Background
Breast cancer is the most common type of cancer and the leading cause of cancer deaths in women worldwide [
1]. In most countries, 3% to 12% of breast cancers are advanced or metastatic at diagnosis [
2]. Most breast cancers are hormone receptor–positive (HR
+), with 75% to 83% of breast cancers expressing estrogen receptor (ER)-α and/or progesterone receptor [
3‐
5]. Likewise, approximately 86% to 87% of breast cancers are negative for overexpression of human epidermal growth factor receptor 2 (HER2
−) [
6,
7].
Currently, endocrine therapy is recommended as initial therapy for patients with HR
+, HER2
− advanced breast cancer [
8,
9]. Aromatase inhibitors are the standard of care for postmenopausal patients [
9]. However, only 20% to 40% of patients respond to first-line therapy, and approximately one-half of responders relapse within 8–14 months [
10]. Most patients eventually relapse because currently available treatments are not curative [
11]. Second-line endocrine therapy (e.g., fulvestrant, aromatase inhibitors, tamoxifen) is recommended following relapse, but the response is generally short-lived. For example, the duration of response (DOR) to second-line fulvestrant or exemestane is approximately 3–5 months [
12].
Several mechanisms of endocrine therapy resistance have been described, including activation of receptor tyrosine kinases (e.g., fibroblast growth factor receptor [FGFR]) and their downstream signaling pathways (e.g., phosphoinositide 3-kinase [PI3K]/Akt/mechanistic target of rapamycin [mTOR]), as well as activation of the cyclin-dependent kinases 4 and 6 that regulate cell cycle progression [
13]. Efforts to improve outcomes and reduce endocrine therapy resistance have led to the development of combination therapies that included targeted agents against these resistance pathways. Positive results from the phase III Breast Cancer Trials of Oral Everolimus 2 (BOLERO-2) trial led to the approval of everolimus, an mTOR inhibitor, in combination with second-line exemestane in postmenopausal women with HR
+, HER2
− advanced breast cancer that progressed during prior nonsteroidal aromatase inhibitor therapy [
14,
15]. Later, positive results from the Palbociclib: Ongoing Trials in the Management of Breast Cancer (PALOMA) studies led to the approval of palbociclib (a cyclin-dependent kinase 4 and 6 inhibitor) as first-line therapy in combination with letrozole and as second-line therapy in combination with fulvestrant [
16,
17].
Aberrant regulation of fibroblast growth factor (FGF) and FGFR signaling is associated with tumorigenic activity [
18], an increased risk of developing breast cancer [
19‐
21], and resistance to endocrine therapy [
13]. Amplifications in
FGFR1 and
FGFR4 are found in 9% to 10% and 10% of primary breast cancers overall, respectively [
22‐
25].
FGFR1 amplification is more frequently associated with luminal B cancer, whereas
FGFR4 amplification is more common in HR
+ tumors and a subset of HER2
+ tumors [
22,
26,
27].
FGFR2 amplification is present in 4% of triple-negative breast cancers [
28]. Overexpression of
FGFR family members is associated with poor prognosis, including reduced overall survival (OS), disease-free survival, and relapse-free survival [
24,
29‐
31]. FGFR overexpression is also associated with resistance to hormone therapy [
26,
32] and chemotherapy [
33,
34]. Importantly, FGFR1-induced tamoxifen resistance can be reversed by inhibiting
FGFR1 expression [
32]. Aberrant PI3K/Akt/mTOR signaling is also seen in cells with
FGFR1 overexpression and amplification [
26], and response to the PI3K inhibitor alpelisib is reduced in ER
+/
PIK3CA-mutant breast cancer cells that overexpress
FGFR1 [
35]. Taken together, these results provide rationale for the investigation of FGFR inhibitors in breast cancer therapy.
