Surgically resected tumor samples (kidney and bladder cancer) were retrieved from the Mayo Clinic operating rooms by dedicated laboratory assistants immediately upon removal from the patient. The sample was rapidly processed for clinical care and tumor in excess of diagnosis was collected for this study. Tumor samples were cryopreserved by dicing into 1 mm cubes placing them in 1.5mL cryovials and suspending them in 1.0 mL CryoStor media (07955; Stemcell) per tube. The vials were placed in freezing containers (5100-0001; Thermo Fisher Scientific), and cooled the cells at a rate of -1°C/min until reaching -80°C. After overnight incubation at -80°C, the cryopreserved tumors were stored in liquid nitrogen for later use.

The cryopreserved tumors were mechanically and enzymatically dissociated using the Miltenyi Biotec tumor dissociation kit (#130-095-929) and the GentleMACS ™ Dissociator (#130-093-235) based on the manufacturer’s protocol. To generate the multicellular microcancer (immunotumoroid), 1.5 x 10 ⁴ cells/well were loaded on a 96-well GravityPLUS Hanging Drop plate (InSphero, Schlieren, Switzerland) and incubated at 37°C. Cells were seeded at 1.5 x 10 ⁴ cells/well to increase the inclusion of non-tumor cells, such as stromal cells and tumor-infiltrating lymphocytes. After 4 days of incubation (day -4), the aggregated microcancers were dropped into a 96-Well clear round bottom ultra-low attachment (ULA) plate (#7007 Corning) and cultured in DMEM/F12 media supplemented with 10% heat-inactivated human AB sera, primocin, glutamax, hepes, N-Acetyl-L-cysteine, Nicotinamide, B-27 Supplement, A83-01, insulin, EGF, and the therapeutic agents. The total volume of media per well was 240 μL. Cultures were incubated in a humidified atmosphere of 5% CO2 at 37°C.

Initial cell viability was measured on day 0, after dropping the immunotumoroid but before drug dosing, and then evaluated at later time points by Incucyte Cytotox Red staining (#4632 Sartorius). Incucyte Cytotox Red is a highly sensitive nucleic acid dye for real-time quantification of cell death. As cell health begins to decline, the integrity of the plasma membrane is compromised, allowing the dye to enter the cell. Fluorescence increases 100-1000-fold upon binding to DNA. For each drug, we applied 5 log concentrations and just the C_max concentration indicated in the main figures of the paper. The response to the other drug concentrations can be found in the Supplementary Figures . The C_max of the drugs was obtained from relevant publications (7) and also the Micromedex website (https://www.micromedexsolutions.com/). The microcancer response to treatment was recorded by the Incucyte S3 Live-Cell Analysis Instrument (Sartorius, Germany) using the Incucyte Spheroid Analysis Software Module (# 9600-0019).

Cisplatin (Fresenius Kabi USA, LLC), gemcitabine (Athenex Pharmaceutical Division, LLC), and paclitaxel (Athenex Pharmaceutical Division, LLC) were used as chemotherapy drugs. For ICI drugs, pembrolizumab (Merck Sharp & Dohme LLC), nivolumab (E.R. Squibb & Sons, LLC), and durvalumab (AstraZeneca Pharmaceuticals LP) were used.

Immunohistochemical staining of 5 μm thick paraffin-embedded tissue sections was performed on an automated BenchMark ULTRA IHC/ISH Staining Module (Ventana Medical Systems, AZ) as follows. CD3 clone LN10 at 1/250 dilution, pretreatment with CC1 (Catalog # 950-224) 32 min at 100C, antibody for 32min at 36C, Optiview DAB detection (Catalog # 760-700). CD4 clone SP35, predilute, pretreatment with CC1 36 min at 95C, antibody for 32min at 36C, Ultraview DAB detection (Catalog # of 760-500). CD8 clone C8/144B at 1/250 dilution, pretreatment with CC1 32 min at 100C, antibody for 16 min at 36C, Optiview DAB detection. CD68 clone PG-M1 at 1/75 dilution with CC1 36 min at 95C, antibody for 32 min at 36C, Ultraview DAB detection. Vimentin clone V9, prediluted, pretreatment with CC1 32 min at 100C, antibody for 16 min at 36C, Optiview DAB detection. PD-L1 clone 22C3 at 1/50 dilution, pretreatment with CC1 64 min at 100C, antibody for 24min at 36C, Optiview +Amp DAB detection (Catalog # 760-099). PD-1 clone NAT105 at 1/300 dilution, pretreatment with CC1 32 min at 100C, antibody for 32 min at 36C, Optiview +Amp DAB detection.

