June 2002
Cover Story
Karen Titus
Managing patients with advanced sepsis has just gotten a little bit easier. That’s the good news. The bad news? The task remains a little on the order of skirting Niagara Falls in a canoe.
Physicians now have a drug—Eli Lilly’s Xigris, or drotrecogin alfa (activated), also known as recombinant activated protein C—that has been shown to significantly reduce mortality in patients with severe sepsis. The drug, approved by the FDA last November, showed an absolute reduction in the risk of death of 6.1 percent and a reduction in the relative risk of death of 19.4 percent in a multicenter clinical trial involving 1,690 randomized patients (Bernard GR, et al. N Engl J Med. 2001;344:699-709).
The enthusiasm generated by the PROWESS study comes as no surprise. Rick Dew, RPh, a pharmacist at Massachusetts General Hospital, Boston, has seen more than his share of failed sepsis agents over the years. "Historically, none of the other trials have worked. Over 30 of them have failed. This is the first one to have shown a benefit and made it onto the market. So it’s very exciting, after participating in so many negative trials, to finally have something that works."
The real surprises lie elsewhere, in the largely unexpected links between sepsis and inflammation, coagulation, and fibrinolysis. Laboratories and clinicians must now take on a tangle of delicate decisions as they try to puzzle out who will benefit from APC—and, just as important, who won’t.
For starters, Xigris appears to be associated with an increased risk of bleeding.
"People with severe sepsis who have a profound coagulopathy are at a higher risk of bleeding, with or without Xigris. They’re also at a higher risk of mortality," says John Brandt, MD, medical advisor at Lilly and a member of the CAP’s Coagulation Resource Committee. "So you’re caught in a little bit of a dilemma in that people with a very marked coagulopathy are perhaps the ones who need the drug most." Retrospective analysis of the data from the PROWESS trial suggests that patients with more severe coagulopathy, if it is due to the sepsis, may have a positive risk benefit.
Moreover, some patients, such as those with cirrhosis, may be unable to upregulate the production of clotting factors and may not be candidates for Xigris. "The risk of bleeding must be weighed against the benefits of Xigris in each patient," Dew says.
In the clinical trial, notes Dew, a subset of less-sick patients, as determined by APACHE II scores, actually had three percent higher mortality when given the drug. "So the fear is that somebody who is very clearly septic, but not septic enough, if you will, might actually be harmed"—a concern that will be addressed in another large clinical trial, Dr. Brandt reports.
Some septic patients actually get better without treatment. "You don’t want to give them the drug and expose them to the risk of bleeding," Dew says. Then there’s the expense of the treatment—about $7,000.
[Interestingly, a front-page article in the Wall Street Journal May 17 addressed the costliness of Xigris. The story, part of an ongoing series on drug prices, noted that several studies, though far from conclusive, have hinted that cheap, common steroids may also halt sepsis, by normalizing the inflammatory process.
[One article, published in Chest (Meduri GU, et al. 1998;114:355b-360b), proposed that low doses of glucocorticoid might be effective. Another, published in JAMA (Meduri GU, et al. 1998;280:159-165), looked at prolonged administration of methylprednisolone in patients with acute respiratory distress syndrome, which can result from sepsis. It should be noted that none of the studies have numbers anywhere similar to those in the PROWESS study; the JAMA study, for example, had only 24 patients.]
None of which diminishes the cheers that have greeted Xigris. Quite simply, it saves lives. And, as Michael Laposata, MD, PhD, director of clinical labs at Mass General, puts it: "Who would have thought that regulating coagulation in a septic patient could be a life-saving maneuver?"
Consider the approval of APC to be the alpha, not the omega, of understanding the pathophysiology of sepsis. No one knows for sure why Xigris is effective, and the theories put forth are rigorously debated. "We know that it has anticoagulant properties," Dr. Brandt says. "We believe it has significant anti-inflammatory properties, and that it probably restores the fibrinolytic system toward its normal state."
Are all three pathways important? "There are some of us who think so," says Dr. Brandt. Though some suggest that Xigris’ anticoagulant properties are the linchpin, Dr. Brandt hesitates.
