Introduction

Patient-derived Xenograft (PDX) models are grafts of human cancer tumors implanted into various transplantation sites in a different species, typically mouse or rat.  PDX models are essential tools for evaluating experimental therapeutics and for serving as cohorts of patient avatars when conducting larger scale pre-clinical trials. Selection of PDX models for pre-clinical studies has, to date, mostly been limited to those available via existing collaborations, or those obtained through commercial vendors rather than via a rational search for the most appropriate models with which to conduct a given study. Therefore development of a publicly accessible PDX Portal (https://pdxportal.research.bcm.edu/) is a natural next step to promote the translational application of more recently developed and less well-used PDX models. With the support of a CPRIT Core Facility Support Grant, the Baylor College of Medicine (BCM) Patient-derived Xenograft (PDX) Core Facility created the Texas PDX Portal to provide collection management and technical support as well as computational infrastructure for PDX-based research.

Methods

To manage PDX collections effectively, de-identified patient clinical and omics data, as well as PDX experimental and omics data are curated and stored. DNAseq, RNAseq, and proteomics data are processed through standardized pipelines and made accessible on the portal. PDX gene expression (mRNA/protein), copy number alterations, and mutations can be queried in combination with clinical markers to identify models potentially useful as a PDX cohort. Samples can be requested from one or more model development laboratories either at BCM or at institutions across the state of Texas.

Web-based data self-service tools were implemented to load patient and model data via standardized templates that can be used by PDX developers. The inclusion of the data import and editing tools creates a self-service data management framework that enables PDX development groups to track, analyze and share PDX models for collaborations or study services independently, without external assistance.

Results

The distinguishing characteristics of this PDX Portal are the PDX collection management functions which provide individual PDX development groups real-time access to standardized, databased, and searchable data vs. using a traditional time consuming, labor intensive spreadsheet approach. Currently the PDX Portal represents Bladder, Breast, Colorectal, Leukemia, Liver, Lung, Pancreatic, Prostate, and Sarcoma collections, from Baylor College of Medicine, MD Anderson Cancer Center, Texas Children’s Hospital, and UT Austin. Eventually, 95% of all PDX programs in the state will be represented on the PDX Portal.

According to Google Analytics, the PDX portal has been viewed by over 10,000 unique individuals from 64 different countries, notably 2,000 users originating from China, Europe, Australia, and Canada, thus promoting awareness of the PDX model resources within Texas. Users of the portal have reported that models identified on the portal have supported 37 grants, representing awards equaling $52,400,000. Forty-six publications have been generated from the 788 specimens which have been distributed for PDX research from the portal to 89 investigators at 39 institutions since the PDX portal was made available to the public.

Conclusion

The BCM PDX Portal is a highly effective PDX collection management tool allowing tightly coupled clinical and genomic data access in a visually intuitive manner, thereby enhancing the utility of PDX collections. Cost benefits, ease of data sharing, increased utilization of the models, and establishment of a central resource with critical mass for drug studies have been realized. As additional Texas PDX groups join the consortium, we expect increased recognition of the existing domain expertise within Texas and increased collaborations with clinical research organizations and industry partners.