Organoids were obtained from totally anonymized specimens. However, biobank informed consent for research purposes was obtained to collect the samples at the National Cancer Institute (CRO) of Aviano.

The ascitic fluids were centrifuged at 1,000 rpm for 10 min and the pellets of cells were washed twice with HBSS (Gibco, Massachusetts, United States). In order to lyse erythrocytes present in the fluid, cold red blood cell lysis buffer (Roche Diagnostics, Basel, Switzerland) was added and kept on ice under stirring for 10 min. The pellet was centrifuged at 1,000 rpm for 10 min and resuspended in Cultrex RGF BME, Type 2 (Bio-techne, Minnesota, USA). Solid tumor tissues were incubated in Dulbecco’s modified Eagle’s medium/Nutrient Mixture F-12 Ham implemented with Levofloxacin 100 μg/mL, Vancotex 25 μg/mL, Ciproxin 5 μg/mL, Gentamicin 200 μg/mL, and Fungizone 5 μg/mL for 30 min. The tissues were minced into approximately 0.5–1 mm diameter pieces with fine dissection scissors, and 1 mL of 4 mg/mL collagenase IV (Gibco, Massachusetts, United States) solution was added, incubated at 37°C for no more than 45 min, and mechanically dissociated by pipetting. The clusters of cells were centrifuged at 1,000 rpm for 10 min, resuspended in an appropriate volume of Cultrex RGF BME, Type 2 (Bio-techne, Minnesota, United States) and seeded on a 24-well cell culture plate. After the polymerization of Cultrex, 500 μL of organoid culture medium, as described by Kopper et al., was added to each well and refreshed every 3 days (Kopper et al., 2019). The organoids were cultured at 37°C and 5% of CO₂.

Sections of formalin-fixed, paraffin-embedded ascites and solid tumor patient-derived organoids were used for histopathological analyses. Organoids were recovered from BME using ice-cold cell recovery solution (Corning, New York, United States) in accordance with manufacturing protocols, fixed in phosphate-buffered 10% formalin, and embedded in 500 μL of Bio-Agar (Bio-Optica Milano Spa, Milano, ITA). Five μm sections were stained with hematoxylin and eosin (H&E) using a Leica ST5020 multistainer and 2 μm sections were cut for IHC analysis. The IHC was performed with an UltraVision LP Detection System HRP DAB kit (Thermo Scientific, Waltham, USA). Heat-induced antigen retrieval was performed using 10 mM citrate buffer pH 6.0. The following antibodies were used to characterize HGOC patients-derived organoids and parental tumors: PAX8 (ProteinTech Group, Germany, EU), WT1 (Abcam, U.K.), and CA-125 (Santacruz Biotechnology, TX).

The organoids were characterized by immunofluorescence following the protocol of Dekkers et al. (Dekkers et al., 2019). The following primary antibodies were used to characterize HGOC patients-derived organoids culture in static and passive flow conditions: PAX8 (ProteinTech Group, Germany, EU), WT1 (Abcam, U.K.), and CA-125 (Santacruz Biotechnology, TX). The secondary antibodies that were used are Goat anti-Rabbit IgG (H + L) Alexa Fluor™ Plus 488 (Thermo Fisher Scientific Waltham, Massachusetts, United States) and Goat anti-Mouse IgG (H + L) Alexa Fluor™ Plus 488 (Thermo Fisher Scientific Waltham, Massachusetts, United States). The images were acquired with an EVOS FL Auto 2 (Thermo Fisher Scientific Waltham, Massachusetts, United States).

The Mimetas 2-lane OrganoPlate^® consists of 384 well plates composed of 96 independent tissue culture chips. The medium perfusion is provided by gravity through the microfluidic channels of the OrganoFlow^®, a pump- and tube-free perfusion basculating platform with an adjustable time and tilting angle and interval rocking settings. In order to establish better conditions for cultured organoids in the Mimetas 2-lane OrganoPlate^®, a comparison with the static culture in the 96-wells plates was done. Clusters of organoids were mixed in an appropriate volume (approximately 10,000 clusters of cells/mL) of Cultrex RGF BME, Type 2 (Bio-techne, Minnesota, United States) previously diluted 1:1 with DMEM/F12 without phenol red (SIGMA-ALDRICH, Missouri, United States), and 2 uL of this mixture in five replicates was seeded in 96-wells plates and every chip of the Mimetas OrganoPlate^® 2-lane 96. In particular, the mixture of organoids, Cultrex, and DMEM/F12 was loaded in the “Gel Inlet” channel of the OrganoPlate^® 2-lane 96, and after the polymerization of the mixture, 50 μL of the organoid medium was loaded in the “Medium inlet” channel and 50 μL in the “Medium outlet” channel (Figure 1). Flow and inclination rates were evaluated and compared with the static cell cultures in the 96-wells plates. There were three different tilting angles of 7°, 15°, and 20° each with four different intervals of rocking every 30 s; 1, 8, 40 min, and continuous rocking were compared. The viability of organoids in dynamic and static conditions was evaluated at 0, 24, 72, and 96 h with the PrestoBlue HS Assay (Invitrogen Corporation, San Diego, United States) in accordance with manufacturing protocols. The viability was performed in seven replicates and the p-value was calculated using two-way ANOVA with GraphPad Prism (La Jolla, CA, United States). The p-values are expressed as follows: * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001.

