A phase 1 and pharmacodynamic study of decitabine in combination with carboplatin in patients with recurrent, platinum-resistant, epithelial ovarian cancer


  • Fang Fang MD, PhD,

    1. Medical Sciences Program, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
    • The first two authors contributed equally to this article.

  • Curt Balch PhD,

    1. Medical Sciences Program, Indiana University School of Medicine, Indianapolis, Indiana
    2. Melvin and Bren Simon Cancer Center, Indiana University, Bloomington, Indiana
    3. Department of Cellular and Integrative Physiology, Indiana University, Bloomington, Indiana
    Search for more papers by this author
    • The first two authors contributed equally to this article.

  • Jeanne Schilder MD,

    1. Melvin and Bren Simon Cancer Center, Indiana University, Bloomington, Indiana
    2. Department of Obstetrics and Gynecology, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • Timothy Breen MS,

    1. Division of Biostatistics, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • Shu Zhang MD, PhD,

    1. Medical Sciences Program, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • Changyu Shen PhD,

    1. Division of Biostatistics, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • Lang Li PhD,

    1. Melvin and Bren Simon Cancer Center, Indiana University, Bloomington, Indiana
    2. Division of Biostatistics, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • Carol Kulesavage,

    1. Melvin and Bren Simon Cancer Center, Indiana University, Bloomington, Indiana
    Search for more papers by this author
  • Anthony J. Snyder BS,

    1. Molecular and Cellular Biochemistry Department, Indiana University, Bloomington, Indiana
    Search for more papers by this author
  • Kenneth P. Nephew PhD,

    Corresponding author
    1. Medical Sciences Program, Indiana University School of Medicine, Indianapolis, Indiana
    2. Melvin and Bren Simon Cancer Center, Indiana University, Bloomington, Indiana
    3. Department of Cellular and Integrative Physiology, Indiana University, Bloomington, Indiana
    4. Department of Obstetrics and Gynecology, Indiana University School of Medicine, Indianapolis, Indiana
    • Medical Sciences Program, Jordan Hall 302, 1001 East Third Street, Bloomington, IN 47408
    Search for more papers by this author
  • Daniela E. Matei MD

    Corresponding author
    1. Melvin and Bren Simon Cancer Center, Indiana University, Bloomington, Indiana
    2. Department of Obstetrics and Gynecology, Indiana University School of Medicine, Indianapolis, Indiana
    3. Division of Hematology/Oncology, Indiana University School of Medicine, Indianapolis, Indiana
    4. VA Roudebush Hospital, Indianapolis, Indiana
    • Indiana University School of Medicine, Division of Hematology and Oncology, Joseph E. Walther Hall, Room C218D, 980 W. Walnut St, Indianapolis, IN, 46202
    Search for more papers by this author

  • We thank Eisai Co., Ltd. (Tokyo, Japan) for generously providing decitabine and Mrs. Amber Allen, Nancy Menning, and Jessica Roy for coordination of clinical trial activities.



Aberrant DNA methylation is a hallmark of cancer, and DNA methyltransferase inhibitors have demonstrated clinical efficacy in hematologic malignancies. On the basis of preclinical studies indicating that hypomethylating agents can reverse platinum resistance in ovarian cancer cells, the authors conducted a phase 1 trial of low-dose decitabine combined with carboplatin in patients with recurrent, platinum-resistant ovarian cancer.


Decitabine was administered intravenously daily for 5 days, before carboplatin (area under the curve, 5) on Day 8 of a 28-day cycle. By using a standard 3 + 3 dose escalation, decitabine was tested at 2 dose levels: 10 mg/m2 (7 patients) or 20 mg/m2 (3 patients). Peripheral blood mononuclear cells (PBMCs) and plasma collected on Days 1 (pretreatment), 5, 8, and 15 were used to assess global (LINE-1 repetitive element) and gene-specific DNA methylation.


Dose-limiting toxicity (DLT) at the 20-mg/m2 dose was grade 4 neutropenia (2 patients), and no DLTs were observed at 10 mg/m2. The most common toxicities were nausea, allergic reactions, neutropenia, fatigue, anorexia, vomiting, and abdominal pain, the majority being grades 1-2. One complete response was observed, and 3 additional patients had stable disease for ≥6 months. LINE-1 hypomethylation on Days 8 and 15 was detected in DNA from PBMCs. Of 5 ovarian cancer-associated methylated genes, HOXA11 and BRCA1 were demethylated in plasma on Days 8 and 15.


