Introduction: While children diagnosed with the most common pediatric malignancy, acute lymphoblastic leukemia (ALL), now experience close to a 90% likelihood of cure, the outcome for several high-risk subtypes remains poor. Furthermore, since standard-of-care drugs are extremely effective in this disease, and there are relatively few patients eligible for early phase clinical trials, only the most promising new agents are advanced for clinical evaluation following rigorous preclinical testing. However, conventional preclinical testing of novel agents is not sufficiently resourced to be able to encompass the vast heterogeneity of pediatric ALL, and new approaches to preclinical testing are required in this disease. The purpose of this study was to evaluate the utility of a single mouse trial (SMT) platform for preclinical assessment of novel agents on an almost clinical trial scale, to encompass the broad heterogeneity of pediatric ALL in a single experiment, and to identify molecular biomarkers associated with in vivo drug responses when carried out using molecularly-annotated patient-derived xenografts (PDXs).

Methods: Eighty pediatric ALL PDXs broadly representative of all pediatric ALL subtypes were characterized in terms of engraftment kinetics in immune-deficient NSG mice, and molecularly annotated by RNA-seq, exome-seq and DNA copy number analysis. Between 2-5 million cells from each PDX were inoculated via the tail vein into 2 NSG mice/PDX. Starting at 2 weeks post inoculation engraftment was monitored by flow cytometric enumeration of the proportion of human CD45+cells in the murine peripheral blood (%huCD45+). When the %huCD45+ for each PDX reached >1% one mouse/PDX was treated with the established drug and topoisomerase I inhibitor topotecan (Tpt, 0.6 mg/kg IP daily x 5 x 2 weeks, repeated at 21 days) and the other mouse was treated with the experimental drug and second mitochondria-derived activator of caspases (SMAC)-mimetic birinapant (Bpt, 15 mg/kg IP every 3 days x 5). Treatment response was monitored using stringent objective response criteria modeled after the clinical setting, by mouse event-free survival (EFS where an event was defined as 25% huCD45+), and by waterfall plots comparing the maximum decrease in %huCD45+at any point post treatment initiation. The authenticity of each PDX was verified using a 60-allele SNP array both at the time of inoculation and at relapse post drug treatment for all mice.

Results: Retrospective analysis of 1,000 samples of "single mouse" data previously showed that the single mouse results predicted the overall group response from conventional testing 75.3% of the time, which increased to 94.3% if a deviation of ± one response category was allowed. SMT results were achieved for 71 (88.8%) and 73 (91.3%) of the intended 80 mice for Tpt and Bpt, respectively. Waterfall plots revealed that 60/71 (84.5%) and 30/73 (41.1%) of PDXs achieved regressions in response to Tpt and Bpt treatment, respectively. When compared with historical objective response measures from conventional drug testing carried out by the Pediatric Preclinical Testing Consortium (PPTC) the SMT results showed high concordance for both Tpt (R=0.904; P=0.014; n=7 PDXs) and Bpt (R=0.804; P<0.0001; n=20 PDXs). Moreover, the SMT confirmed the preferential in vivo efficacy of Bpt against B-ALL compared with T-ALL, early T-cell precursor ALL (ETP-ALL) and ALL with mixed linage leukemia gene rearrangements (MLLr-ALL) that was previously revealed by conventional testing carried out by the PPTC. Analysis of divergent responses observed within the MLLr-ALL subpanel (n=9 PDXs) to Tpt revealed a potential 40 gene signature model, with up-regulated genes in Responders being associated with transcription regulation, cellular proliferation and differentiation.

Conclusions: This study has shown that SMTs provide an accurate and cost-effective platform for preclinical drug testing in pediatric ALL on an almost clinical trial scale. Moreover, SMTs can almost encompass the heterogeneity of pediatric ALL in a single experiment, and are likely to be useful for large-scale correlations with in vitro drug sensitivity data. Finally, SMTs have the power to identify molecular biomarkers of in vivo response to established and novel drugs in pediatric ALL when combined with exome-seq, RNA-seq and DNA copy number analysis of molecularly-annotated PDXs. Supported by U01CA199000 from the NCI.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.