EFSA AFC Panel 2008 — Safety of Aluminium from Dietary Intake

The European Food Safety Authority Panel on Food Additives, Flavourings, Processing Aids and Food Contact Materials (AFC) Scientific Opinion on the Safety of Aluminium from Dietary Intake (EFSA Journal 2008;754:1-34, Question Nos EFSA-Q-2006-168 and EFSA-Q-2008-254), adopted on 22 May 2008, established a tolerable weekly intake (TWI) of 1 mg aluminium per kilogram body weight per week, derived from combined evidence on neurotoxicity, testicular toxicity, embryotoxicity and developmental neurotoxicity in mice, rats and dogs after dietary administration of aluminium compounds. The TWI was set as a rounded value between the TWI of 0.7 mg/kg b.w./week obtained by applying a 100-fold uncertainty factor to a developmental-neurotoxicity NOAEL of 10 mg Al/kg b.w./day in mice and the TWI of 1.2 mg/kg b.w./week obtained by applying a 300-fold uncertainty factor (100 × 3 for LOAEL-to-NOAEL) to the corresponding LOAEL of 50 mg Al/kg b.w./day. The opinion replaced the 1990 Scientific Committee for Food endorsement of the prior JECFA provisional tolerable weekly intake of 7 mg Al/kg b.w./week and aligned EFSA with the revised PTWI of 1 mg/kg b.w./week adopted by JECFA at its 67th meeting (2006). The Panel concluded that the TWI of 1 mg/kg b.w./week is likely to be exceeded in a significant part of the European population, particularly among consumers of foods containing aluminium-containing food additives, infants fed certain formulae, and adults consuming aluminium-containing medications. Subsequent regulatory action on aluminium-containing food additives was issued by the European Commission as Regulation (EU) No 380/2012 amending Annex II to Regulation (EC) No 1333/2008.

Key numbers

Reference points (Section 7, Conclusions):

Lowest NOAELs and LOAELs from dietary administration of aluminium compounds in mice, rats and dogs (Section 7):

Dietary aluminium exposure across European adults compiled from duplicate-diet studies (Netherlands, Hungary, Germany, Sweden, Italy) and market-basket/total-diet studies (UK, Finland, France) (Section 4):

Potential dietary aluminium exposure in infants 0–3, 4–6, 7–9 and 10–12 months from infant formulae and other foods manufactured specially for infants, FSA 2006 menu-based estimates assuming maximum recommended food amounts (Section 4):

Estimated dietary aluminium exposure in 3-month-old infants (6.1 kg b.w.) from a variety of infant formulae, based on Navarro-Blasco & Alvarez-Galindo (2003) analytical data for reconstituted formulae available on the Spanish market (Section 4):

The Panel noted that in some individual formulae (both milk-based and soya-based) the aluminium concentration was around four times higher than the means above, leading to roughly four-fold higher potential exposure in infants fed those formulae loyally (Section 4; brand identities not reported in the opinion).

Total dietary exposure as fraction of the TWI:

Aluminium chemistry and bioavailability (Sections 1, 5.1):

Reported aluminium concentrations in food categories (Section 4, compiled from German, French, UK, Irish and Spanish surveys; total aluminium, not speciated):

EU-authorised maximum levels for aluminium-containing food additives selected from Table 3 (Directive 95/2/EC as modified):

Aluminium-containing additives authorised in food-contact plastics under Directive 2002/72/EC (Table 2) include aluminium calcium hydroxide phosphite hydrate (Ref 34475), aluminium fibres, flakes and powders (Ref 34480), aluminium hydroxide (Ref 34560), aluminium hydroxybis[2,2′-methylenebis(4,6-di-tert-butylphenyl)phosphate] (Ref 34650, SML = 5 mg/kg), aluminium magnesium carbonate hydroxide (Ref 34690), aluminium oxide (Ref 34720) and silicic acid, lithium aluminium salt (Ref 85760, SML(T) relative to lithium).

Methods (brief)

