Study design and participants

Leukocytes were obtained from the buffy coat blood product material from healthy donors collected by NHS Blood and Transplantation (NHSBT; n = 4, London, UK). Cell-free urine supernatants were obtained from healthy donor (HD; n = 2) or untreated bladder cancer patients (n = 2) (Fidelis Research, Sophia, Bulgaria).

Non muscle invasive bladder cancer (NMIBC) patients (n = 5) receiving Bacillus Calmette-Guérin (BCG) therapy at Lausanne University Hospital (CHUV), Switzerland, were prospectively recruited for this study (Table S1). Due to the exploratory nature of the study, no formal hypothesis on the sample size was made. At the time of inclusion, NMIBC patients were starting standard BCG therapy involving 6-weekly intravesical instillations of OncoTICE® BCG (2 to 8 × 108 colony forming units; Merck Sharp & Dohme AG) in line with European Association of Urology (EAU) guidelines and at the shared decision of both the physicians and patients. Urine samples were self-collected at the Lausanne University Hospital 2–3 h following BCG instillation (second urination) (Table S1). All patients provided written informed consent before taking part.

The study was approved by the Health Research Authority, London—Stanmore Research Ethics Committee (19/LO/0179; 257743) and the Ethics Committee of the Canton de Vaud in Switzerland (#2019–00564).

Isolation of leukocytes from healthy donors and spiking into cell-free urine supernatant

Buffy coat was recovered from the NHSBT bag into a 50 mL tube and centrifuged (800xg, 10 min, room temperature (RT)). The leukocyte layer and plasma were collected and washed with Hanks’ Balanced Salt Solution (HBSS; no Ca2+ , no Mg2 +) by centrifugation (300xg, 10 min, RT). Red blood cells (RBC) were removed by lysis with RBC lysis buffer (Biolegend®). Leukocytes were washed again and re-suspended in Dulbecco’s phosphate buffered saline (D-PBS, Gibco) with 0.5% bovine serum albumin (BSA; Sigma-Aldrich) for cell counting.

Cell-free urine supernatant was thawed at 4°C. Before adding leukocytes, cell-free urine was centrifuged (300xg, 10 min, 4°C), then filtered with a 0.2 μm syringe filter to remove any protein aggregates or cell debris. The leukocytes were spiked into urine cell-free supernatants at three different concentrations (103, 104 or 105 cells/mL of urine supernatant) and analysed by flow cytometry either immediately or after 24, 48 or 72 h at 4℃ (Fig. 1a).

Processing of urine samples from BCG-treated NMIBC patients

Upon collection, patient urine samples were split into two tubes followed by either immediate processing or overnight refrigeration for 24h at 4℃ (Fig. 2a). Both tubes were centrifuged (1500 revolutions per minute (RPM), 5 min) and the resulting cell pellet was counted, stained (as per below) and fixed before flow cytometry acquisition. The urine cell-free supernatant was frozen at -80℃ for batch proteomic analysis (Fig. 2a).

Flow cytometry analysis

Cell-surface antigens were stained for 20 min at 4°C in the staining buffer (PBS with 0.2% BSA and 2 mM ethylenediaminetetraacetic acid [EDTA]) and an amine reactive dye (aqua live/dead stain kit, Life Technologies, Carlsbad, CA, USA) or propidium iodide (PI) (Invitrogen™) was used for dead cell exclusion. Fc-Receptor Blocking Reagent (Miltenyi Biotec) was used to increase staining specificity by blocking non-specific antibody binding.

Spiked leukocytes were stained with anti-human CD45-BV605™ (BioLegend®, mouse IgG1, clone HI30), anti-human CD3-PE (BioLegend®, mouse IgG1, clone UCHT1), anti-human CD14-BV421™ (BioLegend®, mouse IgG1, clone HCD14) and anti-human CD15-AF700 (BioLegend®, mouse IgM, clone HI98).

