Advances in B Lymphblastic Leukemia MRD. Brent Wood MD PhD Departments of Laboratory Medicine and Pathology University of Washington.

Transcription

1 Advances in B Lymphblastic Leukemia MRD Brent Wood MD PhD Departments of Laboratory Medicine and Pathology University of Washington

2 Measures of Response Clinical outcome OS, EFS, RFS, etc. Blast count Remission = < 5% by morphology Minimal Residual Disease Cytogenetics, Flow, Molecular

3 Pediatric AML Blast count does not correlate with MRD Inaba, et al (2012) JCO 30:

4 Pediatric AML Blasts < 5% MRD- MRD+ Blasts 5% Morphologic blast count is not predictive Inaba, et al (2012) JCO 30:

5 AML Pre-Transplant MRD- Blasts < 5% Pre-Tx Blasts > 5% = MRD+ Blasts < 5% Araki, et al (2015) JCO (in press)

6 ALL Morphology vs. Flow 0.8% M1 but Flow > 5% 6.5% M1 but Flow > 5% Gupta, et al (2016) ASH 2016

7 ALL Morphology vs. Flow B-ALL T-ALL Gupta, et al (2016) ASH 2016

8 Need a new definition of remission Morphologic blast counts correlate poorly with response in AML Shouldn t use in clinical trial design Assess residual disease by other methods

9 Reproducibility

10 Sources of Variability Identification (false positive or negative) Insufficiently informative reagents Improper assay validation Immunophenotypic shift Inexperienced interpreters Quantitation Too few events acquired Denominator effects (2 fold) Sample degeneration Hemodilution

11 Borowitz, Wood, et al (2015) Blood 126(8):964-71

12 % of cases MRD >.01% Flow MRD on AALL03B d29 d15* d8 blood JHU(n=2282) UW(n=1947) * day 8 M1 patients excluded Unpublished data, courtesy Mike Borowitz

13 Day 29 Flow MRD on AALL0232 FIGURE S3 COMPARISON OF MRD RATES BETWEEN THE EASTERN AND WESTERN REFERENCE LABORATORIES % 0.1-1% >1% overall MRD level % of cases JHU (n=1204) UW (n=1228) Figure S3. Comparison of MRD Borowitz, rates between Wood, the et al two (2015) reference Blood laboratories. 126(8): The

14 Correlation between labs Borowitz, Wood, et al (2015) Blood 126(8):964-71

15 CCR probability Correlation between labs AALL % <= Day 29 MRD < 1.0% Eastern Region (n = 121) Western Region (n = 114) Years P = Borowitz, Wood, et al (2015) Blood 126(8):964-71

16 67% 67% 67% % Attempted 67% % Attempted 5 5 # Positive # Positive # Positive # Negative # Negative # Negative False POS False POS False POS False NEG False NEG False NEG % 87% 60% 60% 87% 80% 90% 60% 87% 90%(%) Concordance Overall 60% 90% 60% 90% 90% Log (log) % 90% 60% 90% 90% 60% 87% Interpretive Variability 26 60% 87% 70% 73% 53% 60% 80 70% Outside 0.5log 80Concordance 90%(%) Overall Concordance (%) Outside Log (log) 0.5log Concordance (%) Last 2 rounds.tc s. re First 3 rounds 90% 53% s. in Overall 87% Concordance (%) Outside Log (log) 0.5log 87% 73% Concordance (%) 87% FIGURE 2B: Dist the results from the rounds of sendouts reference laborat review of the first and educational se MRD analysis. w FIGURE 2A: Distribution of FIGURE 2A: Distribution of the 3results from the first 3 the results from the first of sendouts from rounds of sendouts rounds from the reference laboratorythe reference laboratory w %) Experience with MRD Testing in B- ALL By Flow Cytometry Does Not Prevent Interpretative Discordance w %) % Attempted Keeney, Wood et al. (2017) Cytometry B

17 Reproducibility Flow cytometry is capable of reproducible MRD detection and enumeration Lack of standardization in implementation is the source of variability in current practice

18 Opportunities

19 Anti-CD19 Immunotherapy Targets primary gating reagent Identifying B cells difficult Eliminates expression of CD19 Rapidly after administration Selects for CD19 (-) subset of leukemia CD19 (-) MRD and relapse Need alternate strategy CD22, CD24 + CD66b, ccd79a Now routine assay

20 CD24 APC CD10 BV421 CD38 A594 CD24 APC CD22 PE CD19 PE-Cy7 Anti-CD19 Immunotherapy Viable Viable CD66b FITC CD66b FITC CD10 PE B cells B cells B cells CD22 PE CD20 PE-Cy7 CD10 BV421 CD19-negative MRD with background hematogones B Cells HP Cherian, et al (2016) Clin Cyto Part B

21 MRD by Flow Cytometry Advantages Fast Relatively inexpensive Large instrument base Reproducible Disadvantages Subjective interpretation Immunophenotypic drift after therapy Moderate sensitivity Poorly standardized

