Thursday, November 16, 2017
Advancing the Need to Reduce Unnecessary Antibiotic Treatment by Using the Biomarker Procalcitonin
Editors: Kelly Maurelus, MD FAAEM and Michael C. Bond, MD FAAEM
Originally Published: Common Sense January/February 2017
ED physicians frequently treat and admit patients for infectious diseases. Judicious use of antimicrobial therapy is important in order to avoiding the development of antimicrobial resistance and adverse drug effects. Procalcitonin (PCT) is one of several bbiomarkers which may be useful in decreasing unnecessary antibiotic therapy. Specifically, PCT levels should be low for viral, as opposed to bacterial, infections. Procalcitonin has been studied as both a diagnostic and prognostic marker in various types of systemic and organ-specific infections. The potential for PCT to reduce unnecessary antimicrobial therapy has been shown in several observational and randomized controlled trials performed in outpatient, inpatient, and ICU environments. The most robust evidence is in sepsis and pulmonary infections. Here we review some of the evidence behind the use of PCT in acute infectious disease management.
Bouadma L et al. Use of procalcitonin to reduce patients’ exposure to antibiotics in intensive care units (PRORATA trial): a multicentre randomised controlled trial. Lancet. 2010, 375:463-474.
In this multicenter randomized, prospective, parallel-group, open-label clinical trial, critically ill patients were randomized into two groups. In one group, antibiotic administration was driven by a predefined algorithm using PCT cutoffs (n=307), while antibiotic management in the second group was driven by local and international guidelines (n=314). In the PCT group, the initial PCT level was used to guide whether or not antibiotics should be prescribed and for those receiving antibiotics, serial PCT values were used to guide the timing of discontinuation of antibiotics. In patients who did not meet PCT criteria for the initiation of antimicrobial therapy, clinical reassessments and serial PCT values were used to guide the decision to initiate treatment later in the course of care. The infectious etiologies varied, with the majority attributable to pulmonary infections (71% in the PCT group versus 74% in the control group). Most others sources of infection were the urinary tract (9% in the PCT group versus 6% in the control group) and intra-abdominal sites (5% in the PCT group versus 7% in the control group).
The primary endpoints were 28-day and 60-day mortality rates, as well as the number of days without antibiotics. During the 28 days following inclusion into the study, the PCT group had significantly more days without antibiotics compared to the control group (14.3 days versus 11.6 days). The between-group absolute difference revealed 2.7 fewer days of antibiotic use in the PCT group (p<0.0001). Yet, there was no statistical difference in 28-day and 60-day mortality between the two groups, as determined by non-inferiority analysis using a 10% margin of non-inferiority. The hazard ratio between the two groups in terms of survival over 60 days was 0.96 (90% CI 0.84-1.09). Of note, at 28 days from inclusion the sequential organ failure assessment (SOFA) score in the PCT group compared to the control group was 1.5 versus 0.9 with a between-group absolute difference of 0.6 (0.0 to 1.1 p=0.0370).
While this study demonstrated that the use of PCT may lead to fewer days of antibiotics, and that PCT is non-inferior to standard management, there are several important limitations. In the PCT group, 53% of the patients were not treated per the PCT-driven protocol because of either earlier discharge from the ICU or because of physician prescription of antibiotics despite what was recommended by the PCT cutoffs. In addition, while non-inferiority was demonstrated by a 10% margin in terms of mortality, the PCT group actually showed slightly increased 28 and 60 day mortality as well as slightly worse SOFA scores.
Wacker C et al. Procalcitonin as a diagnostic marker for sepsis: a systematic review and meta-analysis. Lancet Infect Dis. 2013;13:426-435.
In this meta-analysis, Wacker and colleagues aimed to determine the ability of PCT to diagnose sepsis in critically ill patients. Specifically, the authors examined the ability of PCT to differentiate patients with sepsis from those with a systemic inflammatory response syndrome (SIRS) of non-infectious origin. The authors only included studies that confirmed the presence of infection either microbiologically or based on high clinical suspicion, defined as the presence of at least one of four criteria: a) leukocytes in a normally sterile bodily fluid; b) presence of a perforated viscus; c) radiographic evidence of pneumonia and purulent sputum; or d) presence of a syndrome associated with high risk of infection (i.e., ascending cholangitis). The authors excluded studies on healthy patients, patients without infection, and children under 28 days old.
