Increase in platelet count in older, poor-risk patients with acute myeloid leukemia or myelodysplastic syndrome treated with valproic acid and all-trans retinoic acid




The authors investigated the efficacy and safety of the histone deacetylase inhibitors valproic acid (VPA) and all-trans retinoic acid (ATRA) as differentiation agents in a cohort of older, poor-risk patients with acute myeloid leukemia (AML) or myelodysplastic syndrome (MDS).


Twenty older patients with recurrent or refractory AML or MDS were treated in a Phase II protocol with sequential VPA and ATRA therapy. VPA was started at a dose of 10 mg/kg per day and then escalated to achieve the serum concentration of 45–100 μg/mL. ATRA was added at 45 mg/square meters (sm) per day when VPA reached the target serum concentration. Only patients treated continuously for ≥ 2 months were considered evaluable.


Hematologic improvement, according to World Health Organization criteria, was observed in 6 of 20 patients enrolled in the protocol but in 6 of 11 considered evaluable. In five patients, a major platelet response was observed, achieving platelet transfusion independence. Three of these five patients also exhibited a minor erythroid response. A sixth patient showed both a minor erythroid response and a platelet response. The median duration of response was 189 days (range, 63–550 days). No significant reduction in the blast count was observed. Grade 3 neurocortical toxicity was observed in four patients. Severe bone pain was experienced by 4 patients (2 Grade 4 and 2 Grade 3) and was associated with an increase in the peripheral blast cell count. Treatment with ATRA did not modify the response observed with VPA alone.


Differentiation therapy with VPA was of clinical benefit in approximately 30% of elderly patients with AML and MDS of the refractory anemia with excess of blast type with unfavorable prognostic features. A striking platelet transfusion independence lasting several months may be obtained in some patients, reducing the burden of palliative care and improving the quality of life. Cancer 2005;. © 2005 American Cancer Society.

Acute myeloid leukemia (AML) predominantly affects older adults with a median age at diagnosis of 68 years.1 Unfortunately, treatment outcome in these older patients is disappointing, with results remarkably inferior to the results obtained for patients < 60 years. The long-term, disease-free survival rate is < 10% for patients treated with standard chemotherapeutic protocols, but this percentage does not include a large proportion of patients (approximately 30%) for whom age and comorbidities preclude any chemotherapeutic approach. Furthermore, even for patients treated with intensive chemotherapy, the percentage of patients with complete disease remission is approximately one-half (40%) of that observed for patients < 60 years.2 The well known unfavorable biologic and cytogenetic profile of AML in older patients accounts for this unsatisfactory result.3, 4 Thus, less toxic and more specific or targeted therapies are urgently needed for this group of patients, as well as for younger patients with recurrent or refractory disease.

After the success of all-trans retinoic acid (ATRA) in patients with acute promyelocytic leukemia,5 new classes of drugs able to induce differentiation of leukemic blasts have been developed in the last few years. Among these, the histone deacetylase (HDAC) inhibitors induce acetylation of the NH2-terminal tails of core histones, allowing recruitment of transcription factor complexes and leading to derepression of transcription.6, 7 Valproic acid (VPA) is a well known antiepileptic drug that inhibits HDAC at the same serum concentrations that control seizures.8 Particularly, VPA inhibited HDAC1 in HeLa cells with an IC50 of 0.4 mM (57.7 μg/mL), which is well within the therapeutic range for humans.9 In a large series of preclinical studies, VPA induced growth inhibition and differentiation in several neoplastic cell lines.10 Furthermore, leukemic blasts isolated from patients with newly diagnosed AML differentiated in vitro at therapeutic concentrations of VPA.8 In addition, differentiation induced by HDAC inhibitors was strengthened in vitro by ATRA. On the basis of these promising preclinical results, as well as the well established VPA safety profile, we tested the efficacy and feasibility of combination VPA/ATRA therapy in older, poor-risk patients with recurrent/refractory AML and refractory anemia with excess of blasts (RAEB) in a Phase II clinical trial.