22 V IgH and TCR Diversity 54 2 V D J V D RSS RSS RSS RSS N D N J 13 J

23 MRD by NGS Faham et al (2012) Blood 120:

24 Day 0 Cancer clone Frequency Day 29 Cancer clone Frequency PT_44 PT_63 PT_48 PT_9 PT_7 PT_15 PT_59 PT_36 PT_33 PT_53 PT_52 PT_5 PT_50 PT_1 PT_32 PT_8 PT_30 PT_49 PT_66 PT_39 PT_65 PT_41 PT_61 PT_51 PT_6 PT_47 PT_4 PT_58 PT_2 PT_40 PT_14 PT_55 PT_57 PT_3 PT_12 PT_54 PT_45 PT_60 PT_68 PT_56 PT_13 PT_42 PT_11 PT_44 PT_63 PT_48 PT_9 PT_7 PT_15 PT_59 PT_36 PT_33 PT_53 PT_52 PT_5 PT_50 PT_1 PT_32 PT_8 PT_30 PT_49 PT_66 PT_39 PT_65 PT_41 PT_61 PT_51 PT_6 PT_47 PT_4 PT_58 PT_2 PT_40 PT_14 PT_55 PT_57 PT_3 PT_12 PT_54 PT_45 PT_60 PT_68 PT_56 PT_13 PT_42 PT_11 Individual ETP netp Flow Cytometry TCRBSEQ Wu, et al (2012) Sci Trans Med 4: 1-7

25 93 / 98 (95%) with IgH rearrangement Remainder: 2 with D-J, 1 with clonal D-J + many VH, 2 none Wu, et al (2014) Clin Cancer Res 20:4540-8

26 MRD AND TRANSPLANT Flow vs NGS Pulsipher et al Blood 2014;123:2017 Pulsipher et al Blood 2015;125:3501 Better definition of low risk through higher sensitivity

27 NGS A SWOG 0333 C" D B" Flow Sala-Torra et al (2017) In Press

28 B-LL End of Induction MRD Flow and HTS are equivalent at 0.01% Use of cutoff < 0.01% does not improve risk stratification

29 B-LL End of Induction MRD Standard Risk High Risk Lower MRD more informative for high risk

30 B-LL End of Induction MRD Absence of detectable MRD by NGS = Excellent outcome Standard risk patients

31 B-LL End of Induction MRD Absence of detectable MRD by NGS for High risk patients Standard risk patients

32 3.25 All patients 7.2 Standard risk High risk 10E-7 Cut Point Analysis Hazard Ratios 10E-3 p-values MRD 10E-7 10E-3 MRD Optimal cut point for combined patients is ~0.01% p = 0.05 p = 0.05

33 B-LL End of Induction MRD Discordant Flow and HTS has intermediate outcome Similar result for Flow+/HTS- (N=17) Discordant = Concordant MRD- at lower HTS cutoff of 0.001%

34 SmMIPS Minimal error rate for via single molecule tagging Targets multiple genomic regions of interest Simple, scalable protocol Modular and cost-effective target enrichment Low sample input requirements

35 Abnormal myeloid blasts (%) N = 353 AML MRD NPM1 by NGS NPM1 mutant VAF (%) Zhou Y, et al (2017) submitted

36 Relapse free survival (%) AML MRD NPM1 by NGS Non-transplant N = 37 MRD MFC POS MRD NGS HI MRD LO/NEG Time (months) N = 5 N = 7 N = 25 Zhou Y, et al (2017) submitted

37 x The Cancer Genome Atlas Research Network (2013) NEJM 368:

38 Gene Capture Design AML MRD NGS Nucleotides sequenced (bp) 32 gene panel of AML mutations Coverage for ~85% of AML Sensitivity of 0.011% Number smmip probes Gene Capture Design Nucleotides sequenced (bp) Number smmip probes ABL1 Hotspot NT5C2 Hotspot ALK Hotspot PHF6 Full Gene BRAF Hotspot PIK3CA Hotspot CEBPA Full Gene PPM1D Full Gene DNMT3A Hotspot PTPN11 Full Gene EZH2 Full Gene RAD21 Full Gene FAM5C Full Gene RET Hotspot FLT3 Hotspot ROS1 Hotspot HNRNPK Full Gene RUNX1 Full Gene IDH1 Hotspot SMC1A Full Gene IDH2 Full Gene SMC3 Full Gene JAK2 Hotspot STAG2 Full Gene KIT Hotspot TET2 Full Gene KRAS Hotspot TP53 Full Gene NPM1 Hotspot 89 1 U2AF1 Hotspot NRAS Hotspot WT1 Full Gene Cell line variants Hotspot Total N/A Salipante, et al (2017) Haematologica 102(9):

39 AML MRD NGS Good correlation between multigene assay and standard NPM1 Salipante, et al (2017) Haematologica 102(9):

40 NGS Conclusions NGS is capable of MRD detection Can reproduce risk groups Sensitivity = NGS (10-6 ) > ASO-PCR > Flow Uses Standardize testing Simplify sample requirements Define MRD(-) low-risk group early Define MRD(+) high-risk group late (EOC)? better for targeted immunotherapy TAT of > 1 week currently

41 Subclonal CRLF2 CRLF2 deletion P2RY8-CRLF2 Harvey et al. (2016) submitted

42 Molecular Cytometry The ability to measure multiple molecular features at the single cell level for entire populations of cells

43 Tumor Heterogeneity Ma, et al (2015) Nature Communications DOI: /ncomms7604

44

45

46 Conclusions Diagnosis is immunophenotypic, classification is genotypic MRD detection provides important unique prognostic information in acute leukemia Is standard of care for B-ALL MRD by flow cytometry can be done reproducibly MRD by NGS is feasible for B-ALL, T-ALL and AML Single cell molecular methods are the next frontier

47 Acknowledgements Hematopathology Laboratory at UWMC Michael Borowitz MD PhD Entire COG ALL subcommittee Steven Hunger, Bill Carroll, Mignon Loh Adaptive Biotechnologies Harlan Robins (FHCRC)