Dovitinib (TKI258), a small-molecule inhibitor of FGFR1, FGFR2, and FGFR3 and other receptor tyrosine kinases [
36], has shown preclinical activity in
FGFR-expressing breast cancer models in vivo and in vitro [
37]. Dovitinib inhibited cell proliferation in
FGFR-amplified cell lines and showed antitumor activity in
FGFR-amplified xenograft models [
38]. In a phase II trial of single-agent dovitinib, encouraging clinical activity was observed in patients with HR
+, HER2
−
FGF pathway–amplified breast cancer [
38].
FGF pathway amplification status was determined using in situ hybridization as part of the eligibility criteria (
FGFR1 only) and using quantitative polymerase chain reaction (qPCR) as an exploratory analysis (
FGFR1,
FGFR2, and
FGF3). Correlative studies between
FGF pathway amplification markers and antitumor activity indicated that dovitinib activity was higher in patients who had
FGF pathway amplification measured by qPCR, particularly in those who had higher levels of
FGFR1 amplification (i.e., at least six copies of
FGFR1) [
38]. The combination of fulvestrant and dovitinib could potentially overcome resistance to endocrine therapy, thereby reducing the need for cytotoxic chemotherapy in relapsed patients. Together, these data and hypotheses prompted the initiation of this phase II, placebo-controlled trial of dovitinib plus fulvestrant in postmenopausal patients with HR
+, HER2
− locally advanced or metastatic breast cancer.
The primary objective of this study was to determine the effect of treatment with dovitinib in combination with fulvestrant vs placebo plus fulvestrant on progression-free survival (PFS) in postmenopausal patients with HR+, HER2− breast cancer that had progressed during or after prior endocrine therapy in all evaluable patients, regardless of FGF pathway amplification status, and in patients with FGF pathway amplification (as measured by qPCR using a cutoff of at least six copies of FGFR1, FGFR2, or FGF3). The key secondary objective was overall response rate (ORR). Additional secondary objectives included DOR, OS, safety, and pharmacokinetics of dovitinib.
Methods
Study design
We conducted a phase II, multicenter, international, randomized, double-blind, placebo-controlled trial (ClinicalTrials.gov identifier: NCT01528345) designed to evaluate the efficacy and safety of dovitinib in combination with fulvestrant in postmenopausal women with HR+, HER2− locally advanced or metastatic breast cancer who had evidence of disease progression. Enrolled patients were randomized in a 1:1 ratio to receive dovitinib plus fulvestrant or placebo plus fulvestrant, stratified by FGF pathway amplification status (amplified vs nonamplified) and presence of visceral disease (yes vs no). All patients received fulvestrant 500 mg (intramuscular injection once every 4 weeks, with an additional dose 2 weeks after the initial dose) and dovitinib (500 mg) or placebo orally following a weekly 5 days on and 2 days off schedule until death, loss to follow-up, disease progression, or consent withdrawal.
Patients and medications were randomized using automated systems. At the time of initial screening, enrolled patients received a patient number, which was used as the primary identifier for the patient throughout the study. The interactive response technology provider generated a randomized patient list, using a validated automated system, by randomly assigning patient numbers to randomization numbers. Each randomization number was linked to a treatment arm and a medication number. Patients were randomized 1:1 to each of the study arms, with 45 FGF pathway–amplified and 30 FGF pathway–nonamplified patients planned in each arm. Medications were separately randomized by the study sponsor using a validated automated system that randomly assigned medication numbers to medication packs containing each of the study treatments. In this double-blind study, patients, investigators, study team members, and anyone involved in the conduct of the study remained blinded to the identity of the treatment from the time of randomization until database lock. The study medication and placebo had identical packaging, labeling, appearance, and administration schedules to conceal the identity of the treatments.