In order to assess the viability of the organoids to evaluate differences in proliferation and necrosis between the static and the dynamic cultures, the LIVE/DEAD^® Viability/Cytotoxicity Kit (Thermo Fisher Scientific Waltham, Massachusetts, United States) was used. The viability of the organoids was evaluated with or without 50 μM Carboplatin at 72 and 96 h. In particular, the organoids were stained for 40 min with Calcein AM 0,50 μM, washed twice with HBSS, and stained with EthD-1 0,25 μM for 40 min. After being washed twice with HBSS, 100 μL DMEM/F12 without phenol red was applied and the fluorescence was observed in a Nikon Eclipse Ts2R. The green (live organoids cells) and red (dead organoids cells) pixels were calculated and were normalized on the blue pixels (total number of cells) with ImageJ.

To evaluate the penetration of the drug into the ECM in static and dynamic conditions, organoids were treated with 1 μM Oregon Green™ 488 Paclitaxel (Thermo Fisher Scientific Waltham, Massachusetts, United States). The Fluo-Paclitaxel was monitored over time at 3, 24, 48, 72, and 96 h. The organoids were stained with Hoechst 33,342 (Thermo Fisher Scientific Waltham, Massachusetts, United States) for 20 min, washed twice with HBSS, and stained with 5 μg/mL of Propidium iodide (Thermo Fisher Scientific Waltham, Massachusetts, United States) for 15 min. After being washed twice with HBSS, 100 μL DMEM/F12 without phenol red was applied and the fluorescence was observed in a Nikon Eclipse Ts2R.

The experiments were performed in triplicates and the p-value was calculated using a two-tailed Student’s t-test with GraphPad Prism (La Jolla, CA, United States). The p-values are expressed as follows: * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001.

A triplicate of 2 μL of 1:1 Cultrex/DMEM-F12 was seeded in a 96-wells plate and in the Mimetas OrganoPlate^® 2-lane 96. After the polymerization, a solution of 10 μM Oregon Green™ 488 Paclitaxel in DMEM/F12 was added. Every 24 h, the solution was completely removed and the fluorescence of the Oregon Green™ 488 Paclitaxel that remained in the supernatant was measured with a BioTek Synergy H1. The p-value was calculated using a two-tailed Student’s t-test with GraphPad Prism (La Jolla, CA, US), and the p-values are expressed as follows: * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001.

Clusters of organoids were mixed in an appropriate volume of Cultrex RGF BME, Type 2 previously diluted 1:1 with DMEM/F12 without phenol red, and 2 μL of this mixture was seeded in 96-wells plates and every chip of the Mimetas OrganoPlate^® 2-lane 96. The viability was measured with the PrestoBlue HS Assay every 24 h for 1 week (168 h) in accordance with manufacturing protocols. The viability was performed in four replicates and the p-value was calculated using a two-tailed Student’s t-test with GraphPad Prism (La Jolla, CA, United States). The p-values are expressed as follows: * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and **** p ≤ 0.0001.

After the viability assays were performed, the organoids were treated with six different concentrations of Carboplatin, ATRA (all-trans Retinoic acid), Paclitaxel (200, 40, 8, 1.6, 0.32, and 0.064 μM), Doxorubicin (50, 10, 2, 0.4, 0.08, and 0.016 μM), and Caelyx (100, 20, 4, 0.8, 0.16, and 0,032 μM). After 96 h, cell viability was measured using three different methods. ATP-based assay: the CellTiter-Glo 3D (Promega, Madison, WI, United States) was used to calculate the viability, and the luminescence was acquired with BioTek Synergy H1. Reducing-based assay: the PrestoBlue HS assay was used to calculate the viability, and the fluorescence was acquired with BioTek Synergy H1. Florescence imaging-based assay: the organoids were stained with 100 μL of 0,1 μg/mL of Hoechst 33,342 (Thermo Fisher Scientific Waltham, Massachusetts, United States) in DMEM/F12 without phenol red and incubated for 20 min. After the incubation, the organoids were washed twice with HBSS and stained with 100 μL of 5 μg/mL of Propidium Iodide (Thermo Fisher Scientific Waltham, Massachusetts, USA) in DMEM/F12 without phenol red and kept for 15 min. The medium was removed, and the organoids were washed twice with HBSS, so 100 μL of DMEM/F12 without phenol red was applied and the fluorescence was observed in a Nikon Eclipse Ts2R. The colocalization of blue areas (nuclei of organoids) and red areas (dead cells of organoids) were evaluated with Pearson’s correlation coefficient using the ImageJ software.

