Tumor samples were obtained from adult patients aged ≥18 years with acute myeloid leukemia (AML) or colorectal, ovarian or renal cancer at Akademiska Hospital (Uppsala University Hospital) (Uppsala, Sweden) by bone marrow or peripheral blood sampling, intraoperatively or by diagnostic biopsy in accordance with the approval from the Regional Ethical Committee in Uppsala (approval no. 2007/237). The samples were collected January 2008-December 2014, prepared as detailed below and then cryopreserved until use in 2022 for experiments in the present study. Written informed consent for sample collection and further analysis was provided by patients before sampling. Inclusion criteria for sampling were: i) Patients aged ≥18 years; ii) localized or advanced untreated or previously treated disease; and iii) the sampling procedure was considered to be safe and with a prospect that drug sensitivity testing could be of potential patient benefit. Sex distribution among the included patients was 70% (n=155) women and 30% (n=67) men.

The cell preparation protocol for solid and leukemic tumors has been described in detail elsewhere (32). Briefly, for the solid tumors, the samples were manually minced using sterile scissors and dispersed in collagenase [collagenase type I, 1.5 mg/ml (Sigma-Aldrich; Merck KGaA) and DNase type I, 100 µg/ml (Sigma-Aldrich; Merck KGaA) in CO₂ Independent Medium, pH 7.35–7.45 (Gibco; Thermo Fisher Scientific, Inc.)] for 4 h at 37°C. The samples were then purified by density gradient centrifugation [200 × g, 5 min, at room temperature (RT)] (Histopaque^®−1077; Sigma-Aldrich; Merck KGaA) followed by the determination of cell viability and count using a Bürker chamber. The cells were diluted in complete culture medium [RPMI 1640 (Thermo Fisher Scientific, Inc.) containing 10% fetal calf serum (Sigma-Aldrich; Merck KGaA) and 2% glutamine/penicillin-streptomycin] in a 37°C humidified atmosphere containing 5% CO₂ before seeding in plates for further analysis.

The leukemic cells were isolated by Ficoll-Paque density gradient centrifugation (400 × g, 30 min, RT) and then counted using a Bürker chamber after staining for viability in Türks solution (33). The cells were then diluted in complete culture medium until experimental seeding.

Isolated cells were stained using the May-Grünwald Giemsa staining method. Cells were placed on slides then first stained with May-Grünwald solution for 5 min at RT, washed thoroughly in deionized water and then stained in Giemsa solution for 10 min at RT, followed by another wash. Slides were air-dried and evaluated for tumor representativity, based on cell size, nucleus to cytoplasm ratio, chromatin staining pattern, presence of nucleoli and mitoses and cellular adhesion tendency, by a trained cytopathologist. A representative image of May-grünwald Giemsa stained isolated cells from a colorectal tumor is shown in Fig. 1. Only samples with ≥70% tumor cells were included in the study.

The fluorometric microculture cytotoxicity assay (FMCA) described in detail previously (34), was used for the measurement of cell survival. In the FMCA, living cells with an intact membrane hydrolyze fluorescein diacetate to fluorescein. By measuring fluorescence, cell survival is expressed as the survival index % (SI), calculated as: SI=(F_sample-F_blank)/(F_control-F_blank) ×100, where F_sample represents the fluorescence in wells with the drug, F_control the fluorescence in wells with cells and medium and F_blank the fluorescence in wells with medium alone. The FMCA is a total cell death assay, the results of which correspond well with the much more laborious clonogenic assay (35).

Briefly, 5,000 cells/well (solid tumor samples), or 20,000 cells/well (hematological tumor samples) were seeded in 384-well Nunc culture plates (cat. no. 164688; Thermo Fisher Scientific, Inc.) using the pipetting robot, Biomek 4000 (Beckman Coulter, Inc.). Drugs were added immediately after cell seeding using an ECHO 550 system (Beckman Coulter, Inc.). The plates were incubated for 72 h in a 37°C humidified atmosphere containing 5% CO₂ and then transferred to an integrated system for automated fluorescence scanning and data calculation. A successful assay was defined by a fluorescence signal in control wells of ≥5 times the mean blank and a coefficient of variation of cell survival in control cultures of ≤30 and ≥70% tumor cells before incubation and/or on the assay day.

