Prospective evaluation of ex vivo organ culture (EVOC)-based functional precision oncology in advanced malignancies: Updated analysis from the CURESPONSE study

CURESPONSE

Paziente / Popolazione Intervento / Esposizione Confronto Esito

In patients with advanced or metastatic malignancies undergoing biopsy before systemic therapy, ex vivo organ culture testing of the planned treatment predicted objective radiological response at 2 months with 71% sensitivity, 82% specificity and a 91.7% negative predictive value.

N

103pazienti

Disegno

Prospective multicentre study in patients with advanced or metastatic malignancies undergoing biopsy before systemic therapy. Fresh tumour samples were processed into ex vivo organ culture (EVOC) models and exposed to the planned treatments, and EVOC predictions were compared with radiological outcomes assessed by RECIST.

Esito

Predictive performance of EVOC at the first clinically actionable imaging assessment (2 months), compared with RECIST radiological outcome. In the initial analysis, objective response was defined as CR/PR.

Risultato103 eligible patients were enrolled, 84 underwent tumour biopsy and 41 yielded evaluable EVOC and clinical outcome data. At the 2-month assessment, 34 patients had paired EVOC and RECIST assessments. Against objective response (CR/PR), EVOC had 71% sensitivity, 82% specificity, 50% PPV and 91.7% NPV. EVOC generation was affected by protocol deviations, including use of 18G rather than 16G needles and failure to transport samples under specified conditions. When protocol-specified conditions were followed, successful EVOC generation was 90%. An EVOC score (≥45) was developed to predict clinical benefit encompassing SD, and the dataset is being re-analysed using disease control (CR/PR/SD vs PD) as the endpoint.

Krell JM, Tookman L, Clark J, et al. Presented at ESMO Congress 2026, abstract LBA95, 26 October. Not yet published.

Discussione e critica

This prospective multicentre study shows that EVOC predictions can be compared with RECIST outcomes, with a high negative predictive value but a PPV of only 50% against objective response. The evaluable sample is small: only 34 of 103 enrolled patients had paired assessments, and protocol deviations in biopsy and sample transport reduced EVOC generation. The abstract gives no confidence intervals, no breakdown by tumour type or treatment, and no performance figures for the EVOC score or for the disease control re-analysis, which is still under way. The study tests prediction of short-term radiological response only and does not show that EVOC-guided treatment selection improves survival or other clinical outcomes; it was funded by Curesponse LTD, with which several authors declare affiliations.