FDA vs NMPA vs EMA: How they differ on mAb biosimilar guidance in 2026

The three agencies that matter for monoclonal antibody biosimilars in 2026 — the US FDA, China NMPA, and EU EMA — have spent the past five years converging on the science. They have not converged on the paperwork. If your biosimilar program targets all three markets, the cost is no longer in the analytics. It is in the regulatory project management.

This post compares the 2026 picture across eight dimensions that show up in every mAb biosimilar review meeting. It is written for the regulatory affairs lead who has the FDA guidance bookmarked but wonders what the NMPA actually wants this quarter.

1. The reference product question

The FDA still requires that the reference product be licensed in the US. The EMA accepts non-EU comparators only if they are sourced from a regulatory environment with comparable standards (typically the US, Switzerland, Japan, Australia, Canada). The NMPA's 2024 update accepts US- or EU-licensed reference products, but with strict bridging studies.

Practical implication: if you are building a single comparability program for all three, source the reference product from the US — it satisfies all three with the cleanest paperwork trail.

2. Analytical similarity (Tier 1 / 2 / 3)

All three agencies have converged on a tiered analytical similarity framework. The differences are in how strict the tiering is and which attributes go into which tier.

• FDA: Tier 1 (clinically relevant CQAs) → equivalence testing with formal statistical bounds. Tier 2 → quality range. Tier 3 → side-by-side visual comparison.
• EMA: Similar tiered approach but allows more sponsor justification for tier assignment.
• NMPA: Adopted the FDA tiered model in 2023. In 2026, Tier 1 CQA lists have been substantially harmonized with FDA.

Where they still diverge: the cutoff for what constitutes a Tier 1 vs Tier 2 attribute remains a judgment call. The EMA tends to accept more attributes into Tier 2; the NMPA tends to follow the FDA position closely.

3. Clinical comparability — extrapolation

Extrapolation across indications has become the single biggest accelerator for mAb biosimilar programs.

• FDA & EMA: Both accept extrapolation across indications with a strong scientific justification (mechanism of action, PK/PD, immunogenicity).
• NMPA: Accepted extrapolation in principle as of 2023 but reviews case-by-case. In 2025–2026 reviews, we have seen extrapolation accepted for oncology mAbs but still contested for autoimmune indications.

Practical implication: plan PK/PD studies that support extrapolation upfront. NMPA reviewers want to see this discussed proactively, not buried in a justification section.

4. Immunogenicity

The three agencies converge on immunogenicity testing approach — three-tier ADA assay strategy, in-study sampling, post-marketing surveillance. They diverge on how much pre-marketing immunogenicity data is enough.

• FDA: Generally accepts pivotal comparability data plus a comparative PK/PD study.
• EMA: Often wants a confirmatory clinical efficacy study even for highly characterized programs.
• NMPA: Pre-2024 typically wanted a Phase III efficacy comparison. Since 2024 has accepted PK/PD-only programs for some highly characterized molecules.

5. Naming and prescriber information

The naming convention is one of the few areas with active divergence in 2026.

• FDA: Four-letter suffix attached to the non-proprietary name (e.g., adalimumab-atto).
• EMA: INN with no suffix; brand name carries the distinguisher.
• NMPA: Following the FDA suffix approach since 2023.

This is rarely a CMC problem but matters for labeling, supply chain, and pharmacovigilance.

6. Manufacturing changes post-approval

Biosimilars are particularly exposed to post-approval CMC changes because the reference product itself may have drifted.

• FDA: Comparability protocols are the standard tool. Major changes require BLA supplement.
• EMA: Similar variation classification system (Type IA / IB / II / Extension).
• NMPA: Uses a parallel variation system; in 2026 has accepted more comparability protocol filings rather than full re-comparability studies.

7. Interchangeability

This is a US-only concept. Neither the EMA nor the NMPA has an interchangeability designation. The FDA's interchangeable biosimilar designation, however, often drives commercial strategy even for programs that target all three markets.

8. Where they all converge: PIPL / GDPR / cross-region clinical data

If your biosimilar program runs Phase III in China and the US, you have a data residency problem the agencies do not solve for you. Patient data collected in China cannot leave China under PIPL without specific transfer mechanisms. The FDA accepts the data; the NMPA wants to keep it.

This is a CRO / sponsor architecture problem, not a regulatory one — but if you have not designed for it, your filing strategy will hit it.

What this means for your filing strategy

Three concrete planning recommendations:

1. Build one analytical similarity package that satisfies the most stringent of the three Tier-1 lists (typically the FDA's). The EMA and NMPA will accept a superset. The reverse is not true.
2. Plan extrapolation justification proactively — especially for the NMPA. Treat it as a primary regulatory deliverable, not an afterthought.
3. Decide your immunogenicity strategy by market mix. A US-only program can often skip a confirmatory efficacy study. A China-plus-EU program usually cannot.

The convergence trend is real. The divergence on paperwork is durable. Plan for both.

Want a single answer to "what does {FDA / NMPA / EMA / ICH} say about {topic}?" — InPrism compares the four agencies side-by-side with paragraph-level citations. Try one.