Abstract P1-01-10: Exome sequencing of circulating tumor cells in metastatic breast cancer

Aim: We interrogated whether Circulating Tumor Cells (CTCs) can complement metastatic biopsies for genomic analyses.

Patients and Methods: We compared single nucleotide variants (SNVs) and copy number aberrations (CNAs) identified using whole exome sequencing (WES) of DNA from frozen tumor tissue (primary/metastasis), amplified DNA from CTCs and normal DNA from 3 metastatic breast cancer (BC) patients (pts). All samples of the same patient were collected at the same timepoint. CTC isolation was performed using CellSearch and DEPArray systems followed by whole genome amplification (Ampli1 kit). WES was performed using the Illumina HiSeq2000 with 200X targeted coverage. Reads were aligned using bwa. SNVs had to be called by both Haplotype Caller (vs. reference genome) and Strelka (vs. paired normal). CNAs were determined by counting reads in 1MB windows and by comparing tumor/CTC samples with normal DNA. Pairwise concordance of CNAs profiles of different samples from the same patient was assessed using Spearman correlation (ρ). Significance of ρ differences between pts was obtained by Kruskal-Wallis test. Orthogonal validation for selected SNVs was performed.

Results: We studied 3 patients from the 3 major BC subtypes, patient (pt)1 with ER-/HER2+ BC (samples collected at diagnosis, initially metastatic disease), pt2 with triple-negative BC (samples collected 2 years from diagnosis) and pt3 with ER+/HER2- BC (samples collected 8 years from diagnosis).

We first compared tumor tissue and CTCs for SNVs. For pt1, of the 77 SNVs identified in the tumor, 51 were found on at least one of 12 CTCs samples. For pt2, of the 62 SNVs identified in the tumor, 19 were found on at least 1 of 11 CTCs samples. For pt3, of the 225 SNVs identified in the tumor, 48 were found on at least 1 of 3 CTCs samples. Interestingly, by increasing the number of CTCs analyzed, we increased the % of identified SNVs from synchronous tumor tissue. SNVs with high variant allele fraction (VAF) in tumor tissue were detected significantly more often in CTCs: 22% of the SNVs with VAFs <20% were found at least once, compared to 53% and 74% of SNVs with VAFs >20% and >40%, respectively (p=10-12, Fisher exact test).

Then, we compared tumor tissue and CTCs for CNAs. As time from diagnosis of metastatic disease to samples collection increased, we observed significantly higher heterogeneity within CTCs from the same patient (median ρ between CTCs was 86% for pt1, 84% for pt2 and 28% for pt3, p<0.01) and between CTCs and tumor tissue from the same patient (median ρ was 78% for pt1, 67% for pt2 and 21% for pt3, p<10-4). Interestingly, in pt3 one CTC was more similar to the metastasis than the other 2 (ρ of 53%, 21% and 21%). When a phylogenetic tree was constructed for pt3 by combining SNVs and CNAs data, three clones were identified: one clone with an AKT1 (E17K) and a TP53 (R248W) mutation and a 8p deletion, a second clone with the above profile plus an 8q amplification and a third clone with an AKT1 and an ESR1 (Y537N) mutation and 1p deletion. The metastasis was similar with the first clone.

Conclusions: These data suggest that tumor tissue and single CTC exome sequencing analyses provide complementary information to map tumor heterogeneity. Further validation for potential clinical applications is needed.

Citation Format: Ignatiadis M, Rothé F, Peeters D, Rouas G, Smeets D, Haan J, Lambrechts D, Campbell P, Piccart M, Voet T, Dirix L, Venet D, Sotiriou C. Exome sequencing of circulating tumor cells in metastatic breast cancer [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-01-10.

Abstract P6-07-02: CDK4 phosphorylation status and corresponding gene expression profile predict sensitivity to Palbociclib

Although the specific CDK4/6 inhibitor PD0332991 (Palbociclib) was recently approved by the FDA to treat advanced ER+ breast tumors, there is yet no reliable sensitivity prediction tool. Cyclin D-CDK4/6 are the first CDK complexes to be activated in G1 phase in response to oncogenic pathways. They phosphorylate and inactivate the central cell cycle/tumor suppressor pRb. CDK4 activity requires its binding to a cyclin D (CCND1-3 genes) with which INK4 CDK4 inhibitors such as p16 (CDKN2A-D genes) compete. Although the assembly of the CDK4-cyclin D complexes was considered to be the main level of CDK4 activity control, we have shown that the activating T172-phosphorylation of CDK4 is actually the central rate-limiting event that initiates the cell cycle decision and signals the presence of active CDK4.

