China on the Move: China Reprices Clinical Data (Part Two) — The Rebuilt GCP and What Must Change Before Sept. 1

On June 8, 2026, four central agencies jointly published a revamped Good Clinical Practice (GCP) standard, effective Sept. 1, 2026. It is materially shorter than the standard it replaces, and that compression is the point: China’s GCP now incorporates International Council for Harmonisation (ICH) E6(R3) by reference rather than restating it, and it reallocates responsibility among sponsors, service providers, investigators, and sites in ways that may require contract amendments before the effective date.

Part 1 addressed what a China dataset is now worth. This GT Advisory addresses what it takes to generate one, and the work required before Sept. 1.

I. Overview of China’s 2026 GCP Revision: Scope and Applicability

The revised standard was issued under Announcement No. 50 of 2026, dated May 21 and published June 8, 2026, takes effect Sept. 1 and repeals the 2020 GCP. The 2020 version had two signatories; the 2026 revision is issued jointly by four — the NMPA, the National Health Commission, the National Administration of Traditional Chinese Medicine, and the National Disease Control and Prevention Administration. That brings vaccine and traditional medicine trials inside one framework and embeds the health authority’s jurisdiction over hospital ethics committees. Enforcement expectations may be correspondingly aligned.

The reach extends beyond Chinese sites. Article 10 of the revised Implementing Regulations provides that drug development conducted outside China for the purpose of China registration must comply with the Drug Administration Law, the Regulations, and relevant standards and specifications — and the GCP is one of those specifications. The implication, which several China practitioners have drawn, is that a multiregional trial run entirely offshore with China filing intent may meet China’s requirements. This has not been tested in inspection practice, but it may warrant a deliberate decision at the protocol design stage rather than an assumption. Stakeholders may wish to identify which China-specific requirements the whole program would meet, document the decision, and be prepared to explain it. The requirements most likely to differ from E6(R3) practice elsewhere relate to consent content, the prohibition on testing biological samples outside the approved protocol, and rules governing direct access and investigational-product sample retention.

II. How China’s 2026 GCP Incorporates ICH E6(R3) and Key Compliance Dates

The 2026 GCP comprises six chapters and 54 articles, compared to nine chapters and 83 articles in the 2020 version. Three substantive chapters were deleted outright — Trial Protocol, Investigator’s Brochure, and Essential Documents Management — while the terminology chapter was folded into the supplementary provisions and a Data Governance chapter was added. Article 54 refers undefined terms to the Chinese E6(R3) glossary, and the NMPA’s question-and-answer document confirms operational detail was deliberately left out because China implements E6(R3) in full.

Compliance in China therefore includes the 2026 GCP and the complete text of ICH E6(R3), read together. A sponsor whose China procedures are keyed to the GCP alone may have gaps in exactly the three deleted areas, and a gap analysis run against the GCP in isolation might produce a false clean result there.

The two standards commence five months apart. ICH E6(R3) has applied to trials conducted after March 31, 2026, under NMPA Announcement No. 125 of 2025; the revised GCP takes effect Sept. 1, 2026. Between these dates, trials would operate under E6(R3) and the 2020 GCP — meaning risk-based quality management obligations apply while the deleted chapters remain in force. Sponsors running studies across either boundary may wish to record in the trial master file which standard governed which period, with dated protocol, procedure, monitoring, and quality plan versions. That record is easier to create contemporaneously than to reconstruct on inspection.

III. Sponsor and Investigator Responsibilities Under China’s New GCP: What Has Changed?

• The sponsor is the ultimate responsible party. Article 32 so designates it; Article 36 permits delegation to qualified service providers but requires the sponsor to supervise the provider and any subcontracting and to bear ultimate responsibility. This is non-delegable residual accountability, not joint and several liability alongside the vendor — a well-drafted CRO indemnity may reallocate economic loss but not regulatory exposure.
• Covered vendors are broader than CROs. Because the revised GCP refers broadly to “service providers,” sponsor oversight may need to extend beyond CROs to include central laboratories, data capture and randomization vendors, and imaging core labs. Oversight programs scoped only to CROs may need widening.
• The principal investigator (PI) is ultimately responsible at the site. Article 19 so provides; Article 20 makes the PI’s trial decisions, key confirmations, and formal reports to the ethics committee, the institution, and the sponsor non-delegable in principle. Note it names those three recipients — not regulatory authorities — and that “in principle” leaves room for justified, documented exceptions.



