ICH Just Wrote the First Global Rulebook for Leachables. Your US Comments Closed January 30.

Three numbers to anchor on

Jan 30, 2026 is the US comment deadline for the first global leachables rulebook (FDA Docket FDA-2025-D-4678). It is closed. The framework has moved into the post-consultation phase.

1.5 µg/day is the ICH M7 threshold of toxicological concern (TTC) that Q3E adopts for mutagenic leachables at lifetime (greater than 10 years) exposure. The same number anchors the Class 1 high-concern compounds.

12 µg/day is the Q3E Qualification Threshold for non-mutagenic systemic toxicity by the parenteral route at chronic exposure. The oral equivalent is 48 µg/day. That roughly four-fold gap is the route correction in a single comparison, and it is one of the most consequential numbers in the document.

Read together, the three numbers say one thing. Q3E sets your leachable safety floor as the lower of a mutagenicity number (the M7 TTC) and a systemic-toxicity number (the Qualification Threshold), and the systemic number depends on how the drug gets into the patient.

The 20-year patchwork it replaces

Before Q3E, extractables-and-leachables work ran on a stack of partial references that you stitched together by hand.

USP chapters <1663> and <1664> covered the assessment and qualification framework for the final container-closure system. USP <665> and <1665> were drafted for the polymeric components used upstream in manufacturing. The PQRI Parenteral and Ophthalmic Drug Products (PODP) working group published the operational thresholds many toxicologists actually derived against, including the 5 µg/day floor for sensitisers and irritants. ICH Q3A through Q3D governed impurities, and ICH M7 governed mutagenic impurities. Not one of them was written for the leachable that migrates out of a rubber stopper into a biologic over 24 months of storage.

So you assembled the assessment yourself: a threshold from PQRI, a route correction from Q3C, mutagenicity logic from M7, a read-across approach from whatever your toxicologist trusted. Then you hoped the reviewer in your filing region agreed with the assembly. File the same dossier in another region and the same molecule might draw a different question, because the expectation lived in guidance interpretation rather than in one cited document.

That fragmentation is what Q3E targets. Per the ICH Step 2 presentation, its stated aims include minimizing the uncertainty in meeting global regulatory expectations and encouraging a holistic, science- and risk-based control strategy built on quality-by-design principles. It is meant to complement and stay consistent with the existing ICH impurities guidelines and to align the diverse regional pharmacopoeias. The same math, defensible everywhere.

What is genuinely new

Q3E is more than a tidy-up of prior guidance for three reasons.

• One risk-based framework, with ICH Q9 at the core: The risk model is explicitly multifactorial. On the pharmaceutical-quality side it weighs dosage form, storage temperature, the quantity of extractables, leaching propensity, the surface-area-to-volume (SA:V) ratio, and contact duration. On the safety side it weighs route of administration, treatment duration, leachable dose, and patient population. Risk climbs as you move from oral toward injectable and inhalation, from acute toward chronic, and from healthy adults toward vulnerable groups such as neonates, children, the elderly, and patients already burdened by disease.
• ICH M7 mutagenic-impurity logic, folded straight into the leachables workflow: Most teams underestimate this part. For any leachable that is an M7 Class 1, 2, or 3 mutagenic impurity, the safety assessment now routes it through M7 (reduce it below its Acceptable Intake or the TTC). Non-mutagenic species route through the Qualification Threshold pathway instead. Q3E is the first harmonized E&L guideline to write that mutagenic-versus-non-mutagenic fork into a single safety-assessment flow. Where daily exposure exceeds 1 mg/day, genotoxicity studies should be considered per Q3A and Q3B.
• One harmonized threshold table that varies by route and duration: You no longer reconstruct the staircase from M7 plus a route adjustment you derived yourself. The numbers are in the guideline.

Worth stating plainly: Q3E applies to new drug products, including cell and gene therapy products, and to drug-device combination products requiring marketing authorization. It is not applied retrospectively to already-approved products, nor to excipients, nor to contamination or adulteration, nor to products still in clinical development. It targets new filings. If you have a commercial product, this is about your next submission, not a forced re-qualification of the last one.

The qualification math, decoded

Q3E organizes the safety question into three layers. Get these straight and the rest of the document reads cleanly.

Layer 1: the AET (Analytical Evaluation Threshold)

The AET is a reporting trigger, not a control limit. It is the concentration above which an extractable or leachable must be identified (its structure elucidated), quantitated, and reported to a toxicologist. It is study-specific, derived from the maximum daily dose and the applicable Safety Concern Threshold, and semi-quantitative methods carry an uncertainty factor so you do not underestimate concentrations. The algebra carries the lesson: because the AET sits downstream of dose, a high-dose or high-volume product drives a lower reporting threshold, so more species cross the line and demand qualification.

