Pfizer's $10B Innovent bet meets the China-data reckoning

Pfizer paid Innovent $650 million upfront for ex-China rights to twelve oncology candidates last week, in a deal Fierce Biotech values at up to $10 billion and BioPharma Dive frames as one of the largest multinational–China alliances on record. It lands inside a week in which a House bill would extend the COINS Act’s outbound-investment regime to biotech, and in which the first China-only Phase 3 readouts — Akeso’s ivonescimab and Merck-Kelun’s sac-TMT — were taking plenary slots at ASCO 2026. The biometrics question that has been parked for two years is now the gating risk on tens of billions of pipeline value.

The deals have outrun the data policy

The Pfizer-Innovent structure is the now-familiar one: Western sponsor takes ex-China rights, Chinese originator runs the early trials at speed, and the global dossier inherits whatever the NMPA package looks like. BMS did a version of this last year. The reason it keeps happening is the one SR One’s Simeon George described in our archive — de-risked, clinically validated assets at a fraction of US discovery cost. The reason it is now a problem is that FDA Oncology Center of Excellence leadership has publicly flagged concern about the volume of trials conducted exclusively in China, arguing that genuine multiregional participation is what builds reviewer confidence in those sites. That posture is being tested in real time. HARMONI-6, the ivonescimab squamous NSCLC trial that posted a ~34% reduction in risk of death and a simultaneous Lancet publication, was conducted entirely in China; Merck-Kelun’s sac-TMT, which cut progression risk 65% versus Keytruda monotherapy in first-line PD-L1+ NSCLC, was also a single-country Phase 3. These are the test cases that will set the bridging precedent for the twelve Pfizer-Innovent assets behind them.

What changes in the SAP and the SDTM

For biometrics teams the work is concrete and largely under-scoped. Bridging strategy under ICH E17 is not optional any more for these programs — it is the design question. Expect FDA to push for either a confirmatory MRCT arm or a pre-specified ethnic-sensitivity analysis with a credible extrapolation argument; “the China trial was positive” will not carry the day on its own, particularly where standard-of-care control arms diverge from US practice. Estimand specification under E9(R1) is where the differences in concomitant medication, subsequent therapy and censoring conventions surface; teams inheriting an NMPA dossier should assume the estimand will need to be rebuilt, not translated.

Data-standards work is the other half. NMPA submissions are increasingly CDISC-aligned but the mapping from Chinese CRO outputs into FDA-acceptable SDTM and ADaM — controlled terminology, AE coding versions, lab unit harmonisation, Define-XML lineage — is rarely clean on the first pass. Audit-trail completeness and ALCOA+ posture at Chinese sites are what FDA reviewers will probe; sponsors acquiring assets mid-stream should budget for a data-integrity diligence pass that goes beyond the standard transition checklist.

The policy overlay

Then the legal layer. The Moolenaar-Dingell House bill would add biotechnology to the COINS Act, and Treasury is being lobbied to expand the codified framework — currently semis, quantum and AI — to cover precisely the licensing structures Pfizer just used. Nothing has passed. But sponsors are reportedly writing termination and data-portability clauses into new China deals on the assumption something will, and biometrics leads should expect contract language requiring TMF and dataset repatriability on short notice. That has operational consequences: where are the SAS grids, who holds the Define-XML, which CRO contracts have China-resident-data clauses that conflict.

None of this stops the deals. It does mean the unglamorous bridging-and-standards work is now on the critical path for a meaningful share of the oncology pipeline. The ASCO plenary outcomes will be read by reviewers as much as by investors.

Protocol read: The China-deal wave has moved the bridging-study and CDISC-mapping question from a regulatory-affairs hypothetical to a portfolio-level execution risk; teams whose sponsors are buying these assets need a bridging playbook before the next one closes, not after.

What to do now:

• Pre-specify ICH E17 bridging and E9(R1) estimand strategy for any in-licensed China asset before the pre-IND, not at the dossier stage.
• Run a data-integrity and SDTM-conformance diligence pass on inherited Chinese CRO datasets as a standard transaction workstream — treat audit-trail and ALCOA+ posture as gating, not cosmetic.
• Ask legal to surface China-deal contracts for data-portability and TMF-repatriation clauses now, in case COINS Act expansion forces unwind scenarios.