1. The Hierarchy of Somatic Mutations in Follicular Lymphoma
Michael R. Green, Andrew Gentles, Ramesh Nair, Jonathan Irish,
Ron Levy, Ash Alizadeh.

2. Follicular Lymphoma (FL)
Follicular lymphoma histology
black stain = T Cells (CD3)
FL flow cytometry 3X
B Cells
(follicular structures)
Lymphoma B cell receptor
Ig light chain (k)
T Cells
(infiltrating tumor)
Tumor-infiltrating T cells (CD3)
10X
Lymphoma B cell receptor
Ig light chain (k)
BCL2
CD20

3. Follicular Lymphoma (FL)
Clonally rearranged immunoglobulin
Characterized by t(14;18)(q32;q21) translocation
Incurable using conventional therapy
Good candidate for molecularly-targeted therapies
Frequent mutation of MLL2 histone methyltransferase
Recurrent mutation of CREBBP histone acetyltransferase
Genetic “constants”
Adapted from WHO 2008
Solal-Celigny et al. Blood 2004;104:1258

4. Histone Modification by MLL2 and CREBBP
CREBBP/EP300
inactivating mutation
MLL2/3
inactivating mutation
HAT
MLL
ING H3 H4
K27 K4

5. Targeted Therapy: Hitting the Achilles Heel of Cancer
7 July 2012
8 July 2012
9 July 2012

6. The Theory of “Personalized Oncology”
Roychowdhury et al. Sci. Transl. Med. 2011;3:111
MacConaill and Garraway J. Clin. Oncol. 2010;28:5219

7. The Reality of “Personalized Oncology”
Mutation 1
Mutation 2
Mutation 3
Catalogue of Mutations
Peter C. Nowell (1976)
Science.194(4260):23-8.
Mutation 1
Mutation 2
Mutation 3
DRUG
RELAPSE

8. The Reality of “Personalized Oncology”
Mutation 1
Mutation 2
Mutation 3
Catalogue of Mutations
Mutation 1
Mutation 2
Mutation 3
DRUG
RELAPSE

9. Premise, Aim and Approach
Premise: Early genetic events are likely to be clonally dominant and represent good targets for mutation-directed therapy
Aim: To identify the hierarchy of genetic events in FL
Approach: Identify clonally dominant mutations
Consistently represented between intratumoral subpopulations
Maintained from diagnosis to relapse

10. Experimental approach
FACS
T-cells
CD20int
CD20hi
DNA Extraction
Whole Exome
Sequencing
t(14;18) qPCR Tumor Purity Measurement
IgHV cloning/sequencing
Sanger Validation
Genetic “Constants”

11. Exome Sequencing Methodology
Constructed libraries from 3ug of DNA
Captured exome with with Nimblegen SeqCap (v2)
Indexed with Illumina barcodes
4-plexed samples on a single HiSeq 2000 lane (2x101bp)

12. Mutation Calling GATK score of ≥250 in B-cells
Called somatic nucleotide variants (SNVs) with stringent implementation GATK:
GATK score of ≥250 in B-cells
GATK score of <50 in T-cells
Filtered silent mutations and those in dbSNP/1000genomes
Only considered cSNVs with:
≥20X coverage in both T-cells and B-cells
<5% variant allele frequency (VAF) in T-cells
≥5% VAF in B-cells
 96% validation rate

13. Exome Sequencing and Mutation Detection
In 10 tumors from 8 cases, identified 877 coding SNVs in 572 unique genes
95% of genes mutated in only 1/8 cases
Previously, populations frequencies were only way of assessing

14. Majority of Mutations in FL are Subclonal

15. Assessing sub-population skew
Interrogated minor allele frequencies of 232 germ-line coding SNPs/patient (1856 total)
Some noise around VAFs of heterozygous SNPs
By definition, variation in germline SNPs are false-positives*
Set thresholds to obtain confident calls
At 16% VAF deviation, 85 false-positives
4.58% error
At 33% VAF deviation, 18 false-positives
0.97% error
*Possibility of LOH  over-estimating error

16. Mutation Frequencies in Tumor Subpopulations

17. More mutations, more clonal divergence

18. Illustrative Case of Diagnosis/Relapse Comparison
A 40 year old woman with enlarged lymph nodes and fevers found to have advanced follicular lymphoma
Diagnosis (1996)
Histology: FL grade 1
Stage: 4B
Time to first treatment = 362 days
First treatment = CVP (1997) achieved Complete Remission (CR)
Second treatment = Id-vac (1998)
Relapse (1999)
Treatment: Fludarabine + Cyclophosphamide, CR
Second relapse in 2003, treated
Patient alive as of Feb 2013

19. Interrogation of FL Relapses: Case 128
Previously, populations frequencies were only way of assessing

20. Genetic Evolution of Case 128

21. Conclusions
The majority of mutations in FL are not recurrent and are subclonal.
MLL2 and TNFRSF14
Skewed distribution in tumor cell subpopulations
Lost between diagnosis and relapse in LP-J128
CREBBP
Equally represented in tumor cell subpopulations
Maintained between diagnosis and relapse
Subclonal =
Late event
Previously, populations frequencies were only way of assessing
Clonally dominant =
Early event

22. Conclusions

23. Acknowledgments Prof. Ron Levy Prof. Ash Alizadeh Prof. Jonathan Irish
Dr. June Myklebust
Dr. Itai Kela
Prof. Ash Alizadeh
Shingo Kihira
Dr. Chih Long Liu
Prof. Sylvia Plevritis
Dr. Andrew Gentles
Dr. Ramesh Nair
Prof. Hanlee Ji
Dr. Eric Hopmans