Cammilleri et al. 2025 — Toxic metals and metalloids in fish and jellyfish collagen supplements

Summary

This is a primary occurrence survey of toxic metals and metalloids in collagen supplements of marine origin, covering both fish-derived and jellyfish-derived products. The authors measured arsenic, chromium, cadmium, lead, and mercury in 120 samples spanning 4 species-labeled brands (Atlantic cod, Pangasius, jellyfish Rhizostoma pulmo, Tilapia) sold on Italian e-commerce and retail channels during 2022–2023, plus a lab-extracted Scomber scombrus (mackerel) skin collagen for comparison. The load-bearing findings are that (1) arsenic is the dominant contaminant in fish-derived collagen — total As reached 1.11 mg/kg in the Atlantic cod brand and showed significant between-brand variation; (2) the Pangasius brand carried the highest mean concentrations of lead, cadmium, and chromium (with the highest Pb and Cd maxes; Tilapia held the Cr max); and (3) the jellyfish-derived (R. pulmo) brand and the in-house S. scombrus skin extract showed non-detectable concentrations for every metal measured, identifying jellyfish and mackerel-skin substrates as potential low-contaminant alternatives. No sample exceeded EU Regulation 2023/915 maximum levels (Cd 3 mg/kg, Pb 3 mg/kg, Hg 0.1 mg/kg for food supplements), and all estimated average daily intakes were below the relevant tolerable daily intakes used by the authors. For the Index, this establishes arsenic as the priority analyte for fish-derived marine collagen supplements, with the fish/marine origin as the plausible route for the arsenic burden, and identifies jellyfish-origin collagen as an empirically lower-contaminant substrate within the same product category.

Key numbers

ICP-MS for As, Cr, Cd, Pb on UltraWAVE-digested samples; direct mercury analyzer (DMA-80, atomic absorption) for Hg. Concentrations expressed as mg/kg wet weight of supplement. Arsenic is total As; the authors apply the EFSA iAs reference (0.3 µg/kg/day BMDL01) to the measured total As for the risk-assessment step but do not perform iAs speciation. Chromium is total Cr (no Cr-VI speciation).

Pooled totals across the dataset (Table 3, last row):

Between-brand pattern (Table 3, by labeled species):

• Atlantic cod (Gadus morhua, Brand 1): As mean 0.912 ± 0.16, max 1.11 mg/kg — significantly higher mean As than other brands (p = 3 × 10⁻⁷); Pb mean 0.123 ± 0.023, max 0.038 as printed in Table 3 (this value is internally inconsistent with the stated mean and the Figure 1 boxplot — see Verification notes); Cd mean 0.002 ± 0.002, max 0.006; Cr mean 0.054 ± 0.02, max 0.08.
• Pangasius (P. hypophthalmus, Brand 2): highest Pb (mean 0.136 ± 0.03, max 0.201), Cd (mean 0.003 ± 0.003, max 0.013), Cr (mean 0.053 ± 0.01, max 0.08), and the only brand with detectable Hg (max 0.0018 mg/kg in 2 samples); As mean 0.484 ± 0.16, max 0.64.
• Jellyfish (Rhizostoma pulmo, Brand 3): n.d. for all 5 metals measured.
• Tilapia (Tilapia spp., Brand 4): As mean 0.390 ± 0.2, max 0.71; Pb mean 0.125 ± 0.01, max 0.14; Cd mean 0.002 ± 0.001, max 0.011; Cr mean 0.040 ± 0.03, max 0.14.
• S. scombrus skin extract (in-house, not commercial): n.d. for all 5 metals measured.

Compliance and intake context:

• No sample exceeded EU Reg 2023/915 maximum levels for food supplements (Cd 3 mg/kg, Pb 3 mg/kg, Hg 0.1 mg/kg).
• Average daily intakes (ADDs) at 5 g and 10 g daily doses for a 67.7 kg adult woman were below the corresponding TDIs for iAs (0.3 µg/kg/day BMDL01, EFSA), Pb (0.5 µg/kg/day BMDL01, EFSA), and Cd (0.36 µg/kg/day from TWI 2.5 µg/kg/week, EFSA). Table 4 reports per-brand and pooled ADDs (e.g., total Pb ADD 1.4 × 10⁻² µg/kg/day; total iAs ADD 4.9 × 10⁻² µg/kg/day; total Cd ADD 3 × 10⁻⁴ µg/kg/day) and shows all ADD-as-percent-of-TDI values rounding to 0 at the precision printed. The authors qualitatively flag Pb as “alarming” in the Conclusions despite the formal compliance.

