Zvěřina et al. 2025 — Essential and toxic elements in plant-based dairy alternatives (Czech Republic market)

Zvěřina and colleagues at Masaryk University, the Czech Academy of Sciences, and Comenius University in Bratislava measured four essential elements (Ca, P, Fe, Zn) and two toxic elements (Cd, Pb) in 54 plant-based dairy alternative (PBDA) samples drawn from the Czech market and from a laboratory-prepared homemade oilseed-milk panel. Calcium, phosphorus, iron and zinc were quantified by high-resolution continuum source flame atomic absorption spectrometry (HR-CS FAAS, Ca) or electrothermal atomic absorption spectrometry (HR-CS ETAAS, Fe, P, Zn); cadmium and lead were quantified by HR-CS ETAAS on the same digests. Non-fortified PBDAs were markedly poorer in Ca, Fe and Zn than the cow’s-milk reference range, with median Ca in store-bought plant milks 25-fold lower than cow’s milk (4.71 vs roughly 120 mg/100 mL) and median Ca/P ratios below the 1.1–2 ESPGHAN window in every PBDA category except cheese alternatives. Lead concentrations were below the 7 µg/kg limit of detection in every sample (store-bought and homemade alike). Cadmium concentrations were below the 0.7 µg/kg LOD in store-bought milks, creams, yogurts, cheeses, and in sesame-seed homemade milks, but were elevated in tofu (median 14.5 µg/kg) and in three of the homemade oilseed milks (median Cd: flax 18.3 µg/kg, pumpkin 1.2 µg/kg, hemp 1.0 µg/kg). Homemade poppy-seed milk carried the highest Cd burden by an order of magnitude (median 58.8 µg/kg, range 13.4 to 85.0 µg/kg), reflecting the well-documented cadmium-accumulation behaviour of Papaver somniferum.

Why this matters

• It is one of the few studies that measures the full PBDA matrix family (milk, cream, yogurt, cheese, tofu, plus homemade oilseed seed milks) in one analytical frame with documented CRM validation and matched detection limits.
• It is the primary modern occurrence-data point for homemade oilseed milks — a category not covered by retail-only PBDA surveys and not yet routed in the wiki taxonomy.
• It connects soy-product (tofu) cadmium occurrence in the Czech market to the EFSA TWI exposure framing, with the authors computing that 200 g/day of tofu contributes about 10 % of a 75 kg adult’s Cd TWI.
• The Czech-market homemade-poppy-milk Cd signal (full cow’s-milk substitution at the Czech per-capita poppy-milk equivalent of 1.12 L/week reaches roughly 35 % of the Cd TWI) is a row-specific exposure signal that the plant-milks-non-soy-non-rice product page should track if homemade-seed-milk subcategories are added.

Key numbers

All concentrations are reported in the units the source uses. Cd and Pb values in Table 3 are in µg/kg of the PBDA as analyzed (homogenized liquid or wet semi-solid product); the abstract restates the homemade poppy-seed milk Cd median in mg/kg (“0.052 ± 0.04 mg/kg” — see Verification notes for the small numerical discrepancy with Table 3’s 58.8 µg/kg median). Essential-element values from Fig. 1 are reported in mg/100 g of the PBDA as analyzed; Ca/P ratios in Table 2 are unitless (mean ± SD). Limits of detection and quantification (Table 1) are in µg/L for the digestate solutions; LOQ for Cd in product matrix corresponds to 0.7 µg/kg, and LOQ for Pb to 7 µg/kg, given the 1 g sample digested into 10 mL final volume.

Sampling and analytical frame (source Materials and Methods, pp. 2008–2009)

Analytical performance (source Table 1, p. 2009)

CRM trueness (bias relative to certified value): Cd −9 %/−5 %, Fe +6 %, P −10 %/+6 %, Pb −11 %/−7 %, Zn −4 %/−2 %, Ca −4 %/−4 % across INCT-TL-1 and NIST 1570a.

Toxic-element concentrations by PBDA category (source Table 3, p. 2013)

Values are µg/kg of product as analyzed; “median (min, max)“. Values reported as “<0.7” or “<7” denote the LOQ for Cd and Pb in the product matrix respectively (Cd LOQ 0.7 µg/kg, Pb LOQ 7 µg/kg). No sample exhibited Pb above the LOQ.

