additionally were used as covariates. A statistical significance level of α=0.05 was used. Pairwise comparisons in the MANOVA and correlation analyses of neuropsychological and MRI metrics with metal exposure measures were corrected for multiple comparisons using the Stepdown Bonferroni method (Holm 1979) to control the family-wise error rate (FWER). We report uncorrected raw p-values but indicate significant results after FWER-correction. SAS 9.4 or R was used for statistical analyses.

Data Availability

Data is available upon request.

Results

Demographics and exposure features

Exposed subjects were older (p=0.048) and had lower education years (p<0.001) than controls. There was no significant group difference in UPDRS-III or MoCA scores (p=0.326 and p=0.797, respectively; Table 1.A). Exposed subjects displayed overall significantly greater exposure history and higher whole blood metal levels [Pillai’s Trace=0.547, F(12, 58)=4.00, p<0.001]. Blood Cu, Fe, K, Mn, Pb, and Se levels remained significant after FWER-correction (F’s>9.15, R2’s>0.117, p’s<0.004). There were no group differences in both R1 and R2* values of any ROI as well as in PI (p’s>0.442; Table 1.B).

Table 1.

Summary statistics for demographics (a) and overall exposure measures (b) for controls and exposed subjects (welders)

Controls (N=31) Exposed subjects (N=42) Raw p-values
A. Demographics
Age (y) 43.6 ± 11.4 48.9 ± 10.7 0.048
Education (y) 16.2 ± 2.2 12.9 ± 1.6 < 0.001
MoCA 26.2 ± 2.5 26.3 ± 2.1 0.797
UPDRS-III 1.5 ± 2.1 2.0 ± 2.5 0.326
B. Metal exposure measures
 Overall metal exposure history MANOVA: Pillai’s Trace’s=0.791, F=63.37, p=<0.001
HrsW90 (hours) 0 ± 0 241 ± 200 < 0.001
E90 (mg-days/m3) 0.003 ± 0 2.3 ± 2.0 < 0.001
YrsW (y) 0 ± 0 26.2 ± 10.9 < 0.001
ELT (mg-y/m3) 0.001 ± 0.0003 1.1 ± 0.7 < 0.001
 Overall blood metal levels MANOVA: Pillai’s Trace’s=0.453, F=4.00, p=<0.001
Ca (mg/L) 57.8 ± 8.75 58.4 ± 9.16 0.758
Cr (μg/L) 7.91 ± 20.5 29.1 ± 148 0.443
Cu (μg/L) 749 ± 133 894 ± 124 < 0.001*
Fe (mg/L) 498 ± 76 556 ± 53 <0.001*
K (mg/L) 1,770 ± 340 2,090 ± 390 0.001*
Mg (mg/L) 35.7 ± 5.17 38.4 ± 6.48 0.060
Mn (μg/L) 8.9 ± 2.5 11.0 ± 3.2 0.004*
Na (mg/L) 2,040 ± 370 2,190 ± 390 0.121
Pb (μg/L) 8.5 ± 4.8 22.7 ± 18.5 < 0.001*
Se (μg/L) 149 ± 69.8 220 ± 105 0.003*
Sr (μg/L) 13.9 ± 5.17 14.0 ± 6.26 0.907
Zn (μg/L) 5.62 ± 1.57 6.44 ± 1.47 0.019
 MRI T1 metrics
PI 109.02 ± 1.85 109.66 ± 2.50 0.442
MANOVA: Pillai’s Trace’s=0.01, F=0.13, p=0.970
Pallidal R1 0.87 ± 0.07 0.89 ± 0.06 0.664
Parahippocampal R1 0.64 ± 0.06 0.66 ± 0.08 0.610
Entorhinal R1 0.56 ± 0.05 0.57 ± 0.04 0.513
Hippocampal R1 0.51 ± 0.04 0.52 ± 0.04 0.609
MRI T2 metrics MANOVA: Pillai’s Trace’s=0.04, F=0.68, =0.606
Pallidal R2* 36.01 ± 5.19 36.72 ± 4.65 0.827
Parahippocampal R2* 21.41 ± 3.43 21.43 ± 3.74 0.924
Entorhinal R2* 28.71 ± 8.96 29.40 ± 9.03 0.671
Hippocampal R2* 17.28 ± 2.14 18.40 ± 2.99 0.150
Open in a new tab

Data represent the mean ± SD for each measure. Groups were compared using one-way analysis of variance (ANOVA). Metal exposure measures were compared using multivariate analysis of variance (MANOVA). Abbreviations: y = years; MoCA =Montreal Cognitive Assessment; UPDRS = Unified PD Rating Scale; HrsW90 =Hours welding, 90 days; E90 = Cumulative 90 day exposure; YrsW =Years welding, lifetime; ELT = Cumulative exposure, lifetime;

*

indicates significant result at FWER=0.05.

Group comparison of medial temporal MRI and neuropsychological metrics

There were no group differences in morphological metrics including hippocampal volume and cortical thickness in entorhinal and parahippocampal cortices (p’s>0.117; Figure 2.A.-B). Exposed subjects displayed higher diffusivity (MD, AxD, and RD) throughout all ROIs (Pillai’s Trace’s <0.768, F’s>6.36, p’s<0.001; Figure 2.C.-E). The overall group differences in FA values were not significant (Pillai’s Trace=0.918, F(3, 66)=1.95, p=0.131), although exposed subjects had lower FA values in entorhinal and parahippocampal cortices (F’s>4.75, R2’s>0.161, p’s <0.033; Figure 2.F).

Figure 2.

Figure 2.

Open in a new tab

Morphologic MRI measures A. hippocampal volume; B. cortical thickness in entorhinal and parahippocampal cortices and diffusion tensor imaging (DTI); C. mean (MD); D. axial (AD); E. radial (RD) diffusivity; and F. fractional anisotropy (FA) of medial temporal structures for exposed subjects (welders) and controls; * indicates significant result at FWER=0.05.

Exposed subjects trended towards overall lower cognitive performance than controls (p=0.063; Table 2). In domain-wise analyses, exposed subjects performed worse in processing/psychomotor speed, executive function, and visuospatial processing (Pillai’s Trace <0.133, F’s >2.57, p’s <0.046), but similarly on language (p=0.487) and attention/working memory (p=0.126) tests. They also trended to perform worse on learning/memory tasks (p=0.055). For individual subtests, exposed subjects performed worse on Stroop-Color, Stroop-Word, and Symbol Search processing/psychomotor speed and Symbol-Digit Coding, Phonemic Fluency, and Trail Making-B executive function tasks (F’s>4.70, R2’s>0.063, p’s<0.034). They also scored lower on Complex Figure-Copy visuospatial processing and Delayed Story Recall learning/memory tasks (F’s>4.00, R2’s>0.083, p’s <0.049). There were no significant group differences in the remaining individual subtests (p’s >0.131).

