prediction of nsclc survival with the pam50 breast cancer signature
TRANSCRIPT
Volume 90 � Number 5S � Supplement 2014 Poster Presentations S71
263Noninvasive, Highly Specific Detection of T790M-MediatedAcquired Resistance in EGFR Mutant NSCLC Using a Novel DropletDigital PCR (ddPCR)eBased Cell-free Plasma DNA (cfDNA) AssayTranslational Research
A.G. Sacher,1,2 S. Mach,1 M. Messineo,3 D.M. Jackman,1 P.A. Janne,1,3
C.P. Paweletz,3 and G.R. Oxnard1; 1Dana-Farber Cancer Institute, Boston,
MA, 2University of Toronto, Toronto, ON, Canada, 3Belfer Institute for
Applied Cancer Science, Boston, MA
Purpose/Objective(s): The T790M mutation is the most common cause
of acquired resistance to EGFR kinase inhibitors, occurring in more than
50% of patients. Promising mutant-specific third generation EGFR kinase
inhibitors are being developed. However, selection of candidates for these
therapies requires invasive re-biopsy for tissue genotyping. We recently
reported on a new assay for plasma genotyping using ddPCR of cfDNA.
This plasma genotyping assay has the potential to noninvasively detect
T790M and accelerate delivery of appropriate therapies e potentially
obviating the need for biopsy entirely. We hypothesize that this assay can
be used to accurately detect the presence of the EGFR T790M resistance
mutation compared to standard tissue genotyping.
Materials/Methods: We identified patients (pts) with advanced EGFR
mutant NSCLC and acquired resistance to erlotinib that underwent
rebiopsy and plasma collection on 4 IRB-approved protocols. Rebiopsy
specimens underwent clinical EGFR genotyping. Plasma was collected in
EDTA tubes, cfDNA extracted, and EGFR genotype quantified using
ddPCR assays for L858R, exon 19 del, and T790M. Serial plasma geno-
typing on treatment was performed for a subset of pts with acquired
resistance initiating new treatment.
Results: A total of 41 pts undergoing rebiopsy for acquired resistance to
erlotinib had plasma available for analysis. Standard tissue genotyping on
these specimens detected the T790M resistance mutation in 26 pts (63%).
Plasma genotyping for T790M was concordant with rebiopsy T790M
status in 32/41 pts (78%). Plasma genotyping for T790M exhibited a
specificity of 93% (14/15 pts - false positive rate 7%), sensitivity of 70%
(18/26 pts) and a positive predictive value of 95% (18/19 pts) in this cohort
of patients. Posttreatment plasma specimens were available for 11 pts with
detectable T790M; the 6 pts with a partial response on imaging had a
significant decrease in plasma concentration compared to the 5 pts without
a response (mean 373 copies/mL decrease vs 22 copies/mL increase,
PZ.05). Treatment included platinum-doublet (1 pt), second generation
EGFR TKI (2 pts), and third-generation EGFR TKI (7 pts).
Conclusions: Plasma genotyping of cfDNA with ddPCR can detect
potentially actionable resistance mutations in EGFR mutant NSCLC with
acquired resistance to kinase inhibitors. The high specificity and positive
predictive value of this assay support its further development as a tool for
therapy selection despite its modest sensitivity. The quantitative nature of
this assay may also allow for early prediction of treatment failure. Clinical
development of this assay is ongoing.
Author Disclosure: A.G. Sacher: None. S. Mach: None. M. Messineo:
None. D.M. Jackman: G. Consultant; Genentech, Foundation Medicine.
P.A. Janne: G. Consultant; Pfizer (uncompensated). K. Advisory Board;
BI, Roche, Genentech, Astra-Zeneca, Sanofi, Chugai, Merrimack, Forma,
Clovis. C.P. Paweletz: None. G.R. Oxnard: K. Advisory Board; Clovis
Oncology, Genentech, BI, Novartis, Astellas, Chugai, Astra-Zeneca. Q.
Patent/License Fee/Copyright; Patent pending plasma genotyping assay.
264Tumor Biomarker Evaluation of 6785 Patients for CombinationTreatment Strategies in NSCLCTranslational Research
R. Feldman,1 J. Xiu,1 and R. Salgia2; 1Caris Life Sciences, Phoenix, AZ,2University of Chicago, Chicago, IL
Purpose/Objective(s): Non-small cell lung cancer (NSCLC) exhibits
activation of multiple tumor pathways. Presence of multiple aberrations
may account for drug resistance as well as strategies for combination
therapies. We examined concurrent aberrations of biomarkers in NSCLC to
present an overview of potential patient cohorts who may benefit from
such combinations.