"My personal opinion is that there must be something else additional going on," he says. His reasoning: Heparin has been tried in severely septic patients, particularly those with a coagulopathy, with no effect on mortality. Granted, says Dr. Brandt, those clinical trials took place in the 1970s and were not all that sophisticated; they might not pass muster with today’s more rigorous trial standards. "But there’s no suggestion that heparin as an anticoagulant had an effect on mortality."
Tissue factor pathway inhibitor and antithrombin can also shut down coagulation, but they too have failed to lower death risk in large phase III clinical trials. So while the anticoagulant properties of Xigris are important, particularly in controlling thrombosis in the smaller blood vessels and microvasculature, "I think it [the drug] must be doing more than just that," Dr. Brandt says. "That’s why I think the anti-inflammatory properties are of potential interest."
Until APC came along, the link between coagulation and inflammation had not been seen as intense. "That was a surprise," says Dr. Laposata, former vice chair of the Coagulation Resource Committee. "We knew that minimizing inflammation was important; we didn’t know one of the ways to do it was to give an anticoagulant drug."
In the past, the search for a sepsis agent had taken a different tack, focusing on cytokines and other mediators of inflammation to either eliminate the underlying organism or minimize the inflammatory effects produced by the infection. "If anybody guessed what you really need for sepsis, it would be something that kills the organism, right? Or some major inflammatory drug," says Dr. Laposata.
Dew underscores that point, recalling the many years researchers spent trying to manipulate the balance between pro- and anti-inflammatory responses. One early hope was that blocking the endotoxin component of gram-negative bacteria would shut down the proinflammatory response. It didn’t.
Another approach was to look at two proinflammatory cytokines, interleukin-1 and tumor necrosis factor, which are released by gram-positive bacteria. "We figured if we gave patients antagonists to that, it would shut the whole thing down," says Dew. This approach didn’t work either.
Researchers then turned their attention to the endothelial lining and its role in transporting white blood cells. As it turned out, upregulating tissue factor—which is not expressed on normal endothelium—activates the coagulation pathway.
Most researchers were at first skeptical of proposals to replace coagulation proteins. The original hypothesis was that activated protein C would shut down the tissue factor-activated extrinsic coagulation pathway, eliminating the fibrin clots responsible for organ failure in severe sepsis cases. As studies continued, researchers discovered that activated protein C also works to inhibit inflammation as well as the coagulation pathway.
The latter is accomplished as APC binds to and shuts down factors Va and VIIIa. Other factors that have failed in recent sepsis trials, such as tissue factor pathway inhibitor and antithrombin, inhibit coagulation, but at other points in the pathway. "So there might be something special about inactivating factors Va and VIIIa that makes this drug work where others have failed," Dr. Laposata proposes.
In addition, APC appears to enhance fibrinolysis (which is suppressed in sepsis) by neutralizing plasminogen activator inhibitor and by accelerating tissue plasminogen activator-dependent clot lysis.
These emerging links between coagulation, inflammation, and fibrinolysis have launched new rounds of questions. Does inflammation start the coagulation cascade, or vice versa? How do the three pathways fit together? Moreover, the fibrinolysis genes in particular are becoming targets for genomics and proteomics research aimed at sorting out whether a genetic predisposition toward sepsis exists.
As if all this weren’t complicated enough: "I’ll tell you what—there’s another biology on top of all this," says Liliana Tejidor, PhD, director of research and development in hemostasis at bioMérieux. That would be apoptosis. "People are starting to say, OK, this is not just the intersection of inflammation and coagulation, but also cell death. And there are so many genes and so many proteins involved in programming cells to die."
Not that everything is a complete muddle. Says Tim Hayes, MD, DVM, chief of clinical pathology at Maine Medical Center, Portland: "We realize coagulation is activated early on in this course. You don’t have to go into septic shock before you’ve activated the coagulation cascade." With Xigris moving off the shelf into clinical practice, more answers will be forthcoming.
As of late April, Dr. Laposata and his colleagues at Massachusetts General had experience with seven patients using Xigris. "We’re seven for seven," says Dr. Laposata, explaining that all seven did well on the drug. It’s unlikely, he adds, that any or all would have improved without APC. "We estimate the risk of mortality was on the order of 40 percent before the drug," he says. Indeed, the death rate in sepsis patients is significant, between 30 and 50 percent by most estimates.
