Master 1

1
Ovarian Cancer Early Detection Panel
Nicole Urban Fred Hutchinson Cancer Research
Center Canary Foundation Science Team
2
Cancer BiomarkersOpportunities for Impacting
Clinical Care
Ludwig JA Nat Rev Cancer Nov 2005
3
Preclinical blood samples are useful because the
behavior of a marker prior to diagnosis
determines its utility as a diagnostic (A), early
detection (B), or risk (C) marker
4
A longitudinal algorithm is useful if marker
levels vary among healthy women, and rise as
cancer develops
5
FDA Approved Biomarkers for Gynecological Cancers

• Ovarian Cancer
• CA 125- monitoring response to therapy
• Cervical Cancer
• HPV DNA - stratify women with abnormal cytology
  for further follow-up identify HPV infection
• Breast Cancer
• HER2/NEU protein tissue- prognosis and treatment
  selection
• HER2/DNA tissue -prognosis and treatment
  selection
• HER2/NEU protein serum- monitoring response to
  therapy
• ER/PR protein tissue- treatment selection

6
Novel markers of ovarian cancer HE4 has been
studied independently by two institutions and
found to be promising for early detection

• Hellstrom I, et al. The HE4 (WFDC2) protein is a
  biomarker for ovarian carcinoma. Cancer Res. 2003
  Jul 163(13)3695-700
• Drapkin R, et al. Human epididymis protein 4
  (HE4) is a secreted glycoprotein that is
  overexpressed by serous and endometrioid ovarian
  carcinomas. Cancer Res. 2005 Mar 1565(6)2162-9.

7
Additional markers have been reported to be
promising by academic centers

• Mesothelin (MSLN SMR)
• McIntosh MW, et al. Combining CA 125 and SMR
  serum markers for diagnosis and early detection
  of ovarian carcinoma. Gynecol Oncol. 2004
  Oct95(1)9-15.
• MUC1 (CA15-3)
• Cramer DW, et al. Conditions associated with
  antibodies against the tumor-associated antigen
  MUC1 and their relationship to risk for ovarian
  cancer. Cancer Epidemiol Biomarkers Prev. 2005
  May14(5)1125-31.
• Osteopontin
• Kim JH, et al. Osteopontin as a potential
  diagnostic biomarker for ovarian cancer. JAMA.
  2002 Apr 3287(13)1671-9.
• DcR3
• Wu Y, et al. Clinical significance of detecting
  elevated serum DcR3/TR6/M68 in malignant tumor
  patients. Int J Cancer. 2003 Jul 10105(5)724-32.

8
Methods for multiplexing and for using change
over time in a marker facilitate progress

• Multiplexing
• Gorelik E, et al. Multiplexed immunobead-based
  cytokine profiling for early detection of ovarian
  cancer. Cancer Epidemiol Biomarkers Prev. 2005
  Apr14(4)981-7.
• Longitudinal algorithm (PEB)
• McIntosh MW, Urban N, Karlan B. Generating
  longitudinal screening algorithms using novel
  biomarkers for disease. Cancer Epidemiol
  Biomarkers Prev. 2002 Feb11(2)159-66

9
What are we doing to make early detection of
ovarian cancer a reality?
Identify candidate markers that perform well in
clinical blood samples obtained prior to
treatmentAnalyze them in preclinical blood
samples obtained 1 years prior to diagnosis of
ovarian cancerSelect the best markers for a
panel Lead time (earliness of
signal) ComplementarityValidate new panel in
an independent set of preclinical samples

10
To evaluate candidate markers as diagnostic
markers, a Triage Set of clinical serum samples
from FHCRC was used

• 200 blinded serum specimens from ovarian cancer
  cases healthy women
• 41 healthy controls from a screening study
• 20 contributed blood two times, a year apart
• 47 otherwise healthy women undergoing surgery
• 24 benign surgical controls
• 68 cases
• Stage 1 11 Serous 34
• Stage 2 5 Endometrioid 7
• Stage 3 39 Mucinous 3
• Stage 4 11 Clear Cell 2
• Stage N/A 2 Other 17
• 
  Undifferentiated 5

11
Markers that complement CA125 and/or show
stability over time are good candidates for a
panel

• Contribution to a panel including CA125
• 6 markers improved the sensitivity of CA125 at
  .95 specificity
• Stability over time within women can improve
  performance using a longitudinal algorithm
• 8 markers showed correlation gt .5 for samples
  taken 1 year apart from the same woman

12
Exploiting stability over time Parametric
Empirical Bayes (PEB) algorithm

• At each screen, a womans serum is tested for
  deviance from her own normal value of each marker
  
• X of women are referred for further work-up at
  each screen
• Sensitivity is maximized within desired
  specificity
• Womens characteristics are accounted for using
  the parametric empirical Bayes (PEB) rule
• McIntosh et al, 2002

13
This work led to a SPORE/EDRN/PLCO collaboration
that will provide access to pre-clinical PLCO
samples for the development of an early detection
panel

