Multiple distinct molecular mechanisms influence sensitivity and resistance to MDM2 inhibitors in adult acute myelogenous leukemia

The therapeutic outcome in adult acute myelogenous leukemia (AML) remains unsatisfactory, and novel treatment approaches are needed to improve the prognosis of affected patients. One promising approach involves chemical activation of p53 through use of drugs that interfere with the binding of p53 and MDM2 (MDM2 inhibitors): a nongenotoxic approach of inducing cancer cell apoptosis. Various compounds that directly interfere with the binding of p53 and MDM2, including the Nutlins and the MI-series of MDM2 inhibitors, have been developed.^1-5  Currently available evidence indicates that induction of p53 through MDM2 inhibition by Nutlins or MI-series compounds results in the elevation of p53 protein levels, followed by p53-mediated apoptosis or p53/p21-mediated cell cycle arrest.^3,6-9  For reasons that remain largely unknown, noncancerous cells are relatively resistant to MDM2 inhibitor-mediated apoptosis and usually undergo transient cell cycle arrest.^10,11  Equally unclear is the nature and exact contribution of various p53 network/effector molecules to MDM2 inhibitor-induced apoptosis; thus, it remains unknown whether individual p53 effector genes or signaling pathways are absolutely necessary for MDM2 inhibitor-induced apoptosis to occur.^6,12-15  Evidence for involvement of intrinsic and extrinsic apoptosis pathways in MDM2 inhibitor-induced apoptosis as well as direct effects of the p53 protein on mitochondrial apoptosis molecules has been provided, and it is thus possible that MDM2 inhibitor-mediated apoptosis uses functionally redundant apoptotic pathways.^16-21 

Studies into resistance mechanisms to MDM2 inhibitors in various cell systems have proven that intact p53 is necessary for MDM2 inhibitor-induced apoptosis to occur.^10,22,23  What is less clear is how often and under what cellular circumstances wild-type p53 status is alone sufficient as a predictor for sensitivity, or what other sensitivity/resistance determinants may be operational.^24-26  Two of the proposed regulators of p53-mediated apoptosis are MDM2 and MDMX, and elevated levels of these proteins have been shown to influence MDM2 inhibitor sensitivities in various experimental settings. Nonetheless, the available evidence in support of a critical role of these proteins is neither conclusive nor consistent across experimental systems; thus, it remains possible that these p53 regulatory molecules are not critical determinants of MDM2 inhibitor efficaciousness in all human tumors.^27-29 

As part of broad-based efforts to study the therapeutic potential of MDM2 inhibitors in hematologic malignancies (which are generally characterized by only a small fraction of cases with mutated p53), we have studied ex vivo apoptosis induction in more than 100 human AML samples. Through the studies detailed herein, we have identified a previously unsuspected large fraction of primary human AML samples that displayed primary resistance to MDM2 inhibitors despite wild-type p53 exons 5-9 gene status. Initial investigations into this phenomenon provide evidence for multiple distinct mechanisms of resistance: one centered on insufficient p53 protein induction and another centered on defective p53 protein or defective p53-regulated effector pathways. These novel findings substantially complicate transition of MDM2 inhibitors into clinical AML applications and motivate further study to achieve optimal efficaciousness of these drugs in the clinical setting. Finally, through correlative analysis, we have identified, for the first time, a significant and strong association between mutated Flt3 status (presence of Flt3-ITD) and heightened sensitivity to MDM2 inhibitors, thus providing a novel and practical rationale for MDM2 inhibitor trial design, patient subgroup selection and trial data interpretation in AML.

The 109 AML cases analyzed in this study were enrolled at the University of Michigan Comprehensive Cancer Center between March 2005 and October 2009. The study was approved by the University of Michigan Institutional Review Board (IRBMED 2004-1022), and written informed consent was obtained from all patients before enrollment in accordance with the Declaration of Helsinki. Samples from consecutively enrolled patients were analyzed as long as sufficient material was available for the various analyses described herein.

