Dr. Hermans presented the following schema used by the Trace platform to develop PDX models.
Tumor samples are taken from consented patients as part of their normal treatment plan. Her group has had good success developing PDX models from a variety of tumor types, with an overall take rate of ~60%. She has successfully developed more than 120 PDX models from the following tumor types:
- Head & neck
- Endometrial / uterine sarcoma
Dr. Hermans included specific details regarding annotation and quality control of PDX models, both very important steps which add significant value to these tools. Her group compares the originating tumor to the PDX using SNP-based fingerprinting, histology, genomics as well as pharmacological evaluation. She highlighted examples for each type of characterization.
For example, histological characterization may go beyond simple morphological study to comparison of specific markers such as progesterone receptor, HER2 and so on. Genomic characterization using both shallow sequencing and whole exome sequencing shows high degree of similarity between the primary human tumor and the PDX.
In terms of pharmacological characterization, Dr. Hermans showed examples of similar response to therapy between patients and PDX models developed from their tumors. The PDX model developed from a patient who responded well to taxol/carboplatin therapy was also sensitive to carboplatin. In contrast, a tumor from a patient with acquired resistence to taxol/carboplatin showed progression under carboplatin treatment, as expected from the clinical history.
Applications of the Trace PDX platform include preclinical drug screening and biomarker analysis, co-clinical trials and generation of mice engrafted with both tumors and human immune systems. Dr. Hermans concluded that PDX models can narrow the gap between the preclinical and clinical phase. PDX models may permit cost-savings in R&D, but more importantly improve patient outcomes.