The arsenic-Alzheimer's disease link: Emerging evidence and unanswered questions

Contamination of arsenic in groundwater remains a major global health challenge, particularly in low-income, rural, and historically exposed communities.^1,2 Arsenic has been linked to cancer,^3,4 cardiovascular disease, ⁵ metabolic disorders like diabetes, ⁶ and its neurotoxic effects are thought to increase risk for neurodegenerative disorders. ⁷ For much of the twentieth century, private wells supplied drinking water for a substantial portion of the Swedish population, where groundwater arsenic concentrations vary regionally due to geology and historical mining activity. ⁷ This geographic heterogeneity effectively created a nationwide natural experiment: people born and raised in different regions experienced systematically different arsenic exposures through drinking water while sharing the same nationalized healthcare and social system.

In a new study, Wei et al. ⁸ leverage this geographic variation in groundwater arsenic across Sweden, using birthplace and adult residence to approximate early and late-life arsenic exposure. While this registry-based design assesses Alzheimer's disease (AD) diagnoses based on arsenic exposure earlier in life, its interpretability relies on several key assumptions. Exposure is inferred rather than directly measured, which introduces the potential for misclassification, and the reliance on education as a direct measure of socioeconomic status risks residual confounding. Migration further complicates all of this if individuals from high arsenic regions are relocating for healthcare access, potentially biasing associations. Despite these possible limitations, the patterns observed in the Swedish cohort provide important insight into how timing of exposure shapes AD risk.

Examining rates of AD by birthplace arsenic exposure led to striking results with a clear dose-response relationship; those born in the highest arsenic exposure areas had a 156% higher AD risk than those born in low arsenic exposure regions, while AD risk for those born in high-middle and low-middle exposure areas were 98% and 47% higher, respectively. Wei et al. also examined effects of early adulthood exposure, with similar dose-response findings but overall blunted effects on AD risk compared to the birthplace exposure analysis. Particularly interesting is the comparison between birthplace and young adulthood exposure. That is, compared to individuals in low exposure regions at both birth and early adulthood, those with high exposure at birth but living in low exposure areas by early adulthood had an adjusted hazard ratio of 2.36, while the converse (low birthplace but high early adulthood exposure) had an adjusted hazard ratio of 1.73. This contrast between neonatal and early adulthood exposure suggests that high arsenic during critical developmental windows may have disproportionate effects on long-term brain health, shaping brain development, cognition, or metabolic pathways in ways that predispose to neurodegeneration, whereas adult exposure has less dramatic effects.

At the same time, this study opens important avenues for deeper inquiry: because national registries capture causes of mortality, future analyses could assess whether arsenic-high regions have higher rates of cardiovascular disease, cancer, or metabolic disorders, as well as other neurodegenerative disorders, alongside AD. This could help clarify not only whether arsenic exposure predisposes to overall adverse health outcomes versus specifically to AD, but also whether there may be shared biological pathways that underlie these conditions.

One previous study provides some clues, suggesting that arsenic exposure is a particularly strong driver of AD specifically. Work done in the Hamadan Province of Iran analyzed mortality data from over 8000 deaths in a region where groundwater arsenic levels far exceeded the international safety guidelines. ⁹ Poisson regression revealed that, among all causes of death examined, AD showed the strongest correlation with arsenic exposure. Interestingly, diabetes also demonstrated a strong association with arsenic exposure, and because diabetes itself is associated with increased AD risk,^10,11 this finding may indicate an overlapping metabolic or vascular pathway that warrants further investigation. Since the temporal and causal relationships among arsenic exposure, diabetes, and AD remain unclear, future epidemiologic studies should consider diabetes as a potentially relevant covariate when evaluating the relationship between arsenic exposure and AD.