For all three methods, luminescence, fluorescence, and Pearson’s correlation coefficient outcome were used to establish the logistical dose-response curves and calculate the IC₅₀ with GraphPad Prism (La Jolla, CA, US).

As the first step, the experiments were set up to demonstrate the superiority of dynamic versus static cultures with organoids directly derived from HGSOC patients for the first time. Figure 2 shows a flowchart of the designed experiments with the intended goals.

Different research laboratories have shown that microfluidic technology could outperform static conditions due to a better exchange of media, elimination of waste products, and mechanical stimulation of cell lines and primary cells (Legendre et al., 2013). Mimetas OrganoPlate^® 2-lane 96 was utilized with one in-gel culture channel and one passive perfusion channel (Lanz et al., 2017). The plate allows for the testing of multiple drugs with one patient or multiple patients with a few drugs. To better understand in which experimental conditions the organoids reached the “optimum growth rate”, different inclination degrees (7°, 15°, and 20°) and intervals of rocking including 30″, 1′, 8′, 40′ minutes, and continuous rocking were tested. As highlighted by the dotted lines in Figures 3A–C, continuous and short intervals of rocking (30″ and 1′) produced a very fast flow inside the microchannels, which probably generated high friction that destroyed the ECM. With these culture conditions, it is not possible to grow organoids as shown by the small viability values. An optimal condition for ECM and organoids was set up with an interval rocking time of 8′ and 40′ minutes (Figures 3D, E) and a maximum inclination of 7°, 15°, and 20°. In particular, the highest growth rate was reached with a tilting angle of 15° and an interval of rocking every 8 min. It is important to underline that the organoids, cultured in the above-mentioned conditions, grow better compared to a no-rocking condition (Figure 3F).

Cellular viability is mainly influenced by the rate of proliferation or the rate of cell death. In order to assess this point, organoids were stained with Calcein AM (live cells in green) and EthD-1 (dead cells in red) and observed under fluorescence microscopy. As demonstrated in Figure 4A, the microscopic analysis showed that passive flow cultures show fewer red cells and more green cells compared to static cultures. The quantification of the fluorescent signals (number of red or green pixels normalized on the number of blue pixels—Total number of cells) confirms that microfluidic technology could better reproduce the physiological conditions (Figures 4A, B).

The consistency, repeatability, and variability of the technology were evaluated in six HGOC patients. Patient 1 is an ascites derived from HGSOC. Patients 2, 3, and 5 are HGSOC chemonaive tumors, and patient 4 is a primary tumor of a high-grade endometroid ovarian cancer. All the patients were sensitive to platinum-based chemotherapy.

The established organoids captured the histological characteristics of the primary tumours. The immunohistochemistry (IHC) can be accessed as follows: Patients 1 and 4 (Adeel et al., 2021), patients 2 and 3 (paper under evaluation), and patient 5 (Supplementary Figure S3). Hematoxylin and eosin together with PAX8, WT-1, and CA-125 HGOC markers were evaluated by the pathologist. All the tissues were positive for the markers and resembled HGOC. The organoids were also characterized by immunofluorescence after being seeded in the microfluidic chip (Figure 5). PAX8, WT-1, and CA-125 HGOC markers were evaluated, and no differences were observed between static and passive flow culture conditions, indicating that organoids grown in the Mimetas 2-lane Organoplate are still similar to the parental tumors.

The viability of the organoids seeded in the Mimetas 2-lane Organoplate was acquired for up to 7 days by measuring the power of living cells to convert resazurin to fluorescent resorufin using the PrestoBlue HS Assay. Patient 1’s growth was fast in static conditions in the first days of the culture, but after 5 days, the growth rate increased in the passive flow, and at 7 days there were no significative differences despite there being a better growth trend in passive flow conditions, even if it is not statistically significant. After 7 days, patients 2 to 5 showed a better growth rate in passive flow conditions than in static conditions (Figures 6A, B), confirming the preliminary evidence that was obtained during the setup of the culture (see Figure 3).