Mbz was purchased from Sigma-Aldrich (Merck KGaA) whereas cisplatin, irinotecan and gemcitabine were either from commercial clinical preparations or obtained from Selleck Chemicals or LC Laboratories. Mbz was tested at concentrations of 45, 15, 5, 1.67 and 0.56 µM, cisplatin at 30, 10, 3.3, 1.12 and 0.37 µM, irinotecan at 180, 60, 20, 6.6 and 2.2 µM, and gemcitabine at 90, 30, 10, 3.34 and 1.12 µM. The procedure of drug dissolution and storage has been described in detail by Blom et al (32).

CT26 murine colon carcinoma cells (ATCC) were diluted in complete culture medium (RPMI 1640 supplemented with 10% fetal calf serum and 2% glutamine/penicillin-streptomycin) and seeded at 500 cells per well into 384-well Corning plates (cat. no. 3764; Thermo Fisher Scientific, Inc.). Plates were incubated for 24 h at 37°C in a humidified atmosphere containing 5% CO₂. Mbz, cisplatin and gemcitabine were tested at the concentrations detailed above and were dispensed using an Echo 650 (Backman Coulter, Inc), followed by a 72 h incubation period in a 37°C humidified atmosphere containing 5% CO₂. After incubation, the plates were centrifuged (200 × g, 5 min, RT) and the supernatant was removed using an ELx405 plate washer. CellTiter-Glo^® 2.0 reagent (25 µl/well; cat. no. G9243; Promega Corporation) was added using the Biomek 4000 liquid handling system. Plates were sealed, wrapped in foil, shaken (600 rpm for 2 min) and incubated in the dark for 10 min. Luminescence was then measured using a FLUOstar Omega microplate reader (BMG Labtech GmbH). The ATP-based viability method was chosen over the FMCA, as CT26 cells exhibit a pronounced fibroblast-like phenotype that may interfere with fluorescence-based viability measurements (36).

The interaction between Mbz and cisplatin (0.37–30 µM), irinotecan (2.2–180 µM) or gemcitabine (1.1–90 µM) in solid tumor samples was assessed by comparing the SI for each drug separately and when combined with Mbz at 5 µM, a concentration that induced modest cytotoxicity alone. The drug concentration ranges used were previously shown to induce a suitable range of cytotoxic effects in solid tumor samples.

For the assessment of drug interaction for each drug and sample SI, area under the curve (AUC) was calculated. AUC_i was defined as the AUC_Combo/AUC_Single ratio, where AUC_Combo denotes the area under the concentration-SI curve for the drug combined with Mbz and AUC_Single denotes the AUC for the cytotoxic drug alone, as illustrated in Fig. 2. Thus, AUC_i ratios ≤1 indicate positive interactions (such as sub-additive or synergistic), whereas ratios >1 indicate negative interactions (such as antagonistic). The results are illustrated by heatmaps plotted using the ‘pheatmap’ (version 1.0.12) package in R (https://www.r-project.org; version 4.1.2), with rows and columns ordered by hierarchical clustering (average linkage and Euclidean distance).

To further detail the interactions between Mbz and irinotecan or cisplatin, FMCA experiments were performed using 4 CRC samples and analyzed in SynergyFinder 3.0 (https://synergyfinder.fimm.fi), an online web-application that enables analysis and visualization of multi-drug combination screening data (37,38). The application offers interaction analysis with different models: Loewe additivity (39), Bliss excess (40), Zero Interaction Potency (ZIP) (41) and Highest Single Agent (HSA) (42). The interactions are visualized as two-dimensional heatmaps and then summarized over a dose-response matrix as an interaction score calculated for each interaction model. The synergy experiments were performed in triplicate.

The animal care and in vivo experimental procedures were performed by Adlego Biomedical AB (Solna, Sweden) according to Good Laboratory Practice standards and per ethical approval by the regional animal ethics committee in Stockholm, Sweden (approval no. 12201-2019). The study started late 2021 and was finalized, including subsequent analyses in our laboratory, late 2022.

