Transcript Evaluation of Investigator’s Progress

Early Detection of Clinically Significant
Cancer – Are We There Yet?
Christos Patriotis, PhD
Cancer Biomarkers Research Group
Division of Cancer Prevention, NCI
Early Detection Can Help Reduce Mortality
from Cancer
Mortality rates from cancers where screening tools
are available are considerably lower than from
cancers for which no viable screening tools exist
(e.g., colon, breast, lung)
 Better risk assessment and early detection of cancer when
treatment is effective can help reduce mortality
 Addressing the issue of over-diagnosis can further improve
overall survival by reducing unnecessary morbidity from overtreatment
Early Detection Research Network EDRN
EDRN is a national infrastructure supported by NCI to
help move promising biomarkers for early cancer
detection and risk assessment from the laboratory
bench to clinical validation
Mission and Goals
• Discover, develop and validate biomarkers for cancer risk
assessment and the early detection of clinically significant
cancer (molecular diagnosis and prognosis)
• Conduct correlative studies and trials to validate
biomarkers as indicators of early, pre-invasive cancer or
cancer risk
• Develop quality assurance programs for biomarker testing
and evaluation
• Forge public-private partnerships
EDRN and Biomarker Research

EDRN established in 2000

Focused solely on early detection biomarker
research

Provides guidelines on reference samples,
study designs and validation

Unparalleled, fully operational, open-access
network

Strong scientific collaborations, e.g., Associate
Membership Program for translational research
Biomarker Development Pipeline
EDRN
Investigator-Driven Consortium
Assay Development
(Phase II & III)
Discovery
(Phase I & II)
U24
(RFA-CA-09-019)
U01
(RFA-CA-09-017)
U01
(RFA-CA09-018)
Validation
(Phase II & III)
U24
(RFA-CA-09-020)
Building Teams via Collaborative Groups
EDRN Defines Phases of Biomarker
Development
EDRN
Source: Pepe et. al., J. Natl. Cancer Inst. 93, 1054-1061, 2001
Vertical Integration of Biomarker Discovery,
Development and Validation
Discovery
(BDLs and others)
Candidate Biomarker Identified
(BDLs, CVCs, BRLs)
NO
Development
(BDLs, CVCs, BRLs)
YES
NO
Re-evaluate
biomarker for utility
as a predictive,
prognostic or target
marker
NO
Pre-validation
(BDLs, CVCs, BRLs,
DMCC)
Validation
(BDLs, CVCs, BRLs,
DMCC)
YES
Re-evaluate
biomarker use in
clinical setting
NO
Cost Effective?
YES
Milestones
Study design and sample selection to ensure
that candidate biomarker clinically useful
 Robust and reproducible assay
 Protocol for sample collection is fully
established and valid for cross-laboratory
validation
 Biomarker distinguishes cases from
controls with statistical validity
 Cost-effectiveness
 Biomarker is reproducible across a wide
spectrum of cases and controls collected
prospectively from retrospective cohorts
of samples – e.g. such samples may come
from recently concluded trials that have
appropriate case and control groups
 Phase IV: Biomarker is further tested and
refined in asymptomatic population and
found to be effective (i.e., reducing
incidence or mortality of disease) –
beyond the scope of EDRN
Utilization of biomarker in clinical care
setting
Foundational Studies on Phases and Study
Design for Biomarker Development
Phases of Biomarker Discovery and Validation
Preclinical
Exploratory
PHASE 1
Promising directions identified
Clinical Assay
and
Validation
PHASE 2
Clinical assay detects established
disease
Retrospective
Longitudinal
PHASE 3
Biomarker detects preclinical disease
and a “screen positive” rule defined
Prospective
Screening
PHASE 4
Extent and characteristics of disease
detected by the test and the false
referral rate are identified
Cancer
Control
PHASE 5
Impact of screening on reducing
burden of disease on population is
quantified
PRoBE Study Design:
Prospective-specimen collection,
Retrospective Blinded Evaluation
RESOURCES:
Sample Reference Samples
 Clinical validation studies are expensive

