Made in China, Approved in the U.S.: How real-world data can bridge the evidence gap

China now accounts for roughly one third of new molecules in global pipelines — up from 4% in 2014 — and oncology leads the way.¹ Large pharma is buying in: the 186 out-licensing deals from China in 2025 reached $137.7 billion in disclosed value, nearly ten times the 2021 total.² Many involve newer modalities, including antibody-drug conjugates (ADCs), bispecifics, and T-cell engagers.

 

While early results on many of these assets are strong, the question remains: how will the data translate to an ex-China population?

For approval, the FDA requires foreign data to apply to U.S. patients; China-only data often doesn’t

 

The FDA can approve a drug on foreign data alone, but only if the data apply to the U.S. population and U.S. medical practice (21 CFR 314.106).³ Its 2024 draft guidance on oncology multiregional trials goes further, asking sponsors to plan U.S. enrollment up front and use control arms that match U.S. care.⁴ More recently, leaders of the FDA’s major review centers and its Oncology Center of Excellence warned that the agency may refuse foreign data unless it can access sites, records, and source data, and that trials run outside the U.S. should expect inspection.⁵

 

Sintilimab (2022) is a clear example of where a China-only trial fell short. ORIENT-11 met its progression-free survival (PFS) endpoint in first-line NSCLC. However:

 

• Every patient was Asian.
• 35% were never smokers, compared with about 15% of U.S. patients with nonsquamous NSCLC.⁶
• 76% were male, while the U.S. population is split approximately even male/female.⁶
• The control arm was chemotherapy alone, which does not reflect U.S. practice.

 

The FDA’s Oncologic Drugs Advisory Committee (ODAC) voted 14 to 1 that more data were needed, and the FDA issued a complete response letter asking for a multiregional trial against U.S. standard of care.⁶

 

Other relevant examples include surufatinib (2022) and glofitamab (2025), which similarly faced challenges with regulatory review because their trials enrolled few U.S. patients or did not reflect contemporary U.S. standards of care.^7,8

 

What drives these rulings is not where the drug is from, but how closely the trial’s patients and comparator match U.S. patients and practice. Sponsors holding a strong China result need to answer two key questions early: What level of additional evidence is needed? And, if a new trial is required, how should it be designed (patients, comparators, endpoints)?

Three questions every sponsor should answer

 

For an asset whose clinical evidence comes mostly from China, diligence comes down to three questions:

 

1. Patients: Do U.S. patients differ from the trial population in ways that change outcomes?

 

2. Biology: Are the drivers of disease the same in U.S. and Chinese patients?

 

3. Practice: How does the U.S. standard of care differ, and what are the outcomes?

 

Each can be answered using data from real-world populations to help sponsors identify evidence gaps, prioritize additional clinical trials, and make better informed development decisions.

 

Patient: Do patients differ in ways that change outcomes?

 

In China-only small cell lung cancer (SCLC) trials, more than 80% of patients are male, versus about half in U.S. trials.⁹ A large gap, but a difference like this matters only if it changes outcomes. Among more than 1,000 U.S. patients in the Tempus real-world dataset with extensive-stage SCLC that progressed after first-line platinum therapy, 46% were male, in line with U.S. trials.¹⁰ In that cohort, overall survival did not differ meaningfully by sex, therefore the male skew in the trial population is unlikely to distort the baseline, de-risking the decision to move to a global trial.

 

Other differences may matter more. Hepatitis B accounts for about 85% of patients in China-only hepatocellular carcinoma (HCC) trials, versus roughly 40% in trials run elsewhere.⁹ Etiology shapes prognosis and possibly response, too. In an exploratory IMbrave150 analysis, the survival benefit of atezolizumab plus bevacizumab looked larger in HBV patients (HR 0.58) versus non-viral HCC (HR 1.05).¹¹ This raises a question: How well does a mostly-HBV trial carry over to a U.S. population? Real-world data offers a way to look at this: how the U.S. etiology mix compares with a given trial’s, and how survival on current therapy varies by etiology. Viral status often goes beyond structured EHR fields alone, so a dataset that includes the full patient record, including unstructured de-identified clinical notes, becomes necessary to uncover the full patient characteristics.

 

Biology: Are the drivers of disease similar?

 

Germline differences matter as well. The UGT1A1*6 variant, which affects clearance of SN-38, the payload of sacituzumab govitecan, is far more common in people of East Asian ancestry.¹² Linked clinical and molecular data lets a sponsor measure biomarker prevalence in U.S. patients, relate it to outcomes on current therapies, and identify ancestry-linked differences.

 

A common obstacle is that relevant subgroups look too small. Race is often missing from real-world data records and depends on self-reporting, so patients of East Asian ancestry can appear too few to study. Genetic ancestry can be inferred from single nucleotide polymorphism (SNP) markers already captured by tumor sequencing, turning a small heterogeneous subgroup question into one with a usable sample size.

 

Practice: How does the U.S. standard of care differ, and what are the outcomes?

