Survival Stratification In Acute Myeloid Leukemia By Single Cell Signal Profiling

Dysregulation and mutations in signaling genes of cancer cells characterize more than half of the acute myeloid leukemia (AML) patients, and contribute to chemoresistance through regulation of cellular processes including apoptosis and DNA repair. We investigated if determination of single cell basal phosphorylation of signaling proteins reflected mutation of FLT3 and NPM1, cytogenetics, response to first course of chemotherapy or overall survival.

We employed flow cytometric single cell analysis of phosphoproteins central to signal transduction pathways in myeloid cancer cells and analyzed peripheral blood leukocytes from 93 acute myeloid leukemia (AML) patients. Blood samples from consecutively diagnosed AML patients with high peripheral blood blast counts (>7x109/L, >70% blasts) were collected after informed consent was given and biobanked by cryo-preservation. AML samples was thawed and equilibrated for one hour in a defined serum-free medium (StemSpan SFEM medium, Stem Cell Technologies) which include insulin and transferrin. All samples were viability controlled and validated for growth factor response. Basal phosphorylation was determined using 17 phosphorylation specific antibodies: All staining panels contain the same 4 surface antibodies and a live dead discriminator; CD33(P67.6) PerCP-Cy5.5, CD38(HB7) PE-Cy7, CD34(581) PE, CD45(MEM-28) PE-Dynamics590 and phospho c-PARP(Asp214) Alexa Flour 700. Two phosphospecific antibodies were added to each panel with the respective direct conjugated dye Alexa Flour 488 and Alexa Flour 647; p38(pT180/pY182) and ERK2(pT202/pY204), SRC(pY418) and Akt(pT308), PDK1(pS241) and Akt(pS473), STAT1(pY701) and ribosomal protein S6(pS235/36), STAT3(pY705) and STAT5(pY694), CREB(pS133) and STAT3(pS727), ribosomal protein S6(pS240) and NFkB(pS529), 4EBP(pT37/pT45) and STAT6(Y641), without p-antibody and JNK(pT183/pY185). The lowest median signal for each phosphoprotein in AML cells and lymphocytes, respectively, were used as reference value for calculation of basal phosphorylation. Hierarchical clustering with the use of complete linkage were created using TM4, and Principal Component Analysis was carried out using Unscrambler X (CAMO Software).

Unsupervised clustering revealed two distinct signature clusters based on low or elevated phosphorylation level among AML cells. A similar cluster signature was absent in endogenous non-leukemic lymphocytes from the same patients. No correlations between basal phosphorylation and prognostic mutations (FLT3 or NPM1), cytogenetics or response to first course of chemotherapy were found. In AML patients treated with intensive chemotherapy (n=45) the cluster with low phosphorylation level (n=19) correlated with significant (p=0.007) shorter overall survival. Principal component analysis verified the cluster analysis and guided a reduction to only three phosphoproteins (STAT3, 4EBP1, ribosomal protein S6) with statistically significant (p=0.014) stratification of survival.

Leukemic cells demonstrated a phosphorylation profile that reflected survival of the intensively treated patients, but surprisingly not correlated with mutational status of FLT3, NPM1, cytogenetics or first course remission status. This suggests that phosphoprotein determination in leukemic cells provide prognostic information so far not available with current diagnostics. The robust and relatively simple method of single cell signal profiling should be tested in clinical trials to examine its feasibility in therapy response prediction. More extensive mutational and epigenetic analyses are needed in search for the molecular origin of the low/high signal profiles.

(A) Unsupervised hierarchical cluster analysis based on three phospho protein analysis stratified in two distinct clusters of patients receiving standard intensive induction chemotherapy (n=45; daunorubicin + cytarabine or idarubicin + cytarabine (3+7) similar to HOVON AML protocols 103 and 102, respectively). (B) Kaplan-Meier plot was performed calculating P value with the use mantel-Cox log-rank test.

Figure 1

View largeDownload slide

Phosphorylation status of STAT3(S727), 4EBP1(T37/Y45) and ribosomal protein S6 (S235/6), a diagnostic tool for estimating overall survival.

Figure 1

View largeDownload slide

Phosphorylation status of STAT3(S727), 4EBP1(T37/Y45) and ribosomal protein S6 (S235/6), a diagnostic tool for estimating overall survival.

Close modal

No relevant conflicts of interest to declare.