Genetic Polymorphisms Identify the Likelihood of Bone Disease in
Myeloma: Correlations with Myeloma Cell DKK1 Expression and High
Risk Gene Signatures
Brian GM Durie, MD1*, Brian Van Ness, PhD2*, Christine Ramos, PhD2*, Owen
Stephens3, Majda Haznadar2*, Antje Hoering, PhD4*, Jeff Haessler, PhD4*,
Michael S Katz, BS5*, Gregory R. Mundy, MD6*, Robert A. Kyle, MD7*, Gareth J
Morgan, MD,PhD8, John Crowley, PhD4*, Bart Barlogie, MD3*, John D.
Shaughnessy, Jr., PhD3*
1Cedars-Sinai Cancer Centre, Los Angeles, United States; 2University of Minnesota, Minnesota, United
States; 3Myeloma Institute for Research and Therapy, University of Arkansas, Little Rock, Arkansas,
United States; 4Southwest Oncology Group Statistical Centre, Cancer Research and Biostatistics, Seattle,
United States, WA 98109; 5International Myeloma Foundation, North Hollywood, California, United
States, 91607-3421; 6Vanderbilt Center for Bone Biology, Nashville, TN, United States; 7Mayo Clinic
College of Medicine, Rochester, Minnesota, United States; and 8The Institute of Cancer Research,
London, Surrey, United Kingdom, SM2 5NG
Goals of DNA SNP Correlation Studies
in Bone Disease
Identify Top DNA
SNPs
Test and validate
within TT2
dataset and
external data
sets.
Assess bone
biology and
predisposition.
Patients and Methods
Peripheral blood DNA used
282 patients in TT2 Trial evaluated
Primary classification of bone disease
Limited disease
· X-rays Negative
· 0-7 MRI Focal Lesions
Definite/ extensive disease
· X-rays Positive: focal lesions and/or osteopenia/fractures
·MRI > 8 Focal Lesions
DNA studies using "Custom Affymetrix SNP Chip" (BOAC): 3400
DNA SNPs
Evaluate correlations using univariate and recursive partitioning
methods.
Bone Disease Classification
Based upon Focal Lesions on X-ray
and/or
MRI
Pattern of Bone Disease in TT2 Dataset
Total of 282 Patients
Number of Patients
Limited bone disease
X-ray/MRI Negative
65
99
X-ray Negative
34
MRI 0-7 focal lesions
Extensive bone disease
X-rays Positive
171
183
X-rays Negative
12
MRI > 8 focal lesions
Total = 282 Patients
"Top 30" SNPs: Univariate Correlation
Bone Related
Recursive Model
Recursive Partitioning using "Top SNPs"
EPHX1
MAP4K5
DDX18
TNFSF8
Number of MRI Lesions and SNP Pattern
P value < 0.001
05
0
Pre-treatment
04
Lesions
Mean=8.66
03
MRI
246 Patients
of
Extensive Bone
36 Patients
02
Disease
1
mean=8.66
Limited Bone
O
)
Mean=3.33
EPHX1 GG
Disease
Number
mean=3.33
o
O
0
MAP4K5 AG/AA
EPHX1 GT/TT
DDX18 CG/CC
MAP4K5 GG
Bone Disease Predicted
TNFSF8 CT/TT
DDX18 CC
TNFSF8 CC
Overall Correlation Between GEP, SNP, and
Clinical MRI/X-ray Data in TT2 dataset
Stepwise Multivariate Regression Analysis
EPHX1 SNP
Univariate
MAP4K5 SNP
DDX18 SNP
TNF SF8 SNP
GEP DKK-1
GEP DKK-1
Multi-
EPHX1 SNP
variate
MAP4K5 SNP
TNFSF8 SNP
ROC Curve
Recursive Partitioning Using "Top SNPs"
X-Ray Only Model
EME1
·Excellent model
·New SNPs identified
ABCC1
CYP2C9
·Linked to predisposition
CYP1B1
·However, original model
better using X-ray plus
IL-4R
MRI versus X-ray only
Testing of Recursive Partitioning
X-ray Plus MRI Model
Percentage of Patients Correctly Classified (%)
Biologic Correlates of Top SNPs
Two Major Categories of SNPs
Bone Biology
Osteoblast
Osteoclast
Local/Systemic microenvironmental factors
Predisposition
Toxin Metabolism
· Dioxins/PBDEs/other "HAHs"
· Benzene/Diesel Exhaust/other "PAHs"
Micro Environment and Bone Biology
Systemic
Local
Top SNPs Linked to Bone Biology
SNP
Rank
MAP4K5 (GCK R)
2
1GF1R (Insulin growth factor)
6, 9
CYP19A (P450 family)
4
ADRB3 (Adrenergic 3)
11
BDKRB (Bradykinin)
10, 14
APOB (Apolipoprotein)
3, 5
LRP-1 (Lipoprotein 1)
26
SLC (Solute family)
1, 4
IL-4 7 SNPs in top 100
20
Details of Wnt Signaling
MAP 4K5 (GCK R) Binds GSK3i
·Stabilizes -catenin
·Synergistic with bone
mechanical loading
·Activated by RAS oncogene
·Co-localizes with oncoproteins
·Triggers "stress activated
protein kinase (SAPK)" pathway
·Cross activates cyclin D1
pathway
·Triggered by Bcr-Abl in CML
Target for Thalidomide Limb
Deformities
Crucial Link Between Osteoblast and
Adipocyte
Obesity Linked to Increased Myeloma
Links to Toxin Metabolism
Dioxin PBDEs
Enzyme Induction
PAHs:
Benzene
CYP1B1 other CYP enzymes
Diesel Exhaust
Activation of Carcinogens
EPHX1
Toxic Intermediaries
EME1
DNA Damage
Reduced
Adducts
immunity/viral
activation
DDX18
DNA Repair
Viral Helicase
What's Next?
The statistics are strong
The biologic links are plausible
Reasonable targeted hypotheses can now be
generated
Assessing all relevant SNPs (e.g. >27 EPHX1
variants) and determining functional significance
in myeloma patients
Potential use of SNP panel to stratify bone care
Follow through with epidemiologic studies with
Dalsu Baris and NCI team
Combine targeted and genome-wide SNP
screening
Work towards personalized molecular
approaches and ultimately prevention
Summary Conclusions
1.
Very strong statistical linkage to "Top SNPs"
2.
Key SNPs associated with Bone Biology
MAP4K5 Wnt pathway; Thalidomide Target
1GF1R- Osteoblast and myeloma cell activation
Several other bone associates SNPs: IL-4; ADRB2; BDKRB1; LRP1...
3.
Key SNPs linked to toxin metabolism
CYP1B1
P450 system
EPHX1
Epoxide hydrolase
EME1
DNA repair
DDX18
Viral Helicase
4.
Focus for further studies
Test additional data sets
Explore key SNPs in more detail especially regarding function
Initiation targeted biology/predisposition/epidemiology studies
Acknowledgements
Special thanks to CRAB
Antje Hoering, Jeff Haessler, John Crowley
John Shaughnessy & Bart Barlogie for allowing
use to test TT2 patients' DNA
SWOG and ECOG for setting up collaborative
arrangements for DNA testing in clinical trials
IMF for financial and administrative support