Hepatocellular carcinoma (HCC) is the most common primary malignancy of the liver and it is the third leading cause of cancer-related deaths worldwide. Recently, aberrant signaling through the FGF19/FGFR4 axis has been implicated in HCC. Here, we describe the development of FGF401, a highly potent and selective, first in class, reversible-covalent small-molecule inhibitor of the kinase activity of FGFR4. FGF401 is exquisitely selective for FGFR4 versus the other FGFR paralogues FGFR1, FGFR2, FGFR3, and all other kinases in the kinome. FGF401 has excellent drug-like properties showing a robust pharmacokinetic/pharmacodynamics/efficacy relationship, driven by a fraction of time above the phospho-FGFR4 IC90 value. FGF401 has remarkable antitumor activity in mice bearing HCC tumor xenografts and patient-derived xenograft models that are positive for FGF19, FGFR4, and KLB. FGF401 is the first FGFR4 inhibitor to enter clinical trials, and a phase I/II study is currently ongoing in HCC and other solid malignancies.

This article is featured in Highlights of This Issue, p. 2183

Hepatocellular carcinoma (HCC) is the third leading cause of cancer-related mortality worldwide (1). After sorafenib, other multikinase inhibitors targeting angiogenesis have recently been approved for patients with advanced disease, but median overall survival (mOS) remains approximately 1 year. Advances in the immuno-oncology (IO) field have also led to the approval of nivolumab but only 20% of the overall population benefits from it (2). Thus, additional therapeutic breakthroughs are still needed.

Genomic and functional studies have highlighted FGF19 as a potential driver gene in subsets of HCCs (3). FGF19 is an ileum-derived postprandial hormone that has a fundamental role in the enterohepatic bile acid/cholesterol system. In order for FGF19 to bind and activate its unique receptor FGFR4 it requires the coreceptor KLB, a transmembrane glycoprotein without enzymatic activity. FGFR4 and KLB are coexpressed in hepatocytes and mediate FGF19 transcriptional suppression of CYP7A1, the rate-limiting enzyme in the bile acids synthesis pathway, thereby lowering the bile acids pool (4, 5). Work from various laboratories including ours suggests that FGF19 is also an oncogene in HCC. First, FGF19 amplification and aberrant expression occurs in some HCCs and HCC cell lines, where it likely activates FGFR4 in an autocrine fashion (6–8). Second, transgenic mice expressing FGF19 in the skeletal muscle develop HCC, which is abolished in an FGFR4-null background and by FGFR4 and FGF19 neutralizing antibodies (9–11).

Because FGF19 has unique specificity for FGFR4 (12), we hypothesized that selective FGFR4 kinase inhibitors would offer a novel therapeutic modality to target FGF19-driven HCC with potentially better tolerability than the current chemotherapies or multi-targeted agents. FGFR4 has a cysteine at position 552 (Cys552) in the middle-hinge region of the ATP-binding site, which is poorly conserved across the kinome and provides an excellent opportunity to achieve selectivity. Recently, BLU-99331, an acrylamide-containing irreversible covalent FGFR4 inhibitor targeting Cys552, has been described (13) and its successor, BLU-554, entered clinical trials (NCT02508467). Following a similar approach, a second irreversible FGFR4 inhibitor, H3B-6527 (14), has also entered clinical studies (NCT02834780). Because of the short resynthesis rate of FGFR4 in HCC cell lines, we considered a reversible-covalent strategy a more promising one as compared with an irreversible binder (15).

Our approach resulted in the identification of FGF401, a highly potent and selective, first in class, reversible-covalent small-molecule inhibitor of the kinase activity of FGFR4 (16, 17). Given its selective activity in a subset of HCC cells, positive for FGF19/FGFR4/KLB and robust oral pharmacokinetic properties, FGF401 holds promise as a new treatment option in FGF19-dependent HCC.

Compound profiling in biochemical and cell-based assays

FGF401 was prepared within Novartis Pharmaceuticals following the procedures described in the international application WO 2015/059668 (example 83) published under the patent cooperation treaty, and the characterizing data are provided as Supplementary File FGF401 characterizing data.

For in vitro profiling assays, FGF401 was dissolved at 10 mmol/L in 100% DMSO. Biochemical kinase assays were conducted as described previously (16, 17). The KINOMEscan profiling was performed at DiscoverRx (18). Cellular FGFR auto-phosphorylation assays were performed using HEK293 cells transiently overexpressing each of the FGFR paralogs. FGFR tyrosine-phosphorylation was quantified by ELISA assay as described previously (19).

BLU-554 and H3B-6527 were acquired from ABCR GmbH & Co. and ChemShuttle Inc., respectively, and prepared as 10 mmol/L solutions in 100% DMSO.

Automated cell proliferation assays were conducted with an ultra-high throughput screening system as described previously (20). For manual cell proliferation assays, cells were seeded in 96-well plates in triplicates and treated with the indicated drugs in 8 point dose-response assays starting at 10 or 3 μmol/L and DMSO for 72 hours. Cell viability was determined with methylene blue staining.

Cell lines were obtained from ATCC, DSMZ, and HSSRB and cultured in RPMI or DMEM with 10% FBS (Invitrogen) at 37°C, 5% CO2. The cell line identity was confirmed by SNP genotyping. Their genomic characterization has been described elsewhere (20), https://portals.broadinstitute.org/ccle. The RNAseq data for FGF19, KLB, FGFR4, and gene copy number for FGF19 are depicted in Supplementary Table S2. To analyze gene promoter methylation profiles we used standard RRBS.

Inducible knockdown of KLB and FGF19 in HCC cells

We cloned short hairpin RNAs (shRNA) targeting FGF19 and KLB in pLKO-Tet-On vector to produce replication-incompetent lentiviruses. Upon lentivirus infection of HUH7 cells, stable cell lines were generated by selection with puromycin (1.5 μg/mL) for 5 days. For cell proliferation and colony formation assays we seeded cells in 96-well plates and 6-well plates, respectively, and induced shRNA expression with 50 ng/mL doxycycline. Cell proliferation and colony growth were evaluated by methylene blue staining. shRNA sequences are listed in the Supplementary Section.

In vivo studies in rodents

The experimental procedures involving animal studies strictly adhered to the Association for Assessment and Accreditation of Laboratory Animal Care International guidelines as published in the Guide for the Care and Use of Laboratory Animals, and to Novartis Corporate Animal Welfare policies. Studies with cell line–derived xenografts were performed at Novartis facilities and were conducted under licenses BS-2498 and BS-2499, approved by the Cantonal Veterinary Office Basel-Stadt. Subcutaneous tumors were induced by injecting cells in HBSS containing 50% BD Matrigel in the flank of athymic nude mice (RH30 4 × 106; Hep3B, HuH7 5 × 106) or nude rats (HuH7 3 × 106). Most of the studies with patient-derived tumor xenografts (PDX) were contracted at GenenDesign, and some conducted at Novartis facilities. The details of pharmacokinetic/pharmacodynamic studies, antitumor efficacy studies and pharmacodynamics analyses are described in the Supplementary Section.

FGF401 is a potent, selective, first-in-class, reversible-covalent inhibitor of FGFR4 kinase

FGF401 is a potent and highly selective FGFR4 kinase inhibitor optimized for oral delivery, the structure of which is shown in Fig. 1A (16). Selective inhibition of FGFR4 is achieved through a reversible–covalent binding interaction, in which the 2-formyl tetrahydronaphthyridine moiety of FGF401 reacts with a cysteine residue at position 552 in the kinase domain of FGFR4, to form a hemithioacetal addition product. Figure 1B depicts the X-ray structure of the addition product resulting from the reaction between FGFR4 and FGF401. This unusual mode of kinase inhibition was revealed by an unbiased high throughput screening approach aimed at identifying inhibitors of FGFR4 showing selectivity over FGFR2 (16). Cys552, a residue situated two positions beyond the gatekeeper (GK+2) in the middle-hinge region of the ATP pocket, is poorly conserved across the human kinome. Only four other kinases have a cysteine residue at this position, and the other FGFR family members contain a larger tyrosine residue at the GK+2 position (Fig. 1C; ref. 21). In biochemical kinase assays, FGF401 inhibited FGFR4 with an IC50 of 2 nmol/L and displayed >2,900-fold selectivity against the other 64 kinases tested (Supplementary Table S1). In FGFR biochemical and mechanistic cellular assays, FGF401 inhibited only FGFR4 (Fig. 1D) and in a KINOMEscan (456 kinases) FGF401 at 3 μmol/L displayed binding affinity only for FGFR4 (Fig. 1E; ref. 22). Thus, FGF401 shows excellent selectivity for the inhibition of FGFR4 versus the other FGFR paralogs and across the human kinome.