Patients
Eligible patients were postmenopausal women with HR
+, HER2
− locally advanced or metastatic breast cancer who had evidence of disease progression. Progression was defined as at least one measurable lesion per Response Evaluation Criteria In Solid Tumors (RECIST) version 1.1 or at least one nonmeasurable lytic or mixed bone lesion in the absence of measurable disease. Progression could have occurred during or after prior endocrine therapy, within 12 months of the end of adjuvant endocrine therapy, or within 1 month of the end of any endocrine therapy for localized advanced or metastatic breast cancer. Eligible patients had confirmed postmenopausal status (i.e., aged ≥55 years with ≥1 year of amenorrhea, aged <55 years with ≥1 year of amenorrhea in the absence of ovarian suppression with an estradiol assay result of <20 pg/ml, or surgical menopause with bilateral oophorectomy), Eastern Cooperative Oncology Group (ECOG) performance status ≤2, and available archival or fresh tumor tissue for
FGF pathway status determination in the primary tumor by the central laboratory.
FGF pathway amplification was determined by a designated Clinical Laboratory Improvement Amendments–certified laboratory using a TaqMan PCR assay (Life Technologies, Carlsbad, CA, USA), as previously described [
38]. Positive amplification for each
FGF pathway marker tested (i.e.,
FGFR1,
FGFR2, or
FGF3) was defined as a copy number ≥6; copy number was quantified by comparison with a reference gene (human ribonuclease P RNA component H1) and calculated using CopyCaller software (version 1.1; Applied Biosystems, Foster City, CA, USA). Samples were considered to be
FGF pathway–amplified if they had positive amplification of
FGFR1,
FGFR2, and/or
FGF3. Given that the previously defined amplification cutoff of at least six copies for
FGFR1 was associated with higher sensitivity to dovitinib monotherapy [
38], the same cutoff was used in this combination therapy study. Exclusion criteria included HER2 overexpression (assessed by immunohistochemistry), prior therapy with fulvestrant (as a single agent or in combination with other therapies) or FGFR inhibitors, or chemotherapy or more than one line of any prior hormone therapy for locally advanced or metastatic breast cancer. An estimated 1000 patients were expected to be screened for
FGF pathway amplification in order to identify and randomize a total of 150 patients stratified by
FGF pathway amplification and presence of visceral disease.
Assessments
Radiographic assessments (computed tomography, magnetic resonance imaging, or radiography) were performed at screening, on day 5 of weeks 8 and 16, before fulvestrant administration every 8 weeks for the remainder of study treatment, and at the end of treatment (if not assessed within 8 weeks before visit). Safety assessments were performed continually until 30 days after the last study treatment. No additional tumor assessments were required to confirm response (complete response [CR] or partial response [PR]) outside the protocol-specified 8-week tumor assessment.
Patients who did not discontinue study treatment owing to disease progression or death, or who were not lost to follow-up or did not withdraw consent, were assessed every 8 weeks for disease status, ECOG performance status, and patient-reported outcomes until the start of new anticancer therapy, disease progression, death, loss to follow-up, or consent withdrawal. Survival follow-up was performed every 3 months until death, loss to follow-up, or consent withdrawal for patients who discontinued the treatment.
Study endpoints
The coprimary endpoints were PFS in the overall patient population regardless of FGF pathway amplification status and PFS in the subgroup of patients with FGF pathway amplification. PFS was defined as the time from date of randomization to the date of first radiologically documented, investigator-assessed disease progression per RECIST v1.1 or death due to any cause. The key secondary endpoint was ORR, defined as the percentage of patients with best overall response of CR or PR. Additional secondary endpoints included DOR, OS, safety, and pharmacokinetics of dovitinib. Safety analysis, by treatment arm, was based on the frequency of adverse events (AEs), summarized by system organ class, severity (based on the Common Terminology Criteria for Adverse Events version 4.03), type, and relationship to study treatment.
Analysis sets
The full analysis set, which consisted of all patients who were randomized and assigned study treatment, was the primary population for the efficacy endpoint analyses. The safety set consisted of all randomized patients who received at least one dose of any compound of the study treatment (dovitinib plus fulvestrant or placebo plus fulvestrant).
