Constitutive phosphoinositide 3-kinase/Akt activation represents a favorable prognostic factor in de novo acute myelogenous leukemia patients

Several recent articles have highlighted that the class IA phosphoinositide 3-kinase/serine-threonine kinase (PI3K)/Akt pathway is activated in acute myelogenous leukemia (AML), both in blast cells and in the immature CD34 ⁺, CD38^Low, and CD123 ⁺ population.^1-4  Most PI3K activity in AML is caused by the p110δ isoform of PI3K, the activity of which can be specifically inhibited by IC87114.^5,6  The mechanisms leading to PI3K activation are unclear and no activating mutation in the PI3KCD gene coding this isoform has been identified.⁷ 

Two studies have suggested that the prognosis of AML patients with an upregulated PI3K/Akt pathway is worst, suggesting that this pathway could constitute a therapeutic target.^2,8  To ascertain this hypothesis, we prospectively studied constitutive PI3K/Akt/FOXO3A activation in 92 de novo AML patients.

Ninety-two patients with de novo AML treated in the AML-2001 trial from the Groupe Ouest-Est des Leucémies et Autres Maladies du Sang (GOELAMS) were included, after approval by the GOELAMS Institutional Review Board and receipt of signed informed consent in accordance with the Declaration of Helsinki. The patients received cytosine-arabinoside (200 mg/m², days 1-7) and daunorubicin (60 mg/m², days 1-3) or idarubicin (8 mg/m², days 1-5), and a second course (daunorubicin 35 mg/m², days 17-18 or idarubicin 8 mg/m², days 17-18, and cytosine-arabinoside 1 g/m² every 12 hours days 17-20) if bone marrow blasts were more than 5% at day 15. All patients in complete remission received low-dose consolidation with daunorubicin (60 mg/m², days 1-2) or idarubicin (12 mg/m², days 1-2), and cytosine-arabinoside (100 mg/m², days 1-7). Patients with HLA-identical siblings underwent allogenic stem cell transplant (up front until 50 years old; after a consolidation and with a reduced intensity conditioning after 51 years old). Other patients received one consolidation (idarubicin 12 mg/m² or daunorubicin 60 mg/m², days 1-2 and cytosine-arabinoside 3g/m² every 12 hours days 1-4), followed by autologous stem cell transplantation or high-dose chemotherapy in favorable-risk karyotypes.

Fresh blast cells from the bone marrow were studied at diagnosis for PI3K/Akt activation and FOXO3A (a Forkhead transcription factor) phosphorylation. Highly infiltrated samples (75 [81%] of 92) were studied by Western blot, and others (17 [19%] of 92) by flow cytometry. We previously reported that both methods gave identical results when tested on the same samples.⁴  Blast cells were starved for 4 hours in cytokine and serum-free medium before testing for constitutive or FLT3-ligand (a growth factor for hematopoietic progenitors)–stimulated Akt and FOXO3A phosphorylation by Western blot analysis.

Antibodies used were anti-p-Akt (Ser 473 and Thr 308, Cell Signaling Technologies, Danvers, MA), anti-pFOXO3A (Thr 32; Upstate Biotechnology, Charlottesville, VA), anti-beta-actin (Sigma, St Louis, MO), and anti-Akt1/2 (Cell Signaling Technologies).

Dual color flow cytometry of Akt Ser 473 phosphorylation in the CD45_low leukemic population was performed as described.⁴ 

These analyses were performed as described.^7,9-11 

Statistical analysis is described in Figure 1B and 1C.

We first determined that a 4-hour starvation clearly divided AML patients into 2 groups: one with PI3K constitutive activation (PI3K ⁺) and the other without PI3K activation (PI3K ⁻; Figure 1A). In patient 1 (PI3K ⁺ ), Akt (Ser 473) and FOXO3A (Thr 32) phosphorylations were detected with the same intensity during 8 hours of starvation. LY294002 inhibited both phosphorylations. Patient 2 is representative of the PI3K ⁻ group and had no detectable Akt or FOXO3a phosphorylation. The functionality of the PI3K/Akt pathway was detected by stimulating the cells with FLT3-ligand as shown in patient 2. All PI3K ⁻ samples presented an activation of the pathway after cytokine stimulation. In 15 samples tested, a perfect correlation between Akt phosphorylation on Ser 473 and on Thr 308 residues was found (data not shown). Foxo3A phosphorylation was only detected in PI3K ⁺ patients. The PI3K ⁺ group represented 46 (50%) of 92 patients.