 

Recent China-only phase 3 trials in first-line NSCLC, including OptiTROP-Lung05 (sacituzumab tirumotecan plus pembrolizumab) and HARMONi-2 (ivonescimab), tested against pembrolizumab alone in patients with PD-L1 TPS ≥1%. Both reported strong PFS results.^13,14

 

U.S. practice looks different. Among more than 4,000 PD-L1-positive patients with advanced NSCLC adenocarcinoma in the Tempus real-world database, about three in four receive immunotherapy plus chemotherapy — not immunotherapy alone. A platinum doublet plus a checkpoint inhibitor has been the dominant first-line regimen every year since 2018 for TPS 1%–49%. In an analysis of these patients, median real-world PFS was 9.3 months on immunotherapy plus chemotherapy compared to 7.6 months on immunotherapy alone.¹⁵

 

Beating pembrolizumab alone in China isn’t the same as beating what most U.S. patients receive. Real-world data lets a sponsor map U.S. regimens by line of therapy and across PD-L1 subgroups, estimate outcomes as a realistic benchmark, and provide context for interpreting China-only trial data as well as informing the selection of comparators in U.S. trials.

Answer these questions early and build a bridge

 

These questions are usually asked late, ahead of a filing or advisory committee, and by then the options are limited. Nearly three quarters of China out-licensing deals tracked in 2025 involved preclinical or phase 1 assets, where the acquirer is tasked with evaluating the signal and designing a pivotal trial for U.S. relevance.¹⁶ For later-stage assets, where the key trial has already run in China, the same questions show how far the existing data will carry. A few analyses can help chart an initial path:

 

1. Build a replica cohort: Apply the trial’s eligibility criteria to U.S. patients starting the same line of therapy to see how many would qualify and how they differ from those who wouldn’t

 

2. Compare the populations: Line up the trial’s baseline characteristics against the U.S. replica cohort to capture imbalances across clinical and molecular characteristics and test their impact on outcomes

 

3. Re-anchor the comparator: Benchmark the result by measuring how often U.S. patients receive the trial’s control regimen and compare the effect size against current U.S. SoC

 

Real-world data won’t replace a well-designed multiregional trial. It can, however, offer an early indication as to whether a China signal is likely to hold in U.S. patients, where a trial needs U.S. enrollment, and how to design a bridging program the FDA may be more likely to accept. With one of the largest libraries of linked clinical, molecular, and outcomes data in oncology, Tempus helps sponsors answer these strategic questions with U.S. real-world evidence.

 

To learn how Tempus can help you evaluate a China-originated asset, contact our Life Sciences team.

References

 

1. ING. The next Pfizer will be Chinese. ING THINK, May 27, 2026.
2. Reuters. Analysis: China biotech licensing boom to hit record in 2026 as pipeline swells. February 13, 2026. Data from PharmCube; Greater China (mainland China, Hong Kong, Macau, and Taiwan) out-licensing deal count and disclosed potential value, 2021 and 2025.
3. U.S. Code of Federal Regulations, 21 CFR 314.106, Foreign data.
4. U.S. Food and Drug Administration. Considerations for Generating Clinical Evidence from Oncology Multiregional Clinical Development Programs. Draft guidance, September 2024.
5. U.S. Food and Drug Administration. Good Clinical Practices Are Not Optional. FDA Voices, September 2, 2026.
6. FDA Oncologic Drugs Advisory Committee. Briefing document and minutes, BLA 761222 (sintilimab), February 10, 2022; Eli Lilly and Company, press release, March 24, 2022.
7. HUTCHMED. HUTCHMED Receives Complete Response Letter from the U.S. FDA for Surufatinib for the Treatment of Advanced Neuroendocrine Tumors. Press release, May 2, 2022.
8. FDA Oncologic Drugs Advisory Committee. Briefing document and minutes, glofitamab (STARGLO), May 20, 2025.
9. Tempus internal analysis of published clinical trial data, October 2026. Data on file.
10. Internal analysis conducted by Tempus scientists using the Tempus real-world oncology dataset. U.S. patients with extensive-stage SCLC whose disease progressed after first-line platinum-based therapy (n>1,000). Data on file.
11. Cheng AL, Qin S, Ikeda M, et al. Updated efficacy and safety data from IMbrave150: Atezolizumab plus bevacizumab vs. sorafenib for unresectable hepatocellular carcinoma. J Hepatol. 2022;76(4):862-873. doi:10.1016/j.jhep.2021.11.030
12. gnomAD v4.1 allele frequencies; TRODELVY (sacituzumab govitecan-hziy) U.S. prescribing information; Chauhan D, Dello Russo C, Thompson J, et al. UGT1A1 genotype testing for irinotecan: a guideline developed by the UK Centre of Excellence in Regulatory Science and Innovation in Pharmacogenomics (CERSI-PGx). Br J Clin Pharmacol. 2026. doi:10.1002/bcp.70659.
13. Xiong A, et al. Sacituzumab tirumotecan plus pembrolizumab versus pembrolizumab in PD-L1-positive advanced NSCLC (OptiTROP-Lung05). Lancet. 2026. doi:10.1016/S0140-6736(26)00968-2
14. Xiong A, et al. Ivonescimab versus pembrolizumab for PD-L1-positive non-small cell lung cancer (HARMONi-2): a randomised, double-blind, phase 3 study in China. Lancet. 2025;405(10481):839-849.
15. Internal analysis conducted by Tempus scientists using the Tempus real-world oncology dataset. U.S. patients with PD-L1-positive advanced NSCLC adenocarcinoma (n>4,000). Data on file.
16. PitchBook. Q1 2026 Analyst Note: The Chinese Biopharma Landscape: Where New Assets Are Born. As reported in Fidler B. China’s edge in early-stage drugmaking ‘likely to persist,’ PitchBook says. BioPharma Dive, January 26, 2026.