2.1 Patient Samples

In this study, we used patient samples of all main childhood acute leukemia types (T-ALL, B-ALL, and AML) as well as samples of ALAL. Patient samples were obtained with the informed consent of the patient's guardians. The study was approved by the Ethical Committee of the Second Faculty of Medicine, Charles University in Prague (ref. no. EK-602.32/22; agreement emitted on May 15, 2022) and conducted in accordance with the Declaration of Helsinki. An overview of samples used in experiments presented in this study could be found in Table S1.

2.2 DRP Assay Design and Setting

The general pipeline for the DRP assay was adapted from the method described in the publication of Frismantas et al. [5] and it is summarized in Figure S1. The DRP assay was performed on 1536-well plates (Corning, 3838, NY, USA). The source of leukemia cells was a liquid nitrogen-stored density gradient enriched bone marrow or peripheral blood diagnostic sample containing a minimum of 80% leukemia cells. As a supporting feeder layer, an immortalized human bone marrow mesenchymal cell line—hTERT cell line (MSC; Cat. No. T0523, Applied Biological Materials Inc., Richmond, Canada) was used.

DRP was assessed in MSC-leukemia cell cocultures in serum-free screening medium (CTS AIM-V, Thermo Fisher Scientific, MA, USA). Prior to the setup of the DRP experiments, MSC were grown in culture medium (Advanced RPMI 1640, Gibco, Thermo Fisher Scientific, MA, USA; supplemented with 10% inactivated FBS, 1% Penicillin–Streptomycin, and 1% GlutaMAX [both Gibco, Thermo Fisher Scientific, MA, USA]) on 10 cm tissue culture dishes (TPP, Trasadingen, Switzerland). After reaching the cell number required for the DRP assay, MSC were harvested using Trypsin–EDTA (Gibco, Thermo Fisher Scientific, MA, USA), washed with PBS, and resuspended in screening medium to a density 225 cells/μL. MSC (900 cells per well) were then plated in 4 μL of screening medium per well and left for 24 h to attach. Subsequently, leukemia cells (5000 cells per well) were added in 2 μL of screening medium to the wells with attached MSC, and the coculture was left for another 24 h. Chemical compounds dissolved in dimethyl sulfoxide (DMSO) were then added in several concentrations (in the range from 3 × 10⁻¹⁴ M to 1 × 10⁻⁴ M, each concentration point in duplicates or triplicates) into individual wells of the 1536-well plate using acoustic liquid transfer (Echo, Beckman Coulter, CA, USA) and the cells were incubated with the compounds for 72 h. Despite the MSC–leukemia cells being cocultured for a long period of time (in total 96 h), no special treatment of MSC to reduce their proliferation and consequently to prevent feeder layer overgrowth was required. The serum-free screening medium significantly reduced the proliferation of MSC, a phenomenon previously described by others [6]. The reduced proliferation of MSC did not prevent their ability to fulfill their function as a support for leukemia cells during the DRP experiment. The average survival rate of untreated leukemia cells at the end of the DRP experiment was 52.4% in coculture with MSC and only 3% without MSC. After 72 h of incubation with chemical compounds, living cells were stained by CyQUANT Direct Cell Proliferation Assay (Thermo Fisher Scientific, MA, USA) and automatically imaged (Operetta, PerkinElmer/Revvity, MA, USA). Acquired images were analyzed using Columbus software (PerkinElmer/Revvity, MA, USA). For each compound, a response curve was calculated using relative inhibition (proportion of maximal efficacy) as a readout and the LC50, concentration when half of the maximal efficacy was reached (relative inhibition = 0.5), served as the main parameter for describing drug potency.

2.3 Production and Use of Conditioned Medium

MSC at a density of 0.5 × 10⁶ cells/mL were seeded in culture medium on 10 cm dishes 6 days prior to the start of the DRP experiment. When 70% confluency was reached (about 3 days after seeding), culture medium was discarded and replaced with serum-free screening medium after a brief rinse of the MSC with PBS. After 3 days, the medium was collected, centrifuged to remove any detached MSC, and directly used to dilute freshly defrosted leukemia cells. The leukemia cell suspension was then added to the pre-attached MSC culture on the 1536-well plate, resulting in a mixture of fresh and conditioned screening medium (volume ratio 2:1), a setup referred to as conditioned medium in this paper. The setting referred to as unconditioned medium used fresh medium.