Here, using 2D-gel electrophoresis to separate the modified forms of CDK4, we found in breast cancer cell lines that only the CDK4 T172-phosphorylation correlates with the sensitivity to PD0332991. The only exception was in the rare case of combined CCNE1 amplification and CDKN2A loss wherein combination of PD0332991 with a CDK2 inhibitor is required to block entry in the cell cycle. Additionally, three types of CDK4 modification profile were identified by 2D-gel electrophoresis in 56 breast tumors. In the first profile, the phosphorylated CDK4 was undetectable as in normal breast samples despite a high KI67 index. In the second and third profiles, the CDK4 phosphorylation was detectable and its intensity was either above or below 90% of the intensity of a second yet unidentified form of CDK4, respectively. The proportions of these profiles differ among breast tumors according to their clinic-pathological characteristics, molecular subtypes and risk. Finally, we identified a 11-gene expression signature that faithfully predicted the CDK4 modification profiles of breast tumors and cell lines (concordance rates of 84% and 100% in the 56 analyzed breast tumor samples or cell lines respectively). All three CDK4 modification profiles were evaluated in a merged independent dataset of 4034 published gene expression profiles. In these 4034 patients, 70% of triple-negative tumors, 18% of HER2-positive tumors and 5% of ER-positive tumors were predicted to have the first CDK4 profile wherein CDK4 phosphorylation is undetectable and to be completely unresponsive to CDK4 inhibitors. The phosphorylated CDK4 was predicted to be the major modified form in 26% of triple-negative tumors, 48% of HER2- positive tumors and 56% of ER-positive tumors. These patients should benefit the most from treatment with CDK4 inhibitors. Therefore, prediction of the CDK4 modification profile may allow extending treatment with Palbociclib to presently ineligible patients. As tumors with the third CDK4 modification profile generally present low grade and low OncotypeDX risks, the added value of including CDK4 inhibitors in their treatment compared to surgery and hormone therapy alone is questionable.

In conclusion, we identified CDK4 phosphorylation as the most direct biomarker of CDK4 inhibitor sensitivity in breast cancer and developed a promising 11-gene based surrogate marker to guide their use in the clinic.Although the specific CDK4/6 inhibitor PD0332991 (Palbociclib) was recently approved by the FDA to treat advanced ER+ breast tumors, there is yet no reliable sensitivity prediction tool. Cyclin D-CDK4/6 are the first CDK complexes to be activated in G1 phase in response to oncogenic pathways. They phosphorylate and inactivate the central cell cycle/tumor suppressor pRb. CDK4 activity requires its binding to a cyclin D (CCND1-3 genes) with which INK4 CDK4 inhibitors such as p16 (CDKN2A-D genes) compete. Although the assembly of the CDK4-cyclin D complexes was considered to be the main level of CDK4 activity control, we have shown that the activating T172-phosphorylation of CDK4 is actually the central rate-limiting event that initiates the cell cycle decision and signals the presence of active CDK4.

Here, using 2D-gel electrophoresis to separate the modified forms of CDK4, we found in breast cancer cell lines that only the CDK4 T172-phosphorylation correlates with the sensitivity to PD0332991. The only exception was in the rare case of combined CCNE1 amplification and CDKN2A loss wherein combination of PD0332991 with a CDK2 inhibitor is required to block entry in the cell cycle. Additionally, three types of CDK4 modification profile were identified by 2D-gel electrophoresis in 56 breast tumors. In the first profile, the phosphorylated CDK4 was undetectable as in normal breast samples despite a high KI67 index. In the second and third profiles, the CDK4 phosphorylation was detectable and its intensity was either above or below 90% of the intensity of a second yet unidentified form of CDK4, respectively. The proportions of these profiles differ among breast tumors according to their clinic-pathological characteristics, molecular subtypes and risk. Finally, we identified a 11-gene expression signature that faithfully predicted the CDK4 modification profiles of breast tumors and cell lines (concordance rates of 84% and 100% in the 56 analyzed breast tumor samples or cell lines respectively). All three CDK4 modification profiles were evaluated in a merged independent dataset of 4034 published gene expression profiles. In these 4034 patients, 70% of triple-negative tumors, 18% of HER2-positive tumors and 5% of ER-positive tumors were predicted to have the first CDK4 profile wherein CDK4 phosphorylation is undetectable and to be completely unresponsive to CDK4 inhibitors. The phosphorylated CDK4 was predicted to be the major modified form in 26% of triple-negative tumors, 48% of HER2- positive tumors and 56% of ER-positive tumors. These patients should benefit the most from treatment with CDK4 inhibitors. Therefore, prediction of the CDK4 modification profile may allow extending treatment with Palbociclib to presently ineligible patients. As tumors with the third CDK4 modification profile generally present low grade and low OncotypeDX risks, the added value of including CDK4 inhibitors in their treatment compared to surgery and hormone therapy alone is questionable.

In conclusion, we identified CDK4 phosphorylation as the most direct biomarker of CDK4 inhibitor sensitivity in breast cancer and developed a promising 11-gene based surrogate marker to guide their use in the clinic.

Citation Format: Raspe ES, Coulonval K, Pita J, Paternot S, Rothé F, Larsimont D, Van Laere S, Piccart M, Ignatiadis M, Sotiriou C, Roger PP. CDK4 phosphorylation status and corresponding gene expression profile predict sensitivity to Palbociclib [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P6-07-02.