Authors’ synthesis: “both store-bought and homemade products did not contain Pb amounts above the detection limit” and “in the commercially available PBDAs besides tofu, levels of cadmium were found to be very low, in most cases under the limit of detection.”

Essential-element headline findings (source Fig. 1, p. 2010, and Table 2, p. 2012)

Per-seed homemade-milk Ca medians: poppy 50.7, flax 41.2, sesame 19.4, hemp 16.0, pumpkin 10.3 mg/100 mL (Kruskal–Wallis p < 0.05 across seed types; Dunn’s post-hoc significant only between poppy and pumpkin). Per-seed homemade-milk Ca/P ratios: flax 0.6 ± 0.1, hemp 0.1 ± 0.01, pumpkin 0.1 ± 0.02, poppy 0.4 ± 0.1, sesame 0.2 ± 0.04.

Cow’s milk reference range cited by authors: Ca 107 to 134 mg/L with a typical median around 120 mg/L (107 to 134 mg/L in the cited literature [refs 10, 17, 22–25]); Ca/P ratio in cow’s milk approximately 1.2 (cited refs 22, 23, 25).

Hypothetical exposure scenarios computed by the authors

• Calcium intake under PBDA substitution (source p. 2011): replacing three daily servings of dairy (target intake 950 mg Ca per day per Czech and EFSA reference values for adults ≥25 y, ref 36) with non-fortified PBDAs at the average, worst-case (lowest observed), and best-case (highest observed) Ca contents yielded modeled daily Ca intakes of 300 mg (mean), 10 mg (worst case), and 850 mg (best case). The worst case provides “a mere 10 mg” daily Ca, roughly 1 % of the reference intake.
• Cadmium exposure from tofu (source p. 2013): “For a 75 kg adult, a daily consumption of 200 g of tofu would contribute by 10 % to the TWI for Cd,” where TWI = 2.5 µg/kg bw/week (EFSA, ref 61).
• Cadmium exposure from full poppy-seed milk substitution (source p. 2013): “a full substitution of cow’s milk (average consumption of 1.12 L per week in the Czech Republic) would account for 35 % of the TWI for Cd” using the homemade poppy-seed milk median Cd content.

Methods (brief)

Fifty-four PBDA samples were acid-digested in Teflon vessels with 5 mL HNO3 (Analpure, Analytika, Czech Republic) plus 1 mL H2O2 after a 1 h pre-reaction, then microwave-digested in a Multiwave Go plus system (Anton Paar, Austria) with a 20 min ramp to 180 °C held for 10 min, and topped up to 10 mL final volume. Digests were analyzed by high-resolution continuum source atomic absorption spectrometry on two instruments: ContrAA 800G (Analytik Jena GmbH, Germany) in electrothermal mode (HR-CS ETAAS) for Cd, Fe, P, Pb, and Zn; ContrAA 300 (Analytik Jena) in flame mode (HR-CS FAAS) with an acetylene-nitrous oxide flame for Ca. Pd/Mg(NO3)2 (1 g/L Pd in 0.6 g/L Mg(NO3)2; Analytika, Czech Republic) served as the matrix modifier for Cd, P, Pb, and Zn; CsCl (Suprapur, Ca ≤ 0.05 mg/kg; Merck, Germany) and 5-sulfosalicylic acid (99.5 %, p.a.; Merck) were used for Ca. Calibration standards were AAS-grade single-element standards (Ca, Cd, Fe, P, Pb, Zn; Analytika, Czech Republic). Reference materials INCT-TL-1 Tea Leaves (INCT, Poland) and NIST SRM 1570a Spinach Leaves were digested in 250 mg portions per batch; no significant CRM bias was detected by t-test (p > 0.05), and spike recoveries on PBDA-matrix samples ranged from 94 to 110 %.

Method R² values ranged from 0.995 (Ca, FAAS) to 0.9999 (Cd, Fe, P, Pb by ETAAS). Limits of detection were 0.02 µg/L (Cd), 1 µg/L (Fe), 0.54 mg/L (P), 0.2 µg/L (Pb), 2 µg/L (Zn), and 0.02 mg/L (Ca) on the digestate solution, with LOQ taken as three times the LOD’s underlying standard deviation. Translated to the 1 g sample / 10 mL final volume product matrix, the in-product LOQ is 0.7 µg/kg for Cd and 7 µg/kg for Pb (as Table 3 reports). Quadratic calibration was used for P (y = 0.0005 + 0.0023x − 0.00001 × x²) per the source’s Table 1 footnote. Statistics were computed in OriginPro: non-parametric Mann–Whitney U for two-group comparisons (e.g., store-bought vs homemade seed milks) and Kruskal–Wallis with Dunn’s post-hoc for multi-group comparisons across PBDA categories or across seed types, given that the data did not follow a normal distribution. Significance level α = 0.05.