Methods (brief)

UltraWAVE microwave digestion (3 mL 60% HNO₃ + 5 mL ultrapure water; ~0.1 g supplement) followed by 0.45 µm filtration to 50 mL final volume. ICP-MS on Agilent 7700x for As, Cr, Cd, Pb; the LOD by 3σ and LOQ by 10σ approach. Hg measured separately by Milestone DMA-80 direct mercury analyzer (atomic absorption at 250 nm, no digestion required). 7-point calibration with r² > 0.999; CRM DORM-5 (fish protein) verified accuracy; recovery 96–105% across three concentration points. Repeatability and within-laboratory reproducibility assessed across three days with operator changes. Not-detected values imputed at LOQ/2 for statistical analysis (Shapiro–Wilk normality, Kruskal–Wallis between-brand). Risk step used EFSA inorganic-As assumption applied to total As; reference body weight 67.7 kg (woman).

Implications

• Establishes total arsenic as the priority toxic metal for fish-derived marine collagen supplements within Ingredient class — protein/collagen powders, with Atlantic cod showing the highest As burden and Pangasius the highest co-burden of Pb, Cd, and Cr. Complements the protein-powder cadmium signal from bandara2020-protein-powder-heavy-metals — collagen powders track arsenic (and a Pangasius-specific multi-metal burden), protein/whey powders track cadmium.
• Jellyfish-derived collagen (R. pulmo) and mackerel-skin (S. scombrus) extract show non-detectable toxic metals across the full panel, identifying jellyfish biomass as an empirically lower-contaminant substrate. Relevant to supply-chain screening for collagen sourcing and to any future product subcategory that distinguishes source-organism within collagen powders.
• Contributes occurrence evidence to Arsenic (1.11 mg/kg max, fish-marine route), Lead (0.201 mg/kg max, Pangasius-driven), Cadmium (low-level but near-ubiquitous in fish-derived; n.d. in jellyfish), Chromium (low-level, total Cr only, no Cr-VI speciation), and Mercury (largely absent except in Pangasius).

Wiki pages this source may touch

• Ingredient class — protein/collagen powders — fish-derived collagen carries As-dominant toxic-metal profile with a Pangasius multi-metal signature; jellyfish-derived collagen empirically clean.
• Arsenic — dominant contaminant; fish/marine-origin route; 1.11 mg/kg max in cod brand.
• Lead — 0.201 mg/kg max, Pangasius-driven.
• Cadmium — near-ubiquitous in fish-derived (98% detection), n.d. in jellyfish.
• Chromium — total Cr only, low-level occurrence.
• Mercury — largely absent (Pangasius-only trace).

Verification notes

Merge-enhanced 2026-06-09 from the 2026-06-08 ingest. The prior version of this page contained the following defects, corrected against the source PDF in this revision:

