Just last week, I mentioned that Eli Lilly and Company, as a condition of approval for their blockbuster drug, was required to conduct post-marketing studies of Xigris. Yesterday, Lilly unexpectedly announced the withdrawal Xigris from the market. This was not due to adverse events or side effects, but was the result of exactly such a post-marketing study.

Xigris, known generically as recombinant activated protein C or drotrecogin alfa activated, was approved by the FDA in 2001—by an evenly split vote—and was approved by the European Medicines Agency in 2002 for the treatment of severe sepsis.

Xigris is not an antibiotic. It is an anticoagulant, similar to the body’s own activated protein C. The theory behind many sepsis protocols was that much of the end-organ damage in septic patients was due to microscopic clots, and that the experimental drug could prevent this tissue damage. The specific site of action is still unclear. Curiously, bleeding is seen as a dreaded complication from disseminated intravascular coagulation (DIC) in late sepsis, as proteins needed for normal clotting are consumed. Yet an anticoagulant seemed to be beneficial for treating sepsis. As you might expect, bleeding was the major side effect of Xigris.

Xigris cost approximately $8,000 per treatment course. In a coup for Lilly, the Centers for Medicare and Medicaid Services (CMS) approved Xigris “as the first and only medical product to be granted new technology status for the substantial improvement in treatment it offers Medicare patients with life-threatening severe sepsis,” providing hospitals with reimbursement for the drug. Mathematical modeling showed Xigris was cost-effective—but only for younger patients with very severe sepsis (defined as an APACHE II severity of illness score > 25). However, “The cost effectiveness of treating patients with an APACHE II score of 24 or less increased to $575,054 per life-year gained when the FDA's estimates of effectiveness were considered.” Pretty amazing how you can play with numbers.

Xigris’ use was indicated only for patients with severe sepsis and failure of two organ systems—so patients with an expected mortality of ~40%.

In the first Xigris study, “PROWESS,” a small reduction in mortality was found, particularly in patients with more severe illness. Having a risk for bleeding is a major contraindication to its use. Subsequent studies did not confirm the same magnitude of benefit, prompting the post-marketing study of Xigris in septic shock. Since the “PROWESS-SHOCK” study showed the Xigris did not achieve the expected reduction in 28-day all-cause mortality, Lilly pulled the plug on their flagship drug.


“Sepsis” describes a syndrome that goes along a spectrum of severity:

Sepsis syndrome is quite explicitly defined and includes documented infection plus two signs of an abnormal systemic inflammatory response system, known as SIRS:

  • Temperature >38.5ºC or <35ºC
  • Heart rate >90 beats/min
  • Respiratory rate >20 breaths/min or PaCO2 <32 mmHg
  • WBC >12,000 cells/mm3, <4000 cells/mm3, or >10 percent immature (band) forms

SIRS is seen not only from infection, but commonly from trauma and pancreatitis, as well.

Severe sepsis is worse, and also requires presence of organ dysfunction:

  • Areas of mottled skin
  • Capillary refilling requires three seconds or longer
  • Urine output <0.5 mL/kg for at least one hour, or renal replacement therapy
  • Lactate >2 mmol/L
  • Abrupt change in mental status
  • Abnormal electroencephalographic (EEG) findings
  • Platelet count <100,000 platelets/mL
  • Disseminated intravascular coagulation
  • Acute lung injury or acute respiratory distress syndrome (ARDS)
  • Cardiac dysfunction (i.e., left ventricular systolic dysfunction), as defined by echocardiography or direct measurement of the cardiac index

Further on in the progression is septic shock, which adds inadequate blood pressure despite fluid resuscitation or need for pressors, drugs to support the blood pressure. There is a high predicted mortality for sepsis, with “stepwise increases in mortality rates in the hierarchy from SIRS, sepsis, severe sepsis, and septic shock: 7%, 16%, 20%, and 46%, respectively.

Why care? The face of septic shock:


I participated as my site’s principal investigator on many of the major sepsis protocols between 1995 and 2005—studies for Pfizer, Knoll, Novartis, Chiron, and Lilly.

Each trial had rigorous inclusion and exclusion criteria for patient participation, as above. Additionally, patients had to be identified and started on the investigational medication within 24-48 hours of admission. Because the criteria were so difficult, target enrollment was one-two patients per month per site, or approximately 18–24 months to accrue the requisite number of patients.

