Seafood arsenic risk cannot be judged by total arsenic alone, review finds

Seafood is one of the healthiest foods humans eat, and also one of the main ways people are exposed to arsenic. In the United States, about 90 percent of dietary total arsenic intake has been attributed to seafood, and in a Spanish cohort seafood accounted for 98 percent of total arsenic intake yet less than 2 percent of intake of the toxic inorganic form. That asymmetry sits at the heart of a new narrative review arguing that seafood safety assessments built on total arsenic measurements are increasingly inadequate, and that risk evaluation must shift toward measuring which specific arsenic compounds are actually present.

The review, published in Frontiers in Ocean Sustainability by Rachel Ann Hauser-Davis of the Oswaldo Cruz Foundation in Rio de Janeiro and Oswaldo Luiz de C. Maciel of the Fisheries Institute of the State of Rio de Janeiro, synthesizes current knowledge on arsenic occurrence and speciation in seafood, analytical methods, toxicology, and human exposure assessment. Its central claim is that arsenic toxicity depends strongly on chemical form rather than total concentration, and that the two simplifications underlying traditional seafood arsenic assessment, treating seafood arsenic as a single harmless pool and treating seafood consumers as a single population, no longer survive the accumulated evidence.

Most arsenic in marine finfish occurs as arsenobetaine, an organic compound generally regarded as low in toxicity and rapidly excreted in humans, typically making up 70 to 90 percent or more of total arsenic. Inorganic arsenic, primarily arsenite and arsenate, is a different matter entirely: it is classified as carcinogenic, with established links to lung, urinary bladder and skin cancers, and the 2025 U.S. EPA toxicological review concluded that the human evidence for circulatory system disease is robust, identifying ischemic heart disease as a critical effect. The Joint FAO/WHO Expert Committee on Food Additives at its 101st meeting likewise selected ischemic heart disease as the critical endpoint, deriving a benchmark dose lower limit of 0.3 micrograms per kilogram of body weight per day.

The review also documents that the old assumption of organic arsenicals being uniformly harmless is being tested. Methylated intermediates such as monomethylarsonous acid and dimethylarsinous acid have demonstrated cytotoxicity and genotoxicity in some cases exceeding that of their inorganic precursors. Among arsenolipids, arsenic-containing hydrocarbons proved up to five-fold more potent than arsenite in a blood-brain barrier cell model, and in a Caenorhabditis elegans study only the arsenic-containing hydrocarbons were substantially metabolized and induced toxicity, affecting survival and development. Thiolated arsenicals add another complication: thio-DMA(V) was about five-fold more cytotoxic than arsenite in human lung cells and roughly 100-fold more cytotoxic than DMA(V) in urothelial cells, yet no seafood occurrence data for these compounds have been submitted to the European Food Safety Authority, partly because they convert to their oxo-analogs during extraction and are likely under-reported.

Elasmobranchs, the sharks, rays and skates, illustrate why total arsenic can be a misleading indicator. These animals can accumulate very high total arsenic concentrations in edible tissues, in some cases exceeding food-safety limits, but arsenobetaine commonly accounts for 50 to 95 percent or more of total arsenic in their tissues, apparently tied to their distinctive ureosmotic physiology and its reliance on methylamine osmolytes. Caribbean sharpnose sharks sampled off Rio de Janeiro exceeded the Brazilian maximum permissible limit, which is based on total arsenic, but the review stresses that such an exceedance does not by itself establish a health risk, because the Brazilian limit does not distinguish among arsenic species with vastly different toxicities. Conversely, the authors caution that a high arsenobetaine fraction is not a safety argument either: in long-lived predators, even a small non-arsenobetaine fraction can represent a substantial absolute concentration of toxicologically relevant species.

The regulatory landscape is changing. Commission Regulation (EU) 2025/1891, applicable from 8 October 2025, established maximum levels specifically for inorganic arsenic in fish and other seafood, with taxon-specific wet-weight limits ranging from 0.05 milligrams per kilogram for cephalopods to 1.5 milligrams per kilogram for langoustine and rock lobster, and a general fish limit of 0.10 milligrams per kilogram, rising to 0.50 for sharks, rays and several other taxa. China has had inorganic arsenic limits for fish since mid-2023, and Australia and New Zealand also apply inorganic arsenic-based limits. Brazil, by contrast, still regulates total arsenic only, meaning products dominated by harmless arsenobetaine may be non-compliant. Codex Alimentarius sets no maximum level for arsenic in most seafood commodities. The review notes that the challenge is shifting from the absence of standards to ensuring the analytical infrastructure, validated methods, certified reference materials and laboratory capacity, can support consistent enforcement.

