Download the ESMO 2012 Abstract Book - Oxford Journals

BRAF mutations. Circulating free DNA (cfDNA) is present at low levels in plasma of
healthy individuals, whereas its increased concentration was observed in patients
with malignant tumors including non-small cell lung cancer (NSCLC). Cancer gene
somatic mutations can be detected in cfDNA.
Methods: This is a randomized trial, comparing G+ docetaxel (D), to D in 2 nd line
advanced NSCLC patients. This study aimed at investigating the potential usefulness
of plasma DNA concentration in NSCLC patients for efficacy of therapy. We
performed a prospective exploratory analysis to identify serum biomarkers as
predictors of improved outcomes with G. Plasma samples were collected from 160
pts at baseline prior to initiation of treatment, and at end of cycles 1 and 2. cfDNA
was evaluated using the Ion Torrent platform to survey mutations in 46 cancer
genes. Serum levels were correlated with progression free survival (PFS) and overall
survival (OS) in both treatment arms. Upfront enrichment using both nanoparticle
capture and phosphoprotein capture technologies were coupled to high resolution
Orbitrap-LTQ MS/MS analysis followed by spectral counting for each samples.
Results: By early May approximately half of the 300 planned patients were enrolled.
Baseline characteristics were balanced. Full proteomic profiling and cfDNA analysis
is ongoing, and will be correlated with the efficacy outcomes from a planned interim
analysis in early September, including disease control rate, PFS, and OS.
Disclosure: I. El Hariry: stock ownership. V. Vukovic: stock ownership. F. Teofilovici:
stock option. V. Reichert: stock options. All other authors have declared no conflicts
of interest.
1665P INTER-OBSERVER AGREEMENT OF THE RESPONSE TO
THERAPY ASSESSMENT IN ADVANCED LUNG CANCER
WITHIN A NORMATIVE MEASUREMENT ENVIRONMENT
H. Beaumont 1 , E. Oubel 1 , A. Iannessi 2 , D. Wormanns 3
1 Science, MEDIAN Technologies, Valbonne, FRANCE, 2 Radiology, Centre
Antoine Lacassagne, Nice, FRANCE, 3 Department of Radiology, ELK Berlin
Chest Hospital, Berlin, GERMANY
Purpose: Image-based biomarkers play an increasing role in the assessment of
response to therapy. The value of a biomarker comes, in part, from its ability to
guarantee reproducibility in a varying context. This study aims at evaluating the
impact of workflow normalization and automation on the reproducibility of
volume-based response assessment. This impact is measured in terms of inter-reader
agreement (IRA).
Method: A retrospective study was performed on 10 patients with Non-Small Cell Lung
Cancer (NSCLC) lesions followed over 7 time points (TP) on average with Computed
Tomography. Five imaging scientists measured sequentially the volume of each lesion at
each TP and the time required to perform segmentations. We relied on a software
providing semi-automatic segmentation capabilities and follow-up (FU) display. After 6
months, a second reading session used no automation for segmentation. The response
to treatment was assessed according to +/-30% thresholds as recommended by the
Quantitative Imaging Biomarker Alliance (QIBA). The IRA was measured by using
Kappa coefficient. From the initial reading, where the same reader reviewed
consecutively all TPs from the same patient, additional IRA assessments where
performed by random mixing of measurements from different readers. Different types
of mixing patterns simulated several deviations from normalization, this corresponding
to FUs involving more than one radiologist or method.
Results: The IRA of a normalized and automated workflow yielded a significantly
higher kappa = 0.69 [0.59; 0.79] compared to mixed manual segmentations where
kappa was 0.24 [0.06; 0.42]. Analyzed separately, both single-reviewer assessment and
semi-automated segmentation led to higher reproducibility. The recourse to
semi-automated segmentation reduces the average segmentation time by a factor of 4.
Conclusions: Normalization and automation of the measurements improved
significantly the IRA. Both normalization and automation contribute to improved
reproducibility. Even small deviations from a normalized review may impair the
global reliability of FUs. Single-reviewer reading and automation must be considered
for a highly reproducible assessment of response to therapy.
Disclosure: All authors have declared no conflicts of interest.