A Phase II protocol with sequential VPA and ATRA treatment was undertaken at our institution (Division of Internal Medicine and Hematology, Department of Clinical and Biological Science, University of Turin) and included patients with AML or myelodysplastic syndrome (MDS) (RAEB) who experienced recurrent or refractory disease after standard therapies were administered. The study was conducted in accordance with the Declaration of Helsinki and was approved by our institutional ethics committee. Written informed consent was obtained from all patients. Inclusion criteria were age 60–85 years, histopathologic diagnosis of AML or MDS of the RAEB type, recurrent or resistant disease to standard chemotherapy without matched sibling or unrelated bone marrow donor, and patients not suitable for intensive chemotherapy because of their age, performance status, and comorbidities. Patients with human immunodeficiency virus infection, advanced-stage liver disease, and porphyria or documented allergy to protocol drugs were excluded.

In the treatment protocol, VPA was started at standard dose of 10 mg/kg per day to be administered in divided doses at 8 a.m., 2 p.m., and 8 p.m. Patients were monitored weekly for side effects and the serum VPA concentration was recorded. Serum VPA concentrations were measured with a commercially available latex inhibition assay (Biokit, Barcelona, Spain). On the basis of in vitro activity,8, 9 the optimal serum VPA concentration able to simultaneously induce histone hyperacetylation and avoid most hematologic and nonhematologic toxicities11, 12 was established at 45–100 μg/mL (311–693 μM). This level coincides with the optimal concentration suggested to treat patients with seizures. Retinoic acid at 45 mg/square meters (sm) per day was started when VPA reached target serum concentrations. No dose escalation was performed. In case of an increasing leukocyte count > 10,000 cells per microliter, cytoreductive therapy with low-dose cytosine arabinoside (ara-C) or hydroxyurea (HU) was added. Other medications as well as supportive therapy, including parenteral nutrition and transfusions of red blood cells (RBC) and platelets, were given at the discretion of the treating physician. Whenever possible, therapy was administered on an outpatient basis. During treatment, clinical findings and blood chemistry levels were monitored at least weekly. Toxic effects were graded according to World Health Organization criteria. To be evaluable, patients had to be treated for 2 months continuously at least with VPA. Response to therapy was evaluated according to WHO criteria.13


Bone marrow biopsy specimens were buffered in formalin and decalcified in ethylenediaminetetraacetic acid (EDTA). Paraffin-embedded sections were stained with hematoxylin-eosin and giemsa.

Acetylation studies were carried out on bone marrow sections as follows: 5-μm paraffin sections were rehydrated in xylene and graded alcohols. Endogenous peroxidase activity was blocked with 0.3% peroxide in methanol (15 minutes, room temperature) and washed 3 times with distilled water. Sections were boiled in 10 mmol/L EDTA buffer solution (pH 8.0) for 15 minutes using a microwave heater for antigen retrieval, cooled, and rinsed in phosphate-buffered saline. Purified goat polyclonal antibody raised against a peptide corresponding to an amino acid sequence containing acetylated Lys-9 and Lys-14 of H3 histone of human origin (Santa Cruz Biotechnology, Santa Cruz, CA) was added for 1 hour at room temperature. After washing, the sections were incubated with a rabbit anti-goat antibody for 40 minutes at room temperature (Dako Cytomation, Glostrup, Denmark). After washing, the sections were incubated with an enzyme-labeled polymer preconjugated with anti-rabbit (En Vision Plus System, Dako Cytomation; a biotin-free detection system) at room temperature for 30 minutes. Finally, peroxidase activity was demonstrated with 3-3′diaminobenzidinetetrahydrochoride.

Cytogenetic G or R-banding analysis was performed with standard methods. The criteria used to define chromosome aberration followed the 1995 International System for Human Cytogenetic Nomenclature.