Table 2.

Neuropsychological test results for controls and exposed subjects

Test Controls (n=31) Exposed subjects (n=42) P-value
Overall Cognition 0.063
Processing/Psychomotor speed: MANCOVA: Pillai’s Trace’s=0.133, F=2.57, p=0.046
Stroop-Colorb −0.35 ± 1.16 −0.80 ± 1.03 0.034
Stroop-Wordb −0.03 ± 1.05 −0.84 ± 1.19 0.004*
Symbol Searcha 0.88 ± 0.99 0.43 ± 0.67 0.011*
Trail Making-Ab 0.69 ± 0.79 0.50 ± 0.75 0.203
Executive function: MANCOVA: Pillai’s Trace’s=0.261, F=6.02, p=<0.001
Color-wordb 0.12 ± 0.58 0.06 ± 0.70 0.512
Symbol-Digit Codinga 0.46 ± 1.06 −0.10 ± 0.78 0.010*
Phonemic Fluencyc 0.36 ± 0.87 −0.31 ± 0.73 <0.001*
Trail Making-Bb 0.51 ± 0.83 −0.35± 1.41 0.004*
Language: MANCOVA: Pillai’s Trace’s=0.021, F=0.73, p=0.487
Picture Naminga 0.24 ± 1.13 0.59 ± 0.24 0.468
Semantic Fluencya 0.04 ± 1.06 −0.23 ± 0.99 0.365
Visuospatial Processing: MANCOVA: Pillai’s Trace’s=0.086, F=3.24, p=0.045
Judgement of Line Orientationa 0.69 ± 0.94 0.55 ± 0.82 0.858
Complex Figure Copya −0.22 ± 0.86 −1.09 ± 1.35 0.013*
Learning/Memory: MANCOVA: Pillai’s Trace’s=0.148, F=2.30, p=0.055
List Learninga −0.64 ± 0.93 −0.76 ± 1.00 0.493
Immediate Story Recalla 0.06 ± 0.93 −0.42 ± 1.04 0.151
Delayed List Recalla −0.73 ± 0.98 −0.52 ± 1.01 0.192
Delayed Story Recalla 0.13 ± 0.90 −0.46 ± 1.04 0.049
Delayed Figure Recalla −0.13 ± 1.10 0.01 ± 1.16 0.261
Attention/working memory: MANCOVA: Pillai’s Trace’s=0.080, F=1.98, p=0.126
Letter Number Sequencinga 0.30 ± 1.00 0.03 ± 0.94 0.906
Digit Spana 0.81 ± 1.11 0.09 ± 0.83 0.131
Spatial Spana 0.34 ± 0.65 −0.10 ± 0.93 0.176
Open in a new tab

Data represents the mean ± SD. Test scores were converted to z-scores based on age- (denoted as a), age- and education- (denoted as b), or age-, education-, and race- (denoted as c) adjusted norm (t- or scaled) scores. Higher scores indicate better performance. Multivariate analysis of covariance (MANCOVA) was conducted with Group as the independent variable and individual subtests of each cognitive domain as the dependent variables. Bold numbers indicate significant result at p<0.05. Education level was additionally adjusted for the cognitive domains language, visuospatial processing, learning/memory, and attention/working memory; * indicates significant result at FWER=0.05.

Associations among exposure measurements in exposed subjects

Higher HrsW values were correlated with lower hippocampal R2* values (R=−0.313, p=0.047). Metal exposure history was not correlated with any blood metal levels or any remaining MRI R1 and R2* measures (−0.288<R’s <0.278; p’s >0.068; Supplementary eTable S1a–b). Several blood metal levels were associated with MRI R1 metrics: Higher blood Cu levels were associated with lower R1 values in the entorhinal and parahippocampal cortices (R=−0.456, p=0.007 and R=−0.464, p=0.006, respectively). Higher blood Fe levels were associated with lower R1 values in all ROIs and PI (R’s < −0.348, p’s<0.044). Higher blood K levels were associated with higher parahippocampal R1 values (R=0.395, p=0.023). Higher blood Mn levels were associated with higher R1 values in the entorhinal and parahippocampal cortices (R=0.395, p=0.021 and R=0.372, p=0.030, respectively). Higher blood Se levels were associated with lower entorhinal R1 values (R=−0.385, p=0.025; Supplementary eTable S1b). The negative associations of blood Cu and Fe with entorhinal and parahippocampal R1 values remained significant after FWER correction. There were no significant correlations between blood metal levels and R2* values in any ROI (R’s < 0.272, p’s>0.119; Supplementary eTable S1b).

The associations of metal exposure with MRI and neuropsychological metrics in exposed subjects

Metal exposure history as the exposure of interest

Diffusion MRI:

Multiple linear regression analysis showed that greater YrsW was a significant predictor of lower parahippocampal FA (ß=−0.001, t=−2.61, p=0.013; Table 3.A and Figure 3.A). Higher ELT was a significant predictor of higher hippocampal RD (ß=0.021, t=2.06, p=0.046; Table 3.A).

Table 3.

Multiple linear regression analyses evaluating effects of exposure measures on medial temporal MRI and neurobehavioral metrics for exposed subjects

Outcome variables Selected Predictors Linear ß t-value p-value
Exposed subjects (N=42)
    A. Exposure measures: metal exposure history
Parahippocampal FA YrsW −0.001 −2.61 0.013
Hippocampal RD ELT 0.021 2.06 0.046
Symbol-Digit Coding E90 −0.163 −2.54 0.015
    B. Exposure measures: blood metal levels
Entorhinal MD Mn 0.163 2.14 0.040
Entorhinal AD Mn 0.180 2.16 0.038
Entorhinal RD Mn 0.153 2.11 0.043
Entorhinal FA Pb 0.013 2.13 0.040
Hippocampal AD Fe −0.337 −2.33 0.027
Phonemic Fluency Pb −0.572 −2.13 0.041
Trail Making-B Zn 4.657 2.78 0.009
List Learning Fe 4.713 2.41 0.022
Immediate Story Cu −3.779 −2.50 0.018
Recall
Delayed Story Recall Cu −3.881 −2.87 0.007
    C. Exposure measures: brain metal accumulation (MRI PI, R1, & R2*)
Entorhinal MD Entorhinal R1 1.048 2.07 0.045
Entorhinal R2* −0.006 −3.54 0.001
Entorhinal AD Entorhinal R2* −0.006 −3.02 0.005
Entorhinal RD Entorhinal R1 1.047 2.21 0.034
Entorhinal R2* −0.006 −3.82 <0.001
Entorhinal FA Entorhinal R2* <0.001 3.87 <0.001
Parahippocampal R2* 0.002 2.17 0.036
Parahippocampal MD Entorhinal R2* −0.003 −2.30 0.027
Parahippocampal RD Entorhinal R2* −0.003 −2.58 0.014
Parahippocampal FA Entorhinal R2* 0.001 3.45 0.001
Stroop-Color Parahippocampal R1 6.291 2.71 0.010
Symbol Search Parahippocampal R1 4.135 2.72 0.010
Trail Making-B Parahippocampal R1 8.432 2.53 0.016
Delayed List Recall Entorhinal R2* 0.036 2.47 0.019
Open in a new tab

Data represent parameter estimates and corresponding t- and p-values of exposure measurements to predict MTL MRI and neurobehavioral metrics.