Figure 1.

FGF401: chemical structure, X-ray structure in complex with FGFR4, and kinase profile. A, Chemical structure of FGF401. B, X-ray structure of FGF401 in complex with the tyrosine kinase domain of FGFR4 at 2.13 Å resolution. C, Amino acid sequence alignment of the middle-hinge region of the ATP-pocket of FGFR1, FGFR2, FGFR3, and FGFR4 highlighting the differentiating cysteine at the GK+2 position (FGFR4 numbering). D, Activity of FGF401 in biochemical FGFR kinase assays and cellular FGFR auto-phosphorylation assays. IC50s in nmol/L. E, Kinome profiling of FGF401 at 3 μmol/L against 456 kinases showing FGFR4 as the only kinase exhibiting >35% inhibition of control binding (red dot, 0.65% control binding; selectivity score 0.003).

Figure 1.

FGF401: chemical structure, X-ray structure in complex with FGFR4, and kinase profile. A, Chemical structure of FGF401. B, X-ray structure of FGF401 in complex with the tyrosine kinase domain of FGFR4 at 2.13 Å resolution. C, Amino acid sequence alignment of the middle-hinge region of the ATP-pocket of FGFR1, FGFR2, FGFR3, and FGFR4 highlighting the differentiating cysteine at the GK+2 position (FGFR4 numbering). D, Activity of FGF401 in biochemical FGFR kinase assays and cellular FGFR auto-phosphorylation assays. IC50s in nmol/L. E, Kinome profiling of FGF401 at 3 μmol/L against 456 kinases showing FGFR4 as the only kinase exhibiting >35% inhibition of control binding (red dot, 0.65% control binding; selectivity score 0.003).

Close modal

FGF401 activity in cancer cells

To define the pattern of cellular activity for FGF401 we conducted a high throughput proliferation screen across 436 molecularly characterized cancer cell lines from the cancer cell line encyclopedia (CCLE; ref. 20). We generated dose–response curves and derived IC50, EC50, and the maximal activity value reached within a model (Amax) for each cell line (Supplementary Table S2).

These data revealed a pattern dominated by FGF401-insensitive cell lines with EC50 > 30 μmol/L and/or Amax > −50%. A small number of drug-sensitive outliers unveiled four HCC cell lines (JHH7, HUH7, HEP3B217, and SNU878), one gastric cancer cell line (FU97), and one breast cancer cell line (MDAMB453; Fig. 2A). As revealed by RNAseq expression, the four HCC cell lines and the gastric cancer cell line were the highest coexpressers of FGF19 and KLB across 933 cancer cell lines (Fig. 2B). Among the HCC cell lines for which RNAseq was available (n = 24), JHH7, HUH7, HEP3B217, and SNU878 were the highest expressers of FGF19, and all but SNU878 were outlier expressers for KLB (Supplementary Fig S1A). FGFR4 expression was generally not distinct except for being highly overexpressed in the breast cancer MDAMB453 cell line (Fig 2B; Supplementary Fig S1A), likely contributing to its known dependency on FGFR4 (23) and thus, sensitivity to FGF401.

Figure 2.

FGF401 activity in cancer cells and comparison with CRISPR-Cas9 screen. A, Scatter plot showing EC50 and Amax values of FGF401 in high-throughput cell viability assays for 436 cancer cell lines across multiple lineages. B, Scatter plot of 933 cancer cell lines distributed according to FGF19 and KLB transcript expression (RNAseq) and highlighting the FGF401-sensitive cell lines. Symbol size reflects FGFR4 transcript levels (RNAseq). Circles, cell lines included in the proliferation screen; squares, cell lines that were not included in the screen. C and D, Waterfall plots of FGFR4 and KLB dependencies in 426 cancer cell lines from the genome-scale CRISPR-Cas9 screen. Outlier cell line drop-outs for each gene are colored in red and ID is provided. E, Scatter plot of 26 HCC cell lines and one gastric cancer cell line analyzed in manual proliferation assays and distributed according to FGF401 EC50 and Amax. Scatter plots illustrating the distribution of cell lines according to KLB and FGF19 protein levels in cell lysates (F), and in the supernatants (G). Horizontal and vertical dotted lines indicate the FGF19 LOD and the KLB LOQ, respectively. Color coding in A, B, E, F, and G indicates primary site of the cell line. H and I, HUH7, JHH7, and Hep3B2.1-7 cells treated with FGF401 at the indicated concentrations or DMSO (−) for 1 and 72 hours. Bar plots show the ratio of Tyr-phosphorylated (pFGFR4) over total FGFR4 (tFGFR4) measured by ELISA assay (H). Western blots (WB) show phosphorylated FGFR4 upon total pTyr pull down and IgG-HC as loading control; total FGFR4, pFRS2, pERK, and pAKT with ERK, AKT, and α-actinin as loading controls (I). IP, immunoprecipitation.

Figure 2.

FGF401 activity in cancer cells and comparison with CRISPR-Cas9 screen. A, Scatter plot showing EC50 and Amax values of FGF401 in high-throughput cell viability assays for 436 cancer cell lines across multiple lineages. B, Scatter plot of 933 cancer cell lines distributed according to FGF19 and KLB transcript expression (RNAseq) and highlighting the FGF401-sensitive cell lines. Symbol size reflects FGFR4 transcript levels (RNAseq). Circles, cell lines included in the proliferation screen; squares, cell lines that were not included in the screen. C and D, Waterfall plots of FGFR4 and KLB dependencies in 426 cancer cell lines from the genome-scale CRISPR-Cas9 screen. Outlier cell line drop-outs for each gene are colored in red and ID is provided. E, Scatter plot of 26 HCC cell lines and one gastric cancer cell line analyzed in manual proliferation assays and distributed according to FGF401 EC50 and Amax. Scatter plots illustrating the distribution of cell lines according to KLB and FGF19 protein levels in cell lysates (F), and in the supernatants (G). Horizontal and vertical dotted lines indicate the FGF19 LOD and the KLB LOQ, respectively. Color coding in A, B, E, F, and G indicates primary site of the cell line. H and I, HUH7, JHH7, and Hep3B2.1-7 cells treated with FGF401 at the indicated concentrations or DMSO (−) for 1 and 72 hours. Bar plots show the ratio of Tyr-phosphorylated (pFGFR4) over total FGFR4 (tFGFR4) measured by ELISA assay (H). Western blots (WB) show phosphorylated FGFR4 upon total pTyr pull down and IgG-HC as loading control; total FGFR4, pFRS2, pERK, and pAKT with ERK, AKT, and α-actinin as loading controls (I). IP, immunoprecipitation.

Close modal

We inspected a genome-scale CRISPR-Cas9 screen performed in over 400 cancer cell lines (https://depmap.org/portal/) (24), which included four of the FGF401-sensitive ones (HUH7, JHH7, FU97, and MDAMB453). Consistent with the FGF401 screen, we identified FGFR4 and KLB to be selectively essential to the growth of HUH7, JHH7, and FU97, and FGFR4 to also be essential for MDAMB453 and two additional breast cancer lines (Fig. 2C and D), which were resistant to FGF401 (Supplementary Table S2). A greater impact of knocking out FGFR4 versus inhibiting its kinase could account for this difference. Unlike FGFR4 and KLB, the depletion of FGF19 did not reveal a selective outlier profile, maybe due to less optimal performance of the sgRNAs (https://depmap.org/portal/).

FGF401 inhibits proliferation and downstream signaling of FGF19/KLB-positive cell lines

To verify the activity of FGF401 in HCC cell lines that are FGF19/KLB/FGFR4-positive, first we analyzed protein expression of the three markers and found a good correlation with RNAseq in the three instances (Supplementary Fig. S1B). Next, we performed one-by-one, manual proliferation assays with the HCC cell lines from the CCLE (n = 26) and included the gastric cancer FU97 cells. In keeping with the high throughput screen, HUH7, Hep3B, JHH7, SNU878, and FU97 were the only FGF401-sensitive cell lines (Fig. 2E; Supplementary Table S2). We confirmed coexpression of FGF19, FGFR4, and KLB protein in lysates of these cell lines, and detected FGF19 in the media supernatants (Fig 2F and G). As FGF401, BLU-554, and H3B-6527, the two additional clinical FGFR4 inhibitors were also active against the FGF401-sensitive HCC cell lines, with slightly higher GI50 (Supplementary Table S3).