From these and other early patients, Dr. Laposata and his colleagues are attempting to refine the use of activated protein C, which needs to be treated with the healthy respect given to any anticoagulant.
The laboratory has taken a rather aggressive approach in monitoring Xigris, measuring the full panel of coagulation factors each of the four days the patient receives the drug and looking for major changes that might indicate a predisposition to bleed. "We have some early hints," is all Dr. Laposata will reveal.
Prior to a patient’s receiving the drug, the laboratory looks at all the usual starting points—platelet count, PT, PTT, D-dimer. Each measure has its own cutoff value, which would vary among institutions. Says Dr. Laposata: "The bottom line is, there are thresholds for the PT and the PTT above which we would say, ’No, the patient can’t get the drug, they’re already too compromised.’ And there’s a threshold for platelets below which we say, ’There are not enough platelets, it is not safe to give the drug.’ Because we realize we are going to compromise the coagulation even more, and if they’re already predisposed to bleed, then that’s a problem."
A coagulation workup has not been routine in sepsis patients. In phase III of the clinical trial, Lilly did do extensive coagulation testing, Dr. Brandt reports, which demonstrated how common coagulopathies were among patients with severe sepsis. Nearly everyone, for example, had increased levels of D-dimer. "And these tended to be fairly high levels of D-dimer," he says. More than 90 percent of patients had prolonged PTs, and about two-thirds had prolonged APTTs. Thrombin-antithrombin complex and prothrombin fragment 1.2 were also elevated.
The clinical trial had its own criteria for excluding patients at high risk of bleeding: platelet count less than 30,000/µL, recent surgery, recent trauma, history of intracranial pathology, congenital bleeding disorder, and recent gastrointestinal bleeding. In the real world, says Dew, caregivers have the flexibility to consider stringent safety criteria without being restricted by them. "We can correct for a lot of these things while they’re happening," he points out. "We can give them platelets, FFP, blood."
At the same time, says Dr. Laposata, "We are extremely cautious." At Mass General, approval for the drug comes by way of a committee, which consists of a pharmacist and a clinician, either an intensivist or a pulmonary care physician. "These two people call the shots," he says. "If the patient qualifies, then they bring the rest of us in."
Yet there is room for judgment. If a patient looks like a good candidate despite a borderline platelet count, for example, the committee will ask other members of the care team, including the lab, to weigh in with their opinions. "This is where Dr. Laposata’s expertise has really come into play," Dew says.
Since most intensivists don’t have extensive hematological backgrounds, Dew says, the laboratory has stepped up to the plate, interpreting the data in real-time to determine if a transfusion is needed, and, if so, what kind. "It’s one thing to look at the data from a trial in a population-based study," Dew says. "We’re arguing that this drug can benefit some people who were probably excluded from the trial—and we need the lab to help figure that out."
Other challenges lie in wait.
Laboratories don’t need to routinely monitor the drug’s administration, says Dr. Brandt. "It’s administered as a constant-rate infusion," he explains. "The half-life, once you stop the infusion, is quite short. And generally within two hours you can no longer measure APC in the plasma from these patients." But activated protein C does have an effect on the APTT, which is related to the concentration of APC. The APTT is also dependent on the underlying coagulopathy that affects so many of these patients. Another caveat: The APC is gradually neutralized in the test tube between the time the sample is drawn and the time it’s tested in the laboratory. Thus, the APTT does not provide much helpful information on clinical samples. "If someone were trying to follow the underlying coagulopathy, rather than the drug effect, the prothrombin time would probably be a better test to use," Dr. Brandt says. And given that factor VIII assays are measured using the APTT, "You can get spurious results on a factor VIII assay. It can be very problematic."
Laboratories may also be called on to help monitor the impact of the drug, though the particulars are a question mark. "Certainly there is a great desire to limit the severe consequences of an intracerebral hemorrhage, and the lab may be able to help with that," says Dr. Laposata. "But we are not there yet."
The number of unknowns is hardly shocking, given that both the definition of sepsis and the tests used to identify it are undergoing makeovers.
