miRNAs and biomarkers

1
miRNAs and biomarkers

• Gabriella Sozzi

2

• diagnostic microRNAs in lung tumors
• stratifying lung cancer molecular subtypes (
  Landi L. et al)
• prognostic microRNAS in tumors
• miRNA expression profiles to predict clinical
  outcomes of resectable SCLC patients (Nan Bi et
  al)
• MicroRNAs associated with survival in malignant
  pleural mesothelioma patients (Kirschner et al)
• diagnostic microRNAs in biological fluids
• sputum miRNA expression profiles for the
  detection of non-small cell lung cancer (Razzak
  et al)
• plasma miRNA test for lung cancer screening
  (Sozzi et al)
• Biomarker-Driven Programs for Lung Cancer
  Screening
• General considerations (Massion P.)

3
small noncoding RNAs that regulate gene
expression by binding complementary sequences of
target mRNAs and inducing their degradation or
translational repression Evolutionary
conserved One miRNA has multiple targets
microRNA a new class of biomarkers
4
Diagnostic/prognostic miRNAs in lung cancer
Diagnostic miRNA signatures
Prognostic miRNAs
let-7a, miR-221, miR-137, miR-372 miR-182 Yu
et al., 2008
?let-7a, -34a, 34c, 25, -91
Landi et al., 2010
5
Lung cancer meta-signature miRNAs
Urmo Vosa Int. J. Cancer 2013

• Urmo Võsa
• Int. J. Cancer 2013
• 20 published miRNA studies
• 598 tumor and 528 non-cancerous samples
• 15 miRNA metasignature
• robust rank aggregation method

6
microRNA plasma/serum-based biomarkers for
cancer detection?

• Blood-based miRNA studies are in their infancy
• miRNA remain rather intact and stable in
  plasma/serum
• Simple universally applicable assay for
  quantification (i.e. qRT-PCR)
• In lung cancer plasma/serum levels of miRNAs
  might have diagnostic (Silva, ERJ 2010 Shen, Lab
  Invest 2010 Foss, J TO 2011 Boeri PNAS 2011
  Bianchi EmboMolMed 2011 Hennessey, PLoS One
  2012) and prognostic value (Hu, JCO, 2010).

 
 
 
miRNAs have been found packaged in exosomes
derived from multivesicular bodies (7) or be
exported in the presence of RNA-binding proteins
(i.e. Ago-2)(8) or might be exported
microvesicles shed during membrane blebbing (9).
Once in the extracellular space, these miRNAs
could be taken up by other cells, degraded by
RNases, or excreted(10).
7
MO 16.01 Different Micro-RNA Expression In Lung
Adenocarcinoma With Molecular Driver Events
Lorenza Landi1
Pierluigi Gasparini2, Stefania Carasi 2, Carmelo
Tibaldi1 , Luciano Cascione2, Greta Alì3, Armida
DIncecco1, Jessica Salvini1, Gabriele Minuti1 ,
Antonio Chella3 , Gabriella Fontanini3, Federico
Cappuzzo1 and Carlo M.Croce2 1 Istituto Toscano
Tumori, Dipartimento di Oncologia, Livorno
Italy 2 The Ohio State University, Comprehensive
Cancer Center, Department of Molecular Virology,
Immunology and Medical Genetics, Columbus, OH,
USA 3 Azienda Ospedaliera Universitaria Pisana,
Pisa, Italy  
8

• Trial Background
• Oncogenic driving mutations identify lung
  adenocarcinoma with different prognosis and
  sensitivity to targeted therapy
• Recent studies have suggested that miRNAs could
  be useful for stratifying lung cancer subtypes,
  however miRNAs deregulation in NSCLC with ALK
  translocation, EGFR or KRAS mutations is largely
  unknown
• Aim
• Identify miRNA signature differences according to
  the presence of specific oncogenic driver
• Methods
• Retrospective analysis of a cohort of 67 NSCLCs
  matched with 17 normal lung tissues
• RNA was isolated from FFPE using the Recover ALL
  kit (Ambion) and miRNA levels were analyzed using
  the NanoString miRNA V2 panel
• Data were processed according to manufacture
  guidelines. We used Limma to test for
  differential expression analysis data
• The miRNAs expression between tissues for all
  RT-qPCR was analyzed using the parametric t-test
  (unpaired,2-tailed for validation)

