Comparison of the cancer gene targeting and biochemical selectivities of all targeted kinase inhibitors approved for clinical use

Abstract

The anti-proliferative activities of all twenty-five targeted kinase inhibitor drugs that are in clinical use were measured in two large assay panels: (1) a panel of proliferation assays of forty-four human cancer cell lines from diverse tumour tissue origins; and (2) a panel of more than 300 kinase enzyme activity assays. This study provides a head-on comparison of all kinase inhibitor drugs in use (status Nov. 2013), and for six of these drugs, the first kinome profiling data in the public domain. Correlation of drug activities with cancer gene mutations revealed novel drug sensitivity markers, suggesting that cancers dependent on mutant CTNNB1 will respond to trametinib and other MEK inhibitors, and cancers dependent on SMAD4 to small molecule EGFR inhibitor drugs. Comparison of cellular targeting efficacies reveals the most targeted inhibitors for EGFR, ABL1 and BRAF(V600E)-driven cell growth, and demonstrates that the best targeted agents combine high biochemical potency with good selectivity. For ABL1 inhibitors, we computationally deduce optimized kinase profiles for use in a next generation of drugs. Our study shows the power of combining biochemical and cellular profiling data in the evaluation of kinase inhibitor drug action.

Figure 1. 2D structures of the kinase inhibitors profiled in this study.

All are kinase inhibitors that were approved for clinical use at Nov. 2013.

Figure 2. Cellular profiling of marketed kinase inhibitors.

A: Tissue origin of cell lines in the Oncolines panel. B: Frequency of cancer gene changes in the cell panel, i.e., mutations, translocations and copy number changes in the COSMIC Cell Line Project . C: Hierarchical clustering of profiling data of marketed kinase inhibitor drugs in the 44-cell line panel. Unscaled ¹⁰logIC₅₀s were used. Doxorubicin_123 is a triplicate profiling for control. Non-kinase inhibitors are coloured red. D: Kinase inhibitors have a greater selectivity in the cell panel than classic cytotoxic agents (5-fluorouracil, cisplatin, vincristine, doxorubicin, etoposide, docetaxel and bortezomib).

Figure 3. Biochemical profiling of marketed kinase inhibitors.

A: Hierarchical clustering of inhibitory profiles of all kinase drugs in a panel of more than 300 biochemical kinase assays (%-inhibition at 1 μM inhibitor concentration). Trametinib, everolimus and temsirolimus show only minor inhibition, as mTOR and MEK kinase assays are not included in the panel. B: Potent biochemical IC₅₀s on the biological target correlate with more potent cellular IC₅₀s. C: Biochemical selectivity leads to a more selective response in the cell panel. Biochemical selectivity was quantified by selectivity entropy and the selectivity of targeting cell growth was expressed by the average IC₅₀ in the cell panel. Non-oncology drugs fasudil and tofacitinib were deleted from the analysis because of lack of response. Open circles: the mTOR and MEK inhibitors everolimus, temsirolimus and trametinib, respectively.

Figure 4. Anova analysis reveals novel drug response markers.

A: the MEK inhibitor trametinib and B: the EGFR inhibitor afatinib. The volcano plots show the average IC₅₀ shift between mutant and non-mutant cell lines (x-axis) and the significance from the Anova test (y-axis). Significance was corrected for multiple-testing and all associations above the threshold level (dotted line) are coloured green. Areas of circles are proportional with the number of cell lines carrying mutations.

Figure 5. Comparison of the targeting efficacy of marketed inhibitors.

Each circle represents a marketed kinase inhibitor and its targeted cell growth inhibition. A: Cell lines that overexpress EGFR. B: Cell lines containing the BRAF(V600E) mutation. C: Cell lines containing aberrant ABL1 signalling. Compounds in the upper left corner of the plots have superior targeting. Statistically relevant associations after correction for multiple testing are coloured blue. D: Quantitative comparison of inhibitor targeting by standardization of IC₅₀ shifts between sensitive and non-sensitive cell lines.

Figure 6. Determining the optimal kinome profile of an ABL inhibitor.

A: Biochemical components of kinase inhibitors that contribute to the specific targeting of ABL1 dependent cell growth. The circle labelled ABL1 refers to biochemical ABL1 inhibition. B: Inhibitors with equal ABL1 and ABL2 affinity in an independent dataset are better in targeting ABL1-dependent cell growth than inhibitors with ABL1 activity alone. Poor ABL2 affinity signifies binding K_d differences between 4 and 26-fold compared to ABL1. Equal affinity signifies binding K_d differences between 0.5 and 4-fold.