16.3 International Regulatory Harmonization: OECD, ICH Guidelines & EU REACH

Key Takeaways

  • OECD Mutual Acceptance of Data (Council Decision C(81)30, 1981) requires member countries to accept, for assessment, chemical-safety data generated in another member (or MAD-adherent) country when both OECD Test Guidelines and OECD GLP were followed. Interpretation and extra testing remain national prerogatives.
  • ICH Safety, selected Quality (Q3), and Multidisciplinary (M3, M7) guidelines time and scope pharmaceutical nonclinical work across the US, EU, and Japan; they do not replace EPA OCSPP or FDA Redbook food playbooks.
  • EU REACH (EC 1907/2006) is no data, no market at ≥1 tonne/year. Annexes VII–X escalate with 1–10, 10–100, 100–1,000, and ≥1,000 t/y bands; SVHCs can move from the Candidate List to Annex XIV authorization; Annex XVII imposes restrictions. DNEL is a threshold dose-descriptor/assessment-factor value; DMEL is a non-threshold risk construct (Chapter 13).
  • Green chemistry and benign-by-design cut hazard at the molecule (IV.7 A). Sustainable substitution can still be regrettable (for example replacing a solvent with 1-bromopropane). EPA Safer Choice, USDA BioPreferred, and NSF/ANSI certifications answer different questions (hazard criteria versus biobased content versus drinking-water standards).
  • Product stewardship (IV.9) uses tiered toxicology, labeling, and manufacturing controls to manage intended use, reasonably foreseeable misuse, and environmental release — not a single chronic rat study for every SKU.
Last updated: September 2026

Why “international” is an applied-toxicology skill, not a travel footnote

Handbook I.A.2 already asked you to pick OECD, FDA, EPA, ICH, EMA guidelines when designing a study. Domain IV asks you to live inside those systems when the product is a medicine, a tonne of industrial chemical in the EU, or a consumer cleaner claiming to be greener. Independent OpenExamPrep teaching in this section covers OECD Mutual Acceptance of Data (MAD) and Test Guidelines, ICH safety guidelines, EU REACH registration / evaluation / authorization / restriction, a recap of DNEL versus DMEL from Chapter 13, green chemistry and regrettable substitution (IV.7 A), EPA Safer Choice, USDA Biobased, and NSF International (IV.7 C), and product stewardship (IV.9). It is not an OECD, ICH, ECHA, EPA, USDA, NSF, or ABT product and does not claim official approval, review, or partnership.

OECD MAD and Test Guidelines

The Organisation for Economic Co-operation and Development Council Decision C(81)30(Final) (12 May 1981) is the MAD core. Data generated in the testing of chemicals in an OECD member country in accordance with OECD Test Guidelines and OECD Principles of GLP shall be accepted in other member countries for assessment and other uses relating to protection of health and the environment. C(89)87 covers GLP compliance monitoring. C(97)114 lets non-member countries adhere to MAD. The bargain is no duplicate animal testing when the method and the quality system match. It is not a promise that Country B will copy Country A’s RfD, classification, or ban. Interpretation, extra endpoints, and risk management stay national. A non-guideline investigative study, or a guideline study run outside GLP when GLP was required, is not MAD currency.

The ABT study list flags health-effects TGs such as 408, 413, 414, 420, 432, 437, 438, 442, 443, 471, 492, and 497, and biotic-system TGs 202, 203, and 207. Those numbers are methods. MAD is the acceptance rule that makes a Dutch GLP TG 408 usable in a U.S. industrial-chemical file — still subject to EPA’s own decision framework.

ICH safety guidelines (pharmaceuticals)

ICH (International Council for Harmonisation) Safety (S), selected Quality (Q3 impurities), and Multidisciplinary (M3, M4, M7) documents are the ABT study-list pharmaceutical set. They coordinate timing and scope among the US, EU, and Japan so sponsors do not run three incompatible nonclinical programs. They do not govern a TSCA solvent or a Redbook flavoring.

