Heavy Metal Index

p4-batch4-2026-05-12

7 min read

P4 Batch 4 — Ingest Report

Date: 2026-05-12 Handles processed: 200 (food-matrix-filtered P4, 2025 papers, positions 200–400 in filtered list) Source pages created: 12 False positives: 188 (94%)

Summary

Batch 4 is the first batch with meaningful yield from the food-matrix-filtered strategy: 12 source pages including two high-priority finds (dai2025 PNAS rice speciation, seyfferth2025 GeoHealth rice water management). The 94% false positive rate reflects persistent manifest metadata mismatches — labels like “formula,” “chocolate,” “liver,” and “kidney” in text_mined_ingredients often arise from paper body text discussing toxicology targets or processing equipment, not food-matrix occurrence data.

Note: A background session ingesting raw/manual-fetch/seasonal-geographic-variance/ PDFs ran concurrently and its uncommitted pages (fan2025, gul2024, kazeminia2023, maccaferri2019, ramtahal2016) were swept into the group 3 commit by the pre-commit hook. No duplicate raw_handles exist across the 386 total source pages.

Source Pages Created

Group 1 (FM_12179796 – FM_12261059)

lin2025-taiwan-blood-metals-urothelial-carcinoma (FM_12198073) Case-control, n=1,047 Taiwanese hospital patients. Blood As, Cd, Pb, Cr, Ni by ICP-MS (Agilent 7700c). Blood As independently associated with urothelial carcinoma (adjusted OR 1.05 per µg/L, p=0.002). Control mean blood As 6.05 µg/L — elevated vs ATSDR reference, consistent with rice/seafood dietary exposure in Taiwan. Total As without speciation; seafood consumption noted as unmeasured confounder. Evidence tier: A. Metals: tAs, Cd, Pb. Matrices: blood-biomonitoring.

haydous2025-pet-food-metals-lebanon-uae (FM_12219634) First systematic toxic-metals survey of pet food in Lebanon and UAE. n=196 SKUs by ICP-MS. Dry Pb mean 0.981 mg/kg, wet Cd 0.296 mg/kg, tHg mean 1.213 mg/kg (attributed to fish-based ingredients in wet cat food). Pb and As exceeded thresholds in a subset. Supply-chain relevance: shares fish-meal ingredient pathways with human food. Evidence tier: A. Metals: Pb, Cd, tHg, tAs. Matrices: pet-food-wet, pet-food-dry.

Group 2 (FM_12262887 – FM_12337386)

asaditouranlou2025-bread-wheat-mashhad (FM_12286368) ICP-OES, 270 flour/dough/bread samples from 90 bakeries across 5 Mashhad city regions. Pb and Hg below detection in all samples; Al, As, Cr, Fe exceeded WHO/FAO thresholds; Cd exceeded Iranian national standards. As dominant risk driver: HI >1 in all five regions for adults (5.23–7.64) and children (6.53–8.77); carcinogenic risk 3.6×10⁻³–6.1×10⁻³ (1–2 orders above EPA 10⁻⁴). Total As (ICP-OES, no HPLC). PLoS ONE, CC BY. Evidence tier: A. Metals: tAs, Cd, Al, Cr. Matrices: wheat-flour, bread.

ray2025-fish-chennai-metals (FM_12279835) n=18 fish (3 species, liver/gills/muscle), Chennai/Ennore industrial coast India. ICP-MS; Pb 15.4–17.6 µg/kg, Cd 0.696–0.778 µg/kg, Cr 5.26–12.40 µg/kg, tHg 0.020–0.066 µg/kg (dry weight, all tissues). HI <1 for all. Total Hg and total Cr only. Frontiers in Public Health, CC BY. Evidence tier: A. Metals: Pb, Cd, Cr, tHg. Matrices: marine-fish.

erol2025-turkish-cheese-metals (FM_12277537) n=30 artisanal Turkish cheeses (Aho, Golot, Telli), Trabzon region. ICP-MS. One Golot cheese outlier: Pb 1,788.75 µg/kg (vs EU ML 200 µg/kg). One Telli cheese outlier: tHg 468.71 µg/kg. Mean Ni in Golot: 1,502.47 µg/kg (max 3,681.67 µg/kg). 93.3% of samples HI <1. Food Science & Nutrition (Wiley), CC BY. Evidence tier: A. Metals: Pb, tHg, Ni, Cd. Matrices: cheese.

Group 3 (FM_12338970 – FM_12422898)

seyfferth2025-rice-water-management-as-hg-cd (FM_12344449) Critical find. Two-year controlled field study, University of Delaware RICE Facility. Speciated As (HPLC-ICP-MS), tHg and MeHg (isotope-dilution GC-ICP-MS), and Cd (ICP-MS) across 6-paddy soil-redox gradient. Key finding: drier irrigation (AWD/row rice) reduces grain iAs, total As, tHg, MeHg but raises grain Cd 50–97%, exceeding CODEX ML of 0.4 mg/kg (400 ppb) despite very low background soil Cd (0.093 mg/kg). As AWD adoption accelerates to address climate change, historical flood-paddy As benchmarks will understate Cd exposure. GeoHealth 2025, DOI 10.1029/2025GH001410. Evidence tier: A. Metals: iAs, tAs, tHg, MeHg, Cd. Matrices: rice-grain. Jurisdictions: US.