Figure 3.

Figure 3.

Open in a new tab

Scatter plots show A. adjusted parahippocampal FA values (y-axis) vs. YrsW (x-axis); B. adjusted Symbol-Digit Coding scores (y-axis) vs. E90 (x-axis) in exposed subjects (welders); C. adjusted of entorhinal AD (y-axis) vs. log-transformed blood Mn (x-axis); D. adjusted Delayed Story Recall (y-axis) vs. log-transformed blood Cu (x-axis) in both exposed subjects (welders) and controls. Adjusted values indicate values after controlling for age and education level.

Neuropsychological tests:

Multiple linear regression analysis revealed that higher E90 was a significant predictor of lower Symbol-Digit Coding scores (ß=−0.163, t=−2.54, p=0.015; Table 3.A and Figure 3.B).

Whole blood metal levels as the exposure of interest

Diffusion MRI: multiple linear regression analysis revealed that higher blood Mn was a significant predictor of higher entorhinal diffusivities (ß’s>0.153, t’s>2.11, p’s <0.043; Table 3.B and Figure 3.C). Higher blood Pb significantly predicted higher entorhinal FA values (ß=0.013, t=2.13, p=0.040), whereas higher Fe level predicted lower hippocampal AxD (ß=−0.337, t=−2.33, p=0.027).

Neuropsychological tests:

Multiple linear regression analyses revealed that higher blood Pb levels were a significant predictor of poorer Phonemic Fluency (ß=−0.572, t=−2.13, p=0.041; Table 3.B). Lower blood Zn and Fe were significant predictors of lower Trail Making-B and List Learning scores (ß=4.657, t=2.78, p=0.009 for Zn; ß=4.713, t=2.41, p=0.022 for Fe). Higher blood Cu was a significant predictor of poorer Immediate and Delayed Story Recall (ß=−3.779, t=−2.50, p=0.018 and ß=−3.881, t=−2.87, p=0.007, respectively; Table 3B and Figure 3.D).

Metal mixture effects using BKMR analyses:

When examining single metal exposure effects on brain DTI and neuropsychological metrics, higher blood Mn levels significantly predicted higher entorhinal AxD, MD, and RD values (p’s <0.05). Higher blood Fe levels were a significant predictor of better List Learning (p<0.05), whereas higher blood Cu levels were a significant predictor of poorer Delayed Story Recall (p<0.05; Supplementary eFigure S2). None of the blood metal levels were significant predictors of hippocampal DTI metrics.

When examining overall effects of metal mixture, there were no significant overall metal mixture effects on MTL DTI or neuropsychological metrics (p’s >0.05; Supplementary eFigure S3). When examining pairwise interaction effects among multiple metals, none of pairwise interactions among multiple metals were significant (p’s>0.05; Supplementary eFigure S4).

Brain metal accumulation metrics (MRI PI, R1 and R2*) as the exposure of interest

Diffusion MRI:

Multiple linear regression analyses revealed that higher entorhinal R1 but lower R2* values predicted higher entorhinal MD and RD values (ß’s >1.05, t’s >2.07, p’s <0.045 for R1 and ß’s <−0.006, t’s <−3.54, p’s <0.001 for R2*). Higher entorhinal R2* values predicted lower entorhinal AxD (ß=−0.006, t=−3.02, p=0.005 for AxD). Higher entorhinal and parahippocampal R2* values predicted higher entorhinal FA values (ß=0.001, t=3.87, p<0.001 for entorhinal; ß=0.002, t=2.17, p=0.036 for parahippocampal). Higher entorhinal R2* values predicted higher parahippocampal MD and RD but lower FA values (ß’s <0.003, t’s <2.30, p’s<0.027 for MD and RD; ß=0.001, t=3.45, p=0.001 for FA; Table 3C).

Neuropsychological tests:

Multiple linear regression analyses revealed that higher parahippocampal R1 values predicted lower performance on Stroop-Color, Symbol Search, and Trail Making-B tasks (ß’s >4.135, t’s >2.71, p’s<0.016). Higher entorhinal R1* values predicted better performance on the Delayed List Recall task (ß=0.0036, t=2.47, p=0.019)

The associations of MTL DTI with neuropsychological test scores

Multiple linear regression analyses revealed that higher entorhinal diffusivities were significant predictors of Delayed List Recall scores (ß’s<−3.594, t’s<−2.74, p’s <0.010; Table 4 and Figure 4a). Higher parahippocampal diffusivities and higher entorhinal MD were significant predictors of poorer Immediate Story Recall scores (ß’s<−3.426, t’s<−2.08, p’s <0.044; Figure 4.B). Lower parahippocampal FA values were a significant predictor of lower Symbol Search and Trail Making-A scores (ß=16.058, t=2.39, p=0.022 and ß=16.908, t=2.26, p=0.030, respectively; Figure 4.C-D). Higher hippocampal AxD values were a significant predictor of lower List Learning scores (ß=−4.728, t=−2.18, p=0.035; Figure 4.E).

Table 4.

Multiple linear regression analyses evaluating effects of medial temporal structural MRI on neurobehavioral metrics for exposed subjects

Outcome variables Selected Predictors Linear ß t-value p-value
Exposed subjects (N=42)
MTL MRI metrics
Symbol Search Parahippocampal FA 16.058 2.39 0.022
Trail Making-A Parahippocampal FA 16.908 2.26 0.030
Immediate Story Entorhinal MD −3.426 −2.08 0.044
Recall
Parahippocampal MD −6.542 −2.95 0.005
Parahippocampal AD −6.604 −3.11 0.004
Parahippocampal RD −7.561 −3.44 0.002
List Learning Hippocampal AD −4.728 −2.18 0.035
Delayed List Recall Entorhinal MD −4.076 −2.97 0.005
Entorhinal AD −3.594 −2.95 0.006
Entorhinal RD −3.707 −2.74 0.010
Open in a new tab

Data represent parameter estimates and corresponding t- and p-values of MTL MRI metrics to predict neurobehavioral metrics.

Figure 4.

Figure 4.