Layer 2: the SCT (Safety Concern Threshold)

This is the safety floor. Q3E sets it as the lower of two numbers: the ICH M7 TTC (for mutagenicity) and the Q3E Qualification Threshold (for non-mutagenic systemic toxicity). It is product-specific, because it depends on the toxicological endpoint, the route, and the duration. Class 1 high-concern leachables are the exception; they are controlled below a Permitted Daily Exposure (PDE).

Layer 3: compound-specific limits (Acceptable Intake or PDE)

When a leachable exceeds the Qualification Threshold or is Class 1, you derive a compound-specific Acceptable Intake or PDE from toxicology data, or from read-across when direct data are missing.

The harmonized threshold table

These values are read directly off the ICH Q3E Step 2 presentation slide for systemic and local toxicity thresholds. Treat them as authoritative.

Systemic toxicity thresholds (µg/day)

* The Qualification Threshold values for inhalation and dermal/transdermal are provisionally set equal to the parenteral values and flagged for update in a later revision. ICH scheduled a March 2026 Threshold Project meeting to finalize them.

Local toxicity thresholds

The TTC columns are the ICH M7 less-than-lifetime staircase: 1.5 µg/day for lifetime exposure, 10 µg/day for greater than 1 to 10 years, 20 µg/day for greater than 1 month to 1 year, and 120 µg/day for 1 month or less. Shorter exposure buys a higher allowable intake. That bracket structure has governed mutagenic impurities under M7 for years; Q3E makes it the explicit default for leachables too.

Route correction, in one comparison

Toxicology points of departure are often drawn from oral studies. The leachable in a pre-filled syringe does not get swallowed, though. It gets injected, with near-complete bioavailability, so an oral point of departure overstates the safe injected dose. Q3E and its safety workflow require route-appropriate thresholds, which is why the parenteral chronic Qualification Threshold (12 µg/day) sits at roughly a quarter of the oral value (48 µg/day). An assessment that quietly carried an oral threshold into a parenteral product was running four times too permissive. The new table makes that error harder to commit and easier for a reviewer to catch.

Potency classes for leachables

Q3E Appendix 4 sorts leachables into three potency classes, with named example compounds.

• Class 1 (high concern): M7 Class 1 with an Acceptable Intake below 1.5 µg/day and a cohort-of-concern member. These should be avoided or given a compound-specific acceptable level. Named examples include benzo(a)pyrene (from carbon black) and bisphenol A (from polycarbonate and epoxy resin).
• Class 2 (default): The standard case, where the mutagenicity TTC and the systemic Qualification Threshold are considered sufficiently protective. This covers M7 Class 1 with an Acceptable Intake at or above 1.5 µg/day, M7 Class 2 and 3 impurities, and non-mutagenic leachables that are not Class 1 or 3.
• Class 3 (low potency): A chronic parenteral PDE of 1 mg/day or higher, qualified up to 1 mg/day regardless of route or duration (or to the compound-specific PDE). Named examples include BHT, erucamide, 4-tert-amylphenol, C8 to C22 fatty acids, and a C21H40 rubber oligomer.

Where it bites hardest

Q3E touches every contact material, but the pressure lands unevenly.

The high-risk routes carry the tightest thresholds: parenterals, ophthalmics, inhalation products, and the cell and gene therapy products that Q3E explicitly brings into scope. These are the products where the parenteral Qualification Threshold (12 µg/day) and the local thresholds (20 ppm ophthalmic, 50 ppm subcutaneous and intradermal) apply, and where a single under-derived threshold can stall a filing.

The biologics shift compounds the problem. Single-use systems in biomanufacturing (the bags, tubing, connectors, and filters that touch the drug) are the structural reason E&L exposure keeps growing. The concern is old. In a 2018 BioPlan Associates survey, 73.3% of bioprocessing professionals agreed that extractables and leachables are a major concern that may limit single-use adoption (per BioPharm International). That figure is dated, a long-standing industry signal rather than a current measurement, but the trajectory it describes has only steepened.

The timing converges too. USP <665> and <1665>, covering polymeric components and systems used in manufacturing, take effect May 1, 2026. They do not apply to the final container-closure system, which stays under USP <1663> and <1664>. The manufacturing-side standard and the container-closure-side standard now bracket the same drug product from both ends, with Q3E sitting above both as the harmonized safety framework. Your upstream single-use train and your final vial-and-stopper are both in scope, under different pharmacopoeial chapters, answering to one risk logic.

Why "no data" is no longer a defense

For years, the honest answer to a leachable with thin toxicology was a paragraph of hand-waving. Q3E removes that option, and it tells you what to do instead.