Integrative proteomic and gene expression analysis identify potential biomarkers for adjuvant trastuzumab resistance: analysis from the Fin-her phase III randomized trial

Trastuzumab is a remarkably effective therapy for patients with human epidermal growth factor receptor 2 (HER2) - positive breast cancer (BC). However, not all women with high levels of HER2 benefit from trastuzumab.

By integrating mRNA and protein expression data from Reverse-Phase Protein Array Analysis (RPPA) in HER2-positive BC, we developed gene expression metagenes that reflect pathway activation levels. Next we assessed the ability of these metagenes to predict resistance to adjuvant trastuzumab using gene expression data from two independent datasets.

10 metagenes passed external validation (false discovery rate [fdr] < 0.05) and showed biological relevance with their pathway of origin. These metagenes were further screened for their association with trastuzumab resistance. An association with trastuzumab resistance was observed and validated only for the AnnexinA1 metagene (ANXA1). In the randomised phase III Fin-her study, tumours with low levels of the ANXA1 metagene showed a benefit from trastuzumab (multivariate: hazard ratio [HR] for distant recurrence = 0.16[95%CI 0.05–0.5]; p = 0.002; fdr = 0.03), while high expression levels of the ANXA1 metagene were associated with a lack of benefit to trastuzmab (HR = 1.29[95%CI 0.55–3.02]; p = 0.56). The association of ANXA1 with trastuzumab resistance was successfully validated in an independent series of subjects who had received trastuzumab with chemotherapy (Log Rank; p = 0.01).

In conclusion, in HER2-positive BC, some proteins are associated with distinct gene expression profiles. Our findings identify the ANXA1metagene as a novel biomarker for trastuzumab resistance.

The current use and attitudes towards tumor genome sequencing in breast cancer

There is increasing availability of technologies that can interrogate the genomic landscape of an individual tumor; however, their impact on daily practice remains uncertain. We conducted a 28-item survey to investigate the current attitudes towards the integration of tumor genome sequencing in breast cancer management. A link to the survey was communicated via newsletters of several oncological societies, and dedicated mailing by academic research groups. Multivariable logistic regression modeling was carried out to determine the relationship between predictors and outcomes. 215 physicians participated to the survey. The majority were medical oncologists (88%), practicing in Europe (70%) and working in academic institutions (66%). Tumor genome sequencing was requested by 82 participants (38%), of whom 21% reported low confidence in their genomic knowledge, and 56% considered tumor genome sequencing to be poorly accessible. In multivariable analysis, having time allocated to research (OR 3.37, 95% CI 1.84–6.15, p < 0.0001), working in Asia (OR 5.76, 95% CI 1.57 – 21.15, p = 0.01) and having institutional guidelines for molecular sequencing (OR 2.09, 95% 0.99–4.42, p = 0.05) were associated with a higher probability of use. In conclusion, our survey indicates that tumor genome sequencing is sometimes used, albeit not widely, in guiding management of breast cancer patients.

Abstract P2-08-08: Circulating tumor cells count-based nomograms to predict survival of metastatic breast cancer patients: Results from the European pooled analysis

Background: The European Pooled Analysis of CTC (EPAC) in metastatic breast cancer, based on 1,944 individual data from patients with various tumor types and clinical settings (Bidard et al, Lancet Oncol 2014), has established CTC count (CellSearch) at baseline and during therapy as a level of evidence 1 independent prognostic biomarker and demonstrated its superiority over serum blood markers. As part of the study pre-planned objectives, we sought to establish nomograms allowing accurate individual survival predictions.

Methods: Using individual data from 17 centers, we built simplified multivariate prognostic models taking into account the independent prognostic clinico-pathological (CP) characteristics including CTC count, dichotomized using the 5CTC/7.5ml threshold, at baseline and at 3-5 weeks after the start of a new treatment regimen, and derived nomograms for progression-free survival (PFS) and overall survival (OS) prediction at baseline and after 3-5 weeks of treatment. We report here the internal validation of these nomograms. Discrimination of the models was assessed using the c-index estimated by a jackknife procedure and the calibration was visually assessed through 10-fold crossvalidated calibration plots at 1,2,3 years for OS and 1,2 years for PFS.

Results: Multivariate models at baseline for PFS and OS were fitted on 1501 and 568 individual patient data with CTC count at baseline and CTC count at baseline and after 3-5 weeks, respectively. Models include tumor subtype, the number of previous chemotherapy lines (0/1/≥2), PS, age (&lt;=50/&gt;50-65/&gt;65 years), metastasis-free intervals (0/&gt;0-3/&gt;3 years), metastatic sites (liver and CNS) and CTC count at baseline and eventually at 3-5 weeks of treatment. The C-index increased from 0.722 to 0.755 (increase in C-index:0.033, 95% CI [0.019;0.045]) when adding baseline CTC to the CP only model for OS (n=1501). For those patients with CTC values at 3-5 weeks (n=568), there was an additional increase in the C-index when adding CTC at 3-5 weeks to a model with already CP and baseline CTC from 0.731 to 0.743 (increase in C-index 0.013, 95% CI [-0.004;0.025]). The model with CP and baseline CTC counts showed a good calibration for OS at 1,2,3 years and the model with CP, baseline CTC and CTC count at 3-5 weeks a moderately good calibration. Similar results were obtained for PFS.