In total, 54 female BALB/c mice aged 6–7 weeks at arrival were obtained from Charles River Laboratories, Inc. The animals were kept 4–5 animals/cage in individually ventilated cages (type IVC 4). Bedding was provided by BeeKay Scanbur AB and all of the cages were enriched with happy homes, play tunnels, sizzle nest and soft paper wool (Scanbur AB). The mice were kept at 22±3°C with 50±20% humidity. Lighting was regulated to 12-h light and dark cycles and the animals had free access to food and water (RM 3; Special Diet Services). Animals weighed 17.0–18.4 g at arrival and 19.3–21.2 g on day 20 with no obvious differences between groups.

CT26, a murine colorectal carcinoma cell line derived from the BALB/c strain, was purchased from ATCC. Irinotecan was purchased from Fresenius Kabi AG and Mbz was purchased from Recipharm AB. On Day-11, animals were inoculated with 3×10⁵ CT26 cells (diluted in complete RPMI 1640 medium to a total volume of 100 µl), subcutaneously in the right rear flank. Tumor volumes, body weights and the animal general health status were measured 3 times weekly by designated animal house technicians. In addition, the team conducting the experiments monitored the animals on each day of treatment and during the endpoint procedures. Tumor volumes (in mm³) were measured using a caliper and calculated by the following formula: Length × width × width ×0.44.

On Day-1, when the tumors had reached a mean volume of 46.5±1.97 mm³, the animals were stratified into 6 experimental groups, with 9 animals per group. Treatment was then started at day 0, as detailed in Table I. The dosing schedule was aimed to be as clinically relevant as possible with Mbz considered to be a drug continuously administered orally and irinotecan intermittently intravenously, each at doses giving effects allowing for analysis of interactions (5,43). Treatment was to be continued for a maximum of 41 days after which the animals were euthanized by cervical dislocation. Animals with tumors >2,000 mm³ and/or tumor-related wounds, were to be euthanized pre-term in accordance with the ethical permit governing these experiments. For this reason, most animals were euthanized around day 20, thus tumor growth data are presented up until that day. Survival was defined as time from stratification to pre-term termination or day 41. Following euthanasia, tumors were extracted and placed in medium (CO₂ Independent Medium) with 5% heat-inactivated fetal calf serum (Sigma-Aldrich; Merck KGaA) and 1.1% glutamine/penicillin-streptomycin consisting of equal parts 200 mM L-glutamine (Sigma-Aldrich; Merck KGaA) and penicillin-streptomycin (10,000 U/ml penicillin and 10 mg/ml streptomycin in 0.9% NaCl; Sigma-Aldrich; Merck KGaA) and shipped directly to our laboratory for preparation of the cells for flow cytometric evaluation.

Immune cell infiltration in the mouse tumors was assessed by flow cytometry. Mouse tumor tissues were prepared by mechanical disintegration and collagenase digestion as aforementioned. Cells were then enriched for CD45⁺ expression using an AutoMACS Pro (Miltenyi Biotec GmbH).

Prior to flow cytometry analysis, the cells were washed and diluted with PBS (Thermo Fisher Scientific, Inc.) and human serum albumin (Sigma-Aldrich; Merck KGaA) (PBS-HSA) 0.1% for 5 min at 4°C. Gating antibodies were added to the cells according to the following: i) Macrophage panel: Lymphocyte antigen 6 complex, locus G FITC (cat. no. 127605; Biolegend, Inc.), CD45 Pacific Orange (cat. no. MCD4530; Thermo Fisher Scientific, Inc.), CD163 PE (cat. no. 156704; Biolegend, Inc.), CD38 Pacific blue (cat. no. 102720; Biolegend, Inc.), F4/80 PE/Cy7 (cat. no. 123114; Biolegend, Inc.) and CD11b Violet 711 (cat. no. 101241; Biolegend, Inc.); and ii) T cell panel: CD8 FITC/CD4 PE/CD3 PC7 (cat. no. 558391; BD Biosciences) and CD45 Pacific Orange. All antibodies were diluted in PBS-HSA 0.1%. The cell and antibody mixture was incubated for 20 min in the dark, washed with cold 1% PBS-HSA and centrifuged at 400 × g for 5 min at 4°C. Finally, flow cytometry was performed using a Navios flow cytometer (Beckman Coulter, Inc.) and data were analyzed using Navios software v1.2 (Beckman Coulter, Inc.).

