Genetic alterations in the fibroblast growth factor receptor (FGFR) pathway are promising therapeutic targets in many cancers, including intrahepatic cholangiocarcinoma (ICC). The FGFR inhibitor BGJ398 displayed encouraging efficacy in patients with FGFR2 fusion-positive ICC in a phase II trial, but the durability of response was limited in some patients. Here, we report the molecular basis for acquired resistance to BGJ398 in three patients via integrative genomic characterization of cell-free circulating tumor DNA (cfDNA), primary tumors, and metastases. Serial analysis of cfDNA demonstrated multiple recurrent point mutations in the FGFR2 kinase domain at progression. Accordingly, biopsy of post-progression lesions and rapid autopsy revealed marked inter- and intra-lesional heterogeneity, with different FGFR2 mutations in individual resistant clones. Molecular modeling and in vitro studies indicated that each mutation lead to BGJ398 resistance and was surmountable by structurally distinct FGFR inhibitors. Thus, polyclonal secondary FGFR2 mutations represent an important clinical resistance mechanism that may guide development of future therapeutic strategies.

Purpose: The selective MET inhibitor capmatinib is being investigated in multiple clinical trials, both as a single agent and in combination. Here, we describe the preclinical data of capmatinib, which supported the clinical biomarker strategy for rational patient selection. Experimental Design: The selectivity and cellular activity of capmatinib were assessed in large cellular screening panels. Antitumor efficacy was quantified in a large set of cell line-or patient-derived xenograft models, testing single-agent or combination treatment depending on the genomic profile of the respective models. Results: Capmatinib was found to be highly selective for MET over other kinases. It was active against cancer models that are characterized by MET amplification, marked MET overexpression, MET exon 14 skipping mutations, or MET activation via expression of the ligand hepatocyte growth factor (HGF). In cancer models where MET is the dominant oncogenic driver, anticancer activity could be further enhanced by combination treatments, for example, by the addition of apoptosis-inducing BH3 mimetics. The combinations of capmatinib and other kinase inhibitors resulted in enhanced anticancer activity against models where MET activation cooccurred with other oncogenic drivers, for example EGFR activating mutations. Conclusions: Activity of capmatinib in preclinical models is associated with a small number of plausible genomic features. The low fraction of cancer models that respond to capmatinib as a single agent suggests that the implementation of patient selection strategies based on these biomarkers is critical for clinical development. Capmatinib is also a rational combination partner for other kinase inhibitors to combat MET-driven resistance.

No abstract

<div>AbstractPurpose:<p>The selective MET inhibitor capmatinib is being investigated in multiple clinical trials, both as a single agent and in combination. Here, we describe the preclinical data of capmatinib, which supported the clinical biomarker strategy for rational patient selection.</p>Experimental Design:<p>The selectivity and cellular activity of capmatinib were assessed in large cellular screening panels. Antitumor efficacy was quantified in a large set of cell line– or patient-derived xenograft models, testing single-agent or combination treatment depending on the genomic profile of the respective models.</p>Results:<p>Capmatinib was found to be highly selective for MET over other kinases. It was active against cancer models that are characterized by <i>MET</i> amplification, marked <i>MET</i> overexpression, <i>MET</i> exon 14 skipping mutations, or MET activation via expression of the ligand hepatocyte growth factor (HGF). In cancer models where <i>MET</i> is the dominant oncogenic driver, anticancer activity could be further enhanced by combination treatments, for example, by the addition of apoptosis-inducing BH3 mimetics. The combinations of capmatinib and other kinase inhibitors resulted in enhanced anticancer activity against models where <i>MET</i> activation co-occurred with other oncogenic drivers, for example <i>EGFR</i> activating mutations.</p>Conclusions:<p>Activity of capmatinib in preclinical models is associated with a small number of plausible genomic features. The low fraction of cancer models that respond to capmatinib as a single agent suggests that the implementation of patient selection strategies based on these biomarkers is critical for clinical development. Capmatinib is also a rational combination partner for other kinase inhibitors to combat MET-driven resistance.</p></div>

