Introduction

The implantation of patient tumor samples into immunocompromised mice for creation of patient-derived xenograft (PDX) models is a well-established method. However, the details of how the tissues are implanted can greatly affect the success of forming a tumor in vivo. In this study, we directly compare two variables: mouse host strain and implantation site. In this study we were able to use the same patient tissue specimen for each comparison. NSG mice are the gold standard for establishment of PDX models due to the lack of a functional immune system while SCID/Bg mice are a cheaper alternative but retain some functional immune components.  The evaluation of implantation site is especially important for cancers where orthotopic transplantation is difficult. While subcutaneous (SQ) implants are easy to perform, access to the rodent’s vascular system and a supportive microenvironment is lacking. These issues are overcome in the mammary fat pad (MFP) which is also easily accessible.

Methods

Tumor samples collected from patients diagnosed with bladder, lung, ovarian, or head and neck (H&N) cancer were implanted fresh or viably frozen for implantation at a later date. Each specimen was dissected to 1-2mm³ pieces for implantation into NSG, SCID/Bg, or both strains of mice. One fragment was implanted into the right, inguinal MFP and a second fragment was implanted subcutaneously into the left flank. Mice were monitored for tumor growth for six months. Tumors that formed were harvested and tissues banked for future use. Histology was verified by a pathologist for each cancer site. Short tandem repeat (STR) analysis was performed to confirm that the PDX tumor matched the patient germline DNA. When possible, matched patient tumor and PDX tissue were submitted for WES, RNAseq, and mass spectrometry proteomics for comparison.    

Results

For comparison of tumor take rate in NSG versus SCID/Bg animals, 26 lung and H&N cancer specimens were implanted. Half failed to grow in either strain. Of the remaining 13 tumors, 7 grew in both strains, and 3 grew in NSG only. The 3 remaining tumors grew out as human lymphomas and were not considered further in the analysis. Lung cancer samples formed PDX models equally well in both strains with a take rate of 31%. H&N cancer specimens showed an enhanced take rate in NSG versus SCID/Bg (50% versus 20%). 

Bladder, ovarian, H&N, and lung cancer specimens were used to compare implantation in the MFP versus the SQ space. Tumor samples from 22 bladder and 8 ovarian cancer patients were implanted into SCID/Bg mice.  For lung cancer, 20 patient samples were implanted into NSG (5), SCID/Bg (2), or both strains (13). H&N cancer had 9 patients with 6 implanted in both strains and 3 in SCID/Bg only. These conditions resulted in a total of 78 comparisons which led to 24 outgrowths in the MFP and 6 in the SQ space. The 6 that grew in SQ space also grew in the MFP (5 from lung cancer and 1 from bladder). 

Histology and STR verification of each PDX model was confirmed. Those models whose patient and PDX tumor have been sequenced by RNAseq cluster together in an unsupervised hierarchical clustering analysis.

Conclusion

Using breeding colonies at BCM, we were able to compare the take rate of lung and H&N cancer in NSG and SCID/Bg strains of mice. Similar to published results for breast cancer, lung cancer specimens grew equally well in both strains, while H&N cancer grew more robustly in NSG mice. The numbers of specimens implanted thus far is quite small and more samples need to be implanted to evaluate these trends. A majority of the lung specimens that formed PDX models were from mesothelioma patients, an aggressive cancer which may explain their ability to grow in either strain and in the SQ space.

Evaluation of the implantation site clearly showed that the MFP is more permissive for PDX growth than the SQ space in all cancer sites evaluated.  Although SQ implants are technically easier to perform, transplantation into the MFP is a minimally invasive and easy to learn procedure which enhances establishment of PDX models.