Mononuclear cells from blood or marrow from AML patients were isolated by Ficoll gradient centrifugation (GE Healthcare), aliquoted into fetal calf serum with 10% dimethyl sulfoxide (DMSO), and cryopreserved in liquid nitrogen. For purification of AML blasts using negative selection, cryopreserved cells were washed and recovered by centrifugation and then treated with anti–human CD3 (Miltenyi Biotec 130-050-101), anti–human CD14 (if blasts were negative for CD14 expression; Miltenyi Biotec 130-050-201), anti–human CD19 (if blasts were negative for CD19 expression; Miltenyi Biotec 130-050-301), and anti–human CD235a (Miltenyi Biotec 130-050-501) microbeads per the manufacturer's recommendations. Cell suspensions were run through Miltenyi MACS separation LS columns (130-042-401) to negatively enrich for AML blasts. All blast preps were analyzed by cytospins for purity. This schema always resulted in greater than 90% blast purity.

AML blast DNA used for SNP array, Version 6.0 profiling was extracted from samples that were further purified as follows: post-Miltenyi column samples were washed and stained with fluorescein isothiocyanate–conjugated anti-CD33, phycoerythrin-conjugated anti-CD13, and allophycocyanin-conjugated anti-CD45 (all antibodies: eBioscience). After final washing, propidium iodide (PI) was added to a concentration of 1 μg/mL to discriminate dead cells. Sorting of cells was done on a FACSAria high-speed flow cytometer (BD Biosciences). Live cells (PI-negative) were gated for blasts by identifying those cells with intermediate-intensity staining for CD45 and low- to moderate-intensity side scatter.³⁰  CD33 and CD13 were then used to further discriminate blasts versus erythroid lineage and mature myeloid lineage cells.

Blasts enriched to more than 90% purity using methods detailed in “Cell purification” were incubated in serum-supplemented RPMI medium at 2.5 × 10⁵ cells in 100 μL final volume in the presence of various concentrations of the MDM2 inhibitors MI-219 and MI-63 (range, 0.625-20μM) for 40 hours. Apoptosis and necrosis were measured for each treated blast aliquot using annexin V/PI flow cytometry–based readouts, and values were subsequently normalized to spontaneous death rates in untreated parallel cultures according to the formula (% alive = % mean alive treated samples/% mean alive paired nontreated samples).

Blasts enriched to more than 90% purity using methods detailed in “Cell purification” were incubated with either DMSO or 0.5μM 5-azacytidine (A2385; Sigma-Aldrich) for 48 hours (with 5-azacytidine replenished every 24 hours). During the last 12 hours of incubation, blasts were further aliquoted and treated with either 0.3μM trichostatin A (9950; Cell Signaling Technology) or DMSO. At the end of the 48-hour incubation, each of the 4 differentially treated subgroups of blasts was treated with MI-219 at final concentrations of 0, 2.5, 5, and 10μM for 40 hours, followed by annexin V/PI FACS-based analysis of apoptosis. Aliquots of blasts in parallel were cultured in a 48-well plate at 10⁶ cells per well in 1 mL of medium and treated with 10μM MI-219 or solvent for 8 hours. Blasts were harvested, lysed, and protein prepared for immunoblotting as described.

RNA was prepared from FACS-sorted blasts from AML cases using the Trizol reagent and resuspended in 50 μL diethyl pyrocarbonate-treated water. A total of 20 μL complementary DNA was made from approximately 50 ng of RNA using the Superscript III first strand synthesis kit (Invitrogen) and random priming. Primers and TaqMan-based probes were purchased from Applied Biosystems (Primers-on-demand). Primer/probe mixtures included: p53 (Hs_00153349_m1), MDM2 (Hs_01066930_m1), MDM4 (Hs_00159092_m1), and Hu PGK1.

Duplicate amplification reactions included primers/probes, TaqMan 2x Universal PCR Master Mix, No AmpErase UNG, and 1 μL of cDNA in a 20-μL reaction volume. Reactions were done on an ABI 7900HT machine. Normalization of relative copy number estimates for the mRNA of the gene of interest was done with the threshold cycle (C_t) values for PGK1 as reference (C_t mean gene of interest − C_t mean PGK1). Comparisons between AML subgroups were performed though subtractions of means of normalized C_t values.