Data analysis
The primary endpoint, PFS, was evaluated in each of the treatment arms using three sets of comparisons, following a Bayesian design: (1) all patients regardless of
FGF pathway amplification status, (2)
FGF pathway–amplified, and (3)
FGF pathway–nonamplified (Additional file
1). Patients who did not have a PFS event at the time of analysis or who had received further antineoplastic therapy were censored at the time of the last tumor assessment. Kaplan-Meier plots were generated by treatment arm for the full population,
FGF pathway–amplified subpopulation, and
FGF pathway–nonamplified subpopulation. The HR of PFS in the full population was estimated using a Cox proportional hazards model stratified by
FGF pathway amplification status and presence of visceral disease (yes vs no).
Efficacy of dovitinib plus fulvestrant over placebo plus fulvestrant was established if the estimated HR was <0.68 for the full population (i.e., improvement of approximately 3.0 months in median PFS) or <0.65 for the FGF pathway–amplified subpopulation (i.e., improvement of approximately 3.5 months in median PFS). Futility criteria in the FGF pathway–nonamplified subpopulation was determined if the posterior probability (HR >0.81) was >50% (i.e., improvement of <1.5 months in median PFS). The number of PFS events needed for the final analysis was calculated by assuming a 10% prevalence of FGF pathway amplification and a median PFS of 6.5 months with fulvestrant and placebo. To achieve the required number of PFS events for the final analysis (≥90 in the full population and ≥50 in the FGF pathway–amplified subgroup, whichever occurred later), a total of 150 patients had to be randomized as follows: 75 patients per treatment arm (45 FGF pathway–amplified and 30 FGF pathway–nonamplified).
Separate interim analyses were planned for patients with and without FGF pathway amplifications, owing to the faster enrollment expected for the FGF pathway–nonamplified subgroup. The first interim analyses occurred when 36 PFS events had been documented in the FGF pathway–nonamplified subgroup, and the second interim analysis occurred when ≥10 (20%) of 50 PFS events had been documented in the FGF pathway–amplified subgroup. The intent of these interim analyses was to assess the efficacy or futility of the study treatment. If the futility criteria were met (HR >0.81 in the first interim analysis; HR >0.7 in the second interim analysis), the study could be terminated early by the data monitoring committee.
The key secondary endpoint, ORR, was summarized as a percentage rate with 95% CI. OS was estimated using Kaplan-Meier analysis for each treatment arm; patients still alive at the time of analysis were censored at the last contact date.
Discussion
In this randomized, double-blind trial, we evaluated the safety and efficacy of dovitinib plus fulvestrant compared with placebo plus fulvestrant in postmenopausal patients with HR+, HER2− advanced breast cancer that progressed during or after prior endocrine therapy. The final analysis was initially planned to occur when 90 PFS events were recorded in the full population, including ≥50 PFS events in the FGF-amplified subgroup. However, the study was terminated early because of slow enrollment in the FGF-amplified subgroup.
In this study, patients in the
FGF pathway–amplified subgroup who received dovitinib plus fulvestrant had prolonged median PFS (10.9 vs 5.5 months), with an estimated 36% risk reduction compared with patients who received placebo plus fulvestrant. However, a similar trend in risk reduction with dovitinib plus fulvestrant treatment (vs placebo plus fulvestrant) was seen in all patients (32%) and in patients without
FGF pathway amplification (31%). This suggests that dovitinib plus fulvestrant may have antineoplastic activity regardless of
FGF pathway amplification status in the evaluated patient population, although the estimated risk reduction reached statistical significance (as defined in the study protocol) only in the
FGF pathway–amplified cohort. Furthermore, patients in the dovitinib plus fulvestrant arm had a higher ORR than patients in the placebo plus fulvestrant arm (27.7% vs 10.0%), regardless of
FGF pathway amplification status (ORR 20.0% vs 12.5% in
FGF pathway–amplified subgroup; ORR 31.3% vs 8.8% in
FGF pathway–nonamplified subgroup). Nevertheless, these data should be interpreted cautiously. First, the small sample size in the
FGF pathway–amplified subgroup contributed to a lower-than-expected number of PFS events and very large CIs. Second, we cannot exclude that dovitinib had an effect regardless of
FGF pathway amplification status, given that the number of events for the full study population was 64 (30 in the dovitinib arm and 34 in the placebo arm), which was less than the 90 planned events. One potential explanation for the activity of dovitinib plus fulvestrant regardless of
FGF pathway amplification status is that, as a multitargeted tyrosine kinase inhibitor, dovitinib targeted other pathways [
36], such as signaling through vascular endothelial growth factor receptor or c-Kit, which are overexpressed in 10% to 11% and 11% to 17% of breast cancers, respectively [
39,
40].