High-yield S/M/Q map for this chapter (detail of S5, S1, S7 lives in Chapter 6):

  • M3(R2) — what to have when, relative to clinical phase and marketing (section 16.1).
  • S1 — carcinogenicity need and design; S2 — genotoxicity battery; S3A/S3B — toxicokinetics and repeat-dose TK; S4 — chronic duration (historically 6-month rodent / 9-month non-rodent as the chronic pair).
  • S5 — reproductive and developmental; S6 — biotechnology-derived pharmaceuticals (duration logic differs from small-molecule M3 tables).
  • S7A/S7B — safety pharmacology / hERG; S8 — immunotoxicology; S9 — anticancer pharmaceuticals (abbreviated package); S10 — photosafety; S11 — pediatric support.
  • M7 — mutagenic impurities (1.5 µg/day lifetime TTC for most; Chapter 13).
  • Q3A/Q3B — drug-substance/product impurities; Q3C — residual solvents (PDE); Q3D — elemental impurities.

VICH is the veterinary analogue; do not grade a livestock drug on human M3 tables without looking at VICH.

EU REACH: registration bands, SVHC, authorization, restriction

REACH is Regulation (EC) No 1907/2006 — Registration, Evaluation, Authorisation and Restriction of Chemicals — run by the European Chemicals Agency (ECHA) and Member States. The slogan is no data, no market. A manufacturer or importer generally registers a substance at ≥1 tonne per year per legal entity (with listed exemptions). Non-EU manufacturers appoint an EU Only Representative. Polymers are largely exempt as polymers, but monomers may still need registration.

Tonnage bands and Annexes (information requirements escalate; waivers and read-across exist but must be justified):

Tonnage (t/year)Core annexTypical human-health incrementChemical Safety Report
1–10VIIPhyschem; in vitro irritation/sensitization; Ames; acute oral; basic eco (Daphnia, algae, ready biodegradation)Not the default CSR trigger
10–100VII + VIIIIn vitro mammalian genotox; 28-day repeated dose (OECD 407) or combined 421/422; screening reproduction; more eco/fateCSR required at ≥10 t/y (Article 14)
100–1,000+ IX90-day (OECD 408); prenatal developmental (OECD 414); EOGRTS/two-generation (OECD 443) as specified; bioaccumulation and higher-tier ecoCSR
≥1,000+ XChronic/carcinogenicity and further long-term eco when triggered; remaining higher-tier studiesCSR

Evaluation has two limbs: dossier evaluation (ECHA completeness and compliance) and substance evaluation (Member State, Community rolling action plan).

SVHC → authorization. Substances of Very High Concern meet Article 57 criteria: CMR category 1A/1B, PBT, vPvB, or equivalent level of concern (endocrine disruption is the usual example). Listing on the Candidate List triggers article communication at >0.1% w/w (Article 33) and notification duties. Moving a substance to Annex XIV imposes authorisation: after the sunset date, placing on the market or use requires a granted authorisation based on adequate control of threshold risks or a socio-economic analysis when residual risk remains. Restriction (Annex XVII) can limit or ban manufacture, placing on the market, or use without going through Annex XIV — a different tool for unacceptable risk.

DNEL / DMEL recap (Chapter 13). A derived no-effect level (DNEL) is a threshold value: dose descriptor (usually NOAEL/BMDL) divided by assessment factors, derived per route, duration, and population (worker versus general). Risk characterisation ratio RCR = exposure / DNEL (Chapter 14). A derived minimal-effect level (DMEL) is a non-threshold, risk-based construct for effects such as mutagenic carcinogenicity. ECHA R.8 discusses illustrative extra-risk figures on the order of 10^-5 for workers and 10^-6 for the general population; those figures are guidance illustrations, not an EU statute and not a Superfund NCP copy-paste. Do not call a DMEL “a DNEL that is 10-fold tighter.”

Green chemistry, substitution, and designations (IV.7 A, IV.7 C)

IV.7 A names green chemistry, sustainable substitution, regrettable substitution, and benign by design. Anastas and Warner’s 12 principles (1998) are the usual list; the DABT-relevant cluster is prevention (do not make waste), atom economy, less hazardous chemical synthesis, designing safer chemicals, safer solvents and auxiliaries, design for degradation, and inherently safer chemistry for accident prevention. Benign by design means building low intrinsic hazard into structure and life-cycle before relying on PPE and permits. Sustainable substitution replaces a chemical of concern with a functionally adequate, lower-hazard alternative after checking the alternative’s own tox, fate, and use pattern. Regrettable substitution is the failure mode: 1-bromopropane (n-propyl bromide) as a “drop-in” for methylene chloride or ozone-depleting solvents later became a TSCA high-priority substance with unreasonable risk findings; BPS as a BPA thermal-paper replacement kept endocrine activity in play; some early PFAS replacements kept persistence. A Safer Choice-looking trade name is not a hazard assessment.