Primary AML cases with wild-type p53 exons 5 to 9 were ranked according to the 50% inhibitory concentration (IC₅₀) values for MI-219 and 15 cases with high IC₅₀ values (IC₅₀ > 10μM) and 15 cases with low IC₅₀ values (IC₅₀ < 2μM) selected for further analysis. Blasts were purified as outlined in “Cell purification” and subsequently cultured for 8 hours with 10μM MI-219, 10μM Nutlin-3, solvent, or treated with 5 Gy of ionizing radiation. Cells were harvested after treatment, lysed in detergent lysis buffer (50mM Tris, pH 7.5, 100mM NaCl, 2mM ethylenediaminetetraacetic acid, 2mM ethyleneglycoltetraacetic acid, 1% Triton X-100, 20mM NaF, 1mM sodium orthovanadate [13721-39-6 Alfa Aesar], 1mM phenylmethylsulphonylfluoride [Pierce Chemical], phosphatase inhibitor cocktail I [P2850; Sigma-Aldrich], and protease inhibitor cocktail [P8340; Sigma-Aldrich]), protein fractionated, and prepared for immunoblotting with antibodies directed against p53 (Ab-6, clone DO-1; Calbiochem) and actin (AC-15; Sigma-Aldrich). Positive control lysates were generated from the AML cell line MOLM-13 treated with MI-219 at 10μM for 8 hours, and aliquots of these lysates were run side-by-side with lysates of the primary cases on every immunoblot. Thus, these MOLM-13 lysates served as internal standards for blot-to-blot band intensity comparisons. Leftover lysates from these experiments were subsequently prepared for immunoblotting using antibodies against human MDMX (A300-287A; Bethyl Laboratories), MDM2 (Ab-1, clone IF2; Calbiochem), p21 (clone SX118; BD Biosciences), and actin.

The SNP 6.0 assay was performed following the manufacturer's recommended protocols. Affymetrix CEL files for each blast and buccal sample were analyzed using Affymetrix Genotyping Console software for initial quality control, followed by use of the Affymetrix “Birdseed” algorithm to generate tab-delimited SNP call files in text format. Call rates for the entire group of samples included in this report were between 94.93% and 99.45%, with a mean call rate of 98.33%. Sample copy number heatmap displays were obtained from CEL files through use of the freely available software dChip (build date February 25, 2010)³¹  adapted to a 64-bit operating system environment. To generate functional and practical displays of loss of heterozygosity (LOH), a 2-step, internally developed, Java-based software analysis system was used. The Pre-LOH Unification Tool served to align all individual patient SNP calls to their respective dbSNP rs ID numbers and genomic physical positions before incorporation into the LOH tool, Version 2, an updated version of the LOH tool able to accommodate Affy SNP 6.0 array data.³²  For LOH analysis between paired samples, a filter setting within the LOH tool, Version 2 was used, allowing visualization of individual paired SNP calls as LOH only if present within 3000 bp of another such call. This step filtered out many false, sporadically distributed single LOH calls because of platform noise.

Primers to amplify and sequence exon 12 of human NPM1, exons 13 to 15 and 20 of human Flt3, and exons 5 to 9 of human p53 and adjacent intronic sequences were designed using the primer 3 program (http://frodo.wi.mit.edu/primer3/). Polymerase chain reaction (PCR) products were generated using Repli-g (QIAGEN)–amplified DNA from highly pure blast cells as templates. Amplifications were done using Taq polymerase. PCR amplicons were prepared for direct sequencing with internal nested sequencing primers using the exonuclease/shrimp alkaline phosphatase method (USB). Mutation Surveyor (SoftGenetics LLC) software was used to compare experimental sequences against Refseq GenBank or genomic sequences as well as by visual inspection of sequence tracings. Mutations were confirmed using paired patient buccal DNA as templates.

Associations between binary classifications (eg, between drug sensitivity and gene mutation status) were assessed using log odds ratios. Mean IC₅₀ values were compared between groups of samples using 2-sample t tests. Results for all statistical tests are reported as Z-scores and 2-sided P values.