Patient characteristics according to their PI3K status are reported in Table 1. There was no difference between the 2 groups for age, sex, FAB subtypes, cytogenetic subgroups, myelodysplastic syndrome (MDS) features, white blood cell counts, LDH levels, bone marrow blast percentage, or tumor burden. The rates of N-RAS/K-RAS and NPM mutations were similar between PI3K ⁺ and PI3K ⁻ patients. FLT3-ITD were slightly more frequent in the PI3K ⁻ group (P = .048; Table 1).

Complete remission was obtained in 74 (80%) of 92 patients with no difference between the 2 groups (P = .60). With a median follow-up of 26 months (range, 3 days to 37 months), 48 patients died. Overall survival was 45% (95% confidence interval [CI], 34%–58%) for the entire population. Figure 1B shows a clear nonproportional effect between the 2 groups, confirmed by the test by Grambsch et al¹⁴  (P = .002). A 2-phase effect is present (Figure 1B), with 2 hazard rates crossing each other at approximately 1 year. When testing for short-term and long-term effects, the test is highly significant (P = .014), with PI3K activation being associated with a better long-term survival. For an actually unknown reason, slightly more patients in the PI3K ⁺ group died from infection before 12 months without AML relapse (short-term effect, P = .027). Nevertheless, the overall survival rate at 30 months was 56% (95% CI, 41%-76%) in the PI3K ⁺ group and 33% (95% CI, 21%-52%) in the PI3K ⁻ group.

Thirty patients relapsed. Crude cumulative incidence curves for relapse are presented in Figure 1C. Relapse free survival was 72% (95% CI, 54%-89%) in the PI3K ⁺ group and 41% (95% CI, 24%-58%) in the PI3K ⁻ group (P = .001).

Because there was more FLT3-ITD in the PI3K ⁻ patients, and a nonsignificant imbalance in term of unfavorable cytogenetics (PI3K ⁺ 16% and PI3K ⁻ 30%), explorative multivariate models were performed. The PI3K activation still played a significant prognostic role on overall survival and relapse-free survival after adjustment for FLT3-ITD (P = .01 and P = .007, respectively) and cytogenetics (P = .014 and .008, respectively).

Two studies have correlated Akt phosphorylation to an adverse overall survival in AML.^2,8  In their study, Kornblau et al⁸  did not starve blast cells before testing for Akt phosphorylation, and they did not test for constitutive PI3 activation. Accordingly, they observed that 91% of blast samples presented Akt phosphorylation, whereas we observed constitutive activation in 50% of AML samples only. Moreover, the populations studied were not similar, because all patients in our study had de novo AML, whereas 32% in Kornblau's study had AML secondary to MDS or chemotherapy.⁸  In their study, Min et al²  observed that Akt was undetectable in pAkt-negative patients. Clearly, their samples are different from those tested in our study in which pAkt could be induced in all PI3K ⁻ samples stimulated by FLT3-ligand.

One hypothesis for the lower relapse rate of PI3K ⁺ patients may be that PI3K drives immature leukemic populations in the S phase, increasing their susceptibility to cell cycle-dependant chemotherapies. Overall, these results suggest that precise identification of Akt targets is needed before targeting this pathway. As recently reported in cancer models, inhibiting the Akt pathway may lead to unexpected dark side effects.^12,13 

The authors thank all participating investigators from the GOELAMS (Groupe Ouest-Est d'Etude des Leucémies et Autres Maladies du Sang).

This work was supported by l'Association Laurette Fugain and by grants from the Association pour la Recherche contre le Cancer (ARC).

Contribution: J.T., V.B., and N.C. performed research. C.E. and P.B. performed the statistical analysis. S.P., B.L., N.I., F.W., and F.D. provided patient samples. P.C.L., V.H., and O.B. performed molecular analysis. P.M., C.L., and D.B. designed research and wrote the paper.

J.T. and C.E. contributed equally to this work.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Didier Bouscary, Département d'Hématologie, Institut Cochin, 27 rue du Faubourg Saint-Jacques, F-75014 Paris, France; e-mail: bouscary@cochin.inserm.fr.