All the other HCC cell lines were insensitive to FGF401 with EC50 values > 10.000 nmol/L and/or Amax values > −10% and most of them had undetectable FGF19 and/or KLB protein. HUH1 and SNU761 expressed FGF19 and KLB but lower levels than the sensitive cell lines, potentially not reaching a threshold of expression needed to sufficiently activate the FGFR4 pathway and confer dependency. SNU886 scored slightly positive for FGF19 by RNAseq (Supplementary Fig S1C), however, the protein was only found in the supernatant. KLB levels in this cell line were near the background levels of the assay, suggesting that KLB expression might be too low to allow FGF19 binding to FGFR4, thus providing a plausible explanation for the cell lysate being FGF19 negative. Within the HCC lineage, FGF19 expression was the best correlate with response to FGF401 (Supplementary Fig S1D). FGF19 lies within the 11q13 amplicon. Some publications suggested that FGF19 is overexpressed in HCC because of gene amplification and that this amplicon could be a response biomarker to anti-FGF19 therapies (6). In our analysis we found that only JHH7 and SNU878 cells have FGF19 copy-number gain above five copies. Conversely, other HCC cell lines with FGF19 copy number >5 were insensitive to FGF401 and with the exception of SNU886, FGF19 protein was below LOD or very low, in-line with low RNAseq reads. In one of these cell lines, Li, high methylation content at the FGF19 promoter revealed by RRBS may explain the absence of FGF19 mRNA and protein despite gene amplification (Supplementary Fig. S1C).

We analyzed FGF401-mediated inhibition of FGFR4 and downstream signaling in Hep3B, JHH7, and HUH7 cells. As shown in Fig. 2H and I, FGF401 inhibited FGFR4 tyrosine phosphorylation at compound concentrations needed to inhibit cell proliferation. Consistent with this result, in cells treated with FGF401 less FGFR4 was pulled down in anti-phosphotyrosine immunoprecipitation assays, although total FGFR4 levels were not significantly impacted by treatment. In agreement with FGFR4 inhibition, pFRS2 and pERK were also inhibited. Modulation of pAKT was generally minor, with some reduction observed only in Hep3B cells. The low baseline pAKT levels in JHH7 cells precluded further conclusions. A differential gene expression analysis revealed downregulation of MAPK target genes including DUSP5, DUSP6, SPRY4, ETV4, and ETV5 after 18-hour treatment with FGF401 (Supplementary Fig. S1F), as well as upregulation of components of the bile acids synthesis pathway including CYP7A1 and glucose metabolism. Thus, our data suggests that HCC cells have retained some properties of hepatocytes and can in part regulate the bile acids and glucose pathways. While FGF19 and FGFR4 were unaffected, KLB showed a statistically significant increase of 2.5 fold (Padj = 1.5 × 10−11), potentially as a compensatory mechanism to FGFR4 inhibition.

Proliferation of FGF19/FGFR4/KLB-positive HCC cell lines depends on each of the three markers

We have shown that FGF401 almost exclusively inhibits proliferation of cell lines positive for FGF19 when coexpressed with FGFR4 and KLB (Fig. 2A–C). The genome-wide CRISPR screen suggested that triple-positive cell lines also depend on KLB. To investigate FGF19 dependency and verify KLB dependency, we engineered HuH7 cell lines to express inducible shRNAs targeting FGF19, KLB, nontargeting control shRNAs (NT-1 and NT-2), and PLK1-targeting shRNA as a positive control for profound cell growth inhibition.

Effective doxycycline-induced knockdown of FGF19 by four different shRNAs (F-2, F-4, F-5, and F-6) led to inhibition of the FGFR4 pathway as exemplified by reduced pFRS2, and to inhibition of cell proliferation. shRNAs F-1 and F-3 were less efficient in knocking down FGF19 and thus, pFRS2 and cell proliferation were less impacted. Nontargeting shRNAs did not affect pFRS2 nor cell proliferation (Fig. 3A and B). Similarly, efficient KLB protein knockdown by shRNAs K-1 and K-4 led to a reduction of pFRS2, proliferation, and colony formation, whereas ineffective anti-KLB shRNAs and nontargeting controls did not (Fig. 3C–E). Owing to optimal KLB knockdown achieved with K-1, we utilized the HUH7/K-1 cells to examine FGFR4 activity in the absence of KLB. Doxycycline-induced KLB knockdown caused profound reduction of baseline pFGFR4 and prevented FGF19-induced FGFR4 phosphorylation as compared with non-doxycycline–treated cells and cells expressing nontargeting shRNA NT-1, which displayed higher pFGFR4 and remained sensitive to exogenous FGF19 treatment (Fig. 3F).

Figure 3.

FGF19 and KLB knockdown in HUH7 HCC cells stably transduced with lentiviruses expressing doxycycline-inducible shRNAs. A and C, Western blots showing FGF19, KLB, and pFRS2 levels in HUH7 cells harboring nontargeting shRNAs (NT-1 and NT-2), FGF19-targeting shRNAs (F-1–F-6), or KLB-targeting shRNAs (K-1–K-4), with and without doxycycline (Dox) induction for 4 days. β-tubulin and β-actin Western blot confirm equal loading. B, D, and E, Cell viability assays with HUH7 cells harboring nontargeting shRNAs, FGF19 shRNAs, KLB shRNAs, and PLK1 shRNA with and without doxycycline induction. B, Cells were plated on 6-well plates and cell growth was monitored using methylene blue staining after 10 days of doxycycline induction. D, Cells were plated on 96-well plates in quadruplicates and stained with methylene blue 10 days post-doxycycline induction. Data are expressed as percentage of control, no doxycycline-treated cells and each bar is the average of four replicates ± SD. E, Cells were seeded on 6-well plates and induced with doxycycline for 21 days. Colony formation was monitored with methylene blue staining. F, HUH7 cells with NT-1 shRNA or K-1 shRNA were treated with doxycycline for 6 days and/or FGF19 (50 ng/mL) for 1 hour. Tyr-phosphorylated (pFGFR4) and total FGFR4 (tFGFR4) were quantified by ELISA assay and expressed as a ratio. G, Cells were treated with doxycycline or FGF401 50 nmol/L for the indicated time and cell lysates analyzed by Western blot using β-tubulin as a loading control. H, Cells were treated with doxycycline for 4 days and FGF19 levels in cells as well as in supernatant were quantified by ELISA assay.

Figure 3.

FGF19 and KLB knockdown in HUH7 HCC cells stably transduced with lentiviruses expressing doxycycline-inducible shRNAs. A and C, Western blots showing FGF19, KLB, and pFRS2 levels in HUH7 cells harboring nontargeting shRNAs (NT-1 and NT-2), FGF19-targeting shRNAs (F-1–F-6), or KLB-targeting shRNAs (K-1–K-4), with and without doxycycline (Dox) induction for 4 days. β-tubulin and β-actin Western blot confirm equal loading. B, D, and E, Cell viability assays with HUH7 cells harboring nontargeting shRNAs, FGF19 shRNAs, KLB shRNAs, and PLK1 shRNA with and without doxycycline induction. B, Cells were plated on 6-well plates and cell growth was monitored using methylene blue staining after 10 days of doxycycline induction. D, Cells were plated on 96-well plates in quadruplicates and stained with methylene blue 10 days post-doxycycline induction. Data are expressed as percentage of control, no doxycycline-treated cells and each bar is the average of four replicates ± SD. E, Cells were seeded on 6-well plates and induced with doxycycline for 21 days. Colony formation was monitored with methylene blue staining. F, HUH7 cells with NT-1 shRNA or K-1 shRNA were treated with doxycycline for 6 days and/or FGF19 (50 ng/mL) for 1 hour. Tyr-phosphorylated (pFGFR4) and total FGFR4 (tFGFR4) were quantified by ELISA assay and expressed as a ratio. G, Cells were treated with doxycycline or FGF401 50 nmol/L for the indicated time and cell lysates analyzed by Western blot using β-tubulin as a loading control. H, Cells were treated with doxycycline for 4 days and FGF19 levels in cells as well as in supernatant were quantified by ELISA assay.