IV. How China’s 2026 GCP Affects Existing Clinical Trial Agreements

The GCP contains no transitional provision preserving earlier contracts, but it also imposes no obligation to re-execute them. It governs the conduct of trials from Sept. 1, 2026, not the date agreements were signed.

Every trial conducted from that date must comply, regardless of when the agreement was executed. A sponsor cannot rely on a legacy clause permitting subcontracting on notice to justify a subcontract entered into after Sept. 1. But the obligation attaches to conduct and may be met by operational control while contracts are remediated. That supports a proportionate program: new agreements carry the revised language now; for active agreements, issue a standing written instruction confirming that subcontracting and off-site delegation require prior written consent from Sept. 1 notwithstanding contrary terms, and establish an internal process to grant it at operational speed; then amend on a rolling basis prioritized by exposure — providers that actually subcontract, sites where off-site delegation occurs, bioequivalence studies, and budgets lacking fair-market-value support — using the next scheduled protocol amendment or budget revision as the vehicle. Conduct concluded before Sept. 1 does not require remediation.

V. Potential Clinical Trial Contract Amendments Checklist for China’s 2026 GCP Compliance

Amendment Checklist — Governing Conduct from Sept. 1, 2026
Subcontracting (Art. 36). Replace notice-based or pre-approved-affiliate subcontracting with a mandatory sponsor prior written consent gate.
Off-site delegation (Art. 20). Add express prior consent where the PI or institution delegates functions outside the institution.
Fair market value (Art. 36). Trial fees must be reasonable and conform to market practice — a fair-market-value standard inside a clinical quality instrument, connecting to the State Administration for Market Regulation commercial bribery guidelines covered in our July 2025 GT Advisory. Support site budgets and investigator payments with documentation.
Pre-commencement execution (Art. 36). Contracts signed with the PI, the institution, and all service providers before trial activities begin, specifying roles, responsibilities, rights, obligations, and conflicts.
Direct access (Art. 49). Secure in the protocol or contract that monitors, auditors, ethics reviewers, and regulatory inspectors may access source records directly — and confirm the matching consent language.
Retention samples (Art. 28). Bioequivalence samples held at least two years post-marketing, and not returned to the sponsor or any interested third party.
Cost and insurance (Arts. 45; Regulations Art. 8). Trial drug free of charge, trial-related testing costs borne by the sponsor, insurance proportionate to risk excluding site fault, and no charges to participants.
Change of sponsor (Art. 50). NMPA approval is required, decided within 20 working days of acceptance under Article 9 of the revised Implementing Regulations — a codified period short enough to sit as a condition precedent if the application is prepared before signing. Allocate responsibility for prosecuting it, and schedule the written notices to the PI, institution, and ethics committee as closing items.




VI. Ethics Review and Participant Protection Requirements Under China’s New GCP

Chapter 2 is the chapter that may require action at site level, rather than sponsor level. Stakeholders may wish to confirm the following four points with China sites before Sept. 1.

• Follow-up review is risk-based with a 12-month ceiling. Article 14 sets the frequency by degree of risk, with the interval not exceeding 12 months. This is not simply an annual requirement — confirm the interval set for each site is documented and defensible on risk grounds.
• Consent forms may not contain waivers or liability exclusions. The committee must confirm no provision requires participants to waive lawful rights or relieves the PI, institution, sponsor, or service providers of their responsibilities. Exculpatory language that is unremarkable in other jurisdictions might not pass.
• Certain events require prompt, prioritized review, including serious adverse events, deviations made to eliminate an urgent hazard, serious or persistent non-compliance, changes increasing risk, and new information adversely affecting safety — with the manner of review proportionate to the urgency required.
• Re-consent triggers are broader than protocol amendment. Under Article 27, new information that may affect willingness to continue, a minor participant becoming able to consent, and a participant regaining civil capacity each require consent to be taken afresh.