Repetitive low-dose decitabine is tolerated when combined with carboplatin in ovarian cancer patients, and demonstrates biological (ie, DNA-hypomethylating) activity, justifying further testing for clinical efficacy. Cancer 2010. © 2010 American Cancer Society.

Epithelial ovarian cancer (EOC) is the most lethal gynecologic malignancy1 with a 5-year survival rate below 25% for patients diagnosed with stage III-IV.1, 2 Most advanced stage patients respond to cytoreductive surgery and platinum-based chemotherapy; however, >70% of women relapse, and platinum-resistant EOC is uniformly fatal.3, 4

Similar to other malignancies, the progression of ovarian cancer is positively correlated with accumulation of aberrant gene promoter DNA cytosine methylation.5-7 As promoter DNA methylation is intimately associated with transcriptional gene silencing, this epigenetic modification can substitute for genomic lesions (eg, deletions or mutations) in providing the first or second inactivating hit to tumor-suppressor genes (TSGs).8 In EOC, promoter DNA methylation-associated silencing is frequently detected for TSGs such as RASSF1A, BRCA1, DAPK, OPCML, and hSulf-1, and for development-associated transcription factors HOXA10 and HOXA11.7, 9-11 Correspondingly, DNA hypomethylating agents can restore gene expression of TSGs silenced by promoter DNA methylation. Hypomethylating agents have demonstrated clinical activity for the treatment of hematologic malignancies, particularly myelodysplastic syndromes.8, 12-14

Several preclinical studies of ovarian and other cancers have indicated that hypomethylating agents, including the deoxycytidine analog decitabine, can reverse platinum resistance in chemoresistant cancer cell lines and mouse-engrafted tumors.11, 15-18 This chemosensitization is hypothesized to be because of the derepression of TSGs that were previously silenced by promoter DNA methylation, resulting in restored TSG gene expression and the reestablishment of chemotherapy drug response cascades.11, 15, 16, 19 Consequently, few clinical studies have examined combinations of DNA methylation inhibitors, such as decitabine (5-aza-2′-deoxycytidine), with conventional chemotherapies, including platinum-based drugs.20-23 Several preclinical and clinical studies in hematologic cancers have shown that although decitabine (and other DNA methylation inhibitors) is cytotoxic at high doses, lower, nontoxic doses more effectively induce DNA hypomethylation and gene re-expression.12, 13

Because of the difficulty of obtaining tumor samples from cancer patients at relapse, there has been an increasing interest in the development of predictive plasma or serum markers.24 In particular, using a highly sensitive fluorimetric (polymerase chain reaction [PCR]-based) assay,25 specific methylated DNA sequences (believed to be shed by membrane-ruptured tumor cells) can be detected in the plasma of cancer patients.26 These plasma-associated DNA methylation biomarkers correlate well with disease stage and therapy response, representing a promising diagnostic/prognostic tool.9, 26, 27 In ovarian cancer in particular, several plasma-methylated genes have demonstrated high sensitivity and specificity for disease detection and/or prognosis, including BRCA1, RASSF1A, and SFRP.9, 27

In this report, we describe results of a phase 1 trial testing repetitive, low-dose decitabine, in combination with carboplatin, in patients with recurrent, platinum-resistant EOC. We show that this regimen is well tolerated and exhibits biological activity in vivo, as evidenced by global (in peripheral blood cells) and gene-specific (in plasma) DNA hypomethylation, thus establishing a basis for its further investigation in a phase 2 study now underway.


