A biobank of pediatric patient-derived-xenograft models in cancer precision medicine trial MAPPYACTS for relapsed and refractory tumors

Abstract

Pediatric patients with recurrent and refractory cancers are in most need for new treatments. This study developed patient-derived-xenograft (PDX) models within the European MAPPYACTS cancer precision medicine trial (NCT02613962). To date, 131 PDX models were established following heterotopical and/or orthotopical implantation in immunocompromised mice: 76 sarcomas, 25 other solid tumors, 12 central nervous system tumors, 15 acute leukemias, and 3 lymphomas. PDX establishment rate was 43%. Histology, whole exome and RNA sequencing revealed a high concordance with the primary patient's tumor profile, human leukocyte-antigen characteristics and specific metabolic pathway signatures. A detailed patient molecular characterization, including specific mutations prioritized in the clinical molecular tumor boards are provided. Ninety models were shared with the IMI2 ITCC Pediatric Preclinical Proof-of-concept Platform (IMI2 ITCC-P4) for further exploitation. This PDX biobank of unique recurrent childhood cancers provides an essential support for basic and translational research and treatments development in advanced pediatric malignancies.

Fig. 1. Workflow for PDX development in MAPPYACTS and established models.

a 2–5 mm³ solid tumor fragments were collected from resection or biopsy samples and transplanted (subcutaneously (SC) or orthotopically, see Supplementary Table S1 and Supplementary Data 1 for details) in immunocompromised mice (NSG, NSG-IL, SCID-beige, Swiss nude). For central nervous system (CNS) tumors, samples were digested and cell homogenates were injected subcutaneously and/or intracerebrally (IC) into the caudate nucleus, cerebellum or pons of nude mice. To generate leukemia PDX, 0.05–1 × 10⁶ cells from bone marrow or peripheral blood were injected into sub-lethally irradiated (2.5 Gy) NSG mice by intrafemoral injection. Tumor take was monitored during 6–18 months or 1 year for leukemias. Model characterization was performed using morphological and molecular analyses from passages (P1 to P4) and validated by comparing it to the patient’s tumor. b Plot represents numbers of tumor samples transplanted (left bars), samples with first tumor take in P0 (middle bars) and established models in ≥P2 or P1 in some leukemias (right bars). c Graph shows numbers of established models per histological tumor type and the research center in which they were established. B-ALL/T-ALL acute lymphoblastic leukemia, AML acute myeloid leukemia, ALCL anaplastic large cell lymphoma, HD Hodgkin lymphoma, BL Burkitt lymphoma, NHL non-Hodgkin lymphoma, OS osteosarcoma, EWS Ewing sarcoma, RMS rhabdomyosarcoma, RT rhabdoid tumor, BCOR/CIC BCOR or CIC-translocated sarcoma, SS synovial sarcoma, MPNST malignant peripheral nerve sheath tumor, DSRCT desmoplastic small round cell tumor, ES epithelioid sarcoma, US undifferentiated sarcoma, RS renal sarcoma, NB neuroblastoma, NPB nephroblastoma, CA carcinoma, HB hepatoblastoma, PPB pleuropulmonary blastoma, MM melanoma, HGG high-grade glioma, LGG low grade glioma, MB medulloblastoma, EP ependymoma, ATRT Atypical teratoid rhabdoid tumor, CEA Commissariat à l'énergie atomique et aux énergies alternatives, CLB Centre Léon Bérard, NAN CHU Nantes, IC Institut Curie, STR CHU Strasbourg, UL Nancy Lorraine University, GR Gustave Roussy, XEN Xentech, VHIR Vall d’Hebron Research Institute, INTMI Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy.

Fig. 2. PDX tumor growth.

Tumor growth of established subcutaneous alveolar, embryonal or other rhabdomyosarcoma, Ewing sarcoma, other soft tissue sarcoma, neuroblastoma, and nephroblastoma PDX. Tumor volumes depicted were obtained from 2 to 14 animals per PDX model. Tx: transplantation; * - eRMS, ** - other RMS. Error bars represent ± standard error of mean (SEM).