And each of these sepsis trials failed to show a statistically demonstrable benefit to the use of the study drug compared to a placebo in addition to standard care. Ironically, it appears that, in part, these failures were because each trial had slow enrollment. It seems that advances in supportive care during the enrollment period resulted in enough improvement in the patients’ conditions to wipe out any evident differences between the active study drug and placebo groups. At my hospital, we started with a 40% mortality in this specific patient population, as expected. Yet our mortality promptly dropped to ~10%--not because any of these drugs worked or even, in some periods, because of other great advances, but because our supportive care improved. (In fact, one study’s medical director said that, had he known before that we had such a low mortality, he would not have included our site in the study, as it would have skewed their data and have been statistically impossible to show a response to their drug). We identified septic patients much earlier on the spectrum, and treated them much more aggressively, preventing end-organ damage, such as renal failure, in many cases. But there were some patients who we all expected to die, yet didn’t, and that we felt was a result of the drug.

Eli Lilly has been faulted—justifiably so, I believe—for its incredibly aggressive and successful marketing campaign and for insinuating itself into practice guidelines. They provided a “$1.8 million grant to form the Values, Ethics, and Rationing in Critical Care (VERICC) Task Force, purportedly to address ethical issues raised by rationing in the intensive care unit. Finally, the Surviving Sepsis Campaign was established, in theory to raise awareness of severe sepsis and generate momentum toward the development of treatment guidelines." The strategy of focusing on rationing led them to partner with bioethecists and academicians, using them as key opinion leaders to generate enthusiasm for their drug.

FDA and Post-Marketing Studies

The FDA has the regulatory authority to require post-marketing studies as a condition of approval for drugs. This is often done for drugs receiving accelerated approval, as drugs for life-threatening indications often do. This also occurs with therapies that are approved based on a surrogate endpoint, as HIV drugs (e.g., rise in CD4 count, reflecting improved immunity) rather than demonstrated survival. According to the Office of Inspector General, from 1990 through 2004 “forty-eight percent of new drug applications approved…involved at least one postmarketing study commitment.” Yet this authority had no teeth, in the FDA had no corresponding ability to penalize the drug company sponsors for failing to complete the agreed to studies. In a rather scathing report in 2006, the OIG noted that even if the sponsors submitted reports, they often went unvalidated by the FDA, which claimed it lacks adequate resources to provide such oversight. In a 2009 draft guidance, the FDA was given the authority to impose modest financial penalties for violations.

Post-marketing drug studies are also critically important for geriatric patients, who are often excluded from Phase 1-3 clinical trials, and so there is little real-world experience in this group until the drug has already received FDA approval.

Post-marketing drug withdrawals are sometimes for adverse events or safety issues; the withdrawal of Xigris for lack of efficacy was rather unusual. Another high profile case is Avastin, whose use both the FDA and EMEA want to restrict, again due to lack of efficacy in improving survival for some cancers.

Sometimes, only dosage recommendations have been altered, as with the case of Interferon alfa-2b, which was originally recommended for a six month course of therapy for hepatitis C. Subsequently, a twelve month course was shown to be more efficacious.

A number of drugs have been removed due to the discovery, post-approval, of serious side effects. As explained in the last post, on phases of trials, not enough patients participate in most studies to detect serious side effects that only occur infrequently. So later, after broad use of the drug, serious problems may be discovered and the drug will then be withdrawn or severely restricted in use. Examples include thalidomide, temafloxacin (Omniflox), and the anti-diabetic drug troglitazone (Rezulin).

Lessons Learned

There are some important lessons here. First post-marketing studies have value and the FDA should be more stringent in enforcing its demands for them as a condition of drug approval. Clinical trials involve only small numbers of patients, often in controlled (rather than real world) conditions; it is important to test the drug’s efficacy and safety in a broader population, and particularly under more real-world conditions (such as elderly patients on multiple medications).

Second, and of grave concern, as some physicians noted, this is one more disincentive to drug companies interested in developing a treatment for sepsis. The trials are enormously costly and labor-intensive. And, as for antibiotics, the successful drug for sepsis will only be used for a very short term, so is not likely as profitable as another me-too “lifestyle” drug or med for a chronic disease.

Thus post-marketing studies form a “damned if you do and damned if you don't” or “Catch-22” scenario. On the one hand, they are necessary to uncover infrequently occurring side effects, but they pose a huge risk for the pharmaceutical sponsor in question and for investment in future studies for serious, but short-lived, conditions.