The following monoclonal antibodies were applied at pre-determined optimal concentrations to stain leukocytes in urine samples from BCG-treated patients: anti-human CD45-PE (eBioscience™, mouse IgG1, clone 2D1), anti-human CD15-PerCP/Cy5.5 (BioLegend®, mouse IgG1, clone W6D3), anti-human CD14-PE-Cy7 (eBioscience™, mouse IgG1, clone 61D3), anti-human CD3-AF700 (BioLegend®, mouse IgG2a, clone HIT3a), anti-human CD4-APC (BioLegend®, mouse IgG2b, clone OKT4) and anti-human CD8-AF488 (BioLegend®, mouse IgG1, clone SK1).

Stained samples were acquired on a Gallios (Beckman Coulter™) or Attune (Thermo Fisher Scientific) flow cytometer followed by analysis on FlowJo™ version 10.8.1 software (FlowJo LLC, Ashland, OR, USA).

Protein quantification

The Quick Start™ Bradford Protein Assay Kit 2 (Bio-Rad) was used to measure total protein in urine according to manufacturer’s instructions. Briefly, urine was diluted 1:10 and 1:50 with sterile D-PBS (Gibco). A standard curve was built using the albumin standard. The 1 × dye reagent provided in the kit was added to each diluted sample and reference standard in a microplate format, incubated at room temperature for not less than 10 min, and absorbances were measured at 595nm with a Spark microplate reader (Tecan).

Creatinine quantification

A creatinine assay kit was used to measure creatinine in urine according to manufacturer’s instructions (Oxford Biomedical Research®). Briefly, urine was diluted 1:10 with sterile distilled water and a standard curve was prepared in a microplate using creatinine. Two sequential absorbance readings were performed at 490 nm using a Spark microplate reader (Tecan) to determine the creatinine concentration (in mg/dL) in the samples. The first reading was performed after adding alkaline picrate solution (initial reading) and a second reading was performed after addition of acid reagent (final reading). The difference in absorbance upon subtracting the final reading from the initial reading for each standard and sample was directly proportional to the creatinine concentration.

Multiplex assay

Multiplex protein immunoassays were conducted using both the 48-plex (Merck Millipore) and 3-plex TGF-beta (Merck Millipore) assay kits according to manufacturer’s instructions. For the 48-plex assay, urine was diluted 1:3 with assay diluent. For the 3-plex TGF-beta assay, urine was acidified with 1N HCl for 10 min and the reaction was stopped by the addition of 1N NaOH. Acidified urine was diluted 1:3 with assay diluent. Samples were read on the Luminex™ xMAP™ INTELLIFLEX system (Merck, Germany). Acquired multiplex data were initially analysed using Belysa® software (Millipore Merck, version 1.0.19). After excluding any values below the limit of detection from analysis and accepting paired samples with a minimum of seven samples, the following analytes were excluded from the analysis: IL-1β, IL-7, IL-17E/IL-25, IL-22 and PDGF-AB/BB. To facilitate comparison between analytes, concentrations were normalised in the range of 0 to 1 using the Python numpy package (Supplementary online content).

Statistical analysis

GraphPad Prism version 10.0.3 (275) for Windows (GraphPad Software, USA) was used for statistical analysis of flow cytometry, total protein and urine creatinine data, as well as for plotting of raw bead array data.

Comparisons between groups for immune cell analysis to distinguish the change in cell frequencies with refrigeration time were conducted using a one-way ANOVA mixed-effects model with Dunnett’s multiple comparison test.

Correlation analysis was performed using the Spearman non-parametric r test with a two-tailed P value. The normality assumption for correlation analysis was assessed via the Shapiro–Wilk normality test.

Group analysis for multiplex assays was conducted using a mixed-effect model (REML) with Sidak multiple comparisons test. For proteomic correlation analysis, the Spearman rho test with two-tailed P test was calculated using the Python stats module of the SciPy library (1.11.1). Correlation for each analyte was done using Jupyter Notebooks (version 6.5.4 executed on Python 3.10.12) with the support of the packages: pandas (2.0.3), scipy (1.11.1), matplotlib (3.7.2) and numpy (1.25.1). The full python code is provided in Supplementary online content.