The effects of the drugs were evaluated with three different methods: the ATP and Resazurin-based method and the propidium iodide-Hoescht cell staining. In particular, the ATP method requires lysing the cellular matrix and cells of the organoids, but this did not always happen correctly due to the narrow capillaries that make up the chip structure, often generating an inadequate output. On the contrary, the Resazurin-based method is not affected by these problems because it is used directly on the culture medium. But like the method mentioned above, it requires great precision in always plating the same number of cells in order to obtain a precise (low SD) measurement of the viability. However, this is not feasible since it is not always possible to count the exact number of cells that make up a cluster of organoids, and it can happen that larger clusters can remain partially stuck in the thin channels of the Mimetas 2-lane Organoplate, making it difficult to dispense the same volume and number of organoids in the chip. In particular, the last assay was developed to overcome these problems, allowing the selection of only organoids, avoiding single cells and contaminants, and the viability is not influenced by the metabolism of the cells but only by the number of live and dead cells. It is possible to make a dose-response curve and obtain IC₅₀ values even from a highly inhomogeneous sample (Supplementary Figures S1, S2). In fact, a variable number and amount (between 2 and 5 µm) of organoids were sown and treated in a serial dilution of Paclitaxel. The propidium iodide-Hoescht cell-staining method allowed us to obtain a value of IC₅₀ with a much smaller standard deviation than the conventional methods (ATP-based and Resazurin-based).

In Figure 7A, an example of PI/DAPI staining with two chemotherapeutic drugs (carboplatin and paclitaxel) is shown. In particular, for each drug concentration and each of the four replicates, TRITC-DAPI colocalization was calculated using Pearson’s coefficient. When Pearson’s Coefficient tends toward one, there is a perfect colocalization between blue (nuclei) and red (dead cell) and it means that there is 0% viability. When Pearson’s coefficient tends toward zero, it means that a lower number of red dots (dead cells) colocalizes with blue dots (nuclei), and it means that there is 100% viability. When Pearson’s coefficient tends toward minus one, it means that there is no colocalization between the red and blue dots, meaning that the paired images are not right and they get reported, which makes them particularly suitable for use in an automated system.

As shown in Figure 7B, each patient responded differently to the drugs, but there was the same consistency between the data obtained in the Mimetas 2-lane Organoplate and those obtained in the static culture. It is interesting to note that the IC₅₀ values are generally smaller in the passive flow culture compared with the static conditions; a graphical representation of this observation is shown in Figure 7C (Jang et al., 2015). Of interest is the analysis of the liposomal formulation of doxorubicin (Caelyx). As a nanodrug, it improves in vivo pharmacokinetics and biodistribution values in the efficacy/toxicity ratio (Gabizon et al., 2003). In vitro, Doxil could permeate the cellular membrane slowly and then free doxorubicin, resulting in less activity, and the IC₅₀ values are higher (Toffoli et al., 2015). Even the ECM represents a barrier for nanodrugs that negatively contribute to their local distribution. In this study, it was demonstrated that the passive flow could improve the activity of Doxil, and on average, the IC₅₀ values are 20 times lower than in the static culture. Similarly, we tested the ATRA, which was recently discovered as a potent inhibitor of the prolyl isomerase Pin1 (Wei et al., 2015) and a target of interest for our laboratory in the therapy of ovarian cancer (Russo Spena et al., 2018). As observed for chemotherapeutic drugs, the activity of this targeted molecule improves under passive flow culture conditions.

To investigate the mechanism of drug distribution in the passive flow microfluidic, the FITC-paclitaxel was utilized, and the fluorescence was measured in the supernatant (Figure 8A) over time.

In the first 24 h, no difference was detected between the static and passive flows. But after 48 h and up to 72 h, the paclitaxel penetrated better in the ECM during passive flow conditions (Huang et al., 2012). At 96 h, again, there was no difference probably due to a saturation effect. The variation in the amount of paclitaxel that remained in the supernatant (and, consequently, the amount that penetrated the ECM) explains the data obtained in Figure 8B. To demonstrate that this effect is important for the efficacy of the drug, the IC ₅₀ values were calculated at different time points. In the static condition, the IC ₅₀ value was calculable only at 96 h. However, in the passive flow condition, the IC ₅₀ value was appreciable after 48 h and decreased at 72 and 96 h; the viability was calculated with the PrestoBlue HS Assay.

To directly visualize the effect of paclitaxel on organoids, Calcein AM (green, live cells) and EthD-1 (read, dead cells) were utilized for staining. As shown in Figure 8C, at 72 h, there was green staining only in static conditions. The red staining became prominent at 96 h. In passive flow conditions, green staining was not detectable but only dead cells. The quantification of green and red pixels was done as explained above and the results are reported in Figure 8D.

Lastly, as shown in Figure 9, FITC-paclitaxel was utilized to treat organoids. As shown in Figure 9A, there was a greater entry of paclitaxel over time and a consequent greater cell death in dynamic conditions compared to static conditions. The amount of FITC-Paclitaxel (green) within the organoids and dead cells (red) was quantified as pixels and normalized on the numbers of the blue pixels (nuclei). Figure 9B shows that there is a correlation between a greater entry of paclitaxel into the organoids (green signal) and an increase in mortality (red signal), which is also confirmed by the viability of the organoids (Figure 9C) that decreases more over time in the dynamic cultures.