Scientific opinion synthesising the chemistry, sources, regulated uses, dietary exposure and biological/toxicological data on aluminium in food. The opinion does not generate primary measurements; it compiles and re-evaluates existing literature, EU food-additive directives (95/2/EC, 94/36/EC, 2002/72/EC, 96/77/EC, 95/45/EC), drinking-water and mineral-water regulations (Directives 98/83/EC and 2003/40/CE), JECFA assessments (1985, 1989, 2006 67th meeting), Codex monographs, IPCS Environmental Health Criteria 194 (WHO 1997), and individual peer-reviewed studies on absorption, distribution, excretion, acute and subchronic toxicity, genotoxicity, carcinogenicity, reproductive and developmental toxicity, and neurotoxicity. Dietary-exposure estimates are taken from duplicate-diet studies (Netherlands, Hungary, Germany, Sweden, Italy) and market-basket/total-diet studies (UK, France, Finland, Ireland, Spain), and from the UK Food Standards Agency 2006 Survey of Metals in Weaning Foods and Formulae for Infants (FSA Food Survey Information Sheet FSIS 17/06). Infant-formula exposure estimates rely on Navarro-Blasco & Alvarez-Galindo (Food Additives and Contaminants 20:470–481, 2003) for Spanish-market formulae. Reported concentrations are total aluminium; the underlying analytical methods (typically ICP-MS or GFAAS in the cited surveys) are described in the original references, not in this opinion. Toxicological reference points are derived from dietary studies in mice, rats and dogs that used soluble aluminium salts (lactate, chloride, nitrate, sulphate) or sodium aluminium phosphate (SALP, acidic and basic).

Limitations stated by the Panel: dietary studies determine total aluminium content of food, not individual aluminium compounds or species, so source attribution between natural background, food-additive use, and migration from food-contact materials is not possible; most biochemical and toxicological studies did not measure background aluminium in the basal diet, so stated doses likely underestimate total aluminium exposure; few studies report NOAELs and LOAELs in the dose ranges relevant to dietary exposure; and there are no specific toxicological data on most aluminium-containing food additives apart from SALP (acidic).

Implications

Certification: this opinion contributes the EU-level TWI of 1 mg Al/kg b.w./week (total aluminium) and a quantified exposure distribution against that TWI — soya-based-infant-formula-fed 3-month infants at ~110% of the TWI under the Navarro-Blasco high-consumption scenario, UK toddlers (1.5–4.5 y) at 230% of the TWI at the 97.5th percentile, UK elderly in care at 114% of the TWI, and country-mean adult exposures spanning 20–150% of the TWI. These figures are the regulatory-baseline evidence on aluminium exposure for the matrices the opinion covers (infant formulae, bakery and cereals, drinking water, foods using aluminium-containing additives, and foods in contact with aluminium foils/cookware) and should be read alongside any later EFSA aluminium publications (e.g., the 2021 Q&A summarised at efsa2021-aluminium-food-qa) when those become contributing sources to the synthesis pass.

Courses: regulatory-affairs course materials cite this opinion as the European derivation of the 1 mg/kg b.w./week TWI and as the source for the “TWI is likely to be exceeded” conclusion that drives precautionary reformulation pressure on bakery, dairy, infant-formula, and food-additive sectors.

App: aluminium exposure estimates in the app for infant-formula consumers should use the per-form values reported here (mean 0.6 mg/kg b.w./week milk-based, 0.75 mg/kg b.w./week soya-based; high 0.9 and 1.1 mg/kg b.w./week respectively) until superseded by EFSA’s later opinions or by the FSA 2013 update. The Panel’s observation that some individual formulae were ~4× higher than the mean is an important variance signal for the high-end exposure tail.

Microbiome: not applicable. The opinion does not address gut-microbiome effects of aluminium.

Verification notes

Merge-enhanced 2026-06-03 from the prior 2026-04-25 revision (raw_path 9d04359da…) using the re-downloaded copy in the Kimi June 3 Folder ( 2.pdf, sha256 183affd8…) under the ingest-next-manual-fetch-pdf skill. Enhancements:

