Measurable/Minimal Residual Disease in Patients with Acute Myeloid Leukemia Undergoing Transplantation

Measurable/Minimal Residual Disease in Patients with Acute Myeloid Leukemia Undergoing Transplantation June 9, 2018 Dave Sanford The Leukemia/Bone Marrow Transplant Program of British Columbia

Disclosures Advisory boards: Jazz, Novartis, Pfizer, Shire Honoraria: Jazz, Novartis, Pfizer, Shire Clinical trial involvement: AMGEN, Daiichi-Sankyo, KITE, Novartis

Overview MRD Methodology Risk stratification using MRD following induction chemotherapy MRD Pre-transplant MRD Post-transplant

Complete Response in AML <5% blasts by morphology, no Auer rods Platelets >100 x 10 9 /L, Neutrophils > 1 x10 9 /L, no circulating blasts No extramedullary disease Minimal/Measurable Residual Disease Detectable disease in a sample that otherwise shows a complete morphologic remission Cheson B, et al. J Clin Oncol. 2003;21(24):4642-9.

Value of Detecting a Deeper Remission in AML Objectively measure the depth of remission status Improve prognostication and inform post remission treatment Early detection of relapse with early intervention Use of MRD response as surrogate endpoint in clinical trials 1.Szczepański T, et al. Lancet Oncol. 2001;2(7):409-17. 2. Schuurhuis GJ, et al. Blood. 2018;131(12):1275-1291.

MRD Methodology

Qualities of an Ideal MRD Marker Sensitive Present in all leukemia cells Specific Absent in normal cells 1. Viny AD, et al. N Engl J Med. 2016;374(23):2282-4. 2. Hourigan CS, et al. Leukemia. 2017;31(7):1482-1490. 3. Selim A, et al. J Mol Diagn. 2018: S1525-1578(17)30485-3.

Qualities of an Ideal MRD Marker Sensitive Present in all leukemia cells Specific Absent in normal cells 1. Viny AD, et al. N Engl J Med. 2016;374(23):2282-4. 2. Hourigan CS, et al. Leukemia. 2017;31(7):1482-1490. 3. Selim A, et al. J Mol Diagn. 2018: S1525-1578(17)30485-3.

Qualities of an Ideal MRD Marker Sensitive Present in all leukemia cells Specific Absent in normal cells Stable Present at relapse (and remission) Stemness Present in population of cells capable of relapse 1. Viny AD, et al. N Engl J Med. 2016;374(23):2282-4. 2. Hourigan CS, et al. Leukemia. 2017;31(7):1482-1490. 3. Selim A, et al. J Mol Diagn. 2018: S1525-1578(17)30485-3.

Multi-Parameter Flow Cytometry (MFC) Limit of Detection ~ 1/1000 (10-3 ) -1/10,000 (10-4 ) abnormal cells Rapid turn-around Readily available technology Applicable to most cases of AML Limitations Shifts in immunophenotype Less sensitive than optimized RT-qPCR method Sensitivity differs between cases Requires Expertise

Leukemia Associated Immunophenotypes Diagnosis SSC CD45 CD34 CD33 Follow-up FSC SSC SSC HLA-DR Ossenkoppele GJ et al. Br J Haematol. 2011;153(4):421-36.

Different from Normal Normal Abnormal Jaso JM et al. Bone Marrow Transplant. 2014;49(9):1129-38.

Schuurhuis GJ, et al. Blood. 2018;131(12):1275-1291.

ELN Recommendations on Flow MRD Sampling BM over PB Methodology DfN and LAIP approach 8 color flow Same tubes for Dx and F/U (LAIP) Recommendation for panels Quantification Relate MRD to CD45+ leukocyte population excluding CD45- erythroblasts Thresholds Early threshold of 0.1% to distinguish MRD pos vs. MRD neg Recommend reporting out detectable disease <0.1% Suggested Reporting Template Schuurhuis GJ, et al. Blood. 2018;131(12):1275-1291.

RT-qPCR for MRD Limit of Detection ~ 1 in 100,000 (10-5 ) abnormal cells Potentially rapid. Batched to be economically viable Fusion gene transcripts CBF transcripts (RUNX1-RUNX1T1, CBFB-MYH11), MLL, t(6;9) - DEK-NUP214 Mutations NPM1 Abnormal expression WT1

RT-qPCR for MRD MRD assay for transcripts formed by translocations/inversions 10 2 AML1 (21q22) ETO (8q22) QPCR FWD PRIMER 3 4 5 2 3 TAQMAN PROBE QPCR REV PRIMER Standard Curve MT fusion transcript (i.e. RUNX1- RUNX1T1) formed by translocation or inversion Reference gene transcript (i.e. ABL1) Background (non-target) transcripts Cruz NM, et al. Int J Lab Hematol. 2017;39 Suppl 1:53-60. RUNX1-RUNX1T1 Copies