Detailed clinical and genomic characteristics of the 109 AML patients analyzed in this study are summarized in Table 1 and supplemental Table 1 (available on the Blood Web site; see the Supplemental Materials link at the top of the online article). Of the 109 AML cases analyzed, 90 (83%) were previously untreated and 19 (17%) were relapsed at study enrollment (postanalysis review of relapsed cases demonstrated that none of the cases displayed complex karyotypes, probably as a consequence of the small sample size). A total of 66%, 16%, and 18% were primary, secondary or treatment-related AML, respectively, and 12 cases had p53 exons 5 to 9 mutations.

To evaluate the efficacy of MDM2 inhibitor-mediated apoptotic cell kill in AML blasts ex vivo, we purified blasts from 109 AML specimens (97 with wild-type p53 and 12 with mutant p53 by exon 5-9 exon sequence analysis) to more than 90% purity and incubated cell aliquots for 40 hours with escalating concentrations of the MDM2 inhibitors MI219 (N = 109) and MI63 (N = 60). The apoptotic cell fraction in treated samples was subsequently quantitated through annexin V-PI–based FACS analysis and normalized to measurements in paired untreated cells.

As can be seen in Figure 1, all AML cases with mutant p53 exons 5 to 9 (red) or absent p53 mRNA expression (green) displayed resistance to MDM2-inhibitor treatment, consistent with previous findings.^20,22  Although many AML cases with wild-type p53 exons 5 to 9 (black) were very sensitive (IC₅₀ < 2μM; 32 of 97 = 33%) or sensitive (IC₅₀ ≥ 2μM to < 5μM; 33 of 97 = 34%) to MI-219 or MI-63, a substantial fraction of cases with wild-type p53 displayed various resistance levels (MI219 IC₅₀ > 5μM for 32 of 97 = 33% and IC₅₀ > 10μM for 21 of 97 = 22% of cases, respectively). Thus, unlike the situation in CLL, these data demonstrate that a substantial subset of AML cases with wild-type p53 exons 5 to 9 sequence displays primary resistance to MDM2 inhibitors ex vivo.

Further, the mean IC₅₀ values for MI219 in primary, secondary, and treatment-related AML (exclusive of cases with p53 exon 5-9 mutations), were 6.1μM, 7.9μM, and 4.8μM, respectively. The mean IC₅₀ values for MI219 in previously untreated versus relapsed AML cases (exclusive of cases with p53 exon 5-9 mutations) was 6.6μM versus 4.4μM, respectively.

Given that we had used highly purified blasts in the analysis outlined in this subsection, we became interested in measuring MDM2 inhibitor IC₅₀ values for AML bone marrow-derived mononuclear cell mixtures (after Ficoll gradient purification). The analysis of 10 such cases disclosed invariably higher IC₅₀ values than found in paired highly pure blasts, indicative of substantial resistance of nonblast cells (including nonmalignant bone marrow cells) to MDM2 inhibitor-induced apoptosis. These data are summarized in supplemental Figure 1 and supplemental Table 2.

Next, we assessed the ability of MI-219 to induce apoptosis in 19 AML-derived cell lines. Data are summarized in Figure 2 and Table 2. As can be seen in Figure 2, all AML cell lines with mutant p53 (red) were resistant to MI219, whereas the cell lines with wild-type p53 (black) displayed various degrees of sensitivity/resistance to MI-219, reminiscent of the findings in primary AML blasts presented under “Primary resistance to MDM2 inhibitor treatment is common in adult AML.”

Given the central importance of intact p53 to MDM2 inhibitor sensitivity, we proceeded with analysis of p53 protein expression levels in primary AML blasts. We ranked all primary AML cases with wild-type p53 exons 5 to 9 according to the IC₅₀ values for MI-219 and selected 15 cases with high IC₅₀ values (IC₅₀ > 10μM) and 15 cases with low IC₅₀ values (IC₅₀ values < 2μM) for further analysis. Purified blasts were either left untreated or treated for 8 hours with MI219 (10μM), Nutlin 3 (10μM), or a one-time dose of 5 Gy of external irradiation. Cellular lysates made from these blasts were prepared for immunoblotting with anti-p53 and antiactin antibodies. Further, aliquots of lysates from the MI219-treated AML cell line MOLM13 were analyzed on each blot to permit blot-to-blot comparisons of band intensities.

