National Surgical Adjuvant Breast and Bowel Project (NSABP) Foundation Annual Progress Report: 2011 Formula Grant

Transcription

1 National Surgical Adjuvant Breast and Bowel Project (NSABP) Foundation Annual Progress Report: 2011 Formula Grant Reporting Period July 1, 2012 June 30, 2013 Formula Grant Overview The NSABP Foundation received $851,360 in formula funds for the grant award period January 1, 2012 through December 31, Accomplishments for the reporting period are described below. Research Project 1: Project Title and Purpose Markers and Mechanisms of Trastuzumab Resistance and Cardiotoxicity The purpose of this project is to identify molecular changes associated with treatment failure in Her2-positive breast cancer patients treated with trastuzumab and chemotherapy. Specifically, specific molecular changes that have been implicated in preclinical models to be responsible for trastuzumab treatment failure will be investigated. Identification of molecular changes that are associated with treatment failure helps to identify those patients who may need additional treatment and may help to identify those pathways that are most critical to trastuzumab response and to understanding treatment success as well as failure. Anticipated Duration of Project 1/1/ /31/2014 Project Overview The broad research objective is to improve treatment of breast cancer patients. We propose the following specific aims: 1) Identify DNA sequence alterations associated with resistance to trastuzumab chemotherapy in Her2-positive (+) breast cancer; 2) Identify DNA sequence alterations associated with trastuzumab cardiotoxicity; 3) Determine the molecular mechanism responsible for acquired trastuzumab resistance in node-positive, Her2 (+) breast cancer patients. Several preclinical models implicate specific molecules and pathways as responsible for trastuzumab resistance. Resistance to trastuzumab can be inherent, meaning that the tumor does not respond to treatment, or it can be acquired, meaning that the tumor initially responds to treatment, and then later recurs. We propose to explore both types of resistance. To explore acquired resistance mechanism(s), we will examine molecular/genetic changes between primary and recurrent tumors in patients treated with trastuzumab and chemotherapy and in patients given NSABP Foundation Formula Grant Page 1

2 only chemotherapy. The changes to be investigated are ones that have been implicated in preclinical models as responsible for trastuzumab or other targeted antibody resistance. Changes that appear only in recurrent tumors from patients treated with trastuzumab and not in recurrent tumors from those given only chemotherapy would be likely candidates to be at least partly responsible for trastuzumab-acquired resistance. To explore inherent resistance mechanisms and to predict associated cardiotoxicity, we will examine white blood cells for single nucleotide polymorphisms (SNPs) for FCγ receptors and ERBB2 in all B-31 patient samples available for analysis, approximately 1400 cases. Previous reports indicated that certain polymorphisms in FCγ and ERBB2 receptors were associated with response to trastuzumab and to trastuzumabassociated cardiotoxicity, respectively. Our goal is to validate or dispute this initial finding by using all available cases in NSABP clinical trial B-31, which established the effectiveness of trastuzumab in early-stage breast cancer. If such results were conclusive, they would provide for a simple test to determine the responsiveness of breast cancer patients to trastuzumab before treatment is given and could suggest that other therapies be tried. Principal Investigator Katherine Pogue-Geile, PhD Assistant Director of Molecular Profiling NSABP Foundation, Inc Federal St Suite 303 Pittsburgh PA Other Participating Researchers Patrick Gavin, BS employed by the NSABP Foundation, Inc. Expected Research Outcomes and Benefits The expected research outcomes will determine whether specific SNPs are associated with trastuzumab resistance or cardiotoxicity. This information may benefit patients with Her2- positive breast cancer by providing information that could improve the selection of therapy. These SNP tumor markers could identify patients who are likely to be resistant to trastuzumab or who may suffer serious cardiotoxicity. Patients with trastuzumab-resistant tumors could be treated with other Her2-targeted therapies, such as lapatinib. Patients who are likely to suffer from cardiotoxicity from the trastuzumab/anthracycline regimen, which is the most effective treatment, may opt for a trastuzumab/carbotaxol regimen, which is not associated with cardiotoxicity. Other expected research outcomes will include answers to several important questions concerning trastuzumab response and resistance. On the basis of preclinical models, many mechanisms have been proposed to explain trastuzumab resistance but none of these proposed mechanisms have been shown to be relevant in women. DNA sequence and RNA expression analyses of the genes that have been proposed to be responsible for acquired resistance to trastuzumab may provide potential biomarkers to monitor the success or failure of trastuzumab treatment. NSABP Foundation Formula Grant Page 2