MO 16.01 Different Micro-RNA Expression In Lung
Adenocarcinoma With Molecular Driver Events -
Landi L
9
Patient Characteristics
other histology included patients with clear
cell carcinoma EGFR wild type (wt) included
patients EGFR wt and KRAS wt and ALK negative
Codon 12 exclusively defined by break-apart
FISH assay.
MO 16.01 Different Micro-RNA Expression In Lung
Adenocarcinoma With Molecular Driver Events -
Landi L
10
Results
hsa-miR-515 family expression in normal versus
tumor and according to molecular events
hsa-miR-515 family Normal ALK pos EGFR WT EGFR mut KRAS mut
miR-520d-5phsa-miR-518a-5phsa-miR-527 5.3 5.2 2.7 2.0 1.4
miR-520h 4.9 5.3 2.8 3.0 2.0
miR-548d-3p 5.7 5.0 2.6 2.5 1.3
miR-548q 5.5 3.7 2.0 1.8 1.3
miR-549 7.1 5.7 2.5 1.8 1.5
p lt 0.001
MO 16.01 Different Micro-RNA Expression In Lung
Adenocarcinoma With Molecular Driver Events -
Landi L
11
Conclusions

• miRNAs profile significantly differs in lung
  cancer patients with ALK translocation, EGFR
  mutations and KRAS mutations
• Prognostic and predictive role of several miRNAs
  are currently under investigation
• miRNAs expression could represent an useful tool
  to refine diagnosis of oncogene addicted NSCLC
• Targeting miRNAs could represent a potential
  strategy to modulate sensitivity to biological
  agents

MO 16.01 Different Micro-RNA Expression In Lung
Adenocarcinoma With Molecular Driver Events -
Landi L
12
MicroRNA Signature Predicts Survival in
Resectable Small Cell Lung Cancer
Nan Bi, Jianzhong Cao, Yongmei Song, Jie Shen,
Wenyang Liu, Jing Fan, Guogui Sun, Tong Tong, Jie
He, Yuankai Shi, Xun Zhang, Ning Lu, Qimin Zhan,
and Luhua Wang Cancer Hospital and Cancer
Institute, Chinese Academy of Medical Sciences
Peking Union Medical College, Beijing 100021,
China
13
Study Design
Training set (n42)
R
RNA isolation
Testing set (n40)
Patients
Tissue
Normal Lung (n3)
Micro- array
Identifying miRNA signature associated with
overall surivival
42 patients
miRNA analysis
Good prognosis
Internal validation
qRT- PCR
Bad prognosis
40 patients
14
Results (3)-Training Set (N42)
The expression levels of miR-886-3p and miR-150
are lower in SCLC tumors than those in normal
lung tissues.
P0.05
P0.09
NL
SCLC
SCLC
NL
N3
N3
N42
N42
15
Results (3)-Training Set (N42)
miRNA signature 0.545microRNA150
0.617microRNA886-3p
P0.02
Low risk (N21) MST not reached
High risk (N21) MST 12.6 months
D
16
Results (5)-Test Set (N40)
P0.005
Low risk (N20) MST not reached
High risk (N20) MST 18.9 months
17
Results (6)- MiRNA Signature Predicts PFS in
both Training and Test Groups
B
A
Low risk
Low risk
High risk
High risk
Test group (N40)
Training group (N42)
18
Results (7)- Multivariate Regression Analysis of
MiRNA Signature and Survivals in Test Set (N40)
Variable Hazard Ratio 95CI P value
Overall survival
miRNA signature (low risk vs high risk) 0.26 0.10-0.69 0.007
Age (60 vs lt60) 1.96 0.77-5.02 0.16
Gender (female vs male) 1.12 0.11-11.71 0.92
Smoking status (no smoking vs smoking) 0.57 0.07-4.49 0.60
Progression-free survival Progression-free survival Progression-free survival Progression-free survival
miRNA signature (low risk vs high risk) 0.36 0.15-0.86 0.02
Age (60 vs lt60) 1.45 0.60-3.52 0.41
Gender (female vs male) 1.15 0.11-11.73 0.91
Smoking status (no smoking vs smoking) 0.45 0.06-3.46 0.44
19
Conclusion

• A miR-150/miR-886-3p signature was correlated
  with the survivals in 42 resectable SCLCs and
  validated independently with another 40 SCLC
  cases.
• The expression levels of both miR-150 and
  miR-886-3p were lower in tumors than in normal
  lung tissues, indicating both of them could serve
  as tumor suppressor genes in SCLC.
• MicroRNAs may serve as promising prognostic
  markers as well as noval therapeutic targets for
  SCLC.
• Larger sample size and function studies are
  warranted to validate our ?ndings.