"We’ve not had real sensitive tools. They’re kind of blunt instruments to detect sepsis," says Maine Medical Center’s Dr. Hayes, pointing to dropping platelet counts, a drop in fibrinogen, and elevated fibrin degradation products and/or D-dimer.
At the same time, the definition of inflammation itself has changed considerably in recent years, with efforts to grade the stages of sepsis. "Up until now there were a lot of terms out there that were poorly defined; now the intensivists and clinical doctors are using the new criteria, and we’re finally starting to all speak the same language," Dr. Hayes says.
SIRS, or systemic inflammatory response syndrome, is a cornerstone of this new lingua franca. Patients must have two of the four criteria—abnormal body temperature, either elevated or lowered; tachycardia; increased respiratory rate; and an abnormal white count, either high or low—to meet the definition of SIRS. Patients with SIRS plus a confirmed infectious process meet the criteria for sepsis; those with sepsis and organ failure, such as respiratory failure, hypotension, or hypoperfusion, are considered to have severe sepsis. Nonresponsive shock (hypotension despite fluid resuscitation) translates into septic shock.
"It’s only in the last few years that we’ve begun to start understanding the relationship between SIRS, and the various forms of it, and coagulation," says Dr. Hayes. "We’ve always known that people who have fulminant gram-negative sepsis often end up in DIC. What we’re realizing now is that early on in sepsis we’re activating the coagulation cascade. DIC is kind of the end stage of that process."
For that matter, perceptions of DIC are also evolving, Dr. Hayes says. "Most of us, when we think of DIC, we’re thinking of patients who are bleeding out—it looks like a hemorrhagic disorder." The multitude of fibrin deposition microthromboses, which impair organ perfusion and often lead to the multiorgan system failure, have been less obvious.
"We’re finding out that before you can really see those changes, you oftentimes have a severe derangement of the coagulation system," Dr. Hayes says. "When you use more sensitive tools, like the thrombin-antithrombin complex, or some of the activation peptides in prothrombin, we find these things are oftentimes elevated very early on in sepsis. In a sense, they’re sort of subclinical, both from a laboratory perspective and a clinical perspective. They’re really, really subtle changes."
It would be nice, says Dr. Hayes, if laboratories could "go to clinicians before the cat’s out of the bag—before we’ve got overt DIC—and give them a heads-up because we’ve got early indicators of a pending disaster."
Some are turning a hopeful eye toward the biphasic waveform; bioMérieux uses its Waveform Analysis Technology on its MDA analyzers to alert the operator to the presence of this abnormal waveform. A flag, called A2, is reported with the patient’s APTT clot time. Early investigations of the biphasic waveform, done by Cheng Hok Toh, MD, a hematologist with The Royal Liverpool University Hospitals and the University of Liverpool, United Kingdom, showed a high correlation between the biphasic waveform and DIC.
"I personally have found it useful, in cases where the diagnosis is a little less clear," says Elizabeth Van Cott, MD, director of the coagulation laboratory at Massachusetts General. "When the waveform was strongly abnormal, it pushed our diagnostic decision-making toward DIC." Her lab has started using the waveform in Xigris patients, and she and her colleagues plan to continue evaluating the test.
Others who are familiar with the waveform offer similar anecdotal information. Dr. Hayes weighs in with mixed findings.
"We’ve had some very interesting cases where the A2 flag was an early predictor of bacterial infection. And the laboratory was aware of it before the clinicians were. That’s very clear," he says.
"Interestingly, we’ve also seen some fulminant cases of meningococcal sepsis that didn’t have an A2 flag, even though the patient was in fulminant DIC," he continues. "We’re still trying to understand the nuances of this tool." The questions are lining up: What is its sensitivity? Specificity? What are the positive and negative predictive values of the assay in the ER? The ICU? "A lot of this is just trying to get the bugs out and understand the performance characteristics in different settings," Dr. Hayes says.
The waveform’s first phase reflects the precipitation of a complex between C-reactive protein and lipoproteins, primarily very low-density lipoproteins; the second phase represents the clotting reaction. The lipoprotein component is especially complicated, with the literature offering no clear-cut answers about the precise role lipoproteins play in sepsis and infection. "I think some of our understanding has been limited by the technology used to measure the lipoproteins and triglycerides," bioMérieux’s Dr. Tejidor says. "Perhaps the measurement of the complex quantifies a unique subfraction of lipoproteins that is clinically important." They have evidence, she says, that suggests lipoproteins in patients with biphasic waveforms are qualitatively different. "What we don’t know yet is whether these qualitative differences can be attributed to the lipoproteins that form the lipoprotein complexed CRP complex."