Safety data were consistent with the known safety profile of dovitinib [
38,
41‐
44], with no new safety concerns identified with the use of dovitinib in combination with fulvestrant in patients with HR
+, HER2
− advanced breast cancer. The use of FGFR inhibitors in breast cancer merits further investigation because other studies of single-agent FGFR inhibitors showed encouraging results in patients with breast cancer [
38,
45‐
47]. Resistance to hormone therapy (i.e., tamoxifen) is potentially mediated by FGFR signaling through activation of the mitogen-activated protein kinase (MAPK) and PI3K pathways [
26,
32]. For example, resistance to tamoxifen has been associated with constitutive activation of MAPK and the subsequent expression of cyclin D1 in
FGFR1-amplified breast cancer cell lines [
26]. Similarly, in ER
+ cell lines, activation of
FGFR3 reduced sensitivity to tamoxifen and fulvestrant through activation of MAPK and PI3K signaling pathways [
32]. Furthermore, the combination of dovitinib and the dual PI3K/mTOR inhibitor dactolisib (BEZ235) showed strong inhibition of PI3K pathway activation in vitro and in vivo, as well as antitumor activity in
FGFR-expressing breast cancer models [
48]. Currently, researchers in a phase Ib trial (ClinicalTrials.gov identifier: NCT01928459) are investigating the pan-FGFR inhibitor BGJ398 in combination with the selective PI3K inhibitor alpelisib (BYL719) in patients with solid tumors with
FGFR1,
FGFR2, and
FGFR3 alterations and phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit α mutations. Thus, further exploration of the use of FGFR inhibitors in combination with other agents is warranted.
The present study was terminated early because of slow accrual. The rates of
FGF pathway amplification observed in this study were lower than previously reported. In the previous phase II monotherapy study, 10% of the patients screened who were enrolled in the study had
FGFR1 amplification [
38], whereas approximately 5% of patients screened and randomized in this study had
FGF pathway amplifications. In the phase II monotherapy study, many of the patients were prescreened by the French cooperative group, which reduced the overall number of patients needed to be screened. In addition, the eligibility criteria allowed more heavily pretreated patients to be enrolled, which expanded the pool of potential patients. Conducting clinical trials in molecularly selected patient populations is challenging, particularly because the screening failure rate is high with current trial designs [
49], accrual can be slow when the molecular aberration is very rare, and patient dropout rates [
50] and costs [
51] can be high. In this study, accrual of
FGF-amplified patients was very slow and resulted in the early termination of the trial, thereby confounding interpretation of the results. Several strategies have been proposed to overcome these challenges. A novel idea to increase rapid recruitment of patients with rare molecular markers is to develop molecular screening programs that screen several genes at a time in a large patient pool, using next-generation sequencing assays, and then to guide patients to specific clinical trials on the basis of their specific biomarkers [
52,
53]. One example is the National Cancer Institute’s Molecular Analysis for Therapy Choice (NCI MATCH) trial (ClinicalTrials.gov identifier: NCT02465060); patients are screened for approximately 200 genes, assigned to a study arm on the basis of a molecular abnormality, and followed for response and PFS [
52]. The NCI MATCH study currently includes 24 arms, in one of which investigators are evaluating the FGFR inhibitor AZD4547 in patients with FGFR pathway aberrations (
FGFR1–FGFR3 amplification, mutation, or translocation). Following progression, patients may be rescreened and enrolled in a second study arm; patients may also receive their screening results and decide, together with their doctor, to receive alternative therapy [
52]. It remains to be seen whether new trial designs will have widespread support [
50].