Close modal

Knockdown of KLB also led to reduction of FGF19 in cells (Fig. 3C). To investigate whether FGF19 decrease was due to FGFR4 inhibition (as measured by reduction in pFGFR4; Fig. 2F) we treated HUH7/K-1 cells with doxycycline or FGF401. As before, KLB knockdown resulted in reduced pFRS2 and FGF19, while FGFR4 was unaffected. Instead, FGF401 inhibited pFRS2 but did not affect FGF19 levels (Fig. 3G). We also measured FGF19 by ELISA assay in cell lysates and supernatant (Fig. 3H). In-line with experiment in Fig. 2A, HUH7/F-2 and HUH7/F-4 cells and supernatants thereof were negative for FGF19 upon doxycycline treatment. FGF19 was detected in the supernatants of HUH7/K-1 cells after doxycycline treatment but not in the cell lysates. These data favor a model in which loss of KLB expression prevents FGF19 from binding to FGFR4 resulting in its accumulation in the media.

In summary, our data confirm that FGF401-sensitive HCC cell lines are also dependent on FGF19 and KLB for proliferation, and suggest that disruption of the signaling complex by targeting any of these proteins may inhibit tumorigenesis.

Single-dose pharmacokinetics and pharmacodynamics in RH30 tumor-bearing nude mice

We evaluated the pharmacokinetic properties of FGF401 in mice following a single dose of 1 mg/kg i.v. or 3 mg/kg orally (Supplementary Fig. S2A). Although clearance in mice is low, FGF401 shows a relatively short half-life of 1.4 hours. The corresponding oral bioavailability was 21%.

To study the pharmacokinetic/pharmacodynamics relationship of FGF401, we used the RH30 rhabdomyosarcoma xenograft, which constitutively overexpresses activated FGFR4. Supplementary Fig. S2B shows the means (n = 2) of total drug concentration in plasma and phospho-FGFR4 inhibition in tumor over 24 hours post single oral administration of FGF401 at 1 up to 100 mg/kg. FGF401 exhibited a quick absorption at all doses with Cmax achieved at 2 hours (Tmax). FGF401 plasma exposure was linear over the five dose levels with normalized AUCs of 0.33–0.52 μmol/L × hour and a dose proportional increase in Cmax (Supplementary Fig. S2C). Tumor phospho-FGFR4 normalized to total FGFR4 (p/tPFGFR4) was inhibited in a dose-dependent manner (Supplementary Fig. S2B).

Total drug concentration in plasma versus p/tFGFR4 in tumor was plotted (Fig. 4A) to calculate the in vivo PD IC50 (2.1 nmol/L) and IC90 (52.1 nmol/L), which are highly consistent with the in vitro PD IC50 (10 nmol/L) and IC90 (25 nmol/L). Increasing doses of FGF401 led to more sustained p/tFGFR4 inhibition where doses ≥10 mg/kg resulted in plasma drug concentration above IC50 for 12 hours, and doses ≥30 mg/kg in plasma drug concentration above the IC90 for 8 hours (Fig. 4B).

Figure 4.

Pharmacokinetic/pharmacodynamic and efficacy of FGF401. A, Pharmacokinetic/pharmacodynamic relationship of FGF401 after one single oral treatment of RH30 tumor-bearing nude mice. Total drug concentration in plasma was plotted versus p/tFGFR4 in tumor for each individual animal to show the correlation between pharmacokinetic and pharmacodynamic. In vivo IC50 and IC90 values were calculated from the curve. B, Time was plotted versus total drug concentration in plasma. The IC50 and IC90 concentrations of FGF401 are indicated with dotted lines. C and D, Female nude mice bearing Hep3B subcutaneous xenografts were treated with FGF401 or vehicle control orally (p.o.) twice daily (bid). Values are mean ± SEM, n = 5 mice per group. *, P < 0.05 versus vehicle (Veh) using Kruskal–Wallis (Dunn post hoc) on ΔTVol (C) and one-way ANOVA (Dunnett) on ΔBW (D). E, Female nude mice bearing HuH7 subcutaneous xenografts were treated with FGF401 or vehicle control orally twice a daily. Values are mean ± SEM, n = 6 mice per group. *, P < 0.05 versus vehicle using Kruskal–Wallis (Dunn post hoc). F–I, Pharmacodynamic analyses of FGF401 in HUH7 tumors. F, Tumor-bearing mice were treated with one single dose of FGF401 or six repeated doses and sacrificed at 2 and 12 hours (vehicle 2 hours only) post-dose to analyze tumors by pERK IHC. G, Analysis of Cyp7a1 mRNA transcript at 2 and 12 hours post-last dose (n = 3). H, 7α-hydroxy-4-cholesten-3-one (C4) measured by LC/MS in tumor (n = 3). I, Cyp7A1 activity measured in tumor at 4 and 8 hours post-last dose (n = 3). J and K, Female nude mice bearing subcutaneous NIH3T3 xenografts constitutively expressing the mutant forms of FGFR4 and FGFR4/V550E (J) and FGFR4/N535K (K), were treated with vehicle or FGF401 orally twice a daily. Values are mean ± SEM, n = 6 mice per group. *, P < 0.05 versus vehicle using one-way ANOVA (Dunnett post hoc; J) or Kruskal–Wallis (Dunn post hoc; K) on ΔTVol. ΔBW, change in body weight; ΔTVol, change in tumor volume. *P < 0.05, 0.001 < **P < 0.01, ***P < 0.001 using the two-sample unequal variance, two-tailed distribution (hetero-scedastic Student's t-test).

Figure 4.

Pharmacokinetic/pharmacodynamic and efficacy of FGF401. A, Pharmacokinetic/pharmacodynamic relationship of FGF401 after one single oral treatment of RH30 tumor-bearing nude mice. Total drug concentration in plasma was plotted versus p/tFGFR4 in tumor for each individual animal to show the correlation between pharmacokinetic and pharmacodynamic. In vivo IC50 and IC90 values were calculated from the curve. B, Time was plotted versus total drug concentration in plasma. The IC50 and IC90 concentrations of FGF401 are indicated with dotted lines. C and D, Female nude mice bearing Hep3B subcutaneous xenografts were treated with FGF401 or vehicle control orally (p.o.) twice daily (bid). Values are mean ± SEM, n = 5 mice per group. *, P < 0.05 versus vehicle (Veh) using Kruskal–Wallis (Dunn post hoc) on ΔTVol (C) and one-way ANOVA (Dunnett) on ΔBW (D). E, Female nude mice bearing HuH7 subcutaneous xenografts were treated with FGF401 or vehicle control orally twice a daily. Values are mean ± SEM, n = 6 mice per group. *, P < 0.05 versus vehicle using Kruskal–Wallis (Dunn post hoc). F–I, Pharmacodynamic analyses of FGF401 in HUH7 tumors. F, Tumor-bearing mice were treated with one single dose of FGF401 or six repeated doses and sacrificed at 2 and 12 hours (vehicle 2 hours only) post-dose to analyze tumors by pERK IHC. G, Analysis of Cyp7a1 mRNA transcript at 2 and 12 hours post-last dose (n = 3). H, 7α-hydroxy-4-cholesten-3-one (C4) measured by LC/MS in tumor (n = 3). I, Cyp7A1 activity measured in tumor at 4 and 8 hours post-last dose (n = 3). J and K, Female nude mice bearing subcutaneous NIH3T3 xenografts constitutively expressing the mutant forms of FGFR4 and FGFR4/V550E (J) and FGFR4/N535K (K), were treated with vehicle or FGF401 orally twice a daily. Values are mean ± SEM, n = 6 mice per group. *, P < 0.05 versus vehicle using one-way ANOVA (Dunnett post hoc; J) or Kruskal–Wallis (Dunn post hoc; K) on ΔTVol. ΔBW, change in body weight; ΔTVol, change in tumor volume. *P < 0.05, 0.001 < **P < 0.01, ***P < 0.001 using the two-sample unequal variance, two-tailed distribution (hetero-scedastic Student's t-test).

Close modal

Overall, FGF401 has good pharmacokinetic properties and based on the murine pharmacokinetic data we predicted it would require a twice daily schedule to achieve sustained target inhibition, which is key for maximal efficacy.

