AUY922 in NSCLC Supplementary Material

AUY922 in NSCLC Supplementary material

Patients continue to receive AUY922 70 mg/m2 as monotherapy in 21-day treatment cycles, until RECIST-defined disease progression, unacceptable toxicity, patient withdrawal, death, or discontinuation from the study for any other reason.

The study protocol and all amendments were reviewed and approved by the Institutional Review Board for each investigational site and written informed consent was obtained from each patient prior to the initiation of any study-related procedures. The study was conducted according to the ethics of the Declaration of Helsinki.

Patients

Eligible patients had histologically or cytologically confirmed, advanced (stage IIIB or IV) NSCLC and were grouped into 1 of the 5 strata based on tumor molecular etiology: ALK-rearranged, EGFR mutant, less pretreated EGFR cohort (EGFR < 2), KRAS mutant, and a stratum for all remaining NSCLC patients that did not have an activating mutation in the KRAS, ALK, or EGFR gene that was termed wild-type. All patients, except those in the EGFR < 2 stratum must have received > 2 prior lines of therapy. Protocol amendment was done to include EGFR < 2 stratum due to the observation that a high number of EGFR-mutant patients discontinued the study participation due to early disease progression prior to completing first radiological assessment (6 ±1weeks) reflecting the fact that most patients had advanced disease, brain metastasis, and had received multiple previous lines of therapies. Prior platinum-based treatment was not required for patients with EGFR-activating mutations and ALK rearrangements. In addition, patients with an EGFR mutation had to have received a prior EGFR tyrosine kinase inhibitor (TKI)-based treatment unless they had known denovo resistance to an EGFR TKI; there was no requirement of ALK TKI in patients with the ALK rearrangement as this was not the standard of care during most of the trial. Patients in the EGFR < 2 stratum were also required to have a baseline biopsy prior to the study treatment. Other key inclusion criteria included a World Health Organization (WHO) performance status ≤ 2 (≤ 1 in the EGFR < 2 stratum) and ≥ 1 measurable lesion at baseline as defined by Response Evaluation Criteria in Solid Tumors (RECIST), v1.0. Patients with asymptomatic or neurologically stable brain metastases at baseline were eligible (these patients were excluded from the EGFR < 2 mutant stratum).

Study Assessments

Efficacy Analysis

Tumor responses (CR/PR [ORR], or in absence of CR/PR a SD at 18 weeks, or PD) were assessed by computerized tomography or magnetic resonance imaging at screening/baseline (within 21 days prior to Cycle 1, Day 1) and at the end of every 2 treatment cycles (6 weeks ± 1 week).

The primary assessment of efficacy was performed using investigator-assessed radiologic data, with centrally reviewed radiologic data used in a supporting secondary analysis. The secondary endpoints in this study were PFS and OS as assessed by RECIST.

Patients who had not progressed, died, or withdrawn consent were required to undergo an 18-week assessment in order to provide survival or disease progression status.

Safety Analysis

Adverse events (AEs) were assessed according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE), v4.14

Pharmacokinetic Analysis

A full plasma PK profile was obtained from all patients on Day 1 of Cycle 1 to characterize the disposition of AUY922 and its metabolite BJP762 after 1 hour intravenous infusions. Plasma samples were assayed for AUY922 and BJP762 concentrations using validated liquid chromatography-tandem mass spectrometry.

Molecular Tumor Analyses

All patients had to provide archival or baseline tumor tissue for central analysis of its molecular status to enable correct allocation into one of the 5 study strata. Retrospective primary analysis was performed in clinical laboratory improvement amendments (CLIA)-certified setting at Foundation Medicine (FM), Cambridge, MA, wherein the samples were tested for mutations, copy number alterations, and rearrangements in 189 known cancer-related genes using next-generation sequencing (NGS).