VII. Data Governance, Privacy, and Cross-Border Transfer Rules for Clinical Trials in China

Chapter 5 is the new material and the most operationally demanding, allocating data governance across the sponsor, the PI, and the institution across the full lifecycle, with the specific requirements under Articles 51–53 varying by role. Article 51 requires metadata, including audit trails for data from any source, documented error-correction processes, and validated transfer between systems. Article 53 requires procedures for system configuration and use, documented training, lifecycle security with backup and contingency, validation maintained throughout the trial, problem logging and periodic review, and user, permission, and audit-trail management with electronic signatures meeting applicable local laws and regulations — with permissions matched to role, blinding status, and organizational affiliation. Article 52 treats blinding integrity as a data governance obligation, requiring unblinding procedures defined before the trial and any unblinding recorded and assessed.

Two provisions tighten the perimeter around exploratory work. Article 9 requires participant privacy and personal information security to be protected in accordance with China’s requirements, importing the Personal Information Protection Law into GCP inspection scope. Under Article 37, the sponsor must select qualified laboratories and supervise sample management, testing, transport, storage, and destruction, and prohibits sample testing unrelated to the ethics-committee-approved protocol, naming genetic testing as an example. The consent form must state whether residual samples are retained or used after the trial, for how long, and on what terms they may be shared.

A biomarker or genomic sub-study outside the approved protocol is therefore prohibited at GCP level, whatever consent the sponsor holds. Layered on the Human Genetic Resources (HGR) regime covered in earlier GT Advisories, and on cross-border transfer obligations, a China exploratory analysis now needs protocol and ethics coverage, HGR clearance, and a valid personal information transfer mechanism, each verified separately. Adding exploratory endpoints by late amendment should be reconsidered for China sites, since the amendment must clear ethics review before any sample is tested.

Cross-Border Transfer

Trial data captured in China and transferred to a sponsor, central laboratory, or data capture environment abroad is a cross-border transfer of personal information, and health data is sensitive personal information under the Personal Information Protection Law (PIPL). Both are governed by the applicable privacy and data security regulatory framework, including PIPL. One of three mechanisms applies unless an exemption does — a CAC security assessment, a filed China Standard Contract, or personal information protection certification, the last of which became fully usable with measures effective Jan. 1, 2026. Because trial data is sensitive personal information, the sensitive thresholds govern; most single trials fall below the security assessment threshold, but volumes are counted cumulatively from Jan. 1 each year across the same exporter, so sponsors running multiple China programs might assess the aggregate. The contract-necessity exemption should not be assumed to apply: the CAC confirmed in October 2025 that exemptions are construed narrowly, and a clinical trial agreement between sponsor and site is not a contract to which the participant is a party. Note also that the amended Cybersecurity Law, effective Jan. 1, 2026, raised penalty ceilings and removed the expectation of a warning before a fine.

VIII. Enforcement Risks, Penalties, and Data Rejection Under China’s 2026 GCP

Exposure	Basis and Consequence
GCP non-compliance	Under Article 126 of the Drug Administration Law: order to correct and a warning; if not corrected, RMB 100,000 to 500,000; where serious, RMB 500,000 to 2,000,000; suspension of operations, and up to revocation of approval certificates and licenses. Trial institutions may be barred for five years. Responsible individuals face confiscation of income earned during the violation, fines of 10% to 50% of it, and 10-year to lifetime industry bans.
Safety reporting failures	Under Article 127 of the Drug Administration Law, failure to adjust the protocol, to suspend or terminate the clinical trial, or to report on discovering a safety problem attracts an order to correct and a warning, and a fine of RMB 100,000 to 500,000 if not corrected.
Trial suspension or termination	The NMPA may require protocol adjustment, suspension, or termination where risks are identified. In an enrolling pivotal program this might exceed any fine in cost.
Rejection of the data	The Provisions on the Administration of Drug Clinical Trial Institutions provide that the NMPA will not accept data completed by an institution that has not been filed as required, and a filing may be cancelled for concealing or submitting false information. Data that cannot be verified on inspection has no registration value.