Appendix 5 explicitly encourages, when direct data are missing, scientific justification through in silico analysis and read-across to structurally similar surrogate compounds. It states that such alternative approaches should be considered before commissioning new in vitro or in vivo studies. "We have no data" is not a stopping point anymore. It is the trigger for a documented, structure-based justification, and the expectation holds in every region that adopts the guideline.

That raises the bar in a specific way. The assessment a reviewer now expects is route-corrected, endpoint-specific, and read-across-justified, with the arithmetic shown. The target is written down. The catch is that improvising it across dozens of extractables per study, each needing identity, a worst-case exposure, a route correction, a threshold lookup, and a margin of safety, is exactly the repetitive, citation-heavy derivation work that drains a CMC team's calendar.

How DeepC maps to the Q3E workflow

The structure of the guideline and the structure of the product line up almost exactly here. Q3E defines two terms cleanly. Extractables come off a manufacturing, packaging, or device component under specified forcing laboratory conditions, and are potential leachables. Leachables actually migrate into the drug product under established manufacturing and labeled storage conditions. DeepC ships two regulatory-safety specialists that mirror that split one to one.

Both agents produce submission-structured regulatory PDFs: a stamped cover page, a Klimisch-scored reliability annex on every primary toxicity citation, numbered references, and a Qualified Person sign-off block left unsigned for your QP. DeepC qualifies substances and derives thresholds. It does not run the wet-lab extraction or migration study, and it never invents a toxicological value.

The Extractables Agent: the worst-case stress side

The Extractables Agent qualifies substances identified in container-closure stress-extraction studies (USP <1663>) for parenteral, ophthalmic, inhalation, and other risk-tier products. It produces ICH Q3E and USP <1664> reports anchored to PQRI PODP recommendations.

That last row matters for timing. You do not need finished analytical data to start. The agent can run a threshold-only derivation now and slot the concentrations in when the lab returns them.

The Leachables Agent: the in-use migration side

The Leachables Agent derives per-substance safety thresholds for chemicals migrating into the drug product under normal in-use conditions, distinct from the worst-case stress conditions the Extractables Agent handles. It produces ICH Q3E and PQRI PODP risk assessments with Klimisch-scored toxicological dossiers.

Notice the fourth row. Q3E's SCT logic, the lower of the M7 TTC and the Qualification Threshold, is the same margin-of-safety calculation the Leachables Agent already runs against the lower of the substance-specific threshold and the PQRI floor. The guideline formalized a comparison the agent was built to make.

The clean read

• Two definitions, two agents: Extractables under forcing lab conditions, leachables under real storage and use: the two Q3E definitions map to the two DeepC agents, worst-case side and in-use side.
• The safety floor is the margin-of-safety logic: Q3E's “lower of two thresholds” safety floor is the Leachables Agent's margin-of-safety logic.
• Read-across, operationalized: Q3E Appendix 5's read-across encouragement is operationalized by both agents: named class analogs, similarity criteria (same functional groups, log P within one log unit, molecular weight within 30%), a PAINS and structural-alert parity check on both target and analog, and an F6 = 10 uncertainty factor by default.

For the cases that spill past these two agents, the wider DeepC safety suite picks them up: a Safety Assessment dispatcher that routes the question, a Nitrosamines Agent for cohort-of-concern leachables, and a data stack that includes ECHA IUCLID REACH study records, EPA CompTox, all 153 ICH guidelines as section-addressable text, and eCFR 21 CFR 211.94 packaging requirements.

The clock you are running against

Step 2 is a draft, not Step 4. The anticipated Step 4 adoption, after which Q3E implements in each regional regulatory system, is around June 2027, roughly a year out. The EU and US comment windows are closed, the Threshold Project finalization meeting is behind us, and the framework is stable enough to design to even though it is not yet legally binding.

Treat that gap as the opportunity. The control strategy you can document today, route-corrected, endpoint-specific, read-across-justified, in the structure Q3E prescribes, is the one that survives the transition to binding guidance. File now and your qualifications already speak the reviewer's language when adoption lands. Wait, and you will reverse-engineer compliant assessments under deadline pressure, against the same threshold table, with less time.

The reference, in one paragraph

The first global rulebook for extractables and leachables exists. It puts one threshold table, one ICH Q9 risk model, and the M7 mutagenicity fork into a single document that reviewers in every ICH region will read the same way. The parenteral systemic floor is 12 µg/day, the mutagenic lifetime TTC is 1.5 µg/day, and "no data" now means "build a read-across justification," not "skip the assessment."

You have until roughly June 2027 before it is mandatory. Use the window to qualify your container leachables to the Q3E table now, in submission format, while it is still a choice rather than a scramble. DeepC's Extractables and Leachables Agents derive the route-corrected, read-across-justified limits the guideline asks for, structured for submission, with the QP sign-off left for your Qualified Person.