Open in a new tab

Scatter plots show A. adjusted Delayed List Recall scores (y-axis) vs. adjusted entorhinal MD values (x-axis); B. adjusted Immediate Story Recall scores (y-axis) vs. adjusted parahippocampal RD values (x-axis); C. Symbol Search scores (y-axis) vs. adjusted parahippocampal FA values (x-axis); D. adjusted Trail Making-A scores (y-axis) vs. adjusted parahippocampal FA values (x-axis); E. adjusted List Learning scores (y-axis) vs. adjusted hippocampal AD values (x-axis) in both exposed subjects (welders) and controls.

Causal mediation analyses: from exposure to neurobehavior via MTL MRI feature.

Metal exposure history as the exposure of interest

There were significant indirect effects of higher YrsW on lower Symbol Search and Trail Making-A scores that were mediated by lower parahippocampal FA values (ß=−0.025, z=−2.29, p=0.022 for Symbol Search and ß=−0.028, z=−2.27, p=0.023 for Trail Making-A) but no significant direct or total effects (p’s>0.055). There were significant total and direct effects of higher E90 values on lower Symbol-Digit Coding scores (ß’s<−0.153, z’s<−2.44, p’s<0.014) but no significant indirect effects (p’s>0.647). There were no significant direct or indirect effects of other metal exposure history on neuropsychological scores (p’s>0.065).

Whole blood metal levels as the exposure of interest

There were significant total (ß=6.401, z=3.85, p<0.001) and indirect effects of higher blood Fe on higher List Learning scores via lower hippocampal AxD (ß=7.567, z=2.11, p=0.035) without significant direct effects (p=0.726). Blood Mn levels had significant total effects (ß’s<−1.230, z’s <2.24, p’s <0.025) on Delayed List Recall scores without significant direct or indirect effects via MTL DTI metrics (p’s>0.109). Blood Cu levels had significant total and direct effects (ß’s<−3.1633, z’s<−1.97, p’s<0.049) on lower Symbol-Digit Coding and Delayed Story Recall scores without significant indirect effects via MTL DTI metrics (p’s>0.366). Blood Pb levels had significant total and direct effects (ß’s<−0.614, z’s<−2.23, p’s<0.026) on lower Phonemic Fluency scores without significant indirect effects via MTL DTI metrics (p’s>0.366).

Brain metal accumulation metrics (MRI PI, R1 and R2*) as the exposure of interest

There were significant indirect effects of higher entorhinal R2* values on higher Symbol Search via higher parahippocampal FA values (ß=0.022, z=2.34, p=0.019) or via higher parahippocampal MD and RD values (ß’s <0.019, z’s>1.96, p’s<0.049) without significant total or direct effects (p’s >0.050). There were significant total and direct effects of parahippocampal R1 values on Symbol Search and Trail Making-B scores (ß’s>4.377, z’s>2.12, p’s<0.034).

Discussion

This is the first in-depth examination of the relationship between mixed metal exposure, neuropsychological status, and brain structural features in several MTL regions assessed via multimodal MRI. We found significant differences in MRI diffusion features (higher AxD, RD, and MD and lower FA) across MTL areas, MRI features often observed in AD at-risk populations. Along with the absence of morphometric differences (markers of significant macroscopic neuronal loss), these findings suggest that these MTL DTI metrics may serve as biomarkers for gauging early pathological processes that may reflect changes in microstructural integrity. Subjects with mixed metal exposure had lower performance in select processing/psychomotor speed and executive function domains, as well as the Delayed Story Recall task, despite having no detectable deficits in the broader learning/memory domain.

This is also the first comprehensive systematic analysis of metal exposure-brain health outcome relationships. Our data predict that long-term mixed metal exposure history (YrsW and ELT) is related to altered MTL DTI metrics (lower parahippocampal FA and higher hippocampal RD). We also observed robust associations of higher whole blood Mn and Cu levels with higher entorhinal diffusivity and lower Delayed Story Recall performance. These features often are seen in populations with AD and related disorders. MTL DTI metrics (reflecting microstructural differences) could mediate, at least partially, the effects of mixed metal exposure on cognitive performance.

MTL features in exposed subjects resembles Alzheimer’s disease at-risk populations

The hippocampus and entorhinal/parahippocampal cortices are key structures for learning/memory tasks. The parahippocampal gyrus receives converging inputs from widespread associative areas including frontal, temporal, parietal, occipital, and cingulate cortices (Aminoff et al. 2013). It projects to the entorhinal cortex that, in turn, has reciprocal connections with the hippocampus. Thus, the entorhinal cortex serves as a major information relay station to the hippocampus and then back to the cortical association areas via the parahippocampal gyrus (Wixted and Squire 2011) for longer term memory storage. These pathways parallel the fimbria-fornix system of the hippocampal projections to diencephalic and other limbic system structures particularly subserving related memory consolidation processes (Eichenbaum and Lipton 2008). The entorhinal outputs then are projected back to various cortical regions through the parahippocampal cortex.

Consistent with its role in learning/memory, MTL areas appear to be among the first regions affected by AD-related pathological changes (Devanand et al. 2007; Echavarri et al. 2011). Decreased hippocampal and entorhinal volume and/or thickness have been reported as predictors for conversion to AD (Apostolova et al. 2006). Recent studies suggested other imaging metrics such as shape, texture, functional networks, or water molecule diffusion in neuronal tissues may be even more sensitive in capturing early AD-related alterations (Leandrou et al. 2020; Weston et al. 2020). A recent study noted altered cortical MRI diffusion metrics in Aβ-positive subjects while they had no clear atrophy in the areas of Aβ accumulation (Spotorno et al. 2023). The chronic but low-level characteristics of neurotoxin exposure (as our asymptomatic welders in this study) may provide one explanation why there were no overt morphological differences in MTL regions detected in those previous studies. The robust group differences in MTL DTI metrics, however, support the idea that MTL may be vulnerable to neurotoxic metal exposures and microstructural changes.

Neuropsychological features in exposed subjects resembles Alzheimer’s disease at-risk populations

Subjects with chronic metal exposure are known to perform worse on several neuropsychological tests (Chang et al. 2009; Meyer-Baron et al. 2013). AD-related neuropsychological deficits comprise various cognitive functions including executive, attentional, and language functions. Difficulties with learning/memory tasks, however, have the most robust association with early AD and related disorders (Backman et al. 2001; Tromp et al. 2015). Learning and memory are crucial functions of the hippocampus (Tulving and Markowitsch 1998). MTL areas, such as parahippocampal and entorhinal cortices, are known to provide major inputs to the hippocampus (Eichenbaum and Lipton 2008). Previous studies reported that lower entorhinal cortex volume, but not hippocampal volume, was associated with lower Delayed List and Story Recall scores in Alzheimer’s patients (Di Paola et al. 2007; Eichenbaum and Lipton 2008). Furthermore, MTL DTI metrics have been associated with memory scores in both healthy elderly and Alzheimer’s patients (Carlesimo et al. 2010; Mayo et al. 2018). Together, these prior studies suggest MTL involvement in learning/memory function and early AD processes (Di Paola et al. 2007). Our findings of robust associations of higher entorhinal and parahippocampal diffusivities with lower (worse) learning/memory performance are consistent with this notion.

