Measurable Residual Disease Detection of IDH1 or IDH2 Mutations Post but Not Prior to Allogeneic Hematopoietic Stem Cell Transplantation Show High Prognostic Relevance in Patients with Acute Myeloid Leukemia

Introduction: Even after allogeneic hematopoietic stem cell transplantation (HSCT) relapse remains a main driver of mortality in acute myeloid leukemia (AML) patients (pts). In overt relapse, treatment options are limited and survival is significantly shortened. Early detection of impeding relapse by monitoring measurable residual disease (MRD) could improve outcomes by identifying pts that would benefit from preemptive treatment. Most studies analyzing molecular MRD used mutation (mut) panels and did not differentiate between different genes. However, distinct genes may provide distinct prognostic relevance, as shown for ASXL1, DNMT3A, or TET2 (“DTA”) mut, which do not function as MRD markers after chemotherapy. Mut in IDH1 and IDH2 occur in up to 20% of AML pts, and cluster at three hotspots, making them potential easy targets for sensitive PCR-based MRD detection. So far, no larger study analyzed their relevance for MRD detection after HSCT.

Methods: We screened 462 AML pts for IDH mut by NGS (MiSeq platform, Illumina) or Sanger sequencing and detected IDH mut in 100 pts (IDH1 R132: n=44; IDH2 R140: n=37; IDH2 R172: n=18; IDH1 R132 & IDH2 R140: n=1). At HSCT, IDH mut pts were in first complete remission with or without count recovery (CRc, 70%), second CRc (15%), or relapsed/refractory (15%). Median age at HSCT was 64 (range 20-76) years (y). Conditioning regimens were myeloablative (10%), reduced-intensity (30%) or non-myeloablative (60%). MRD analyses were performed by custom designed mut specific digital droplet PCR probe assays as published previously. IDH MRD positivity (MRD^pos) was defined as VAF ≥ 0.05%. Median follow up after HSCT was 3.6 y.

Results: Compared to all others, presence of IDH1 or IDH2 mut at diagnosis did not significantly associate with outcomes (cumulative incidence of relapse [CIR] P=.13, overall survival [OS] P=.40). In CRc prior to HSCT (n=66), 59% of IDH1 mut, 81% of IDH2 R140 mut and 38% of IDH2 R172 mut pts remained IDH MRD^pos, which did not associate with CIR (P=.17) or OS (P=.80). In CRc after HSCT, the IDH MRD status was available for 57 pts (IDH1 R132: n=25; IDH2 R140: n=26; IDH2 R172: n=7) with a median of 4 samples/pt (range 1-21). 34% of IDH mut pts relapsed during follow-up, and all pts with available relapse material (n=16) were positive for their known IDH mut (median VAF at relapse 18.7%, range 0.09-29%). In all relapsing pts with samples available, relapse was preceded by at least one MRD^pos sample (median VAF in first pos sample 0.23, range 0.07-19.9%). Median time from first MRD^pos sample to overt relapse was 51 (range 8-341) days (IDH1: median 40.5 [range 8-119] days, IDH2: median 91 [range 13-341] days).

Of the pts in ongoing CRc with post-HSCT samples available (n=40), 3 pts had 1 MRD^pos sample (IDH1: 2 pts; IDH2 R140: 1 pt) and 1 pt had two MRD^pos samples (IDH2 R172). In 2 pts these positive samples occurred directly after HSCT and converted to MRD negativity (MRD^neg), 1 pt died from an acute infection 96 day after testing MRD^pos and 1 pt converted to MRD^neg after suffering from grade 3 acute graft-versus-host disease (GvHD). ROC curves showed that the IDH1 and IDH2 MRD status after HSCT were highly predictive of relapse within the next 56 days (AUC_IDH1=.98 and AUC _IDH2=.93) and 84 days (AUC _IDH1=.91 and AUC _IDH2=.90). Of note, ROC comparison showed that IDH1 and IDH2 MRD was not inferior in predicting relapse than the NPM1 MRD status within 56 days (AUC_NPM1=.95, compared to IDH1: P=.19 and IDH2: P=.81) and 84 days (AUC_NPM1=.81, compared to IDH1: P=.27 and IDH2: P=.35) in NPM1 mutated pts.

Pts with at least one IDH MRD^pos sample within the first year after HSCT had a significantly higher CIR (P<.001) and shorter OS (P=.001). This was also seen when IDH1 and IDH2 mut were regarded separately (IDH1: CIR P<.001 and OS P=.005; IDH2: CIR P<.001 and OS P=.07).

Conclusion: Neither the IDH mut status at diagnosis nor IDH-based MRD in CRc prior to HSCT associated with outcomes. In contrast, after allogeneic HSCT, IDH1 and IDH2-based MRD reliably predicted relapse in AML pts and was not inferior to NPM1 mut based MRD. Our data indicates that all IDH mut only function as MRD marker after allogeneic HSCT, which resembles mut associated with clonal hematopoiesis, like DTA. This should be taken into account when assessing MRD in clinical practice. The conversion of MRD^pos to MRD^neg in pts early after HSCT or after suffering GvHD implies that immunological effects can eradicate MRD after HSCT.