Conclusion: From the largest database with individual CTC data, we were able to build PFS and OS survival nomograms, with satisfactory discrimination and calibration. Our planned next step is to validate the nomogram in an additional cohort.

Citation Format: Bidard F-C, Peeters D, Fehm T, Nole F, Gisbert-Criado R, Mavroudis D, Grisanti S, Generali D, Garcia-Saenz JA, Stebbing J, Caldas C, Gazzaniga P, Manso L, Zamarchi R, Fernandez de Lascoiti A, de Mattos-Arruda L, Ignatiadis M, van Laere SJ, Meier-Stiegen F, Sandri M-T, Vidal-Martinez J, Politaki E, Consoli F, Bottini A, Diaz-Rubio E, Krell J, Dawson S-J, Raimondi C, Rutten A, Janni W, Munzone E, Carañana V, Agelaki S, Almici C, Dirix L, Solomayer E, Zorzino L, Reis-Filho JS, Squifflet P, Pantel K, Beije N, Sleijfers S, Pierga J-Y, Michiels S. Circulating tumor cells count-based nomograms to predict survival of metastatic breast cancer patients: Results from the European pooled analysis. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P2-08-08.

The evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer: recommendations by an International TILs Working Group 2014

BACKGROUND

The morphological evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer (BC) is gaining momentum as evidence strengthens for the clinical relevance of this immunological biomarker. Accumulating evidence suggests that the extent of lymphocytic infiltration in tumor tissue can be assessed as a major parameter by evaluation of hematoxylin and eosin (H&E)-stained tumor sections. TILs have been shown to provide prognostic and potentially predictive value, particularly in triple-negative and human epidermal growth factor receptor 2-overexpressing BC.

DESIGN

A standardized methodology for evaluating TILs is now needed as a prerequisite for integrating this parameter in standard histopathological practice, in a research setting as well as in clinical trials. This article reviews current data on the clinical validity and utility of TILs in BC in an effort to foster better knowledge and insight in this rapidly evolving field, and to develop a standardized methodology for visual assessment on H&E sections, acknowledging the future potential of molecular/multiplexed approaches.

CONCLUSIONS

The methodology provided is sufficiently detailed to offer a uniformly applied, pragmatic starting point and improve consistency and reproducibility in the measurement of TILs for future studies.

Circulating tumor cells and circulating tumor DNA for precision medicine: dream or reality?

Next-generation sequencing studies have provided further evidence to support the notion that cancer is a disease characterized by Darwinian evolution. Today, we often fail to capture this evolution and treatment decisions, even in the metastatic setting, are often based on analysis of primary tumor diagnosed years ago. Currently, this is considered a major reason for treatment failures in cancer care. Recent technological advances in the detection and characterization of circulating tumor cells and circulating tumor DNA might address this and allow for treatment tailoring based on real-time monitoring of tumor evolution. In this review, we summarize the most important recent findings in the field, focusing on challenges and opportunities in moving these tools forward in clinical practice.

The evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer: recommendations by an International TILs Working Group 2014

BACKGROUND

The morphological evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer (BC) is gaining momentum as evidence strengthens for the clinical relevance of this immunological biomarker. Accumulating evidence suggests that the extent of lymphocytic infiltration in tumor tissue can be assessed as a major parameter by evaluation of hematoxylin and eosin (H&E)-stained tumor sections. TILs have been shown to provide prognostic and potentially predictive value, particularly in triple-negative and human epidermal growth factor receptor 2-overexpressing BC.

DESIGN

A standardized methodology for evaluating TILs is now needed as a prerequisite for integrating this parameter in standard histopathological practice, in a research setting as well as in clinical trials. This article reviews current data on the clinical validity and utility of TILs in BC in an effort to foster better knowledge and insight in this rapidly evolving field, and to develop a standardized methodology for visual assessment on H&E sections, acknowledging the future potential of molecular/multiplexed approaches.

CONCLUSIONS

The methodology provided is sufficiently detailed to offer a uniformly applied, pragmatic starting point and improve consistency and reproducibility in the measurement of TILs for future studies.

Plasma circulating tumor DNA as an alternative to metastatic biopsies for mutational analysis in breast cancer

BACKGROUND

Molecular screening programs use next-generation sequencing (NGS) of cancer gene panels to analyze metastatic biopsies. We interrogated whether plasma could be used as an alternative to metastatic biopsies.

PATIENTS AND METHODS

The Ion AmpliSeq™ Cancer Hotspot Panel v2 (Ion Torrent), covering 2800 COSMIC mutations from 50 cancer genes was used to analyze 69 tumor (primary/metastases) and 31 plasma samples from 17 metastatic breast cancer patients. The targeted coverage for tumor DNA was ×1000 and for plasma cell-free DNA ×25 000. Whole blood normal DNA was used to exclude germline variants. The Illumina technology was used to confirm observed mutations.