Patient Population

Patients with advanced, histologically documented EOC or primary peritoneal carcinomatosis who progressed during or recurred within 6 months after platinum-based chemotherapy were eligible. Eligibility included both measurable and detectable disease. Measurable disease was defined according to Response Evaluation Criteria in Solid Tumors (RECIST).28 Patients with nonmeasurable disease could enroll if they had clinically or radiologically detectable disease and a pretreatment serum CA 125 level at least twice the upper limit of normal. All patients were at least 18 years old with an Eastern Cooperative Oncology Group performance status of 0 to 1 and life expectancy of at least 3 months. Eligibility criteria allowed an unlimited number of prior cytotoxic therapies, and required adequate hematologic, hepatic, and renal function. Key exclusion criteria included prior history of allergic reaction to carboplatin, history of brain metastases, grade 2 or higher neuropathy, or ongoing uncontrolled medical problems. All patients signed written informed consent, and the protocol was approved by the Indiana University Institutional Review Board.

Treatment Plan

Treatment consisted of escalating doses of decitabine (Eisai, Tokyo, Japan) given as an intravenous (IV) infusion daily for 5 days, and carboplatin (Bristol Meyers Squibb, New York, NY) administered IV on Day 8 at a dose delivering an area under the curve of 5. Decitabine was escalated from 10 mg/m2 (dose level 1) to 20 mg/m2 (dose level 2), using the standard 3 + 3 dose escalation.29 The rationale for this dose schedule was based on previous trials in hematologic cancers demonstrating biological activity in this dose range.12, 13

Dose-limiting toxicity (DLT) was defined as any occurrence during the first 4 weeks of therapy of grade 4 thrombocytopenia, grade 3 or 4 neutropenic fever, grade 4 neutropenia lasting >5 days, grade 3 or 4 nonhematologic toxicity, or failure to recover blood counts by Day 42. Each cycle was 4 weeks, and treatment was continued until disease progression or intolerable toxicity. Growth factor support was not permitted during Cycle 1, and intrapatient dose escalation was not allowed.

Efficacy and Toxicity Assessments

Adverse events were assessed on Day 1 of each cycle and graded according to National Cancer Institute Common Terminology Criteria for Adverse Events version 3.0. When possible, tumor burden was evaluated by clinical examination at baseline and before each cycle. Alternatively, tumor burden was evaluated radiographically at baseline, before each odd cycle, when clinically indicated, and at the end of treatment. Investigator-determined best overall response was defined using RECIST 1.0 criteria for patients with measurable tumors. CA 125 measurements were obtained from all patients on Day 1 of each cycle, and modified Rustin criteria were to be used to assess response for patients with detectable disease.30

Blood and/or Ascites Collection

Samples of whole blood (20 mL) were collected in 2 10-mL ethylenediaminetetraacetic acid tubes. After mixing, samples were centrifuged (∼1500 × g at 4°C for 15 minutes) to separate the plasma from the blood cell layers. Plasma and buffy coats were securely stored at −80°C until use. For the first 2 cycles, blood samples were collected at baseline (pretreatment), on Day 5 (after the last day of decitabine infusion), on Day 8 (before carboplatin infusion), and on Day 15 (to assess DNA methylation). To examine time-dependent DNA demethylation, methylation levels of blood collected on Days 5, 8, and 15 were compared with baseline levels (Day 1). When available and accessible, ascites fluid or tumor material obtained through core biopsies under radiographic guidance were collected on Days 1 (pretreatment) and 8 (before carboplatin infusion).

DNA Extraction and Methylation Analysis

DNA was extracted from 200 μL of buffy coat, 1 mL of plasma, or 200 μL ascites using a QIAmp DNA Blood Mini Kit (Qiagen, Valencia, Calif) according to the manufacturer's instructions, with the column elution repassaged through the column 4× to increase DNA yield. Universal methylated human reference DNA was purchased from Zymo Research (Orange, Calif). Sodium bisulfite conversion of genomic DNA and cleanup were performed using an EZ DNA Methylation kit (Zymo Research), according to the manufacturer's instructions. After bisulfite conversion (unmethylated cytosine deamination to uracil), the treated DNA was analyzed by MethyLight, a highly sensitive, quantitative methylation-specific PCR technique,25 using the primers and probes listed in Table 1. In addition to the primers and probe sets designed for each gene of interest, triplicate assessments of an internal reference PCR product (using primers and probes to a consensus CpG-devoid region of an ALU repetitive element, ALU-C4) were used to normalize for input DNA.25 DNA methylation levels of LINE-1 repetitive elements were expressed as percentages of fully methylated reference molecules, determined by dividing the GENE:ALU-C4 ratio (ie, using ALU-C4 as the reference product) of the experimental sample by the GENE:ALU-C4 ratio of universally methylated human DNA (Zymo Research), and multiplying by 100.25, 31