• Removed the legacy ## Summary heading per the post-2026-05-08 schema; converted that section into the un-headed opening prose under the H1 (CLAUDE.md Part 6).
• Removed the legacy ## Wiki pages updated on ingest heading; routing destinations are now derived from frontmatter metals, ingredients, products, matrices and jurisdictions arrays by the routing audit (CLAUDE.md Part 5b).
• Added raw_handle: MFK_efsa-journal-2008-safety-of-aluminium-from-dietary to record the Kimi-folder origin of the re-downloaded copy.
• Added near_duplicates with the two byte-distinct copies in the June 3 Folder (sha256 3ed580ef… and 183affd8…); the 3.pdf is byte-identical to the 2.pdf so a single near_duplicates entry covers both.
• Added products array (umbrella scopes: infant-formula-dairy, infant-formula-dairy-free, bread-and-baked-goods, bottled-drinking-water, food-packaging-foils-wraps, cookware-metal-alloy) — the opinion covers infant formulae split by milk/soya base, the bread-and-baked-goods category as the leading bakery-additive-exposure driver, drinking-water under Directive 98/83/EC, and aluminium migration from food-contact foils/wraps and cookware. This fixes the products: [] advisory recorded against this source in data/evidence/routing_malformed.csv (row 218 as of 2026-06-03).
• Added matrices: [dietary-intake] consistent with the convention used on the EFSA CONTAM 2009 and 2010 opinion pages.
• Added broader ingredient slugs that the opinion discusses materially (cereals, bread, tea, cocoa, herbs-and-spices, vegetables) on top of the prior water and infant-formula-ingredients.
• Replaced the thin parameter table with explicit exposure tables (adults by country, children/toddler 97.5th percentile, infants by age group, infants by formula form including the 4× brand-loyal variance) and the LOAEL/NOAEL table by endpoint that anchor the TWI derivation. All values verified against the PDF pages 1–4 (Summary, Conclusions), 16–18 (Section 4 Dietary Exposure), 21–23 (Sections 5.3 / 5.6 / 5.7 Subchronic, Reproductive, Neurotoxicity) and 28–29 (Section 7 Conclusions).
• Replaced the line “JECFA also subsequently lowered its PTWI to 1 mg/kg b.w./week aligning with EFSA” — the chronology is the reverse of what that line implies. JECFA established the revised 1 mg/kg b.w./week PTWI at its 67th meeting (June 2006), and EFSA AFC aligned in this 2008 opinion. Verified against the Background section, p. 6: “Recently at its sixty-seventh meeting, JECFA re-evaluated aluminium from all sources … and established a PTWI of 1 mg/kg bw which is 7 times lower than the previous PTWI.”
• Brand-firewall check (CLAUDE.md Part 12): the opinion itself does not name brands; it cites Navarro-Blasco & Alvarez-Galindo (2003) for Spanish-market formulae but reports only milk-based vs soya-based aggregates and a “some individual formulae … around 4 times higher” variance note. The merged page preserves the form descriptors (milk-based vs soya-based) and the 4× brand-loyal variance without naming any product line, consistent with the 2026-05-17 stricter brand-firewall reading.
• Added a Methods (brief) section per current schema (CLAUDE.md Part 6); the opinion is a regulatory synthesis so the section documents what it compiles rather than primary analytical methods.
• Added an Implications section per current schema (Certification / Courses / App / Microbiome), keeping it descriptive of the source’s own findings — no HMT&C threshold proposals or wiki-side synthesis claims (CLAUDE.md Part 2).

Audit subagent (2026-06-03, fresh-context general-purpose agent invoked via the ingest-next-manual-fetch-pdf skill v2.0) verdict REVISE. Findings applied:

• Check 1 ⚠️ on “10th-decile dietary exposure” for the 1988 former-West-Germany duplicate diet row — verified against PDF p. 17 (“10% of children aged 5-8 years had an exposure higher than 0.38 mg/kg bw/week”). The original phrasing was ambiguous (a reader could read “10th decile” as either top or bottom). Rewrote as “90th-percentile dietary exposure (the value above which 10% of children fell)” to anchor the statistic explicitly.
• Check 1 ⚠️ on an unanchored “2025 dietary-exposure update” reference and a [[sources/efsa-cef-aluminium-additives-2013]] wikilink in Implications — verified independently: there is no wiki/sources/efsa-cef-aluminium-additives-2013.md, so the wikilink was an orphan. Removed both the 2013 wikilink and the 2025 reference; replaced the synthesis-pointer with [[sources/efsa2021-aluminium-food-qa]], which exists in wiki/sources/ and is the natural later-EFSA Al publication to read against this page.
• Check 4 ⚠️ on the Implications/Certification paragraph leaning into HMT&C-ratcheting framing (“the literature anchor for any ratcheted aluminium standard tighter than current regulatory ceilings”) — verified against CLAUDE.md Part 2: the original phrasing crossed the wiki/HMT&C boundary by naming ratcheting as the use case. Rewrote to descriptive source-contribution language (“contributes the EU-level TWI … and a quantified exposure distribution against that TWI”) that reports what the opinion supplies without invoking HMT&C policy logic.

Findings noted but not changed:

• Check 2 ⚠️ on matrices: [dietary-intake] — the auditor flagged this as not strictly a “matrix” slug. Left in place because it follows the established cross-page convention for whole-diet EFSA regulatory opinions (e.g., efsa-arsenic-contam-2009.md, efsa-cadmium-contam-2009.md). If Karen decides the matrices vocabulary should distinguish exposure-pathway opinions from per-commodity studies, the change would apply uniformly across those sibling pages.

All five audit-check categories were verified. Numerical fidelity ✅ across ~25 spot-checked values (TWI, prior PTWI, NOAELs/LOAELs by endpoint, country-level adult and child 97.5th-percentile exposures, infant exposures by age band and formula type, chemistry/bioavailability reference points, drinking-water parametric value, EU additive maxima from Table 3). Brand firewall ✅. Speciation discipline ✅ (total aluminium throughout; no invented speciation). Wiki/HMT&C firewall now ✅ after the Implications rewrite.

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.