RESULTS

Evaluable NGS results were obtained for 60 tumor and 31 plasma samples from 17 patients. When tumor samples were analyzed, 12 of 17 (71%, 95% confidence interval (CI) 44% to 90%) patients had ≥1 mutation (median 1 mutation per patient, range 0-2 mutations) in either p53, PIK3CA, PTEN, AKT1 or IDH2 gene. When plasma samples were analyzed, 12 of 17 (71%, 95% CI: 44-90%) patients had ≥1 mutation (median 1 mutation per patient, range 0-2 mutations) in either p53, PIK3CA, PTEN, AKT1, IDH2 and SMAD4. All mutations were confirmed. When we focused on tumor and plasma samples collected at the same time-point, we observed that, in four patients, no mutation was identified in either tumor or plasma; in nine patients, the same mutations was identified in tumor and plasma; in two patients, a mutation was identified in tumor but not in plasma; in two patients, a mutation was identified in plasma but not in tumor. Thus, in 13 of 17 (76%, 95% CI 50% to 93%) patients, tumor and plasma provided concordant results whereas in 4 of 17 (24%, 95% CI 7% to 50%) patients, the results were discordant, providing complementary information.

CONCLUSION

Plasma can be prospectively tested as an alternative to metastatic biopsies in molecular screening programs.

Abstract OT1-3-02: The treat CTC trial – A new approach targeting circulating tumor cells (CTC) in early breast cancer (EBC)

Background:

The presence of CTC in metastatic BC is associated with an impaired prognosis. Recent data show a reduced disease-free survival and increased risk of death in the presence of CTC in EBC. Therefore, patients with persisting CTC after (neo)adjuvant chemotherapy might benefit from additional systemic treatment.

Recent data have reinforced the hypothesis that trastuzumab can eliminate tumor cells by antibody dependent cell cytotoxicity (ADCC) or other immune mechanisms. Preclinical data have provided evidence that the benefit of trastuzumab may be associated with targeting cancer stem cells in a HER2 independent model (Ithimakin et al Cancer Res 2013). Trastuzumab eliminated CTC, irrespective of the HER2 status of the primary tumor and of CTC and this was associated with reduced relapses(Georgoulias et al Ann Oncol 2012).

Trial Design:

Treat CTC trial is a multicenter European randomized phase II trial, sponsored by the EORTC and run under the BIG umbrella. It will assess the efficacy of trastuzumab in eliminating persisting CTC after the completion of (neo)adjuvant chemotherapy and surgery in patients with HER-2-negative EBC. Eligible patients will be randomized in a 1:1 ratio to either 6 cycles of trastuzumab or observation. Patients’ peripheral blood will be tested again for CTC after 18 weeks.

Main Eligibility criteria:

- Adequately excised HER2-negative EBC

- Evidence of CTC detection using the CellSearch technology after completion of (neo)adjuvant chemotherapy

- Completion of adjuvant chemotherapy for node-positive disease or neoadjuvant chemotherapy with residual invasive disease in breast or lymph nodes (no complete pathological response)

- Histological Grade &gt; 1 and primary tumor size &gt; 1 cm

Specific aims:

The primaryobjective of the trial is to evaluate whether trastuzumab decreases the detection rate of CTC in patients with HER2-negative EBC by comparing the trastuzumab treated arm to the observation arm. Furthermore, clinical outcomes as measured by Recurrence Free Interval (RFI), Invasive Disease Free Survival (IDFS), Disease Free Survival (DFS) and Overall Survival (OS)) between the trastuzumab and observation arms will be compared.

Present accrual and target accrual:

Treat CTC started patient screening in April 2013 in Belgium. It is estimated that 2175 women will be registered to include 174 patients eligible for randomization. Accrual is expected to be completed in 2 years.

Methods:

The primary test will be a one-sided test to compare the trastuzumab arm to the observation arm for the CTC detection rate at week 18 (superiority test). The comparison for the primary endpoint will be performed on the intention-to-treat population using a one-sided test with overall a of 0.1. The odds ratio and its confidence interval will be estimated using a logistic regression model. The comparison of RFI, IDFS, DFS and OS will be done using a two-sided test in a proportional hazards model for cause specific hazard, adjusted for the stratification factors.

Perspectives:

Given the prognostic relevance of CTC in BC, the Treat CTC trial will be the first multicenter, randomized trial in which CTC are used to guide treatment decisions in EBC. The results of this trial will help to clarify the clinical utility of CTCs in early disease.

Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr OT1-3-02.

Abstract P1-04-03: Genome-wide analysis of copy number variations and mutation profiles of single circulating tumour cells using massively parallel paired-end sequencing

INTRODUCTION

Recent advances in single cell isolation techniques and next generation sequencing (NGS) have paved the way for the genome-wide molecular analysis of individual circulating tumour cells (CTCs) in patients with metastatic carcinomas. Here we present the results of a pilot study evaluating the feasibility and reliability of NGS of single CTC from whole blood samples.

