Patient-derived organoids as a potential model to predict response to PD-1/PD-L1 checkpoint inhibitors

Selection of cancer patients for treatment with immune checkpoint inhibitors remains a challenge due to tumour heterogeneity and variable biomarker detection. PD-L1 expression in 24 surgical chordoma specimen was determined immunohistochemically with antibodies 28-8 and E1L3N. The ability of patient-derived organoids to detect treatment effects of nivolumab was explored by quantitative and qualitative immunofluorescence and FACS analysis. The more sensitive antibody, E1L3N (ROC = 0.896, p = 0.001), was associated with greater tumour diameters (p = 0.014) and detected both tumour cells and infiltrating lymphocytes in 54% of patients, but only 1–15% of their cells. Organoids generated from PD-L1-positive patients contained both tumour cells and PD-1/CD8-positive lymphocytes and responded to nivolumab treatment with marked dose-dependent diameter reductions of up to 50% and increased cell death in both PD-L1-positive and negative organoids. Patient-derived organoids may be valuable to predict individual responses to immunotherapy even in patients with low or no immunohistochemical PD-L1 expression.


BACKGROUND
Immunotherapies targeting the programmed cell death-1 receptor (PD-1) and its ligand-1 (PD-L1) yielded impressive clinical results in advanced cancers expressing high levels of PD-L1. 1,2 However, novel treatments for rare cancers are limited by insufficient patients and trials to establish treatment benefits. This is particularly true for chordomas, rare malignant tumours predominantly located in the spinal axis with a high local recurrence rate (43-85%) and a low tendency for distant metastasis. 3 Chordomas are resistant to chemotherapy (standard treatment: surgery and carbon ion-radiotherapy 4 ), and are candidates for immunotherapy because they express more PD-1/PD-L1 than healthy bone tissues. [5][6][7] Clinical trials evaluated the efficacy of targeting the PD-L1 axis with nivolumab alone or in combination with ipilimumab 8 and a trial combining nivolumab with stereotactic radiosurgery is ongoing (NCT02989636), but no data have yet been published on the correlation of PD-L1 expression and outcomes. In terms of clinical strategy, a selection of cancers/patients sensitive to anti-PD-L1 blockage therapy on the basis of PD-L1 expression in their tumour cells and infiltrating lymphocytes would be highly desirable. Here, we compared PD-L1 recognition by two antibodies, separately assessing expression in tumour cells and tumour-infiltrating lymphocytes of whole surgical specimens, and correlated results with clinical parameters. Because the potential advantages of organoids over cancer cell cultures are increasingly recognised, 9,10 we also generated patient-derived organoids and determined the dose-dependent effects of nivolumab by quantifying diameters, apoptosis, and PD-L1 expression, to establish the potential of this approach for the prediction of treatment responses.

Patients
Twenty four primary chordoma patients treated at the G. Pascale Institute and the G. Pini Institute. Sections of surgical specimens were stained with two monoclonal antibodies to PD-L1, E1L3N and 28-8 (Cell Marque) according to manufacturer's instructions, using BenchMark XT kits and an automatic immunostainer (Ventana Medical Systems). PD-L1-positive cancer cells and lymphocytes were determined as percentages of positive cells in all section, 11  Organoids were fixed and incubated with specific antibodies for 1 h (see Supplement for details). Fluorescence imaging was performed on a LSM 700 confocal microscope (Zeiss, Germany). Twenty six layers (z-projection) of each image were scanned and analysed with FV10-ASW 4.2 software. Results were reported as percentages of DAPI-stained cells. Organoid diameters were determined by ImageJ (National Institutes of Health, Bethesda, MD). Statistical analyses were performed with SPSS v21.0 (Chicago, IL, USA). Significance was set at p < 0.05.