Materials/Methods: 6785 NSCLC cases were evaluated. Specific testing
was performed and included a multiplatform approach: sequencing
(Sanger, NGS), protein expression (IHC), and gene amplification (CISH/
FISH).
Results: The EGFR mutation (MT) rate in this population was 12.7%
(135/1059), of which 57% (55/96) overexpressed EGFR (IHC) and 61%
(169/277) had EGFR gene amplification (FISH). This demonstrates the
dependence on the EGFR pathway and potential importance of dual in-
hibition with cetuximab and EGFR TKIs. Sixty-six percent (63/96) and 7%
(6/90) of EGFR MT patients were MET high (IHC) and amplified (CISH),
respectively, suggesting potential benefit from dual targeting of EGFR and
MET. Interestingly, TP53 mutations were observed in 54% (70/129) of
EGFR MT which has important implications for resistance to EGFR TKIs
and possible cross-resistance to radiation therapy. ALK translocations were
observed in 101 of 3611 (2.8%) patients, among which 19% (7/36), 3% (1/
29) and 2% (1/48) carried concurrent EGFR, METand HER2 amplification
(ISH), respectively, suggesting the potential for combining crizotinib with
agents such as cetuximab, onartuzumab, or trastuzumab. A BRAF muta-
tion was observed in 3.3% (34/1061), among which EGFR and MET
expression was high by IHC (both have been implicated in resistance to
BRAFi in other tumor types) in 58% (15/26) and 48% (12/25), respec-
tively, indicating benefit from combination of newly approved dabrafenib
with cetuximab or onartuzumab.
Conclusions: This is one of the largest analyses in the United States of
biomarkers in NSCLC. Our study shows the nonexclusive activation of
multiple pathways. These results also show the importance of multiplat-
form testing as both protein, gene copy number, and genetic alterations
were detected.
Author Disclosure: R. Feldman: A. Employee; Caris Life Sciences. J.
Xiu: A. Employee; Caris Life Sciences. R. Salgia: None.
265Prediction of NSCLC Survival With the PAM50 Breast CancerSignatureTranslational Research
J.M. Siegfried,1,2 Y. Lin,2 S. Dacic,2 L.P. Stabile,2 B. Diergaarde,2 H. Lin,2
T. Nukui,2 and M. Romkes2; 1University of Minnesota, Minneapolis, MN,2University of Pittsburgh, Pittsburgh, PA
Purpose/Objective(s): Lung tumors often express the estrogen receptor b(ERb) and progesterone receptor (PR). We hypothesized that breast cancer
survival signatures may also predict lung cancer survival.
Materials/Methods: We examined the relationship between disease-free
survival (DFS) and gene expression using 2 breast cancer survival signatures
(Oncotype DX and PAM50) in 104 early-stage (IA and IB) non-small cell
lung cancers. Criteria for cases were: fresh-frozen tumor tissue fromprimary
T1N0 or T2N0 adenocarcinoma or squamous cell lung cancer with no
neoadjuvant therapy; smoking history, age, sex, tumor size, and outcome
were known.mRNAexpressionwas examinedwith the IlluminaHumanHT-
12 v4BeadChip.Datawere subjected to background subtraction and quantile
normalization. Supervised Principal Component Analysis was used to
evaluate prediction of DFS for each gene panel as a group and by using the
top genes showing differential expression in the lung cancer cohort. All re-
sults were subjected to 10-fold cross-validation. A validation cohort of 64
lung cancer cases of all stages was also examined.