Table 1. Repetitive Element Primers and TaqMan Probe Sequences Used in the MethyLight Analyses
GeneForward Primer (5′ to 3′)Reverse Primer (5′ to 3′)TaqMan Fluorescent Probe (5′ to 3′)

Statistical analysis

This was an open-label, single-center phase 1 study conducted to determine the safety and the biologic activity of the decitabine/carboplatin combination. The primary objectives were to determine the safety and tolerability of decitabine administered IV daily for 5 days before carboplatin. The secondary objective was to measure the biologic activity by evaluating hypomethylation of target sequences. Demographic and baseline characteristics were summarized using medians (with ranges) for continuous variables, and counts and proportions for categorical variables. Toxicity of the regimen was assessed by the frequency/percentage and severity of reported AEs. To account for the correlation of measurements from the same subject, levels of global or gene-specific methylation were analyzed by linear random-effects models. Factors included in the model were days, dose level, cycle number, and subject. If these factors were statistically significant, we included 2-way interaction terms. Marginal effects were estimated by averaging over other factors. Correlations between different methylation levels were evaluated by Pearson correlation coefficient. A P value of .05 was considered statistically significant.



Ten consenting patients were enrolled in cohorts of 3, with 9 patients completing at least 1 cycle of therapy. Two dose levels were used: dose level 1 (10 mg/m2) and 2 (20 mg/m2), and the escalation followed a standard 3 + 3 design.29 One patient enrolled at dose level 1 deteriorated rapidly because of progressive disease and received only part of the first treatment cycle. Therefore, this patient was only partly evaluable for assessment of toxicity and was replaced. As 2 DLTs were recorded among 3 patients treated at dose level 2, an expanded cohort of 3 additional patients was open for dose level 1, to assess tolerability of the regimen, before proceeding to the phase 2 trial.

Table 2 indicates that all patients had measurable and platinum-resistant recurrent EOC or primary peritoneal carcinomatosis. The median age was 62.5 years (range, 51-71 years). Nine (90%) patients had EOC, and 1 (10%) had primary peritoneal carcinomatosis. Serous papillary carcinoma was the most common histology (9 [90%] patients). This was a heavily pretreated group of patients who had received a median of 5 prior regimens (range, 2-9).

Table 2. Demographics of the Study Patient Cohort
Patients EnrolledCharacteristics
  1. ECOG indicates Eastern Cooperative Oncology Group; PS, performance status; RECIST, Response Evaluation Criteria in Solid Tumors.

Age, y 
Race, white10
Primary site of tumor 
 Primary peritoneal1
Histological subtype 
 Serous papillary9
Stage at diagnosis 
Number of prior therapies 
Platinum sensitivity 
Measurable disease (RECIST)10
Detectable disease0

Treatment administration and safety

Among all patients, 38 cycles were administered. Seven patients were treated at dose level 1, including an expansion cohort, and 3 patients at dose level 2. The median number of cycles completed was 2.5 (range, 1-9+). Causes for treatment discontinuation were: progressive disease (3 patients), withdrawal of consent (1 patient), death (1 patient died of gastrointestinal obstruction while on treatment; the attribution to treatment was considered highly unlikely), and other causes (1 patient because of events unrelated to treatment). Two patients are still receiving treatment.

Table 3 lists treatment-related adverse events (AEs). The most common AEs were nausea (80%), allergic reactions (60%), neutropenia (70%), fatigue (50%), anorexia (50%), vomiting (40%), and abdominal pain (40%), the majority being grades 1-2. Grade 3-4 toxicities affecting >1 patient included neutropenia (n = 4) and carboplatin allergic reaction (n = 3). At dose level 2, 2 of the 3 enrolled patients experienced grade 4 neutropenia with fever, events considered DLT. There were no DLTs at dose level 1. No treatment-related deaths were recorded.

Table 3. Toxicities
Toxicity, CTCAE v3Grade 3, No.Grade 4, No.Decitabine Dose LevelAll Grades, No. (%)
  • CTCAE indicates Common Terminology Criteria for Adverse Events; GI, gastrointestinal.