FGF401 shows antitumor activity in HCC xenografts

We addressed whether the FGF19/FGFR4/KLB-positive HCC xenografts HUH7 and Hep3B were sensitive to FGF401 in vivo. On the basis of the pharmacokinetic profile and pharmacokinetic/pharmacodynamic relationship (Fig. 4A and B; Supplementary Fig. S2), doses of FGF401 of 10, 30, and 100 mg/kg in an oral, twice daily schedule were chosen to assess its antitumor effect. FGF401 induced approximate tumor stasis at 10 mg/kg twice daily and tumor regression at 30 and 100 mg/kg twice daily, and was well tolerated on the basis of body weight increase (Fig. 4C–E). The antitumor effect correlated with the time of FGF401 plasma exposure over the in vivo pharmacodynamic IC90 (Fig. 4B). Notably, the antitumor effect of FGF401 in HUH7 xenografts was superior to that of sorafenib, first-line standard of care, dosed at 30 mg/kg every day (Supplementary Fig. S3A), which has been shown to induce tumor stasis in a variety of xenograft models and to be similar to 60 mg/kg every day (25, 26).

To investigate pharmacodynamic biomarkers for FGF401, first we monitored pERK levels by IHC in HUH7 (Fig. 4F) and Hep3B (Supplementary Fig. S3B) tumors. pERK was almost completely abolished at 2 hours post-dose and recovered to almost vehicle levels at 12 hours post-dose, irrespective of whether FGF401 was administered as a single dose or as multiple treatments. Because FGF19 signaling impacts CYP7A1 transcription and bile acids synthesis, we investigated CYP7A1 modulation and 7alpha-hydroxy-4-cholesten-3-one (C4) levels, a downstream product of CYP7A1 and precursor of bile acids (4, 5). We detected an increase of CYP7A1 mRNA in both HUH7 tumors (Fig. 4G) and mouse liver (Supplementary Fig. S3C) at 2 and 12 hours post-dose. Correspondingly, C4 was also elevated in HUH7 tumors (Fig. 4H) treated with 100 mg/kg FGF401 at 2 and 12 hours post-dose, in mouse liver (Supplementary Fig. S3D) at all doses at 2 hours post-dose and to a lesser extend at 12 hours post-dose, and in plasma (Supplementary Fig. S3E). Notably, CYP7A1 mRNA and C4 levels appeared significantly lower in HUH7 tumors as compared with mouse liver, likely reflecting its suppression induced by the pathologically activated FGFR4 pathway in the tumor. Finally, we used lysates of HUH7 tumors treated with FGF401 at 100 mg/kg twice daily to measure CYP7A1 activity in a liver microsome assay. Total CYP7A1 activity was increased at 4 and 8 hours post-dose compared with vehicle (Fig. 4I), which likely accounts for the increased C4 levels.

These results are concordant with the observed suppression of MAPK pathway genes and CYP7A1 induction in HCC cell lines treated with FGF401 in vitro (Supplementary Fig. S1E).

Activity of FGF401 against mutationally activated FGFR4

FGFR4 kinase domain mutations V550E and N535K occur in rhabdomyosarcoma, transform NIH3T3 cells, and enhance the metastatic phenotype of rhabdomyosarcoma cells (27). Because no preclinical models of rhabdomyosarcoma with FGFR4-activating mutations are known, we examined the activity of FGF401 against NIH3T3 xenografts engineered to constitutively express FGFR4/V550E and FGFR4/N535K mutants. Tumor-bearing nude mice were treated with FGF401 at 30 or 100 mg/kg orally twice daily. Whereas FGF401 at both dose levels could only moderately delay the growth of tumors harboring the V550E mutation, tumors harboring the N535K mutation regressed (Fig. 4J and K). This data suggest that FGF401 might be a therapeutic option for certain forms of FGFR4-mutated cancers like rhabdomyosarcoma with FGFR4-N535K.

FGF401 efficacy is driven by a fraction of time above the p/tFGFR4 IC90

The dose-dependent sensitivity of HUH7 and Hep3B xenografts to treatment with FGF401 (Fig. 4C and E) suggested a requirement for FGF401 plasma trough (Cmin) exposure near the IC90 (Fig. 4B) for most of the dosing interval to achieve strong antitumor activity (i.e., Cmin ∼ IC90). We addressed this hypothesis in rat using the HUH7 xenograft model, because the FGF401 half-life in rat is significantly longer than in mouse (4.4 hours in rat vs. 1.4 hours in mouse; Supplementary Figs. S2A and S4A).

We first conducted a dose-range efficacy study to identify the minimal efficacious FGF401 dose. We treated HUH7 tumor-bearing nude rats with FGF401 in an oral twice daily schedule at doses of 1, 3, 10, and 30 mg/kg. FGF401 induced approximate tumor stasis at 1 mg/kg twice daily (T/C = 18%), and tumor regression at −52%, −82%, and −96% for the 3, 10, and 30 mg/kg twice daily doses, respectively (Supplementary Fig S4B). In keeping with the mouse efficacy studies, the antitumor effect in rats correlated with FGF401 blood exposure over time near the IC90 (Supplementary Fig. S4C).

As 3 mg/kg twice daily was the lowest dose tested resulting in tumor regression, we conducted a dose fractionation study where the total FGF401 dose was kept constant and divided into either 3 mg/kg twice daily, 6 mg/kg every day, or 12 mg/kg every 2 days dosing regimens in the HuH7 xenograft model. The 3 mg/kg twice daily dose was found to result in the best tumor growth inhibition (regression = −84%) compared with the same daily dose of 6 mg/kg given in a once daily schedule (6 mg/kg every day; T/C = 1%) or in an every other day schedule (12 mg/kg every 2 days; T/C = 23%; Fig. 5A and C). The efficacy correlated well with the fraction of time of FGF401 exposure over the IC90 (52 nmol/L, total concentration in blood) during the dosing interval: 60% for 3 mg/kg twice daily, 42 % for 60 mg/kg every day, and 28% for 12 mg/kg every 2 days (Fig. 5B and C). Taken together, these data suggest that the fraction of time above the p/tFGFR4 IC90 value is a key determinant associated with FGF401 efficacy in vivo.

Figure 5.

Antitumor activity of FGF401 is driven by a fraction of time above the p/tFGFR4 IC90. A, Female nude rats bearing Huh7 subcutaneous xenografts were treated with FGF401 or vehicle control orally at the indicated dose schedules. Values are mean ± SEM, n = 8 mice per group. *, P < 0.05 versus vehicle using Kruskal–Wallis (Dunn post hoc). B, Blood samples were collected at indicated timepoints post 8 days of treatment to determine drug concentrations. These were plotted versus time. Data show mean values ± SD (n = 2) for total drug concentration. IC90 (52.1 nmol/L) level from the in vivo pharmacodynamic model is indicated with a dotted line. C, Efficacy and pharmacokinetic parameters. Antitumor efficacy of FGF401 as percent T/C or percent regression (Reg). Drug exposure is calculated as mean AUCs0–48h (n = 2); Cmax is the maximal achieved drug concentration; Cmin is the drug concentration before the next dose. For exposure at day 8, fraction of time above the IC90 (0.052 μmol/L) is provided. bid, twice daily; qd, every day; q2d, every 2 days.

Figure 5.

Antitumor activity of FGF401 is driven by a fraction of time above the p/tFGFR4 IC90. A, Female nude rats bearing Huh7 subcutaneous xenografts were treated with FGF401 or vehicle control orally at the indicated dose schedules. Values are mean ± SEM, n = 8 mice per group. *, P < 0.05 versus vehicle using Kruskal–Wallis (Dunn post hoc). B, Blood samples were collected at indicated timepoints post 8 days of treatment to determine drug concentrations. These were plotted versus time. Data show mean values ± SD (n = 2) for total drug concentration. IC90 (52.1 nmol/L) level from the in vivo pharmacodynamic model is indicated with a dotted line. C, Efficacy and pharmacokinetic parameters. Antitumor efficacy of FGF401 as percent T/C or percent regression (Reg). Drug exposure is calculated as mean AUCs0–48h (n = 2); Cmax is the maximal achieved drug concentration; Cmin is the drug concentration before the next dose. For exposure at day 8, fraction of time above the IC90 (0.052 μmol/L) is provided. bid, twice daily; qd, every day; q2d, every 2 days.