• Title was truncated to “Toxic metals in marine collagen supplements: occurrence and risk assessment”; corrected to the actual paper title “Toxic metals and metalloids in collagen supplements of fish and jellyfish origin: Risk assessment for daily intake” (paper cover page).
• Chromium was missing from the metals frontmatter and Key numbers; the paper measures 5 elements (Pb, Cd, Cr, Hg, As — Abstract, Methods §2.3, Table 3). Added Cr.
• Cadmium maximum was reported as 0.0018 mg/kg; this value is the mercury maximum (Results §3, “mercury was found in only two samples from brand 2, showing a maximum of 0.0018 mg/kg”). The actual Cd total max is 0.013 mg/kg (Table 3, total row). Corrected the value transposition.
• Lead maximum was missing; Table 3 reports total Pb max 0.201 mg/kg, min 0.094 mg/kg. Added.
• The 1.11 mg/kg As maximum was attributed to a Pangasius (catfish) brand in the prior page; the paper assigns this maximum to Brand 1 (Atlantic cod, Gadus morhua) — Table 3 and Results §3 (“As was the most abundant trace element, with a maximum concentration of 1.11 mg/kg in a marine collagen sample of brand 1”; “Mean concentrations of As in brand 1 were significantly higher than the other brands”). Pangasius (Brand 2) is the worst performer for Pb, Cd, and Cr but not As. Corrected the species attribution.
• Jellyfish was omitted entirely from matrices and sample_population; Brand 3 is jellyfish (Rhizostoma pulmo) per Methods §2.2 and Table 3. Added jellyfish-collagen matrix and updated sample_population.
• The non-detection finding for jellyfish and S. scombrus skin extract was omitted; this is a load-bearing finding for collagen-substrate selection and is added to the Key numbers and Implications sections.
• Sampling year range was reported as 2022; Methods §2.2 states purchases were made “during 2022 and 2023”. Corrected to 2022-2023.
• Authors list was truncated with “et al.”; expanded to the full 11-author list from the paper cover.
• Mercury analytical method (DMA-80 direct mercury analyzer, atomic absorption) was conflated with the ICP-MS measurement of the other four elements; clarified in Methods and Key numbers.
• EU regulatory compliance finding (no samples exceeded Reg 2023/915 limits) and the per-element ADD-vs-TDI risk assessment outcomes were omitted; added to Key numbers for regulatory context.

Brand-firewall: the paper uses anonymized “brand 1/2/3/4” labels with only species attribution disclosed (no commercial brand names). Per Part 12, species-level results are reproduced (species are a supply-chain category, not a brand identity); the underlying anonymized brand-by-brand sample counts are not reproduced beyond what is necessary for the species-level findings.

Hg detection-percentage note: the abstract states “Hg was detectable in only 12% of the samples” while the Results section narrates “mercury was found in only two samples from brand 2.” These two figures cannot both be true under the stated 4-brand × 30-sample design (2/120 = 1.7%; 2/30 = 6.7%); neither matches 12%. The Results-section narrative (Pangasius-only, 2 samples, max 0.0018 mg/kg) is reported as canonical because it is more specific and matches the species pattern of all other metals; the abstract 12% figure is treated as a likely abstract reporting error rather than substantive disagreement.

Audit pass 2026-06-09 — applied corrections from the fresh-context audit subagent (REVISE verdict):

• Brand 1 Pb max: prior version reported 0.138 mg/kg (which is the total-row Pb mean, not the cod-brand max). Table 3 prints Brand 1 Pb max as 0.038 mg/kg, which is itself internally inconsistent with the stated Brand 1 mean (0.123 ± 0.023) and with the Figure 1 boxplot (Brand 1 Pb whiskers extending to ~0.17 mg/kg). Treated as a likely typo in the published Table 3; the printed value is reported with the inconsistency flagged. Audit subagent (2026-06-09) flagged “0.138” as wrong; verified against source — corrected to the printed “0.038” with explicit paper-internal-inconsistency caveat.
• ADD-as-percent-of-TDI: prior version stated “highest ADD-to-TDI ratio was Pb at the cod brand (~14% of TDI at the upper dose)“. This figure is not in the paper — Table 4 shows the percent-of-TDI parenthesized values rounding to (0) at the precision printed; back-calculation from Pb TDI 0.5 µg/kg/day and total Pb ADD 1.4 × 10⁻² µg/kg/day gives ~2.8%, not 14%. Audit subagent flagged the 14% as unsupported; verified against source — removed and replaced with the actual Table 4 ADD values and the authors’ qualitative “alarming Pb” wording from Conclusions.
• Cr “consistent worst performer” attribution to Pangasius: Pangasius has the highest Cr mean (0.053) but Tilapia has the highest Cr max (0.14, vs Pangasius 0.08); the paper Discussion §3 explicitly says “Higher mean Cr, Pb, and Cd values were found in… Pangasius”. Audit subagent flagged the wording; verified — Summary revised to “highest mean concentrations of lead, cadmium, and chromium (with the highest Pb and Cd maxes; Tilapia held the Cr max)“.
• HMTc framing in Summary (“within the HMTc supplements-protein-collagen-powders scope”) and in Implications (“any HMTc category that distinguishes source-organism within collagen powders”) was a Part 2 boundary concern. Audit subagent flagged both; applied — Summary rephrased without HMTc reference; Implications changed to “any future product subcategory that distinguishes source-organism within collagen powders”.

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.