• Scientific Leadership (coordinating centers that
  will also provide specimens and measure
  biomarkers)
• Brigham and Womens Hospital PI Daniel Cramer,
  M.D.
• Fred Hutchinson Cancer Research Center PI
  Nicole Urban, ScD
• Specimen Sites
• Fox Chase Cancer Center PI Andrew Godwin, PhD
• MD Anderson Medical Center PI Robert Bast, M.D.
• University of Alabama, Birmingham PI Edward
  Partridge, M.D.
• Assay Performance Site
• Univ. of Pittsburgh Cancer Institute PI Anna
  Lokshin

14
The Prostate, Lung, Colon and Ovary (PLCO) trial
is a valuable source of preclinical samples
15
PLCO study Hypothesis

• A panel of biomarkers will have better
  performance characteristics as a screening test
  for pre-clinical ovarian cancer than any single
  marker, and yield a longer lead time than CA125
  alone.

16
Specific aims Replicate performance of the
diagnostic panel using standard assays

• Evaluate candidate biomarkers in a new
  independent set of clinical samples representing
  four institutions
• Identify a consensus panel based on performance
  as diagnostic markers
• Evaluate the reproducibility, concordance with
  standard ELISA, and performance of bead-based
  (Luminex) assays requiring less serum for use in
  PLCO samples

17
Year One Specific Aims

• Evaluate candidate biomarkers in a new SPORE set
  of
• 160 cases (80 early-stage and 80 late-stage)
• 160 surgical controls
• 480 general population controls
• serial samples collected one year apart in 40
  healthy controls.
• Evaluate the reproducibility, concordance with
  standard ELISA, and performance of the bead-based
  (Luminex) assays.

18
Year One Inter-SPORE collaboration

• Identify a consensus panel of biomarkers that are
  
• informative on their own
• complementary when used together
• Require lt than .3 ml of serum using Luminex, and
  lt 5 additional biomarkers that can be measured
  adequately only by standard ELISA.

19
Year Two PLCO collaboration

• Estimate the lead time of each individual marker
  and establish the best marker combination
• Markers that fail to show elevation within a
  year prior to diagnosis will be eliminated
• Remaining markers will be evaluated using the
  entire pre-clinical history to estimate the lead
  time for each marker and the marker panel
• Use lt 1 ml of sera from pre-diagnostic specimens
  from PLCO subjects and 10 matched controls per
  case

20
Year Two includes discovery

• Using residual sera from false positive and false
  negative cases
• Apply high throughput proteomic discovery
  platforms
• Identify a new set of sequenced (identified)
  biomarker candidates to complement the existing
  panel.

21
The PLCO/SPORE/EDRN study will be critical

• Pooling of resources by SPORE, EDRN, and PLCO
  investigators will yield a single best marker
  panel for early detection of ovarian cancer.
• Independent preliminary work of the investigators
  provides the best current information on
  biomarkers for ovarian cancer.
• Lead times of individual markers will be
  estimated to ensure the best marker panel.
• Collaboration among leading NCI groups will
  ensure translational potential of the marker
  panel.

22
Research challenges

• Estimation of marker lead time requires access to
  preclinical samples in order to identify the best
  panel
• Specimen quantities are small in stored samples
  from trials
• Presence or absence of disease cannot be
  documented from preclinical samples
• Bead-based assays are needed to facilitate panel
  identification and validation research

23
Bead-based assays have been developed for use in
preclinical samples where specimen quantity is
limited

• Available antibodies pairs were tested on a
  bead-based platform to assess
• Assay feasibility (affinity)
• Accuracy in assessing known antigen
  concentrations
• CA125 (two assays) and HE4 were measured using a
  bead-based assay in a triage set of ovarian cases
  (n66) and controls (n138)
• Bead-based assays were evaluated for
• Reproducibility
• Validity
• Screening performance
• Scholler N, Clin Ca Res, 2006

24
CA125 performs well as a bead-based assay
Available commercially 4 CA125 monoclonal
antibody pairs Two bead-based assays were
optimized and evaluated for reproducibility,
validity and performance in serum samples Best
bead-based assay uses antibodies from RDI CV
18.2 in negative controls 18.9 in positive
controls Correlation between replicates 0.99
overall 0.83 in healthy controls Correlation
with CA125II 0.95 overall 0.64 in screening
controls AUC based on 64 cases, 55 screening
and 70 surgical normals AUC 0.86 for
cases vs screening controls (CA125II AUC 0.90)
AUC 0.83 for cases vs surgical controls
(CA125II AUC 0.84)
25
HE4 also performs well as a bead-based assay
1 HE4 monoclonal antibody pair available through
collaboration A bead-based assay was optimized
and evaluated for reproducibility, validity and
performance in serum samples CV 32.6 in
negative controls 26.6 in positive
controls Correlation between replicates 0.95
overall 0.86 in healthy controls Correlation
with ELISA 0.95 overall 0.86 in screening
controls AUC based on 64 cases, 55 screening
normals, 70 surgical normals 0.93 for cases
vs screening controls (CA125II AUC 0.90) 0.86
for cases vs surgical controls (CA125II AUC
0.84)
26
Diagnostic performance of HE4 and each of 2 CA125
bead-based assays is comparable to the CA125II
RIA assay (Triage set 64 cases vs 125 controls)
Scholler, Clin Ca Res, 2006
27
The bead-based assays were used to explore the
complementarity of CA125 and HE4

