Researchers Claim Sepsis Has Four Faces So Tailored Therapy May Hold Promise

By Gene Emery

May 22, 2019

(Reuters Health) - Doctors in Pittsburgh say they have developed a system to classify sepsis into 4 categories that may help doctors develop better treatments for a condition that is the leading killer of hospitalized patients.

It may also prompt reexamination of proposed therapies that appeared to have failed when all sepsis cases were lumped together.

"This is a new way of thinking about sepsis - that not all patients are the same," Dr. Christopher Seymour, associate professor of critical care medicine at the University of Pittsburgh's School of Medicine and the chief author of the study told Reuters Health in a telephone interview.

"We can image that a lot of folks will be looking at their own trial datasets" to see if a therapy that failed overall might hold promise for one of the four phenotypes.

The findings were reported online May 19 in the Journal of the American Medical Association and at the American Thoracic Society's international conference in Dallas.

Their statistical analysis of more than 63,000 cases looked at over 29 variables ranging from basic demographic information to markers of inflammation.

The analysis found that the cases seemed to break down into four phenotypes, designated Alpha, Beta, Gamma and Delta.

Alphas, accounting for 33% of patients with a death rate of 2%, showed the fewest abnormal laboratory test results. Betas, who accounted for 27% of cases (mostly among older patients), had a death rate of 5% and tended to suffer from chronic illnesses and kidney dysfunction. Gammas, who also made up 27% of sepsis patients, had a death rate of 15% and were most notable for having elevated measures of inflammation and primarily pulmonary dysfunction. Patients in the Delta phenotype accounted for 13% of sepsis cases, had a death rate of 32% and were far more likely to have liver dysfunction and shock.

How reliably can an individual patient be put in a category?

When the Seymour team assessed the "chances of individual patients being members of their assigned group versus being members of others, it was only about six out of 100 patients who were on the margin," he said. "It was very reassuring that the group members had very high probabilities" and it was all based on "the data in front of you at the bedside."

The team used 20,189 cases of sepsis at a dozen Pennsylvania hospitals -- analyzed with statistical, machine learning and simulation tools -- to derive the four phenotypes. They were validated using 43,086 additional cases.

Dr. Seymour said the classification system has not produced any immediate strategies for treating the different phenotypes.

"There's a lot more to learn about the biology underlying these groups," he said. "We really have to move forward with new trials to see if simple interventions like fast antibiotics or more complex interventions like immune therapy will or will not benefit patients of certain subtypes."

But doctors need to get past the one-size-fits-all model of sepsis, Dr. Seymour said, because currently "our choice of antibiotics for a patient with sepsis and organs under stress is the absolute same regardless of whether that patient came with a urinary tract infection that had descended into the kidney, or whether that patient had community-acquired pneumonia."


JAMA 2019.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: