The Public Repository of Xenografts Enables Discovery and Randomized Phase II-like Trials in Mice

Abstract

More than 90% of drugs with preclinical activity fail in human trials, largely due to insufficient efficacy. We hypothesized that adequately powered trials of patient-derived xenografts (PDX) in mice could efficiently define therapeutic activity across heterogeneous tumors. To address this hypothesis, we established a large, publicly available repository of well-characterized leukemia and lymphoma PDXs that undergo orthotopic engraftment, called the Public Repository of Xenografts (PRoXe). PRoXe includes all de-identified information relevant to the primary specimens and the PDXs derived from them. Using this repository, we demonstrate that large studies of acute leukemia PDXs that mimic human randomized clinical trials can characterize drug efficacy and generate transcriptional, functional, and proteomic biomarkers in both treatment-naive and relapsed/refractory disease.

Figure 1. Integrative analysis of leukemia and lymphoma PDXs

(A) Unsupervised hierarchical clustering over expression of 1000 genes with the greatest variance-to-mean ratios among 107 PDXs representing all WHO diagnostic categories encompassed by our repository. (B) Key clinical characteristics of patients and their tumors from which PDXs were derived. Patient age in years reflects the time when the xenografted tumor specimen was obtained. Phases of treatment are defined as: Untreated, prior to therapy directed at the xenografted tumor (n.b., does not include therapy directed at prior malignancies); Primary refractory, failed to respond to all tumor-directed therapy to date; Relapse, recurred by standard disease-specific criteria after achievement of a complete remission; Refractory, disease that is progressing during or shortly after the administration of salvage therapy for relapsed disease; Progression, specific to lymphomas progressing by clinical or radiographic criteria after a period of stable disease or partial remission. (C) Binary matrix of prior therapies to which patient was exposed prior to sampling of the referenced tumor, if known. (D) Selected cytogenetic features of patient tumors from which PDXs were derived. (E) OncoPrint of selected mutations detected in PDXs by targeted exon sequencing of a panel of 205 genes (Odejide et al., 2014). See also Figure S2 and Tables S1–S3.

Figure 2. Integrative analysis of B-ALL PDXs, primary samples and cell lines

(A) Unsupervised hierarchical clustering of RNA expression profiles among 60 B-ALL PDXs (PDX cohort), 19 primary pre-B ALL samples (SRP058414), and 10 B-cell leukemia cell lines (CCLE), using the same methods as for Figure 1. (B) Key clinical characteristics of cell lines and primary samples, including those from which PDXs were derived. (C) Binary matrix of prior therapies to which patient was exposed prior to sampling of the referenced tumor, if known. (D) Selected cytogenetic features of cell lines and primary samples, including those from which PDXs were derived. (E) OncoPrint of selected mutations detected in PDXs by targeted exon sequencing, or reported in cell lines by ATCC or DSMZ. Mutation data for the primary samples (SRP058414) were not available. See also Table S4.

Figure 3. The MDM2 inhibitor CGM097 extends survival of mice engrafted with TP53-wild-type B-ALL PDXs in a randomized phase II-like trial

(A) Kaplan-Meier survival analysis for overall survival of mice engrafted with 8 B-ALL PDXs, with 3 animals per treatment arm or 1 randomly selected animal from each arm. (B) Kaplan-Meier survival analysis for overall survival of vehicle- and CGM097 treated cohorts of mice engrafted with TP53 wild-type (n=20) and TP53 mutant (n=4) B-ALL PDXs. (C) Time to sacrifice after start of treatment for vehicle and CGM097 treated animals for each PDX. (D) Kaplan-Meier analysis of PDXs derived from patients who had relapsed or progressed on any form of therapy (relapsed/progressed) versus patients who had not received treatment (untreated). See also Figure S3.

Figure 4. Transcriptional, protein-based and functional biomarkers for response and resistance to CGM097 in B-ALL PDXs

(A) Agglomerative clustering by gene expression of CGM097-induced genes in each treatment group. Samples from censored animals indicated by ^. PDX samples with TP53 mutations are indicated by *. Samples used for immunoblotting in Figure 4F are indicated with lower-case letters (a–g). (B) Immunoblotting performed on purified splenic B-ALL cells at the 26 hour timepoint. Transcript levels of CDKN1A (encodes p21) were determined on the same samples. (C) Immunohistochemistry for p53 and p21 in vehicle and CGM097-treated spleens collected on day 6 of treatment. Scale bars depict 0.1mm. (D) Δ% priming in 26 hour vehicle- and CGM097-treated groups was determined by dynamic BH3 profiling using a PUMA BH3 peptide in 11 PDXs. The 4 “non-responder” models harbor TP53 mutations and had no improvement in survival between CGM097- and vehicle-treated mice. Error bars represent standard error of the mean (S.E.M.) (E) R.O.C. curve analysis for dynamic BH3 profiling using the PUMA BH3 peptide. (F) Immunoblotting for p53 and p21 in CGM097- and vehicle-treated splenic B-ALL cells collected at the SUR timepoint. See also Figure S4, Tables S5–S7.