Objective

In cancer drug discovery, lead compounds found to possess the most anti-tumor efficacy in vitro need to be rapidly and economically prioritized for further analysis in conventional xenograft testing. To bridge this gap, Hollingshead et al. (1995) developed the in vivo Hollow Fiber Model (HFM), a pre-screen predicting xenograft efficacy. The screen is based on drug- (but not cell-) permeable fibers loaded with tumor cells which are implanted into mice. Following drug treatment, the number of viable cells in the fibers provides information about the drugs’ potential in vivo anti-tumor efficacy. Since one animal can receive up to three intraperitoneally and three subcutaneously implanted fibers, one lead can be simultaneously evaluated against up to three cell lines.

To evaluate the predictability of the HFM two target-specific kinase inhibitors were selected: Crizotinib and VX-680. Crizotinib is an FDA-approved drug for NSCL which inhibits the kinases ALK, MET and ROS1, whereas VX-680 inhibits the AURORA kinase family. Both drugs were run in a proliferation assay against a large panel of tumor cell lines to determine the IC₅₀ value. Subsequently, 9 cell lines representing different inhibitor potencies (high, medium, low/none) were chosen for comparison of the proliferation, HFM and xenograft tumor model results. For Crizotinib the cell lines MKN-45, MiaPaca2 and KARPAS-299 were chosen with highly potent inhibition whereas for VX-680 the cell lines HCT-116, Molm-13 and Colo-205 are most potently inhibited in the proliferation assay. Cell lines SKOV-3, LN229 and HT-29 were chosen as non-inhibited cell lines for both kinase inhibitors in the in vitro proliferation assay.

For each of the 9 cell lines, a subcutaneous xenograft mouse model was established and the effect of both Crizotinib and VX680 treatment on tumor growth determined. In parallel, all cell lines were tested in three separate HFMs with 3 cell lines simultaneously. The implanted hollow fibers were analyzed via MTT assay 14 days after therapy initiation.

The correlation between the cell-based inhibitory activity and xenograft tumor model results and the correlation between HFM and xenograft tumor model results will be presented. Finally, the implementation of the HFM for the tumor model selection process in cancer drug discovery will be discussed. In case the HFM is a highly predictive in vivo prescreen for target-specific inhibitors, the HFM will reduce costs, time and most importantly mice consumption according to reduction and refinement, criteria of the three rules (3Rs), guiding principles for more ethical use of animals in drug testing.