Only few biomarkers may succeed
 There is a need for a triage system that
allows “Go or No-Go” decision
 EDRN has developed a mechanism through
which biomarkers are first tested in
Standard Reference Samples constructed
for a specific intended clinical use
 If initial evaluation is successful, then a
large, multi-center validation study is
planned
Reference Samples are
sets of cases and
controls collected based
on PRoBE principles and
statistically powered to
allow rapid assessment
of technologies and
biomarkers discovered
through a wide variety of
technology platforms.
First-ever concept originated and implemented within
EDRN for rapid evaluation of technologies and
biomarkers
Breast Standard Reference Set
Clinical Application:
• An assembled set of well-characterized and annotated specimens for
testing biomarkers which, in conjunction with mammography, can
detect and discriminate breast cancer from benign conditions.
Prospective (pre-diagnostic; PRoBE) collection of samples for early
detection of BC:
•
•
Samples and specified Common Data Elements (CDEs) are collected from
eligible individuals referred to diagnostic radiology after mammography
(and/or another imaging modality, e.g., US), but prior to surgery and
diagnosis
Samples collected at mammography screening clinic and confirmed as
cases or controls following referral, biopsy, and pathology report
Reference Set: serum and plasma in 200 ul aliquots:
•
•
•
•
•
207 Incident ICA
55 DCIS/LCIS
63 Benign pathology with atypia (ADH)
231 Benign pathology without atypia
276 Healthy controls (no evidence of breast disease by screening
mammography (BI-RADS 1 or 2)
EDRN Reference Sets
• Lung Cancer (3)
• Prostate Cancer
• Liver Cancer (2)
• Colon Cancer
• Pancreatic Cancer
• Bladder Cancer
Request EDRN Standard Reference Set
1. Obtain application form from EDRN public portal
(http://edrn.nci.nih.gov/resources/sample-reference-sets) and prepare
proposal as per provided requirements.
2. Submit application to NCI Program Officer for evaluation and for
review of preliminary biomarker performance data by DMCC.
3. Approval Decision:
•
•
•
Not approved - denied;
Additional information required; and/or
Approved
4. Notify approved investigator
•
•
•
NCI prepares MTA to be executed with receiving institution
DMCC creates pull-list, shipment list and ensures set is blinded
NCI-Frederick ships out reference set
5. Investigator conducts assays (4 months) and deposits data with
DMCC.
6. DMCC provides sample ID to investigator. After 3 months data is
posted on secure EDRN website. After 12 months data is posted on
EDRN public portal (unless peer-reviewed publication is imminent).
Common Data Elements
Common Data Elements provide a
set of standard terms and values
for a given domain
•
•
•
•
They are classified into organ,
epidemiological, and specimen
CDEs
Critical to coordinating and
managing a multi-center
biomarker validation study
EDRN CDE Ontology necessary
for integrating all data
warehouses, study management
systems, and informatics tools
into one harmonized knowledge
environment
Based on ISO/IEC 11179
(standard for data elements)
Captured by EDRN and
maintained by the EDRN DMCC
and IC (FHCRC/JPL)
EDRN Knowledge Environment
Access to
science data
sets
Access to biomarker
data and results
Access to specimen
information
Access to study data
http://cancer.gov/edrn (operational)
http://edrn.jpl.nasa.gov (beta; emerging capabilities)
Validation Study Management
VSIMS: Online study management system
supporting all EDRN validation studies. Built
upon the EDRN CDE repository and using
reusable modules to speed development for