MATERIALS & METHODS

Single cells of the human breast cancer cell line HCC38 were harvested from spiked blood samples in a semi-automated workflow consisting of immunomagnetic enrichment using the CellSearch system and dielectrophoretic cell sorting using the DEPArray system. DNA was isolated and amplified using the Ampli1 whole genome amplification (WGA) kit and subjected to low-coverage genome-wide paired-end sequencing for copy number variation (CNV) analysis and targeted re-sequencing of 200 cancer-related genes for somatic mutation analysis.

RESULTS

Single-cell WGA products of four HCC38 cells were subjected to whole genome sequencing for CNV analysis. Average coverage depth was 0,68x. At a binning window of 50 kb, detection results of CNVs in single-cell samples were highly consistent (&gt;81% copy number concordance per bin genome wide) with CNV profiles from non-amplified multi-cell samples of the same cell line. We could demonstrate that part of the discordance was due to the acquisition of novel DNA-rearrangements in the single cells. Three of the single-cell WGA products were additionally subjected to targeted re-sequencing for mutation analysis of 200 selected genes, of which the analysis is currently ongoing.

DISCUSSION

Our study demonstrates the feasibility of a comprehensive genome-wide CNV analysis and targeted mutation analysis using NGS of single tumour cells isolated from whole blood samples in a highly automated isolation workflow. This approach provides a robust framework for the study of intercellular heterogeneity within the CTC population in blood samples of patients with (metastatic) breast cancer. In addition, our results document the extent of WGA-induced bias of a recently commercialized PCR-based WGA kit.

These authors contributed equally to the data presented in this abstract.

Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P1-04-03.

Abstract PD6-5: Pooled analysis of circulating tumor cells in metastatic breast cancer: Findings from 1944 individual patients data

Background: Clinical validity of CTCs (CellSearch®) in metastatic breast cancer (MBC) patients has previously been assessed in studies with limited statistical power. We aimed to pool all European studies to obtain high-level evidence on the prognostic value of CTCs, to investigate their effects across different clinico-pathological characteristics and therapies and to further validate the MD Anderson/Institut Curie/Fox Chase CTC-based prognostic nomogram established in first-line treated MBC patients (Giordano et al, Clin Cancer Res 2013).

Material and methods: Methods were predefined in a written protocol. In December 2012, we searched for eligible studies that accrued patients in 2003-2012. We contacted all European laboratories using CellSearch®. We used likelihood ratio tests (LR) in Cox regression models stratified by study to assess the independent prognostic value of CTC when added to a clinicopathological (CP) model for progression-free (PFS) and overall survival (OS). Landmark analyses were used to assess the prognostic effect of early changes in CTC. The CTC-based nomogram (http://cancernomograms.com/CTCOnline.html) score was retrieved for every patient; we calculated C-indices, drew calibration plots and Kaplan-Meier curves according to quintiles of the nomogram score.

Results: We collected individual data of 1944 MBC patients, from 20 different studies (some unpublished), from 17 centers in 7 European countries. We observed 1507 PFS events and 929 deaths. Baseline CTC count was significantly associated with several patient characteristics, such as performance status (PS, p&lt;10-4), synchronous metastasis (p&lt;10- 2) tumor subtype (p&lt;10-4), liver & bone metastases (p&lt;10-4), CEA & CA15-3 levels (p&lt;10-4). The CP model for OS included PS, MBC subtypes, number of previous lines of treatment, patient's age, metastasis-free interval, metastatic sites (p&lt;0.01 for all). In a multivariate analysis containing the CP model parameters and CTC count at baseline, elevated CTC count (≥5) was a significant independent predictor of OS (n = 1444, HR = 2.7, 95%CI [2.2-3.2], LR p&lt;10-4). Baseline serum markers added either no or marginal effect to the CP plus baseline CTC model for OS. In contrast, early changes in CTC status at week 3-5 significantly added prognostic information for OS to the model with CP factors and baseline CTC+ (n = 569, HR = 1.8 [2.2-3.2], LR p&lt;0.001). In the population of interest (MBC treated by first line chemotherapy, n = 402 patients, 176 deaths), the CTC-based nomogram exhibited a good C-index for OS (0.69), was well calibrated and showed clear separation of the survival curves. Additional results, including subgroup analyses by tumor subtype and treatments will be presented at the meeting.

Conclusions: This pooled analysis is the largest study ever reported on CTC in MBC, with a previously unreached statistical power. It provides a clear level-of-evidence 1 on the independent prognostic value of CTCs before and during treatment in MBC. Also, the CTC-based prognostic nomogram is independently validated.

Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr PD6-5.

Genomic grade adds prognostic value in invasive lobular carcinoma

BACKGROUND

The prognostic value of histologic grade (HG) in invasive lobular carcinoma (ILC) remains uncertain, and most ILC tumors are graded as HG2. Genomic grade (GG) is a 97-gene signature that improves the prognostic value of HG. This study evaluates whether GG may overcome the limitations of HG in ILC.

METHODS

Gene expression data were generated from frozen tumor samples, and GG calculated according to the expression of 97 genes. The prognostic value of GG was assessed in a stratified Cox regression model for invasive disease-free survival (IDFS) and overall survival (OS).

