Eltrombopag With or Without Hypomethylating Agent After Hypomethylating Agent Failure For Patients With Myelodysplastic Syndrome (MDS)

Inclusion Criteria:

1. Signed, informed consent must be obtained prior to any study specific procedures.
2. Subjects with a histologically confirmed diagnosis of MDS by FAB criteria, including both MDS and RAEB-T (AML with 20-30% blasts and multilineage dysplasia) by World Health Organization (WHO) classification are eligible.
3. Patients must have completed at least 4 cycles of hypomethylating agent therapy (e.g azacitidine or decitabine) with failure to achieve at least a partial response, or with the presence of ongoing cytopenias per International Working Group (IWG) (platelet count \< 100x10\^9/L, hemoglobin \<11g/L or Absolute Neutrophil Count (ANC) \<1x10\^9/L). Patients with progressive disease on HMA-therapy prior to this time point are also eligible at the time of documented progression. Therapy with decitabine analogs (i.e. SGI-110) will be considered as decitabine for the purposes of this study.
4. Platelet count \<100x10\^9/L
5. Low, intermediate-1, intermediate-2 or High-risk category by International Prognostic Scoring System (IPSS)
6. Eastern Cooperative Oncology Group (ECOG) performance status of 0 to 2
7. Adequate liver function, as evidenced by a serum bilirubin \</=2x the ULN (except for patients with a confirmed diagnosis of Gilbert's Disease) and an Alanine Aminotransferase (ALT) or Aspartate Aminotransferase (AST) \</=3x the laboratory Upper Limit of Normal (ULN).
8. Serum creatinine \</=2x upper limit of normal
9. Subjects must be\>/= 18 years of age at the time of informed consent, because no dosing or adverse event data are currently available on the use of eltrombopag in children.
10. Subject is practicing an acceptable method of contraception (documented in chart). Female subjects (or female partners of male subject) must either be of non-childbearing potential (hysterectomy, bilateral oophorectomy, bilateral tubal ligation or post-menopausal \> 1 year), or of childbearing potential and use one of the following highly effective methods of contraception (i.e. Pearl index \< 1.0%) from 2 weeks prior to administration of study medication, throughout the study, and 28 days after completion or premature discontinuation from the study: - Complete abstinence from intercourse; - Intrauterine device (IUD); - Two forms of barrier contraception (diaphragm plus spermicide, and for males condom plus spermicide); - Male partner is sterile prior to entry into the study and is the only partner of the female; - Systemic contraceptives (combined or progesterone only).
11. Patients must have recovered from acute toxicity (to grade 1 or less) of all previous therapy prior to enrollment. Treatment may start earlier if necessitated by the patient's medical condition (e.g. progressive disease) following discussion with the Investigator.

Exclusion Criteria:

1. Subjects with any prior exposure to a thrombopoietin-receptor agonist
2. Any prior or co-existing medical condition that in the investigator's judgment will substantially increase the risk associated with the subject's participation in the study
3. Psychiatric disorders or altered mental status precluding understanding of the informed consent process and/or completion of the necessary study procedures
4. Active uncontrolled serious infection or sepsis at study enrollment
5. Clinically significant gastrointestinal disorders that may interfere with absorption of drug.
6. History of arterial thrombosis (i.e. stroke) in the past year
7. History of venous thrombosis currently requiring anti-coagulation therapy
8. Unstable angina, congestive heart failure (New York Heart Association (NYHA) \> Class II), uncontrolled hypertension (diastolic blood pressure \> 100mmHg), or recent (within 1 year) myocardial infarction
9. Subjects with a QTc \> 480 msec (QTc \> 510 msec for subjects with Bundle Branch Block) at baseline
10. Pregnant or breast-feeding because there are no adequate and well-controlled studies of eltrombopag use in pregnancy and it is unknown whether eltrombopag is excreted in human milk.
11. Subjects with known history of human immunodeficiency virus (HIV) or active infection with hepatitis C virus (HCV) or hepatitis B virus (HBV), because eltrombopag is hepatically cleared, and underlying hepatic impairment may lead to an increased risk of hepatotoxicity. Eltrombopag has not been evaluated with combination antiretroviral regimens.