Dataset Information

Chemogenomic profiling of breast cancer patient-derived xenografts reveals targetable vulnerabilities for difficult-to-treat tumors

ABSTRACT: Breast cancer is a heterogeneous collection of disease arising from the breast with distinct molecular and phenotypic features. Certain subsets of patients have tumors that are particularly difficult-to-treat, which include triple-negative breast cancer (TNBC), metastatic/recurrent disease and rare histological variants. To delineate the underlying biology and identify therapeutic candidates for these patients, a series of 37 breast cancer patient-derived xenografts (PDX) from both chemo-naïve and pre-treated specimens was generated from 81 transplant attempts. Whole-genome and transcriptome sequencing revealed marked fidelity of the molecular landscape for the majority of PDXs in comparison to parental tumors. Reverse-phase protein array analysis of PDXs further identified potential therapeutic targets. Metastatic potential varied between PDXs, where low-penetrance lung micrometastases was the most common pattern of dissemination and observed in 34.5% of models. Three PDXs recapitulated the metastatic localization seen in the corresponding patients, including two lines with high-frequency metastases to multiple vital organ systems, while another PDX displayed tropism to the skull-base. Chemosensitivity profiling was performed in vivo with standard-of-care agents (doxorubicin, cisplatin, gemcitabine or paclitaxel), where multi-drug chemoresistance was found in 60.0% of PDXs and 64.7% of responses were concordant with pre-engraftment responses observed in the patient. Consolidating chemogenomic data identified potentially actionable features in 97.2% of PDXs, and marked regressions were seen in vivo when a subset of these underwent proof-of-concept functional studies. This included FGFR inhibitor sensitivity in a FGFR1-amplified lobular carcinoma, mTOR inhibitor sensitivity in a recurrent neuroendocrine breast cancer with proteomic evidence of mTOR/PI3K activation, and platinum sensitivity in a TNBC with BRCA1 germline mutation predicted as benign. Together, this clinically-annotated PDX library with comprehensive molecular and phenotypic profiling serves as a resource for both discovery and validation preclinical studies on difficult-to-treat breast tumors.

ORGANISM(S): Homo sapiens

PROVIDER: GSE142767 | GEO | 2019/12/31

REPOSITORIES: GEO

ACCESS DATA

Similar Datasets

Project description:Osteosarcoma (OS) and Ewing’s sarcoma (EW) are the two most common pediatric solid tumors, after brain tumors. Multimodal treatments have significantly improved prognosis in localized disease but outcome is still poor in metastatic patients, for whom therapeutic options are often inadequate. Preclinical drug testing to identify promising treatment options that match the molecular make-up of these tumors is hampered by the lack of appropriate and molecularly well-characterized patient-derived models. To address this need, a panel of patient-derived xenografts (PDX) was established by subcutaneous implantation of fresh, surgically resected OS and EW tumors in NSG mice. Tumors were re-transplanted to next mice generations and fragments were collected for histopathological and molecular characterization. A model was considered established after observing stable histological and molecular features for at least three passages. To evaluate the similarity of the model with primary tumor, we performed a global gene expression profiling and tissue microarrays (TMA), to assess tumor specific biomarkers on tissues from OS/EW tumors and their PDXs (1st and 3rd passage). Moreover, we verified the feasibility of these models for preclinical drug testing. We implanted 61 OS and 29 EW samples: 14/38 (37%) primary OS and 9/23 (39%) OS lung metastases successfully engrafted; while among EW, 5/26 (19%) primary samples and 1/3 (33%) metastases were established. Comparison between patient samples and PDXs, highlighted that histology and genetic characteristics of PDXs were stable and maintained over passages. In particular, correlative analysis between OS and EW samples and their PDXs was extremely high (Pearson’s r range r=0.94-0.96), while patient-derived primary cultures displayed reduced correlation with human samples (r=0.90-0.93), indicating that in vitro adaptation superimpose molecular alterations that create genetic diversion from original tumors. No significant differentially expressed gene profile was observed from the comparison between EW samples and PDXs (fold change > 2, adjusted p <0.05 at paired t-test). In OS, the comparison between OS patient-derived tumors and PDX indicated differences in 397 genes, mostly belonging to immune system functional category. This is in line with the idea that human immune cells are gradually replaced by murine counterparts upon engraftment in the mouse. As proof-of concept, two EW PDX and one OS PDX have been treated with conventional and innovated drugs to test their value in terms of drug-sensitivity prediction. Overall, our study indicated that PDX models maintained the histological and genetic markers of the tumor samples and represent reliable models to test sensitivity to novel drug associations.

Project description:Triple negative breast cancer (TNBC) is an aggressive subtype of breast cancer, with approximately 50% of patients having residual disease after neoadjuvant chemotherapy (NACT) and worse prognosis with higher residual cancer burden (RCB). Patient-derived xenograft (PDX) models of breast cancer are an effective discovery platform and tool for preclinical pharmacologic testing and biomarker identification. There is a need for PDX models encompassing the heterogeneity of TNBC both pre- and post- therapy and for identifying clinical and molecular features of patient’s tumors that are associated with success in establishing these models. Using fine-needle aspirates (FNAs), we established orthotopic PDX models of TNBC from the primary breast tumors of patients prior to and following neoadjuvant systemic therapy while enrolled in the ARTEMIS trial (NCT02276443). ARTEMIS is a prospective neoadjuvant clinical trial for women with primary TNBC who are treated with four cycles of Adriamycin combined with cyclophosphamide (AC) followed by taxane +/- different experimental therapies. Serial biopsies were obtained from patients prior to treatment (pre-therapy), from poorly responsive disease after four cycles of AC (mid-NACT), and in cases of AC-resistance, after a three-month course of different experimental therapies and/or additional chemotherapy (post-NACT). Our study cohort includes a total of 269 FNA biopsies from 217 women, generating a total of 62 successful PDX models (overall success-rate = 23%). This cohort includes five serial pre- and mid-NACT PDX pairs, as well as one serial pre-, mid-, post-NACT triplet. Success of PDX engraftment was generally higher from cancers that proved to be treatment-resistant: whether poor response to AC determined by ultrasound measurements mid-NACT (p=0.063), RCB II/III status after NACT (p=0.046), or metastatic relapse within two years of surgery (p=0.008). TNBC molecular subtype determined from gene expression microarrays of pre-NACT tumor revealed no significant association with PDX engraftment rate (p=0.877). Finally, we developed a statistical model predictive of PDX engraftment using percent Ki67 positive cells in the patient’s diagnostic biopsy, positive lymph node status at diagnosis, and low volumetric reduction of the patient’s tumor following AC treatment. This novel bank of 62 PDX models of TNBC provides a valuable resource for biomarker discovery and preclinical therapeutic trials aimed at improving neoadjuvant response rates for patients with TNBC.

Dataset Information

Chemogenomic profiling of breast cancer patient-derived xenografts reveals targetable vulnerabilities for difficult-to-treat tumors

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure