Key Accomplishments

  1. Eighty-one new PDX models of solid tumors have been established as xenografts in NSG mice. Germline sequencing has been completed for most samples. Models have been derived from predominantly Hispanic patients and thus represent a unique cohort for preclinical drug development. Hepatoblastoma models (n=15) have been incorporated into the NCI/PIVOT pediatric in vivo testing program. It is proposed to incorporate many of the solid tumor PDX models into the European ITCC-P4 program to develop new drugs for pediatric cancer. Molecular characterization can be obtained from the PCAT portal at GCCRI: The portal can be accessed at:
  2. We have developed 79 new patient-derived xenograft models of childhood leukemia, 63% from Hispanic patients with poorer outcomes than White children. Notable is that germline sequencing was conducted on 50 samples allowing unambiguous identification of somatic mutations and patient race/ethnicity. PDX models represent high- and low-risk Pre B precursor ALL, T-cell ALL, and AML. It is proposed to incorporate many leukemia models into the European ITCC-P4 program to develop new drugs for pediatric cancer. Patient demographics for all PDX models and molecular characterization of leukemia PDX models can be accessed at:
  3. We have sequenced germline, patient tumor, and its respective PDX model for over 150 models, thus representing a unique cohort of pediatric preclinical models. In each case, whole exome or whole genome sequencing was completed on germline, patient tumor, and PDX. For RNA sequencing, patient tumor and PDX were analyzed. For 10 leukemia models, DNA methylation profiles were obtained from the patient tumor and PDX. Of the 82 successfully engrafted leukemic PDXs, 51/82 (62%) were obtained from patients of Hispanic ethnicity. PDX models faithfully reflected somatic mutations, gene copy-number alterations, RNA expression, gene fusions, whole-genome methylation patterns, and immunophenotypes found in PT samples.70 solid tumor PDX models have been fully characterized, with analysis of 11 still ongoing. We show that PDXs generally recapitulate the genomic and transcriptomic features of the matched PTs. However, 20-30% of the PDXs show evidence of clonal selection. The clonal selection was seeded by a PT subclone that can be either closely or distantly related to the major clone. The patient tumors that underwent clonal selection also showed increased antitumor immunity. These data collectively suggest more complex, immunosuppressed tumors may be prone to clonal shifts when engrafted into immunodeficient mice.
  4. A database with a web portal, “Pediatric PDX Explorer,” has been developed and deployed to the public domain at The portal allows users to view all of the PDX models that have been established and explore the associated patient demographic and clinical, molecular, and pathologic characteristics. Currently, the database hosts data from 193 patients across four institutions from whom PDX models have been established. Users can filter and identify the patients/PDXs of interest based on various demographic, clinical, sample, genomic, and imaging variables (24 variable filters in total). The associated data and metadata of biospecimen samples (n = 833), PDX samples (n = 155), and pathology images (n = 1214) are also made accessible to registered users. A dedicated team manages and updates them in real-time using the built-in Sample Management module. The links between patients, biospecimen-generating procedures, derived samples, and pathology images are established. High-resolution pathology images can be visualized through a built-in image viewer. The portal also provides online visualization and analysis of the genomic data (WES, WGS, RNAseq, and DNA methylation). Users can request samples of interest and access the project-generated genomic data through the portal.

Request for PDX samples (Academic Research):
Please complete the attached file and return it to Upon return, our office will send an MTA to be signed by an appropriate official at your institution.

Download MTA-Outgoing-Form

The Texas Pediatric Testing Core (TPTC)

Despite significant advances, long-term disease control in many children remains elusive; improved survival has been largely driven by dose escalation and the introduction of new agents. However, these therapies are coupled with long-term life-threatening sequelae in many survivors.  Increased understanding of the molecular pathology for many childhood cancers and the growing awareness that ‘homogeneous’ diseases are, in fact, quite heterogeneous at a molecular level provides the basis for the identification of drugs/biologics targeted at specific molecular drivers and the development of effective therapeutics. This Core will provide a service for testing new agents/combinations against a large panel (>340) of molecularly characterized Patient-Derived Xenografts (PDX’s; childhood cancers grown in mice) representing both childhood and adolescent/young adult hematologic and solid tumors. The Core will facilitate pediatric cancer drug development within the context of the Research to Accelerate Cures and Equity for Children Act. The Act requires the Food and Drug Administration (FDA) to develop a list of known and new drugs/biologics molecular targets. If agents are determined to be substantially relevant to the growth and progression of pediatric cancer, this may trigger the requirement for pediatric investigations by both Pharma and academic centers. Unique aspects of this Core are:

  • Bioinformatics-driven selection of molecularly appropriate tumor models;
  • PDX models that encompass molecular heterogeneity of a cancer type;
  • Testing approaches that facilitate incorporation of clinical heterogeneity.

Our strengths include the extensive experience of the Core Leaders in the development and characterization of pediatric cancer models, the successful development of novel agents that have been integrated into standard-of-care clinical protocols, and the development of novel approaches to drug testing that encompass molecular heterogeneity of pediatric cancers.

Our objective is to provide reproducible, high-quality in vivo data to guide the clinical development of novel agents and combinations. It is anticipated that this CPRIT Core facility will be developed within the context of the Research to Accelerate Cures and Equity for Children Act (RACE for Children Act). For preclinical testing, this will necessitate having sufficient models with the appropriate genetic alterations in the context of the appropriate childhood cancer types. The RACE Act requires the FDA to develop a list of molecular targets of known and new drugs/biologics. If agents are determined to be substantially relevant to the growth and progression of pediatric cancer, this may trigger the requirement for pediatric investigations. This expectation applies to drugs/biologics being developed by Pharma and academic centers. Our primary focus will be to use PDX models established and characterized under the previous CPRIT grant (RP160716) representing underserved and Hispanic children.

New approaches to developing novel therapies.

Although preclinical PDX models have been valuable in identifying agents and combinations of drugs that have subsequently shown activity against solid tumors and leukemias, the traditional approach to preclinical testing has been restricted to relatively few models for each tumor type and has failed to incorporate the genetic/epigenetic diversity of the clinical disease. Our retrospective analysis of >2100 drug/tumor experiments from the Pediatric Preclinical Testing Program (PPTP) showed that the use of 1 mouse (rather than 10 per treatment group) gave essentially the same result (78% identical, 95% concordance within one response classification). This ‘single mouse’ testing (SMT) design allows the incorporation of many more models, thus incorporating far greater genetic/epigenetic heterogeneity for a particular disease. Prospective testing and analysis of PPTC data have confirmed and validated the SMT approach.

Single Mouse Testing (SMT):
A. Across tumor histiotypes: It is unknown whether a particular cancer type may be sensitive to many agents based on molecular characteristics. Conversely, a particular characteristic could impart sensitivity in one context but not in another (e.g. BRAF mutations in melanoma vs colon cancer(52, 53)). Using the SMT approach, an agent can be screened against various tumor types, and those sensitive ones can be further validated. An example is shown for the antibody-drug conjugate DS-8201a (trastuzumab-deruxtecan, Figure 1). As shown in Fig. 4A (right), Event-Free Survival (EFS) can be used to assess the intrinsic sensitivity of each model, allowing classification as poor-, intermediate-, and exceptional-responders. Of note, because the SMT design was used, allowing for greater inclusion of models across histiotypes, it was found that malignant rhabdoid tumors (MRTs) were highly sensitive to this agent, with 5 models remaining in Complete Remission (CR) at week 25. An example of SMT where combining a novel CDK4/6 inhibitor with an SoC agent (cyclophosphamide) was antagonistic (Figure 1B).

B. Within a tumor type: An alternative approach, where a particular characteristic being targeted is to evaluate the antitumor activity of the agent/combination against a large number of PDX/CDX models derived from a particular tumor type (e.g. neuroblastoma, rhabdomyosarcoma, etc.) where molecular characterization has identified molecular subsets. We have established baseline antitumor activity for SoC agents and drug combinations used in high-risk protocols for (vincristine/actinomycin D/cyclophosphamide [VAC] and vincristine/irinotecan [/V]I) in rhabdomyosarcoma, relapsed anaplastic Wilms tumor and MRT models, thus are in a good position to assess combinations of SoC agents with new agents (54, 55).

C. Conventional testing to validate SMT in heterotopic (sub-cutaneous) models. Conventional testing (using 10 mice/treatment group) can thus be focused on tumor models found to be highly sensitive to agents tested in the SMT study to validate the result. Definitions of tumor response and statistical analysis are given below and are an evolution of analyses used in the Pediatric Preclinical Testing Program (PPTP(18)) and are presented under Goal 3.

Figure 1A. SMT evaluation of trastuzumab-deruxtecan (DS-8201a). (A) Tumor volume response for 31 xenograft models treated with DS-8201a. Each curve represents the growth of an individual tumor. (B) Time to the event is replotted as a Kaplan-Meier EFS Probability curve, allowing classification of tumors as poor-, intermediate- or exceptional-responders. Note 5 malignant rhabdoid tumor (MRT) models were maintained CR at week 25.

Fig. 1B EFS for 19 pediatric tumor models in mice treated with cyclophosphamide or CDK4/6 inhibitor 30 min before cyclophosphamide administration. Data plotted as Kaplan-Meier EFS Probability curve. Cyclophosphamide (black), Combination (red). Median EFS is 9 weeks for cyclophosphamide and 6 weeks for combination treatment.









D. Incorporating genetic heterogeneity into the screening. We now know that substantial molecular and genetic diversity exists within what has historically been considered homogeneous histological subtypes of childhood cancer(56) (57) (58, 59). With Institutional support ($100K/year from Greehey CCRI and $80k/yr from the Long School of Medicine [see letters attached]), we plan to catalog scRNAseq profiles for all PDX models at passage 3 (priority for CPRIT-derived PDXs) as a reference point to determine whether drugs lead to changes in baseline heterogeneity. While it may not be practical or feasible to recapitulate every rare mutation/fusion, the generation of large numbers of ‘omically’ characterized PDX models can recapitulate major subtypes with a tumor type allowing for drug evaluation(33). We have developed >340 PDX/CDX xenograft models representing solid tumors, brain tumors, and leukemias, Table 1 (at the end of the reference section).

E. Orthotopic models: The group has expertise in the use of orthotopic models, including:

(i) Tibial implantation model Ewing sarcoma (EwS) and osteosarcoma.

(ii) Pulmonary metastasis model for osteosarcoma and rhabdomyosarcoma PDX models.

(iii) Intrahepatic model. For orthotopic hepatoblastoma models, we will use the approach described recently for direct intrahepatic inoculation of tumor cells into the liver(60), using ultrasound-guided implantation (VisualSonics Vevo 2100).

(iv) Subrenal Capsule model. For kidney tumors, we will use implantation to the subrenal capsule(61, 62)

(v) Intracranial model. Will be used for brain tumors

F. Assessment of response using orthotopic or disseminated models: We will determine responses using Kaplan-Meier transformations of event-free survival (EFS) for subcutaneous models, then apply normalization for drug effects.  For example, log cell kill estimates based on EFS as described by Wild et al(63). For orthotopic models, where tumor volume cannot be measured, time to event can be assessed by an investigator(s) blinded to treatment assessing at what point a mouse should be euthanized because of disease progression. Such data can be used to compute EFS and compared to the subcutaneous model. These approaches using EFS or log cell kill will be applied to evaluate single agents or drug combinations and may be used to determine whether there are significant differences in the response of orthotopic models compared to subcutaneous models.

G. Use of the Core testing facility: We want to make the core both accessible and affordable for academic researchers. We propose the following price structure:

  1. Established (funded) investigators:
    The cost per cage of 5 mice (@$62 each) + per diem ($0.77/day) for an average of 11 weeks approximates ($370). For SMT, this would cover 5 tumor lines; for one tumor line, control + 4 treatment groups.
  2. For conventional testing (n=10 mice/group) and Control/treatment (20 mice), the cost would be ($1240/mouse) + $237 per diem costs.
  3. We have budgeted 100 mice/month from CPRIT funds. These will be used to offset costs for testing agents for New Investigators without extramural funding covering testing.
  4. For-profit entities: We would build in a 100% surcharge. This would cover some of the unfunded studies requested.

A website will be created where a ‘Testing Proposal’ form can be completed. Information to be provided will include target identification, proposed testing method (SMT/Conventional), data showing target validation, and chemical structure. This information will be used to prioritize testing by the SC.

Table 1. CPRIT Derived Patient Derived Solid Tumor Xenograft (PDX) Models available for distribution.

Table 2. CPRIT Derived Patient Derived Leukemia Xenograft  (PDX) Models available for distribution.

Table 3: A list of all PDX models with patient demographics available for distribution and drug testing in the TPTC