Not surprising but hardly well taken by investors was the second setback with Onconova Therapeutics Inc.’s intravenous (I.V.) rigosertib for higher-risk myelodysplastic syndromes (MDS), which failed to meet its primary endpoint of overall survival (OS) in the phase III trial known as ONTIME.

Newtown, Pa.-based Onconova’s dual PI3K/PLK inhibitor also has faltered in pancreatic cancer, and rigosertib is partnered with Baxter International Inc., of Deerfield, Ill., for both indications.

On word of the MDS fizzle, shares of Onconova (NASDAQ:ONTX) closed Thursday at $8.79, down $5.07, or 36.6 percent.

Rigosertib missed the OS goal in MDS patients who had progressed on, failed or relapsed after therapy with hypomethylating agents (HMAs).

The trial enrolled 299 patients, including 199 in the rigosertib-plus-best-supportive-care arm, and 100 in the best-supportive-care-only arm. Median OS was 8.2 months in the former compared to 5.8 months in the latter, but the compound did not turn up a statistically significant survival benefit (HR = 0.86, p = 0.27).

But a posthoc analysis that was not pre-specified showed promise in a subset of patients: those who had progressed on or failed treatment with HMAs – that is, had not responded to HMAs. Here, median OS was 8.5 months with the dual regimen vs. 4.7 months with best supportive care alone (HR = 0.67, p = 0.022).

The subset included 184, or about 62 percent, of the 299 enrolled patients. The remaining 115 were those who had relapsed after responding to previous treatment with HMAs. That subset did not show a statistically significant survival benefit.

“As we understood from the very beginning, we are dealing with a very heterogeneous population of patients,” said CEO Ramesh Kumar during a conference call with investors. “Even the front-line patients are really a mixture of different types of pathologies. The relapsed patients who have had long, sometimes many, cycles of previous HMA therapies are known to be different.” Therefore, contrasting results between one patient group and a subset is “striking, and probably reflects a fundamental difference in these two groups of patients,” he said.

The company “remain[s] committed” to advancing rigosertib, Kumar said, and a phase III trial with the oral version in low-risk MDS patients will begin “as soon as possible.” Talks are under way the FDA.

Low-risk and high-risk MDS are “really different diseases,” he noted. “They are united by one common theme – both types of MDS have cytopenias.” High-risk MDS is “driven by [the presence of] blasts, which are non-existent or very low in the low-risk population.”

Results from a phase II trial with the oral version were reported at last year’s meeting of the American Society of Clinical Oncology (ASCO), showing a 53 percent overall response rate, with 36 percent of patients achieving independence from transfusions.

RBC Capital Markets analyst Glenn Novarro earlier had assigned a probability of less than 50 percent that rigosertib would succeed in the MDS I.V. trial.

Gauging prospects from the point of view of Onconova’s European partner Baxter, Novarro wrote in a research report that the likelihood rigosertib will be approved in the European Union without another trial is less than 10 percent.

“The safety analysis [from the latest study] indicated that rigosertib was generally well tolerated,” wrote Novarro, who had doubts about the OS subset analysis. “Despite the lack of alternatives for high-risk MDS patients who fail HMAs, we believe regulatory agencies are likely to be skeptical of these results as well,” in Novarro’s view.

Onconova is due to offer details on secondary endpoints and more at this year’s ASCO meeting, which starts in late May in Chicago, and an update on the planned phase III trial with oral rigosertib in lower-risk MDS should come then or sooner.

In December, Onconova gave up on the phase II/III trial with rigosertib in pancreatic cancer. The study tested I.V. rigosertib plus gemcitabine in front-line metastatic pancreatic cancer. An interim analysis by the data safety and monitoring board of the ONTRAC experiment determined that the combination was unlikely to demonstrate a statistically significant improvement in OS, compared to gemcitabine alone. (See BioWorld Today, Dec. 19, 2013.)

At the time, Novarro wrote in a research report that expectations for rigosertib against pancreatic cancer had been low, and high-risk MDS was the indication with “the best probability of success” with the compound.

Rigosertib is by far the most advanced of Onconova’s candidates. The pipeline also includes ON 013105, designed to suppress cyclin D1 accumulation in cancer cells. Onconova is testing that compound in a phase I trial against relapsed or refractory lymphomas and acute lymphoblastic leukemia.

Four phase I trials have been completed with recilisib, a small molecule with radiation protection properties that is being developed as a medical countermeasure to treat the effects of acute radiation sickness, specifically cytopenia. A subcutaneous version has undergone three phase I studies, and an oral form has been tested in one phase I experiment.