Specifically, biotinylated RNA “baits” and streptavidin-coated magnetic beads were used to extract sequences of the exons of interest from the SureSelect buffer. Washed, digested, and amplified cDNA was subsequently sequenced using an Illumina HiSeq 2000 instrument. The analysis pipeline at FM used Bayesian analysis for point mutations, local assembly for short insertions and deletions, process-controlled matched normal for copy number alterations, and analysis of chimeric reads for rearrangements and translocations. Detected genomic lesions were subsequently filtered based on state-of-the-art public resources, such as Catalogue of Somatic Mutations in Cancer (COSMIC). The subsequent analyses focused only on the subset of functionally known and likely alterations which were performed in R programming language.

There are multiple classifications and nomenclatures that describe specific junctions between a particular exon of ALK and an exon of EML4 in ALK+EML4 fusions. These naming conventions emerged with the availability of NGS data. For example “variant 1” or V1 refers to a fusion of ALK sequence beginning at exon 20 to EML4 ending after exon 13. To avoid ambiguity, whenever EML4-ALK variants are mentioned, the following convention will be used: “V1 – [E13-A12]”. To calculate the exon breakpoint on the locations obtained through NGS, specific genomic coordinates of ALK transcript (NM_004304) and EML4 transcript (NM_019063) and their exon coordinate annotation based on the reference human genome (HG19) are available at and accessible using the R/BioConductor packages (BSgenome.Hsapiens.UCSC.hg19 and GenomicFeatures), to obtain the per-sample variant classification for the samples where a particular variant of ALK+EML4 was detected by NGS.

Statistical Analysis

A sample size of 140 patients (30 patients in each of the following 4 strata: EGFR mutant, EGFR < 2, KRAS mutant, and wild type; and 20 patients in the ALK-rearranged stratum) was planned with the primary efficacy endpoint analyzed using a Bayesian, partially exchangeable, multinomial model.

The sample size is based on the operating characteristics of the Bayesian partially exchangeable multinomial model.

Two scenarios were considered. One in which AUY922 treatment was efficacious in all 5 strata, with response (R, complete or partia]), stable disease at 18 weeks (SD18w), and no clinical benefit (NCB rates) corresponding to H1 (ie, R = 0.10, SD18w = 0.30, NCB = 0.60 for the exchangeable strata, R = 0.20, SD18w = 0.40, NCB = 0.40 for the ALK-rearranged stratum, and R = 0.15, SD18w = 0.30, NCB = 0.55 for the EGFR < 2 mutant stratum), and a second scenario in which AUY922 treatment was nonefficacious in all 5 strata with R, SD18w, and NCB rates corresponding to H0 (ie, R = 0.05, SD18w = 0.10, NCB = 0.85).

The sample size was chosen such that:

1.If AUY922 treatment is efficacious (under first scenario above):

-For the exchangeable strata P (R > 0.075 or NCB < 0.65) ≥ 0.7 for at least one stratum is > 0.7

-For the ALK-rearranged translocation stratum P (R > 0.175 or NCB < 0.45) ≥ 0.7 is > 0.7

-For the EGFR < 2 mutant stratum P (R > 0.125 or NCB < 0.60) ≥ 0.7 is > 0.7

2.If AUY922 treatment is nonefficacious (under second scenario above):

-For the exchangeable strata P (R > 0.075 or NCB < 0.65) ≥ 0.7 for at least one stratum is < 0.06

-For the ALK-rearranged translocation stratum P (R > 0.175 or NCB < 0.45) ≥ 0.7 is < 0.06

-For the EGFR < 2 mutant stratum P (R > 0.125 or NCB < 0.60) ≥ 0.7 is < 0.02

The EGFR mutant, KRAS mutant, and wild type strata (without the ALK-rearranged stratum) were considered exchangeable to allow for the a priori specification of those strata that were expected to show similar efficacy, but also allowed for the a priori expectation of differences between strata to be acknowledged. ALK-rearranged and EGFR < 2 strata were considered nonexchangeable with all other study strata. For each stratum, joint posterior distribution of probability of response and NCB was derived.

Median PFS, median OS, and the PFS and OS rates at fixed time points were estimated using Kaplan–Meier methods.

PK parameters were estimated from plasma drug concentration–time data by noncompartmental analysis using Phoenix®, v6.2.