Statistical analysis was performed using GraphPad Prism™ 6 (Dotmatics). Between group differences were first analyzed by one-way ANOVA and if statistically significant, comparisons between groups were made using Dunnett's post hoc test. P<0.05 was considered to indicate a statistically significant difference. The data are presented as mean ± SD throughout.

In total, 222 patient samples fulfilled the FMCA quality criteria including AML (n=24) and ovarian (n=78), colorectal (n=71) and renal (n=49) cancer samples. The single-drug effect of Mbz was modest across all tumor types, with AML showing the highest sensitivity, followed by ovarian, colorectal and renal cancer (Fig. 3). Statistical analysis confirmed a significant difference between AML and CRC at the two highest concentrations (P<0.001). Moreover, a significant difference was observed between AML and renal cell carcinoma at all tested concentrations except the lowest (P<0.001 for the two highest concentrations and P<0.05 for the remaining concentrations). Mbz had a modest effect at 5 µM (mean SI% 72, 76, 83 and 88 in AML, ovarian, colorectal and renal cancer, respectively) and this concentration was selected for the combination experiments.

Drug interactions notably differed between samples and tumor types (Fig. 4). In all of the samples (n=82), regardless of diagnosis, a positive interaction following the addition of Mbz was observed in 68% of the samples (50/74) for cisplatin, 77% (58/75) for gemcitabine and in 88% (72/82) for irinotecan. The corresponding numbers for renal cancer were 73% (14/19), 79% (15/19) and 85% (18/21), for ovarian cancer 64% (25/39), 68% (28/41) and 90% (38/42) and for CRC 84% (16/19), 69% (12/16) and 93% (14/15). Notably, some drug-sample combinations were not tested in all experiments (indicated in grey in Fig. 4), which explains why the sum of subgroup analyses does not equal the total number of samples.

To investigate the observed interactions between Mbz and irinotecan or cisplatin in more detail, a new set of FMCA experiments with 4 CRC samples were performed using the SynergyFinder application with four models for drug interactions. According to the application, interaction scores in the −10 to +10 range suggests additive interactions, >+10 synergistic and <-10 antagonistic interactions (36). The interaction scores for each patient and each interaction model are shown in Fig. 5. Overall, the interaction scores for the Mbz/irinotecan combination were higher across most interaction models compared with the Mbz/cisplatin combination, suggesting the Mbz/irinotecan combination to be the most favorable. Therefore, this combination was selected for the in vivo study. The majority of synergy scores for both combinations fell within the range of −10 to +10, indicating additive effects. However, the Mbz/irinotecan combinations occasionally exceeded +10 in certain models and samples, suggesting potential synergistic interactions.

Using Mann-Whitney test, a statistically significantly higher overall mean score was observed for the Mbz/irinotecan combination compared with the Mbz/cisplatin combination [mean synergy score for Mbz/irinotecan, 2.63 (Bliss excess, 2.11; HSA, 3.48; Loewe additivity, 2.63; ZIP, 2.32) vs. mean synergy score for Mbz/cisplatin, −2.69 (Bliss excess, −4.11; HSA, −1.16; Loewe additivity, −2.12; ZIP, −3.37); P<0.0001; Fig. 5].

The low synergy scores for the Mbz/cisplatin combination in sample 1 likely reflect the intrinsic drug sensitivity and resistance of that particular tumor sample. Individual tumor samples exhibit substantial heterogeneity in drug response, which underscores the value of using a patient-derived cell model to reflect drug activity in the clinic.

