ICH M12: The Numeric Cutoff That Decides Whether You Need a Clinical DDI Study

Until May 2024, a sponsor running a global drug-interaction program was really running three of them. FDA, EMA, and Japan's PMDA each published their own drug interaction study guidance, with different in vitro cutoffs, different modeling expectations, and different thresholds for when an in vitro signal became a clinical obligation. ICH M12, adopted at Step 4 on May 21, 2024, replaced that with one set of numbers. The harmonization is the headline. The numbers are the part that changes what you actually have to do.

What actually changed

ICH M12 harmonizes the design, execution, and interpretation of enzyme- and transporter-mediated pharmacokinetic drug interaction studies across ICH regions. That is a real change in mechanics, not just a paperwork alignment: a sponsor building a CMC and regulatory strategy around DDI risk now designs one in vitro package against one decision framework, instead of running FDA's cutoffs, checking whether EMA's differ, and padding the plan to cover both. The cutoffs themselves also moved in places — a study that a pre-2024 FDA-only screen would have flagged for a clinical follow-up can land differently under ICH M12's basic model, in either direction.

The cutoff that does the actual work

For reversible CYP-mediated inhibition, ICH M12's basic model compares the unbound maximum plasma concentration at steady state (Cmax,u) against the unbound inhibition constant (Ki,u). A ratio below 0.02 lets a sponsor exclude a clinically relevant interaction on in vitro data alone — no clinical study, no PBPK model, no further work on that enzyme. That single number is the difference between a DDI program that closes in a DMPK lab and one that adds a clinical arm to the nonclinical development package.

• Below 0.02. Risk of a clinically relevant CYP interaction can be excluded on the basic in vitro model; no further evaluation is required for that pathway.
• 0.02 to 0.1. ICH M12 recommends considering PBPK (or an adequately verified mechanistic static) model rather than moving straight to a clinical DDI study — the intermediate zone is where a modeling investment can substitute for a trial.
• At or above the relevant threshold. A clinical DDI study, or a PBPK prediction verified against the compound's own clinical data, is generally warranted; ICH M12 sets a tighter default (0.1) for parenteral products and metabolites than the oral-CYP baseline.
• Transporters run on separate cutoffs. P-gp and BCRP substrates and inhibitors in orally administered drugs use an R-value cutoff of 10; OATP1B1/1B3, OAT1/3, and OCT2 use a tighter cutoff keyed to whether hepatic or renal elimination is a significant clearance pathway.

A sponsor who built its exclusion argument on FDA's 2020 cutoffs is not automatically covered by ICH M12's numbers — the basic model moved, and a study designed to the old framework needs to be re-checked against the new one before it goes in a submission. Why re-running the basic model matters now

Where this bites a global program

The transporter cutoffs are where most teams under-scope the work. ICH M12 doesn't just ask whether your compound is a P-gp or OATP substrate in vitro — it asks whether that transporter sits on a clearance pathway that matters clinically: intestinal absorption limitation for P-gp/BCRP, or a hepatic elimination fraction material enough to make an OATP1B1/1B3 interaction clinically relevant. Two compounds with identical in vitro transporter data can land on opposite sides of the clinical-study line depending on their elimination pharmacology, which is a bioanalytical cross-validation and PK-modeling question as much as a regulatory one. Teams that treat the transporter panel as a checkbox — run the assay, file the result — miss the step where the elimination-pathway analysis decides whether that result triggers anything.

None of this is exotic modeling work — it is a bounded exercise in re-running a decision tree against the current numbers before a study protocol locks in. Programs that carry forward a pre-2024 exclusion argument without re-checking it against ICH M12's basic model are the ones that discover the gap at an NDA filing review, not before it. Building the DDI strategy around the current cutoffs, and treating the transporter elimination-pathway analysis as a real deliverable rather than a formality, is what keeps that discovery from happening at all.