Surface antigen expression was determined by direct fluorescent techniques on whole blood or bone marrow samples using a FACS scan flow cytometer (Becton Dickinson, San Jose, CA) and Cell Quest software (Becton Dickinson). The following panel of antibodies was employed to detect clusters of differentiation of myeloid lineages: CD34 (HPCA2), CD33(P67.6), CD45(2D1), CD13 (L138), CD117 (104D2), CD16(NKP15), and CD11b (D12) (Becton Dickinson).


Twenty patients were enrolled in our protocol: 13 with AML, 2 with chronic myelomonocytic leukemia, and 5 with RAEB2. The median age of the patients at diagnosis was 70 years (range, 63–80 years). Table 1 lists their clinical and biological profile. However, only 11 patients were evaluable (i.e., they were treated continuously for ≥ 2 months). The remaining 9 patients were not able to accomplish the minimum period of treatment required for various reasons and are not fully evaluable (Table 2). However, they were included in the group of patients who failed to respond. Hematologic improvement, according to WHO criteria, was observed in 6 of 20 patients (30%). Of these 6 patients, 5 exhibited a major platelet response with platelet transfusion independence for ≥ 2 months, and 3 of them also experienced a minor erythroid response with a reduction of approximately 50% of the RBC transfusion requirement (Table 3). One patient (RA in Table 3) exhibited both a minor erythroid response and a minor platelet response. Usually, the platelet increase occurred early in the first 2 weeks of VPA treatment and was strictly dependent on attaining the therapeutic VPA serum level (Table 4) with the remarkable exception of patient GA, who achieved and maintained an almost normal platelet count with very low VPA serum concentrations (Table 4). In all patients, the subsequent addition of ATRA did not modify the response observed with VPA alone. A time course diagram of platelet counts in the 6 VPA-responding patients is represented in Figure 1. The median duration of response was 189 days (range, 63–550 days), and the platelet response lasted > 6 months in 3 patients (PMT 18, AG 8, and VPS 9 months respectively; Table 3). The platelet count was quite stable in these patients except during episodes of fever and infection. When these occurred, we observed a decrease in the platelet count associated with a decrease in the serum VPA level below the target concentration of 45 μg/mL. Moreover, in Patients PMT and AG, ATRA was stopped after 13 and 6 months, respectively, without a change in their blood counts.

Table 1. Clinical and Biologic Profile of Patients with MDS/AML at the Start of VPA-ATRA Treatment
PatientsGenderAgeFABCytogeneticLeukocyte count (109/L)Hb level (g/dL)PLT count (109/L)BM blasts (%)Previous therapyECOG status
  1. F: female; M: male; RAEB: refractory anemia with excess of blasts; CMML: chronic myelomonocytic leukemia; MDS: myelodysplastic syndrome; AML: acute myeloid leukemia; noc: not otherwise categorized; VPA: valproic acid; ATRA: all-trans retinoic acid; Hb: hemoglobin; PLT: platelet; BM: bone marrow; ECOG: Eastern Cooperative Oncology Group; FAB: French–American–British classification; G-CSF: granulocyte–colony-stimulating factor; NA: not available; Cy: cytosine arabinoside; E: etoposide; I: idarubicin; D3: daunorubicin for 3 days; Cy7: cytosine arabinoside for 7 days; HU: hydroxyurea; T: thalidomide; CyA: cyclosporine; EPO: erythropoietin; Auto-PBSCT: autologous peripheral blood stem cell transplant; Allo-PBSCT: allogeneic peripheral blood stem cell transplant.