Results: Genes from the Oncotype DX panel showed no differential
expression and had no ability to separate cases based on DFS. For the
PAM50 gene panel, 3 probes (CXXC5, FGFR4, and FOXC1) showed
significant differential expression (P<.05) and 23 probes showed dif-
ferential expression at P<.1. The PAM50 gene panel was able to
separate lung cancer cases based on DFS (HR 1.9, PZ.03); PAM50
genes with differential expression at P<.1 also separated the cohort into
4 prognosis groups (PZ.008 for trend). The worst prognosis group had a
HR of 4.76 (PZ.002). These genes were next analyzed in a subset of 44
International Journal of Radiation Oncology � Biology � PhysicsS72
stage I cases for whom ERb immunohistochemistry (IHC) scores were
known. ERb was detectable in all cases. Comparing ERb high (Allred
score of 4 or more, NZ 20) and ERb low (Allred score of 4 or less, N Z24) cases, the top PAM50 genes were more informative in ERb high
cases (HR 11.7, PZ.0007) compared to ERb low cases (HR 3.38,
PZ.045). Similar analyses in the Oncotype DX gene panel did not
improve DFS prediction. We next validated the top PAM50 genes in a
different cohort of 64 NSCLC cases, which included all stages and
histologies and had ERb IHC scores. The top PAM50 genes were also
able to predict DFS in this cohort: HR 2.19, PZ.034 in all cases and HR
3.24, PZ.042 in ERb high cases. Pathway analysis indicated that the
informative PAM50 genes describe a network that contains the ER, PR,
HER2/HER3, neuregulins, FOXC1, FGFR4, and cyclin E.
Conclusions: These results suggest that genes involved in interactions
between ER, PR and HER signaling may be predictive of NSCLC survival,
especially in early-stage lung cancer that is ERb positive, and provide
further evidence for the importance of hormonal pathways in the biology
of lung cancer.
Author Disclosure: J.M. Siegfried: E. Research Grant; NIH, V Founda-
tion, Somalogic, Inc. K. Advisory Board; Free to Breathe. Y. Lin: E.
Research Grant; NIH. S. Dacic: E. Research Grant; NIH. L.P. Stabile: E.
Research Grant; NIH. B. Diergaarde: E. Research Grant; NIH. H. Lin:
None. T. Nukui: None. M. Romkes: E. Research Grant; NIH.
266Next-Generation Sequencing (NGS) in Advanced Lung Cancer in theCommunity to Prospectively Guide Clinical Trial SelectionTranslational Research
Z. Ma,1 T.M. Bauer,1,2 P.K. Chandra,3 D.J. Haynes,4 J.L. Prescott,3
D.M. Stults,1 S.F. Jones,1 J.D. Hainsworth,1,2 J.R. Infante,1,2
H.A. Burris,1,2 and D.R. Spigel1,2; 1Sarah Cannon Research Institute,
Nashville, TN, 2Tennessee Oncology, Nashville, TN, 3PathGroup,
Nashville, TN, 4Sarah Cannon Research UK, London, United Kingdom
Purpose/Objective(s): We launched a molecular profiling program in 10/
2012 at a single community practice in middle Tennessee. This program
was designed to discover molecular alterations with proven/potential
therapeutic significance to match pts to mutation-selective clinical trials.
Here we report the lung cancer cohort findings.
Materials/Methods: Tissue specimens from advanced lung cancer pts
were tested by NGS with 1000X coverage in a CLIA/CAP-certified lab to
detect hotspot mutations in 35 genes. Results were reported to the treating
physician <12 days of tissue procurement. Results were stored in a
database to enable correlation with clinical outcomes.
Results: From 10/2012-12/2013, 1039 pts were profiled across tumor
types; 238 (23%) were from lung cancer pts. 18% (43/238) of the lung
samples were inadequate for testing. Of the 195 pts with sufficient tissue,
80 (41%) had >1 mutation. The mutation frequency by histology was:
adenocarcinoma 59% (54/91), squamous 22% (10/45), large cell 67% (4/
6), and small cell 18% (5/28). The most frequent mutations were KRAS
and EGFR (16% and 12%, respectively). Fifty-two of 195 (27%) pts
have been prospectively enrolled to clinical trials. Twenty-three of 52
(44%) pts have “actionable” mutations. Mutations were considered
“actionable” if there is a known treatment that specifically targeted the
mutation or a target along the same molecular pathway. The other 29 pts
have either no mutation detected or nonactionable mutations. Among 23
enrolled pts with actionable mutations, 8 pts were enrolled to at least 1
treatment trial of agent(s) matched to the mutations, including 2 of 4
(50%) pts with FGFR mutations, 2 of 7 (29%) pts with MET mutations, 1
of 4 (25%) pts with SMO mutations and 1 of 24 (4%) pts with EGFR
mutations. Fifteen pts were enrolled to trials not matched to their mu-
tations including 7 pts in immunotherapy trials, 3 pts in antibody-drug
conjugate trials, and 4 pts with cytotoxic agents. Response rates and
durations will be presented.