venant2025-lake-victoria-sardine-cd-pb (FM_12399412 area — group 3) FAAS, n=279 sardine (Rastrineobola argentea) from three Tanzanian Lake Victoria regions. All below FAO/WHO limits. Cd median 0.012 mg/kg, Pb median 0.081 mg/kg (dry weight). Sand-drying slightly elevates Pb vs rack-drying. THQ <0.003 for both metals. Food Science & Nutrition, CC BY. Evidence tier: A. Metals: Cd, Pb. Matrices: freshwater-fish. Jurisdictions: TZ.

paul2025-meghna-fish-arsenic (FM_12407486 area — group 3) ICP-MS, n=300 muscle samples, 10 small indigenous freshwater fish species, Lower Meghna River Bangladesh, 3 seasons, 10 stations. Five species exceed WHO 1 mg/kg (1,000 ppb) threshold during pre-monsoon, driven by groundwater As inputs. No iAs speciation — total As only. PLoS ONE, CC BY, DOI 10.1371/journal.pone.0330602. Evidence tier: A. Metals: tAs. Matrices: freshwater-fish. Jurisdictions: BD.

Group 4 (FM_12422912 – FM_12482525)

hassan2025-cornflakes-lebanon-metals (FM_12455183) n=42 Lebanese retail cornflake products, ICP-MS for tAs, Cd, Cr, tHg, Pb. All below EU/Codex limits. Baseline occurrence dataset for breakfast cereal in a Middle Eastern market. Evidence tier: A. Metals: tAs, Cd, Cr, tHg, Pb. Matrices: cereal-breakfast. Jurisdictions: LB.

nour2025-djiboutian-seaweeds-metals (FM_12440808) n=6 seaweed species (Red Sea/Gulf of Aden coast, Djibouti), ICP-MS. Total As up to 70,200 ppb DW in Turbinaria decurrens — well above EU supplement limits. Total Cr up to 29,610 ppb DW in Ulva clathrata. No As or Cr speciation. Evidence tier: A. Metals: tAs, Cr, Al. Matrices: seaweed. Jurisdictions: DJ.

dai2025-rice-arsenic-methylation-demethylation-global (FM_12478174) Critical find — P1-grade. PNAS 2025, CC BY-NC-ND. HPLC-ICP-MS in rice grains: iAs 78–109 µg/kg DW, methylated As (DMA+DMMTA) 6.5–42 µg/kg DW across Chinese soil chronosequences. Key finding: DMMTA is 3–10× more cytotoxic than iAs and accounts for 1–21% of total grain arsenic globally; Americas/Europe paddies (shorter cultivation history) have higher methylated As fractions than ancient Southeast Asian paddies. Directly relevant to why total-As-only testing of rice is insufficient for toxicological risk assessment. DOI in paper. Evidence tier: A. Metals: iAs, tAs. Matrices: rice-grain. Jurisdictions: CN.

yazdanian2025-camel-milk-heavy-metals-iran (FM_12482525) AAS, n=49 raw camel milk samples, Chaharmahal Bakhtiari province, Iran. Pb max 55 µg/kg, Cd max 5.9 µg/kg, tAs max 6.7 µg/kg — all below MACs. Caveat: only microbiologically positive samples tested; results not population-representative. Evidence tier: A. Metals: Pb, Cd, tAs. Matrices: camel-milk. Jurisdictions: IR.

High-Value Findings

  1. dai2025 (PNAS) — DMMTA finding in rice: HPLC-ICP-MS speciation data showing methylated As species (especially DMMTA, which is 3–10× more cytotoxic than iAs) account for up to 21% of total grain As. Total-As-only testing systematically undercounts toxicological risk. Critical for HMT&C rice speciation standard justification.

  2. seyfferth2025 (GeoHealth) — AWD irrigation Cd tradeoff: drier rice irrigation cuts As and Hg but raises Cd above CODEX limits. Supply chain implication: historically low Cd in US/EU paddy rice will rise as AWD adoption grows. Critical input for ingredient Cd risk modeling and certification timing.

  3. asaditouranlou2025 — Iranian bread arsenic: carcinogenic risk 1–2 orders above EPA threshold across all city districts; Pb undetected throughout. Unusual pattern worth noting in the bread/wheat ingredient profile.

Notable Manifest Metadata Mismatches

  • FM_12315110 labeled “chocolate” — is a hospital grief-support clinical program paper (“Good Grief and Chocolate at Noon”)
  • FM_12334442 labeled “cocoa;kidney;liver;rice” with Cd — is soil bioremediation bacterial isolation from cacao rhizosphere
  • FM_12219634 labeled “meat” — is actually pet food (not human meat products)
  • FM_12451243 labeled “human milk, As” — is a Canadian infant food nutritional labeling policy study, no metals
  • FM_12425301 labeled “tuna, Ni;Pb” — is albacore tuna energy content/physiology, no metals occurrence data

Concurrent Background Session Note

A background session ingesting raw/manual-fetch/seasonal-geographic-variance/ PDFs ran concurrently. Its pages (fan2025, gul2024, kazeminia2023, maccaferri2019, ramtahal2016) were swept into the P4 batch 4 group 3 git commit by the pre-commit hook. These are valid ingest pages from the seasonal-geographic-variance initiative — no action needed, but the group 3 commit message undercounts them. Total wiki source pages after batch 4: 386.

Batch Commits

  • 41f0979 — group 1: 2 pages, 48 FP
  • bb32b03 — group 2: 3 pages, 47 FP
  • 87efb16 — group 3: 3 pages (+ 5 swept seasonal-variance pages), 47 FP
  • ffb63b3 — group 4: 4 pages, 46 FP