Abstract

Patients with acute myeloblastic leukemia (AML) are divided according to the French American British (FAB) classification into eight subgroups (M0 to M7) on the basis of their degree of maturation/differentiation. Although immunophenotypical analysis is routinely performed to distinguish myeloblastic and lymphoblastic leukemia, these results are not contributive to the diagnosis of FAB subtype in patients with AML. Here we develop an algorithm based on immunophenotypic markers in order to assist in the diagnosis of FAB subtype. The median age in this patient population was 58 years with 10 patients less than 15 years old and 61 patients over the age of 60. There were 85 males and 54 females. The FAB subtypes most represented were M5 (42 patients, 30%) and M2 (25 patients, 18%), while 11 patients were not classifiable according to FAB criteria. The following markers were analysed in all samples: CD38, CD34, CD117 and HLA-DR (immature cell markers); CD13, CD15, CD33, CD36, CD41, CD61, CD65, CD11c, CD11b, CD13cy, CD14, MPOc (myeloid lineage); cytoplasmic CD3 (cyCD3), cytoplasmic CD22 (cyCD22), cytoplasmic CD79a (cyCD79a), nuclear TdT, CD2, CD4, CD5, CD7, CD9, CD10, CD19, CD56 (lymphoid lineage). Expression of each marker was characterized by the percentage of cells expressing this marker with the fluorescence level exceeding a background threshold, determined with an isotype control. Markers of surface antigens were routinely considered positive when greater than 20% while the cutoff limit for a positive intracellular marker was 10%. All samples except one of the FAB unclassified samples expressed at least two of the “myeloid specific markers” (CD13, CD33, CD65, CD117, MPO) and 124 samples (89%) expressed three or more of these markers. The primary distinction was made between CD36 negative samples (< 20% expression) which contains 89 patients (64%) and CD36 positive patients (≥20%) which contains 50 patients (36%). Used as a single marker, CD36 is significantly different between M5 subtypes and other subtypes with 2/3 of M5 cases positive for CD36 vs. 23% of positive cases in other subtypes (p<0.0001). However CD36 positivity is poorly predictive of the M5 subtype since only 56% of all CD36 positive samples correspond to the M5 subtype. The predictive power of CD36 is enhanced when taking into account cases with more than 50% of positive cells: in this case 78% of the cases correspond to the M5 subtype. Interestingly, all of the pediatric cases (< 15 years) were CD36 negative. CD34 was used as the second marker in the algorithm. We considered three subgroups among the CD36− population according to the level of expression of CD34: CD36−/CD34− samples with less than 20% positivity for CD34, CD36−/CD34+ samples with 20 to 70% positivity for CD34 and CD36−/CD34++ cells with greater than 70% positivity for CD34. These thresholds were chosen to separate subgroups in the most efficient manner. All of the M3 cases were found to be CD36−/CD34− but these M3 cases represented only 25% of all CD36−/CD34− cases. Our results suggest that a sequential analysis of CD36 and CD34 can assist to confirm the diagnosis of FAB subtypes.

Disclosure: No relevant conflicts of interest to declare.

Author notes

*

Corresponding author