Statistical analyses were performed using GraphPad Prism. The unpaired t test was used to determine the statistical significance between the two groups.

This study evaluated the anti-tumor immune response of TILs, recognizing that the immunotumoroid model lacks a complete immune system and that the TILs’ anti-tumor cytotoxicity may be delayed in comparison to T cells in vivo. Additionally, most of the T cells in the TME have an exhausted and dysfunctional phenotype, and only a fraction are reactivated by the ICI treatment (8). Due to the comparatively low immune cell diversity and lack of systemic factors in an immunotumoroid model versus the in vivo TME, reactivation of exhausted cytotoxic T cells and the manifestation of detectable cytotoxicity likely requires some time. We therefore designed our immunotumoroid model to survive extended periods by optimizing the volume and components of the culture medium to support long-term survival (see Methods). Figure 1 depicts a bladder immunotumoroid at multiple time points from day 0 until day 35. The tumor cell suspension is seeded on GravityPlus hanging drop plates on day -4 and the immunotumoroids are dropped into ULA plates on day 0. Even after 5 weeks of incubation without changing the media, the immunotumoroids are healthy with no sign of necrosis and apoptosis compared to the initial time points. By exchanging culture media, the immunotumoroids survived for over 65 days. This long-time survival was exhibited by all 5 cases that successfully formed immunotumoroids in this study, supporting the model’s suitability for ICI response evaluation.

Among 13 collected surgically resected tumor samples, 6 of them had extensive necrosis that prevented the formation of the tumor spheroid (immunotumoroid) in the hanging drop system. Immune checkpoint inhibitors (ICIs) are the blocking antibodies and do not have any cytotoxic effect on tumor cells by themselves; by blocking checkpoint proteins from binding with their ligands ICIs prevent the “off” signal from being sent, allowing the tumor-specific T cells to kill the tumor cells (9). Therefore, the prerequisite of any model to evaluate ICI response is the inclusion of tumor-specific T cells known as tumor infiltrated lymphocytes (TILs) (10–12). However, in our hands, an excess of TILs can prevent the formation of tumor spheroid. One of the collected tumor samples had ~ 40% TILs in the dissociated tumor cell suspension which was sufficient to prevent the formation of immunotumoroid.

The IHC staining of another tumor sample that successfully made tumor spheroid revealed less than 5% TILs in the primary tumor sample which made it ineligible as an immunotumoroid and was excluded from the study.

A model for ICI response evaluation must include immune and non-immune cells of TME that play crucial roles in the response to immunotherapy. Case 1 is a renal tumor with a high T cell (CD3⁺) infiltration. Cytotoxic CD8⁺ and helper/regulatory CD4⁺ T cells were detected by IHC, and the tumor shows high PD-1 and PD-L1 expression ( Figure 2 ). A portion of the tumor was processed and cultured as immunotumoroids before IHC staining which shows that the resulting immunotumoroids retained the different subsets of T cells required for anti-tumor and ICI response evaluation, as well as PD-1 and PD-L1 expression ( Figure 3 ).

The complex interactions between cellular and chemical components of the TME support tumor growth, development, and metastasis; fibroblasts (vimentin) and macrophages (CD68) are essential contributors to tumor development and also contribute to immunotherapy resistance (13–28). Tumor-associated macrophages (TAMs) are the most heterogenous immune cells in the TME, are critical for tumor growth (18, 19), and play a major role in ICI resistance (25–28). Secretion of epidermal growth factor (EGF), platelet-derived growth factor (PDGF), IL-6, IL-8, and IL-10 and activation of the Wnt/β-catenin signaling pathway by TAMs can directly induce tumor cell proliferation (29, 30), while secretion of metalloproteinases (MMPs), growth factors, and angiopoietin-1 promotes angiogenesis to provide nutrient and oxygen support (31, 32). Additionally, TAMs can decrease ICI efficacy by 1) directly blocking CD8⁺ T cells’ anti-tumor functions and tumor infiltration ability (25, 27, 33), 2) expressing alternate immune checkpoints (i.e., VISTA) or 3) sequestering ICI antibodies (25, 27, 34–36). The IHC staining confirms that the immunotumoroid model retained the TAMs from the primary tumor as presented in Figure 3 by CD68 staining.

