Rigosertib is a small molecule anti-cancer agent which inhibits the PI-3K and PLK pathways, promotes G2/M arrest, and selectively induces apoptosis in cancer cells. We have previously reported results of a phase I/II study of rigosertib in patients (pts) with MDS and AML who had relapsed or were refractory to hypomethylating agents (HMA), a population for which there are currently no approved second line therapies. Rigosertib appeared to be well tolerated in this pt population and to have biologic activity with reduction or stabilization of bone marrow (BM) blasts and improvement in the peripheral blood (PB) counts in few treated pts. Reduction in BM blasts by rigosertib was associated with increased survival. In the current analysis, we evaluate pt characteristics that may predict for response.


We analyzed the results of a phase I/II study of Rigosertib that was conducted in pts with MDS and AML. In the phase I component, pts were entered in cohorts of escalating doses in a classic 3+3 design ranging from 650 up to 1700 mg/m2/d continuous IV infusion (CIV) administered for 72 to 144 hours. A MTD of 1375 mg/m2 was identified for the phase II component, and subsequent pts were treated with this dose as a CIV for 72 hours. BMs were performed at baseline, week 4, 8, and then q3 months.


Twenty-two pts with MDS or AML refractory or relapsing to a HMA have been treated with rigosertib. The study cohort comprised pts with a diagnosis of int-2 MDS (2), high risk MDS (6), CMMOL (1), and AML (13 pts all with antecedent MDS). Responses according to IWG 2006 criteria were observed in the BM and PB: marrow CR (4), marrow PR (2). Two pts also had hematologic improvement of the erythroid (1) and platelet (1) lineages. Four pts had stable disease (SD) after treatment but their courses were complicated by infections requiring hospitalization and removal from study. Three pts were deemed to be inevaluable because they received < 2 cycles of treatment or did not have a follow-up BM evaluation. Thus, 10/19 evaluable pts (53%) demonstrated either a BM/PB response (6) or SD (4).

The median overall survival (OS) of pts with marrow CR+PR (n=6) was 12 months versus 1.8 months for those without a BM response (n=9) (p=0.0159, log-rank test). Age was not a predictor of response. 1 out of 6 responders had a major elimination in the size of the clonal population. Non-responders did not have a change in the magnitude of the clone. Prior response to HMA was not a predictor for response to rigosertib. Those who did have a marrow CR or PR responded early with median time to response of 2-4 cycles. Those with higher blast counts were less likely to respond. At study entry, the median blast percentage of responders was 16% versus 44% for non-responders. Less than 20% blasts at study entry was a positive predictor of response (p=0.047). Of those pts who did not respond or were inevaluable, the majority (75%) had AML, many with a proliferative course. Nine of 19 pts developed cystitis manifested by dysuria and/or hematuria as a side-effect of therapy. Among responding patients, 5 of 6 had cystitis [grade (GR) 1 (2); GR 2 (1); GR 3 (2)] compared with 3 of 9 non-responders [GR 1(1); GR 2 (1); GR 3(1)] (p=0.08). In one responding pt with grade 3 cystitis, a cystoscopy was performed which revealed polypoid inflammatory changes of the mucosa with hemorrhage. Biopsy showed neutrophilic inflammation without malignant cells. Upon resolution of symptoms, treatment was restarted at 50% of the original dose without complications. Pts who developed symptomatic cystitis were treated with sodium bicarbonate with improvement. The relationship between cystitis and response is being investigated.


Rigosertib appears to have biologic activity with reduction in BM blasts associated with increased survival and improvement in the PB counts in a subset of treated pts. Pts with <20% blasts at study entry had a greater likelihood of response. Pts with proliferative disease with rapidly rising or high wbc did not respond. Age, cytogenetic profiles, and response to prior therapy were not predictors of response. Cystitis may be a response related biomarker and requires further analysis. A phase III multicenter randomized trial is underway to compare rigosertib to best supportive care with a primary endpoint of OS in pts with higher risk MDS who have failed, progressed, or relapsed after treatment with HMA, and can be used to validate the observations reported here in a larger study.


Reddy:Onconova: Equity Ownership, Research Funding. Holland:Onconova: Research Funding. Wilhelm:Onconova Therapeutics: Employment, Equity Ownership. Silverman:Onconova: Research Funding.

Author notes


Asterisk with author names denotes non-ASH members.