China’s clinical trial regulations received a major update with the release of the 2025 Draft Revision of the Good Clinical Practice for Drugs. It is the biggest shake-up to clinical trial regulations since 2020.
While the 2020 revision narrowed the gap between Chinese and international standards, the new draft goes further, aligning China’s requirements with the upcoming guidelines worldwide.
But it’s more than alignment. It represents a philosophical shift, moving away from box-checking compliance and toward building quality into trial design from the start. For sponsors and CROs, this means rethinking how trials are planned, executed, and monitored.
With that in mind, we will walk you through three major operational changes (around accountability, quality-by-design oversight, and data governance), then explore what these shifts mean for accelerated IND timelines and the often-overlooked details of terminology and translation.
From "unclear accountability" to “sponsor responsibility"
The 2020 GCP left a critical question unanswered: when things go wrong in a trial, who’s ultimately responsible? In China’s layered sponsor-CRO-site structure, that ambiguity created real risk.
The 2025 draft clears it up. The sponsor is now explicitly accountable for trial quality across the entire lifecycle.
Principal Investigators (PI) retain responsibility for site-level medical care and subject safety. But sponsors can no longer “outsource” liability. The draft mandates that sponsors maintain “end-to-end quality responsibility” across all operations, including those delegated to CROs.
This implies that a failure by a CRO (whether in monitoring, data management, or statistical analysis) is treated by inspectors as sponsor failure. The sponsor must demonstrate that they not only hired a vendor but actively oversaw and verified the vendor’s quality. This requires a shift in vendor management strategies from periodic audits to continuous, data-driven oversight.
Quality by design (QbD) and risk-based oversight
The 2025 draft weaves QbD principles throughout the trial lifecycle. Sponsors must now build quality management into every phase, starting at protocol development, not after the fact.
This means identifying upfront which processes and data points are critical to participant safety and result reliability (what ICH calls Critical to Quality, or CtQ factors). Sponsors then assess risks at both the system level and the trial level, embedding controls directly into the protocol, monitoring plan, SOPs, and training programs. Quality becomes structural, not procedural.
The draft also formalizes what risk-based monitoring actually requires. Monitoring efforts should be both prioritized and proportionate—calibrated to the specific risk profile and complexity of each trial. A phase 1 oncology study and a post-marketing surveillance program shouldn’t follow identical monitoring templates.
Perhaps most notably, the draft explicitly endorses hybrid monitoring models. Sponsors can combine on-site visits with centralized monitoring techniques (statistical analysis, sampling-based data checks, automated anomaly detection) to direct resources where risk is highest. The goal is smarter oversight, not just more oversight.
The new "Data Governance" chapter
The 2025 draft introduces an entire chapter dedicated to data governance. That’s a strong signal that data integrity a foundational requirement built into trial infrastructure.
This chapter operationalizes the ALCOA++ principles (Attributable, Legible, Contemporaneous, Original, Accurate, plus Complete, Consistent, Enduring, and Available) by translating them into specific, enforceable controls around validation, audit trails, and lifecycle management. Three requirements stand out:
- System validation: All electronic systems used in trials (e.g., eCRF, ePRO, EDC platforms) must now meet explicit validation requirements. No exceptions, no grandfathering of legacy systems.
- Audit trails: Every data modification must be logged. The ability to track who changed what, when, and why is now baseline compliance.
- Data lifecycle management: Governance extends from data creation through long-term archiving. Sponsors must ensure that trial data remains retrievable and readable for the full mandatory retention period. A requirement that has implications for vendor selection, IT infrastructure, and archiving strategies.
Specific mandates: Bioequivalence and PI roles
The 2025 draft carries forward two requirements from the 2020 version, but sharpens their intent.
- Bioequivalence sample retention: Sponsors must retain BE study samples for at least two years post-marketing and cannot return them to the sponsor or affiliated parties. This isn’t administrative busywork; it gives regulators the ability to conduct retrospective testing if data integrity questions emerge after approval.
- PI accountability: The draft refines the PI definition to emphasize their role as the “site-level responsible person.” While the sponsor owns overall trial responsibility, the PI is accountable for two things:
- The medical care of participants.
- The integrity of data generated at their site.
The draft also clarifies that PIs must have adequate time and resources to actually supervise the trial. This is a direct shot at the “ghost PI” problem (investigators who sign off on studies without meaningful involvement). The expectation now is active oversight, not nominal leadership.
How to accelerate access: The 30-Day IND Pathway
China’s standard IND process operates on a 60-working-day implicit approval mechanism. For eligible innovative drugs, the NMPA now offers an accelerated 30-working-day pathway.
The policy was piloted starting in July 2024 and formalized nationwide through NMPA Announcement [2025] No. 86, effective September 2025.
But this isn’t a blanket acceleration.
It’s a specialized track with strict entry criteria that require high operational maturity from sponsors.
Eligibility Criteria: Who Qualifies?
The 30-working-day pathway applies to INDs for Class I innovative drugs (1类创新药), spanning traditional Chinese medicines, chemical drugs, and biological products. But applications must meet standard dossier submission requirements and satisfy at least one of the NMPA’s qualifying conditions, including:
- Nationally supported key innovative drugs with significant clinical value under national “whole-chain” innovation support policies.
- Eligible pediatric, rare disease, and TCM innovative drugs published by the CDE. In practice, this criterion has been linked to CDE initiatives such as the pediatric anti-tumor “Spark Program” (星光计划) and the rare disease CARE Plan (关爱计划).