RESULTS
The baseline characteristics of patients are shown in (Supplementary Table 1). The median age of patients was 65 years (range 55-79). The mean follow-up period was 73.0 months, Negative Positive with a minimum of 6 months. At the time of analysis, 3 (12%) patients had died, and median overall survival was 50 months (95% CI: 63.8-98.6). Serial sections of chordoma surgical specimens were assessed with two antibodies, E1L3N and 28-8 ( Supplementary Fig. 1 and Supplementary Tables 1-3). PD-L1-positive tumour cells ranged from 1-15% of all tumour cells per section and were mainly localised at the aggressive margin of the tumour (Fig. 1a, b). Staining with both antibodies was correlated by linear regression (R 2 = 0.68, beta = 0.105) (Fig. 1c), but sensitivity and specificity analysed by ROC statistics was higher for E1L3N than 28.8 (p = 0.001) (Supplementary Fig. 1C). This was most evident in samples with low intensity of PD-L1 staining (p = 0.001). E1L3N yielded several distinct staining patterns ( Supplementary  Fig. 2). PD-L1-positive tumour cells detected by E1L3N correlated with positive tumour-infiltrating lymphocytes (R 2 = 0.62 p = 0.001) ( Fig. 1d; Supplementary Fig. 3 and Supplementary  Tables 2, 3), and with larger tumour sizes (p = 0.014) (Fig. 1e and Supplementary Table 1), but not with overall survival or recurrence rate.
Cells of fresh biopsies isolated by collagenase digestion and analysed by FACS confirmed the percentage of PD-L-1 positivity ( Supplementary Fig. 4). Biopsy cells were then cultured on matrigel for 72 h, generating organoids with diameters of 80-200 μm ( Supplementary Fig. 5A, B). Individual organoids contained both CD8 and PD-1-expressing cells (Fig. 1f). PD-L1positive cells determined by confocal immunofluorescence (up to 20%) varied considerably ( Supplementary Fig. 5C-F), However, 60% of organoids obtained from the biopsy of a PD-L1-positive patient were negative for PD-L1 ( Supplementary Fig. 5D), consistent with the immunohistochemical observation that only some tumour cells express PD-L1.

DISCUSSION
The present results confirm the limitations of detecting PD-L1 by immunohistochemistry to select patient sensitive to nivolumab treatment. Comparison of the two antibodies indicated that E1L3N, the more sensitive one, detected PD-L1 expression in only 54% of spinal chordomas. This is less than the 68.5% reported with a different antibody in tissue arrays, 5 possibly as a result of different chordoma stages or aggressiveness. Tumour sizes were greater in PD-L1-positive patients and its expression in tumour cells correlated with expression in infiltrating lymphocytes. 5,6 This is of clinical interest, but does not provide prognostic information. Our results are consistent with those of clinical trials reporting that PD-L1 alone is of limited use to predict response to treatment of chordomas in individual patients. The efficacy of immunotherapy and lower adverse effects than standard treatments has encouraged cancer trials in unselected populations with highly metastatic cancer sarcoma subtypes. 12 Three-dimensional cell culture are revolutionising the study of human diseases and cancer by permitting analysis of patient-derived tissue with non-invasive procedures. 9,10 The present results provide the first evidence that patientderived chordoma organoids allow to test individual responses to treatment. Hundreds of organoids may be generated from fresh tissue to provide a reasonable approximation of tumour heterogeneity. 10 Pools generated from PD-L1-positive patients containing both PD-L1-positive and negative organoids responded to nivolumab with a significant dose-dependent size reduction within 24 h. This further supports the observation that some sarcomas with low or no immunohistochemically detectable PD-L1 expression respond to therapy.
Limitations of the study include the possibility that the original tumour microenvironment may not have been maintained, and that only a few fresh biopsies were available, due to the rarity of chordomas. Nevertheless, results support the notion that patient-derived spheroids may help to identify subsets of chordoma patients who are likely to respond to immunotherapies, and to compare individual sensitivity to various treatments. They may thus constitute a valuable step towards individually targeted treatment of chordomas and other malignancies.

AUTHOR CONTRIBUTIONS
F.d.N. wrote the manuscript and coordinated the team; A.D.C. and G.P. contributed to the conception and design of the research. R.P. contributed to the interpretation of organoid experiments and revised the manuscript; G.S., F.F., M.G., E.A. and R.C. contributed to patient's data collection and analysis; F.C., L.A. and G.C. contributed to data analysis and interpretation.