Supplementary Table 1. Bayesian Analysis of Objective Response Rate and no Clinical Benefit

ALK-rearranged
N=22 / EGFR
mutant
N=35 / EGFR2 mutant
N=31 / KRAS
mutant
N=28 / Wild type N=34
Observed
ORR / NCB / 32% / 59% / 17% / 74% / 10% / 68% / 0% / 93% / 9% / 82%
Probability of improved efficacy / 0.899 / 0.832 / 0.185 / 0.036 / 0.346
Improved efficacy threshold / ORR≥20%
or
NCB≤40% / ORR≥10%
or
NCB≤60% / ORR≥15%
or
NCB≤55% / ORR≥10%
or
NCB≤60% / ORR≥10%
or
NCB≤60%

NCB, no clinical benefit; ORR, overall response rate; PD,progressive disease

Supplementary Table 2.Summary of Best Overall Response as per Investigator Assessment by Study Stratum

ALK-rearranged
N=22 / EGFR
mutant
N=35 / EGFR2 mutant
N=31 / KRAS
mutant
N=28 / Wild type N=34 / Unknown
N=3 / All
patients
N=153
CR, n (%) / 0 / 0 / 0 / 0 / 0 / 0 / 0
PR, n (%) / 7 (31.8) / 6 (17.1)* / 3 (9.7)† / 0 / 3 (8.8) / 1 (33.3) / 20 (13.1)
SD, n (%) / 6 (27.3) / 14 (40.0) / 19 (61.3) / 10 (35.7) / 11 (32.4) / 0 / 60 (39.2)
PD, n (%) / 9 (40.9) / 13 (37.1) / 9 (29.0) / 16 (57.1) / 19 (55.9) / 2 (66.7) / 68 (44.4)
Unknown, n (%) / 0 / 2 (5.7) / 0 / 2 (7.1) / 1 (2.9) / 0 / 5 (3.3)
ORR, n (%)
(95% CI) / 7 (31.8)
(13.9%,
54.9%) / 6 (17.1)
(6.6%,
33.6%) / 3 (9.7)
(2.0%,
25.8%) / 0 (0.0)
(0%,
12.3%) / 3 (8.8)
(1.9%,
23.7%) / 1 (33.3)
(0.8%,
90.6%) / 20 (13.1)
(8.2%,
19.5%)
DCR, n (%)
(95% CI) / 13 (59.1)
(36.4%, 79.3%) / 20 (57.1)
(39.4%, 73.7%) / 22 (71.0)
(52.0%, 85.8%) / 10 (35.7)
(18.6%, 55.9%) / 14 (41.2)
(24.6%, 59.3%) / 1 (33.3)
(0.8%, 90.6%) / 80 (52.3)
(44.1%, 60.4%)

CI, confidence interval; CR, complete response; DCR, disease control rate; NCB, no clinical benefit; ORR, overall response rate; PD, progressive disease; PR, partial response; SD, stable disease

*includes 1 patient with unconfirmed response

†includes 2 patient with unconfirmed responses

Supplementary Table 3.Primary Pharmacokinetic Parameters for Plasma AUY922 and its Metabolite BJP762 Following 1 Hour Infusion at 70 mg/m2 in NSCLC Patients

Pharmacokinetic parameter / n / Mean (SD) / Geometric Mean / Median
[Min; Max]
AUY922
AUC0-∞,ng*h/mL / 119 / 2101 .27 (990.320) / 1950.53 / 1916.89
[955.2; 8846.5]
AUC0-last, ng*h/mL / 139 / 1997.04 (894.903) / 1862.28 / 1860.42
[903.6; 8168.4]
Cmax, ng/mL / 144 / 1130.59 (705.177) / 1002.76 / 1035.00
[79.3; 7150.0]
BJP762
AUC0-∞,ng*h/mL / 117 / 13963.07 (13006.537) / 10183.29 / 9420.9
[2037.4; 66689.2]
AUC0-last, ng*h/mL / 142 / 12496.75 (12035.525) / 8978.28 / 8815.6
[1447.0; 66516.9]
Cmax, ng/mL / 144 / 2332.86 (1377.350) / 1990.03 / 2130.0
[454.0; 7840.0]