November 2010
Feature Story

Karen Lusky

Methodist North Hospital in Memphis uses a “smart” computerized algorithm to detect the first signs of severe sepsis recorded in a patient’s electronic medical record. The computerized rule is bidirectional, so that when an abnormal lab value pops up indicating organ dysfunction, the algorithm probes the EMR to see if the latest vital signs also show a problem, says Paula Jacobs, director of quality and performance improvement at MNH.

If two of the vital signs are also abnormal, bingo—a severe sepsis alert triggers, advising clinicians to take a close look.

You might say the sepsis field itself is increasingly bidirectional these days, with escalating efforts to make the most of what’s long been known to work and forward-thinking research that may soon produce the first personalized therapies.

But for now, hospitals like Methodist North are scoring major points in combating severe sepsis, which Jacobs notes is generally defined as systemic inflammatory response syndrome, or SIRS, coupled with infection and new onset of organ dysfunction.

MNH’s computerized early-detection program has in fact cut the average 30 percent sepsis mortality rate at the hospital by more than half, says Noel Florendo, MD, PhD, director of the clinical laboratory.

“We aren’t looking for the Holy Grail,” Dr. Florendo says. “We’re just looking for two signs of systemic inflammatory response syndrome” and pursuing that in the right clinical setting.

The signs of SIRS in MNH’s computerized severe sepsis detection algorithm are tachycardia and tachypnea, hyper- or hypothermia, a white blood cell count greater than 12,000 or below 4,000, and plasma glucose exceeding 120 mg/dL in the absence of diabetes, says Jacobs, who helped design and implement the early-detection initiative now in place at MNH and the other hospitals in the Methodist Le Bonheur Healthcare system.

MNH emergency department patients who have confirmed or suspected infection and two signs of SIRS automatically receive point-of-care lactate testing to look for organ dysfunction owing to poor perfusion. A severe sepsis alert triggers if the patient’s lactate value exceeds 2.2 mM/L. At that point, “the ED staff immediately administers a minimum two liter fluid bolus,” Jacobs says.

Patients with lactate levels above 4 mM/L go straight to the ICU. Jacobs recounts how the ED evaluated one young man who didn’t appear to be critically ill. But a lactate of 4.4 said otherwise. Diagnosed with severe sepsis, the man spent two to three days in the ICU; he recovered completely.

“Sometimes the lactate may not be elevated” in patients with severe sepsis, Dr. Florendo cautions, which is why the hospital doesn’t rely solely on the marker.

Additional signs of acute organ dysfunction included in the computerized rule are systolic hypotension, elevated creatinine, hyperbilirubinemia, mental status change, and a low platelet count.

The algorithm flags mental status change by looking in the EMR for such words as “agitation, confusion, anxiety, or lethargy,” Jacobs says. “The rule will count ‘lethargic, etc.’ as either SIRS or organ dysfunction, but not both.”

MNH has found mental status change and elevated lactate to be the earliest signs of organ dysfunction, she says. “If we waited for elevated creatinine or low platelets to develop, the patient would be too far into the sepsis cascade.”

Intermountain Healthcare in Salt Lake City uses a computerized sepsis detection method in its hospitals’ ICUs and emergency departments. The protocol is based on both a rules-based algorithm like that of MNH and a Bayesian algorithm of probability that a person has severe sepsis, says Todd Allen, MD, an emergency room physician who helped develop and implement the protocol.

The algorithm includes abnormal vital signs—hypotension, leukocytosis or leukopenia, elevated lactate, and results suggesting a source of infection such as a urinalysis positive for bacteria—or a chest x-ray report signaling pneumonia or an infiltrate, Dr. Allen says.

When positive, the computerized screening tool will send an automatic alert to the clinicians, which helps them avoid the “cognitive traps” in which they fail to entertain sepsis as a diagnosis for those “in-betweener” patients who don’t have clear signs of severe sepsis or septic shock, he says.

In an outcomes study, Dr. Allen and colleagues found that in 2008–2009, when the EDs and ICUs at Intermountain Medical Center and LDS Hospital complied 48 percent of the time with all 11 elements in their severe sepsis and septic shock bundles, their sepsis mortality dropped to 10.3 percent. In 2006–2007, mortality was 17.8 percent, with 33 percent compliance.