One point of precision for foreign sponsors: Article 126 is directed by its terms at marketing authorization holders, manufacturers, distributors, and trial institutions. How it reaches a foreign sponsor that is not yet a China marketing authorization holder is less settled, and the practical exposure may run through trial suspension, data rejection, and consequences visited on the institution and PI than through a direct fine. That is not comfort — it means the sanction that actually threatens the sponsor is the one that destroys the asset, and the one least susceptible to being contracted away.

This is also why the standard matters beyond China. Chinese data is increasingly used in global programs, particularly in rare disease research, and former FDA Acting Commissioner Janet Woodcock put the qualification directly on Shanghai in June 2026: how far the world relies on such data depends on the rigor and quality of what is done in China. Full E6(R3) implementation, a data governance chapter, and clearly allocated non-delegable responsibility are precisely the elements a foreign regulator assesses. Compliance is therefore a precondition of the data’s portability, not only a China obligation — a point to weigh when building on investigator-initiated trial data generated under the previous regime.

Separately, on July 3, 2026, the NMPA released for public comment a draft Announcement on Optimizing the Review and Approval of Cell and Gene Therapy Products — the first time it has addressed this product class specifically. As drafted, qualifying products enter the 30-day channel for innovative drug clinical trial review, and the review period for supplemental applications covering major manufacturing process changes to marketed products that require registration inspection and testing falls from 200 to 130 working days. The draft targets malignant tumors, rare diseases, genetic diseases, immune system diseases, and neurodegenerative diseases, and expressly encourages global simultaneous development and international multicenter trials conducted in China. The CDE issued technical guidelines on the scope and classification of cell therapy and gene therapy products the same day, effective on issuance. The consultation closed Aug. 3, 2026. The draft sits alongside the dual-track system established by State Council Decree No. 818, the Regulations on the Administration of Clinical Research and Clinical Translational Application of New Biomedical Technologies, effective May 1, 2026.

A 30-day clinical trial channel combined with the six-year data protection term might make China an unusually attractive first-filing jurisdiction for first-in-class cell and gene therapies.

IX. Preparing for China’s 2026 GCP Implementation

In light of these changes, stakeholders may wish to consider the following actions.

Owner	Action	When
Legal	Issue a standing written instruction to service providers and sites confirming that subcontracting and off-site delegation require prior written consent from Sept. 1, notwithstanding contrary contract terms, and establish the internal process to grant it at operational speed.	Immediately
Legal	Apply the amendment checklist to all new agreements, then remediate active agreements on a rolling basis prioritized by exposure — providers that actually subcontract, sites with off-site delegation, bioequivalence studies, and budgets lacking fair-market-value support.	New agreements now; rolling thereafter
Legal / Corporate	Add NMPA change-of-sponsor approval to the transaction checklist for any deal touching a China-enrolling trial, as a condition precedent on the 20-working-day statutory clock.	Next transaction
Clinical operations	Run a gap analysis against the 2026 GCP and the full text of ICH E6(R3) together, with particular attention to trial protocol, investigator’s brochure, and essential documents — the three chapters removed from the GCP and now governed only by E6(R3).	Before Sept. 1
Quality / IT	Validate computerized systems against Chapter 5: audit trail completeness, metadata capture, role- and blinding-based access control, backup and contingency, and electronic signature compliance, with the validated state documented as continuing throughout each trial.	Before Sept. 1
Clinical operations	Confirm with each China site that the ethics follow-up interval is risk-based and documented, that consent templates contain no waiver or liability-exclusion language, and that re-consent triggers cover new safety information and changes in participant capacity.	Before Sept. 1
Clinical operations / Legal	Review every protocol with a biomarker, genomic, or residual-sample component against the ethics-approved protocol and confirm HGR clearance and a valid cross-border transfer mechanism are each separately in place.	Before Sept. 1
Clinical operations	Document in the trial master file which standard governed which period for any study running across March 31 or Sept. 1, 2026.	Simultaneously




Conclusion

The revised GCP is not a restatement with new emphases. The standard depends on ICH E6(R3) to be complete, shifts oversight onto sponsors in ways requiring contractual consent gates, and ties clinical conduct to China’s personal information and HGR regimes. With the exclusivity architecture in Part 1, the message is consistent: China data is worth more than it was, and costs more to produce.