Fig. 3. Comparative histological study between primary and PDX tumors.

Concordant cases: all PDX tumors have a suggestive morphological appearance, as seen in the case of synovial sarcoma (SS), identical to the primary, and that of neuroblastoma (NB), with neuropil (*) and the presence of large cells with eccentric nuclei (arrowhead). Note the striking similarities in some cases. In the case of osteosarcoma (OS), the PDX tumor retains the osteoid formation (*) and the presence of giant cells (arrows) observed in the primary; in the case of glioblastoma (HGG), the giant cells (arrows) observed in the primary are visible also in the PDX tumor. In the case of adenocarcinoma (ADC), only a bone biopsy was available for comparison: this likely explains the morphological differences observed between the primary and the PDX tumor, in which, however, the features are highly suggestive that the diagnosis of adenocarcinoma cannot be missed. Discordant cases: the examples illustrate the three main causes of discrepancies between the initial diagnosis (IDg) and the PDX-based tentative morphological diagnosis (TDg). In this case of rhabdomyosarcoma (RMS) suggestive histological features are absent and the PDX tumor looks morphologically undifferentiated (undif.). In this case of nephroblastoma (NBL), the PDX tumor is made only by “blastic” cells and lacks the distinctive epithelial structures present in the primary. In pleuropulmonary blastoma (PPB), the correct diagnosis cannot be achieved by morphology alone. Hematoxylin-eosin-saffron staining; scale bar = 200 µm. Additional abbreviations: EWS Ewing sarcoma, ALCL anaplastic lymphoma, BL Burkitt’s lymphoma.

Fig. 4. Pangenomic profile and comparison of WES and RNA-Seq between matched primary patient tumors and PDX model.

a OncoPrint presents tumor specific alterations and genes considered as actionable or of interest in 39 primary tumor sample and PDX, respectively. b Expression of human genes cluster PDX together with primary tumor histology types. c Microenvironment of murine species in patient and PDX samples that is further specified in Supplementary Fig. S3.

Fig. 5. HLA class I alleles detected in PDX models from pediatric solid cancers.

For each tumor type and molecular subtype, the numbers of alleles in PDX models harboring the corresponding HLA-A, -B and -C alleles are indicated within the boxes (1–11). HLA alleles were grouped according to their supertype-specific binding motifs, unless unclassified (denoted with “NA”), which are indicated at each HLA gene panel, and sorted by descending order of allele frequencies reported from EUR control individuals. Alleles detected only in PDXs derived from non-European (non-EUR) individuals are shown in gray boxes.

Fig. 6. Metabolic functional signatures of 42 PDX models depicted on the background of the whole MAPPYACTS patient population.

a Metabolic functional analysis of PDX and patient tumors using Cellfie, which evaluates the capacity of cancers to carry out 195 metabolic tasks. b Global comparison of PDXs and patients metabolic function using UMAP and correlation distribution between matched and random patient-PDX pairs (Student t test). c Tyrosine to dopamine metabolic task and its binary score. Displayed are matched patient-PDX pairs ordered and colored in green if task is active. d Tyrosine to dopamine task score for all cancer types in patients (circles) with PDX models (triangles). e Synthesis of L-kynurenine from tryptophan task score of patients and PDX models. B-ALL/T-ALL acute lymphoblastic leukemia, AML acute myeloid leukemia, ALCL anaplastic large cell lymphoma, ATRT Atypical teratoid rhabdoid tumor, BCOR/CIC BCOR or CIC-translocated sarcoma, CA carcinoma, CNS central nervous system tumor, CPC choroid plexus carcinoma, EP ependymoma, EWS Ewing sarcoma, HB hepatoblastoma, HGG high-grade glioma, LGG low grade glioma, MB medulloblastoma, NBL neuroblastoma, NPB nephroblastoma, NHL non-Hodgkin lymphoma, NRSTS non-rhabdomyosarcoma soft tissue sarcoma, OS osteosarcoma, PNET primary neuroectodermal tumor, RMS rhabdomyosarcoma, RT rhabdoid tumor, SC sarcoma, ST solid tumor.