Dobrzyńska et al. 2025 — elemental composition of milk formulae on the Polish market

Dobrzyńska and colleagues measured nine basic elements (Na, K, Ca, P, Mg, Fe, Zn, Cu, Mn) and seven toxic or potentially toxic elements (As, Cd, Co, Cr, Hg, Sn, Ni) by ICP-MS in 149 powdered milk formulae available on the Polish market between 2019 and 2023. Cadmium, mercury, and arsenic were below the analytical detection limit in every sample; nickel was detected in 62 of 149 samples (highest 17 µg/kg wet weight in a powder formula) and the estimated daily intake exceeded the EFSA lowest-observed-adverse-effect level of 4.3 µg/kg body weight for eczematous skin reactions in one-month-old infants for every formula category analyzed. Copper exceeded the EU compositional maximum in 18 samples and was below the minimum in 43; potassium exceeded the EU maximum in 63 samples; and manganese varied across roughly two orders of magnitude (1-91 µg/100 kcal). Lead and selenium were excluded from this paper — Pb due to spectral interference in the analytical run, Se reported in a separate publication.

Key numbers

The corpus is 149 powdered formulae purchased on the Polish market between 2019 and 2023 (Materials, p. 386). The 0-6 month infant-formula stratum was 37 samples, the 7-12 month follow-on stratum was 34 samples, and the formulae for special medical purposes (FSMP) stratum was 78 samples. Within FSMP, the breakdown was extensively hydrolyzed n=26, amino-acid-based n=6, manufactured from goat milk proteins n=15, comfort n=12, anti-reflux n=7, lactose-free n=5, preterm n=4, and soy-based n=3.

The reference material was ERM-BD150 (skimmed milk powder, European Reference Materials, Belgium); recovery was 80-120% for most elements. Analyses were performed in triplicate following Yaman et al. ICP-MS sample digestion used a CEM MARS6 microwave system with 10 mL of 69% ultrapure nitric acid (ROMIL, Cambridge, UK) on 0.5 g sample, ramping from ambient to 180 °C over 20 min and holding for 20 min, then diluted to 15 mL with ultrapure water. The propagated uncertainty (RSD) for the whole procedure was below 20%. Limits of detection (µg/L) were As 6.0, Ca 250, Cd 0.45, Co 1.5, Cr 0.77, Cu 2.4, Fe 61, Hg 18, K 140, Mg 150, Mn 1.2, Na 110, Ni 13, P 3100, Zn 5.6, Sn 0.27 (p. 386-387).

Table 9 — toxic and potentially toxic elements, µg/kg wet weight. Cd, Hg, and As were < DL in every infant-formula, follow-on, and FSMP sample (Cd EU limit 10 µg/kg; As EU limit 20 µg/kg; Hg unrestricted under the cited regulation). Sn medians were 0.3 (1st-3rd quartile 0.1-0.5; range 0-2.2) for infant formulae, 0.2 (0.2-0.4; range 0-1.1) for follow-on, and 0.4 (0.2-0.6; range 0-11) for FSMP, far below the Sn maximum of 50,000 µg/kg. Co was 0 (0-0; range 0-1.0) for infant formulae, 0 (0-0; range 0-1.1) for follow-on, and 0 (0-0; range 0-9.6) for FSMP. Cr medians were 0 (range 0-23) for infant formulae, 0 (range 0-13) for follow-on, and 0 (range 0-1.7) for FSMP. Ni medians were 0 (range 0-17) for infant formulae, 0 (range 0-13) for follow-on, and 0 (range 0-12) for FSMP. The text records that Ni was detected in 62 of 149 samples (p. 395).

Estimated daily intake of toxic and potentially toxic elements (Table 10). Sn EDI for the 0-6 month infant-formula stratum was median 1.86 µg/day at 1 month (1st-3rd quartile 0.00-3.11; range 0.00-15.66), 2.37 µg/day at 3 months (range 0.00-19.86), and 2.56 µg/day at 6 months (range 0.00-21.47). For follow-on formulae the Sn EDI was 0.91 µg/day at 7 months and 0.57 µg/day at 10 months. For FSMP the Sn EDI reached 3.31 µg/day at 6 months (range 0.00-104.97), with a stratum maximum of approximately 105 µg/day at 6 months.

Ni EDI (Table 10) for infant formulae (0-6 months, 100% energy coverage) was median 0 µg/day at every age band, but the upper bound of the range was 18.36 µg/day at 1 month, 19.08 µg/day at 3 months, and 19.29 µg/day at 6 months. For follow-on formulae (45% energy coverage at 7 months, 25% at 10 months), Ni EDI ranges reached 6.80 µg/day and 3.84 µg/day respectively. For FSMP (100% coverage at 1 and 6 months), Ni EDI ranges reached 12.49 µg/day at 1 month and 13.13 µg/day at 6 months. The authors converted Ni EDI to µg/kg body weight and compared against the EFSA tolerable daily intake of 13 µg/kg body weight and the LOAEL of 4.3 µg/kg body weight for eczematous flare-up reactions (EFSA 2020). The Ni EDI per kg body weight exceeded the TDI of 13 µg/kg in four FSMP formulae at 6 months and two infant formulae at 1 month; the Ni EDI per kg body weight exceeded the LOAEL of 4.3 µg/kg in all formulae regardless of category for infants aged 1-6 months (Discussion, p. 395-396).