The goal now is for all Intermountain Healthcare hospitals to implement the sepsis bundles 77 percent of the time, Dr. Allen says.

The Methodist Hospital in Houston has a simpler screening tool in its surgical ICU, which has reduced the unit’s sepsis mortality rate from about 34 percent at the outset to 14 percent.

The screen assigns one point for each of four variables: elevated heart rate, too high or low temperature, elevated respiratory rate, and an abnormally high or low white blood cell count, says Laura Moore, MD, assistant professor of surgery, Weill Medical College of Cornell University, The Methodist Hospital.

A score of four is considered positive for SIRS. For purposes of calculating the score, the ICU staff doesn’t obtain a white blood cell count if the patient does not have one, though 99 percent of the patients have daily WBC counts, Dr. Moore says.

“A high white count is worrisome but so is a really low one,” because the latter indicates the person may not be mounting an appropriate immune defense, she says. “Most of the time we see that it’s elevated.”

Patients with a positive SIRS score receive lactate testing to look for organ dysfunction. A physician or physician extender also examines them to identify a potential site of infection. That assessment, which typically takes five to 10 minutes, is important, Dr. Moore says, because the SICU finds a small number of patients who have a positive SIRS score due to bleeding or a cardiac arrhythmia.

To identify who has bacterial infections causing SIRS, some hospital sepsis protocols incorporate procalcitonin, or PCT. Stanford University Medical Center, for example, just started using PCT for that purpose, said James Faix, MD, director of clinical chemistry and immunology at Stanford, in a presentation on sepsis biomarkers last July at the AACC annual meeting.

In his talk, Dr. Faix pointed to a lot of recent evidence, including the results of the PRORATA trial published in Lancet this year, showing that a declining PCT level can be used in the ICU as a marker for when to stop antibiotic therapy safely (Bouadma L, et al. Lancet. 2010;375[9713]:463–474).

Another study, Dr. Faix said, shows that a rising PCT level can identify patients at risk of developing severe sepsis. In the study, “the slope of the change was what was important—not a single level,” he said (Jensen J, et al. Crit Care Med. 2006;34:2596–2602). Even when the initial PCT was elevated, a rising level also predicted mortality.

Dr. Moore in Houston notes that PCT isn’t available at Methodist and probably wouldn’t work well in the surgical ICU because PCT is elevated in surgical patients whether or not they have an infection. (To use PCT in that setting to detect sepsis, a much higher cutoff would have to be established, Dr. Faix says.)

Intermountain Healthcare isn’t using PCT, says pathologist Sterling Bennett, MD, director of the clinical laboratory at Intermountain Medical Center. Clinicians there do not appear to be convinced, based on the literature, that the relatively expensive test adds anything to what they are already doing that would justify the extra cost.

“Some people,” says Mayo Clinic pathologist Brad Karon, MD, PhD, “are very high on using PCT and some aren’t,” and he calls the data on PCT “still very controversial.” He describes PCT as a “promising marker” for early infection, but says sensitivity and reliability have varied in clinical studies. It seems to be “very good at detecting early gram-negative sepsis,” Dr. Karon says, “but perhaps not as reliable when there are other types of bacterial or fungal infections.”

Still unknown, he says, is how many different conditions will result in a rapid rise of PCT and whether the available point-of-care testing applications for procalcitonin are sufficiently sensitive to be valuable.

PCT may well prove to be of value in the future, Dr. Karon adds. Lactate, by contrast, is state of art, “because the quality of the point-of-care test is good enough to correctly categorize patients [as having severe sepsis] when the test is done correctly, and it catches all comers with hypotension/shock and end organ damage, though not as early as would be desired.” (For more on lactate testing, see box at right.)

The bottom line, Dr. Faix says, is that so far “there’s no magic test” to say a person is going to need a lot of attention before he or she has crossed from sepsis to severe sepsis and is “already on the road to a severe outcome.”

In ongoing research, however, Dr. Moore and her colleagues at Methodist in Houston have identified routine lab tests that correlate at baseline with sepsis severity and mortality, some of which haven’t been previously identified in that way, she says.

“For example, we have seen a significantly higher level of BNP at baseline in patients who go on to die from sepsis,” Dr. Moore says. “So when patients come to the SICU, we send a slew of lab tests to document their baseline levels of different variables as part of the standard of care.”