3 Summary of Research Completed The milestones for the reporting period were: Run pilot utilizing the Ion Torrent and the custom primers with cell-line DNA samples Anonymize recurrent cohort DNA samples using honest broker Begin targeted resequencing of recurrent and matching samples Begin extracting DNA from white blood cells (WBC) and from FFPE tissues We have completed the following milestone: Run a pilot utilizing the Ion Torrent and custom primers with cell line DNA and began extracting DNA. No other milestones could be worked on until approval of tissues and blood was completed, which we describe below. Approval was received at the end of April and DNA extraction began at that time. I. Progress for Specific Aims 1) Identify DNA sequence alterations associated with resistance to trastuzumab chemotherapy in Her2-positive (+) breast cancer and 2) Identify DNA sequence alterations associated with trastuzumab cardiotoxicity: Submission and approval for the use of B-31 samples To address aims 1 and 2, we have developed a written proposal for acquiring proper regulatory approval for the use of B-31 blood and tissue samples. The National Cancer Institute (NCI) through the Cancer Therapy Evaluation Program (CTEP) requires that a proposal be written and approved by the NCI for the use of any tissues from studies that were supported at any level with NCI funds. Thus, we have spent much of this report period developing, submitting, and obtaining approval of our protocol titled "Study of SNP markers associated with trastuzumab resistance and cardiotoxicity." The protocol was approved on April 26, As part of protocol development, we have verified the number of specimens for use in our studies through the NSABP Biostatistical Center. For the purpose of testing genotyping of the FCGR (FCγ receptor) genes and benefit of trastuzumab, there are 1698 available B-31 baseline samples from the ACT group (n = 828) or ACTH group (n = 870). Only eligible patients with follow-up who gave consent for Her2 research are included in this analysis. However, we have peripheral blood mononuclear cells (PBMCs) for only 1305 cases because 393 blood samples were compromised during storage at the Baylor Serum Bank facility due to a flood. Thus, for these compromised samples we will use formalin-fixed paraffin-embedded tissue (FFPET). Among the 1305 patients with PBMC samples there were 640 in the ACT group, and 665 in the ACTH group. The 393 FFPE tumor blocks include 188 in the ACT group and 205 in the ACTH group. To test the ERBB2 genotype and tratuzumab-associated cardiotoxicity, there are 776 patients in the ACTH group with available PBMC samples. Among them, 593 have non-compromised samples and 183 have been compromised, so FFPE tumor blocks will be used instead. Power calculations were done for the association study of FCGR2A-158 genotype with trastuzumab benefit. The power was calculated with an α = 0.05 (two-sided), N=1528, Accrual time = 5 years, follow up = 5 years, and assumed genotype frequency to have the value (50%, 60%), λ 11 to have values of (0.01, 0.015, 0.02, 0.025, 0.03, 0.035, 0.04, 0.045, 0.05) and λ 21 to have values of (0.01, 0.015, 0.02, 0.025, 0.03); values with power >.4 are shown in Table 1. The SWOG webtool ( was used for power NSABP Foundation Formula Grant Page 3