20
MicroRNAs miR-17-5p, miR-21 and miR-210 are
associated with survival in malignant pleural
mesothelioma patients undergoing extrapleural
pneumonectomy
Michaela B Kirschner1, Yuen Yee Cheng1, Steven
C Kao1,2, Brian C McCaughan3,4, Nico van
Zandwijk1, Glen Reid1 1Asbestos Diseases
Research Institute, University of
Sydney 2Department of Medical Oncology, Sydney
Cancer Centre 3Cardiothoracic Surgical Unit,
Royal Prince Alfred Hospital Sydney 4The Baird
Institute and Sydney Medical School, University
of Sydney
21
Patient characteristics

• Patients undergoing EPP in Sydney between 1994
  and 2009
• Series previously used to assess NLR and
  Calretinin (Kao et al, JTO, 2011)
• Complete Cohort 85
• Patients with RNA 64

Training Set (microarrayRT-qPCR)
Test Set (RT-qPCR)
Variables Long survivors (N8) Short survivors (N8)
Median Age (range) 51.5 (37 64) 62 (47 70)
Gender Male Female 6 2 6 2
Histological Subtype Epithelioid Biphasic 8 0 8 0
Stage I II III IV 0 1 7 0 0 0 8 0
Induction Chemotherapy Yes No 0 8 0 8
Median survival (mo) 57.2 ( 45.83 90.48) 6.4 (1.94 8.28)
Variables All patients (N48)
Median Age (range) 58 (22 - 74)
Gender Male Female 37 11
Histological Subtype Epithelioid Biphasic 31 17
Stage I II III IV 2 8 32 6
Induction Chemotherapy Yes No 13 35
Median survival (mo) 15.28 (0.07 81.18)
22
Kaplan-Meier and Multivariate Analysis

• Classic prognostic factors (N48)
• Female gender (49.8 mo vs 14.6 mo in males,
  p0.019)
• Epithelioid histology (18.17 mo vs 12.16 mo in
  biphasic, p0.048)

• Cox-Regression for each microRNA combined with
  classic prognostic factors (Histology, age,
  gender, stage)

Hazard Ratio 95 CI p-value
miR-17-5p Low High 2.26 1.04 4.93 1 (ref) 0.041
Factor Hazard Ratio 95 CI p-value
miR-21 Low High 4.12 1.86 9.14 1 (ref) lt0.001
Factor Hazard Ratio 95 CI p-value
miR-210 Low High 1.46 0.73 2.89 1 (ref) 0.283
23
Conclusions and Future Directions

• Lower expression levels of three microRNAs in
  tumour tissue are associated with longer survival
  of patients undergoing EPP
• miR-17-5p and miR-21 remain significant in a
  multivariate model including classic prognostic
  factors

• Those microRNAs have the potential to assist in
  better selection of patients considered for EPP

• Validation in independent samples sets is
  required
• Combination of several microRNAs as potential
  prognostic signature

24
miRNAs in biological fluids

• P2.20-011 A prospective clinical study
  evaluating stage dependent sputum micro-RNA
  expression profiles for the detection of
  non-small cell lung cancer
• Authors  Rene Razzak1, Eric L.R. Bédard1, Julian
  O. Kim2, Sayf Gazala1, Linghong Guo2, Sunita
  Ghosh2, Anil A. Joy2, Tirath Nijjar2, Eric Wong1,
  Wilson H. Roa2 1University of Alberta, Edmonton,
  AB/CANADA, 2Cross Cancer Institute, Edmonton,
  AB/CANADA

25
Our objective was to utilize an efficient,
cost-effective panel consisting of 3 miRNAs
(miR-21, miR-210 and miR-372) for prospective
validation as a potential means of accurately
detecting NSCLC. This panel was selected based on
retrospective analysis of 11 miRNAs our group had
previously undertaken using separate NSCLC and
control cohorts.
26

• 21 early NSCLC ( Stage II) patients, 22 advanced
  NSCLC ( Stage III) patients and 10 control
  subjects were prospectively accrued. A single
  sputum sample was obtained through spontaneous
  expectoration from each study participant.
• miR-21, miR-210 and miR-372 expression was
  conducted on each sputum sample and normalized to
  an endongenous control (U6) relative to a MRC-5
  reference sample, using RNA reverse transcription
  and Quantitative real-time Polymerase Chain
  Reaction (RT-qPCR).
• Statistical evaluation consisted of unsupervised
  hierarchical cluster analysis of the
  experimental-normalized miRNA expression profiles
  using within-group linkage.

