The agency’s staffers raised several issues ahead of a Wednesday meeting to review Nplate, which the biotech hopes to market for a disorder that causes the body to attack its own platelets - the red blood cells that help blood to clot. The condition causes bruising and bleeding after minor injuries, and affects about 200,000 people in the US, the Associated Press writes.

But according to briefing documents on the FDA web site, studies showed several adverse reactions, from bone-marrow abnormalities to dangerous blood clots. The reviewers wrote that Nplate patients showed significantly higher platelet levels than those taking placebo in two, six-month studies conducted by Amgen. Patients received weekly injections of Nplate, a genetically engineered version of the protein that encourages red blood cell production, the AP notes.

The FDA noted that 14 of 204 patients had dangerous blood clots after beginning treatment with Nplate, and agency reviewers wrote that the risk of such clots could increase after patients stop taking the drug, though Amgen’s studies have not yet addressed the issue. They also expressed concern that 8 percent of patients taking Nplate developed resistance to the drug. Only 4 percent of patients taking placebo showed the similar results. And the reviewers noted abnormal bone marrow growth and malignant tumors seen in some patients.

UPDATE: In an investor note this afternoon, Rodman & Renshaw’s Michael King writes that the odds of approval are 60 percent to 70 percent in Amgen’s favor.

“…It is important to point out that bleeding events, which are an important clinical outcomes assessment, were not a prospectively defined endpoint in either pivotal Phase III study. The pooled analysis shows trend in favor of Nplate, but not a significant difference between the groups. While this adds to the risk of a negative panel vote, the inverse correlation between bleeding risk and platelet count (platelet count up, bleeding risk down) is one that is widely recognized in our view. Furthermore, we expect Amgen to present additional data…showing that bleeding is significantly reduced relative to placebo when platelet counts are elevated to therapeutic levels with Nplate.”

The agency will ask the advisory committee whether Nplate should be approved on a limited basis. For example, panelists will be asked if the drug should only be approved for patients who have not responded to other drugs, the AP writes. Lehman Bros. analyst Jim Birchenough told the AP that he believes the FDA will eventually approve the drug because there are few options for patients who have failed other treatments, which include removal of the spleen. And he projects peak sales of $500 million for Nplate.

In its application to FDA, Amgen laid out a plan for managing Nplate’s risks through drug-warning labels, educational outreach and follow-up studies. Under its the proposal, Nplate’s label would warn of safety problems seen in company studies, and Amgen would distribute additional guides on the risks to patients. For doctors, Amgen would hold workshops on proper drug use. And the biotech proposed studying irregular bone marrow growth and other adverse reactions in four follow-up studies. Finally, Amgen offered to refrain from advertising the drug directly to consumers.