Comparison of Data from Fresh and Frozen AML Samples for Functional Drug Testing

Abstract

Functional testing on cancer patient samples is common in cancer research, but the methodology and type of samples used can vary considerably between studies. In 25 recent studies, where functional testing was performed on blood cancers, half used fresh samples, while others included only frozen or both sample types. Here, we investigated how sample handling influences functional testing in primary acute myeloid leukemia (AML) patient samples. To look for systematic changes in cell population induced by sample handling, we first compared cellular function between frozen (n=70) and freshly isolated (n=112) bone marrow mononuclear cell samples from patients with AML using a high-throughput functional drug testing assay with 528 clinically applied and emerging drugs. We then specifically collected paired samples (n=10) before and after cryopreservation using drug testing and flow cytometry as readouts, including a flow cytometry-based drug testing assay for selected compounds.The Z'-factor, an assay quality indicator, was similar between frozen and fresh sample groups. However, overall cell viability and growth rate were significantly lower in the frozen samples. Interestingly, in fresh samples, delays from sampling to the start of the assay had no effect up to 72h. When comparing overall drug efficacies, most of the drugs tested correlated highly between sample types, but 29 out of 528 drugs showed a notable systematic difference in drug sensitivity. Samples that had been frozen were more sensitive to drugs such as kinase inhibitors (amcasertib, dinaciclib), as well as apoptotic modulators (S-63845), and proteasome inhibitors (VLX1570). Additionally, fresh samples were more sensitive to paclitaxel than frozen ones across the paired samples.The cell composition of paired samples was also affected by cryopreservation, with a reduction in the frequency of cells expressing c-KIT (CD117) and a reduction of cells with high granularity (side scatter). Similar effects could be observed for the flow cytometry-based assay where frozen samples showed a cryopreservation-potentiated reduction in CD34+,CD11b+ and CD56+ cells after proteasome inhibitor treatment, and a reduction in CD56+ cells after taxane treatment. In conclusion, drug responses are in general highly correlated between fresh and frozen samples for about 95% of the drugs tested. However, careful consideration should be given to the specific drugs and cell populations of interest before deciding on sample handling methodology. This may be of particular importance when investigating specific drugs and phenotypes affecting cell proliferation and maturation stage, as sample handling may induce systematic differences and confound interpretation.