27
Comparing early NSCLC to controls, the use of
miR-21, miR-210 and miR372 expression yielded a
diagnostic sensitivity of 66.7 and a specificity
of 90.0. Advanced NSCLC patients had an improved
sensitivity of 81.8 with the same specificity of
90.0.
The utilization of miR-21, miR-210 and miR372
sputum expression might provide a sensitive and
specific means of detecting NSCLC. The potential
linkage between their expression and NSCLC stage
may account for the higher sensitivity observed
in the advanced NSCLC group. Future use of this
promising panel on a larger population will be
required to establish its potential application
as a screening tool.
28

Plasma miRNA test for lung cancer screening
Gabriella Sozzi
29
2005 - 2011
Smoking cessation Lung function assessment
blood sampling
4,000 Smokers 50 years
gt 100,000 biological samples

• LDCT

R
Pastorino U. et al., Eur J Cancer Prev. 2012
30
Study Design Aims

• Diagnostic performance of miRNA test (3 levels,
  H-I-L risk MSC classifier) for lung cancer
  detection across LDCT and observational arms
• Combination of LDCT and plasma miRNA test
• Prognostic value of the miRNA assay

Sozzi G. et al., in press
31
Diagnostic and prognostic performance of MSC
Total MSC (risk of lung cancer) MSC (risk of lung cancer) MSC (risk of lung cancer)
Total High (H) Intermediate (I) Low (L)
All subjects 939 63 (6.7) 159 (16.9) 717 (76.4)
No lung cancer 870 32 (3.7) 130 (14.9) 708 (81.4)
Lung cancer 69 31 (44.9) 29 (42.0) 9 (13.0)
performance SE87, SP81, PPV27, NPV99 SE87, SP81, PPV27, NPV99 SE87, SP81, PPV27, NPV99
Lung cancer deaths 19 12 (63.2) 6 (31.6) 1 (5.3)º
Lung cancer, stage I? 37 14 (37.8) 19 (51.4) 4 (10.8)
Lung cancer, stage II-III? 12 5 (41.7) 4 (33.3) 3 (25.0)
Lung cancer, stage IV? 19 11 (57.9) 6 (31.6) 2 (10.5)
SE, SP, PPV and NPV were calculated combining
pre-specified MSC High and Intermediate versus
Low risk. P0.0366, based on the
Cochran-Armitage test for trend in the proportion
of deaths across strata of MSC risk groups among
subjects with lung cancer. º plasma sample
obtained 30 months before disease detection. ?
tumor stage information was not available in one
patient.p0.49 for association of MSC with tumor
stage
Sozzi G. et al., in press
32
Time dependency analysis of diagnostic
performance of MSC, at 6, 12, 18 and 24 months
intervals between blood sampling and lung cancer
diagnosis ( according to Heagerty PJ )
Months from blood sampling to lung cancer detection SE SP PPV NPV
6 83 80 18 99
12 86 81 22 99
18 86 81 23 99
24 87 81 25 99
Sozzi G. et al., in press
33
Modulation of the miRNA signatures in plasma
samples collected pre-disease, at time of disease
and after surgery (disease-free) from 20 pts
miRNAs returning at the normal levels after surgery miRNAs remaining deregulated after surgery
660 451
106a 197
17 142-3p
92a  
320  
486-5p
28-3p
Median time 18 months (4-46)
Median time 20 months (5-28)
Patient developing a second primary lung cancer
At surgery
Post
Patient developing surrenal metastases
LDCT
34
Biomarker-Driven Programs for Lung Cancer
Screening
Pierre P. Massion, MD Thoracic
Program Vanderbilt University Nashville, TN WCLC
Oct 30th, 2013
35
Biomarkers in the natural history of lung cancer
Disease non-measurable
Lung nodules
Diagnosis
Recurrence
Behavior
Diagnostic BM
BM of risk
BM of Response