RT-qPCR in CBF AML RFS by qpcr at end-induction OS by qpcr at end-induction RFS by qpcr at end-consolidation 1.0 1.0 1.0 0.9 0.9 0.9 0.8 0.8 0.8 Relapse free proportion 0.7 0.6 0.5 0.4 0.3 0.2 At remission Total Relapses 5yr OS < 0.1% 43 4 90% 0.1% or above 54 17 66% p = 0.002 Surviving proportion 0.7 0.6 0.5 0.4 0.3 0.2 At remission Total Deaths 5yr-OS <0.1% 43 8 85% 0.1% or above 55 24 61% p = 0.01 Relapse free proportion 0.7 0.6 0.5 0.4 0.3 0.2 End-consolidation Total Relapses median <0.01% 59 9 NR 0.01% or above 28 12 36.5mo p = 0.001 0.1 0.1 0.1 0.0 0 12 24 36 48 60 72 84 96 108 120 132 Time (months) 0.0 0 12 24 36 48 60 72 84 96 108 120 Time (months) 0.0 0 12 24 36 48 60 72 84 96 108 Time (months) Response kinetics and factors predicting survival in core-binding factor leukemia. Leukemia. (in press)

Digital Droplet PCR Reactions occur nanoliter oil droplets Fluorescent oligonucleotide probes for quantification Most droplets have 0-1 targets Quantification with Poisson statistics Advantages: Potentially more sensitive than RT-qPCR Obviates need for multiple mutationspecific assays Cruz NM, et al. Int J Lab Hematol. 2017;39 Suppl 1:53-60.

Custom AS probe assay (ddpcr only) Two AS probe target ins863_864c/tnng (95% of MT-NPM1) AS probe is 100% specific (MT vs. WT) LOD = 10-4 MT-NPM1/ABL1 but potentially could improve with more wells and higher load Added WT-PRBLK to improve signal quality in ddpcr output Probe for other NPM1 exon 12 MT could be potentially added (easily multiplexed) 19

Example of ddpcr output (4 wells) 10% MT-NPM1/ABL1 1% MT-NPM1/ABL1 n=315 n=23,862 n=43 n=2892 MT-NPM1 MT-NPM1 n=994 n=54,667 n=1431 n=80,481 ABL1 ABL1 n=5 0.1% MT-NPM1/ABL1 0.01% MT-NPM1/ABL1 n=287 n=1 n=24 MT-NPM1 MT-NPM1 n=1314 n=78,641 n=1338 n=82,470 ABL1 ABL1 20

Next Generation Sequencing (NGS) Advantages Decreasing cost Genomic DNA or RNA Patient specific Multiple targets Sensitivity influenced by: DNA/RNA quality Target gene Sequencing Coverage Mutation of Interest Point mutation Indel Fusion Relapse OS Jongen-Lavrencic M, et al. NEJM. 2018; 378(13):1189-1199.

ELN Recommendations on Molecular MRD Define Response and Progression Recommend only testing for: CBF transcripts, NPM1 Recommend against following certain genes as single markers FLT3-ITD, FLT3-TKD, NRAS, KRAS, DNMT3A, ASXL1, IDH1, IDH2, MLL-PTD and expression levels of EVI1 or WT1 Positive samples repeat BM and PB testing RT-qPCR is platform of choice NGS and digital PCR future techniques requiring further validation Schuurhuis GJ, et al. Blood. 2018;131(12):1275-1291.

Risk stratification with MRD following induction/consolidation

Impact of AlloSCT by MRD status post-induction Retrospective analysis of HOVON- SAKK trials (N=547) Allo (n=282) Auto (n=105) Chemo (n=160) ~25% of pts had MRD >0.1% (MRD pos ) after induction Relapse @ 4 yrs 32% vs. 54%, p<0.001 OS @ 4 yrs 65% vs. 50%, p=0.002 Versluis J, et al. JCO Precision Oncology. 2017.

Impact of AlloSCT by MRD status post-induction Univariable analysis relapse reduced by AlloSCT in both MRD <0.1% and >0.1% No significant difference in RFS for either group No apparent improvement in OS with allosct in either group MRD neg : 71% (CT/Auto) vs. 60% (Allo) MRD pos : 53% (CT/Auto)) vs. 47% (Allo) MRD<0.1% MRD<0.1% MRD 0.1% MRD 0.1% Versluis J, et al. JCO Precision Oncology. 2017.