The authors describe several limitations themselves: PBDAs from the Czech market only (no breadth across the EU market); selected major elements only (Mn, Cu, I, Se, As, Ni, Hg not measured); total element content only (no speciation, no bioavailability assays); homemade seed milks limited to five seed types and a single household-style preparation protocol.

Implications

• Certification: The Pb-below-LOQ result across all 54 PBDA samples is an unambiguous occurrence-data point for store-bought plant milks, cream alternatives, yogurt alternatives, cheese alternatives, and tofu in the Czech/EU market — the in-product LOQ of 7 µg/kg Pb is fit for purpose for pooling with other plant-based-drink surveys that use comparable detection limits. The Cd signals are routing-relevant: tofu Cd at a median of 14.5 µg/kg contributes occurrence data for the soy-products page, and the homemade-poppy-seed-milk Cd at 58.8 µg/kg is a row-specific signal that any future homemade-seed-milk subcategory will need to track. The yogurt-alternative single high-value outlier (15.1 µg/kg max with median <0.7) flags within-category heterogeneity but is not itself routing-actionable until the source’s per-sample identity is recovered from Supplementary Table S1.
• App: Route to plant-milk, soy-milk, soy, almond, coconut, oat, rice-milk, rice, and milk-and-dairy ingredient pages, and to plant-milks-non-soy-non-rice, plant-milks-soy-based, plant-milks-rice-based, and soy-products product pages. Cd and Pb routing to the metals/cadmium and metals/lead pages is appropriate; the Pb routing carries a non-detect signal that is informative about Pb’s typical absence in this category.
• Courses: Useful for illustrating (a) how a homemade-vs-store-bought comparison can produce a counterintuitive direction (homemade seed milks contained higher Ca than store-bought plant milks but still failed to match cow’s milk), (b) how cadmium-accumulating plant taxa (poppy, flax) drive PBDA contamination signals that depend on the seed source rather than the beverage form, and (c) how a fortification-driven nutrition program can simultaneously address a Ca-deficiency problem and create a brand-vs-brand variability problem because individual fortificants differ in bioavailability.

Limitations

The sampling frame is restricted to the Czech market (commercial PBDAs in Brno supermarkets and health-food stores) plus a laboratory-prepared homemade oilseed-milk panel; the source does not establish how representative these are of the broader EU PBDA market. Only Ca, P, Fe, Zn, Cd, and Pb were measured: Mn, Cu, I, Se, total/inorganic As, Ni, and Hg are absent from this dataset and cannot be inferred. Total element contents were measured with no speciation work (no iAs/tAs split, no inorganic-vs-organic Hg or Cr work) and no bioavailability assays — the calcium-deficiency narrative is therefore based on total Ca contents and inhibitor presence (phytic acid, oxalates, dietary fiber, soy isoflavones, polyphenols) cited from the literature rather than measured in vitro digestibility on these samples. Sample sizes for non-milk categories are small (cream n=5, yogurt n=4, cheese n=4, tofu n=4, three samples per homemade seed type), and the source does not report within-product-type replicate counts beyond the single-mean-per-sample analytical workflow. The homemade-milk preparation protocol used a single seed:water ratio (20 g per 100 mL) and a single soak duration (20 h), so the homemade-milk Cd contents are not directly comparable to commercially-produced seed-milk beverages, which use proprietary extraction processes that may concentrate or dilute heavy-metal content differently. The “1.12 L per week” Czech average consumption figure used in the poppy-milk TWI calculation is the country-wide milk-equivalent consumption from the National Health Information Portal (ref 37), not measured per-capita poppy-milk consumption (which is presumably much lower in real life); the 35 % TWI figure is therefore an upper-bound substitution scenario, not a population exposure estimate.

Wiki pages this source may touch

• plant-milk
• soy-milk
• soy
• almond
• coconut
• oat
• rice-milk
• rice
• milk-and-dairy
• plant-milks-non-soy-non-rice
• plant-milks-soy-based
• plant-milks-rice-based
• soy-products
• cadmium
• lead
• efsa-cadmium-twi

Verification notes

Existing-page check. DOI grep (10.1007/s00217-025-04775-1), raw_handle grep (MFD_zverina2025-plant-dairy-toxic-elements), and cite-key glob (zverina*) over wiki/sources/ all returned no hits on 2026-06-08. This is a NEW source page — no prior version to merge-enhance.

