Identification of Prognostic Immunophenotypes at First Diagnosis in Patients with Acute Myeloid Leukemia (AML) By a Standardized Multicolor Flow Cytometry (MFC) Panel Originally Designed to Detect Measurable Residual Disease (MRD) at Follow-up

Aims

In AML, several risk factors obtained at first diagnosis (FD) have been reported to be associated with shorter RFS and OS. The primary prognostic relevance of multicolour flow cytometry (MFC) has been a matter of debate for years. During follow-up (FU), the prognostic relevance of MRD as detected by MFC is less controversial and MFC is recommended in particular (but not exclusive) for those patients (pts) with no reliable genetic marker. We thought to evaluate the prognostic value at FD of a recently established antigen panel and a corresponding analysis strategy, which had been originally developed for MRD-detection.

Methods

Based on an 8-colour antibody panel (CD45, CD34, CD117, HLA-DR, CD13, CD33, CD7, CD56), we have developed a hierarchical gating strategy with mainly fixed gates. That allows to detect MRD with a high level of standardization and inter-observer reliability (Röhnert M., et al. 25^th EHA 2020). Four distinct categories of aberrations (deficiency of CD13 or CD33, cross-lineage expression of CD7 or CD56) detectable on at least 10% of the myeloid blast population were used to define aberrant phenotypes termed leukemia associated immunophenotypes (LAIP) at FD. These categories were also chosen to define MRD during FU. MRD^pos by LAIP was defined as the (re-)occurrence of an aberrant category already detectable at FD, while MRD^pos by DfN (different from normal) was defined by the de-novo detection of an aberrant category at FU.

The prognostic value of the aberrant phenotypes at FD was examined in a cohort of 528 pts. In 122 pts, we further analysed MRD (LAIP/DfN) after completion of intensive induction chemotherapy (IT). Consolidation therapy consisted of allogeneic hematopoietic stem cell transplantation (n=77) or chemotherapy (n=45).

The bone marrow samples were measured centrally and analysed independently by three different investigators.

Results

The probability to achieve a complete remission (CR) varied between the different aberrant phenotypes (LAIP) at FD. Compared to pts without aberrant phenotype (CR rate=68%, n/N=100/148), pts with CD56^only (the sole aberrant category was a cross-lineage expression of CD56=only) had a significantly lower CR rate (46%, n/N=15/33, p=0.019). The other exclusive aberrant categories did not significantly influence CR rates compared to pts without LAIP: CD13^only (75%, n/N=53/71, p=0.286), CD33^only (64%, n/N=59/97, p=0.28) and CD7^only (62%, n/N=31/50, p=0.472).

In pts with possibly co-occurring aberrant categories (compound aberrant phenotype=comp), the CR rate was significantly higher in CD13^comp compared to all other patients (75% vs. 64%, 107/143 vs. 246/385, p=0.018). The other compound aberrancies did not significantly influence CR rates: CD33^comp (63% vs. 68%, 90/143 vs. 263/385, p=0.244), CD7^comp (66% vs. 67%, 72/109 vs. 281/419, p=0.842) and CD56^comp (68% vs. 66%, 84/123 vs. 269/405, p=0.699). Regarding overall survival (OS), just CD56^only retained its statistical significance (HR 2.5, CI 1.4-4.7, p=0.004). CD13^comp was associated with favourable outcome but without reaching statistical significance (HR 0.7, CI 0.4-1.0, p=0.059).

In the cohort of pts with MRD assessment at the end of IT, 67% were classified as responders (CR n=62, CRi n=19) and 33% as non-responders (PR n=14, refractory n=26) by cytomorphology. By MFC, 71% of these pts were classified as MRD^pos (n=51/36 responders/non-responders) and 29% as MRD^neg (n=30/4). MRD^pos was defined by LAIP only (23%), DfN only (44%) or concordantly by LAIP+DfN (33%). OS of MRD^neg pts was significantly longer compared to MRD^pos patients (HR 4.3, CI 1.0-18.1, p=0.033).

Conclusions

Using our analysis approach originally developed for MRD monitoring, MFC could provide additional information for initial risk stratification. The presence of an isolated cross-lineage expression of CD56 (CD56^only) was associated with a lower CR rate and significant shorter OS. In contrast, CD13^comp (CD13 deficiency ± other aberrant categories) was associated with a higher CR rate and prolonged OS.

Furthermore, MRD^pos as defined by the combined LAIP/DfN strategy provided significant prognostic information.

The presented results are currently refined and validated using genetically defined subcategories. The approach has to be confirmed in an independent cohort of pts.