FGFR Genetic Alterations Predict for Sensitivity to NVP-BGJ398, a Selective Pan-FGFR Inhibitor

NVP-BGJ398 inhibits proliferation of a subset of cancer cell lines. Scatter plot showing IC₅₀ values expressed in μmol/L of NVP-BGJ398 in cell viability assays of cell lines according to cancer type. A total of 5.9% of the cell lines were sensitive to NVP-BGJ398 at concentrations up to 500 nmol/L. Cell lines with the composite “FGFR genetic alterations” feature are indicated in red. Points were jittered horizontally to improve readability.

Predictive modeling of NVP-BGJ398 sensitivity using the CCLE features. Top features of drug response identified by categorical-based predictive modeling. Wilcoxon test or Fisher exact test were conducted for continuous or discrete features, respectively. A, fold change or OR as well as P values and Benjamini–Hochberg corrected P values are reported. The number between parentheses for the GeneSets corresponds to the number of genes in the set. B, heatmap for the top 5 features in the model. NVP-BGJ398 response is shown in dark green for the sensitive cell lines and light green for the insensitive cell lines; dark purple is used for discrete features. For GeneSet expression signatures, continuous Z scores are used. P90 and P10 refer to the 90th and 10th percentiles of the GeneSet scores.

FGFR1 amplification in breast, lung, and osteosarcoma cancer cells is associated with response to NVP-BGJ398. A, box-plot showing FGFR1 copy number expressed as log₂ ratio for the 541 cell lines clustered according to cancer type. B, scatter plot of breast, lung, and osteosarcoma cancer cell lines showing the correlation between DNA copy number and transcript expression of FGFR1. Cell lines are colored according to response to NVP-BGJ398. C, effect of NVP-BGJ398 on FGFR downstream signaling as measured by FRS2 Tyr-phosphorylation and Erk1/2 activation. α-Actinin and total Erk1/2 protein levels are shown as a loading control. DMSO, dimethyl sulfoxide. D, stable G292 cell lines expressing short hairpin RNA (shRNA) under the control of a doxycycline (Dox)-inducible promoter were generated via lentiviral infection and puromycin selection. Western blot analysis shows efficient FGFR1 knockdown, p-FRS2 and p-Erk inhibition with shRNA1237 and shRNA1425, as compared with 2 nontargeting shRNAs (NT sh1 and NT sh2). β-Tubulin Western blot analysis is shown as a loading control. E and F, effect of FGFR1-targeting as compared with nontargeting shRNAs on monolayer cell proliferation (E) and anchorage-independent cell growth assays (F) of G292 cells. For monolayer cell proliferation assay, cell growth was monitored at the indicated days after cell seeding, whereas endpoint measurements are given for the soft agar assay (day 15 after cell seeding). G, FGFR1 copy number in a panel of 17 primary human osteosarcoma samples was analyzed by quantitative real-time PCR. Data are shown as average with SEM (n ≥ 2). Br, breast; Ch, chondrosarcoma; CN, copy number; Co, colorectal; En, endometrial; ERK, extracellular signal-regulated kinase; Es, esophagus; Ew, Ewing sarcoma; Ga, gastric; Gl, glioma; HL, hematopoietic and lymphoid tissue; HN, head and neck; Ki, kidney; Li, liver; Lu, lung; Me, melanoma; Ms, mesothelioma; Nb, neuroblastoma; Os, osteosarcoma; Ov, ovarian; Pa, pancreas; Th, thyroid; UT, urinary tract.

FGFR2 amplification in gastric and colon cancer cell lines is associated with response to NVP-BGJ398. A, box-plot showing FGFR2 copy number expressed as log₂ ratio for the 541 cell lines grouped according to cancer type. B, scatter plot of gastric and large intestine cancer cell lines showing the correlation between FGFR2 DNA copy number and transcript expression. C, effect of NVP-BGJ398 on FGFR downstream signaling as measured by FRS2 Tyr-phosphorylation and Erk1/2 activation. α-Actinin and total Erk1/2 protein levels are shown as a loading control. D, SNU16 tumor xenograft–bearing nude rats received NVP-BGJ398 at the indicated doses or vehicle for 14 consecutive days (n = 6 per group). The changes over time in tumor volume are shown. Statistical analysis was conducted by 1-way ANOVA–Dunnett versus vehicle control (*, P < 0.001). E, tumor tissues were recovered 3 and 24 hours after dose treatment and analyzed for FGFR2 Tyr-phosphorylation by Western blot analysis. Total FGFR2 Western blot analysis was conducted to monitor equal loading. Pharmacodynamic analysis of tumors treated with 15 mg/kg NVP-BGJ398 was not feasible due to insufficient material. Br, breast; Ch, chondrosarcoma; Co, colorectal; En, endometrial; Es, esophagus; Ew, Ewing sarcoma; Ga, gastric; Gl, glioma; HL, hematopoietic and lymphoid tissue; HN, head and neck; Ki, kidney; Li, liver; Lu, lung; Me, melanoma; Ms, mesothelioma; Nb, neuroblastoma; Os, osteosarcoma; Ov, ovarian; Pa, pancreas; Th, thyroid; UT, urinary tract.

FGFR2 amplification in primary human gastric tumors predicts for response to NVP-BGJ398. A, scatter plot of primary human gastric tumors showing the relationship between FGFR2 copy number and transcript expression (n = 49). The gastric tumors CHGA10 and GAM033 with FGFR2 gene amplification [(B) and (C), respectively] were grown subcutaneously in mice. Treatment with NVP-BGJ398 at the indicated doses started when average tumor volume was 150 to 180 mm³ and proceeded for up to 14 or 25 days. Tumor volume changes over the course of treatment are shown. Statistical analysis was conducted by 1-way ANOVA–Dunnett versus vehicle control (*, P < 0.001). D, tumor tissues from primary tumor model GAM033 treated with 15 mg/kg NVP-BGJ398 were recovered 3 hours after the last dose treatment and analyzed for FGFR2 Tyr-phosphorylation and Erk1/2 activation by Western blot analysis. Total FGFR2, Erk1/2, and β-tubulin Western blot analyses were conducted to monitor loading.