Acknowledgements
We thank Alejandro Yovine, Seven Gogov, Andrea Kay, Ying Wang, and Beth McGrain for their involvement in the study design, statistical analysis, and data analysis. Editorial assistance was provided by Pamela Tuttle, PhD, CMPP, and Katherine Mills-Luján, PhD, of ArticulateScience LLC and was funded by Novartis Pharmaceuticals. This study was previously reported in abstract and poster format at the American Society of Clinical Oncology 2012 and 2013 congresses in Chicago, IL, USA, and at the American Association for Cancer Research San Antonio Breast Cancer Symposium in San Antonio, TX, USA, in 2012.
Authors’ contributions
The study was designed by the sponsor (Novartis Pharmaceuticals) and by the study steering committee (chair: FA; members: AM, MC, PN, ND, CHB, JC, and KB). MS, SD, and MMS contributed to the design of the study. As members of the study steering committee, FA, AM, MC, PN, ND, CHB, JC, and KB oversaw the conduct of the study. HS, ZK, and HB contributed substantially to patient recruitment. AM, MC, PN, ND, CHB, JC, KB, HS, ZK, HB, and FA contributed to data collection. AM, MC, PN, ND, JC, KB, HS, MS, YZ, MMS, and FA analyzed and interpreted the data. YZ performed the statistical analyses. AM and FA wrote the manuscript with medical editorial support from ArticulateScience LLC, funded by the sponsor. All authors contributed to draft revisions, had full access to the data, attest to the accuracy and integrity of the data, and read and approved the final manuscript.
Competing interests
FA and CHB have received research funding from AstraZeneca (London, UK), Eli Lilly and Co. (Indianapolis, IN, USA), Novartis Pharmaceuticals, and Pfizer (New York, NY, USA). CHB has also received research funding from AbbVie (Chicago, IL, USA), Abraxis BioScience (Los Angeles, CA, USA), Amgen (Thousand Oaks, CA, USA), Asana BioSciences (Lawrenceville, NJ, USA), BioMarin (Novato, CA, USA), Boehringer Ingelheim (Ingelheim am Rhein, Germany), Bristol-Myers Squibb (New York, NY, USA), Daiichi Sankyo (Tokyo, Japan), GlaxoSmithKline (Brentford, UK), Merck (Kenilworth, NJ, USA), Merrimack Pharmaceuticals (Cambridge, MA, USA), Mylan (Canonsburg, PA, USA), Roche/Genentech (South San Francisco, CA, USA), Sanofi (Paris, France), and Taiho Pharmaceuticals (Tokyo, Japan). ND has received research funding from Genentech, GTx (Memphis, TN, USA), and Novartis. CHB and MC have been consultants for and received honoraria from Novartis Pharmaceuticals and Pfizer. CHB has also been a consultant for and/or received honoraria from AstraZeneca, Boehringer Ingelheim, GlaxoSmithKline, Roche/Genentech, Eisai (Tokyo, Japan), and Samsung Bioepis (Incheon, Republic of Korea). JC has been a consultant for AstraZeneca, Biothera Pharmaceuticals (Eagan, MN, USA), Celgene (Summit, NJ, USA), Cellestia Biotech (Basel, Switzerland), and Roche/Genentech, and he has received honoraria from Eisai, Novartis Pharmaceuticals, and Roche/Genentech. KB has been a consultant for and has received funding from Novartis Pharmaceuticals. HS has been a consultant for AstraZeneca, Biotheranostics (San Diego, CA, USA), and Celgene. MS, YZ, SD, and MMS are current or former employees of Novartis Pharmaceuticals. AM, PN, ZK, and HB declare that they have no competing interests.