- Globally synchronized development products, including Phase I–II globally synchronized trials and Phase III multi-regional clinical trials (MRCTs) led or co-led by principal investigators from domestic clinical trial institutions.
The 12-week initiation commitment
The most operationally demanding requirement is the commitment to initiate the clinical trial within 12 weeks of approval. In the consultation draft, “initiation” was defined as the point when the first subject signs the informed consent form, making the clock operationally real, not symbolic.
Consistent with this timeline, the pathway requires applicants to assess the lead site’s ethics review capability and the PI’s qualifications before submission. Sponsors will likely have to work with the lead institution to run project establishment and ethics review in parallel, not sequentially.
This compresses the traditional workflow and raises execution pressure. If a sponsor secures 30-day approval but can’t activate recruitment quickly, the time advantage evaporates.
The documentation hurdles
To enter the 30-working-day pathway, applicants must flag the application as “30-day pathway” and submit corresponding supporting documentation per CDE dossier requirements. The submission package emphasizes two areas:
- Trial readiness: Meaning documented progress on site contracting and ethics review (not just intent, but evidence of actual coordination).
- Risk management capability: Demonstrated through an R&D-period risk management plan (研发期间风险管理计划 / RMP section), which must be included in the application materials.
Mechanisms and timelines of the review process
The 30-day pathway’s operational workflow prioritizes speed but includes built-in safety valves.
- Acceptance review (within 5 working days of submission): The CDE conducts an acceptance review and issues a decision within 5 working days of submission.
- 30-working-day review and approval: For applications accepted into the pathway, the CDE completes review and approval within 30 working days from acceptance.
- Fallback mechanism: If the review cannot be completed within 30 working days due to complex technical issues, expert consultations, or required meetings, the CDE notifies the applicant via the Applicant’s Window within 20 working days of acceptance. The overall review period then reverts to the standard 60 days.
The pathway rewards operational readiness, but readiness isn’t just about execution, it’s also about documentation. Sponsors must prove compliance on paper and in systems. This is where language becomes a compliance variable: consent forms, privacy notices, contracts, and submission artifacts must be both accurate and usable by their intended audiences.
Linguistic compliance and localization risks
The 2025 draft increases scrutiny on demonstrable quality, i.e., participant protection, traceable data changes, and risk controls calibrated to trial complexity. In this environment, language becomes a compliance concern.
Translation errors can surface as ethics deviations, data integrity findings, or contract scope mismatches during inspections.
Informed consent and back translation
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Ut elit tellus, luctus nec ullamcorper mattis, pulvinar dapibus leo.
The 2025 GCP draft repeatedly emphasizes participant protection, positioning ethics review and informed consent as core safeguards. This makes the accuracy of site-facing Informed Consent Forms (ICFs) a practical compliance risk.
In global trials, a common operational hazard is meaning drift between the English “master” ICF and the Chinese ICF used at sites, particularly around risk language, procedure descriptions, compensation terms, and benefit statements.
If the Chinese ICF omits material risks or introduces benefit claims unsupported by the approved protocol, the consent may no longer be truly “informed.” This creates serious ethics and inspection exposure.
Back-translation isn’t universally mandated by regulators (and the GCP is no exception), but some IRBs and ethics committees recommend it (especially for complex or higher-risk trials) as a defensible quality control to verify conceptual equivalence.
For studies supporting multinational submissions, sponsors need a documented process showing that translated consent materials are prepared by qualified parties and reviewed through ethics/IRB pathways. Some major regulators explicitly address enrollment of participants who don’t understand the source language and expect appropriate translation procedures to be in place.
Understandable language" as a regulatory standard
China’s ethics and GCP framework has long required that informed consent information be presented in language participants can understand, i.e., clear, accessible, and appropriate to their comprehension level. In other words, while a translation can be linguistically correct, it may still fail the “understandable” standard in practice.
At the same time, global regulators and industry standards are placing increasing emphasis on plain-language communication in clinical research. The implication for sponsors is clear: workflows must verify comprehension, not just literal equivalence.
Data privacy terminology risks
A nuanced but high-impact risk lies in how data is categorized across privacy notices, ICFs, and cross-border transfer contracts. Under China’s Personal Information Protection Law (PIPL), terms like biometric information (生物识别信息) and medical health information (医疗健康信息) are explicitly classified as Sensitive Personal Information. Meanwhile, Human Genetic Resources (人类遗传资源, or HGR) fall under a separate regulatory regime, where HGR “information” includes human genes and genome data.
Misclassification creates two distinct risks:
- Over-labeling: Classifying ordinary clinical or health datasets as “genetic” or “HGR” can unnecessarily trigger HGR compliance pathways administered by the State Council science-and-technology authority (MOST). This increases both timeline and operational burden without regulatory justification.
- Under-labeling: Failing to classify genuinely genetic datasets (e.g., exome sequencing or genomic data) as HGR in transfer documentation creates serious exposure if the underlying activity constitutes regulated HGR conduct without required approvals or filings. The Biosecurity Law provides for measures including confiscation of illegal gains and restrictions on responsible persons’ future engagement in the activity in serious cases.
Sponsors should implement a legally vetted data-classification glossary and decision tree that applies consistently across contracts, privacy notices, and ICF language. This framework should align with PIPL’s Sensitive Personal Information definitions and the HGR definitions in the HGR regulation and implementing rules. Cross-border transfer documentation must match the actual data type and the corresponding regulatory pathway.