2.4 Statistics

Statistical significance of differences between cell counts was assessed using unpaired t-test and F-test to compare variances. The significance level of p < 0.05 indicates a statistically significant difference, * denotes p < 0.05, ** denotes p < 0.01, *** denotes p < 0.001, and **** denotes p < 0.0001, while ns signifies no statistical difference. Correlations were assessed using Pearson's correlation coefficient (R squared). All statistical calculations were performed in GraphPad Prism version 10.2.2.

3.1 MSC Detachment in a Coculture of MSC With Leukemia Cells

Coculture of MSC with leukemia cells simulates the bone marrow environment, and data from cocultures reflected in vivo drug responses more closely than without stromal support [7]. However, sometimes, the addition of leukemia cells leads to MSC detachment from the plastic surface of the well bottom. To the best of our knowledge, this phenomenon has not yet been described in the literature. However, it is known in the DRP community to happen rather frequently. In our hands, it occurred in 27.8% of experiments (5 of 18). The detachment could be either mild, with a few larger holes in an otherwise compact MSC layer (11.1% of cases, 2 of 18), or strong, starting with holes in the MSC layer within the first 24 h of coculture and ending with the majority of MSC detached and forming dense patches consisting of a mixture of MSC and leukemia cells (16.7% of cases, 3 of 18; Figure 1A). Leukemia cells indeed appeared to be the trigger of the detachment because we never observed the detachment in control wells containing only MSC (among 18 plates, each containing a minimum of 18 control wells). The tendency to force MSC detachment is an inherent feature of a given specimen, and although most cases were observed with samples from AML patients, it is not limited to any leukemia type (Figure S2A). However, the MSC detachment was not always consistently reproduced when a different aliquot of the same individual was tested under the same experimental conditions (Figure S2B). In the absence of detachment, MSC form a monolayer at the bottom of each well, and automated image analysis, therefore, easily detects and distinguishes them from leukemia cells that lie on top of them. With an increasing degree of MSC detachment, the number of detected MSC drops and variance among replicates significantly increases (Figure 1B). The stronger the MSC detachment, the more variable are also the numbers of detected viable leukemia cells in replicates (Figure 1C). The source of these discrepancies lies primarily in image analysis because individual MSC are challenging to detect or accurately separate from leukemia cells in stretches and dense patches formed after MSC detachment (Figure S3). In addition, the negative impact of the MSC detachment on their role as support for leukemia cells can be expected.

3.2 Use of Conditioned Medium Prevents MSC Detachment

To overcome the problem with MSC detachment, we introduced the utilization of MSC-conditioned medium in the DRP. We observed that replacing one third volume of fresh screening medium with screening medium conditioned on MSC for 3 days positively affected MSC attachment to the well bottom in the coculture, and MSC then remained in monolayer. Parallel DRP experiments using the same patient sample in either unconditioned or conditioned medium showed a markedly reduced MSC detachment in conditioned medium (Figure 2A,B). The evenly distributed monolayer of MSC in conditioned medium facilitated automated image analysis. Thus, more precise numbers of viable leukemia cells could be extracted in comparison with unconditioned medium, where strong MSC detachment occurred in the majority of wells (Figure 2C). This translated into more reliable response curves describing the drug's effect on leukemia cell viability. In general, the correlation of corresponding drug responses measured in both conditioned medium and unconditioned medium was poorer in case strong MSC detachment occurred in unconditioned medium compared to the situation when MSC stayed attached (Figure 2D). A detailed view of differences in dose–response curves using an example of Venetoclax is shown in Figure 2E. This clearly illustrates the negative effect that MSC detachment has on the precision of DRP assay results.

To ensure that conditioned medium does not alter leukemia cell viability per se and that the drug responses were not directly affected, we performed parallel experiments using leukemia cells that did not cause any MSC detachment (Figure 3A). The addition of conditioned medium resulted in a mild change in pH of the medium at the start of the coculture (conditioned medium had pH 7.65, unconditioned medium had pH 7.81), but the overall leukemia cell viability was not affected by the use of conditioned medium. On day 1, cells in unconditioned medium showed slightly higher viability than those in conditioned medium, but on day 4, cells showed a similar survival in both media (Figure 3B). A possible explanation is that conditioned medium contains fewer nutrients and thus provides weaker support for freshly thawed cells, but in later stages of the assay, it supports cells equally well as unconditioned medium. The corresponding response curves calculated from DRPs measured in both media without any MSC detachment were in much better agreement compared to those calculated from experiments where strong MSC detachment occurred in unconditioned medium (Figure 3C–E). In addition, the correlation in this case was very close to the correlation of results from independently repeated DRPs (Figure 3F). Therefore, we can conclude that implementing conditioned medium has no adverse effect on the leukemia cells' viability or their responses to drugs.