1666P UTILITY OF A NOVEL QUANTITATIVE EGFR MUTATED
PROTEIN ANALYSIS USING AQUA SYSTEM FOR EGFR-TKI
TREATMENT IN NON-SMALL CELL LUNG CANCER
PATIENTS
K. Goto 1 , G. Ishii 2 , K. Fujimoto-Ouchi 3 , M. Shirane 4 , H. Ohmatsu 1 , S. Niho 1 ,
K. Yoh 1 , S. Umemura 1 , K. Nagai 5 ,Y.Ohe 1
1 Division of Thoracic Oncology, National Cancer Center Hospital East, Kashiwa,
JAPAN, 2 Department of Pathology, National Cancer Center Hospital East,
Kashiwa, JAPAN, 3 Discovery Pharmacology Department 2, Chugai
Pharmaceutical Co., Ltd., Kanagawa, JAPAN, 4 Medical Affairs Devision, Chugai
Pharmaceuticals Co. Ltd, Tokyo, JAPAN, 5 Department of Thoracic Oncology,
National Cancer Center Hospital East, Kashiwa, JAPAN
Background: Lung cancer carrying EGFR mutation (muEGFR) expresses good
response to EGFR tyrosine kinase inhibitors (TKIs). However, not all patients (pts)
Annals of Oncology
with mutation showed similar clinical benefits from EGFR-TKIs. Heterogeneity of
tumor cells is considered to be one plausible reason. MuEGFR was usually genotyped
in qualitative manner in present clinical practice. In addition, new technology of
AQUA system enables us to measure protein levels objectively with considering
expression heterogeneity in tissues. Here we investigated the correlation between
AQUA score obtained from each specimen and the response and the response
duration by EGFR-TKIs.
Methods: Slice sections from genotyped 105 pts with muEGFR and 33 pts with
wild-type EGFR were stained by E746-A750 deletion (6B6)- and L858R
(43B2)-specific antibodies. Homo- or heterogeneous staining and muEGFR protein
levels of these specimens were evaluated with AQUA system.
Results: Concordance rate between genotypes and AQUA-based phenotypes
(deletion and L858R), with cut-off values based on ROC curve, were 60% and
82%. Low sensitivity of 6B6 was due to lucking of reactivity for other
deletion variants. Alternatively, specificities of both antibodies were 96% and
92%, indicating highly selectivity to muEGFR protein. The AQUA score of 28
pts treated with EGFR-TKIs were varied (5.4 – 798.5) but the falls plot
analysis indicated that pts with high AQUA score showed better objective
response. Specimens of partial response (PR) had significantly higher AQUA
score than those of progression disease (67.4 vs 9.6, p = 0.025). However,
there were no differences of AQUA score between specimens of PR and stable
disease. When divided pts into two groups based on median AQUA score,
the pts with high AQUA score showed a tendency of longer treatment
duration (24 mo vs 15 mo, p = 0.210). Especially, the high AQUA score with
low intratumoral deviation pts showed longer treatment duration in compared
to others. These results indicate that AQUA-based analysis can classify pts
carrying genotyped muEGFR into susceptible and less susceptible ones to
EGFR-TKIs.
Conclusions: We observed that pts with high AQUA score were more sensitive
to EGFR-TKIs. It is important for predicting sensitivity to EGFR-TKIs to
examine not only genotyping but also muEGFR protein levels by using AQUA
system.
Disclosure: All authors have declared no conflicts of interest.
1667P INFLUENCE OF DRUG EXPOSURE AND GENETIC VARIATION
ON PACLITAXEL–INDUCED NEUROTOXICITY
A.M. De Graan 1 , L. Elens 2 , A. Sparreboom 3 , L.E. Friberg 4 , B. van der Holt 5 ,
P.J. De Raaf 1 , E.A.C. Wiemer 1 , J. Verweij 1 , R.H. van Schaik 2 , R.H.J. Mathijssen 1
1 Medical Oncology, Erasmus MC, Rotterdam, NETHERLANDS, 2 Clinical
Chemistry, Erasmus MC, Rotterdam, NETHERLANDS, 3 Pharmaceutical
Sciences, St. Jude Children’s Research Hospital, Memphis, TN, UNITED
STATES OF AMERICA, 4 Pharmaceutical Biosciences, Uppsala University,
Uppsala, SWEDEN, 5 Trials and Statistics, Erasmus MC, Rotterdam,
NETHERLANDS
Objectives: Paclitaxel is used for the treatment of several solid tumors and displays a
high inter-individual variation in exposure and toxicity. Neurotoxicity is a frequent
and sometimes serious adverse event frequently of paclitaxel treatment. This study
aims to find predictive pharmacokinetic (PK) and pharmacogenetic determinants for
the development of neurotoxicity.
Methods: In 261 patients treated with paclitaxel, incidence of neurotoxicity during
all courses (according to CTC 4.0 criteria) and PK parameters were determined. PK
samples were measured by HPLC or LC-MS/MS, and individual PK parameters were
estimated from previously developed population PK models by non-linear mixed
effects modeling in the software NONMEM. Univariate and multivariate stepwise
regression analysis were used to study the influence of co-variables on the
development of neurotoxicity. Variables tested were age, dose, tumor type, ethnicity,
co-medication, smoking status, and genetic variants in the PK-pathway of paclitaxel
(CYP3A4*22, CYP3A5*3, CYP2C8*3, CYP2C8 rs1058932, and ABCB1 3435 C > T,
determined by TaqMan genotyping assays on the ABI PRISM 7500 Fast real-time
PCR System).
Results: Exposure to paclitaxel (logAUC) was predictive for the development
(P