PMTF70M146 XX147010.9900050Cy + E + I1
GLF63M446 XX24807.419,00090None3
AGF72M0No mytosis13207.718,00075D3 + Cy73
GAF72RAEB46 XX123010.635,00015None1
VPSM75M046 XY 5/9 45 X-Y 4/9 mytosis8408.220,000>30None2
CVM73CMML46 XY24,4009.217,00030HU2
CLM70RAEBDel10 5/10 mytosis49208.218,0008CyA, T, EPO, G-CSF2
REM77RAEB46 XY1020917,00010EPO, G-CSF, CyA3
AAM70M5aNA100010.57000>90D3 + Cy73
GGM63AML noc46 XY17408.412,00030D3 + Cy7 + Auto-PBSCT3
RAF71M147 XX +8 4/10 mytosis17308.116,00090D3 + Cy71
MMM70M146 XY30,6009.127,00090Cy + E + I HU4
GVM78AML nocNA30,8008.735,000NANone4
CPM65CMML46 XY 8/10, 47 XY +C 2/10 mytosis21,0008.211,0000.7Cy + E + I Imatinib Allo-PBSCT HU3
ERM63M1NA606011.747,00070D3 + Cy73
BNF63M144 XX, 5q−,−6 17p−,−18 11/18 mytosis 48 XX, 4q+,6q−,17p−+21, +22 7/18 mytosis39707.720,00070%D3 + Cy7 + Auto-PBSCT3
PFF69M2NA21008.639,0006Cy + E + I + Auto-PBSCT2
VTGM63RAEB46 XY 11/15 20q-4/15 mytosis19008.368,0006None2
Table 2. Patients not Evaluable: Reasons for Early Interruption of VPA-ATRA Therapy
PatientsTreatment duration (days)Reason of treatment interruptionOutcome
  1. VPA-ATRA: valproic acid/all-trans retinoic acid; D: deceased; PD: progressive disease; U: unknown.

MM11Neurologic toxicity (Grade 3)D
RI5Neurologic toxicity (Grade 3)D
MM2Neurologic toxicity (Grade 3)D
ER13PD, bone pain (Grade 3)D
PF48Lost at follow-upU
VTG29PD, bone pain (Grade 3)D
Table 3. Hematologic Response to VPA-ATRA Treatment in Evaluable Patients
PatientsLeukocyte peak value (109/L)BM blasts (%)PLTS peak value (109/L)Trasfusional requirement (U/mo)Acetylated H3 histone Blasts (%)Treatment duration (days)Reasons of treatment interruptionOutcomeHI
2 mos before VPA-A treatment2 mos after VPA-A treatment
PLT countRBC countPLT countRBC countBefore VPAAfter VPA
  1. VPA-ATRA: valproic acid/all-trans retinoic acid; NA: not available; ND: not done; PD: progressive disease; D: deceased; L: living; HI: hematologic improvement; HI-E: erythroid response; HI-P: platelet response according to World Health Organization criteria.13; BM: bone marrow; PLT: platelet; RBC: red blood cells.

PMT20,40050103,0004301270550PDDP major E minor
GL472070271,0007603906098PD Bone pain (Grade 4)DP major E minor
AG31,00080102,00033.502.5210238PDDP major
GA107,000> 90259,0000003NDND63PD Bone pain (Grade 4)DP major
VPS139030132,0000.5301NDND287DeathDP major
             E minor
CV23,1005016,0004664.5NDND60No changeDNone
CL38401012,000555.35.3NDND60No changeLNone
RE980NA19,0001.5424NDND60No change Neurologic toxicity (Grade 3)DNone
AA11,800> 9011,000656.56.5NDND60No changeDNone
GG17506023,0000.524.37.7NDND90No changeDNone
RA34505034,000642.523070140PDLP minor E minor
Table 4. PLT Count and VPA Serum Level in Responding Patients
Weeks of therapyPMTVPSGAGLAGRA
PLT count (109/L)VPA (μg/mL)PLT count (109/L)VPA (μg/mL)PLT count (109/L)VPA (μg/mL)PLT count (109/L)VPA (μg/mL)PLT count (109/L)VPA (μg/mL)PLT count (109/L)VPA (μg/mL)
  1. PLT: platelet; VPA: valproic acid.

