Massive Japanese Study Maps Hidden Trace Elements Flowing Through Human Blood

In one of the largest biomonitoring efforts of its kind ever conducted, researchers in Japan have compiled reference intervals for eight trace elements circulating in human serum, drawing on blood samples from 17,523 adults screened across ten collection sites between 2017 and 2020. The study, published in the Journal of Exposure Science & Environmental Epidemiology, provides age-decade– and sex-stratified reference intervals for cobalt, arsenic, selenium, rubidium, strontium, molybdenum, silver, and cesium—elements that previously lacked established population-specific reference ranges for Japanese adults. The work was led by Takazo Tanaka and Tomoko Oguri, with contributions from researchers at Renatech Co., Ltd. and the University of Tsukuba, and was approved by the University of Tsukuba Hospital under approval number R03-127.

The clinical motivation behind the study is deceptively simple. Inductively coupled plasma mass spectrometry, or ICP-MS, can quantify dozens of elements in a single serum sample with extraordinary sensitivity, yet interpreting those numbers requires knowing what a normal range looks like for a given population. Without robust reference intervals—especially those stratified by age and sex—clinicians and researchers have been unable to contextualize whether an individual’s serum concentration of, say, rubidium or cesium is typical or unusual. Earlier Japanese studies had reported serum concentrations for a dozen trace elements in community-dwelling adults, and the Japan Environment and Children’s Study had characterized several blood elements in pregnant women, but no large multicenter effort had established age- and sex-stratified serum reference intervals using a unified Clinical and Laboratory Standards Institute (CLSI EP28-A3c)-oriented framework.

The study drew on the Metallo-Balance screening network, an ICP-MS-based serum profiling system that measures seventeen specific elements after acid digestion, capturing total elemental concentrations regardless of their original molecular binding state. From this predefined seventeen-element panel, the researchers pre-specified eight target analytes for reference interval derivation: cobalt, arsenic, selenium, rubidium, strontium, molybdenum, silver, and cesium. The remaining nine elements were excluded because seven of them—sodium, magnesium, potassium, calcium, iron, copper, and zinc—are already routinely evaluated in clinical practice using established chemistry assays, while total phosphorus and total sulfur, as measured by digestion-based ICP-MS, do not directly correspond to standardized clinical markers such as inorganic phosphate. A recent comparative study showed that total phosphorus measured by ICP-MS is roughly 3.5 times higher than routine serum inorganic phosphorus and only weakly correlated with it, underscoring why those two elements lacked discrete clinical interpretability without complementary biomarkers.

Participants were recruited through an optional serum trace element screening initiative at twelve sites across Japan, spanning Hokkaido to Fukuoka. After sequential exclusions—removing participants under twenty years of age, records with incomplete element panels, and duplicate samples—the final analytic cohort comprised 17,523 unique individuals from ten sites. Blood samples of roughly six milliliters were collected, serum was separated by centrifugation, digested with nitric acid and hydrogen peroxide, and analyzed on an Agilent 7800 ICP-MS system. Cobalt, rubidium, and strontium were quantified using yttrium-89 as an internal standard; arsenic and selenium used tellurium-125; and molybdenum, silver, and cesium used rhodium-103. Selenium was measured in hydrogen reaction mode to minimize polyatomic interferences, while all other target elements were measured in helium collision mode. Quality control was maintained using Seronorm Trace Element Serum Level 1 reference material, and limits of detection and quantification were calculated separately for each of 183 analytical batches.

Extreme values were identified and removed using the LAVE method, a robust outlier procedure based on the median absolute deviation, applied independently within each sex group for each element with up to three iterative passes. The scaling constant 1.4826 renders the MAD-based estimator consistent with the standard deviation under Gaussian assumptions. After element-specific exclusions, retained sample sizes ranged from 14,904 for cobalt to 17,494 for rubidium. Reference intervals were then estimated using Box-Cox transformation to approximate normality, with the optimal transformation parameter estimated by maximum likelihood, and bootstrap 90 percent confidence intervals obtained with 2,000 resamples. Strata with fewer than 20 participants were reported as not calculable, and estimates for strata with fewer than 200 were conservatively flagged to indicate reduced precision.

Before pooling data across collection years and sites, the researchers assessed homogeneity using Kruskal-Wallis tests. Effect sizes across collection years ranged from 0.0007 to 0.0124, with all reference interval width ratios below the EFLM 17 percent criterion, supporting pooled construction. For the site factor, effect sizes were similarly small, though cesium, rubidium, and selenium exceeded the width ratio threshold. Because Kanagawa contributed 60.3 percent of the analytic sample, an additional leave-Kanagawa-out sensitivity analysis was performed, comparing reference limits derived from the non-Kanagawa subset with all-site pooled estimates. Both reference limits differed by less than 5 percent of the pooled interval width for all eight elements, confirming that the pooled reference intervals were robust to site composition.