Several other cohort studies have examined arsenic's effects on cognitive performance rather than AD risk. Adverse effects of arsenic on cognitive and developmental outcomes in children have been demonstrated by numerous studies,^12–14 supporting the idea of the critical impact that early life exposure has not only on AD risk as demonstrated by Wei et al., but also on cognition in general. In adults in China, high arsenic exposure (over 150 μg/L) was associated with poorer Mini-Mental State Examination performance, with cognitive impairment rising from around 10% in the lowest-exposure groups to around 55% in the highest-exposure group. ¹⁵ Conversely, Project FRONTIER tracked the health of older adults in West Texas, where arsenic in groundwater is actually below the US and WHO standard of less than 10 μg/L. Current and long-term exposure were linked to deficits across multiple cognitive domains, including language, executive function and visuospatial skills, while long-term exposure was associated with worse global cognition, processing speed, and immediate memory, ¹⁶ and these associations were strongest among individuals diagnosed with mild cognitive impairment (MCI). ¹⁷ As arsenic levels were largely below the 10 μg/L considered safe in drinking water, these studies suggest that any exposure may have adverse effects. While these studies demonstrated effects of arsenic on cognition, they did not assess whether rates of MCI or AD were associated with arsenic exposure, as the current study did. Taken together, these studies suggest that arsenic impacts cognition across populations with diverse genetics, lifestyles, and environmental contexts, reinforcing its potential role in neurodegeneration.

Mirroring findings by Wei et al. on the importance of early life exposure, perinatal arsenic exposure in animal models produces persistent cognitive and behavioral deficits in offspring.^18,19 Mechanistic evidence increasingly supports a link between arsenic and AD-related pathways. Arsenic induces oxidative stress, ²⁰ mitochondrial dysfunction, ²¹ and inflammatory signaling in glial cells ²² that can lead to adverse effects on neural function, including cortical synaptic changes ²³ and subsequent cognitive deficits. ²⁴ Experimental studies demonstrate that arsenic increases amyloid-β plaques^25,26 and tau phosphorylation ²⁷ in transgenic mouse models of AD, and promotes neuroinflammation and cognitive deficits in both AD and wildtype mice. ²⁶ Arsenic also disrupts the blood-brain barrier (BBB), enabling entry of other harmful factors that may drive neurodegenerative processes. ²⁸ In fact, arsenic joins a broader class of environmental neurotoxicants alongside lead, toluene, ²⁹ microplastics, ³⁰ and air pollution ³¹ that also converge at oxidative stress, inflammatory, and mitochondrial pathways.

While epidemiological studies and animal work together indicate a strong relationship between arsenic exposure and AD, there is a gap in our understanding of how arsenic acts on human cells specifically, especially as mouse models fail to replicate many key features of AD,^32,33 and there are significant species differences in the inherent biological function of AD-relevant cells.^34,35 Human-based experimental systems are needed to determine the mechanisms through which arsenic drives AD risk and induced pluripotent stem cells (iPSCs) offer a unique opportunity to address this. iPSCs can be differentiated into neurons, astrocytes, microglia, and cells making up the BBB, and can be assembled into 3D organ-on-chip platforms. Arsenic-containing media can be introduced through vascular compartments to model exposure across the BBB and examine downstream neural effects. ³⁶ We previously demonstrated the feasibility of modeling environmental risk factor exposure using human iPSCs. ³⁷ Future work should capitalize on these human model systems, including through use of patient-specific lines to determine whether certain genes are associated with increased vulnerability to arsenic. This will ultimately bridge the gap between epidemiologic associations and molecular mechanisms.

Taken together, the literature suggests that early-life arsenic exposure impairs cognitive abilities and increases susceptibility to later AD risk, a pattern that is reflected in the study conducted by Wei et al. Their work strengthens the case for considering arsenic as a potentially-modifiable environmental risk factor for AD, and along with numerous other studies, suggests that current exposure limits may be insufficient to protect against adverse health outcomes. Integrating environmental exposure mapping, longitudinal registries, and human-relevant experimental models will clarify who is most at risk for arsenic-associated adverse health outcomes, including AD risk, what mechanistic pathways it acts on, and ultimately, how we can mitigate effects of arsenic. Unfortunately, arsenic exposure is likely to increase, as a recent study found that climate change is expected to result in higher arsenic bioaccumulation in rice. ³⁸ Meanwhile Wei et al.'s findings add to a growing literature highlighting that there may not be a level of arsenic that can be considered safe and that public health guidelines and standards may need to be reevaluated.