As can be seen in Figure 3, all sensitive AML blasts demonstrated induction of p53 protein after MDM2 inhibitor treatment or external irradiation, albeit to different absolute levels. Importantly, analysis of p53 protein levels in resistant blasts disclosed 2 subsets: (1) blasts with absent or very low p53 expression after MDM2 inhibitor treatment or external irradiation and (2) blasts with baseline and induced p53 levels essentially equal to the levels measured in sensitive blasts. Thus, resistance to MDM2 inhibitors in AML with wild-type p53 exons 5 to 9 is associated with at least 2 distinct molecular defects: (1) low/absent p53 protein expression or (2) apoptotic p53 network defects (including the possibility of aberrant p53 proteins) in the setting of normal p53 protein levels.

To gain further insights into the mechanisms of low/absent p53 expression in resistant AML blasts, we measured normalized p53 mRNA levels in total RNA purified from FACS-sorted AML blast samples. This analysis disclosed that a few AML cases at baseline displayed absent p53 mRNA (see Figure 5E; AML cases 7, 80, and 120). Thus, resistance to MDM2 inhibitors in a small fraction of AML blasts is the result of absent p53 transcription and suggests an acquired p53 gene defect. However, the majority of resistant AML blasts with low/absent p53 protein did express p53 mRNA (AML cases 98, 138, 191, 36, 40, 101, and 100), thus implying posttranscriptional mechanisms for low p53 protein levels.

Attempting to obtain evidence for epigenetic p53 gene silencing in AML with absent p53 mRNA expression, we selected 4 AML cases based on the availability of cryopreserved cells with absent or very low p53 mRNA for further analysis and treated purified blasts with trichostatin A and azacytidine (alone or in combination) followed by treatment with MI219. The readout for these experiments was the fraction of blasts undergoing apoptosis and posttreatment p53 protein levels. As detailed in supplemental Figure 2, evidence for reversible epigenetic p53 gene silencing in these blasts was not found.

The expression levels of MDM2 and MDMX, 2 critical regulators of p53 protein, could account for the observed differences in IC₅₀ values to MDM2 inhibitor treatment in AML cases with wild-type p53 exons 5-9 and for the observed differences in p53 protein levels. To test such hypotheses, we initially measured normalized MDM2 and MDMX mRNA levels across the entire AML cohort. Subsequently, we correlated these mRNA levels with IC₅₀ values to MDM2 inhibitor-mediated apoptosis in all AML cases with wild-type p53 exons 5-9 (Figure 4); as can be seen in Figure 4, neither MDMX nor MDM2 levels correlated with MI-219 IC₅₀ values.

Next, we measured MDMX, MDM2, and p21 protein levels in lysates from blasts derived from the experiment outlined in Figure 3. As can be seen in supplemental Figures 3 and 4, neither MDMX, nor MDM2, nor p21 protein levels demonstrated a clear association with MI-219 IC₅₀ values.

To obtain additional information regarding p53 gene status in AML, we analyzed DNA samples from ultra-pure AML blast populations from 109 AML cases and paired buccal DNA for acquired chromosomal copy number alterations and LOH using ultra-high-density Affymetrix SNP 6.0 arrays.

In Figure 5, we display subchromosomal copy number status at 17p (Figure 5A, buccal DNA; Figure 5B, AML blast-derived DNA; p53 is located at ∼ 7.5 Mb physical position on 17p), LOH at 17p (Figure 5C), p53 exons 5 to 9 sequence data (Figure 5D), and normalized p53 mRNA data (Figure 5E). As can be seen, 17 of 109 (15%) AML cases displayed LOH involving parts or all of 17p that spans the p53 locus. Importantly, paired analysis for copy loss uncovered 2N status for nearly half (8) of these cases (red numbering): examples of copy-neutral LOH or acquired uniparental disomy at 17p. Of note, copy-neutral LOH is undetectable using conventional karyotyping or comparative genomic hybridization and is thus missed in routine clinical practice. Given that copy-neutral LOH is often associated with gene mutations, we compared LOH data with p53 sequence data and p53 mRNA data and found that 6 of 8 of these 17p-associated acquired uniparental disomy cases (red) displayed homozygous p53 mutations (AML 12, 41, 88, 117, 153, and 157, Figure 5D) and 1 of 8 cases (120) had very little p53 mRNA expression. Thus, acquired copy-neutral LOH is common at the p53 locus, is associated with p53 null states in the majority of cases, and is associated with resistance to MDM2 inhibitor treatment. High resolution copy number analysis of the p53 gene and the p53 promoter did not identify homozygous deletions in AML.