In the current study, our mixed metal-exposed subjects also displayed lower processing/psychomotor speed performance. Speedy information processing may be crucial for various cognitive functions requiring attention, encoding and retrieval of to-be-remembered information, reasoning, decision making, and visuospatial perceptions. Several brain areas (e.g., superior/middle/inferior frontal gyri and superior parietal regions) were reported to be involved in processing/psychomotor speed (Hong et al. 2015; Kraft et al. 2020). Significant associations between processing speed and morphological metrics in MTL regions (e.g., hippocampal volume and parahippocampal gyrus thickness) also have been reported, particularly when working with visuospatial materials (O’Shea et al. 2016; Takahashi et al. 2002). This implies that slowed psychomotor speed may serve as an early marker for dementia (also see Andriuta et al. 2019; Kuate-Tegueu et al. 2017). In this respect, it is possible that our current results of lower processing/psychomotor speed performance and its associations with MTL DTI metrics (e.g., parahippocampal FA values) may represent an early sign of AD risk. Collectively, our overall data supports the idea that mixed metal exposure may be associated with cognitive deficits/patterns observed in AD at-risk populations (Backman et al. 2001; Douaud et al. 2013).

From mixed metal exposure to MTL MRI and neuropsychological performance

Recent studies have emphasized the importance of considering individual and collective roles of metal mixtures in neurotoxic processes (Wu et al. 2023). In the present study, we utilized questionnaire-based instruments to assess mixed metal exposure history, whole blood metal, and MRI R1 and R2* metrics to estimate metal exposure dose and multiple statistical methods to explore complex dynamics of metal mixtures with MTL MRI and neuropsychological performance.

First, we found that long-term metal exposure history (e.g., ELT and YrsW) provided significant predictors of hippocampal and parahippocampal DTI metrics, consistent with the notion that MTL microstructural features are affected by chronic metal exposure. Subsequent mediation analyses suggested that YrsW indirectly affects lower processing/psychomotor speed performance (lower Symbol Search and Trail Making-A scores) based on lower parahippocampal FA values.

Second, we noted that short-term metal exposure history that accounts for exposure intensity (i.e., E90) predicted lower Symbol-Digit Coding scores, an executive function task in which subjects matched symbols to designated numbers under time pressure. This task can be demanding and requires significant executive function skills, a challenge particularly for elderly adults. Our data suggest that higher-intensity short-term metal exposure history interferes with proper performance on executive function tasks.

Third, higher entorhinal R1 and whole blood metal levels (e.g., blood Mn and Cu) demonstrated robust associations with higher entorhinal diffusivity and lower episodic memory performance (lower Delayed Story Recall scores). Although associations between metal exposures (e.g., exposure history, blood Mn, or MRI T1W or R2* metrics reflecting brain Mn or Fe, respectively) and altered MRI water diffusion metrics (e. g., lower DTI FA or higher diffusivity) were previously examined in metal exposed subjects (e.g., welders), these studies either limited regions of interest to basal ganglia (BG), white matter structures (e.g., corpus callosum and frontal white matter areas) or hippocampus (Criswell et al. 2012; Kim et al. 2011; Lee et al. 2016; Lee et al. 2019; Lee et al. 2023). Our finding of the association of whole blood Mn and entorhinal diffusivity is a novel addition to the literature. Our mediation analyses also suggested a significant link between higher blood Mn and lower learning/memory performance (lower Delayed List Recall scores), although the exact characteristic of the link (direct or indirect via MTL microstructural features) was inconclusive. This may be due partly to the small sample size and also to the fact that whole blood metal concentrations do not sensitively reflect long-term exposure effects.

The significant association between higher whole blood Cu and lower memory (lower Delayed Story Recall scores) in subjects with multiple metal exposure is also new. Chronic Cu exposure has been associated with increased brain Cu accumulation in the BG and hippocampus (Pal et al. 2013) and increased risk for AD (Patel and Aschner 2021). Cu in brain is known to have a high affinity for β-amyloid plaques, and may enhance inflammatory responses, reactive oxygen species (ROS) generation, and fibril formation leading to increased β-amyloid aggregates, a characteristic feature of AD and related disorders (Bagheri et al. 2017). Future studies need to include exposure to Cu as an important risk factor for neurodegenerative disorders.

Finally, our results also revealed intriguing but contradictory findings: For exposed subjects, higher entorhinal R2* values predicted lower entorhinal and parahippocampal diffusivities and higher scores on Delayed List Recall task; Higher parahippocampal R1 or blood Zn and Fe levels predicted better performance on processing/psychomotor speed performance (lower Stroop-Color and Symbol Search scores), Trail Making-B and List Learning tasks, yet the performance levels on those tasks were lower or comparable to those of controls. Higher whole blood Pb levels also predicted higher entorhinal FA values while being associated with lower executive function performance (e.g., Phonemic Fluency scores). These findings suggest complex dynamics of blood metal levels with other co-exposed metals in competing or synergizing for brain entry and subsequent brain metal accumulation and structural and cognitive metrics.

Conclusions

The present results support that mixed metal exposure is associated with MTL microstructural and neuropsychological features that are observed in AD at-risk populations. Given the ubiquitous nature of environmental mixed metal exposures and growing public health concerns of its link to AD, follow-up studies are warranted. Exposure to environmental metals is a controllable factor, and a clear understanding of how a mixed metal exposure affects higher brain processes may decrease the risk or degree of cognitive damage and AD and/or other age-related neurogenerative disorders. The current findings suggest future studies need to consider both short- and long-term mixed metal exposure history along with exposure intensity, collective dynamics of metal mixtures, as well as separately analyzing selective brain regions and metrics for cognitive outcomes.

Supplementary Material

1
2

Highlights.

  • Exposed subjects had higher diffusion tensor imaging (DTI) mean (MD), axial (AxD), radial (RD) diffusivity values in all medial temporal lobe (MTL) regions of interest (ROI) (hippocampus, entorhinal, and parahippocampal cortices) and lower fractional anisotropy (FA) in the entorhinal and parahippocampal cortices without significant morphologic differences.

  • Long-term mixed metal exposure history predicted altered MTL DTI metrics (lower parahippocampal FA and higher hippocampal RD).

