Our data demonstrate that the PCR-based SF test might represent a rational adjunct tool to the traditional BC method for rapid etiologic diagnosis of BSI in patients after cardiothoracic surgery. SF detects significantly more Gram-negative microorganisms than BC whereas BC was superior regarding Gram-positive pathogens
Early and reliable diagnosis of BSI and identification of bacteria and fungi is essential to initiate appropriate therapy in septic patients within one hour after sepsis as recommended by current guidelines.[6] For decades, detection of pathogen microorganisms in patients with suspected BSI was mainly based on BC. However, the classical BC procedure per se has two intrinsic limitations: Firstly, this method is limited by the delay of identification of the pathogen and the antimicrobial susceptibility profile of 24 to 72 hours in most of the cases. Although blood cultures incubated in modern instrumented systems typically signal positive in a median time of 12–36 h. In addition, approximately 30% of pathogens remain undetected by BC and the time to positivity from collection to detection is longer for some fastidious bacteria, anaerobes, and fungi or under antimicrobial therapy.[14] Thus, there is an urgent need to improve the diagnostic tools for an improved management of patients with BSI or sepsis. Molecular methods offer distinct advantages over blood cultures, including increased sensitivity and rapid diagnosis. However, diagnostic accuracy and cost–effectiveness should be established before implementation in clinical practice. SF runs on the LightCycler® instrument, the first real-time PCR-based system to be awarded a Conformité Européenne (CE) mark for pathogen detection and identification in suspected bloodstream infection and is, to date, most intensively investigated in clinical studies.[7, 15]
A meta-analysis including a total of 34 studies enrolling 6012 patients with suspected sepsis reported a high specificity with a modest and highly variable sensitivity.[16] Recently published studies revealed a low sensitivity of the PCR method accompanied with a limited utility for the diagnosis of healthcare-associated BSI in critical care patients.[17] In contrast, another study including 104 critically ill patients suffering from SIRS showed that in 25 cases (16.9%, n=148) rapid identification of involved pathogens by multiplex-PCR led to adjustment of therapy.[18] A recently published randomized controlled trial enrolling 78 adults with suspected pulmonary or abdominal source of a systemic infection demonstrated a significant reduction in the time required for initial pathogen identification with SF compared with standard BC.[8] Taken together, the results of studies available so far about the usefulness of the SF for rapid detection of BSI in critical ill patients are divergent. According to the review of Dark, the performance of SF in different patient cohorts and clinical settings may vary significantly requiring careful and specific evaluation of the use of SF in respect of the clinical situation.[19] Patients after cardiothoracic surgery significantly differ from other cohorts: The use of cardiopulmonary bypass leads to a damage of the gastrointestinal mucosa, subsequent increased permeability, possible bacteremia, and the activation of a self-limited inflammatory response. The incidence of fungal infections especially in transplant recipients is, due to immunosuppression, higher than in the general ICU population. Commonly used biomarkers for bacterial infection, e.g. procalcitonin, might not work properly in the cardiothoracic population.[20]
However, regarding the utility of SF for the detection of BSI in patients after cardiothoracic surgery there is a paucity of data: Using SF for analyzing heart valve tissue of patients with active infectious endocarditis, the sensitivity of the SF was 100% and SF could identify pathogens in cases where BC tested negative.[21] In a recently published observational study analyzing 130 blood samples from 30 thoracic allograft recipients (23 heart and 7 lung transplants) showed a superiority of PCR compared to BC with a significantly higher number of positive samples in SF.[22]
In accordance with previous studies, the results of the present study demonstrate that SF, compared to BC, provided a better management of contaminants and a lower contamination rate. .[23] In respect of CoNS interpretation and discrimination in BC clinical judgment must be used due to a lack of objective criteria. In contrast, in SF an automated software is used to identify contaminants, which explains the lower rate of contaminants.
In accordance with recently published data, we observed a clear superiority of SF in detecting Gram-negative organisms compared to conventional BC. [24]
The reason for the discrepancy between the detection rate of Gram-negative and Gram-positive pathogens is unclear. Recently published studies could demonstrate that the superiority of SF over BC is particularly observed in very ill patients with severe sepsis.[25] In our cohort, patients with Gram negative BSI had higher concentration of CRP, IL-6 and PCT as well as a higher incidence of AKI with need for RRT compared to those with Gram-positive pathogens. Therefore, it might be hypothesized that SF is superior in detecting Gram-negative pathogens not in general but in critically ill patients with severe infections.
BSI caused by Gram-negative bacteria is associated with a 7-fold increased risk of early mortality after cardiac surgery after adjusting for other covariates, compared with no BSI.[5] In contrast, BSI caused by Gram-positive bacteria other than S. aureus was only associated with a 2.2-fold increased risk of mortality. These data are consistent with other studies reporting an increased mortality associated with BSI due to Gram-negative bacteria and lower attributable mortality associated with BSI due to Gram-positive ones.[26] Against this background, the early detection of Gram-negative bacteria in SF is of tremendous clinical relevance and the identification of additional pathogens with SF might help to improve survival of patients with Gram-negative bacteremia.