“In addition to BNP, lactate is significantly higher in nonsurvivors, as is INR,” she adds. Platelet count (a low count), AST, ALT, and total bilirubin also statistically differ in people who don’t survive sepsis compared with those who do.

Dr. Moore and her co-researchers haven’t broken down the test variables predicting mortality by sepsis, severe sepsis, and septic shock. “Obviously, mortality is going to be highest in septic shock,” she says. “But we have also seen an increase in severity of those baseline labs as patients progress from sepsis to severe sepsis.”

So far, the laboratory values haven’t led to a change in how the clinical team treats patients, Dr. Moore says. “But as we are better able to characterize these patients, it will help us better understand ways to discriminate the severity of illness beyond our current markers.”

Internist John Tayek, MD, who co-presented with Dr. Faix at the AACC meeting, believes albumin can add something to the prognostic picture for sepsis patients, especially those with moderate to high lactate levels. He cited a study of 199 sepsis patients showing that a patient with an albumin level above 2.9 had a 0.7 percent mortality risk. But if it was under 2.9, “you could just call the priest or rabbi, etc.,” he said (Bongard F, Sue D, Vintch J. Current Diagnosis and Treatment Critical Care. 3rd ed. 2008:120).

“In the acute phase of sepsis or injury, albumin turns off as the liver makes acute phase proteins that are critical in sepsis,” Dr. Tayek explains. “Albumin gets out of the way so that the liver can make the good stuff.”

Within the first hour of becoming septic, he says, the body “dumps out” tumor necrosis factor and interleukin 1 and 6, which are “all hugely proinflammatory.” Within a half-hour, the liver releases C-reactive protein, which increases for 12 to 24 hours until it’s at its peak. “CRP blocks too much tissue destruction caused by the proinflammatory substances,” he says, and has an antimicrobial function.

Today, sepsis is increasingly viewed as involving a proinflammatory phase followed by an antiinflammatory one. Among proponents of that view is Edward Abraham, MD, professor and chair of the Department of Medicine at the University of Alabama, Birmingham. In the future, Dr. Abraham predicts, diagnostics will identify where a person is in that continuum and how he or she should be treated.

Richard Hotchkiss, MD, and Steve Opal, MD, in a July 1 New England Journal of Medicine article, “Immunotherapy for sepsis—a new approach against an ancient foe,” write: “Occasionally, patients present with an exaggerated systemic inflammatory response to highly virulent pathogens (such as in cases of meningococcemia) and rapidly succumb” (N Engl J Med. 2010;363:87–89).

Normally, Dr. Hotchkiss tells CAP TODAY, “a healthy individual has a robust proinflammatory response to infection, which heightens the immune response. But as the sepsis progresses, if you don’t eradicate the infection, the person may enter a phase of more profound immune suppression where they get hospital-acquired infections” and die of those. That occurs because the “proinflammatory response wreaks some havoc. The invading microorganisms cause death of a lot of immune cells,” says Dr. Hotchkiss, professor of anesthesiology, medicine, and surgery at Washington University School of Medicine, St. Louis.

In research studies, Dr. Hotchkiss and colleagues are testing for expression of immunosuppressive cell surface proteins such as PD-1 (programmed cell death 1) and CTLA-4 (cytotoxic T lymphocyte antigen 4). The tests, which he says are easy to do, use flow-cytometry–based assays to look at expression on immune cells. “It’s not ready for prime time but it’s progressing and in the future could be used along with other markers of lymphocyte activation to determine if the patient is in the hyper versus the hypo-inflammatory phase of sepsis.”

In their NEJM article, Drs. Hotchkiss and Opal write: “PD-1 impairs immunity by inducing apoptosis, increasing production of interleukin-10 (a key antiinflammatory cytokine increased in sepsis), preventing T-cell proliferation, and causing T cells to become nonresponsive (“exhausted”).”

Knowing a patient is in the hypo-inflammatory phase, a clinician could give the person immunotherapy to boost the immune system, Dr. Hotchkiss says. Potential candidates include a number of drugs in cancer clinical trials, among them IL-7, anti-PD-1, and anti-CTLA-4, which are showing “modest success in treating a variety of malignancies.” If the drugs are proved to be safe, “careful clinical trials” should be considered to test them for sepsis, he says.