023,000 20,000 35,000 12,000 17,000 14,000 
1  54,00021.762,00028.7102,00045.624,00073  
2  74,00037    39,000 27,00034.4
329,00064  129,00010.5184,00054.823,000 15,00048
4  92,00029259,000     900064
547,00057  150,00012.2271,00047  29,000 
6        53,000 11,00045
762,00042108,00051190,000 190,0004249,0008029,00041.4
8      234,00049.5  15,00039.8
9    201,00015.2    34,00061
10  132,00050  90,000 46,00074.324,00047.8
1137,00027.7    31,00038.777,000 18,00054.9
12        92,00050.517,00040.5
1332,00035169,00062  101,000   19,00054.7
1440,000     52,000   15,00059.1
15  113,00053    101,000 15,00071.8
16          29,00083.7
1765,0004595,000     102,000 12,00072.2
18  73,000       17,00061.4
1962,00047        16,00045.7
2055,00080.912,00013.8    73,00062.715,000 
30103,0004994,00064.3    29,000   
Figure 1.

Platelet count of responding patients 3 weeks before the start of valproic acid (VPA) and during VPA treatment (purple wide line). Pink line: Patient PMT; dark blue line: Patient VPT; red line: Patient GA; blue line: Patient GL; brown line: Patient AG; green line: Patient RA (see Tables 1 and 3).

The leukocyte count increased usually in all patients responsive to treatment (Tables 1 and 3). However, no significant reduction in the percentage of blasts was observed. Actually, in two patients (GL and GA in Table 3), a dramatic increase in the peripheral blast cell count occurred early after starting VPA therapy at the same time that the platelet count began to increase quickly. In both patients, the sharp increase in the blast and platelet count was associated with generalized bone pain that required major analgesics. At this time, a bone marrow biopsy sample from Patient GL showed focal necrosis of bone trabeculae with extensive invasion of myeloid blast cells. However, the number of megakaryocytes per high power field increased dramatically (0–6) compared with bone biopsy performed before VPA therapy was initiated (Fig. 2). Bone pain was not responsive to the bisphosphonate, zolendronic acid. Patient GA was treated with HU and ara-C, but subsequently, withdrawal of VPA was required for both patients to control symptoms and blast count.

Figure 2.

Bone marrow biopsy sample of Patient GL (see Tables 1 and 3) at (A) diagnosis and (B) after differentiation therapy. Arrows indicate megakaryocytes. (Hematoxylin-eosin stain, original magnification × 20.)

Cytofluorometric studies of myeloid blast cells were carried out in five of six responding patients. In Patient PMT, a new population of blasts expressing CD11b, a later myeloid antigen of differentiation, was observed after VPA therapy as well as a lower expression of CD34 and CD13 and a higher expression of CD33 molecules (Fig. 3). No analysis of megakaryocyte differentiation markers was performed. No evidence of myeloid differentiation was observed in the remaining patients.

Figure 3.

Patient PMT (see Tables 1 and 3). Cytofluorometric assay of a peripheral blood sample after therapy with valproic acid and all-trans retinoic acid. (A) Blasts at diagnosis (red), new population of blasts (green), and differentiated myeloid cells (blue). The new population of blasts shows (B) higher expression of CD33, (C) lower expression of CD13 and CD34, and (D) expression of CD11b molecules, a later antigen of myeloid differentiation.

Immunohistochemical labeling of core bone marrow biopsy specimens with antiacetylated H3 histone antibody revealed, after VPA therapy, a stronger staining (see Table 3 and Fig. 4) associated with a sharp increase in the percentage of positive cells, with the exception of Patient GL who showed a lower percentage of acetylated blasts but with a brighter staining.

Figure 4.

Bone marrow biopsy sample of Patient PMT (see Tables 1 and 3) before (A) and after (B) differentiation therapy. Immunohistochemical labeling of core bone marrow biopsy samples with antiacetylated H3 histone antibody revealed, after valproic acid therapy, a widespread histone acetylation of bone marrow leukemic blasts. Original magnification × 40.