As a basis for the assessment of drug interactions in vivo, a corresponding set of experiments with in vitro CT26 cells were conducted using Mbz alone and in combination with cisplatin or irinotecan. Mbz alone showed a modest and an almost concentration independent activity (Fig. S1A), whereas both cisplatin and irinotecan induced a dose-dependent cytotoxicity (Fig. S1A and B). Furthermore, Mbz at 5 µM enhanced the cytotoxic effect of cisplatin and irinotecan in an additive/sub-additive pattern without notable differences between these drugs (Fig. S1A-C).

Animal body weight and health status assessments showed that drug administration was not associated with a decrease in overall animal health status. Tumor growth was rapid and some animals (1–3 animals in each group) developed tumor-related wounds and were thus euthanized prior to the final study day. Tumor-related wound development was evenly distributed throughout groups and likely not correlated to drug administration. The maximum measured tumor diameter and volume were 17.6 mm and 2,877 mm³, respectively.

The mean tumor volumes for the control and experimental groups are shown in Fig. 6, divided into a and b for clarity. The values are presented up until day 13 for the control group as after this day most animals in this group had been euthanized pre-term, in accordance with the study protocol due to tumor growth, whereas most animals in the experimental groups were continued to day 20. At day 13, the control group had a mean tumor volume of 651 mm³ compared with 488 mm³ in the low dose Mbz group and 353 mm³ in the high dose Mbz group. The mean tumor volume at day 13 in animals treated with irinotecan was 523 mm³ and when irinotecan was combined with low or high dose Mbz, the volumes were 402 mm³ and 472 mm³, respectively. Differences between the groups were not statically significant. At day 20, the mean tumor volume in the irinotecan group was 1,533 mm³, the low dose Mbz group was 1,133 mm³ and the high dose Mbz group was 517 mm³. The tumor volume for irinotecan combined with low dose Mbz at day 20 was 737 mm³ and 699 mm³ for high dose Mbz. Compared with irinotecan alone, the differences were not statistically significant. At day 20, the difference between all groups was statistically significant according to one-way ANOVA (P=0.048), but only the difference between the mean tumor volumes in the irinotecan and high dose Mbz groups was statistically significant following post hoc test (P=0.04).

Kaplan Meier survival curves are presented in Fig. 7. Drug exposed animals showed a longer survival compared with the control animals. Control animals had a median survival time of 13 days, compared with 20–29 days for those drugs exposed. In agreement with the effect on tumor volumes, a dose-dependent effect of Mbz alone was indicated, with a median survival time of 20 days in animals treated with low dose Mbz compared with 29 days for high dose Mbz. Animals treated with irinotecan combined with low or high dose Mbz had median survival times of 27 days.

Based on our previous ex vivo findings of an immunomodulatory effect of Mbz switching macrophages from the M2 to M1 type (14,16), tumor tissue from the animals was retrieved and analyzed by flow cytometry to detect the presence of various subsets of immune cells. CD45⁺ cells were sorted for CD4⁺/CD8⁺ T cells, and macrophages were then classified further as type M1 (CD38⁺/CD80⁺) or type M2 (CD206⁺). Fig. S2, Fig. S3, Fig. S4, Fig. S5, Fig. S6, Fig. S7 illustrate the gating strategies used to identify macrophages and T cells, for each group separately.

Overall, Mbz slightly increased the infiltration of macrophages and CD4⁺ T cells (Fig. 8A and B). Mbz tended to dose-dependently increase the levels of M1 macrophages and decrease M2 macrophages (Fig. 8D-F). Thus, the mean M1/M2 ratio for low and high dose Mbz were 0.26 (range, 0.15–1.09) and 1.35 (range, 0.12–5.22), respectively, compared with 0.25 (range, 0.08–0.78) for the control group.

Mbz in combination with irinotecan increased the M1/M2 ratio, but not beyond the effect of Mbz alone. Irinotecan alone seemingly increased the macrophage infiltration but otherwise did not notably alter the immune cell infiltration. The tumor immune cell data exhibited substantial variability and no statistically significant differences were observed between groups, except for CD8⁺ T cells. A significant increase in CD8⁺ T cells was observed in the high dose Mbz + irinotecan group compared with the control, low dose Mbz + irinotecan, low dose Mbz and irinotecan alone groups (P=0.01; Fig. 8C).