Our analysis of ex vivo sensitivities to MDM2 inhibitors outlined under “Primary resistance to MDM2 inhibitor treatment is common in adult AML” disclosed many cases that were very sensitive to MDM2 inhibitor-mediated apoptosis. We therefore became interested in identification of markers that would correlate with increased MDM2 inhibitor sensitivity. Focusing on Flt3 and NPM1 (the 2 most commonly mutated genes in AML), we correlated the presence or absence of Flt3-ITD or NPM1 exon 12 mutations with MI-219 IC₅₀ values initially in all AMLs with wild-type p53 exons 5 to 9. We repeatedly dichotomized the AML cohort at either the 25th or 50th percentile (corresponding to threshold IC₅₀ values of 1.78μM and 3.2μM, respectively) and determined Z-scores for the presence of mutated Flt3 (Flt3-ITD) or NPM1, respectively. From this analysis, Flt3-ITD emerged as significantly enriched in sensitive AML cases, with Z-scores of 1.91 (P = .06) and 2.26 (P = .02) for the 25th and 50 percentile analysis, respectively. Eleven of 19 (58%) and 13 of 19 (68%) Flt3-ITD–mutated AML cases had IC₅₀ values to MDM2 inhibitors of less than 2μM and less than 2.25μM, respectively.

We also performed a similar analysis for the comparison of Flt3-ITD positive cases versus all other cases (N = 90; including p53 exon 5-9 mutated cases). From this analysis, Flt3-ITD again emerged as significantly enriched in sensitive AML cases, with Z-scores of 2.42 (P = .02) and 2.64 (P = .01) for the 25th and 50th percentile analysis, respectively.

Analysis of AML cases with both Flt3-ITDs and NPM1 mutations (10 of 19 Flt3-ITD positive cases in this cohort were also NPM1 mutated) as well as cases with NPM1 mutations in the absence of Flt3-ITDs (supplemental Table 1) did not suggest a significant role of NPM1 mutations in conferring MDM2 inhibitor sensitivity (the NPM1 mutation frequency in this cohort was 40% in the subset of AML with normal karyotype).

Next, we graphically displayed IC₅₀ values for all 109 AML cases in 3 mutually exclusive categories: (1) presence of p53 exons 5-9 mutations, (2) presence of Flt3-ITD, and (3) all others (Figure 6). As can be seen in Figure 6, most Flt3-ITD-positive AML cases displayed very low IC₅₀ values and significantly lower mean IC₅₀ values than the Flt3 wt and p53 wt group (P = .02). Thus, the majority of AML blasts with Flt3-ITD mutations are highly sensitive to MDM2 inhibitor treatment. Therefore, this analysis identifies, for the first time, a genomic biomarker for MDM2 inhibitor sensitivity with potential for clinical applications.

This report summarizes detailed studies of the molecular determinants in primary AML blasts (N = 109) and their influence on sensitivity or resistance to MDM2 inhibitor treatment ex vivo. One novel finding of this study is the description and quantitative analysis of primary resistance to MDM2 inhibitors in AML. Within this context, it was demonstrated that: (1) p53 mutations confer resistance to MI219, as expected; (2) low or absent p53 expression in the absence of p53 exon 5-9 mutations exists in a subset of AML blasts and is associated with MDM2 inhibitor resistance; and (3) MDM2 inhibitor resistance exists in subsets of AML despite wild-type p53 and robust p53 protein induction after MDM2 inhibitor treatment or irradiation, thus implying defects in the apoptotic p53 network or in the p53 protein. Together, these various AML blast-intrinsic defects result in primary resistance to MDM2 inhibitors in approximately one-third of all AML cases, a fraction much larger than previously appreciated.