  • Across multiple linear and Bayesian kernel machine regression analyses, higher entorhinal MRI R1 (1/T1; estimate of brain Mn) and whole blood Mn and Cu levels predicted higher entorhinal diffusivity values and lower Delayed Story Recall performance, features that are similar to Alzheimer’s disease at-risk populations.

  • MTL DTI metrics mediate, at least partially, the effects of metal exposure on cognitive performance.

Acknowledgements

We would like to thank Dr. Mike Flynn for conceptualization at the beginning of the study and volunteers who participated in this study. In addition, we are indebted to many individuals who helped make this study possible, including: Melissa Santos, Tyler Corson, Lauren Deegan, and Susan Kocher for subject coordination, recruitment, blood sample handling, and data entry. Important information and help in recruiting came from Pam Susi and Pete Stafford of CPWR; Mark Garrett, John Clark, and Joe Jacoby of the International Brotherhood of Boilermakers; Fred Cosenza and all members of the Safety Committee for the Philadelphia Building and Construction Trades Council; Ed McGehean of the Steamfitters Local Union 420; Jim Stewart of the Operating Engineers; Sean Gerie of the Brotherhood of Maintenance of Way Employees Division Teamsters Rail Conference; and Terry Peck of Local 520 Plumbers, Pipefitters and HVAC.

Funding

This work was supported by NIH grants R01 ES019672, R01 NS060722, U01s NS082151 and NS112008, the Hershey Medical Center General Clinical Research Center (National Center for Research Resources, UL1 TR002014), the Penn State College of Medicine Translational Brain Research Center, the PA Department of Health Tobacco CURE Funds, Basic Science Research Program through the National Research Foundation of Korea [2019R1G1A109957511; Ministry of Education (RS-2023-00247689)].

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Conflicts of interest

The authors report no financial interests that relate to this research.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References