Dr. Brandt terms the waveform "an interesting biologic phenomenon—it’s real. But how it’s going to be used is unclear. For example, would a patient benefit from intervention therapy as soon as they develop a positive A2 flag? I don’t have an answer to that."
No one does. But rest assured, they’re looking. Especially bioMérieux. A multicenter clinical trial just getting underway will look prospectively at the flag’s sensitivity and specificity for sepsis and infection in the ICU setting, says Dr. Tejidor. This trial and others will also clarify the waveform’s role as an early indicator of sepsis.
"We believe that the activated coagulation state and linked inflammatory state, as a background, are the common denominators in the appearance of the biphasic waveform and therefore can be seen in other disease states," such as cancer and trauma, she says. It’s clear that the specificity of the flag is not 100 percent for sepsis. But, Dr. Tejidor contends, sepsis is the most common disease state associated with the biphasic waveform.
There also seems to be a strong relationship between the ongoing appearance of the biphasic waveform in ICU patients and mortality. "One of the things we’re trying to sort out is, Is this just a marker of death, or do you have time to do something for the patient?" Dr. Tejidor says. "In our pilot trial, we were encouraged that the answer is that we do have a window of time for the physician to do something differently, to monitor the patient more closely, order a blood culture, be more proactive even when they don’t suspect infection."
The A2 flag was not part of the Xigris studies. "We want to confirm its utility before we confirm the therapeutic side," says Dr. Tejidor. The flag does not currently have FDA approval for any specific clinical claim.
However, it has been approved for reporting, and some laboratories are doing just that.
Tai Po Tschang, MD, medical director of the laboratory at St. Agnes Medical Center, Fresno, Calif., has been reporting the A2 flag for more than a year, noting "biphasic waveform present" in the lab’s reports. Dr. Tschang calls the flag a value-added service in conjunction with reporting PTT clotting times.
He and his colleagues began looking at the flag two years ago, identifying 43 inpatients whose PTTs had one or more A2 flags. (When the A2 flag function is turned on, it’s run automatically with any panel that has a PTT component.) Out of that group, the mortality rate was 42 percent. The diagnoses ran the gamut: CABG, chest pain, pneumonia, fever, sepsis, pulmonary emboli, edema.
Impressed with those findings, the lab began regular reporting about year and a half ago. In 2001, the lab performed 8,594 inpatient PTTs; out of that group, 169 had an A2 flag present, approximately a two percent incidence. In ER patients, 2,244 PTTs were performed, with 14 A2 flags present (0.6 percent).
Dr. Tschang himself raises the obvious question: What does that mean? "It’s difficult to answer," he says, launching into a mini-Socratic musing. "It’s a bad prognostic sign. But nobody knows what to do with it. Is it useful? Yes. Is it of great importance? At present, the answer is ’No.’ We’re waiting for the research to catch up."
Bernard Fernandes, MD, offers a similar assessment. Dr. Fernandes, associate professor of laboratory medicine at the University of Toronto and head of hematopathology and blood transfusion service at Mt. Sinai Hospital, Toronto, analyzed 100 biphasic waveforms that occurred in 61 patients—40 of whom came from the ER. As at St. Agnes, the flag appeared in a variety of primary diagnoses and was associated with a high degree of mortality (42 percent). A blood culture confirmed infection in 67 percent of these patients and, though all had evidence of altered coagulation, only two met the International Society for Thrombosis and Haemostasis proposed criteria for overt DIC. The lab uses the flag to alert physicians but does not regularly report it.
"This may be another piece of the puzzle, but much of it is speculation at this point," Dr. Fernandes says.
The challenge now is to put all the pieces together, to understand the true pathophysiology of sepsis. And to intervene, says Dr. Hayes, "before the patient ends up in the autopsy suite."
Karen Titus is CAP TODAY contributing editor and co-managing editor.