Conversely, if you knew a person in the early phase of sepsis had too much inflammation, “you could do immunotherapy to modestly down-modulate the response. You don’t want to block it excessively as the hyperactive immune phase helps kill the organisms that are attacking the host.”

Other factors that determine how a person responds to sepsis are the number and types of organisms and their virulence, the site of infection, and whether the person has comorbidities already causing immunosuppression, such as diabetes or cancer, Dr. Hotchkiss says.

In addition, “there are patients who have known genetic changes that predispose them to making more or less of particular cytokines.” And in the future, he says, that information could potentially be used to predict which trajectory a patient is going to be on in sepsis. But so far, researchers haven’t identified the ones that are most predictive.

The first pharmacogenomic test for severe sepsis is, however, more than on the drawing board.

Sirius Genomics, of Vancouver, BC, is working on a pharmacogenomic test as a companion diagnostic for Xigris (recombinant activated protein C), made by Eli Lilly, says Kamran Alam, MBA, senior director of business development for Sirius. Xigris is FDA approved for the treatment of people with severe sepsis who have a high risk of death.

Sirius will soon validate its test, which uses single nucleotide polymorphisms and clinical severity scores to identify subsets of patients with severe sepsis or septic shock who may respond best to Xigris, Alam says. (The company hasn’t yet disclosed what SNPs are involved.) Sirius hopes to complete its validation by the summer of 2011 and have the test available in two years.

If the diagnostic test fulfills its promise, Alam predicts it could “rejuvenate” the use of Xigris, which he notes has fallen short of its expectations commercially, by giving physicians the confidence to prescribe it. Even from a clinical use perspective, Alam says, physicians have had concerns and questions about the drug’s efficacy and safety issues on the label.

Toronto’s Spectral Diagnostics, which makes the FDA-approved endotoxin activity assay, or EAA, is conducting a phase three clinical trial, EUPHRATES. Consisting of 360 patients at 15 medical centers in the U.S., the trial is evaluating the clinical impact of removing high levels of endotoxin in septic shock patients, says Debra Foster, RN, director of the company’s sepsis program.

The researchers are using the EAA to identify patients in septic shock who qualify for the treatment (those with 0.6 EAA units or higher). “The intervention is the polymyxin B hemoperfusion endotoxin adsorption column” made by Toray Industries in Japan, Foster says. About 50 to 70 percent of patients with septic shock have the high levels of endotoxin, she adds.

Endotoxin directly affects many organs and blood vessels, contributing to hypotension in septic patients, Foster explains. “Removing it brings up the person’s blood pressure, and that allows perfusion of organs such as the kidneys, heart, and lungs. Organ dysfunction appears to turn around fairly quickly.”

The design of EUPHRATES is based on a recent Italian study favorable to polymyxin B hemoperfusion and published last year in the Journal of the American Medical Association (Cruz D, et al. JAMA. 2009;301:2445–2452).

The ACCESS global phase three trial is evaluating a drug, eritoran, made by Japanese drug company Eisai. The drug is believed to block activation of the toll-like receptor 4 (TLR4), which is thought to be the receptor for endotoxin and other ligands, says Eisai spokesperson Judee Shuler, of Woodcliff Lake, NJ.

David Klein, MD, MBA, a critical care specialist at St. Michael’s Hospital in Toronto and an investigator in the ACCESS trial, says ACCESS takes patients with severe sepsis based on standard clinical criteria and, within 12 hours, randomizes them to an infusion of eritoran or placebo.

The ACCESS trial completed enrollment at the end of September and has a target date of March 31, 2011 for its regulatory submission, if the study is positive, according to Shuler.

Dr. Klein believes that if the Eisai trial is positive, it will “breathe new energy into and reinvigorate the therapeutic side of sepsis.” And it will “create more momentum for Spectral Diagnostics’ endotoxin activity assay and for the field of sepsis diagnosis.”

Spectral’s EAA tends to engender a “So what do I do with the result?” response from clinicians, he says, which has chilled interest in the diagnostic as a risk stratifier. But that’s historically how it goes for most new diagnostics, including troponin, he adds.

“When troponin came out, it was viewed initially as being just a fancy CK-MB”—until antiplatelet therapy became available to target the high-risk acute coronary syndrome population. That treatment breakthrough “drove uptake of troponin as a diagnostic because it identified a unique population to treat,” Dr. Klein says.

