Intact apoptosis signaling in myeloid leukemia cells determines treatment outcome in childhood AML

Successful treatment of children with acute myeloid leukemia (AML) has improved, reaching overall survival rates of 62%.¹  A considerable number of patients still encounter relapse, however. Because apoptosis plays a key role in regulating tissue homeostasis, defects in apoptosis signaling have been considered to be responsible for treatment failure. Consequently, several apoptosis-regulating molecules have been evaluated for their prognostic impact. Imbalanced expression of Bax and Bcl-2 in favor of the proapoptotic molecule was found to be correlated to good prognosis in AML.²  On the contrary, studies assessing transcript levels of apoptosis signaling molecules showed inverse findings.³  However, expression of single factors may not reflect the efficiency of the cell death machinery. For this reason we previously developed a different approach analyzing the functional integrity of apoptosis signaling in individual cells by simultaneously investigating 2 key apoptogenic events: mitochondrial release of cytochrome c and activation of caspase-3.⁴  Using this method, a prognostic potential of intact apoptosis signaling in acute lymphoblastic leukemia (ALL) cells was described. A new parameter, cytochrome c–related activation of caspases (CRAC), was identified indicating intact or defective cytochrome c–related activation of caspases and implying prognostic impact.⁵  Based on these findings we analyzed 45 pediatric AML cell samples for the significance of intact apoptosis signaling for successful treatment.

The AML Berlin-Frankfurt-Münster (BFM) study was approved by the ethics commission of the board of physicians in Westfalen-Lippe, Germany, and the ethics review board of the faculty of medicine, University of Münster, Germany. Analysis of patient material was carried out in accordance with the guidelines of the AML BFM study.

Forty five diagnostic samples (frozen bone marrow or peripheral blood) obtained before treatment from pediatric de novo AML patients were analyzed. Patients were treated according to the AML BFM-98 or -2004 protocols after informed consent was obtained in accordance with the Declaration of Helsinki. Treatment and diagnosis of relapse were accomplished according to the AML BFM trial criteria. Treatment response or complete remission (CR) were examined by bone marrow morphology (BM) and detection of minimal residual disease (MRD) by flow cytometry (time point [tp] 1, day 15; tp2, days 21-28; tp3, days 42-56; and tp4, days 70-84 of treatment) as described earlier.^1,6  Leukemia cells were analyzed for apoptosis signaling and correlated to clinical data as described previously.⁵  In brief, after induction of spontaneous apoptosis by factor deprivation in culture⁷  (with or without 100 μM zVADfmk), cells were analyzed by flow cytometry and evaluated in the caspase-3 versus cytochrome c plot. The parameters cell death, active caspase-3 (ac), total cytochrome c release (cc^total), caspase-dependent (cc^dep) and -independent (cc^indep) cytochrome c release were quantified as described earlier⁵  (Figure 1A).

Release of cytochrome c and consecutive apoptosome formation lead to activation of downstream effector caspases such as caspase-3, resulting in cells staining low for cytochrome c and positive for activated caspase-3 (Figure 1Bi,ii). Apoptosomal dysfunction results in impaired caspase-3 activation despite release of cytochrome c. Caspases other than caspase-3 acting upstream of mitochondria, such as caspase-8, are also involved in initiating or amplifying the formation or activity of this complex. By incubation in the presence or absence of the pan-caspase inhibitor zVADfmk, cytochrome c release (cc) was observed to be dependent (Figure 1Bi) or independent (Figure 1Bii) on upstream or amplifier caspases (cc^dep, cc^indep). Evaluating the differences in signaling pattern with respect to treatment outcome, the absolute values for cell death (cd), active caspase-3 (ac), and released cytochrome c (cc^total, cc^dep and cc^indep, respectively) were compared (independent samples t test) in patients grouped according to treatment response. No differences were found for the mean values of cd, cc^total and cc^indep. However, differences in mean values for cc^dep were detected comparing patients with good or poor response to treatment (BM; tp1: P = .036, tp3: P = .014 and CR/no CR: P = .016). Higher values for ac were observed in patients encountering relapse (overall relapse, trend P = .057; relapse within 12 months, P = .025 and within 24 months, P = .015) in contrast to not or late relapsing patients. According to the AML BFM protocol, patients are stratified into standard risk (SR) or high risk (HR) groups based on French-American-British (FAB) classification, cytogenetics, and response to induction therapy on day 15. Although in patient groups differentially responding on day 15 to induction therapy, no difference in ac values was found; HR patients showed higher ac values than SR patients (P = .002). Active caspase-3 was identified as an important factor for outcome in pediatric ALL and AML, however, with opposite features. While ALL patients with good response and no relapse revealed high levels of caspase activation, AML patients displayed an association of high caspase-3 activation to high risk and relapse. Likewise, in adult ALL high caspase-3 expression correlated with complete remission,⁸  whereas high levels of activated caspase-3 were associated with impaired apoptosis and decreased survival of adult AML patients.⁹  Together with conflicting reports on the prognostic relevance of caspase-3 in leukemia,^10,–12  these findings strongly suggest that the complex regulated apoptosis signaling system is insufficiently represented if analyzed by one molecule alone. The findings also emphasise the importance to analyze checkpoints or molecules integrating diverse apoptosis signaling pathways, such as cytochrome c, caspase-3, and apoptosome formation.