RESULTS

A total of 166 patients were classified by GG. HG classified 33 (20%) tumors as HG1, 120 (73%) as HG2 and 12 (7%) as HG3. GG classified 106 (64%) tumors as GG low (GG1), 29 (17%) as GG high (GG3) and 31 (19%) as equivocal (cases not classified as GG1 or GG3). The median follow-up time was 6.5 years. In multivariate analyses, GG was associated with IDFS [HR(GG3 vs GG1) 5.6 (2.1-15.3); P < 0.001] and OS [HR(GG3 vs GG1) 7.2, 95% CI (1.6-32.2); P = 0.01].

CONCLUSIONS

GG outperformed HG in ILC and added prognostic value to classic clinicopathologic variables, including nodal status.

Abstract S4-4: Independent validation of Genomic Grade in the BIG 1-98 study

Background: We have reported that a 97-gene signature, the Genomic Grade Index (GGI) can separate histological grade (HG) 2 breast tumors into low vs high GGI tumors with different outcomes. We have also developed a quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) tool including 6 reporter and 3 control genes that can reliably evaluate Genomic Grade (GG) in paraffin embedded tumors.

Methods: We evaluated the qRT-PCR GG in women treated with either tamoxifen or letrozole monotherapy in the Breast International Group (BIG) 1–98 study with 8.1 year median follow-up. The association between continuous GG and distant recurrence-free interval (DRFI) was evaluated in Cox regression models stratified for the 2 vs 4 arm randomization option, chemotherapy and hormonotherapy use, with and without adjustment for clinicopathological characteristics. Estrogen, progesterone receptor (ER, PR), Ki67 and HG were centrally reviewed. The clinicopathological model included age and log2 tumor size, ER and PR as continuous variables, nodal status (N0 vs 1–3 vs &gt;3), HG (1 vs 2 vs 3) and HER2 (negative vs positive). Similar analyses were performed for log2 Ki67 as continuous variable. Added prognostic value was assessed using the likelihood ratio statistic.

Results: We obtained GG in 883 (62%) of 1428 samples. Non evaluable results were due to pre-analytical issues or technical failure. Among the 883 patients, 521 (59%) had N0 disease, 435 (49%) were treated with tamoxifen and 84 (10%) had distant recurrences. The GG classified 502 patients with HG2 tumors as either GG1 (N = 202, 40%), equivocal (N = 220, 44%) or GG3 (N = 80, 16%). In univariate analysis, increasing GG and Ki67 were significantly associated with lower DRFI (Table). When the 773 N0/1-3 patients were analyzed based on HG, the Kaplan-Meier estimate for 10-year DRFI was 98% (95% confidence intervals 96–100%) for the 123 HG1, 92% (88–95%) for the 456 HG2 and 84% (78–90%) for the 194 HG3 patients. In the N0/1-3 population, 10 year DRFI based on GG was 95% (96–100%) for the 290 GG1, 92% (89–95%) for the 309 GG equivocal and 84% (77–90%) for the 178 GG3 patients. Interestingly, when the 456 HG2 and N0/1-3 patients were analyzed based on GG, the 10-year DRFI was 95% (92–100%) for the 185 HG2/GG1 patients, 92% (88–96%) for the 202 HG2/GG equivocal patients, 86% (76–96%) for the 69 HG2/GG3 patients. In all patients either GG or Ki67 as continuous variable significantly improved the clinicopathological model in predicting distant recurrence (Table).

Conclusion: Either Genomic Grade or centrally reviewed Ki67 provides independent prognostic information for risk of distant recurrence beyond clinicopathological characteristics in patients treated with endocrine therapy.

Abstract P6-07-22: Association between SPARC mRNA expression, prognosis and response to neoadjuvant chemotherapy in early breast cancer (BC): a pooled in-silico analysis

Background: SPARC (Secreted Protein Acidic and Rich in Cysteine) is known to regulate cell growth and to inhibit cell-cycle progression. It has high affinity in binding albumin and hence has been suggested to predict benefit of albumin-bound cytotoxic agents. We conducted a pooled analysis to elucidate SPARC expression according to BC molecular subtypes and its association with clinical outcome in early BC.

Methods: We used publically available datasets and normalized microarray data as published by the original studies. Eligible patients were those who received no adjuvant systemic therapy (untreated series), were treated with tamoxifen (tam-treated series) or were treated with neoadjuvant anthracyclines ± paclitaxel or docetaxel (neoadjuvant series). We computed 2 SPARC modules, SPARC7 and SPARC8 that were composed of genes with an absolute correlation above 0.7 and 0.8 with SPARC, respectively. In the untreated series, we examined the expression of SPARC according to BC subtype defined by PAM50. We investigated the correlation with other gene modules representing diverse biological processes; proliferation (AURKA, GGI), immune (STAT1, IRM), and stroma (DCN, PLAU). We investigated the association with relapse-free survival (RFS) in univariate and multivariate models, both in the untreated and tam-treated series. This was performed in all patients and according to BC subtype. In the neoadjuvant series, we investigated the association with pathological complete response (pCR) in all patients and according to BC subtype. All multivariate models were adjusted for tumor size, nodal status, age and histological grade.