The results revealed striking age- and sex-related patterns. Arsenic and molybdenum were generally higher in older age groups in both sexes, and silver and cesium also tended to rise with age, though with non-monotonic decade-specific patterns. Rubidium, in contrast, was generally lower in older groups. Selenium exhibited a sex-dependent trajectory: concentrations remained largely stable across age groups in men, whereas in women they increased up to the 60–69-year age group and declined thereafter. Cobalt was consistently higher in women than in men, with the upper reference limit higher at ages 20–49 and lower at 50–69. Strontium upper reference limits generally increased with age in both sexes, with a more pronounced rise in women aged seventy and above.

The cobalt findings are particularly intriguing from a biological standpoint. The female predominance is consistent with earlier Japanese community-based data, and the midlife decline in women aligns with a proposed mechanism involving the divalent metal transporter 1, or DMT1, which mediates intestinal uptake of both cobalt and iron. Under this framework, cyclical iron depletion associated with menstruation in premenopausal women would sustain upregulation of DMT1 and maintain relatively higher intestinal cobalt absorption, whereas the cessation of menstrual iron loss around menopause may progressively reduce this absorptive stimulus. The researchers emphasize that menopausal status was not recorded in this dataset, so the proposed DMT1–iron mechanism remains speculative and would require direct measurement of DMT1 expression or cobalt absorption for confirmation. Observational data from the Dongfeng-Tongji cohort suggest that higher plasma cobalt is associated with greater physiological dysregulation, a composite index of biological aging, and NHANES data indicate that the adverse renal effects of cobalt may be amplified in older individuals.

The age-related accumulation of arsenic, molybdenum, silver, and cesium may reflect two partially overlapping mechanisms. The first is age-related decline in glomerular filtration rate, estimated at roughly 0.75 to 1.0 milliliters per minute per year in longitudinal studies, which for elements whose homeostasis depends primarily on renal excretion translates directly into blood-level accumulation. The second is tissue-level sequestration, which differs substantially across the four elements. For silver, absorbed deposits in soft tissues predominantly as particles of silver sulfide or silver selenide, considered essentially immobile once formed, and tissue concentrations in the liver and brain cortex have been shown to correlate dose-dependently with the number of dental amalgam fillings. For cesium, which behaves as a potassium analog actively transported into cells via the sodium-potassium ATPase pump, the age-related increase may reflect cumulative dietary loading of a large intracellular reservoir determined primarily by lean muscle mass. The researchers note that the study measures stable cesium-133, not radioactive cesium-137, and that broader food monitoring data in Japan have consistently indicated that radiocesium contamination in the food supply remains well within established regulatory limits.

The selenium story adds another layer of biological complexity. In women, concentrations rose from the youngest age group through the 60–69-year group before declining, while men remained comparatively stable through midlife. A biological basis for this dimorphism may lie in the dynamics of selenoprotein P, the primary selenium transport protein in blood, which accounts for roughly 60 percent of circulating selenium and serves as the principal vehicle for hepatic selenium export. Selenoprotein expression is governed by a tissue-priority hierarchy in which selenium is preferentially allocated to organs with high selenoprotein demand, including the testes, brain, and—during reproductive years—the uterus and placenta. Under this model, the reduction in high-priority reproductive selenium demands at menopause would be expected to release selenium back into the selenoprotein-mediated circulating pool, driving the progressive post-menopausal accumulation observed in the study. The researchers caution that the reference population was drawn from an optional commercial screening initiative rather than a rigorously phenotyped healthy cohort, that the cohort is not nationally representative given the geographic imbalance toward Kanagawa, and that the cross-sectional design precludes causal inference regarding age-related trajectories. The arsenic values represent total serum arsenic without chemical speciation and should not be applied directly to inorganic-arsenic exposure assessment, since dietary arsenic in Japanese populations is predominantly derived from seafood and consists largely of organic, low-toxicity species such as arsenobetaine. Even with these caveats, the study provides a practical reference framework for interpreting ICP-MS-based serum trace element measurements in Japan and offers a foundation for future validation in more comprehensively phenotyped cohorts and other populations.

Subject of Research: Age- and sex-stratified serum reference intervals for eight trace elements in a large Japanese screening cohort

Article Title: Age- and sex-stratified serum reference intervals for eight trace elements in a large Japanese screening cohort

Article References: Tanaka, T., Oguri, T., Shimizu, T., Fujimoto, S., Shiga, M., Nagumo, Y., Kandori, S., Ikeda, A., Kojo, K., & Nishiyama, H. (2026). Age- and sex-stratified serum reference intervals for eight trace elements in a large Japanese screening cohort. Journal of Exposure Science & Environmental Epidemiology. https://doi.org/10.1038/s41370-026-00968-2

Image Credits: AI Generated

DOI: 10.1038/s41370-026-00968-2

Keywords: Age-, sex-stratified, serum, reference, intervals, eight, trace, elements, large, Japanese, screening, cohort