GVL effect similar for MRD pos and MRD neg MV analysis for relapse: ELN, FLT3, # cycles to CR, PRT, MRD MV analysis: reduction in relapse similar with allosct in MRD neg and MRD pos HR MRD neg 0.38, p<0.001 HR MRD pos 0.35, p<0.001 Versluis J, et al. JCO Precision Oncology. 2017.

Impact of AlloSCT by NPM1 mut MRD post-induction Subset analysis of ALFA-0702 trial 229 pts with NPM1 mutations 152 had MRD data in first CR NPM1 mut measured by RT-qPCR in PB/BM samples post induction Balsat M, et al. J Clin Oncol. 2017;35(2):185-193.

Impact of AlloSCT by NPM1 mut MRD post-induction 44 pts underwent AlloSCT in CR1 Benefit of AlloSCT in pts with <4 log reduction DFS HR, 0.25 [95% CI, 0.06 to 0.98], P =.047 OS HR, 0.25 [95% CI, 0.06 to 0.98], P =.047 Balsat M, et al. J Clin Oncol. 2017;35(2):185-193.

AlloSCT for MRD in AML with t(8;21)? 3 log reduction Zhu HH, et al. MRD-directed risk stratification treatment may improve outcomes of t(8;21) AML in the first complete remission: results from the AML05 multicenter trial. Blood. 2013;121(20):4056-62.

AlloSCT for MRD in AML with t(8;21)? CIR DFS OS CIR DFS OS Zhu HH, et al. MRD-directed risk stratification treatment may improve outcomes of t(8;21) AML in the first complete remission: results from the AML05 multicenter trial. Blood. 2013;121(20):4056-62.

Thoughts on early risk-stratification MRD independent prognostic factor in multiple studies suggesting this could be used to allocate to SCT (along with other risk factors) Some evidence that transplant beneficial for high-risk patients based on MRD testing Optimal early MRD threshold, interaction of MRD with other risk factors and magnitude of benefit with MRD guided approach at early time points require further study

Detecting MRD Pre-Transplant

Prognostic Impact of detecting MRD Pre-Transplant Retrospective analysis of 359 consecutive patients at FHCC with AML undergoing MA HSCT in CR1 or CR2 Morphologic remission MRD ve: n=235 (76%) Morphologic remission MRD +ve: n=76 (24%) <0.1%: n=17 0.1% - 1%: n=24 >1%: n=35 Active disease ( 5% blasts): n=48 Twice as many pts with adverse cytogenetics and secondary AML in MRD +ve group Araki D et al. JCO. 2016;34(4):329-36.

MRD Positivity Equivalent to Active Disease in FHCC series Relapse at 3 years MRD (-ve) 22% MRD (+ve) 67% Active disease 65% No difference in outcome by level of MRD Pre-transplant MRD and active disease only predictor of OS in MV analysis HR 3.7 (2.5-5.2, p<0.001) OS Araki D et al. J Clin Oncol. 2016;34(4):329-36.

Not all MRD pre-transplant may be equally bad Study from Fred Hutchinson Cancer Centre (n=59) AML patients with mutated NPM1 who received HSCT Compared MFS to NGS MRD neg ~46% MRD MFC ~20% (All tested +ve by NGS) MRD NGS ~34% (Not +ve by MFC) Zhou Y et al. Biol Blood Marrow Transplant. 2018 [epub]

Not all MRD pre-transplant may be equally bad Retrospective analysis of patients on EORTC/GIMEMA protocols MFC at the end of consolidation Compared outcomes of MRD pos (>0.035%) pts (n=81) who received allosct or autosct 5 yr OS 59 (allo) vs. 27 % (auto) 5 Yr DFS 59 (allo) vs 19% (auto) Disease Free Survival Buccisano F, et al. Bone Marrow Transplant. 2017;52(3):473-475.

Detecting MRD post transplant

Detecting MRD by flow after HSCT Reference Patients Timing (Day post HSCT) Threshold Finding 1. Diez-Campelo (2009) MDS/AML (n=41) +21, +56, +100 </ 10-3 Day +100: 4 year EFS 74 vs. 17% 4 year OS 73 vs. 25% 2. Rossi (2015) AML (n=42) +30, +90, +180 </ 0.05% Day +30 - DFS: PPV (>0.05%): 46% NPV (<0.05%): 94% 3. Bastos-Oreiro (2014) AML (n=28) +30, +90, +180 </ 0.1% MRD post HSCT not predictive of DFS or OS 4. Zhao (2013) AML (n=308) +30, +60, +90, +130, +180, +270 Any level Relapse Rate: 1 +ve test: 14% vs. 82% 2 successive +ve tests: 14% vs. 100% 1. Díez-Campelo M et al. Am J Hematol. 2009;84(3):149-52. 2. Rossi G et al. Leuk Res. 2015;39(2):138-43. 3. Bastos-Oreiro M et al. Eur J Haematol. 2014;93(3):239-46. 4. Zhao XS et al. Ann Hematol. 2013;92(8):1111-9.