A total of 3,487 studies were identified, of which 30 met the inclusion criteria. These studies included a total of 3,244 patients, of which 1,863 (57%) had sepsis and 1,381 (43%) had SIRS of non-infectious origin. Of the 30 studies, 21 (total of 1,173 patients) detailed the severity of sepsis, identifying 499 (42%) with sepsis, 234 (20%) with severe sepsis, and 440 (38%) with septic shock. Most of the studies examined patients either in the ED or ICU. The cohort of studies varied in terms of patient population (pediatric versus adult; medical versus surgical), infection site (blood, urine, lung, abdomen), infection source (nosocomial versus community), and study location (America versus Europe).
The authors calculated that PCT had a pooled sensitivity of 77% (95% CI, 0.72-0.81) and a pooled specificity of 79% (95% CI, 0.74-0.84). After compiling a receiver operating characteristic (ROC) curve from the data points, they calculated the area under the curve (AUC) for all patients and for each sub-category. The pooled AUC was 0.85, but the sub-group analysis showed a trend, without significance testing, toward improved accuracy for surgical patients over medical patients (AUC 0.83 versus 0.79). However, there was no difference in accuracy between adult medical and pediatric patients (AUC 0.85 versus 0.85). Overall, the authors noted significant heterogeneity between the studies (overall I2 for bivariate model 96%, 95% CI 94-99), but they were unable to identify any other significant heterogeneity among the specific covariates.
The authors concluded that PCT is accurate in identifying sepsis, and they support its use in clinical practice. However, several limitations may impact the generalizability and practical application of these results. In comparison to prior meta-analyses on the use of PCT, this report focused on a higher acuity patient population. The majority of the patients were in an ED or ICU setting, and patients were only included if they met SIRS criteria. Although this may be an appropriate population to focus on given the difficulty of differentiating sepsis from SIRS in critically ill patients, the absence of healthy controls does limit the applicability of these findings to healthy patients in both the in-patient and out-patient setting.
In addition, the values of sensitivity and specificity varied greatly among the 30 studies. The confirmation of infection in each study was made using several different criteria, resulting in a heterogeneous approach to diagnosing sepsis. This introduces the potential for interpersonal variation, particularly for diagnoses made based on radiographic findings. As a result, some patients may have been falsely categorized into the sepsis or SIRS group, leading to the increased likelihood of false positive and false negative results.
Finally, there was a high degree of heterogeneity among the cohort of studies included in this meta-analysis. Despite sub analysis of the covariates, there remained a large degree of unexplained heterogeneity, indicating the possibility of additional unrecorded characteristics underlying the variation. Moreover, all of these reports varied greatly in the PCT cutoffs, making it difficult to determine a practical cutoff for use in the clinical setting. The authors suggest a cutoff range of 1-2ng/ml, but indicate that using multiple cutoffs with individual likelihood ratios may actually be superior to using a single cutoff.
Overall, this meta-analysis showed that PCT may be an effective and accurate marker for differentiating sepsis from SIRS. By including a broader range of patients, specifically pediatric patients, the authors further validated the general use of PCT.
Hoeboer SH, et al. The diagnostic accuracy of procalcitonin for bacteraemia: a systematic review and meta-analysis. Clin Microbiol Infect. 2015;21:474-481.
In this systematic review and meta-analysis, Hoeboer and colleagues use data from 58 studies of PCT levels in patients with bacteremia to determine its accuracy. The decision to include patients with bacteremia as opposed to sepsis was that this group is the most “robustly defined.” The authors considered for inclusion both prospective and retrospective studies performed through June, 2014. In total 16,514 patient cases were reviewed of which 3,420 had bacteremia. Only patients with documented bacteremia and a PCT level measured within 24 hours of admission were included. The authors calculated the area under the summary ROC curve and pooled sensitivity and specificity. The authors chose a PCT cut-off of 0.5ng/mL based on manufacturer guidelines and recent research.
The overall area under the SROC curve was 0.79, and the authors calculated that with a cutoff of 0.5ng/mL, the sensitivity and specificity of PCT were 76% and 69%, respectively. The lowest AUC was found in immunocompromised patients (0.71), and the highest AUC was seen in ICU patients (0.88). The lowest sensitivity and specificity both followed the same pattern: lowest in immunocompromised patients (66% and 55%, respectively) and highest in ICU patients (89% and 78%, respectively). Despite the heterogeneity in the overall analysis and in most subgroups, there was no suggestion of a threshold-effect.