Cr EDI was 0.00-210.40 µg/day across the corpus with the highest medians among infant formulae and the lowest among FSMP. The text reports that the chromium EDI did not exceed the National Research Council’s “estimated safe and adequate daily dietary intake” of 0.01-0.04 mg/day, nor the UK Committee on Medical Aspects “adequate level of chromium intake” range of 0.0001-0.001 mg/body weight/day.

Co EDI was below the limit of detection in most preparations. The infant-formula maximum was 6.77 µg/day at 1 month and 9.27 µg/day at 6 months. Follow-on EDI maxima were 4.52 and 2.95 µg/day for 45% coverage at 7 months and 25% coverage at 10 months. FSMP Co EDI reached 64.49 µg/day at 1 month and 6 months.

Essential-element findings against EU compositional norms. For the 37 infant formulae (Table 1), P and Mn were within EU limits for all 37 samples. Na was below the minimum in two samples (11-17% lower). K exceeded the EU maximum in 16 samples by an average of 13% (range 1-32%). Ca was below the minimum in 1 sample (27.5% lower). Mg exceeded the maximum in 2 samples (4-6%). Fe exceeded the maximum in 2 samples (23-69%). Zn was below minimum in 1 sample (20% lower) and exceeded the maximum in 1 sample (27.5%). Cu was inappropriate in 21 of 37 samples: below the minimum in 16 (5-90% lower; one sample below the detection limit) and above the EU maximum in 5 samples.

For the 34 follow-on formulae (Table 2), Mn was within limits for all 34. Na exceeded the maximum in 2 samples (by 2%). K exceeded the maximum in 14 samples (one by 28%). P exceeded the maximum in 7 samples (1-15%, average 15%). Mg exceeded the maximum in 8 samples (1-172%). Fe was below the minimum in 1 sample (50% lower). Zn exceeded the maximum in 4 samples (7-39%). Cu was below the minimum in 1 sample (80% lower) and exceeded the maximum in 7 samples (10-238%).

For the 78 FSMP (Table 3), Ca, Fe, Zn, and Mn were within acceptable limits for all samples. Na exceeded the maximum in extensively hydrolyzed (1 sample, 3%), goat-milk-protein (1 sample, 20%), and preterm (2 samples, 3-72%) formulae; Na was below the minimum in the extensively hydrolyzed and goat-milk formulae. K exceeded the maximum in 33 samples regardless of formulae type (1-98%) except the soya-based formula which met the limits. P exceeded the maximum in 1 goat-milk sample (39%) and 1 soya-based sample (7%). Mg exceeded the maximum in 2 extensively hydrolyzed, 4 amino-acid-based, 2 goat-milk-protein, 1 comfort, and 1 anti-reflux sample. Cu was below the minimum in 26 samples (mostly hydrolyzed) and exceeded the maximum in 6 (maximum 47%).

Across the full 149-formulae corpus: Cu was below minimum in 43 samples and above maximum in 18 samples; K exceeded the maximum in 63 samples; Mn ranged from 1-91 µg/100 kcal across formulae. Cu median was 72 µg/100 kcal across all formulae (66 µg/100 kcal infant; 81 µg/100 kcal follow-on; 68 µg/100 kcal FSMP), all somewhat higher than the European breast-milk reference range of 51-60 µg/100 kcal.

Methods (brief)

Inductively coupled plasma mass spectrometry (ICP-MS), Department of Analytical Chemistry, Adam Mickiewicz University, following the analytical method of Yaman et al. Triplicate analyses per sample, with one to twenty parallel samples per formula depending on availability and study cost. Sample digestion: 0.5 g formula powder + 10 mL 69% ultrapure HNO₃ in a CEM MARS6 microwave (Matthews, NC, USA), 15 min pre-digestion, ramp ambient → 180 °C over 20 min, hold 20 min, dilute to 15 mL with ≥18 MΩ·cm ultrapure water (Milli-Q, Merck Millipore). High-pure (99.999%) Ar, He, and H₂ gases (Linde, Poland). Commercial standards (Romil, UK) and Suprapur® 65% HNO₃ used for calibration and standard preparation. Certified reference material: ERM-BD150 (European Reference Materials, Belgium); recoveries 80-120%. RSD < 20% (propagated uncertainty for the full procedure). Limits of detection are listed in the Key Numbers section above.

EDI was calculated by converting concentrations to mg or µg per 100 kcal and dividing into three age groups (infant 0-6 mo, follow-on 7-12 mo, FSMP). For 7-month-olds the model assumes 45% of energy from formula; for 10-month-olds, 25%. Body weights were taken from current WHO infant percentiles; energy needs from Stan et al. Toxic-element EDI was compared against EU Commission Regulation 2023/915 of 25 April 2023 (repealing Regulation (EC) No 1881/2006) maximum levels and, for Ni, against EFSA’s TDI of 13 µg/kg body weight and LOAEL of 4.3 µg/kg body weight for eczematous flare-up reactions.