Close modal

Antitumor activity of FGF401 in PDXs

The in vitro and in vivo data show that FGF401 inhibits the growth of cell lines and xenografts thereof that coexpress FGF19, its receptor FGFR4, and coreceptor KLB. To corroborate these findings, we assessed the efficacy of FGF401 in 35 HCC and two gastric cancer PDXs that included four HCC and one gastric cancer models that were positive for FGF19, KLB, and FGFR4 mRNA, and one gastric PDX with FGF19 gene amplification but KLB negative. The other HCC PDXs were FGF19 negative (Fig 6A and B; Supplementary Table S4). FGF401 administered at 100 mg/kg twice daily to PDX-bearing nude mice caused a statistically significant antitumor activity only in the five FGF19/KLB/FGFR4-positive PDXs, with −55% regression, −24% regression, 7% T/C, and 3% T/C in the four HCCs and 28% T/C in the gastric cancer PDX. The gastric cancer model with FGF19 gene amplification and overexpression but KLB negative, was refractory to FGF401, with 87% T/C (Fig. 6C; Supplementary Table S4). Further exploration of the FGF19, FGFR4, and KLB RNA profiles showed that within the HCC lineage, FGF19 expression was the best correlate with antitumor efficacy (Fig. 6D). As observed in cell lines, FGF19 copy number was associated neither with FGF19 expression nor with FGF401 efficacy (Fig. 6B and D).

Figure 6.

Antitumor activity of FGF401 in a panel of 38 PDXs. A and B, Scatter plots showing the distribution of PDX samples according to KLB and FGF19 transcript expression (A) and FGF19 copy number (CN) versus FGF19 transcript (B). Color coding in A and B indicates primary site. Symbol size in A indicates FGFR4 transcript expression. C, Response to FGF401 (100 mg/kg orally twice daily) over time of four HCC PDXs and two gastric cancer PDXs (Li, liver lineage; GA, gastric lineage). Values are mean ± SEM, n = 3–4 mice per group. *, P < 0.05 (unpaired t test). D, Scatter plots illustrating the relationship between FGF401 response in HCC PDXs [expressed as % T/C or % regression (Reg)] and FGF19 expression, KLB expression, FGFR4 expression, or FGF19 copy number. The HCC PDXs that respond to FGF401 are highlighted. TV, tumor volume.

Figure 6.

Antitumor activity of FGF401 in a panel of 38 PDXs. A and B, Scatter plots showing the distribution of PDX samples according to KLB and FGF19 transcript expression (A) and FGF19 copy number (CN) versus FGF19 transcript (B). Color coding in A and B indicates primary site. Symbol size in A indicates FGFR4 transcript expression. C, Response to FGF401 (100 mg/kg orally twice daily) over time of four HCC PDXs and two gastric cancer PDXs (Li, liver lineage; GA, gastric lineage). Values are mean ± SEM, n = 3–4 mice per group. *, P < 0.05 (unpaired t test). D, Scatter plots illustrating the relationship between FGF401 response in HCC PDXs [expressed as % T/C or % regression (Reg)] and FGF19 expression, KLB expression, FGFR4 expression, or FGF19 copy number. The HCC PDXs that respond to FGF401 are highlighted. TV, tumor volume.

Close modal

In this study, we describe the identification of FGF401, a first in class, highly selective, and potent FGFR4 inhibitor, which might provide benefit to patients with HCC whose tumors present with aberrant FGF19 expression. FGF401 has been the first FGFR4 inhibitor to enter clinical trials, and it is currently in a phase I/II study (NCT02325739). FGF401 covalently binds in a reversible manner to FGFR4 utilizing a unique Cys552 in the middle-hinge region of the ATP-binding site and enabling a high level of kinase selectivity. The selective nature of FGF401 also translated across safety pharmacology assays, it was well tolerated in pharmacologic studies at efficacious doses, and accordingly it is demonstrating good tolerability in patients (28, 29).

Several studies have implicated FGFR4 in various cancer types besides HCC with FGF19 deregulation and rhabdomyosarcoma with FGFR4 mutations (30–33). To examine these hypotheses and to identify clinically relevant patient selection biomarkers, we tested FGF401 in a high-throughput cell viability in vitro screen consisting of 436 cancer cell lines that are part of our CCLE collection. Only six cell lines were sensitive to FGF401, one breast cancer cell line, four HCC, and one gastric. The breast cancer cell line MDAMB453 is an outlier for FGFR4 overexpression and harbors an FGFR4-activating mutation (23). However, the analysis by next-generation sequencing of 4,853 tumors including 522 cases of breast cancer did not identify this mutation (34), which questions the validity of the concept in the clinic. None of the additional breast cancer cell lines, outliers for FGFR4 expression and for which we had response data (n = 8), were sensitive to FGF401. Rhabdomyosarcoma/alveolar rhabdomyosarcoma (ARMS) characterized by the oncogenic PAX3/FOXO1 fusion leading to FGFR4 overexpression was also not identified as an indication for FGF401 because all the ARMS cell lines with PAX3/FOXO1 fusion in the CCLE (n = 7) were refractory to FGF401 despite high FGFR4 levels. Indeed, Marshall and colleagues showed a lack of any contribution from FGFR4 to PAX3/FOXO1-driven tumorigenesis (35). The more recent FGFR4 inhibitor, H3B-6527, has activity in two PAX3/FOXO1-positive ARMS cell lines. This may be because H3B-6527 is less selective than FGF401 and accordingly it shows activity in additional cell lines and PDX models for which the mechanism is unclear (15).

Oncogenic mutant forms of FGFR4 have been detected only in some cases of rhabdomyosarcomas (27). To evaluate the potential of FGF401 in this disease, we used mechanistic NIH3T3 xenograft models expressing FGFR4/V550E and FGFR4/N535K mutants found in rhabdomyosarcoma. While FGF401 only slightly inhibited V550E tumors, N535K tumors responded to treatment and completely regressed. The impact of these mutations on the activity of FGF401 is consistent with what is anticipated from the interaction of FGF401 within the ATP-binding site of FGFR4. Mutating the gatekeeper residue (V550) to the much larger glutamic acid residue is predicted to hinder FGF401 binding due to unfavorable steric interactions (36). Mutating a key residue of the molecular brake (N535) leads to FGFR4 being displaced toward a higher occupancy of active kinase conformations (37), and FGF401 retaining equivalent activity is consistent with FGF401 being equally active against all activity states of FGFR4 (16, 17). These data supports further exploration of FGF401 in rhabdomyosarcomas that harbor specific FGFR4 oncogenic mutations.

In contrast to previous reports that describe increased FGF19 levels by focal amplification of 11q13 as a driver in HCC (6), our data strongly suggest that rather than FGF19 copy number, FGF19 expression in the presence of FGFR4 and KLB is the determining factor. These results are well aligned with findings published in the context of the two other FGFR4-selective inhibitors, BLU-99331 and H3B-6527 (13, 14). Besides HCC, we identified two gastric cancer models, FU97 cell line and GA180 PDX, that were sensitive to FGF401 and remarkably recapitulated the same biomarker profile as the HCC models with FGF19-high expression in the absence of gene amplification, FGFR4 and KLB coexpression. Interestingly, while this article was being reviewed, Gao and colleagues have shown that instead, in squamous cell carcinoma of the head and neck (HNSCC) FGF19 amplification is associated with FGF19 overexpression, poor patient outcome, and dependency on FGF19/FGFR4/KLB (38). Thus, their work reveals a potential new opportunity for FGF19/FGFR4-targeting therapies in HNSCC with FGF19 amplification.

FGF401 showed favorable pharmacokinetic properties and a consistent pharmacokinetic/pharmacodynamic/efficacy relationship in xenograft animal models with p/tFGFR4 levels in the tumor robustly inhibited in a dose-dependent manner. pERK was also strongly reduced at 2 hours post-dose but unlike pFGFR4 it recovered to almost vehicle levels at 12 hours post-dose. This is likely due to feedback loops in the MAPK pathway that result in pathway reactivation for instance by activating other receptor tyrosine kinases or downstream kinases (39). The inhibition and recovery of pERK was the same after single dose and repeat dosing because FGF401 does not accumulate over time and therefore, the pharmacokinetic/pharmacodynamic relationship remains the same throughout the dosing period. Furthermore, in-line with the role of FGF19 as a regulator of bile acids synthesis and similar to BLU-99331 and H3B-6527, FGF401 treatment led to increased CYP7A1 mRNA and C4 levels in tumor and liver, and increased circulating levels of C4. These findings highlight C4 as an alternative biomarker for FGF401 activity that can be measured in blood and of value to also monitor potential on-target toxicity caused by bile acids elevation.