Regarding the p53 gene status of resistant AML blasts, multiple findings emerged: (1) the p53 exon 5-9 mutation frequency was 10%, which is in line with previous estimates,^33,34  and is insufficient to explain MDM2 inhibitor resistance in the majority of AML cases; (2) p53 mutations frequently (∼ 50% of all p53 mutations) occurred in the setting of acquired uniparental disomy at 17p in AML; (3) some AML cases with 17p deletions that spanned p53 carried wild-type p53 and were sensitive to MI219; and (4) some AML cases with 17p deletions that spanned p53 lacked p53 mRNA expression and thus were resistant to MI219. Therefore, substantial combinatorial molecular diversity exists in the p53 gene status in AML with direct effects on the ability of MDM2 inhibitors to effect apoptotic AML blast death.^20,35 

Focusing on the resistant AML cases with wild-type p53 exons 5-9 and presence of p53 mRNA, 2 subsets emerged that displayed either: (1) low or absent p53 proteins or (2) preserved p53 protein levels. Regarding the molecular basis for low p53 protein in subsets of AML blasts, our initial analysis did not identify supportive evidence for reversible epigenetic changes. We are thus left with the possibility of a defective p53 gene (including alterations in the promoter or epigenetic changes that defied our pharmacologic attempts at reversal), impaired p53 mRNA translation, or reduced p53 protein stability that is independent of MDM2 or MDMX levels.^36-42  Given the scarcity of the primary source material, this was not investigated further but raises questions regarding correlations between p53 status and p53 protein levels that should be evaluated in future studies. Regarding the AML blasts with robust induction of p53 protein after MDM2 inhibitor treatment or external irradiation, the 2 principal molecular defects are: (1) a defective p53 protein, possibly because of aberrant posttranslational modifications of p53, resulting in an altered ability of p53 to activate apoptotic signaling pathways^43-45 ; or (2) defects in the p53-regulated apoptotic network. Regarding the latter possibility of defects in the p53 apoptotic network, it is important to note that a quantitative analysis of the relative importance of various p53 inducible genes in relation to the apoptotic response after p53 induction is not available. In the setting of the treatment of cells with the low-molecular-weight compound RITA (but not Nutlin treatment), it has recently been demonstrated that p21 down-regulation provides a switch between p53-induced apoptosis and cell cycle arrest.⁴⁶  Our analysis of p21 protein levels before and after Nutlin or MI219 treatment also does not provide clear evidence for a unique role of p21 in AML blast fate decision after MDM2 inhibitor treatment. The initial analysis of gene expression changes in sensitive versus resistant blasts after treatment with MI219 (not shown) identified differential induction of a subset of classic p53-responsive genes, but interpretation of these data is hampered by the fact that the critical genes with importance to p53-mediated apoptosis in leukemia are not known. It is thus possible that genes other than classic p53-responsive genes are involved in conferring resistance to MDM2 inhibitors, and future analysis of such genes is of importance for a comprehensive understanding of MDM2 inhibitor effects on myeloid leukemia blasts.

One of the novel results from this analysis was the identification of AML blasts that were very sensitive to MI219 ex vivo. Approximately one-third of AML cases displayed IC₅₀ values to MI219 of less than 2μM. Attempts at identification of determinants of such heightened sensitivity uncovered frequent and significant association with mutated Flt3 (presence of Flt3-ITDs). For instance, 58% and 84% of all AML samples with Flt3-ITD (N = 19) displayed MI219 IC₅₀ values of less than 2μM and less than 5μM, respectively. Therefore, activated Flt3 appears to sensitize AML blasts to MDM2 inhibitor-mediated apoptosis, a finding that could be exploited for clinical applications of MDM2 inhibitors in AML.