new studies.
eSIS: System in development to track
the progress of all EDRN-funded
projects, including timelines, GANTT
charts, phases of development, current
study status.
How Do We Choose Winners Or Losers
Among Markers?
 Clinical Performance Characteristics
 Improved Benefits – Unmet Clinical Needs
 Cost-Effectiveness
Is the Biomarker’s Performance Reproducible
on an Independent Sample Set?
Examples
• TSP1, Kallikreins 2,3,5,11, and EPCA-2 were “no go” for further
validation after they failed on blinded testing using the EDRN
prostate cancer reference set (AUC for TSP1 and EPCA-2 dropped
from >0.95, in preliminary studies, to <0.55 on the reference set).
• In contrast, Percent[-2]proPSA held its performance (AUC=0.69)
on the same reference set. It’s performance was also validated by
a subsequent Phase II study on a separate cohort of samples
from an appropriately recruited population. An IVD is currently
under review by FDA.
Does the Biomarker have Clinical Value?
Example
DCP in preliminary data indicated better sensitivity and
specificity than that of AFP, the conventional clinical biomarker
for HCC. A large, blinded validation study indicated that
although overall, DCP does not perform better than AFP, it has
a better sensitivity and specificity in detecting HCC with viral
etiology
Impact: Biology of non-viral and virus-induced cirrhosis is different; DCP is a
better marker than AFP for early detection of virus-induced liver cancer
Importance of Failures
“As in real life, we often learn more from negative results than we do from
positive ones. It is time that we, as a community, start to regard failures as being
as informative as successes. After all, we do know the difficulty of learning from
positive only experience.”
EDRN
•Accelerator – to drive good markers through to the clinic
•Brake – to use good clinical design to eliminate markers without added
value
EDRN in the News…
Scientific Organization of Phase II/Phase III
OC Biomarker Validation Study
OC Biomarker Validation Study Lessons Learned
While many of the biomarkers can distinguish between cases and controls in
specimens obtained at diagnosis (Phase II), the vast majority of them failed in
distinguishing cancer from healthy control-associated specimens when the
samples were obtained more than 6 months prior to clinical diagnosis (Phase III)
OC Biomarker Validation Study Lessons Learned (continued)
 This outcome emphasizes the importance of using appropriate specimens
for biomarker research - from early discovery stages to clinical validation.
Bias introduced by systematic differences in the case and control specimens
must be maximally avoided by adapting the principles of PRoBE study
design
 Better understanding of the natural history of the disease and the discovery
of biomarkers in lesions, which most likely represent the precursors of
aggressively growing disease can help identify more reliable biomarkers
EDRN in the News…
Therefore, reports from studies funded by the Early
Detection Research Network (EDRN) using prospectively
diagnosed ovarian cancer data from the Prostate, Lung,
Colorectal, and Ovarian Cancer (PLCO) screening trial
have been eagerly awaited. The authors of the 2 reports
in this issue of the journal are to be commended for having
designed and conducted scientifically solid phase 3 studies,
which were nested in a large, randomized
screening trial and will serve as the standard against which
future analyses of this kind should be judged (14, 15).
Major EDRN Accomplishments
Detection/ Biomarker Assay
Discovery
Blood
Refine/
Clinical
Adapt for Clin Use
Validation
Clinical Translation