That infrastructure question is concrete. The European standard method for inorganic arsenic in foodstuffs of marine and plant origin, validated by collaborative trial, covered mass fractions from 0.073 to 10.3 milligrams per kilogram, above the limit-of-quantification levels of 0.033 and 0.067 milligrams per kilogram that EU performance criteria require for the lowest limits. Certified reference materials exist for some species and matrices, including tuna fish tissue, fish protein, lobster hepatopancreas and hijiki seaweed, but coverage of arsenosugars and arsenolipids remains incomplete. High-performance liquid chromatography coupled with inductively coupled plasma mass spectrometry is the reference approach for quantitative speciation, while electrospray tandem mass spectrometry supplies the structural information needed to identify unknown compounds.

Human exposure assessment compounds the problem. The provisional tolerable weekly intake of 15 micrograms of inorganic arsenic per kilogram of body weight was withdrawn by JECFA in 2010, yet the review reports that many studies still use superseded reference values in their target hazard quotient and cancer risk calculations. Consumption patterns also vary enormously: mean per capita fresh fish consumption in Brazil is 13.1 grams per day nationally, ranging from 6.1 grams per day in the Southeast to 45.0 grams per day in the North, while consumption of aquatic products in the Brazilian Amazon has been estimated at 416.4 grams per day, an order of magnitude above any national figure. Coastal, Indigenous and fishing-dependent communities therefore face exposures far beyond what population averages suggest, and genetic variability in arsenic methylation capacity, including well-characterized AS3MT polymorphisms, means the same seafood intake does not produce the same internal dose of trivalent metabolites across consumers.

Cooking and digestion add further uncertainty. Boiling with the cooking water discarded lowered total arsenic by 22 percent in seaweeds, 32 percent in shellfish and 41 percent in fish, with inorganic arsenic lost most readily; in hijiki, boiling removed 65 percent of inorganic arsenic against 50 percent of total arsenic. Yet the evidence on transformation is contested: some studies found no species transformation at preparation temperatures up to 100 degrees Celsius, while mass-balance accounting across boiled bivalves returned a recovery of 435 percent for inorganic arsenic, from which demethylation during boiling was inferred, though the raw inorganic fraction was only 1.1 to 6.9 percent of total arsenic. Bioaccessibility results diverge sharply by matrix and method, from 81 to 99 percent for inorganic arsenic in crayfish to below 1.5 percent for all species except arsenobetaine across five marine taxa.

To organize this complexity, the authors propose a provisional tiered framework. Tier 0 defines the assessment problem: the taxon, tissue, product type, origin and population of interest. Direct speciation is recommended for bivalve molluscs, seaweeds, particularly brown seaweeds, elasmobranchs, organisms from contaminated areas, lipid-rich tissues and fish oils, and products consumed whole or with viscera. Tier 1 compares total arsenic with the applicable inorganic arsenic maximum level, a screening step expected to clear few marine samples because arsenobetaine alone commonly exceeds the lowest limits. Tier 2 requires validated inorganic arsenic determination, and Tier 3 extends speciation to other compounds where needed. Tier 4 characterizes risk against current reference values using consumption data for the population actually being assessed. The review also notes that arsenic has been absent from established seafood risk-benefit frameworks, which weigh methylmercury against omega-3 fatty acids and selenium, because a composite benefit-risk metric requires the exposure term to be expressed as the species that carries the hazard, something total arsenic cannot provide. Speciation, the authors conclude, is a precondition for bringing arsenic into seafood risk-benefit assessment, not merely a refinement of hazard characterization.

Subject of Research: Arsenic speciation and human health risk assessment in seafood

Article Title: Beyond total arsenic: rethinking seafood-related arsenic risk assessments

Article References: Hauser-Davis, R. A., & Maciel, O. L. D. C. (2026). Beyond total arsenic: rethinking seafood-related arsenic risk assessments. Frontiers in Ocean Sustainability, 4, 1951412. https://doi.org/10.3389/focsu.2026.1951412

Image Credits: AI Generated

DOI: 10.3389/focsu.2026.1951412

Keywords: arsenic, seafood, arsenobetaine, inorganic arsenic, arsenolipids, food safety, risk assessment, speciation analysis, elasmobranchs, EU regulation, exposure assessment, toxicology