Phase II Study of AZD4547 in Patients With Tumors Harboring Aberrations in the FGFR Pathway: Results From the NCI-MATCH Trial (EAY131) Subprotocol W

Abstract

Purpose: NCI-MATCH is a nationwide, histology-agnostic, signal-finding, molecular profile-driven trial for patients with refractory cancers, lymphomas, or myelomas. Patients with tumors harboring actionable aberration(s) in fibroblast growth factor receptor (FGFR) 1-3 were treated with AZD4547, an oral FGFR1-3 inhibitor.

Methods: Patients' tumors were screened by next-generation sequencing for predefined FGFR amplification, activating mutations, or fusions. Patients were treated with AZD4547, 80 mg orally twice daily until progression of disease or drug intolerance. A response rate of 16% was considered promising.

Results: Between July 2016 and June 2017, 70 patients were assigned and 48 received protocol therapy and are eligible for analysis. Patients' tumors harbored FGFR1 or FGFR2 amplification (n = 20), FGFR2 or FGFR3 single-nucleotide variants (n = 19), or FGFR1 or FGFR3 fusions (n = 9). The most common primary tumors were breast (33.3%), urothelial (12.5%), and cervical cancer (10.4%).Grade 3 adverse events were consistent with those described in previous clinical trials. Confirmed partial responses were seen in 8% (90% CI, 3% to 18%) and were observed only in patients whose tumors harbored FGFR1-3 point mutations or fusions. Stable disease was observed in 37.5% (90% CI, 25.8% to 50.4%). The median progression-free survival (PFS) was 3.4 months, and the 6-month PFS rate was 15% (90% CI, 8% to 31%). For patients with tumors harboring FGFR fusions, the response rate was 22% (90% CI, 4.1% to 55%), and 6-month PFS rate was 56% (90% CI, 31% to 100%).

Conclusion: Preliminary signals of activity appeared to be limited to cancers harboring FGFR activating mutations and fusions, although AZD4547 did not meet the primary end point. Different FGFR somatic alterations may confer different levels of signaling potency and/or oncogene dependence.

Trial registration: ClinicalTrials.gov NCT02465060.

Conflict of interest statement

The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health, nor does mention of trade names, commercial products, or organizations imply endorsement by the U.S. government.

Figures

FIG 1.

Patient flow diagram of NCI-MATCH Subprotocol W. Reasons that patients did not start protocol therapy were patient refusal (n = 1) and screening laboratories and imaging outside pretreatment window (also ineligible on the basis of pretreatment laboratories, n = 1). Reasons that patients were ineligible for analysis were pretreatment serum creatinine and creatinine clearance outside of eligibility range (n = 1) and multiple separate primary cancers at study entry (n = 1). PD, progressive disease; PS, performance status.

FIG 2.

(A) Pie chart of the primary histologic types (type of malignancy) of the 48 eligible patients. (B) Donut chart of the genomic aberrations of the eligible patients. The inner circle shows the type of genomic aberration (FGFR amplification, single-nucleotide variant, and fusion), and the outer circle shows the specific aberration (genomic aberration). (*) Two fusions present.

FIG 3.

(A) Co-occurring mutations (left column) by histologic tumor type (lowest panel) and best response (lower panel) and their cumulative frequency (left side of the graph). Germline variants not subtracted. Note the high frequency of TP53 and PIK3CA mutations, which did not show a particular histologic predilection. Urothelial carcinomas harbored exclusively FGFR3 aberrations, whereas endometrial carcinomas most frequently had FGFR2 point mutations. (B) Focus on the co-occurrence of PIK3CA mutations and potential association with response. There was a significant association between progressive disease (PD) as best response to AZD4547 and co-occurrence of PIK3CA mutations (P = .03 by Fisher’s exact test). IHC, immunohistochemistry; NSCLC, non–small-cell lung cancer; PR, partial response; SCL, small-cell lung cancer; SD, stable disease; UE, unevaluable.

FIG 4.

(A) Waterfall plot (color coded by type of genomic aberration) of best confirmed response of target lesion(s) according to RECIST1.1 (n = 36). Seven patients were unevaluable and 5 had a new lesion (classified as progressive disease [PD]) without a measurement of their target lesion(s). Also note that 2 patients achieved a > 30% reduction of their target lesions but were coded as PD and stable disease (SD): the first case achieved a 36% reduction of her target lesion(s), but a new lesion was reported at the next scan, and the second case achieved a 42% reduction, but a new lesion appeared after cycle 3. (B) Swimmer’s plot (color coded by type of genomic aberration) of the 22 patients who achieved SD or better as their best response. The time when partial response (PR) was achieved and when disease progressed are indicated with green and yellow triangles, respectively. (C) Overall progression-free survival (PFS). Kaplan-Meier curve shows PFS for all patients eligible for analysis. (D) Kaplan-Meier curve shows PFS by genomic aberration (median 1.8, 3.6, and 10.0 months for amplification, SNV, and fusion, respectively). (*) New lesions. SNV, single-nucleotide variant.

FIG A1.

Lollipop plots. (A, B, D) Lollipop plots of the eligible mutations in (A) FGFR1, (B) FGFR2, and (D) FGFR3. Height on the y-axis is proportionate to the number of eligible base substitutions. (C, E) Lollipop plots of the mutations in (C) FGFR2 and (E) FGFR3 in the study population. Height on the y-axis is proportionate to number of patients harboring the respective mutation. Activating mutations of the FGFR1 are rare in cancer and were not seen in our patient population.

FIG A2.

Type and duration of medical therapies according to tumor type received before enrollment to NCI-MATCH Subprotocol W, as reported by the participating sites. Not shown are one patient with breast cancer who received 14 prior lines of therapy for whom the type and duration of treatments were not recorded, and one patient with salivary gland carcinoma who underwent prior surgery and radiation therapy but had not received any prior medical therapy. Day 0 is the day that the first medical therapy was started. The figure does not include and prior surgery and radiation therapy.

FIG A3.

Lines of therapy, including radiation therapy and surgery, as reported by the participating sites. One patient with cervical cancer harboring an FGFR3-TACC3.F17T8 gene fusion underwent prior surgery and radiation therapy and received 2 months of cisplatin approximately 8.5 years before enrollment.

FIG A4.

(A) Waterfall plot (color coded by tissue of origin) of best confirmed response of target lesion(s) according to RECIST (n = 36). Seven patients were unevaluable and 5 had a new lesion (classified as progressive disease [PD]) without a measurement of their target lesion(s). Also note that 2 patients achieved a > 30% reduction of their target lesions but coded as PD and stable disease (SD): the first case achieved a 36% reduction of her target lesion(s), but a new lesion was reported at the next scan; the second case achieved a 42% reduction, but a new lesion appeared after cycle 3. (B) Swimmer’s plot (color coded by tissue of origin) of the 22 patients who achieved SD or better as their best response. The time when partial response (PR) was achieved and when disease progressed are indicated with green and yellow triangles, respectively. (C) Bar plot of best confirmed response according to RECIST by specific genomic aberration. x-axis, number of patients. (D) Kaplan-Meier curve for the overall survival of the 48 patients eligible for analysis. (*) New lesions. NSCLC, non–small-cell lung cancer; SCL, small-cell lung cancer.