Detection of KMT2A fusion transcripts after induction chemotherapy is strongly predictive of outcome in KMT2A rearranged AML: Results of an international collaborative study Free

KMT2A rearrangements (KMT2Ar) occur in ~5% of adult AML, making adequately powered studies of prognostic factors challenging. We and others have previously shown that detection of KMT2A fusion transcripts after induction is strongly prognostic, however we were unable to examine the impact of MRD within subgroups, or the interaction of MRD with allogeneic transplant in 1^st complete remission (CR1 allo). Moreover, it is unclear if the prognostic impact of MRD is dependent on therapeutic protocol and whether it is reproducible between laboratories. To address these issues, we compiled an international cohort of patients from institutions performing molecular MRD assessment for KMT2A fusion transcripts.

Methods Data were collected from contributing centres in the United Kingdom, Taiwan, Australia, Lithuania and Denmark. Patients treated in the UK NCRI AML17 and AML19 protocols were included, as well as those treated off protocol. Inclusion criteria were previously untreated KMT2Ar AML, receipt of intensive chemotherapy without frontline menin inhibitor, achievement of CR/CRi within 2 courses, and availability of molecular MRD results by RT-qPCR or RT-ddPCR. MRD assessments were performed at one of 6 independent laboratories, using assays with sensitivity of at least 1:10⁴.

Results 151 patients were identified, including 67 enrolled in AML17/19 and 84 treated off study. Median age was 42 (range 12-70) with 16 (11%) over 60 years. Twenty (13%) had therapy-related disease. The fusion gene was t(9;11)/MLLT3 in 61 (40%) with the remaining having ELN adverse disease including 36 (24%) with t(6;11)/AFDN, 25 (17%) t(11;19)/ELL, 4 (3%) t(11;19)/MLLT1, 20 (13%) t(10;11)/MLLT10 and 5 (3%) with other fusions. NGS results were available in 107 of whom 19 (18%) had NRAS and 19 (18%) had KRAS variants.

Induction therapy was with DA or 7+3 in 97 (64%), FLAG-Ida in 43 (28%), CPX-351 in 7 (5%) and other HDAC-induction in 4 (3%). 32 (21%) patients received a single and 119 (79%) a double induction protocol. GO was administered in 33 (22%), venetoclax in 5 (3%) and a FLT3 inhibitor in 6 (4%). With a median follow-up of 3.0 years (y), median OS was 5.6y (95CI 2.7-NR). OS at 3y and 5y was 55% and 52%. 3y cumulative incidence of relapse (CIR) was 47% with 10 patients known to have received a menin inhibitor at relapse.

MRD was assessed after course 1 in 121 patients (115 bone marrow, BM, 48 peripheral blood, PB) and after course 2 (PC2) in 127 (120 BM, 54 PB). Both time points were informative, but the most powerful prognostic information was at PC2 where any detection of MRD in either the PB or BM (35% of patients) was associated with increased relapse (3y CIR 83% vs 27%, HR 5.9, 95CI 3.4-10.1) and poorer survival (3y OS 22% vs 71%, HR 4.4, 95CI 2.4-7.9). In multivariable analyses incorporating baseline factors and chemotherapy regimen, PC2 MRD was the variable most significantly associated with CIR and OS.

The prognostic importance of post C2 MRD was robust across subgroup analyses, with consistent results in all 5 laboratories (heterogeneity p-value=0.75), trial and real-world settings (p=0.46), induction regimens (p=0.38) and for both single and double induction strategies (p=0.51).

CR1 allo was performed in 84 patients (55%). More patients with ELN adverse disease received CR1 allo (60% vs 47%). Using time-dependent regression, CR1 allo was associated with a trend to improved OS (HR 0.7, 95CI 0.4-1.2) and RFS (HR 0.7, 95CI 0.4-1.1) overall. Subgroup analyses suggested that PC2 MRD positive patients were most likely to benefit from CR1 allo (landmarked 3y OS 38% with CR1 allo, 11% without, HR 0.5, 95CI 0.2-1.2) compared to MRD negative patients (3y OS 71% with CR1 allo, 68% without, HR 1.7, 95CI 0.6-4.7). This trend was consistent when stratifying by ELN risk groups and by induction chemotherapy. Patients who were MRD negative after FLAG-Ida had 3y OS of 84% and 100% with and without CR1 allo, while those MRD negative after other regimens had 3y OS 61% and 66% with and without CR1 allo.

The study continues to accrue patients with updated results to be presented at the meeting.

Conclusion In this international collaborative analysis of patients in remission after intensive chemotherapy, we demonstrate the major prognostic importance of tracking KMT2A fusion transcript expression to measure MRD. The method is applicable internationally across treatment protocols and laboratories and may identify patients with the greatest benefit from CR1 allo.