Procalcitonin (PCT), the biologically active precursor of the calcium-modulating hormone calcitonin 2, has been shown in diverse studies to be closely associated with the human host response to bacterial infection 3456. It is elaborated by parenchymal cells throughout the body in response to endotoxin and several pro-inflammatory mediators (in particular TNF-α) and its concentration appears to be roughly linear with the degree of insult 7. The use of circulating PCT measurement to guide antibiotic therapy reduces antibiotic exposure in patients with suspected lower respiratory tract infection in both the inpatient and outpatient setting 8910. The role of PCT in patients with more severe infections such as severe sepsis has yet to be fully elucidated, but it has tantalizing performance characteristics as a biomarker for bacterial infection, showing diagnostic superiority to white cell count, C-reactive protein, and a host of physiologic variables in most reports 11. However, these investigations suffer from the absence of a diagnostic gold-standard, a common problem in studies of infection 12. The use of PCT to diagnose bacteremia or sepsis has been the subject of significant debate and at least three meta-analyses, two supporting 1314 and one discouraging 15 its clinical utility.

In the present single-center randomized controlled trial the authors evaluated a protocol for antibiotic cessation based almost entirely on plasma PCT level, with the primary outcome relating to the duration of antibiotic exposure. Seventy-nine intensive care unit (ICU) patients with suspected severe sepsis or septic shock according to ACCP-SCCM consensus criteria 16 were randomized to either usual care or a protocol arm in which the duration of antibiotics was determined by serial PCT measurements. These patients were quite ill, with ~50% requiring vasopressors and ~80% invasive ventilation. Serum PCT measurements were obtained daily and antibiotic cessation was encouraged on either day 3 or day 5 (depending on the initial PCT level) in intervention patients who experienced a predefined relative or absolute decline in PCT, with the implicit assumption that these patients had resolved their septic focus. The intention-to-treat analysis showed a nonsignificant trend toward reduced duration of antibiotics use. In the per-protocol analysis, PCT-guided therapy not only resulted in significant decreases in duration of antibiotic use, but a 2-day shorter ICU stay. Mortality and infection recurrence rates were similar between groups.

This study does have significant appeal. Rather than focusing on whether PCT can accurately diagnose infection, the authors have instead shown that it can be used as part of a treatment protocol to reduce antibiotic duration in some of the sickest ICU patients. The study does, however, have limitations that deserve consideration. Important exclusion criteria included the presence of certain difficult to eradicate pathogens (notably, Pseudomonas aeruginosa and Acinetobacter baumanni), infections requiring prolonged antibiotic therapy (e.g., endocarditis, deep abscesses) and immunosuppressed subjects, such as those with human immunodeficiency virus, neutropenia, or solid organ transplantation. The ultrasensitive PCT assay used in the study is not yet widely available. Even the standard PCT assay has a turn-around time measured in days, rather than hours, in many academic medical centers, such as our own where PCT is a "send out" lab. The results of the study were only significant in the per-protocol analysis, which was limited to patients with several days of follow-up and without post-randomization death or diagnosis of complicated infection requiring extended antibiotic therapy. This was designed to limit the analysis to subjects in whom a decision to stop antibiotics could actually be taken based on PCT levels. Though this approach is not inherently wrong, the use of serial PCT levels to guide therapy requires levels to be drawn on all patients, not just those in which PCT later proves to be of benefit, which raises issues of cost-effectiveness. Even so, the positive trend seen in the intention-to-treat analysis is reassuring and may have become significant had more subjects been enrolled.

Unfortunately, the main shortcoming of the study is that it was not powered to answer the real question. That is, can antibiotic exposure be safely reduced? Mortality and infection recurrence rates were similar between groups, suggesting that antibiotic use was reduced without harming patients. Yet, as the authors point out, a study powered for these endpoints would require several hundred patients per arm.

There are several large ongoing or recently completed multicenter trials of PCT-guided antibiotic therapy in ICU patients with infection. The PROcalcitonin to Reduce Antibiotic Treatments in Acute-Ill Patients (PRORATA) study, a 630 patient study in adult ICU patients with presumed bacterial infection, completed enrollment May 2008 17. The Procalcitonin and Survival Study (PASS), a 1000 patient study in adult ICU patients with severe sepsis, is expected to complete enrollment in early 2009 18. An additional study in 200 adult ICU patients with suspected infection, but no clear-cut source by clinical or microbiological criteria, is expected to close in late 2009 19. 1

The PCT-based protocol in the study does appear to reduce antibiotic exposure in patients with severe sepsis, but issues of assay availability, generalizeability, safety, and cost-effectiveness must be addressed before we can recommend its routine use.

The authors declare that they have no competing interests.