  • The table lists adverse events occurring during any cycle in >1 patient, the highest-grade toxicity per patient being reported.

  • a

    Dose-limiting toxicity.

  • b

    Toxicity occurring during Cycle 1, unrelated to treatment.

Nausea120 mg8 (80%)
Allergic reaction310 mg6 (60%)
Neutropenia3a1a10 mg & 20 mg7 (70%)
Fatigue1b10 mg5 (50%)
Anorexia120 mg5 (50%)
Vomiting120 mg5 (50%)
Abdominal pain120 mg4 (40%)
Headache1a20 mg2 (20%)
Constipation10 mg2 (20%)
Rash2 (20%)
Fever220 mg3 (30%)
GI obstruction2b10 mg & 20 mg3 (30%)
Back pain20 mg1 (10%)


Efficacy analysis had only an exploratory intent in this part of the study. All patients had measurable disease and were assessed by RECIST. One (10%) RECIST-defined complete response was observed, and 6 (60%) patients had stable disease as their best response. Interestingly, the patient with complete response initially had stable disease, and developed a partial response after Cycle 6 and a complete response at Cycle 8. At 6 months, 4 (40%) patients were without disease progression. With a median follow-up of 8.5 months, 7 patients are alive, of whom 2 are without evidence of progression and continuing on protocol. A waterfall plot of CA 125 levels for patients enrolled in this study is shown in Figure 1. Five patients experienced disease progression during this period.

Figure 1.

A waterfall plot depicts maximal CA 125 level decrease during treatment expressed as percentage of baseline level. CA 125 levels were obtained before each cycle of therapy. Data are available for 8 of 10 enrolled patients. One patient was deceased before completing Cycle 1, and 1 patient did not return for follow-up after Cycle 1.

Pharmacodynamic activity of decitabine, as assessed by DNA hypomethylation

To evaluate the biological activity of the decitabine-based regimen, we assessed its hypomethylating activity in vivo. As a surrogate marker for decitabine biological activity, we assessed post-treatment changes in global (ie, genome-wide) DNA methylation levels in peripheral blood mononuclear cells (PBMCs), using a highly sensitive and fully quantitative methylation-specific PCR assay, MethyLight.25 DNA was isolated from buffy coats and sodium bisulfite-treated to deaminate unmethylated cytosine to uracil. Global DNA methylation levels were assessed by MethyLight assay of LINE-1 (long-interspersed) repetitive elements, which accurately estimates total genomic DNA methylation.25 As shown in Figure 2 (A, B), LINE-1 methylation levels of all patients analyzed demonstrated reduced global methylation on Days 8 and 15, as compared with Day 1. In particular, methylation levels on Day 8 were significantly lower than on Day 1 for Cycle 2, as compared with Day 8 of Cycle 1 (P = .04). The reduction pattern was different (P = .005) between the 2 cycles, partially because of the increase of methylation level at Day 5 for Cycle 1 (P = .04) that was not observed for Cycle 2. No dose effect was observed (P = .36, data not shown).

Figure 2.

In vivo biological activity of 10 mg/m2 decitabine (dose level 1) is shown in peripheral blood mononuclear cells (PBMCs) and plasma collected from patients on Days 1, 5, 8, and 15 of each cycle (averaged over 3 replicates). (A) Methylation levels (expressed as percentage of methylated reference [PMR], percentage of fully methylated reference molecules, normalized by ALU-C4) of LINE-1 repetitive elements are shown in patient PBMCs during Cycle 1. Seven specimens were analyzed from all patients enrolled at dose level 1. (B) PBMC LINE-1 methylation levels are shown for Cycle 2. Six specimens were analyzed from patients enrolled at dose level 1, as 1 patient was discontinued before completing Cycle 1. (C) Methylation of the tumor suppressor gene BRCA1 in patients' plasma during Cycle 1 is shown. (D) Plasma BRCA1 methylation levels for Cycle 2 are shown. (E) A scatter plot shows gene-specific (plasma HOXA11, y axis) and global (PBMC LINE-1, x axis) DNA methylation levels (normalized by ALU-C4) for all subjects. Rho is the Pearson correlation coefficient estimate. For 1A-D, each symbol represents the same patient.