Regarding the direct clinical implications of these findings for clinical trials with MDM2 inhibitors in AML, we offer the following conclusions: (1) p53 mutations confer absolute resistance to MDM2 inhibitors in AML; therefore, patients with p53 mutations should be excluded from such trials; (2) defective p53-mediated apoptosis is substantially more common in AML than p53 mutations; (3) 17p deletions (spanning p53) as detected through karyotyping in AML (or AML-fluorescence in situ hybridization as inferred from the 17p anatomy described here) capture only a small subset of p53 mutated cases; (4) complex karyotype is a good predictor of underlying p53 mutations but identifies only a subset of the total number of AML cases with primary MDM2 inhibitor resistance; (5) a subset of normal karyotype AML cases are absolutely resistant to MDM2 inhibitors; (6) the presence of FLT3-ITDs identifies a subset of AML for which the majority are very sensitive to MDM2 inhibitors (but many additional sensitive AML cases lack such a predictive molecular marker); and (7) no currently used clinical test is able to identify primary MDM2 inhibitor resistance in AML with high sensitivity or specificity. Based on these observations, one may therefore propose that ex vivo testing of purified AML blasts from patients may offer clinical benefits in the setting of MDM2 inhibitor treatment, and such testing may be incorporated into clinical trial designs.

In conclusion, this unique dataset, based on the analysis of more than 100 primary AML cases, quantitatively describes primary resistance to MDM2 inhibitors in AML and provides evidence for multiple distinct molecular mechanisms. These unexpected findings thus substantially complicate transition of MDM2 inhibitors into AML therapy and provide the rationale for further in-depth preclinical studies of resistance mechanisms in AML. Our studies also provide an additional, albeit incomplete, rationale for combination-targeted therapy in AML with primary resistance to MDM2 inhibitors in an attempt to overcome resistance.⁴⁷  Conversely, our novel finding of an association of mutated Flt3 (Flt-ITD) and heightened sensitivity to MDM2 inhibitors was not expected based on published findings⁴⁸  and may be important to AML therapy as: (1) Flt3-ITD AML cases tend to have short remission durations, (2) therapeutic blockage of Flt3 using Flt3 inhibitor monotherapy has not yet resulted in substantial clinical benefits to patients, and (3) application of MDM2 inhibitors to AML with mutated Flt3 and intact p53 may offer clinical benefits.⁴⁹  This novel description of a genomic biomarker for MDM2 inhibitor sensitivity thus introduces the concept of “MDM2 inhibitor sensitizer gene mutations” and justifies ongoing searches for additional genes with similar effects. Finally, such MDM2 inhibitor sensitizer mutations may ultimately also offer explanations for the heightened sensitivity of neoplastic cells to MDM2 inhibitor treatment.

The authors thank the microarray core of the University of Michigan Comprehensive Cancer Center for services provided.

This work was supported by the National Institutes of Health (1R01 CA136537-01) (S.N.M.) and the Translational Research Program of the Leukemia and Lymphoma Society of America (S.N.M.), in part by the National Institutes of Health through the University of Michigan's Cancer Center (support grant 5 P30 CA46592), and in part through the Michigan Institute for Clinical and Health Research (translational grant UL1RR024986).

National Institutes of Health

Contribution: J.L., P.O., and S.N.M. performed the laboratory research; B.P., H.E., D.B., and S.N.M. enrolled patients and contributed and analyzed clinical data; K.S. assisted with statistical analysis; S.N.M. and S.W. conceived the study and supervised the work; and J.L., P.O., S.W., and S.N.M. wrote the paper.

Conflict-of-interest disclosure: S.W. is a shareholder and founder of and a consultant to Ascenta Therapeutics Inc, and the principal investigator of a research contract between Ascenta Therapeutics Inc and the University of Michigan. The remaining authors declare no competing financial interests.

Correspondence: Sami N. Malek, Department of Internal Medicine, Division of Hematology and Oncology, University of Michigan, 1500 E Medical Center Dr, Ann Arbor, MI 48109-0936; e-mail: smalek@med.umich.edu; and Shaomeng Wang, Department of Internal Medicine, Division of Hematology and Oncology, University of Michigan, 1500 E Medical Center Dr, Ann Arbor, MI 48109-0936; e-mail: shaomeng@med.umich.edu.