The most common side effect of VPA therapy was neurologic toxicity. Four patients experienced Grade 3 neurocortical toxicity with tremor, hallucinations, disorientation, and somnolence that required cessation of therapy. In some patients (RI and MM, Table 2), severe toxic effects were observed a few days after the start of therapy before attaining target serum concentrations of VPA. In almost all patients, a mild increase, often transient, in the serum concentration of ammonium was observed, and was usually not associated with neurologic or gastrointestinal symptoms. Severe bone pain was experienced by 4 patients (2 with Grade 4 and 2 with Grade 3 disease). In all patients, the bone pain was associated with an increase in the peripheral blast cell count that sometimes required treatment with HU or ara-C and withdrawal from the trial.


The aim of our trial was to evaluate the safety and the efficacy of a differentiation therapy with the HDAC inhibitor VPA associated with ATRA in a selected group of older patients with AML/MDS. These patients either had recurrent disease that was refractory to standard chemotherapy or they had comobidities that precluded intensive chemotherapeutic regimens. Our data indicate that hematologic improvement, according to WHO criteria, occurred in 30% of all patients enrolled in the trial, but in 6 of 11 evaluable patients as defined earlier in the text. The most striking beneficial effect of therapy was the rapid increase in platelet count observed in all responding patients, which was sustained in three of them and allowed platelet transfusion independence that lasted > 6 months. A minor erythroid response occurred in four patients. Immunohistochemical labeling confirmed that hematologic improvement was associated, at the cellular level, with widespread histone acetylation of bone marrow leukemic blasts. However, no change in the percentage of myeloid blasts was found, although surface markers of differentiation and maturation were detected in at least one patient by cytofluorometric studies. Actually, a dramatic increase in the peripheral blood blast cell count was observed in the first 3 weeks of treatment in 2 responding and in 2 unevaluable patients, usually associated with generalized bone pain requiring additional chemotherapy and withdrawal from the trial. In our series, no patient received erythropoietin, granulocyte or granulocyte-macrophage–colony-stimulating factor, or thalidomide during VPA/ATRA treatment or in the previous months. Furthermore, supportive therapies were received more frequently by nonresponding patients (data not shown). Thus, all the hematologic effects we observed had to be attributed to VPA therapy. Because they appeared before ATRA was added to the scheduled treatment and, subsequently, the withdrawal of ATRA did not change hematologic parameters in patients with lasting response, it is not clear what the contribution of ATRA was, if any, to the hematologic improvement observed. Our results are similar to those reported by Kuendgen et al.14 in a group of 18 patients with MDS and AML secondary to MDS. In that study, eight patients responded to VPA monotherapy with hematologic improvement. Conversely, ATRA treatment was not beneficial when given with VPA at the start of therapy. However, it cannot be excluded that ATRA therapy could have contributed to “consolidate” the hematologic improvement achieved with VPA.

An intriguing issue of our trial is the striking difference in response to VPA among our patients. Of 11 evaluable patients, 2 experienced a rapid increase in their platelet and blast cell count in the first 3 weeks of VPA treatment, whereas 4 others exhibited a moderate and gradual increase in their platelet number without a significant change in their peripheral blast cell count and 5 patients did not respond at all. The marked discrepancy observed could mean that the level of histone deacetylation, leading to differentiation block, is somewhat variable among our patients with AML and MDS or, more probably, that leukemic HDACs exhibit variable sensitivity to VPA-induced inhibition.15 Furthermore, all responsive patients became resistant to VPA treatment after several months (range, 7–19 months) despite remarkably stable serum concentrations of the drug, suggesting that a progressive resistance to VPA could indeed occur.

Neurocortical toxicity was the most serious side effect observed in our population of elderly patients with poor performance status. In 4 of 20 patients, such toxicity was severe enough to warrant withdrawal from the trial.

In conclusion, the results of our trial showed that differentiation therapy with VPA is of clinical benefit in approximately 30% of elderly patients with AML and MDS of the RAEB type with unfavorable prognostic features. Lasting platelet transfusion independence may be obtained in some patients, reducing the burden of palliative care and improving the quality of life. We believe that further studies combining VPA with other differentiation drugs16 should be planned to improve disease control in this poor prognostic group of patients.