Screening programs
Therapeutics
36
Early Lung cancer diagnostic biomarkersSullivan-P
epe, JNCI 2001- EDRN
Candidates Phase 1 Phase 2 Phase 3 Phase 4 Phase 5
  Discovery, Prediction Assay validation Retro-longitudinal Prospective screening Cancer Control
SERUM/PLASMA          
MALDI TOF MS profiling x x x     
Autoantibodies x x x   x  
Specific antigens /proteins x x x     
miRNA x x  x     
DNA methylation Blood x  x      
Circulating Tumor cells x        
TUMOR/airway epith Preinvasive histo/cytology x x x    
DNA methylation x x  x    
RNA airway signature x x  x    
MALDI MS profiling x x      
Chromosome aberrations x  x      
SPUTUM/EBC          
DNA Methylation Sputum x x  x    
DNA CN -FISH x  x      
VOCs x x
37
Indeterminate Pulmonary Nodules (6-15 mm)
High Prob
Low probability
38
How good should the biomarker be?

• Better than standard of care.
• What are the metrics?
• Performance of the test PPV NPV
• ROC curves (TPR vs FPR). C index comparison
• Net reclassification Improvement (NRI) index
• Change in decision making.
• De-emphasize Sensitivity and Specificity
• Irrelevant (except in early phase of marker
  evaluation)
• Not stable across populations
• Require dichotomization of marker values (loss of
  information)
• No information on added value
• Not actionable metric PPV or NPV are.
• Independent of the prevalence of the cancer.

Pecot, CEBP 2012
39
Plasma C4d levels (a stable complement split
product) in early stage lung cancerPhase 2
C4d in plasma samples from early lung cancer
N50
N50
Ajona et al, JNCI 2013
40
C4d levels in screening detected lung cancer
N32
N158
Ajona et al, JNCI 2013
41
A Blood-Based Proteomic Classifier for the
Molecular Characterization of Pulmonary
NodulesPhase 3

• Shotgun Proteomic analysis of tumors.
• Selected candidate proteins for testing in the
  blood
• Developed 13 multiple reaction monitoring MRM
  assays. LRP1, BGH3, COIA1, TETN, TSP1,
  ALDOA, GRP78, ISLR, FRIL, LG3BP, PRDX1, FIBA,
  GSLG1
• Training and testing algorithm.

Li et al. Sci Transl Med. 2013 Oct
42
13 MRM predictor of lung cancer among 247 lung
nodules 4-30 mm (prev 15)
A negative test implies a gt2 fold decrease risk
for cancer. High NPV of the test would obviate
1/4 patients with benign nodules from a biopsy
Li et al. Sci Transl Med. 2013 Oct
43
7 Autoantibody signature Phase 4
CAGE, GBU 45, HER2, p53, c-myc, NY-ES0-1 and MUC1
Boyle, Annals of Oncology 2010 Lam, Cancer Prev
Res 2011 Chapman Tum. Biol. 2012 Jett, Lung
Cancer 2013 in press
44
7 Autoantibody signatureEarlyCDT- Lung Oncimmune
189 nodules tested with the 7 AAB test
Cases Controls   Sensitivity 44.2
Profile 19 17 36 Specificity 88.4
Profile - 24 129 153 PPV 52.8
43 146   NPV 84.3
Prevalence 0.23
RR 3.36
In nodules 8-20 mm, the RR is 4.6
P2.20-006 Autoantibodies to a panel of lung
cancer-associated antigens can provide
significant discrimination between malignant and
non-malignant lung nodules P. Massion
45
Personalizing the management of indeterminate
pulmonary nodules
46
Clinical utility of a diagnostic biomarkerstudy
design
Positive
Biopsy
test
Outcomes Early stage Futile Thorac. Survival Decr
ease cost
Negative
3 mo CT F/U
Randomize IPNs
Biopsy
SOC (Guidelines)
No test
3 mo CT F/U
Randomization of nodules based on the use of a
biomarker test. Proves that biomarker
affects patients outcome Proves that biomarker
test affects patients outcome when compared with
unselected use of same Standard Of Care.
47
Conclusions

• Many early detection candidate biomarkers exist
• Few are validated or tested in preclinical
  setting. Priority to validate existing
  candidates.
• We need to de-emphasize Sensitivity and
  Specificity and emphasize NPV or PPV with change
  in decision making.
• BM should provide knowledge about added value and
  therefore should be integrated to clinical,
  laboratory and imaging routine data.
• To demonstrate clinical utility requires
  significant investment in effort and resources
  towards biomarkers driven clinical trial.