Hoeboer and colleagues conclude that low PCT levels can be used to rule out bacteremia. However, the study had a large amount of heterogeneity. While sensitivity in ICU patients was 89%, authors found a much lower sensitivity in immunocompromised patients (66%) making this test much less useful in this patient population. Lastly, this study focused on patients with bacteremia and cannot be used to draw conclusions regarding PCT levels in those with other severe infections.
Liu D et al. Prognostic Value of Procalcitonin in Adult Patients with Sepsis: A Systematic Review and Meta-Analysis. PloS ONE 10(6):e0129450. Doi:10.1371/journal.pone.0129450.
Liu and colleagues note that mortality rates from sepsis remain as high as 30-60% and that clinicians have few useful tools to risk stratify these patients. Noting that elevated PCT levels have been associated with all-cause mortality in septic patients, the authors performed a systematic review of all publications that assessed the prognostic value of PCT testing in adult patients with sepsis. All patients were age 18 or older. Ultimately, 23 studies from 2000 to 2014 fulfilled eligibility criteria and were included.
Sixteen studies with 3126 patients examined the correlation between a single PCT value and mortality. The authors calculated that the pooled relative risk (RR) of death for patients with an elevated PCT level was 2.6 (95% CI 2.05-3.30). Pooled sensitivity was 0.76 (95% CI, 0.67-0.82) and specificity was 0.64 (95% CI, 0.52-0.74). The positive likelihood ratio was 2.1 (95% CI, 1.6-2.8) and negative likelihood ratio was 0.38 (95% CI, 0.29-0.51). The overall area under the SROC curve was 0.73 (95% CI 0.69-0.77). Because of the large amount of heterogeneity, the authors concluded that while initial PCT levels are associated with mortality in septic patients this has little prognostic value.
Nine studies with 868 patients examined the association between PCT non-clearance and mortality. The pooled RR for mortality in patients with PCT non-clearance was 3.05 (95% CI, 2.35-3.95. Pooled sensitivity was 0.72 (95% CI, 0.58-0.82) and specificity was 0.77 (95% CI, 0.55-0.90). The positive likelihood ratio was 3.1 (95% CI, 1.5-6.3) and negative likelihood ratio was 0.37 (95% CI, 0.25-0.55). The area under the SROC curve was 0.79 (95% CI, 0.75-0.83). There was less heterogeneity between these studies (I2 = 37.9%) than in the prior group. The authors conclude that PCT non-clearance can be a useful index to predict outcomes in sepsis. They also note that the area under the SROC curve (0.79) in this study is higher than values found in meta-analyses of both troponins and lactate clearance for prognostication in septic patients.
The authors conclude that both single PCT levels and PCT non-clearance are strongly associated with all-cause mortality in septic patients and that PCT non-clearance can predict sepsis mortality. However, optimal cut-off values and timelines still need to be defined.
There were many limitations to this review, including substantial heterogeneity between the studies looking at single PCT levels (I2=63.5%). Although most of the studies measured PCT levels within 24 hours of sepsis diagnosis, some of the included studies used levels drawn up to six days after diagnosis. While the PCT clearance studies had less heterogeneity, the sample size (868 patients) was much smaller. In addition, follow-up periods differed significantly, with some studies following patients out to one month, while others only assessed them during their length of stay in the hospital or ICU. Many of these studies were retrospective, and only three were blinded. Lastly, very few studies looked at ED patients, thus limiting the evidence available to support the use of this test as a prognostic marker in the ED. While PCT clearance may add to the overall assessment of septic patients, the exact cut-off values and mortality benefits have yet to be defined.
In conclusion, infectious diseases are common and antibiotic therapy remains a cornerstone of treatment. However, background selective pressures and inappropriate use of antimicrobial therapy are leading to the emergence of multidrug resistant organisms and the potential for patient harm. As a result, there is an increasing need for the identification of biomarkers that can identify bacterial infections and provide prognostic data. Accurate identification of conditions that would not benefit from antimicrobial therapies can decrease antibiotic use and improve patient outcomes. Although limitations exist and more research is needed, the current literature suggests that PCT directed antimicrobial stewardship may play a role in optimizing the care of critically ill patients.