Conclusions: Prospectively identifying lung cancer mutations andmatching
pts to clinical trials by NGS is feasible in the community. Forty-one percent
of ptswith lung cancer had at least onemutation. Twenty-sevenpercent of the
(52/195) pts were prospectively enrolled to clinical trials. Forty-four percent
(23/52) pts have actionable mutations and 8 of them were enrolled to
matched trials. As NGS panels expand and tissue procurement improves,
genomic profiling in the community could potentially broaden treatment and
clinical trial options for patients with lung cancer.
Author Disclosure: Z. Ma: None. T.M. Bauer: None. P.K. Chandra: A.
Employee; PathGroup. F. Honoraria; Pfizer. G. Consultant; Pfizer. M.
Stock; PathGroup. D.J. Haynes: None. J.L. Prescott: A. Employee;
PathGroup. D.M. Stults: None. S.F. Jones: None. J.D. Hainsworth:
None. J.R. Infante: None. H.A. Burris: None. D.R. Spigel: None.
267Genomic Mechanisms Underlying the Pathogenesis of Non-SmallCell Lung Cancer (NSCLC) in Young PatientsTranslational Research
Y. Feng,1 V. Varadan,2 S. Fink,2 W. zhang,3 L. Yin,3 A. Liu,4 K. Guda,2
and P.C. Ma3; 1University Hospital, Case Medical Center, Cleveland, OH,2Case Western Reserve University Comprehensive Cancer Center,
Cleveland, OH, 3Cleveland Clinic Taussig Cancer Institute, Cleveland,
OH, 4Cleveland Clinic Pathology and Laboratory Medicine Institute,
Cleveland, OH
Purpose/Objective(s): The incidence of NSCLC in patients (pts) under 45
years of age is w2% of total cases, with annual newly diagnosed cases
reaching 4500 in United States alone. Majority of these pts are diagnosed
at an advanced stage with poor outcomes. The underlying causative
mechanisms in the pathogenesis and progression in these pts remain
largely unknown, and could be different from elderly pts. The objective of
this study is to examine the genomic features of NSCLC in young pts
which therefore offer new opportunities for the development of novel
prognostic markers and therapeutic avenues.
Materials/Methods: We identified a total of 20 pts (60% female) diag-
nosed with NSCLC at an age of � 45, who underwent surgical resection
for the primary tumors or metastatic lesions from 2000 to 2012. Genomic
DNA from FFPE tumor samples and paired normal lung tissue/peripheral
blood was subjected to whole exome sequencing, and variant calling
performed using GATK and/or SOAPsnp algorithms to identify somatic
mutations in individual tumors. Pathway and protein-protein interaction
(PPI) network analysis on mutant genes was performed using KEGG/NCI-
PID databases and HotNet suite, respectively.
Results: Majority of the tumors had adenocarcinoma (nZ12) or squamous
cell (nZ4) histology. Six pts never smoked, while the others had a median
30 pack-year history. As expected, we found a significantly higher muta-
tion rate in smokers (Median, 3.47/Mb) compared to never-smokers
(Median, 0.76/Mb), with G:C/T:A transversions being more common in
smokers, and C:G/T:A transitions among never-smokers. Key driver
genes such as TP53 (50%) and KRAS (17%) showed mutations exclusively
in smokers, whereas EGFR mutations (14%) were observed specifically in
never-smokers. More important, global pathway/PPI analysis of mutant
genes revealed distinct sub-networks associated with cell adhesion/EMT
processes with a 7-fold enrichment in mutation frequency in these young
pts when compared to their overall frequencies in the COSMIC/TCGA
lung cancer dataset.
Conclusions: Our study identified novel candidate genes/pathways that
may play a key role in young age-at-onset NSCLC. Successful validation
of our findings would eventually lead to tailored therapies benefiting
younger pts.
Author Disclosure: Y. Feng: None. V. Varadan: None. S. Fink: None. W.
Zhang: None. L. Yin: None. A. Liu: None. K. Guda: None. P.C. Ma:
None.
268A Plasma MicroRNA Panel to Diagnose Esophageal Squamous CellCarcinoma and Predict the Effect of Radiation TherapyTranslational Research
Q. Yu,1 B. Li,1 and S. Fu2; 1Department of Radiation Oncology, Sixth
People’s Hospital of Jiao Tong University, Shanghai, China, 2Department