In conclusion, we provide evidence that FGF401 is a potent and highly selective FGFR4 inhibitor with excellent drug-like properties that shows robust antitumor activity in FGFR4-dependent tumor models like hepatocellular carcinomas with aberrant FGF19 overexpression. This prompted us to test its efficacy in relevant cancer patients. FGF401 has been the first FGFR4 inhibitor to enter clinical trials, and a phase I/II study is currently ongoing in HCC (NCT02325739). Since then, two additional selective FGFR4 inhibitors BLU-554 and H3B-6527 have also started phase I studies in advanced HCC (NCT02508467 and NCT02834780), and clinical data for BLU-554 was presented at ESMO 2017 (40). Until now, it is unclear which of the three drugs is superior. Whereas FGF401 was slightly more potent than BLU-554 and H3B-6527 in inhibiting proliferation of FGF19-driven HCC cell lines, multiple factors beyond its intrinsic potency, including human pharmacokinetic properties and tolerability, will influence the clinical outcome. Indeed, BLU-554 shows similar clinical activity to FGF401 in patients with advanced HCC.

Our preclinical analyses of biomarkers for patient selection as well as the results from Hagel and colleagues and Joshi and colleagues (12, 13) implicate that in HCC, a patient selection approach based on tumor FGF19 expression along with FGFR4 and KLB coexpression will likely enrich for responders to FGFR4-selective inhibitors. The results however, do not exclude the possibility that tumors that are only positive for KLB and FGFR4 could be driven in a paracrine manner by FGF19 produced by no-tumor cells and consequently respond to FGFR4 inhibitors (9). These considerations altogether prompted us to implement a molecular prescreening approach in the clinical study to enroll patients with HCC whose tumors are FGFR4 and KLB positive by qRT-PCR, while we monitor FGF19 expression in the tumor in a retrospective manner by both qRT-PCR and IHC. A manuscript describing the design and results of the phase I/II clinical trial of FGF401 will be prepared shortly.

In summary, three novel FGFR4-selective inhibitors, FGF401 described in this article, BLU-554 and H3B-6527 described elsewhere (14, 40) are being evaluated for the treatment of patients with advanced HCC whose tumors display deregulated FGF19/FGFR4 pathway. Because both FGF401 and BLU-554 exhibit promising clinical responses, and given the underlying complexity and molecular heterogeneity of the disease, it will be important to promptly identify the optimal drug combinations to ultimately achieve cures in this patient population.

A. Weiss has ownership interest (including patents) in Novartis. R.A. Fairhurst has ownership interest (including patents) in Novartis. M. Wartmann has ownership interest (including patents) in Novartis Pharma AG. R. Mah has ownership interest (including patents) in a stock. W.R. Sellers is a VP/global head of oncology (paid consultant) at Novartis, has ownership interest (including patents) in Novartis Pharmaceuticals. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Weiss, R.A. Fairhurst, M. Wartmann, J. Kinyamu-Akunda, A. Wolf, Y. Wang, T. Knoepfel, N. Buschmann, C. Leblanc, R. Mah, F. Hofmann, D. Graus Porta

Development of methodology: A. Weiss, R.A. Fairhurst, H.S. Schadt, P. Barzaghi-Rinaudo, J. Blank, D. Graus Porta

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Adler, A. Buhles, M. Kiffe, D. Sterker, M. Centeleghe, H.S. Schadt, P. Couttet, M. Murakami, J. Blank

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Weiss, M. Kiffe, M. Wartmann, H.S. Schadt, P. Couttet, Y. Wang, A. Kauffmann, C. Leblanc, J. Blank, D. Graus Porta

Writing, review, and/or revision of the manuscript: A. Weiss, R.A. Fairhurst, M. Kiffe, J. Kinyamu-Akunda, A. Wolf, C. Leblanc, F. Hofmann, W.R. Sellers, D. Graus Porta

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Kiffe, Y. Wang

Study supervision: A. Weiss, A. Wolf, Y. Wang, F. Hofmann, W.R. Sellers, D. Graus Porta

Other (performed cellular assays and Western blots): C. Stamm

Other (supervision and data review of FGFR4 ELISA setup and execution): M. Wartmann

Other (mechanistical preclinical safety part and review of the manuscript): A. Wolf

