Cancer Research · us
Keeping Tumors Outside the Body for Drug Testing: Organoid Research Expands the Candidate Pool for Precision Cancer Therapy
Two studies bring miniature models of patients’ tumors closer to serving as repeatedly testable treatment surrogates: one spans 15 cancer types, while the other brings lung cancer and immune cells together in the same experimental system, also revealing that current treatment criteria may overlook some potential beneficiaries.
The challenge of precision cancer therapy lies not only in identifying tumor mutations, but also in determining which changes truly affect drug response. “Organoids” grown from patients’ tumors offer a more direct approach: preserving some of the tumor’s structural and molecular features outside the body, then testing its actual response to drugs. Two studies led by teams at Weill Cornell Medicine respectively expanded the range of cancer types covered by such models and attempted to incorporate the immune microenvironment into a lung cancer drug-testing platform.
One study, published in *Science Advances*, used samples from 191 patients to establish 220 organoids spanning 15 cancer types. The researchers reported that the organoids matched the original tumors’ histopathological features at a rate of 93%, had a median concordance rate of 80% for driver mutations, and showed a median correlation coefficient of 0.85 for gene expression. They also retained 85% of dominant tumor clones, and their transcriptional features remained broadly stable after 10 passages.
These figures matter because if organoids rapidly become a different biological system during culture, it becomes difficult to interpret drug-testing results in relation to patients. The study shows that it is feasible to establish a relatively stable organoid library across cancer types while retaining the heterogeneity of the original tumors. However, “similar” does not mean a complete replica: sampling itself may miss minority cell populations within a tumor, and culture conditions may select for clones that survive more readily.
Drug testing produced signals with greater clinical relevance. Among patient-derived organoids considered ineligible under current criteria for treatment with U.S. Food and Drug Administration-approved PARP inhibitors, 58% of the models tested showed a significant response to talazoparib. This does not mean that the same proportion of patients would necessarily benefit after receiving treatment, but it suggests that defining treatment eligibility by a specific cancer type or established biomarker may miss tumors that are drug-sensitive. Combination-drug screening in the study also identified several possible approaches to overcoming drug resistance, particularly in TP53-mutant models.
Another study, published in *Cell Reports Methods*, addressed a commonly criticized gap in organoids: they contain tumor cells but lack the immune environment closely associated with treatment efficacy. From the same non-small cell lung cancer surgical specimen, the team simultaneously cultured tumor organoids and expanded tumor-infiltrating lymphocytes, then used interferon-γ secretion and T-cell killing capacity to evaluate immune checkpoint inhibitors, targeted inhibitors, and their combinations.
The system captured differences in response among patients and also showed that tumor-infiltrating lymphocytes and peripheral blood immune cells from the same patient cannot substitute for one another. The researchers also generated tumor-associated macrophages matched to individual organoids and tested how they affected tumor growth and chemotherapy sensitivity, advancing the model from a simple cancer-cell drug-sensitivity assay toward a multicellular system that more closely resembles the tumor microenvironment.
However, the two studies currently provide evidence mainly for the platforms’ biological fidelity and experimental scalability. They have not yet shown that selecting drugs based on organoid results can extend patient survival or outperform current molecular testing. Clinical application must still contend with culture success rates, the time required to complete testing, specimen quality, standardization across laboratories, and how to translate responses outside the body into verifiable treatment thresholds. The critical next step will be prospective clinical trials: completing testing before treatment decisions are made and examining whether the models’ predictions truly align with patient responses.