  1. Aminoff EM, Kveraga K, Bar M. 2013. The role of the parahippocampal cortex in cognition. Trends Cogn Sci 17(8):379–390. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Andriuta D, Diouf M, Roussel M, Godefroy O. 2019. Is reaction time slowing an early sign of alzheimer’s disease? A meta-analysis. Dement Geriatr Cogn Disord 47(4–6):281–288. [DOI] [PubMed] [Google Scholar]
  3. Apostolova LG, Dutton RA, Dinov ID, Hayashi KM, Toga AW, Cummings JL, Thompson PM. 2006. Conversion of mild cognitive impairment to alzheimer disease predicted by hippocampal atrophy maps. Arch Neurol 63(5):693–699. [DOI] [PubMed] [Google Scholar]
  4. Bäckman L, Jones S, Berger A-K, Laukka EJ, Small BJ. 2005. Cognitive impairment in preclinical alzheimer’s disease: A meta-analysis. Neuropsychology. 19(4):520. [DOI] [PubMed] [Google Scholar]
  5. Backman L, Small BJ, Fratiglioni L. 2001. Stability of the preclinical episodic memory deficit in alzheimer’s disease. Brain. 124(Pt 1):96–102. [DOI] [PubMed] [Google Scholar]
  6. Bagheri S, Squitti R, Haertle T, Siotto M, Saboury AA. 2017. Role of copper in the onset of alzheimer’s disease compared to other metals. Front Aging Neurosci 9:446. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Banerjee P, Sahoo A, Anand S, Ganguly A, Righi G, Bovicelli P, Saso L, Chakrabarti S. 2014. Multiple mechanisms of iron-induced amyloid beta-peptide accumulation in shsy5y cells: Protective action of negletein. Neuromolecular medicine. 16(4):787–798. [DOI] [PubMed] [Google Scholar]
  8. Bateman RJ, Cummings J, Schobel S, Salloway S, Vellas B, Boada M, Black SE, Blennow K, Fontoura P, Klein G et al. 2022. Gantenerumab: An anti-amyloid monoclonal antibody with potential disease-modifying effects in early alzheimer’s disease. Alzheimers Res Ther 14(1):178. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bjorkli C, Sandvig A, Sandvig I. 2020. Bridging the gap between fluid biomarkers for alzheimer’s disease, model systems, and patients. Frontiers in aging neuroscience. 12:272. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Bobb JF, Valeri L, Claus Henn B, Christiani DC, Wright RO, Mazumdar M, Godleski JJ, Coull BA. 2015. Bayesian kernel machine regression for estimating the health effects of multi-pollutant mixtures. Biostatistics. 16(3):493–508. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Brookmeyer R, Evans DA, Hebert L, Langa KM, Heeringa SG, Plassman BL, Kukull WA. 2011. National estimates of the prevalence of alzheimer’s disease in the united states. Alzheimer’s & Dementia. 7(1):61–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Carlesimo GA, Cherubini A, Caltagirone C, Spalletta G. 2010. Hippocampal mean diffusivity and memory in healthy elderly individuals: A cross-sectional study. Neurology. 74(3):194–200. [DOI] [PubMed] [Google Scholar]
  13. Chang Y, Kim Y, Woo ST, Song HJ, Kim SH, Lee H, Kwon YJ, Ahn JH, Park SJ, Chung IS et al. 2009. High signal intensity on magnetic resonance imaging is a better predictor of neurobehavioral performances than blood manganese in asymptomatic welders. Neurotoxicology. 30(4):555–563. [DOI] [PubMed] [Google Scholar]
  14. Criswell SR, Perlmutter JS, Huang JL, Golchin N, Flores HP, Hobson A, Aschner M, Erikson KM, Checkoway H, Racette BA. 2012. Basal ganglia intensity indices and diffusion weighted imaging in manganese-exposed welders. Occupational and environmental medicine. 69(6):437–443. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Crous-Bou M, Gascon M, Gispert JD, Cirach M, Sánchez-Benavides G, Falcon C, Arenaza-Urquijo EM, Gotsens X, Fauria K, Sunyer J. 2020. Impact of urban environmental exposures on cognitive performance and brain structure of healthy individuals at risk for alzheimer’s dementia. Environment international. 138:105546. [DOI] [PubMed] [Google Scholar]
  16. Delgado-Saborit JM, Guercio V, Gowers AM, Shaddick G, Fox NC, Love S. 2021. A critical review of the epidemiological evidence of effects of air pollution on dementia, cognitive function and cognitive decline in adult population. Science of the Total Environment. 757:143734. [DOI] [PubMed] [Google Scholar]
  17. Devanand DP, Pradhaban G, Liu X, Khandji A, De Santi S, Segal S, Rusinek H, Pelton GH, Honig LS, Mayeux R et al. 2007. Hippocampal and entorhinal atrophy in mild cognitive impairment: Prediction of alzheimer disease. Neurology. 68(11):828–836. [DOI] [PubMed] [Google Scholar]
  18. Di Paola M, Macaluso E, Carlesimo GA, Tomaiuolo F, Worsley KJ, Fadda L, Caltagirone C. 2007. Episodic memory impairment in patients with alzheimer’s disease is correlated with entorhinal cortex atrophy. A voxel-based morphometry study. J Neurol 254(6):774–781. [DOI] [PubMed] [Google Scholar]
  19. Douaud G, Menke RA, Gass A, Monsch AU, Rao A, Whitcher B, Zamboni G, Matthews PM, Sollberger M, Smith S. 2013. Brain microstructure reveals early abnormalities more than two years prior to clinical progression from mild cognitive impairment to alzheimer’s disease. J Neurosci 33(5):2147–2155. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Echavarri C, Aalten P, Uylings HB, Jacobs HI, Visser PJ, Gronenschild EH, Verhey FR, Burgmans S. 2011. Atrophy in the parahippocampal gyrus as an early biomarker of alzheimer’s disease. Brain structure & function. 215(3–4):265–271. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Eichenbaum H, Lipton PA. 2008. Towards a functional organization of the medial temporal lobe memory system: Role of the parahippocampal and medial entorhinal cortical areas. Hippocampus. 18(12):1314–1324. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Fischl B, Van Der Kouwe A, Destrieux C, Halgren E, Ségonne F, Salat DH, Busa E, Seidman LJ, Goldstein J, Kennedy D. 2004. Automatically parcellating the human cerebral cortex. Cerebral cortex. 14(1):11–22. [DOI] [PubMed] [Google Scholar]
  23. Guilarte TR. 2010. Aplp1, alzheimer’s-like pathology and neurodegeneration in the frontal cortex of manganese-exposed non-human primates. Neurotoxicology. 31(5):572–574. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Hare DJ, Faux NG, Roberts BR, Volitakis I, Martins RN, Bush AI. 2016. Lead and manganese levels in serum and erythrocytes in alzheimer’s disease and mild cognitive impairment: Results from the australian imaging, biomarkers and lifestyle flagship study of ageing. Metallomics. 8(6):628–632. [DOI] [PubMed] [Google Scholar]
  25. Holm S 1979. A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics. 6(2):65–70. [Google Scholar]
  26. Hong Z, Ng KK, Sim SK, Ngeow MY, Zheng H, Lo JC, Chee MW, Zhou J. 2015. Differential age-dependent associations of gray matter volume and white matter integrity with processing speed in healthy older adults. Neuroimage. 123:42–50. [DOI] [PubMed] [Google Scholar]
  27. Killiany RJ, Hyman BT, Gomez-Isla T, Moss MB, Kikinis R, Jolesz F, Tanzi R, Jones K, Albert MS. 2002. Mri measures of entorhinal cortex vs hippocampus in preclinical ad. Neurology. 58(8):1188–1196. [DOI] [PubMed] [Google Scholar]
  28. Kim Y, Jeong KS, Song H-J, Lee J-J, Seo J-H, Kim G-C, Lee HJ, Kim HJ, Ahn J-H, Park S-J. 2011. Altered white matter microstructural integrity revealed by voxel-wise analysis of diffusion tensor imaging in welders with manganese exposure. Neurotoxicology. 32(1):100–109. [DOI] [PubMed] [Google Scholar]
  29. Kitazawa M, Hsu HW, Medeiros R. 2016. Copper exposure perturbs brain inflammatory responses and impairs clearance of amyloid-beta. Toxicol Sci 152(1):194–204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Kraft JN, O’Shea A, Albizu A, Evangelista ND, Hausman HK, Boutzoukas E, Nissim NR, Van Etten EJ, Bharadwaj PK, Song H. 2020. Structural neural correlates of double decision performance in older adults. Frontiers in Aging Neuroscience. 12:278. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Kress GT, Popa ES, Thompson PM, Bookheimer SY, Thomopoulos SI, Ching CR, Zheng H, Hirsh DA, Merrill DA, Panos SE. 2023. Preliminary validation of a structural magnetic resonance imaging metric for tracking dementia-related neurodegeneration and future decline. NeuroImage: Clinical. 39:103458. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Kuate-Tegueu C, Avila-Funes JA, Simo N, Le Goff M, Amieva H, Dartigues JF, Tabue-Teguo M. 2017. Association of gait speed, psychomotor speed, and dementia. J Alzheimers Dis 60(2):585–592. [DOI] [PubMed] [Google Scholar]