4 calculation. Power calculations were done for the association of the ERBB2-655 genotype with trastuzumab-associated cardiotoxicity (CD) and are shown in Table 2. The maximum sample size is 776 assuming assay success rate is 0.9, N is estimated to be 698. In our analysis, we assumed genotype frequency to have the value (35%, 45%) based on dbsnp database. Other assumptions included an event rate of 18%. Power was calculated with the above assumptions using the 'epicalc' package implemented in R. As shown in Table 2, for a two-sided test, at significance level 0.05, and a genotype frequency of V/V and V/I at 0.35, and a medium odds ratio (OR) of 1.70, the power is 72%. The power is about 83% when OR is 1.84 and 91% when OR is Additionally, it was necessary to obtain internal approval for the method we would use to further anonymize our samples after DNA was isolated. Consent from B-31 patients was obtained for the use of their samples for heritable studies only if the patients were re-consented or if the samples were completely anonymized. For this reason, it was necessary to map out the anonymization process. This process was approved internally by the NSABP Assistant Chair, and NSABP Directors of Regulatory Affairs, the Pathology Laboratory, and the Biostatistical Center. We have now received approval of our anoymization strategy. A list of the PBMC samples has been sent to the Baylor Serum Bank for retrieval. A list of the formalin-fixed, FFPET samples was sent to the NSABP Pathology Laboratory; the PI has received these blocks and isolation of DNA has begun. Experimental Progress We have tested a new kit for the isolation of DNA from PBMCs to optimize our isolation procedure based on the number of cells contained within the samples. We initially planned to use the EZ 96 Blood DNA kit from Omega Bio-Tek for the DNA isolation, but the maximum number of cells that can be used for this procedure is 6 X However, the number of cells in the B-31 PBMCs collected from 10 mls of blood may be greater than 6 X 10 6 cells. We counted the number of PBMCs in a 10-ml blood sample from a commercial vendor and found it contained 38 million cells six times more than what can be used for the Omega Bio-Tek kit. Therefore, we tested a different kit, the Qiagen QIAmp DNA Blood Maxi Kit for PMBC DNA isolation. DNA yields from 10 mls of blood were 136ug, which is much more than we will need for this proposal. II. Progress for Specific Aim 3) Determine the molecular mechanism responsible for acquired trastuzumab resistance in node-positive, Her2 (+) breast cancer patients: For this specific aim, we proposed to conduct sequence analysis of 10 different genes. In order to carry out this specific aim, we needed to be able to determine the sequence of entire genes rather than only at hot spots and so would not be able to use the Sequenom platform (it is only able to interrogate hot spot mutations). The MiSeq Personal Sequencer instrument from Illumina and the Ion [Torrent] Personal Genome Machine (PGM TM ) Sequencer from Life Technologies were considered to be 2 instruments that would provide the appropriate through-put for this project. Furthermore, each company provided targeted sequencing kits for the interrogation of common cancer mutations. The TrueSeq Amplicon Cancer Panel from Illumina and the AmpliSeq Cancer Panel from Life Technologies both interrogate the same 48 common genes and in similar amplicons. Thus, we tested both platforms in a pilot study to determine which of these instruments would work best for DNAs isolated with FFPET. In these comparisons the recommended variant caller settings recommended by the manufacturer were used. NSABP Foundation Formula Grant Page 4