• A composite marker (CM) was defined as a linear
  combination of HE4 and the RDI CA125 assay
• Marker levels were converted to logs and
  standardized
• Logistic regression was used to estimate the
  weights for each marker
• CM 0.56 x RDI 1.20 x HE4
• A similar CM was estimated for HE4 and the
  Fitzgerald bead-based assay CM 0.42 x Fitz
  1.28 x HE4
• 
  Scholler N, Clin Ca Res, 2006

28
Diagnostic performance for 2 composite markers
(CA125 HE4) using bead-based assays is better
than the CA125II RIA assay (Triage set 64
cases vs 125 controls)
Scholler, Clin Ca Res, 2006
29
Canary scientists are working toward completion
of an ovarian cancer early detection marker panel

• Methods have been developed for
• Discovering and prioritizing candidate markers
• Combining markers to form a composite marker
• Using marker history in a longitudinal decision
  rule for early detection
• Developing specimen-efficient bead-based assays
  for validation research
• Several candidate markers have been identified
  that perform well as diagnostic markers
• Additional markers will likely be needed to
  detect ovarian cancer early
• New proteomics technologies make discovery in
  serum possible

30
Classification of Biomarkers

• DNA based
• DNA copy number, translocations, methylation
  changes, mutations, single nucleotide
  polymorphisms (SNPs)
• RNA based
• Differential expression of mRNA, regulatory
  miRNA
• Protein based
• Tumor antigens, post-translational
  modifications, receptors

31
Typical Approach for Biomarker Discovery
Ludwig JA Nat Rev Cancer Nov 2005
32
Candidate markers were identified by comparing
malignant to healthy ovarian tissue (FHCRC) or
and/to healthy tissue from other organs
(Stanford)At FHCRC

• Tissue (and matched blood) samples obtained at
  time of primary surgery
• Multiple rounds of cDNA microarray interrogation
• malignant ovarian tumors 85
• benign ovarian tumors 19
• ovarian tissue from healthy women 66
• RT-PCR to confirm and quantify over-expression
• Schummer et al, Gene, 1999.

33
Using this approach, the Canary Science Team has
identified about 20 candidate biomarkers
Original FHCRC list Stanford additionsCA125 IG
F2 HE4 (WFDC2) CHI3L1 MSLN (Mesothelin) LCN2
MUC1 (CA15-3) BMP7 SLPI TACSTD1 (EPCAM)
CD24 SPONDIN1 SPINT2 (bikunin) GALNT3
PRAME LAMC2 KRT8 CDH6 IFI27 PAX8

34
Four new markers have been evaluated in a
Mini-Triage Set, a subset of the Triage Set used
in previous work

• Purpose of Mini-Triage Set
• to evaluate for the first time in individual
  blinded patient sera the performance
  characteristics of an assay
• Design
• 72 patient samples plus reference standards are
  assayed for the biomarker of interest
• The results are analyzed using ROC curves to
  predict the clinical utility of the assay

35
Cases in Mini Triage Set
36
Specimens in Mini Triage Set
37
Established assays in the Mini-Triage Set Single
Assay ROC Curves
38
Established assays Single Assay ROC Curves
Stages 1(n 8) and 2(n 4) cases only
39
Novel assays in the Mini-Triage Set Single Assay
ROC Curves
40
Novel assays Single Assay ROC Curves Stages 1(n
8) and 2(n 4) cases only
41
Composite markers (CM) using novel markers with
CA125 and HE4 for all cases and early-stage cases
only
42
The challenge Ten-year survival for ovarian
carcinoma is poor unless disease is confined to
the ovary at diagnosis
43
Acknowledgments
Marker identification and development
Institute for Systems Biology Lee Hood FHCRC
Nathalie Scholler, Michel Schummer, Charles
Drescher, Garnet Anderson Cedars-Sinai (LA)
Beth Karlan U of WA Nancy Kiviat, Ingegerd
Hellstrom Stanford Pat Brown BC Cancer
Agency Brad Nelson, Xiaobo Duan Panel
definition and evaluation FHCRC Martin
McIntosh, Garnet Anderson Harvard Dan Cramer,
Steve Skates Funding DOD, NCI/SPORE, Canary
Foundation Institutional support FHCRC, Swedish
Cancer Institute
44
NCI Collaboration Phase II/III Validation Study
of a Consensus Panel of Early Detection Serum
Markers

• Two year study begun in August 2005
• Includes SPORE, EDRN and PLCO investigators