Reported limitations. Pb and Se were excluded from this paper. The authors state that the Pb assessment was limited “due to significant elemental interference in Pb concentrations” and that Se concentrations will appear in a separate publication (p. 396). The corpus is limited to powdered formulae and to the Polish market 2019-2023; ready-to-feed and concentrated liquid forms were not analyzed. Brand identification is not provided in the manuscript.

Implications

Certification: This is routeable occurrence and exposure evidence for powdered infant formula, follow-on, and FSMP forms on the EU regulatory framework (Commission Delegated Regulation 2016/127 for infant and follow-on, 2016/128 for FSMP, and 2023/915 for contaminants). All occurrence data are expressed in the as-placed-on-market powder basis (µg/kg wet weight for toxics; mg or µg per 100 kcal for essential elements); do not pool with reconstituted as-consumed values without applying a documented basis conversion. The Cd, Hg, and As < DL findings constitute a category-level non-detection record for these three analytes in the 149-formula Polish stratum.

Courses: Useful for explaining why nickel — a non-restricted analyte under most infant-formula composition regulations — has emerged as the toxic-metal exposure of concern when the EFSA per-body-weight LOAEL is applied to infant-formula EDI. The same dataset demonstrates how essential-element compositional non-compliance (K above maximum, Cu both below minimum and above maximum, Mn variability) can coexist with regulatory-compliant toxic-element values.

App: Supports infant-formula powder context with categorical evidence on Cd, Hg, and As (non-detect on the Polish market 2019-2023) and on Ni, Sn, Co, Cr (detected at low absolute concentrations but Ni intake exceeding EFSA LOAEL on a per-body-weight basis for 1-month infants). The dataset does not address ready-to-feed or concentrated liquid formula forms.

Wiki pages this source may touch

• infant-formula-powder
• infant-formula-powder-non-soy
• infant-formula-powder-soy-based
• cow-milk-infant-formula
• soy-based-infant-formula
• milk-based-infant-formula
• non-soy-infant-formula
• nickel
• tin
• cobalt
• chromium
• cadmium
• mercury-total
• arsenic-total
• copper
• manganese
• eu2023-contaminants-maximum-levels
• efsa-nickel-tdi

Verification notes

Title, authors, journal, DOI, and publication year were taken from the Springer PDF title page; the journal cites the article as “Biological Trace Element Research (2025) 204:385-400” with “Published online: 17 June 2025.” The cite-key uses 2025 to match the publication-year citation header; the raw filename (and therefore the raw_handle) retains the original “dobrzynska2026” stem assigned at discovery time.

This page records arsenic as tAs (total arsenic) and mercury as tHg (total mercury) because the paper measures total-element concentrations by ICP-MS without speciation. The paper writes “As” and “Hg” without speciation flags throughout. Chromium is recorded as Cr (total chromium); Cr-VI was not measured.

Lead is intentionally not in the metals list. The Discussion section explicitly states that “The assessment of Pb […] concentrations is limited in this study due to significant elemental interference” (p. 396). Selenium is also not included — the authors state it will appear in a separate publication (reference 46).

The 78 formulae for special medical purposes (FSMP) span an HMTc product class not currently represented as a wiki product slug; the routing audit will surface this. Goat-milk-based powdered formula (n=15) is also not represented as a wiki ingredient slug; the routing audit will surface this. Per the ingest-skill rules, this page does not create new ingredient or product pages.

Brand identification is not provided in the paper, so the brand firewall (CLAUDE.md Part 12) is automatically satisfied. Methods-section vendor and instrument names (CEM MARS6, Linde, ROMIL, Milli-Q/Merck Millipore, Romil standards, ERM-BD150) are retained per the 2026-05-17 scientific-method exception.

Audit subagent (2026-06-09) flagged [[ingredients/cow-milk-infant-formula]] and [[ingredients/milk-based-infant-formula]] as invented slugs not present in the taxonomy snapshot. Verified false positive: both pages exist at wiki/ingredients/cow-milk-infant-formula.md and wiki/ingredients/milk-based-infant-formula.md, and the routing audit accepted both wikilinks cleanly with zero unresolved targets for this source. The discrepancy is a staleness gap in docs/gpt-collaboration/taxonomy-snapshot.md, not a defect in this source page. The audit subagent also questioned the Ni FSMP maximum being rendered as 0-12 rather than 0-12.5; verified false positive — Table 9 shows the FSMP Ni range as <0-12> with no decimal, matching the page text. Audit verdict was REVISE; on independent verification no frontmatter or body changes are warranted.

Page history

The five most recent substantive edits to this page. The full version history lives in git; when DOI minting comes online (see schema docs), each entry below will also link to a version-pinned DataCite DOI.