FDA IVD pending review


FDA IVD pending review
proPSA
Urine
PCA3
Urine/TMA assay for T2S:Erg
fusion for Prostate Cancer



CLIA in process
FISH for T2S:Erg fusion for
Prostate Cancer



In CLIA Lab
Aptamer-based markers for Lung
Cancer


In CLIA Lab
Proteomic Panel for Lung Cancer


In CLIA Lab
OVA1TM for Ovarian Cancer


FDA Approved
SOPs for Blood (Serum, Plasma),
Urine, Stool,

N/A
Vimentin Methylation Marker for
Colon Cancer


In CLIA Lab
ROMA Algorithm for CA125 and
HE4 Tests for Pelvic Mass
Malignancies


FDA Approved
Blood/DCP and AFP-L3 for
Hepatocellular Carcinoma


FDA Approved


Together with AFP-L3 used for
monitoring cirrhotic patients for HCC
in China
Blood GP73

Frequently used by biomarker
research community
Over-Diagnosis: Consequence of
Screening and Early Detection
Screened and “cured”
Death unrelated
to cancer
Cancer
Never screened
Sources of Cancer Over-diagnosis
1. The pathologist: diagnostic uncertainty
2. Underlying tumor biology
B. Kramer, 9/15/2011
Cancer Progression – A Heterogeneous
Process
Size
Size at which
cancer causes
death
Size at which cancer
causes symptoms
Fast
Slow
This is
over-Dx.
Very Slow
Non-progressive
Abnormal cell
Time
Death from
other causes
B. Kramer, 9/15/2011
Patterns of Rapid Increase in Cancer Incidence:
True Increase vs. Overdiagnosis
 Indicators of Increased True Occurrence of Cancer vs. Overdiagnosis
New Diagnoses
Number of
new cancer
diagnoses and
deaths
Deaths
Time
Suggests a true increase in
the amount of cancer
New Diagnoses
Number of
new cancer
diagnoses and
deaths
Deaths
Time
Suggests overdiagnosis of cancer
B. Kramer, 9/15/2011
Incidence and Mortality of Five Cancers:
(Surveillance, Epidemiology, and End Results: SEER)
6
New Diagnoses
4
Prostate Cancer
Rate
8
Thyroid Cancer
2
(per 100,000 people)
Rate
(per 100,000 people)
10
Deaths
0
1975 1980 1985 1990 1995 2000 2005
Year
Melanoma
New Diagnoses
Rate
15
10
5
Deaths
0
1975 1980 1985 1990 1995 2000 2005
Year
(per 100,000 people)
Rate
(per 100,000 people)
20
225
200
175
New Diagnoses
150
125
100
75
Deaths
50
25
0
1975 1980 1985 1990 1995 2000 2005
Year
175 Breast Cancer
150
125
New Diagnoses
100
75
50
Deaths
25
0
1975 1980 1985 1990 1995 2000 2005
Year
Rate
(per 100,000 people)
Kidney Cancer
12
10
8
6
4
New Diagnoses
Deaths
2
0
1975 1980 1985 1990 1995 2000 2005
Year
Source: HG Welch, JNCI 2010
Female Breast Cancer Incidence Trends in Connecticut
by Stage and 5-Year Time Periods of Diagnosis
200
Total
Rate per 100,000 Women-Years
100
Localized
Regional
In-situ
Distant
10
1
0.1
1940
1960
Year
1980
2000
WF Anderson, Breast Cancer Res Treat 2006
Strategies to Investigate
Overdiagnosis & Underdiagnosis
Annotate collected specimens with method of
diagnosis
• Molecular patterns of screen-detected cases are
enriched with overdiagnosed cases
• Molecular patterns of true interval cases are enriched
with aggressive cases that we need to prevent (and
target pathways for prevention)
Collect normal organ as well as the tumor
• Study cancer at tissue-level, not simply as a cell-based
disease
• Examples: prostate, breast, esophageal, melanoma
B. Kramer, 9/15/2011
Address Overdiagnosis & Underdiagnosis
• Identify and develop prognostic biomarkers
for early detection to complement screening
modalities
– Indeterminate pulmonary lesions on CT screening
– PSA/biopsy-detected Prostate Cancer
• Identify and develop biomarkers of risk of
progression of precursor lesions to
aggressive cancer
–
–
–
–
Progression of Breast BBD with/without Atypia to IBC
Progression of DCIS/LCIS to IBC
IPMN cyst progression to Pancreatic Cancer
Colon adenoma progression to CRC
Address Overdiagnosis & Underdiagnosis
(continued)
• Develop and validate biomarkers for early
detection of Interval cancers to complement
screening modalities
– Develop blood biomarkers for the early detection of TNBC,
especially for women under the age of 50 and of African
American descent. On positive test refer to more frequent
screening imaging (mammography, MRI, or other newly
developed molecular imaging)
THANK YOU