Detecting MRD by RT-qPCR after allosct WT1 Multiple studies Test characteristics improved by using 2 nd MRD marker Allows for monitoring in PB Core binding Factor AML Serial measurement of RUNX1/RUNX1T1 reported by Chinese centres (n=92) Failure to achieve >3 log reduction with 3 months post allosct associated with higher relapse incidence (~50%) NPM1 mut NPM1 mut /ABL1 > 10% higher risk of relapse post allosct and worse OS 4 year Relapse: 80% vs. 18% (p<0.001) OS: 20% vs. 80% (p<0.001) Median time to relapse 68 days 1. Rautenberg C, et al. Biol Blood Marrow Transplant. 2018. 2. Wang Y, et al. Blood. 2014;124(12):1880-6. 3. Shayegi N, et al. Blood. 2013;122(1):83-92.

Concerns around MRD monitoring post HSCT Labor-intensive Costly Increased anxiety for patients and care-givers Evidence for intervention to prevent relapse and improve outcomes?

Withdrawal of IST and DLI Supportive evidence is based on studies monitoring chimerism Evidence that DLI decreases mixed chimerism, which is associated with lower incidence of relapse

Withdrawal of IST and DLI Largest study of DLI based on MRD reported by Peking University People s Hospital 814 patients with standard risk acute leukemia, high-risk MDS undergoing allosct in CR1 or CR2 MRD by FCM (LAIP) and RT-qPCR for WT1 FCM MRD +ve >0.001%, WT1 MRD +ve >0.6% 3 Groups: (A) MRD-ve (n=709) no treatment (B) MRD +ve (n=49) low-dose IL-2 (C) MRD +ve (n=56) DLI +/- IL2 Yan CH, et al. Blood. 2012;119(14):3256-62.

Withdrawal of IST and DLI Relapse by MRD status Relapse by treatment assignment Yan CH, et al. Blood. 2012;119(14):3256-62.

Hypomethylating agents for MRD RELAZA Trial AML/MDS (n=205) following HSCT (n=147) or chemo (n=58) CD34+ chimerism (n=108) or RT-qPCR for NPM1 mut (n=77) or fusion transcripts DEK-NUP214 (n=1), RUNX1- RUNX1T1 (n=9), CBFB1-MYH11(n=10) Intervention Azacitidine (6 cycles) + withdrawal of IST Outcome 53 pts entered treatment phase Chimerism (n=19), NPM1 mut (n=31), t(8;21) (n=3) 6 Months Response 31/53 patients were still in CR (58%, 95%-CI 44-72%) 42% of pts relapsed after median of 3 cycles of Aza 40% had an MRD or chimerism response 18% had stable MRD or chimerism but no relapse Response rate higher after HSCT (71% vs. 48%) 24/31 pts continued on Aza 8 relapsed at later time-point 1 Year OS and PFS was 76 and 42% 1. Platzbecker U, et al. Leukemia. 2012;26(3):381-9. 2. Platzbecker U, et al. ASH Annual Meeting. Blood. 2017. 130:565.

Novel Approaches to MRD post-transplant Sorafenib Pre-clinical evidence of enhanced GvL in post-transplant setting Non-randomized studies/series report low relapse rates using sorafenib maintenance in FLT-mutated pts Anti-CD123 SL-401 - recombinant IL-3 conjugated to diphtheria toxin Ongoing phase ½ trial of SL-401 in AML in remission post therapy with MRD IDH1/2 Inhibitors Bispecific T-cell Engaging Antibodies 1. Brunner AM, et al. Br J Haematol. 2016;175(3):496-504. 2. Lane AA, et al. ASH Annual Abstracts. Blood 2017;130:2583.

Conclusions MRD appears to likely be useful tool to guide prognostication in transplant decisions understanding the optimal way to incorporate MRD into decision making around transplant is lagging GVL effect may be present regardless of MRD level Pre-transplant MRD may be an adverse risk factor transplant may still be beneficial Post-transplant MRD is an adverse risk factor May be an optimal area to explore novel agents

Opportunities and Challenges in Canada Multiple new agents for the treatment of AML MRD, NGS panels, comprehensive risk scores are improving and complicating prognostication May be benefit to harmonizing MRD testing and incorporation of MRD results into clinical decision making across Canada

Acknowledgements Aly Karsan Donna Hogge VCH Foundation Galloway Leukemia Seed Funding Curtis Hughesman Raewyn Broady Stephen Nantel Kelly McNeil Elaine Law Corrinne Sampson

Questions