We hypothesised that intact apoptosis signaling is characterized by concomitant release of cytochrome c and caspase-3 activation and therefore would be found in the patient groups with favorable outcome. Correlating these 2 events directly revealed that only cytochrome c release dependent on amplifying caspases (cc^dep) significantly correlated with activated caspase-3 and was exclusively detected in patient groups with good response to initial treatment (BM blasts < 5%, tp1-4) or groups exhibiting MRD negativity (tp1-3, tp4 trend) (Table 1). The same finding was made in the groups achieving CR or retaining CR longer than 1 year and in the group without relapse. Remarkably, this correlation was completely absent in the group with poor response to initial treatment, groups not reaching CR or with CR less than 1 year as well as in the group of relapsing patients. This indicates that an intact apoptosome function and active amplifying caspases are elementary for favorable outcome in childhood AML.

The functional integrity of this important apoptogenic checkpoint is subsumed by the parameter “caspase dependent cytochrome c-related activation of caspase-3” (CRAC^dep), which was calculated as the difference of ac and cc^dep in equivalence to the CRAC-value described for ALL.⁵  Proficient signaling results in positive CRAC^dep values in contrast to deficient signaling resulting in negative CRAC^dep values. Division of the patient cohort according to this new parameter resulted in 38 CRAC^dep-positive and 7 -negative patients. The distribution of CRAC^dep values was unrelated to features such as gender, FAB subtype, age or initial blast cell count (Table S1, available on the Blood website; see the Supplemental Materials link at the top of the online article). Maintenance of CR longer than 1 year predisposes for favorable outcome in pediatric AML.^13,14  Thirty of 32 patients sustaining CR for more than 1 year were CRAC^dep-positive (Fisher exact test, P = .015). Analysis of survival and remission duration revealed a superior outcome for CRAC^dep-positive patients overall but also specifically in the HR patient group. Notably, SR stratified patients merely displayed CRAC^dep-positivity (Figure 1C, Tables 2,3). Multivariate analysis revealed an increased risk for relapse and decreased probability of continuous CR for CRAC^dep negative patients in contrast to established risk factors like day 15 remission failure or hyperleucocytosis¹⁵  (Table 4).

Taken together, we clearly found that the propensity to undergo apoptosis of leukemia cells delineated by intact cytochrome c related caspase-3 activation (CRAC^dep) is required for successful treatment of pediatric AML. Analysis of the functional integrity of this apoptosis checkpoint can thus be used for future treatment stratification.

We thank Johanna Pfeil for her excellent technical assistance.

This work was supported by a research fellowship of the European Hematology Association (L.H.M.) and a grant of the Deutsche Forschungsgemeinschaft (K.S. and K.-M. D.).

Contribution: L.H.M. designed research, performed experiments, analyzed data, and wrote the manuscript; M.Q. performed experiments, analyzed data, prepared the figures, and critically read the manuscript; S.M.E. analyzed data and critically read the manuscript; U.C., D.R., and W.-D.L. collected and provided patient samples and clinical data, and critically read the manuscript; K.S. and L.K. designed research and critically read the manuscript; and K.-M.D. designed research, analyzed data, and wrote the manuscript.

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

Correspondence: Klaus-Michael Debatin, University of Ulm, Department of Paediatrics and Adolescent Medicine, Eythstrasse 24, 89075 Ulm, Germany; e-mail: klaus-michael.debatin@uniklinik-ulm.de.