Second, to assess possible tumor response-associated changes in DNA methylation, we performed MethyLight analysis25 on methylated DNA isolated from plasma previously hypothesized to be shed by lysed tumor cells.26, 32 Consequently, we examined 5 genes demonstrated as highly or moderately frequently methylated in ovarian cancer,7, 9, 11 including the tumor suppressors BRCA1, RASSF1A, and WWOX, and the development-associated homeobox genes HOXA10 and HOXA11, recently shown to be highly discriminatory between ovarian cancer and normal tissue.10 Similar to global (repetitive element) DNA methylation levels, demethylation of BRCA1 was more pronounced on Days 8 and 15 compared with Day 1 (Fig. 2C, D) and in patients treated at dose level 2 (data not shown). Of a total of 5 ovarian cancer-associated methylated genes (BRCA1, RASSFA1, WWOX, HOXA10, and HOXA11),7, 9-11HOXA11 was found to be significantly (P < .001) demethylated on Day 15, as compared with Day 1 (not shown), and demonstrated a trend toward demethylation at Day 5 (P = .07). We noted that of the 5 genes, plasma HOXA11 methylation level significantly and strongly correlated with LINE-1 DNA methylation in PBMCs (Fig 2E).

In addition to the composite (ie, multiple patient) methylation responses shown in Figure 2, we also observed, on an individual level, correlations in trends of hypomethylation between plasma and PBMCs. As shown in Figure 3, Patient 1 demonstrated similar decreases in methylation of LINE-1 (PBMCs) and BRCA1 and RASSF1A (plasma).

Figure 3.

Changes in DNA methylation levels from blood collected from Patient 1 during the first 2 treatment cycles are shown: (A) percentage of fully methylated reference (PMR) values for peripheral blood mononuclear cell (PBMC) LINE-1 repetitive elements; (B) PMR values of plasma BRCA1; and (C) PMR values for RASSF1A.

Similar analyses were performed in ascites fluid or tumor biopsies obtained from 2 patients. Ascites was collected from 1 patient (Patient 3). Figure 4 shows a high degree of both global (LINE-1) and gene-specific (BRCA1, RASSF1A, WWOX, and HOXA10) (Fig. 4D) hypomethylation in ascites fluid, as compared with baseline, with similar trends as observed in PBMC and plasma DNA (Fig. 4A-C). Decitabine activity was also measured directly in tumor tissue obtained by core biopsies on Day 1 and Day 8 from 1 patient (Patient 5). We observed LINE-1 and gene-specific hypomethylation (HOXA10, HOXA11, and WWOX) occurring in response to decitabine in tumor tissue (Fig. 5D). Correspondingly, we observed demethylation of HOXA10 and HOXA11 (but not WWOX) after decitabine treatment in this patient's plasma, suggesting at least a partial correlation between methylation levels in tumor and peripheral blood (Fig. 5A-C).

Figure 4.

Changes in DNA methylation levels (expressed as percentage of fully methylated reference molecules [PMR]) from blood and ascites fluid collected from Patient 3 during the first 2 treatment cycles are shown, including PMR values of: (A) peripheral blood mononuclear cell (PBMC) LINE-1 elements; (B) plasma BRCA1; (C) plasma WWOX; and (D) LINE-1, BRCA1, RASSF1A, WWOX, and HOXA10 in DNA isolated from Patient 3's ascites fluid.

Figure 5.

Changes in DNA methylation levels (ie, percentage of methylated reference [PMR] values) from blood and tumor biopsy tissue collected from Patient 5 are shown, including PMR values of: (A) plasma HOXA10; (B) plasma HOXA11; (C) plasma WWOX; and (D) LINE-1, WWOX, HOXA10, and HOXA11 in DNA isolated from biopsy tissue obtained from Patient 5.


This phase 1 trial established the maximum tolerated and biologically active dose of decitabine, given at a low dose daily for 5 consecutive days before carboplatin in patients with recurrent platinum-resistant EOC. On the basis of preclinical studies in various malignancies, including ovarian cancer,15, 16, 18, 33 we hypothesized that administering decitabine could resensitize chemoresistant tumors to carboplatin. One complete response was recorded, and 4 of 10 enrolled patients were free of disease progression at 6 months, indicating clinical activity in platinum-resistant EOC and supporting further investigation in a phase 2 setting.

