Antibiotic Use in Sepsis: How and Why Less Can Really Mean More (Survival)

Timely and accurate antibiotic therapy is essential in the management of sepsis, with delays in the initiation of appropriate antibiotic therapy being a consistent mortality risk factor (1). In this setting, antimicrobial stewardship is needed to avoid the overuse of broad-spectrum therapy while preserving our antibiotic armamentarium and avoiding antibiotic-associated side effects (2). To promote stewardship, clinicians have turned to biomarkers that can identify a bacterial infection and have used serial measurements to guide the discontinuation of antibiotic therapy. One such serum biomarker, procalcitonin (PCT), has been valuable for patients with community-acquired pneumonia (CAP) and is increasingly being used in patients with a variety of infections, including sepsis, both in and out of the ICU (3–5).

PCT is an acute phase reactant that is primarily produced by the liver in response to bacterial infection, with concentrations rising in response to microbial toxins or host responses (IL-1β, TNF-α, and IL-6) and declining when infection is controlled (6). Viral infection–related cytokines attenuate PCT induction, but some elevation can occur with atypical pathogen pneumonia. Based on these concepts, patients with lower respiratory tract infection (LRTI) can often have antibiotics safely withheld when PCT concentrations are low, provided that clinical judgment supplements biomarker measurement. In addition, duration of therapy can be guided by a drop in PCT concentrations. However, it remains uncertain whether routine measurement should be used in patients.

The earliest studies were done in non–critically ill emergency department patients with a variety of LRTIs and then in patients with clinical and radiographic CAP; using an algorithm based on initial and serial serum concentrations, the use and duration of antibiotic therapy was safely reduced (7, 8). PCT concentrations may vary with the time of presentation in the course of illness, with higher concentrations in those presenting within 3 days of the onset of symptoms compared with those presenting later (9). Renal failure can falsely elevate PCT concentrations by interfering with elimination, but certain dialysis membranes can remove PCT and make measurements falsely low.

The value of PCT for antimicrobial stewardship has been demonstrated in ICU patients. In one randomized, open-label study of 621 ICU patients (180 with CAP), PCT guidance reduced antibiotic use for CAP to 5.5 days from 10.5 days in control subjects (P < 0.0001) (4). In another randomized study of 1,546 ICU patients (the SAPS [Stop Antibiotics on PCT Guidance] study), PCT guidance reduced the duration of antibiotic treatment and increased the number of antibiotic-free days compared with control subjects, with 64% of patients having pulmonary infection, 16% having intraabdominal infection, 3% having urinary tract infection, and 1% having bloodstream infection (5). In that study, the number of daily defined doses of antibiotics in the PCT group was 7.5 days versus 9.3 days in the standard-care group (P < 0.0001). However, a significantly higher number of patients in the PCT group versus the standard-care group (5% vs. 3%) needed a second course of antibiotics for reinfection with the initial pathogen. Surprisingly, 28-day mortality was significantly lower with PCT guidance (20% vs. 25.5%; P = 0.012) and remained significantly lower at 1 year (36% vs. 43%; P = 0.018). These impressive results were obtained even though adherence after meeting PCT stopping criteria occurred in only 44% of patients at 24 hours and 53% of patients at 48 hours. The observed mortality reduction was unexpected, and the authors speculated that it could have been due to searching for alternative diagnoses in those with low concentrations and to reevaluating therapy in those with rising concentrations. A systematic analysis of 16 trials in critically ill patients (heavily dependent on the SAPS data) also confirmed a mortality benefit to PCT guidance but noted that the evidence was of low certainty, and there was a high risk of bias because of poor adherence to PCT protocols (10). The concept that antibiotic use can increase mortality if too broad is not new and was also seen in a study of nosocomial pneumonia in 2011 (11). With PCT use in LRTIs, a patient-level meta-analysis also showed mortality benefit (3).

In this issue of the Journal, Kyriazopoulou and colleagues (pp. 202–210) describe a multicenter, open-label study, the PROGRESS (Procalcitonin-guided Antimicrobial Therapy to Reduce Long-Term Sequelae of Infections) trial, of 266 patients meeting sepsis 3 criteria, which compared PCT-guided therapy to standard care and also found a significant benefit in mortality reduction, with a plausible mechanism for this benefit (12). In the study, antibiotic discontinuation was recommended when PCT was below 0.5 μg/L or when there was a reduction of at least 80% from the baseline at Day 5 or later. PCT-guided discontinuation was followed 76.8% of the time in a setting where 98% of all patients received appropriate therapy (presumably because of a 4% rate of multidrug-resistant [MDR] pathogens) and led to a significant reduction in infection-associated adverse events at 180 days. This was the primary endpoint, and it consisted of a composite of new infection or death caused by MDR pathogens or Clostridium difficile. Although fecal colonization rates at 7 and 28 days were similar in both groups, colonized patients, particularly those with C. difficile or MDR pathogens getting standard care (but not those getting PCT guidance) developed a higher rate of infection-associated adverse events. These data suggest that PCT-guided early discontinuation of antibiotics can prevent intestinal colonization from progressing to clinical infection. Patients in the study had a mixture of infections, with 62% because of pneumonia, 37% with pyelonephritis, and 1% with primary bloodstream infection. Although all met sepsis 3 criteria, 7% had septic shock, and 16% had positive blood cultures.

Mortality at 28 days, a secondary endpoint, was significantly lower (15.2% vs. 28.2%; P = 0.02) with PCT guidance, with a trend toward significant reduction in 180-day mortality (P = 0.24). Cost of care and length of antibiotic therapy were also lower in the PCT-guided group. The findings in the PROGRESS study extend the observations of the SAPS study in a number of important ways. They show a mortality benefit of biomarker-guided antibiotic therapy in a population of patients with sepsis who are not exclusively in the ICU and have a lower predicted mortality than in ICU patients. The data also helped to define a potential mechanism for mortality benefit—the avoidance of infection-associated adverse events, particularly the prevention diarrhea and acute renal injury.

Do these new data mean that PCT-guided antibiotics should be used in all patients with sepsis? Certainly the findings are provocative, but the success of the PROGRESS trial was likely dependent on high adherence to biomarker-recommended early discontinuation of therapy in a setting where most patients received initially appropriate therapy. In hospitals in which patients with sepsis have a high rate of resistant pathogens, leading to frequent inappropriate therapy, and longer durations of therapy, this benefit may be less clear. Still, the findings are important and clearly demonstrate that in patients with sepsis, less antibiotic use can equate with more survivors.