IV.7 C designations answer different questions:

  • EPA Safer Choice (formerly Design for the Environment) is a voluntary product certification. Ingredients are screened against Safer Choice criteria; many appear on the Safer Chemical Ingredients List (SCIL) by functional class. Third-party profilers (EPA names NSF International and ToxServices) gather hazard and fate data. Using a SCIL-listed ingredient does not by itself authorize the Safer Choice label. A yellow triangle on SCIL flags residual hazard-profile issues.
  • USDA BioPreferred / certified biobased product reports biobased carbon content (often by ASTM D6866). Biobased is not non-toxic. A 95% biobased solvent can still be an acute toxicant, a VOC, or a sensitizer.
  • NSF International is a third-party standards and certification body. NSF/ANSI 60 addresses drinking-water treatment chemicals; NSF/ANSI 61 addresses drinking-water system components (what can leach into water). NSF also appears as a Safer Choice profiler. NSF is not EPA and not a statute.

Product stewardship (IV.9)

IV.9 A asks for tiered approaches to gathering toxicological information. Match data depth to exposure, use, and hazard flags — the same logic as Redbook concern levels and REACH annexes, not a 2-year bioassay on every fragrance component. A practical ladder: identity and use mappingTTC / read-across / in silico / in vitro at very low exposure → OECD 407/408-class repeated dose when volume or consumer contact rises → developmental, reproductive, and chronic/onco when triggers fire (CMR suspicion, high tonnage, food or inhalation exposure). Stop and refine exposure before defaulting to another in-life study (3Rs plus MAD).

IV.9 B is unintended use and misuse, environmental release, and mitigation through labeling and manufacturing controls. Stewardship scenarios DABT can write without warning: a child drinks a brightly colored cleaner (FHSA label, child-resistant closure, bitterant); a consumer mixes bleach and ammonia (chloramine); a pesticide is applied off-label in a kitchen (FIFRA); a solvent drum is emptied to a storm drain (CWA); a device pouch leaches an untested adhesive (IV.7 B). Labels (HCS SDS, FHSA, FIFRA, CLP in the EU) are controls, not decorations. Manufacturing controls include impurity specifications (ICH Q3), residual-monomer limits, closed handling, change control when a supplier swaps a plasticizer, and extractables programs when the container changes. Stewardship is the system that keeps IV.7–12 from being a stack of unread PDFs.

Scenario

A U.S. company will import 120 tonnes/year of a new plasticizer into the EU, formulate a consumer cleaner, and sell a pediatric oral solution in a multilayer pouch. REACH Annex IX data (90-day, developmental) plus a CSR with DNELs are in play at 100–1,000 t. If the plasticizer is later identified as a reproductive toxicant 1B, Candidate List / Annex XIV logic appears. The cleaner’s “plant-based” claim is a BioPreferred content claim unless the formula also clears Safer Choice hazard criteria — those are not the same badge. The pouch needs a leachables qualification, not a MAD TG 408 pasted onto an SDS. OECD GLP TG studies run in an adherent country should be accepted as data by ECHA; ECHA still decides classification and whether a restriction is needed.

Traps

  • Treating MAD as automatic harmonized classification or a ban/no-ban decision.
  • Using ICH M3(R2) as a REACH annex, or REACH Annex X as an IND.
  • Calling DNEL and DMEL the same number with a 10-fold fudge.
  • Equating biobased content with Safer Choice or with NSF/ANSI 61.
  • Substituting to 1-bromopropane (or another unvetted analogue) and calling it green chemistry.
  • Gathering a full chronic package on a 2-tonne intermediate with no exposure, or skipping labeling because the SDS “looks fine.”
Loading diagram...
REACH evaluation tools feeding product stewardship
Test Your Knowledge

A 90-day oral rat study was conducted in Japan to OECD Test Guideline 408 under OECD GLP. A second country is assessing the same industrial chemical. What does OECD Mutual Acceptance of Data require, and how do ICH safety guidelines differ in role?

A
B
C
D
Test Your Knowledge

Which REACH description correctly pairs tonnage information, SVHC tools, and DNEL versus DMEL?

A
B
C
D
Test Your Knowledge

A formulator replaces methylene chloride with 1-bromopropane, prints “plant-based” on a cleaner because 80% of the carbon is biobased, and skips consumer-misuse labeling because an OECD 408 study was clean. Which stewardship judgment is most accurate for IV.7 A/C and IV.9?

A
B
C
D