Toxicities observed in this trial were mild, consisting mostly of nausea, fatigue, and neutropenia. Grade 4 neutropenia in 2 patients represented the DLT at the 20-mg/m2 dose level. More pronounced rates of myelosuppression were recorded in previous trials examining decitabine-based combinations in lung cancer21 and cervical cancer22 in regimens using bolus decitabine administration. In a phase 1 trial, carboplatin was given with decitabine in patients with solid tumors.23 The schedule of administration was decitabine given as a 6-hour infusion on Day 1 and carboplatin given as an IV bolus on Day 8. The DLT was myelosuppression, and the maximum tolerated dose of decitabine was 90 mg/m2. However, further exploration of this regimen in a phase 2 trial revealed a high incidence of neutropenia, leading to decitabine dose de-escalation to 45 mg/m2 and subsequent termination of the trial.34 In contrast, in our trial, administration of low repetitive doses of decitabine was well tolerated.

Interestingly, we observed a high incidence (60%) of carboplatin hypersensitivity reactions. All allergic events responded to administration of steroids and antihistamines, and subsequent treatment with carboplatin was possible using slower rates of infusion and steroid premedication.35 Given the small number of patients, it is difficult to conclude that decitabine increases the risk of hypersensitivity to carboplatin; however, the rate of allergic reactions observed here appeared higher than previously reported.36, 37

Both dose levels tested were biologically active, as assessed by DNA hypomethylation in PBMCs and plasma, as well as in tumor material and ascites collected from 2 patients (Figs. 4 and 5). In PBMCs, a significant reduction in global methylation (LINE-1 repetitive elements) was observed on Days 8 and 15 (Fig. 2). The reduction pattern was slightly different between the 2 cycles, with increased hypomethylation on Days 8 and 15 of Cycle 2 compared with Cycle 1, consistent with previous studies demonstrating the necessity of multiple treatment cycles.12, 13

In addition to DNA from PBMCs, we also assessed methylation levels of plasma free DNA. Such methylated DNA biomarkers have demonstrated high diagnostic specificity and sensitivity in prostate cancer,38 prognostic utility in colon cancer,39 and predictive value in melanoma.26 It has been hypothesized that circulating free DNA originates from necrotic tumor cells, and paired tumor-plasma samples have demonstrated a moderate degree of correlation.32, 40 Of 5 ovarian cancer-linked methylated genes examined (BRCA1, RASSF1, WWOX, HOXA10, and HOXA11), HOXA11 and BRCA1 were significantly demethylated on Days 8 and 15, as compared with Day 1 (not shown), and BRCA1 hypomethylation was more pronounced during Cycle 2 than during Cycle 1. Significant demethylation of RASSF1 was recorded in 8 of 9 patients (data not shown). Moreover, hypomethylation of both global DNA and gene-specific DNA from selected patients' ascites and tumor biopsies exhibited trends similar to those seen in PBMCs and plasma (Figs. 4 and 5).

These results demonstrate that low-dose decitabine is biologically active up to 10 days after its administration, and DNA hypomethylation can be assessed in PBMCs and plasma. Although in past trials, DNA hypomethylation has not been directly correlated with disease response,12, 13 a recent study of chronic myelomonocytic leukemia suggests that hypomethylation may actually precede response, as measured by tumor cell clearance.41 Consequently, it is plausible that antineoplastic activity may follow decitabine-mediated DNA hypomethylation in EOC, and that these plasma-methylated genes represent candidate epigenetic biomarkers for the bioactivity of this agent. The only patient who experienced a complete response during this study developed unique and substantial demethylation of LINE1, BRCA1, and RASSFI from Cycle 1 to Cycle 2. Although this observation needs confirmation in other patients, perhaps during the phase 2 study, it points to the possibility that the degree of DNA hypomethylation correlates with clinical response.

