Precision oncology researchers have developed a framework for using patient-derived organoids to customize drug-resistant treatment for triple-negative breast cancer (TNBC). The study, published in Cell Reports Medicine, uses clinical data from the I-SPY2 adaptive breast cancer trial to project patient response biology onto organoids, identifying therapies that failed in standard trials but succeed in biomarker-defined groups.
Using clinical data to predict resistance
The approach reverses the traditional workflow of drug discovery. Instead of growing organoids and screening compounds to see if they resemble real patients, investigators led by Laura J. van ’t Veer, PhD, and Jennifer M. Rosenbluth, MD, PhD, began with the I-SPY2 clinical dataset. This trial evaluated 24 treatment regimens in over 2,100 women and generated gene-expression signatures linked to pathologic complete response. The researchers used these clinically validated signatures to identify measurable biomarkers in patient-derived organoids.
A model predicting response to veliparib-platinum therapy in TNBC achieved an excellent AUROC of 0.912 in the clinical dataset. It accurately distinguished sensitive and resistant organoids. This demonstrated that response signatures from patients can be transferred into an experimental system. Only after the models were validated did the drug discovery begin.
Recovering lost treatments
The team focused on a poor-prognosis, platinum-resistant TNBC subtype. They screened 386 compounds and identified several promising candidates. The BCL-2 inhibitor navitoclax (ABT-263) restored cisplatin sensitivity across resistant organoids and showed synergistic activity. The HSP90 inhibitor ganetespib demonstrated potent single-agent efficacy. These results suggest that biomarker-guided patient selection could revive therapies that showed only modest benefit in unselected clinical populations.
Related: GLP-1 Drug May Cut MACE Risk in High-Risk Patients
While current organoid models do not fully capture immune and stromal interactions, they offer a distinct advantage by reframing the research model. Rather than serving as mere miniatures for screening, these systems function as experimental extensions of clinical trials. This allows researchers to investigate resistance mechanisms and discover therapies for specific patient groups.
Historically, the disconnect between lab results and patient outcomes has limited the utility of organoids in clinical settings. This study bridges that gap by anchoring the laboratory work in real-world data. By establishing a direct pipeline from patient signatures to drug testing, the method offers a more reliable path for translating preclinical findings into effective therapies for hard-to-treat cancers.
Extending the framework to other cancers
Patient-derived organoids are already widely used in academia and industry for drug screening and biomarker discovery. Traditionally, however, researchers establish organoids, screen compounds, and only then ask whether the findings translate to patients. This study reverses that workflow. Instead of starting with the organoids, the investigators began with clinically validated biomarkers from the I-SPY2 adaptive breast cancer trial and used them to identify organoids representing clinically meaningful states of drug sensitivity or resistance. Drug discovery therefore occurred within a framework grounded in real patient outcomes rather than laboratory observations alone, potentially accelerating the translation of preclinical findings into the clinic.
The strategy could extend well beyond breast cancer, as organoid biobanks and biomarker-rich clinical trials are becoming increasingly common across oncology. Although current organoids incompletely model immune and stromal interactions and still require prospective clinical validation, the work reframes them as more than miniature tumors for drug screening. Instead, they become experimental extensions of clinical trials, enabling researchers to investigate mechanisms of resistance and discover therapies for biomarker-defined patient populations most likely to benefit.
Leave a Reply