New molecular techniques in infectious disease may also help drive sepsis survival rates to a new high.

Work in that field is moving rapidly, says Dr. Hotchkiss, “and will have a major impact on sepsis by allowing physicians to identify pathogens early and employ specific antimicrobial agents.”

Reports indicate that about half of blood cultures done for suspected sepsis are never positive, though that doesn’t necessarily mean they don’t contain pathogens causing SIRS.

The reasons for the negative results, says Oliver Liesenfeld, MD, director of medical and scientific affairs for microbiology at Roche Molecular Systems, include the small volume of blood tested and the challenges of culturing blood for patients already on antibiotics.

There are also “antimicrobial defense mechanisms of the immune response in the blood, including granulocytes and antibodies/complements [that] complicate the growth of pathogens in blood cultures,” says Dr. Liesenfeld, who is also professor of medical microbiology and infection immunology, Charité Medical School, Berlin. In addition, “sepsis may be characterized by intermittent rather than continuous spread of bacteria from an infectious focus into the bloodstream.”

Roche’s LightCycler SeptiFast test, a DNA detection method used in Europe and other countries, can reportedly detect 25 of the most common bacterial and fungal nosocomial microorganisms in sepsis. Results are available within six hours compared with two days for blood culture. Dr. Liesenfeld says it would be “ideal” to get FDA approval for the test, but a clinical trial for a molecular test that detects 25 organisms poses several challenges that Roche would have to overcome before committing to bringing the test to the U.S. market.

The problem with Roche’s SeptiFast and all other direct-from-patient blood tests thus far has been their sensitivity, says Ellen Jo Bar­on, PhD, professor emer­ita in pathology at Stanford University, and director of medical affairs for Cepheid. “Results have not been close to 100 percent because you can’t use enough of the blood in an amplification reaction to be certain of including organisms present in very low numbers,” she says.

In addition, “you can detect organisms that may not be causing the sepsis,” Dr. Baron adds. For example, “the person may have a urinary tract infection and E. coli organisms in the blood, which are causing sepsis. But you amplify and detect pneumococcal DNA and may miss the E. coli, which takes you in the wrong direction.”

On the upside, Dr. Baron notes that companies have developed methods to concentrate the organisms circulating in the blood. “Once you do that, then you can figure out a rapid way to detect them using molecular techniques.”

Promising newer approaches involve looking directly in blood for microRNA or white blood cell-derived markers of messenger RNA up-regulation in response to the infectious agent, she says. In a study underway at Stanford, Dr. Baron and colleagues are using human white blood cell expression microarrays to analyze whether they can identify a causative infectious organism in ICU patients at the time they initially spike a fever. Currently, the detection approach takes two days but in the future might be able to produce results within hours, Dr. Baron says.

Marc Zubrow, MD, director of critical care medicine at Christiana Care Health System, Wilmington, Del., an early adopter of a sepsis detection and treatment program, isn’t convinced that having molecular tests to identify the causative infectious organism in sepsis will be as helpful as it sounds.

For one, “The run-of-the mill septic patients are typically elderly with urinary tract infection or pneumonia with potentially significant resistance issues,” he points out. And “at 2 AM on a Sunday morning, I’m going to give the patient every ‘gorillacillin’ on the shelf.”

Dr. Zubrow points to research published in September in Critical Care Medicine that shows if a patient does have severe sepsis, “broad-spectrum antibiotics to cover a wide range of organisms improve outcomes more so than giving just one (Kumar A, et al. Crit Care Med. 2010;38:1773–1785).

And even with earlier and better detection of causative organisms, microbial resistance will continue to be a problem, Dr. Zubrow says. In his view, what’s needed are “clever molecular treatments to overcome resistance.” For example, researchers have to “find a way to fool bacteria by putting DNA into them through phages or medications that make them sensitive to antibiotics again.”

There is no shortage of ideas and opportunities for new developments in sepsis care, but, as always, the best path to improvement is in the present. One of the central lessons that MNH’s Paula Jacobs has learned is the need for hospitals to dig into their data to see how they are truly faring in the fight against sepsis.

“Hospitals that don’t have a clue how they are doing will say they are doing great,” she says.

Related Links

• What affects the accuracy of lactate results?