Other (design of FGF401 by molecular modeling): P. Furet

We thank Richard Ducray for his excellent project management, Joerg Trappe and Inga Galupa for conducting the biochemical kinase profiling, the bioinformatics team at C-NIBR for their support with the genomic data of the PDX models, and the FGF401 clinical team first led by Luigi Manenti and afterwards by Andrea Myers, for enabling the clinical PoC of FGF401 in HCC.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Bray
F
,
Ferlay
J
,
Soerjomataram
I
,
Siegel
RL
,
Torre
LA
,
Jemal
A
. 
Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries
.
CA Cancer J Clin
2018
;
68
:
394
424
.
2.
Chen
EY
,
Lopez
CD
,
Vaccaro
GM
. 
Updates in the systemic treatment of hepatocellular carcinoma
.
Oncol Hematol Rev
2018
;
14
:
76
81
.
3.
Repana
D
,
Ross
P
. 
Targeting FGF19/FGFR4 pathway: a novel therapeutic strategy for hepatocellular carcinoma
.
Diseases
2015
;
3
:
294
305
.
4.
Lin
BC
,
Wang
M
,
Blackmore
C
,
Desnoyers
LR
. 
Liver-specific activities of FGF19 require Klotho beta
.
J Biol Chem
2007
;
282
:
27277
84
.
5.
Angelin
B
,
Larsson
TE
,
Rudling
M
. 
Circulating fibroblast growth factors as metabolic regulators–a critical appraisal
.
Cell Metab
2012
;
16
:
693
705
.
6.
Sawey
ET
,
Chanrion
M
,
Cai
C
,
Wu
G
,
Zhang
J
,
Zender
L
, et al
Identification of a therapeutic strategy targeting amplified FGF19 in liver cancer by Oncogenomic screening
.
Cancer Cell
2011
;
19
:
347
58
.
7.
Guagnano
V
,
Kauffmann
A
,
Wöhrle
S
,
Stamm
C
,
Ito
M
,
Barys
L
, et al
FGFR genetic alterations predict for sensitivity to NVP-BGJ398, a selective pan-FGFR inhibitor
.
Cancer Discov
2012
;
2
:
1118
33
.
8.
Ahn
SM
,
Jang
SJ
,
Shim
JH
,
Kim
D
,
Hong
SM
,
Sung
CO
, et al
Genomic portrait of resectable hepatocellular carcinomas: implications of RB1 and FGF19 aberrations for patient stratification
.
Hepatology
2014
;
60
:
1972
82
.
9.
Nicholes
K
,
Guillet
S
,
Tomlinson
E
,
Hillan
K
,
Wright
B
,
Frantz
GD
, et al
A mouse model of hepatocellular carcinoma: ectopic expression of fibroblast growth factor 19 in skeletal muscle of transgenic mice
.
Am J Pathol
2002
;
160
:
2295
307
.
10.
French
DM
,
Lin
BC
,
Wang
M
,
Adams
C
,
Shek
T
,
Hötzel
K
, et al
Targeting FGFR4 inhibits hepatocellular carcinoma in preclinical mouse models
.
PLoS One
2012
;
7
:
e36713
11.
Desnoyers
LR
,
Pai
R
,
Ferrando
RE
,
Hötzel
K
,
Le
T
,
Ross
J
, et al
Targeting FGF19 inhibits tumor growth in colon cancer xenograft and FGF19 transgenic hepatocellular carcinoma models
.
Oncogene
2008
;
27
:
85
97
.
12.
Xie
MH
,
Holcomb
I
,
Deuel
B
,
Dowd
P
,
Huang
A
,
Vagts
A
, et al
FGF-19, a novel fibroblast growth factor with unique specificity for FGFR4
.
Cytokine
1999
;
11
:
729
35
.
13.
Hagel
M
,
Miduturu
C
,
Sheets
M
,
Rubin
N
,
Weng
W
,
Stransky
N
, et al
First selective small molecule inhibitor of FGFR4 for the treatment of hepatocellular carcinomas with an activated FGFR4 signaling pathway
.
Cancer Discov
2015
;
5
:
424
37
.
14.
Joshi
JJ
,
Coffey
H
,
Corcoran
E
,
Tsai
J
,
Huang
CL
,
Ichikawa
K
, et al
H3B-6527 Is a potent and selective inhibitor of FGFR4 in FGF19-driven hepatocellular carcinoma
.
Cancer Res
2017
;
77
:
6999
7013
.
15.
Bradshaw
JM
,
McFarland
JM
,
Paavilainen
VO
,
Bisconte
A
,
Tam
D
,
Phan
VT
, et al
Prolonged and tunable residence time using reversible covalent kinase inhibitors
.
Nat Chem Biol
2015
;
11
;
525
31
.
16.
Fairhurst
RA
,
Knoepfel
T
,
Leblanc
C
,
Buschmann
N
,
Gaul
C
,
Blank
J
, et al
Approaches to selective fibroblast growth factor receptor 4 inhibition through targeting the ATP-pocket middle-hinge region
.
Med Chem Comm
2017
;
8
:
1604
13
.
17.
Knoepfel
T
,
Furet
P
,
Mah
R
,
Buschmann
N
,
Leblanc
C
,
Ripoche
S
, et al
2-formylpyridyl ureas as highly selective reversible-covalent inhibitors of fibroblast growth factor receptor 4
.
ACS Med Chem Lett
2018
;
9
:
215
20
.
18.
Karman
MW
,
Herrgard
S
,
Treiber
DK
,
Gallant
P
,
Atteridge
CE
,
Campbell
BT
, et al
A quantitative analysis of kinase inhibitor selectivity
.
Nat Biotechnol
2008
;
26
;
127
132
.
19.
Guagnano
V
,
Furet
P
,
Spanka
C
,
Bordas
V
,
Le Douget
M
,
Stamm
C
, et al
Discovery of 3-(2,6-dichloro-3,5-dimethoxy-phenyl)-1-{6-[4-(4-ethyl-piperazin-1-yl)-phenylamino]-pyrimidin-4-yl}-1-methyl-urea (NVP-BGJ398), a potent and selective inhibitor of the fibroblast growth factor receptor family of receptor tyrosine kinase
.
J Med Chem
2011
;
54
:
7066
83
.
20.
Barretina
J
,
Caponigro
G
,
Stransky
N
,
Venkatesan
K
,
Margolin
AA
,
Kim
S
, et al
The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity
.
Nature
2012
;
483
:
603
7
.
21.
Leproult
E
,
Barluenga
S
,
Moras
D
,
Wurtz
JM
,
Winssinger
N
. 
Cysteine mapping in conformationally distinct kinase nucleotide binding sites: application to the design of selective covalent inhibitors
.
J Med Chem
2011
;
54
:
1347
55
.
22.
Fabian
MA
,
Biggs
WH
 III
,
Treiber
DK
,
Atteridge
CE
,
Azimioara
MD
,
Benedetti
MG
, et al
A small molecule-kinase interaction map for clinical kinase inhibitors
.
Nat Biotechnol
2005
;
23
:
329
36
.
23.
Roidl
A
,
Foo
P
,
Wong
W
,
Mann
C
,
Bechtold
S
,
Berger
HJ
, et al
The FGFR4 Y367C mutant is a dominant oncogene in MDA-MB453 breast cancer cells
.
Oncogene
2010
;
29
:
1543
52
.
24.
Meyers
RM
,
Bryan
JG
,
McFarland
JM
,
Weir
BA
,
Sizemore
AE
,
Xu
H
, et al
Computational correction of copy number effect improves specificity of CRISPR-Cas9 essentiality screens in cancer cells
.
Nat Genetics
2017
;
49
:
1779
84
.
25.
Wilhelm
SM
,
Carter
C
,
Tang
L
,
Wilkie
D
,
McNabola
A
,
Rong
H
, et al
BAY 43-9006 exhibits broad spectrum oral antitumor activity and targets the RAF/MEK/ERK pathway and receptor tyrosine kinases involved in tumor progression and angiogenesis
.
Cancer Res
2004
;
64
:
7099
109
.
26.
Tang
TC
,
Man
S
,
Xu
P
,
Francia
G
,
Hashimoto
K
,
Emmenegger
U
, et al
Development of a resistance-like phenotype to sorafenib by human hepatocellular carcinoma cells is reversible and can be delayed by metronomic UFT chemotherapy
.
Neoplasia
2010
;
12
:
928
40
.
27.
Taylor
JG
 VI
,
Cheuk
AT
,
Tsang
PS
,
Chung
JY
,
Song
YK
,
Desai
K
, et al
Identification of FGFR4-activating mutations in human rhabdomyosarcomas that promote metastasis in xenotransplanted models
.
J Clin Invest
2009
;
119
:
3395
407
.
28.
Chan
SL
,
Yen
C-J
,
Schuler
M
,
Lin
C-C
,
Choo
SP
,
Weiss
KH
, et al
Phase I/II study of FGF401 in adult patients with hepatocellular carcinoma or solid tumors characterized by FGFR4/KLB expression [abstract]
.
In:
Proceedings of the American Association for Cancer Research Annual Meeting; 2017 Apr 1–5
;
Washington, DC. Philadelphia (PA)
:
AACR
; 
2017
;
77:Abstract nr CT106
.
29.
Chan
SL
,
Yen
C-J
,
Schuler
M
,
Lin
C-C
,
Choo
SP
,
Weiss
KH
, et al
Phase I/II study of FGF401 in adult patients with hepatocellular carcinoma or solid tumors characterized by FGFR4/KLB expression
.
ILCA
2017
;
028
.
30.
Heinzle
C
,
Erdem
Z
,
Paur
J
,
Grasl-Kraupp
B
,
Holzmann
K
,
Grusch
M
, et al
Is fibroblast growth factor receptor 4 a suitable target of cancer therapy?
Curr Pharm Des
2014
;
20
:
2881
98
.
31.
Kothari
V
,
Wei
I
,
Shankar
S
,
Kalyana-Sundaram
S
,
Wang
L
,
Ma
LW
, et al
Outlier kinase expression by RNA sequencing as targets for precision therapy
.
Cancer Discov
2013
;
3
:
280
93
.
32.
Davicioni
E
,
Finckenstein
FG
,
Shahbazian
V
,
Buckley
JD
,
Triche
TJ
,
Anderson
MJ
. 
Identification of a PAX-FKHR gene expression signature that defines molecular classes and determines the prognosis of alveolar rhabdomyosarcomas
.
Cancer Res
2006
;
66
:
6936
46
.
33.
Cao
L
,
Yu
Y
,
Bilke
S
,
Walker
RL
,
Mayeenuddin
LH
,
Azorsa
DO
, et al
Genome-wide identification of PAX3-FKHR binding sites in rhabdomyosarcoma reveals candidate target genes important for development and cancer
.
Cancer Res
2010
;
70
:
6497
508
.
34.
Helsten
T
,
Elkin
S
,
Arthur
E
,
Tomson
BN
,
Carter
J
,
Kurzrock
R
. 
The FGFR landscape in cancer: analysis of 4,853 tumors by next-generation sequencing
.
Clin Cancer Res
2016
;
22
:
259
67
.
35.
Marshall
AD
,
van der Ent
MA
,
Grosveld
GC
. 
PAX3-FOXO1 and FGFR4 in alveolar rhabdomyosarcoma
.
Mol Carcinog
2012
;
51
:
807
15
.
36.
Lesca
E
,
Lammens
A
,
Huber
R
,
Augustin
M
. 
Structural analysis of the human fibroblast growth factor receptor 4 kinase
.
J Mol Biol
2014
;
426
;
3744
56
.
37.
Chen
H
,
Ma
J
,
Li
W
,
Eliseenkova
AV
,
Xu
C
,
Neubert
TA
, et al
A molecular brake in the kinase hinge region regulates the activity of receptor tyrosine kinases
.
Mol Cell
2007
;
27
;
717
30
.
38.
Gao
L
,
Lang
L
,
Zhao
X
,
Shay
C
,
Shull
AY
,
Teng
Y
. 
FGF19 amplification reveals an oncogenic dependency upon autocrine FGF19/FGFR4 signaling in head and neck squamous cell carcinoma
.
Oncogene
2019
;
38
:
2394
404
.
39.
Lake
D
,
Correa
SLA
,
Müller
J
. 
Negative feedback regulation of the ERK1/2 MAPK pathway
.
Cell Mol Life Sci
2016
;
73
:
4397
413
.
40.
Kim
R
,
Sarker
D
,
Macarulla
T
,
Yau
T
,
Choo
SP
,
Meyer
T
, et al
Phase I safety and clinical activity of BLU-554 in advanced hepatocellular carcinoma
.
ESMO
2017
;
28
:
v122
v141
.