5 We prepared libraries for targeted sequencing analysis using FFPE DNAs from a small number of NSABP B-31 samples because it was important to demonstrate that DNAs isolated from routinely processed FFPET could be sequenced with these platforms and to determine the quantity of DNA required. Libraries were prepared following the recommendations of the manufacturer with the exception that with regard to the TruSeq libraries we attempted to use lower amounts of DNA than the recommended 250ng because the amount of tumor cells in B-31 recurrent tumor blocks is very small and the amount of DNA required for sequence analysis is a major consideration in the decision for the sequencing platform. Thus, libraries for the MiSeq instrument were built using 10ng (n=4), 50ng (n=2), 75ng (n=2), 150ng (n=2), 150ng (n=2), and 250ng n=2 of DNA, to determine the minimum amount of DNA needed. While the libraries were constructed in the NSABP laboratory, the DNA sequencing was done as a separate service. The number of variants detected with TruSeq panel using 10ng and 250ng of the same DNA was 581 and 60 (Fig.1). The much larger number of variants detected with 10ng versus 250ng suggested that the MiSeq data yielded a very large number of false-positive mutations with 10ng. In contrast, the AmpliSeq panel yielded considerably fewer variants with 10ng (between 13 and 7 per sample and a total of 39 in 4 different samples). Two of these samples were also run on the MiSeq using the TruSeq panel (at 250ng). There was complete concordance between the 2 platforms with the exception of a variant detected at 7% on the Ion and not on the MiSeq. The same 14 variants were detected at similar frequencies on both platforms with the exception that one variant detected at 7% on the Ion was not detected with MiSeq and one other variant was not detected with MiSeq as it was not included in the TruSeq panel. The 14 variants included 13 SNPs and one PIK3CA mutation. Thus, at 10ng the AmpliSeq panel run on the PGM sequencer seems to yield a more reasonable number of variants, which were confirmed with MiSeq at 250ng. We have compared the results from the AmpliSeq panel to the Sequenom single-base-pairextension reactions obtained on the same samples. In this experiment the AmpliSeq results obtained from a collaborator were compared to results obtained with Sequenom platform run in our laboratory. The Sequenom panel assayed PIK3CA hot spot mutations in 150 breast cancer samples. Both of these instruments give frequency values for the prevalence of a specific mutation within a sample. The frequency of mutations in cancer tissues rarely represent 50% of alleles as would be expected of one mutant allele in a cell line which has a normal copy number for the gene. The prevalence of a somatic mutation in cancer tissue varies due to the heterogeneity of the cancer itself and of the tissue sample, which often contains some normal tissue. We have compared the frequency of PIK3CA mutations in 35 different breast cancer patients using the Sequenom and Ion Torrent platforms. There was reasonably good correlation (R 2 =0.81) between the 2 different sequencing methodologies based the frequency of the mutations in PIK3CA (Fig. 2). While both the MiSeq and the Ion Torrent instruments appeared to perform adequately for sequence analysis of FFPE samples when the recommended amount of DNA was used, we chose to purchase the Ion Torrent PGM instrument from Life Technologies because it required only 10ng of DNA for sequence analysis. We have purchased and installed the PGM Sequencer and ran a pilot study of 10 samples using the AmpliSeqv2 [cancer] panel and our own PGM. To determine the reproducibility of the NSABP Foundation Formula Grant Page 5

6 instrument, we have prepared the libraries and run the same 10 samples at two different times. There is good concordance between the first and second runs in 8 of the 10 samples. In these 8 samples, between 14 and 18 total variants/sample were detected. Between 86% to 94% of the variants were detected in both runs; a representative sample is shown in in Fig. 3, Sample 1. However, 2 of the samples had a much larger number of variants (18 to 67). There was little concordance between the 2 runs (e.g., Sample 2, Fig. 3) and thus minimum overlap in variants among all 10 samples (Fig. 3). Examination of quality matrices did not point to any particular problem with the either of the 2 runs in these 2 samples. This lack of concordance is under investigation. Summary We have received CTEP approval for the use of B-31 samples and have made progress in the sequence analysis of FFPE DNA using the PGM Sequencer. Table 1. Power calculations for association of FCGR2A-158 genotype with trastuzumab benefit FCGRIIIA 158V/V, V/F frequency f 11 f 12 f 21 f 22 λ 11 λ 21 λ 12 λ Power 50% 24.4% 24.4% 25.6% 25.6% % 29.3% 19.5% 30.7% 20.5% NSABP Foundation Formula Grant Page 6

7 Table 2. Power calculations for ERBB2-655 genotype with trastuzumab-associated cardiotoxicity Frequency of ERBB2-655 V/V or V/I 35% 45% CD events for patients with ERBB2-655 V/V or V/I CD events for patients with ERBB2-655 I/I OR Power Number % Number % Variants detected using MiSeq and TruSeq panel PIK3CA variant frequency using Sequenom and PGM (Ion Torrent) platforms NSABP Foundation Formula Grant Page 7

8 NSABP Foundation Formula Grant Page 8