Results: 1008, 393 and 996 patients were included in the untreated, tam-treated and neoadjuvant series, respectively. SPARC expression was highest in luminal-A, small (&lt; 2 cm) and low histological grade tumors (all p &lt; 0.0001). We found high positive correlation between SPARC, SPARC modules and stroma-related modules (r = 0.90) but a negative correlation with proliferation (r = −0.66). No correlation was observed with immune-related modules. In the untreated series, SPARC was not associated with prognosis in the univariate model. However, restricting the analysis to the HER2 molecular subtype, high SPARC expression was associated with short RFS in both univariate and multivariate models (HR: 3; 95% CI [1.5–6.3], p = 0.002). In the tam-treated series, SPARC expression was not associated with clinical outcome neither in luminal-A nor luminal-B. In the neoadjuvant series, SPARC expression was not associated with pCR (p = 0.37) except in the HER2 subtype in which high expression was independently associated with low pCR rates in the multivariate model (OR: 0.32; 95% CI [0.1–0.9], p = 0.043). Similar associations were observed with SPARC7 and SPARC8 modules.

Conclusion: Two main conclusions could be drawn from our analysis 1) SPARC is highly expressed in low proliferative ER+ tumors; 2) High SPARC expression is associated with short RFS and poor response to neoadjuvant therapy in patients with HER2+ BC. These results could be relevant in the clinical development of albumin-bound compounds, in which SPARC expression is believed to identify patients who could benefit from these agents.

Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P6-07-22.

Abstract P6-07-14: Mutational and transcriptomic characterization of breast cancer (BC) arising in young patients (pts) and during pregnancy and their associations with long-term outcome

Background: BC arising in young age is biologically distinct. Whether diagnosis during pregnancy has an impact on prognosis and tumor biology remains to be elucidated. We report for the first time mutational and transcriptomic profiling of BC arising in young pts and the impact of diagnosis during pregnancy. We also correlate these findings with clinical outcome.

Methods: 65 pts with BC during pregnancy diagnosed at the European Institute of Oncology in the period 1999–2009 were matched to 130 BC pts who were diagnosed and treated at the same period/institute. We screened for 84 somatic hotspot mutations on 17 cancer-related genes using mass spectroscopy-based sequencing (Sequenom). We evaluated the pattern of mutations in the two cohorts and according to BC subtype defined using central immunohistochemistry as follows: Luminal A (ER+, HER2−, Ki67 &lt;14%), Luminal-B (ER+, HER2−, Ki67 &gt;14%), HER2+ (HER2+ irrespective of ER), and triple negative (ER−, PgR−, HER2−). Survival endpoints included distant relapse free survival (DRFS) and overall survival (OS).

Results: Median age at diagnosis was 36 years (range: 28–47). At a median follow-up of 74 months (IQR: 42–96), 44 (23%) and 29 (15%) pts developed a DRFS and OS event respectively. Pts diagnosed during pregnancy had inferior DRFS (HR: 3.2 [1.5–6.7]) and OS (HR: 2.9 [1.1–7.9]) after adjusting for pT, pN, grade, BC subtype, and therapy. Mutational profiling was successful in 97% of pts. A total of 57 hotspot mutations (30%) were detected in 51 pts (15 [23%] pregnant and 36 [28%] controls). The differences in mutations between the two groups are summarized in the table.

P1-02-05: Invasive Lobular Carcinoma – A Luminal Breast Cancer Histotype Enriched for Epithelial-to-Mesenchymal Transition Features

Background: Invasive lobular carcinoma (ILC) represent the second most common breast cancer (BC) histotype, but little information is available about its molecular features. We sought to perform a comprehensive molecular classification of ILC and to investigate the molecular features that can differentiate ILC from invasive ductal carcinoma (IDC).

Methods: A consecutive series of 174 ILC tumors from three different centers with mininum 5-year follow-up was selected. Gene expression data were generated using Affymetrix U133 Plus 2.0 arrays. Data were analyzed by using frozen robust multiarray analysis (Matthew et al. 2009). ILC tumors were classified into BC molecular subtypes using the PAM50 classifier. Invasive Disease Free Survival (IDFS) was defined as the interval between the date of surgery and the date of any invasive recurrence or death. Multivariate analyses for 10-y IDFS containing classical clinico-pathologic characteristics were performed for different gene signatures, which were scaled so that 2.5 and 97.5% percentiles equal −1 and 1. A cohort of 90 IDC tumor samples, normalised with the same method, was used to compare expression data between ILC and IDC.

Results: ILCs were mainly characterized as luminal A (LA: 76%, n=133) and luminal B (LB:; 20%, n=34) followed by a minority of HER2−positive (2%, n=3), basal (1%, n=2) and normal-like (1%, n=2) based on Perou's classification. LB had inferior 10-y IDFS than LA in univariate cox regression analysis (HR = 1.99, 95% CI, 1.08−3.67, p=0.026). The EMT gene module was significantly higher expressed in ILC as compared to IDC (p &lt; 0.001). Adjusted hazard ratios (HR) for different gene modules investigated are detailed in the table below.