Abstract vs Table 3 Cd value for homemade poppy-seed milk. The abstract states “homemade poppy seed milk, exhibiting an elevated cadmium level (0.052 ± 0.04 mg/kg)” (= 52 ± 40 µg/kg); Table 3 reports the homemade poppy-seed milk Cd median as 58.8 µg/kg (range 13.4–85.0). The abstract value is most consistent with the mean ± SD across the three poppy-milk samples, whereas Table 3 reports the median; the two are not contradictory (a median of 58.8 with a wide range of 13.4–85.0 is consistent with a sample mean near 52 if the three values span the reported range). Both values are quoted on this page in their respective contexts (Table 3 as median, abstract narrative as mean ± SD) with no silent reconciliation.

Brand firewall (Part 12). The source identifies samples only by product type (store-bought milk alternative, homemade flax-seed milk, etc.) and seed source. No brand names appear in the source text or in the wiki page. The methods section names instrument and reagent vendors (Analytika, Analpure, Anton Paar Multiwave Go plus, Analytik Jena ContrAA 800G/300, Merck, Suprapur, INCT, NIST 1570a, OriginPro) and one supplier-side reference (Czech Statistical Office for the 1.12 L/week consumption figure); these are scientific-method-vendor references permitted under the 2026-05-17 strict-reading exception to Part 12.

Speciation. The source measured total Cd and total Pb only — no isotopic or chemical-form speciation is reported. The metals: frontmatter entries are Cd and Pb accordingly. The paper did not measure As, Hg, Ni, Al, Cr, Sn, Sb, or any other element from the HMTc priority list.

Matrices vocabulary. The matrices array uses plant-based-drink, dairy-alternative, and soy-product per the broad-matrix convention established for begday2026 and good2026. The source covers six PBDA categories (milk alternatives, cream alternatives, yogurt alternatives, cheese alternatives, tofu, homemade seed milks) but only three of these have product slugs in the current taxonomy: plant-milks-non-soy-non-rice, plant-milks-soy-based, plant-milks-rice-based (for the seed-milk and store-bought milk categories), and soy-products (for tofu). Plant-based yogurt, cream, and cheese alternative categories do not have product slugs in the snapshot dated 2026-05-18 and so the cream/yogurt/cheese-alternative findings are not routed to a dedicated product page; they remain associated with the source page itself and will be retrievable when those categories are added to the taxonomy.

Ingredient slug coverage. The five oilseed types used to prepare homemade milks (flax, hemp, poppy, pumpkin, sesame) are not currently in the wiki ingredient taxonomy. The closest available umbrella slug (plant-milk) and the present-but-unrelated sunflower-seeds slug do not capture them. The seed-specific Cd findings are recorded in the Key numbers table on this page; downstream synthesis will need flax-, hemp-, poppy-, pumpkin-, and sesame-specific ingredient pages before per-seed contamination_profile values can be populated. This is a stub-creation candidate for tools/autonomy/create-missing-pages.mjs once the ≥freq-2 threshold is reached (e.g., by ingest of independent flax-seed or poppy-seed contamination studies).

Czech-Republic jurisdiction code. The samples were collected in Brno, Czech Republic; jurisdictions: uses CZ (ISO 3166-1 alpha-2) and EU per the system-prompt vocabulary.

License inference. The article declarations state: “This article is licensed under a Creative Commons Attribution 4.0 International License” and “Open Access publishing supported by the institutions participating in the CzechELib Transformative Agreement.” License recorded as CC-BY-4.0.

Sample-size attribution caveat. The source’s Methods text places “n=35” on the milk-alternatives bucket (20 store-bought + 15 homemade across five seed types). The 15 homemade samples are explicitly described as three samples per seed type for five seed types; the 20 store-bought milk-alternatives count does not enumerate per-base-ingredient (soy/oats/almond/coconut/rice/hemp/flax/poppy/pumpkin/sesame/other) subcounts, so per-store-bought-base-ingredient sample sizes cannot be recovered from the published Methods text alone (Supplementary Table S1 was not retrieved with this PDF and would need separate fetch to recover those per-base counts).

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.