<p>Supplementary text and figures Supplementary Figure 1 A, visualization of the "inflection point" (or EC50) and "Amax" parameters that are shown in Fig. 2A. B, HGF mRNA expression (by RNA-seq) vs HGF protein in culture supernatant for 76 cancer cell lines. Cell lines with high HGF mRNA expression generally led to increased protein levels in supernatant. U87-MG and IM95 are labeled, because in vivo efficacy data are presented for those models. U87-MG shows a remarkably high level of HGF protein secretion despite relatively low HGF mRNA. C, labeled cell lines scored as hits with at least 2 out of 4 MET inhibitors, each screened twice across a subset of the CCLE. Hits were defined as Amax {less than or equal to} â^'25% for all compounds, and inflection point {less than or equal to} 100 nmol/L for capmatinib, {less than or equal to} 1 ï�­mol/L for crizotinib and JNJ-38877605, and {less than or equal to} 500 nmol/L for PF-4217903. The upper panel shows MET vs HGF mRNA expression, while the lower panel shows MET expression vs MET copy number for the same cell lines. Three regions that among themselves contain all hits (amplified, overexpression, autocrine) were defined such that the hit with the lowest value for the respective expression or copy number value sets the cut-off. Number of hits compared to total number of cell lines in the respective regions are indicated in brackets. Colors as in Figure 2. D, Time course of MET phosphorylation after a 5 minute pulse treatment with 100 ng/mL recombinant HGF in 2 cell lines. NCI-H596 cells bear a MET exon14 skipping mutation while A549 cells do not. Supplementary Figure 2 A, MET dependency (x axis) vs MET gene copy number (y axis), downloaded from https://depmap.org. MET gene dependency was estimated by applying the DEMETER2 model to a combination of 3 large-scale RNAi screens (Broad Achilles, Novartis DRIVE, Marcotte et al.) (7). B, MET dependency (x axis) vs HGF mRNA expression by RNA-seq (y axis), downloaded from https://depmap.org. MET gene dependency was determined from pooled CRISPR screening data (Avana 1.0 library, Broad Institute) applying the CERES algorithm (8,9). Supplementary Figure 3 A, potency comparison of several clinical MET inhibitors in cellular proliferation assays with EBC-1 cells. Representative dose-response curves are shown on the left, numerical inflection point values (mean {plus minus} standard deviation, n = 3) are displayed on the right. B, MET mRNA expression measured by Affymetrix human genome U133 Plus 2.0 arrays (x axis) or RNA-seq (y axis) in 67 lung cancer PDX models (Supplementary Table 3). Identifiers of the 3 models with highest expression are indicated. C, same models as in B, but displaying MET mRNA by RNA-seq (x axis) vs MET copy number by Affymetrix SNP 6.0 array (y axis). D, data as in Fig. 3B but showing individual tumor volumes under treatment with capmatinib. E, mouse body weights corresponding to the experiment shown in Fig. 3B. F, data as in Fig. 3C but phospho- MET and total MET values from multi-spot ELISA shown separately. G, repeat of capmatinib efficacy study with lung PDX tumors bearing a MET exon 14 skipping mutation (model LU5381) as in Fig. 3D, but longer treatment duration. An average regression of ~60% was observed on day 12. A vehicle control was not repeated in this study. H, antitumor efficacy of capmatinib (dosed as indicated) against xenografts of the gastric cancer cell line IM95, which expresses HGF. Supplementary Figure 4 A, Loewe excess for data shown in Fig. 4C (upper panel), and % inhibition as well as Loewe excess for a combination matrix of capmatinib and the selective BCL2 inhibitor venetoclax (lower panel), in the cell line NCI-H1993. Percent dead cells were quantified by dual imaging with propidium iodide (PI) and Hoechst 33342. B, Percent dead cells by PI/Hoechst (upper panel) and Loewe excess (lower panel) for EBC-1 cells treated as in Fig. 4C and Supplementary Fig. 4A. C, Loewe excess for data shown in Fig. 4D, cell lines as indicated. Here, data are based on a CellTiter-Glo readout with quantification of seeded cells at time of compound addition ("growth inhibition" calculation). Loewe excess also refers to the "growth inhibition" data. D, treatment of EBC-1 cells with the indicated dose matrix of capmatinib and erlotinib. Experimental setup and analysis as in Fig. 4D (docetaxel combination). Supplementary Figure 5 A, HCC827 or HCC827 GR lung cancer cells were exposed to gefitinib, capmatinib, or combinations in a fixed ratio for 72 hours before measuring cell viability using a resazurin assay. The x axis label corresponds to gefitinib concentrations, while capmatinib was used at 10-fold lower concentrations due to its higher potency. In combination, gefitinib and capmatinib were mixed at a ratio of 10:1. B, HCC827 or NCI-H3255 cells were treated with a dilution series of gefitinib in the presence or absence of 50 ng/mL recombinant HGF. Cell viability was measured after 96 hours (HCC827) or 72 hours (NCIH3255) using a resazurin assay. The initial amount of viable cells was quantified at the time of compound addition (dashed line), and cell growth on the y axis is expressed as a multiple of this value. C, lysates of the lung cancer PDX model X-1787, either treated with vehicle or crizotinib, were incubated on a phospho-RTK array. Phospho-MET is clearly detectable in vehicle-treated lysate, suggesting that the high MET mRNA expression in this model leads to MET protein expression and activation. D, anti-tumor efficacy of crizotinib against X-1787 PDX tumors, characterized by presence of EML4-ALK and high MET expression. N = 4 per arm. E, RKO cells (BRAF V600E-mutant colorectal cancer cells secreting HGF) were exposed to capmatinib and fixed ratios of dabrafenib and trametinib in a dose matrix as indicated. After 3 days, viable cells were quantified by CellTiter-Glo. Inhibition is quantified in 2 different ways, without or with regard of the initial amount of viable cells at the time of compound addition (left and middle panel, respectively). Calculations are explained in detail in Materials and Methods. The right panel shows the Loewe excess for "% inhibition."</p>