  33. Lacalle-Aurioles M, Iturria-Medina Y. 2023. Fornix degeneration in risk factors of alzheimer’s disease, possible trigger of cognitive decline. Cereb Circ Cogn Behav 4:100158. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Lancaster MA, Seidenberg M, Smith JC, Nielson KA, Woodard JL, Durgerian S, Rao SM. 2016. Diffusion tensor imaging predictors of episodic memory decline in healthy elders at genetic risk for alzheimer’s disease. Journal of the International Neuropsychological Society : JINS. 22(10):1005–1015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Leandrou S, Lamnisos D, Mamais I, Kyriacou PA, Pattichis CS, Alzheimer’s D, Neuroimaging I. 2020. Assessment of alzheimer’s disease based on texture analysis of the entorhinal cortex. Front Aging Neurosci 12:176. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Leandrou S, Petroudi S, Kyriacou PA, Reyes-Aldasoro CC, Pattichis CS. 2018. Quantitative mri brain studies in mild cognitive impairment and alzheimer’s disease: A methodological review. IEEE Rev Biomed Eng 11:97–111. [DOI] [PubMed] [Google Scholar]
  37. Lee E-Y, Flynn MR, Du G, Lewis MM, Herring AH, Van Buren E, Van Buren S, Kong L, Mailman RB, Huang X. 2016. Editor’s highlight: Lower fractional anisotropy in the globus pallidus of asymptomatic welders, a marker for long-term welding exposure. Toxicological Sciences. 153(1):165–173. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Lee E-Y, Flynn MR, Du G, Lewis MM, Kong L, Yanosky JD, Mailman RB, Huang X. 2019. Higher hippocampal mean diffusivity values in asymptomatic welders. Toxicological Sciences. 168(2):486–496. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Lee E-Y, Flynn MR, Lewis MM, Mailman RB, Huang X. 2018. Welding-related brain and functional changes in welders with chronic and low-level exposure. Neurotoxicology. 64:50–59. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Lee E-Y, Kim J, Prado-Rico JM, Du G, Lewis MM, Kong L, Kim B-G, Hong Y-S, Yanosky JD, Mailman RB. 2023. Higher hippocampal diffusivity values in welders are associated with greater r2* in the red nucleus and lower psychomotor performance. Neurotoxicology. 96:53–68. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Lee EY, Flynn MR, Du G, Lewis MM, Fry R, Herring AH, Van Buren E, Van Buren S, Smeester L, Kong L et al. 2015. T1 relaxation rate (r1) indicates nonlinear mn accumulation in brain tissue of welders with low-level exposure. Toxicological sciences : an official journal of the Society of Toxicology. 146(2):281–289. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Li H, Habes M, Wolk DA, Fan Y, Initiative AsDN. 2019. A deep learning model for early prediction of alzheimer’s disease dementia based on hippocampal magnetic resonance imaging data. Alzheimer’s & Dementia. 15(8):1059–1070. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Liu B, Moloney A, Meehan S, Morris K, Thomas SE, Serpell LC, Hider R, Marciniak SJ, Lomas DA, Crowther DC. 2010. Iron promotes the toxicity of amyloid beta peptide by impeding its ordered aggregation. Journal of Biological Chemistry.jbc. M110. 158980. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Mayo CD, Garcia-Barrera MA, Mazerolle EL, Ritchie LJ, Fisk JD, Gawryluk JR, Alzheimer’s Disease Neuroimaging I. 2018. Relationship between dti metrics and cognitive function in alzheimer’s disease. Front Aging Neurosci 10:436. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Meyer-Baron M, Schaper M, Knapp G, Lucchini R, Zoni S, Bast-Pettersen R, Ellingsen DG, Thomassen Y, He S, Yuan H et al. 2013. The neurobehavioral impact of manganese: Results and challenges obtained by a meta-analysis of individual participant data. Neurotoxicology. 36:1–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Miu AC, Benga O. 2006. Aluminum and alzheimer’s disease: A new look. Journal of Alzheimer’s disease. 10(2–3):179–201. [DOI] [PubMed] [Google Scholar]
  47. O’Shea A, Cohen RA, Porges EC, Nissim NR, Woods AJ. 2016. Cognitive aging and the hippocampus in older adults. Frontiers in aging neuroscience. 8:298. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Oguz I, Farzinfar M, Matsui J, Budin F, Liu Z, Gerig G, Johnson HJ, Styner M. 2014. Dtiprep: Quality control of diffusion-weighted images. Front Neuroinform 8:4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Pal A, Badyal RK, Vasishta RK, Attri SV, Thapa BR, Prasad R. 2013. Biochemical, histological, and memory impairment effects of chronic copper toxicity: A model for non-wilsonian brain copper toxicosis in wistar rat. Biol Trace Elem Res 153(1–3):257–268. [DOI] [PubMed] [Google Scholar]
  50. Patel R, Aschner M. 2021. Commonalities between copper neurotoxicity and alzheimer’s disease. Toxics. 9(1):4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Rueckert D, Sonoda LI, Hayes C, Hill DL, Leach MO, Hawkes DJ. 1999. Nonrigid registration using free-form deformations: Application to breast mr images. Medical Imaging, IEEE Transactions on. 18(8):712–721. [DOI] [PubMed] [Google Scholar]
  52. Sanders T, Liu Y, Buchner V, Tchounwou PB. 2009. Neurotoxic effects and biomarkers of lead exposure: A review. Reviews on environmental health. 24(1):15–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Shukla A, Tiwari R, Tiwari S. 2023. Structural biomarker‐based alzheimer’s disease detection via ensemble learning techniques. International Journal of Imaging Systems and Technology. [Google Scholar]
  54. Spotorno N, Strandberg O, Vis G, Stomrud E, Nilsson M, Hansson O. 2023. Measures of cortical microstructure are linked to amyloid pathology in alzheimer’s disease. Brain : a journal of neurology. 146(4):1602–1614. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Subedi S, Jeng A, Bush D. 2019. Metal fumes from welding processes and health impact. Virginia Journal of Public Health. 3(1):4. [Google Scholar]
  56. Sugar J, Moser MB. 2019. Episodic memory: Neuronal codes for what, where, and when. Hippocampus. 29(12):1190–1205. [DOI] [PubMed] [Google Scholar]
  57. Swartjes F 2015. Human health risk assessment related to contaminated land: State of the art. Environmental geochemistry and health. 37:651–673. [DOI] [PubMed] [Google Scholar]
  58. Takahashi E, Ohki K, Miyashita Y. 2002. The role of the parahippocampal gyrus in source memory for external and internal events. Neuroreport. 13(15):1951–1956. [DOI] [PubMed] [Google Scholar]
  59. Teleanu DM, Chircov C, Grumezescu AM, Volceanov A, Teleanu RI. 2018. Impact of nanoparticles on brain health: An up to date overview. Journal of clinical medicine. 7(12):490. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Tromp D, Dufour A, Lithfous S, Pebayle T, Despres O. 2015. Episodic memory in normal aging and alzheimer disease: Insights from imaging and behavioral studies. Ageing Res Rev 24(Pt B):232–262. [DOI] [PubMed] [Google Scholar]
  61. Tulving E, Markowitsch HJ. 1998. Episodic and declarative memory: Role of the hippocampus. Hippocampus. 8(3):198–204. [DOI] [PubMed] [Google Scholar]
  62. Valeri L, Vanderweele TJ. 2013. Mediation analysis allowing for exposure-mediator interactions and causal interpretation: Theoretical assumptions and implementation with sas and spss macros. Psychol Methods. 18(2):137–150. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Vaquero JJ, Kinahan P. 2015. Positron emission tomography: Current challenges and opportunities for technological advances in clinical and preclinical imaging systems. Annual review of biomedical engineering. 17:385–414. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Weston PSJ, Poole T, Nicholas JM, Toussaint N, Simpson IJA, Modat M, Ryan NS, Liang Y, Rossor MN, Schott JM et al. 2020. Measuring cortical mean diffusivity to assess early microstructural cortical change in presymptomatic familial alzheimer’s disease. Alzheimers Res Ther 12(1):112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Wixted JT, Squire LR. 2011. The medial temporal lobe and the attributes of memory. Trends Cogn Sci 15(5):210–217. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Wu L, Cui F, Zhang S, Ding X, Gao W, Chen L, Ma J, Niu P. 2023. Associations between multiple heavy metals exposure and neural damage biomarkers in welders: A cross-sectional study. Science of The Total Environment. 869:161812. [DOI] [PubMed] [Google Scholar]
  67. Xu J, Church SJ, Patassini S, Begley P, Kellett KAB, Vardy E, Unwin RD, Hooper NM, Cooper GJS. 2018. Plasma metals as potential biomarkers in dementia: A case-control study in patients with sporadic alzheimer’s disease. Biometals. 31(2):267–276. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Zhou Y, Peng Z, Seven ES, Leblanc RM. 2018. Crossing the blood-brain barrier with nanoparticles. Journal of controlled release. 270:290–303. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2032514-supplement-1.docx (4.4MB, docx)
2
NIHMS2032514-supplement-2.docx (28.1KB, docx)

Data Availability Statement

Data is available upon request.

RESOURCES