Cancer-associated fibroblasts (CAFs) are an essential component of the tumor stroma, with a major role in the physical support of tumor cells and enhancement of tumorigenesis (13–17). CAFs secrete various matrix proteins that play an essential role in the remodeling of tumor ECM (37, 38). Transforming growth factor-β (TGF-β) secreted by CAFs limits T cell infiltration into the tumor (39) while also downregulating the expression of MHC-II, CD80, and CD40 on the surface of dendritic cells in the TME, suppressing anti-tumor immunity (40). Additionally, CAFs promote differentiation and recruitment of Tregs, which (41) inhibit CD8⁺ T cells function and promote T cell exhaustion (42, 43). The Case 1 immunotumoroid model shows the presence of CAFs and TAMs as detected by IHC. Taken together, this data shows that the immunotumoroid model retains the cellularity and the immune checkpoint expression profile of the original tumor ( Figures 2 , 3 ).

To test the model’s functionality and sensitivity to therapeutic agents, the immunotumoroids were treated with chemotherapy drugs cisplatin, gemcitabine, and paclitaxel. Cell killing was quantified by Cytotox Red staining; preexisting dead cells are normalized to untreated controls, and autofluorescence is normalized through untreated-unstained controls.

The response of renal (Case 1) and bladder (Case 2) immunotumoroids to chemotherapy agents at different time points are shown in Figure 4 . Case 1 shows sensitivity to gemcitabine, cisplatin, and paclitaxel ( Figures 4A–C ) compared to the non-treated controls; interestingly, the sensitivity to cisplatin becomes apparent at a later time point and was not detectable on day 6. These experiments were performed on 5 log concentrations and only the C_max concentrations are presented here (see Supplementary Figures for additional drug concentrations). Case 2 was also sensitive to gemcitabine and paclitaxel, but resistant to cisplatin ( Figures 4D–F ).

The immunotumoroid drug screen results were compared to the patient clinical response to the drug when available. The patient in Case 3 responded to paclitaxel and later responded to gemcitabine. The drug screen results for Case 3 align with this clinical response; Figure 5 shows that the immunotumoroid model responded well to both paclitaxel and gemcitabine beginning at day 7. Therefore, the immunotumoroid response to chemotherapeutic agents reflects the patient’s clinical response to these drugs. The patients of Cases 1, 4, and 5 did not receive any chemotherapy drug in their course of treatment. The patient from Case 2 had 1 cycle of gemcitabine and cisplatin but was not able to tolerate this due to renal impairment.

After confirming the immunotumoroid model’s suitability for chemotherapeutic drug screening, we evaluated the model’s response to ICI therapy using PD-1 and PD-L1 inhibitors pembrolizumab, nivolumab, and durvalumab. Pembrolizumab and nivolumab both target PD-1, but we tested both drugs to account for the possibility of different outcomes due to structural differences between the antibodies (44). Nivolumab and durvalumab were used in combination to evaluate possible synergistic effects observed in some cases from targeting both PD-1 and PD-L1 (45).

Pembrolizumab and nivolumab/durvalumab combination therapy induced cytotoxicity in Case 1 (renal) immunotumoroids ( Figures 6A, B ). The response was apparent only at later time points (day 13 and day 20), which validates the need for long-term incubation of immunotumoroids in the context of ICI screening. Case 4 (bladder) immunotumoroids were resistant to both treatments compared to the untreated control ( Figures 6C, D ) Interestingly, Case 4 immunotumoroid viability increased after ICI treatment, which could be due to the hyperprogression phenomenon (46–51).

The patients in Case 3 and Case 5 have been treated with ICIs with different outcomes. Compared to the negative control, Case 5 immunotumoroids show sensitivity to both pembrolizumab and the combination of nivolumab and durvalumab ( Figures 7A, B ). This patient was diagnosed with clear cell renal cell carcinoma and underwent a year of pembrolizumab treatment after radical nephrectomy, and currently shows no evidence of disease. This clinical data correlates with the results of Case 5 immunotumoroid treatment with ICI.

Neither Pembrolizumab nor nivolumab/durvalumab therapies induced cell killing in Case 3 immunotumoroids ( Figures 7C, D ). Interestingly, immunotumoroid viability improved after ICI treatment compared to the untreated controls. The Case 3 patient had been diagnosed with renal carcinoma and progressed after receiving 5 doses of nivolumab. Therefore, the clinical response of Case 3 correlates with the immunotumoroid response to ICIs.

Clinical data and treatment outcomes of the cases are presented as a table in the Supplementary Data.