Paper Details

ABSTRACT

Background: Pediatric relapsed and refractory acute lymphoblastic leukemia (R/R ALL) remains one of the leading causes of cancer-related death in children. Chemotherapy, targeted drugs and immunotherapy are available, but effective biomarkers to individually assign therapy are needed. Functional precision medicine (FPM) addresses this need by directly testing patient-derived tumor cells against potential therapies to predict clinical response. Dynamic BH3 profiling (DBP) offers a rapid, robust readout of treatment-induced apoptotic priming, enabling efficacy prediction within 48 hours of sample collection while preserving sample viability. We first validated DBP in pediatric ALL cell lines and in 21 cryopreserved samples from newly diagnosed patients, where it discriminated good and poor early responders to induction therapy (AUC = 0.81).

Methods: Within the ReALLNet national network, bone marrow aspirates from R/R ALL patients were obtained in affiliated hospitals and sent to two hubs for Ficoll separation and cryopreservation, followed by centralized DBP and next-generation sequencing (NGS). Each sample was tested against a personalized panel of single agents and combinations proposed by the tumor board and the treating clinicians, based on patient characteristics, genetic profile and clinical trial availability; when cell numbers allowed, additional candidate agents were added. Ex vivo efficacy was determined in each experiment by computing a treatment score (DBP score) using the Strictly Standardized Mean Difference (SSMD*) method, a within-patient measure independent of the number of conditions tested. Treatment decisions were made by each treating clinician after tumor board discussion; no randomization was performed and DBP was used as a functional ancillary diagnostic test. The separation threshold was derived by leave-one-out cross-validation (LOOCV).

Results: Forty-three pediatric R/R ALL patients were enrolled; median age 10 years (range 0.6-17); T-ALL represented 28% of cases and 13 of 43 were analyzed at first relapse, the remainder having no standard-of-care option available. Functional results were given to treating clinicians in a median of 6.5 days (range 2-24), and only 4 of 43 samples did not present enough cell viability for quality control (overall DBP efficacy 90.7%). NGS showed a highly heterogeneous genetic background with frequent concomitant alterations: CDKN2A/B (n=12), TP53 (n=9), IKZF1 (n=7), KRAS (n=6) and NRAS (n=3). Only five patients received treatment matched to their genetic background, and DBP predicted the corresponding targeted drug to be ineffective in four of them; the patients receiving genotype-matched trametinib or ruxolitinib were both refractory. Two patients did not fulfill the inclusion criteria, seven could not complete the planned induction therapy due to rapid progression, and three received a regimen not tested by DBP; all were excluded from the efficacy evaluation. DBP showed a strong capacity to stratify patients who would or would not respond to the treatment decided by the clinician (AUC = 0.86). The LOOCV threshold was highly stable across iterations (mean DBP score 1.17±0.04), yielding a sensitivity of 92.58±2.08% and a specificity of 75.00±2.18%. The DBP-positive group achieved an overall response rate of 75%, compared with 7.7% in DBP-negative patients. DBP predicted outcome both at first relapse under standard-of-care regimens and at second or subsequent relapse, across intensive chemotherapy, immunotherapy and targeted agents. DBP-positive patients showed significantly higher event-free survival (median 262 vs 28 days) and better overall survival (median not reached in the DBP-positive group), allowing consolidation with chemotherapy, blinatumomab, CAR-T cells or stem cell transplantation.

Conclusions: Our results demonstrate the predictive value of DBP to guide treatment decisions for pediatric R/R ALL. Genomic profiling remained crucial for diagnosis, risk assessment and identification of targetable lesions, whereas DBP added a complementary functional layer that helped prioritize among available options and avoid ineffective therapies. DBP has not yet been validated for cell-based therapies. To our knowledge, this is the first prospective study demonstrating DBP's worth as a functional biomarker to foster ALL precision medicine and improve clinical outcomes when incorporated as an additional decision layer within a national precision medicine network.