To address our hypothesis of decitabine-mediated platinum sensitization, we examined plasma methylation of hMLH1, a specific gene encoding a mismatch repair enzyme involved in the apoptotic response to platinum-induced DNA damage. However, hMLH1 methylation was not detected in plasma (data not shown), in accordance with results of other studies demonstrating little or no hMLH1 methylation in ovarian carcinomas.42-44 For instance, in 1 study only 7.7% of stage IV ovarian tumors possessed MLH1 methylation,42 and loss of MLH1 was recorded only rarely in recurrent ovarian carcinomas.44

Although the use of DNA hypomethylating agents for platinum resensitization is well supported by preclinical studies,15, 16, 33 we are aware of only 1 recent and 2 earlier trials in solid tumors. In the more recent study, decitabine was administered as a bolus on Day 1, before carboplatin. This regimen induced hypomethylation in vivo, as assessed by fetal globin re-expression and demethylation of the MAGE1A gene promoter in PBMCs, buccal cells, and tumor biopsies.23 At the 90-mg/m2 dose, 1 patient experienced a partial response (after 8 cycles), and 1 had stable disease lasting over 6 cycles.23 A commentary of that trial speculated that the decitabine dosing could be optimized, based on the reproducible success of repetitive low-dose decitabine administration in hematologic malignancies,45 which is accomplished in the trial presented here. Two other trials tested the decitabine/cisplatin combination in nonsmall cell lung cancer21 and in advanced cervical cancer.22 Both trials used bolus decitabine administration for 3 days immediately preceding cisplatin infusion. No responses were observed in the lung cancer study, and clinical activity observed in the cervical cancer study could be attributed solely to cisplatin, as patients enrolled were chemotherapy naive. As decitabine-induced hypomethylation requires several cell divisions before the activation of chemotherapy response genes,14, 46, 47 we speculate that a time delay between the administration of decitabine and of the chemotherapeutic agent might have resulted in greater tumor cell DNA hypomethylation and consequently enhanced antitumor efficacy. The design of the regimen tested in the current study was based on these prior observations.

As platinum is the most active agent used against ovarian cancer, a prognosis of end-stage disease is almost always associated with platinum-resistant tumor relapse.1 Chemotherapy resistance in general, including resistance to platinum drugs, is believed to be associated with several cellular characteristics that permit survival from therapeutic insults.11, 19, 48, 49 The recent, prospective identification of tumor progenitors (cancer stem or cancer-initiating cells) has been reported for several solid tumors, including EOC.50, 51 These progenitors are hypothesized to be largely (or entirely) responsible for chemoresistant tumor relapse, because of the finding that normal stem cells possess many of the phenotypes associated with drug resistance (eg, enhanced DNA repair, diminished apoptotic responses).48, 51 Moreover similarly to normal embryonic or tissue stem cells, cancer stem cells are believed to harbor a significantly altered epigenome, and it has been hypothesized that DNA hypomethylating agents could reset these cells toward differentiation.8 Indeed, several hypomethylating agents were originally characterized as inducers of cancer cell differentiation,47 and 2 of these (including decitabine) are now approved for the therapy of myelodysplastic syndrome and chronic myelomonocytic leukemia.52 It therefore seems reasonable to speculate that hypomethylating agents, in combination with chemotherapeutics, may target drug-resistant EOC initiating cells, possibly leading to tumor eradication.

Based on the above results, we conclude that low-dose decitabine can be safely and effectively combined with carboplatin in ovarian cancer patients and demonstrates biological activity in vivo, as assessed by decreased methylation of genome-wide repetitive elements and of ovarian cancer-associated genes. The 1 complete response and the prolonged disease stability in 3 additional patients are very encouraging, especially considering that this is a selected group of patients with highly chemorefractory tumors. In our now-ongoing phase 2 study, we will determine the clinical efficacy of this regimen in patients with platinum-resistant or -refractory EOC, the primary study endpoint being response rate. Importantly, in that study, clinical benefit will be correlated with DNA methylation changes in PBMC, plasma, and tumor tissue (gene-specific methylation). In summary, based on the results of this work, we believe that hypomethylating agents, in combination with conventional or targeted therapies, represent promising means of reversing chemotherapy resistance in EOC patients.


This work was funded by National Cancer Institute Awards CA133877-01 (to D.E.M.), CA085289 and CA113001 (to K.P.N.), and awards from the Phi Beta Psi Sorority (Brownsburg, Ind) (to K.P.N.) and Ovar'coming Together (Indianapolis, Ind) (to C.B.).