Comparative Assessment Of Minimal Residual Disease In Acute Myeloid Leukemia By Flow Cytometry and Molecular Genetics

Minimal residual disease (MRD) is quantified in patients (pts) with acute myeloid leukemia (AML) by both flow cytometry (FC) and molecular genetics (MG). Advantages und disadvantages for the respective approaches are under debate.

To assess the respective impact of FC and MG for MRD quantification in AML.

420 bone marrow (BM) samples from 252 pts with AML undergoing intensive first-line therapy were analysed by FC and MG in parallel for MRD. FC quantified cells displaying aberrant immunophenotypes identified at diagnosis. Targets for MG included RUNX1-RUNXT1, CBFB-MYH11, EVI1 expression, MLL fusion genes, MLL-PTD, NPM1 mutation, NUP98-NSD1, and PML-RARA. MRD was expressed as log difference of disease level between diagnosis and MRD assessment (LDF for FC and LDM for MG). Four periods were analyzed separately: ≤day 60 after start of therapy, day 61-120, day 121-365, beyond 1 year. Non-MRD factors univariately related to EFS were age (p<0.001, HR 1.25 per decade), WBC count (p=0.010, HR 1.04 per 10 G/L), BM blasts by cytomorphology (p=0.017, HR 1.07 per 10%), and cytogenetics according to MRC [Grimwade et al. Blood 2010] (CG; p<0.001, HR 3.40 per step). Parameters related to OS were age (p=0.002, HR 1.27 per decade), WBC count (p=0.002, HR 1.06 per 10 G/L), BM blasts (p=0.023, HR 1.09 per 10%), and CG (p=0.001, HR 2.34 per step).

LDF and LDM did not differ significantly for the first period (mean 2.94 vs 2.62), but for the other three periods LDF was significantly lower than LDM (3.04 vs 3.90, 3.20 vs 4.20, 3.05 vs 4.06; p<0.001 for each).

In the first period both LDF (Cox, p=0.045, HR 0.80) and LDM (p=0.030, HR 0.84) significantly correlated with EFS. In a multivariate model LDM was the only independent parameter (p=0.028, HR 0.80). EFS in pts with LDM below the 25%ile (1.34) was significantly shorter than in the others (median 8.6 vs 16.3 months, p=0.003).

In the second period LDM was significantly related to EFS (p=0.001, HR 0.74) while LDF was not. In multivariate analysis, LDM was the most important independent factor (p<0.001, HR 0.73) besides CG (p=0.017, HR 2.94) and age (p=0.030, HR 1.28). EFS in pts with LDM below the median (4.04) was significantly shorter than in the others (median 10.3 vs 26.5 months, p<0.001).

In the third period both LDF (p=0.022, HR 0.60) and LDM (p<0.001, HR 0.64) were significantly related to EFS. Multivariate analysis again confirmed LDM as strongest parameter (p<0.001, HR 0.71) followed by CG (p<0.001, HR 3.48) and BM blasts at diagnosis (p=0.038, HR 1.10 per 10%). Separation of pts according to the respective median values resulted in significant differences in EFS, however, separation was more powerful by median LDM (4.57; 3-year-EFS 29% vs 67%, p<0.001) than by median LDF (3.27; 3-year-EFS 38% vs 59%, p=0.046).

In the last period again both LDF (p=0.029, HR 0.66) and LDM (p<0.001, HR 0.60) were significantly related to EFS with only LDM being confirmed independent (p<0.001, HR 0.48) besides WBC count at diagnosis (p=0.005, HR 1.23 per 10 G/L). Separation of pts according to median LDM (4.51) again resulted in significant differences in EFS (3-year-EFS 42% vs 94%, p<0.001).

For all four periods LDM was significantly related to OS (p=0.024, HR 0.79; p<0.001, HR 0.63; p=0.001, HR 0.69; p=0.020, HR 0.68; respectively) while LDF was not. Most importantly, all multivariate analyses of OS for the four periods revealed the respective LDM as the only independent parameter (p=0.012, HR 0.67; p<0.001, HR 0.65; p=0.024, HR 0.74; p=0.014, HR 0.57), except for the last period for which also WBC was independent (p=0.018, HR 1.23). Separation of pts according to the median LDM for the second period resulted in highly significant differences in OS (3-year-OS 32% vs. 75%, p<0.001).

In 34 pts increases of MRD during repeated analyses until relapse were compared between FC and MG. Interestingly, mean increases were similar amounting to 0.276 and 0.281 logs per month, respectively. This was also true when considering disease subgroups with different relapse kinetics like those with NPM1 mutations (FC vs MG, 0.358 vs 0.362) and those with MLL-PTD (0.193 vs 0.206). However, the dynamic range of the lowest MRD level and the level at relapse clearly was higher for MG as compared to FC (4.12 vs 2.69, p<0.01).

Molecular genetics provides the more powerful tool to quantify MRD in AML as compared to flow cytometry mostly due to its higher sensitivity and its broader dynamic range.

Kern:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Rose:MLL Munich Leukemia Laboratory: Employment. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Haferlach:MLL Munich Leukemia Laboratory: Employment, Equity Ownership. Schnittger:MLL Munich Leukemia Laboratory: Employment, Equity Ownership.