Since invasive fungal infections with Aspergillus are frequently associated with high morbidity and mortality, in particular immunocompromised patients benefit from prompt initiation of anti-fungal therapy.[27] However, the Surviving Sepsis Campaign does not recommend the routine use of empirical antifungals, based on the relatively low frequency of fungal causation of sepsis (∼5% of cases), although this is likely to rise. In our cohort of patients, a notably but not significant higher number of Aspergillus amplicons were detected by PCR as compared with blood culture.
Due to the long incubation times required and the lack of sensitivity early diagnosis of fungal infections by BC is often difficult and may delay therapy. Therefore, using SF could improve patient outcome as a result of rapid and accurate fungi detection and the consecutive timely initiation of appropriate therapy. [28] Hence, one important clinical impact of SF seems to be the identification of otherwise undetected fungal BSI.
However, SF was inferior to BC in detecting Gram-positive bacteria including S. aureus representing an important pathogen associated with high mortality.
Coagulase-negative staphylococci (CoNS) are a major constituent of human skin commensal flora, which were once considered relatively apathogen and a likely contaminant. But in patients with prosthetic valves, pacemakers, defibrillators, ventricular assist devices, intravascular catheters, or other foreign bodies these organisms have increasingly been recognized as a cause of clinically significant infections. Due to the propensity of these organisms to colonize foreign material to form a biofilm and to display resistance to multiple antibiotics, infections with CoNS are difficult to treat. Thus, due to the significant number of infections that would be missed, it does not appear SF could replace blood culture for the identification of bloodstream infections, especially in patients after cardiac surgery.
In addition, in SF pathogen identification is restricted to the 25 tested microorganisms and, moreover, susceptibility testing is not possible of and. Therefore, SF cannot replace BC but represents an adjunct tool in combination with BC. Even though in our study antimicrobial therapy was escalated due to the results of SF in eight patients, no de-escalation was done. As most of our patients were already on broad spectrum antibiotics and several blood cultures were drawn before choosing SF as diagnostic tool, empirical antibiotic therapy was considered to be adequate for most of the pathogens detected in SF. It might be hypothesized that the impact on therapy may be even more pronounced if SF would have been used earlier in the treatment course. De-escalation of therapy based on PCR might be difficult in general due to the inability of susceptibility testing.
Recently published studies could demonstrate that use of new PCR based technologies in the management of septic patients lead to a significant reduction in treatment costs with a an average net saving of 9970 € per patient.[29] This economic benefit is mainly based on shortening of intensive care unit stay and the use of fewer antibiotics. However, the costs of SeptiFast (approximately 200-300 USD) are high compared to Blood Culture (approximately 30 USD). Therefore, we assessed the predictive value of clinical and laboratory data to restrict the SF assay to clinical cases with a high probability of positive SF results. There is only limited data available in the literature about predictors for SF positivity:
Mencacci investigated the predictive role of procalcitonin in patients with suspected sepsis for positive test results in BC and PCR and revealed an area under the curve of 0.927 for SF positivity.[30] When applying a cut-off value of 0.37 ng/ml, the number of SF assays could be reduced by 53.9% with identifying 96.4% of pathogens. Leli et al. identified increased procalcitonin or white blood cells, fever >38°C, and low serum albumin as independent predictors of positive SF results in blood samples taken within 12h after the onset of fever in 285 patients.[31] A prospective observational cohort study including neurosurgical patients with external ventricular drainage (EVD) identified the intrathecal concentrations of IL-6 to be a reliable predictor of pathogen identification in SF. [32]
In our cohort, IL-6 as well as CRP was good predictors for SF positivity. Although measurement of PCT concentration are considered to be the gold standard of systemic inflammatory markers for diagnosis as well as for evaluation of the treatment effectiveness, PCT only showed moderate predictive capabilities. The discrepancy could be due to the following: It is well established that aortic cross clamp and cardiopulmonary bypass related perioperative stress is associated with elevated PCT after cardiac surgery.[33] Against this background, several studies showed a poor correlation between elevated PCT concentration and bacterial infections or sepsis after major cardiac surgery.[34] In addition, renal function is a major determinant of procalcitonin concentration and the value of PCT might be limited in patients with renal impairment or RRT.[35] Another aspect is that in our study 12 out of 47 positive SF results identified fungal pathogens. Thus, PCT as marker of bacterial infections is, anyway, not suitable for prediction of SF positivity in our cohort even more. Although the correlation of biomarkers and SF results are not very strong, in respect of the high costs it might be helpful in the decision to perform SF or not.
Limitations
There are several potential limitations to this study. First, our study suffers from the general limitation of a single-center, retrospective investigation: patients were recruited from one cardiothoracic ICU of a university hospital, and consequently, the results may not be applicable to other clinical settings with different patient characteristics, resources, and laboratory procedures. In addition, due to the small number of specific pathogens, the power to detect a difference between the groups is limited.
In interpreting the results of this study heterogeneity in the methods of drawing blood samples for BC must be considered as a limitation. It could not be ensured that all collected samples complied with the guidelines for drawing blood samples for BC, what can affect both for sensitivity and specificity.[36]
A major limitation is the fact that there were no predefined criteria for performing PCR e.g. presence of more than two SIRS criteria. The decision to perform SF was made by the consultant in charge of the ICU at the time or the infectiologist. The algorithms used in this study to differentiate between contamination and infection of BC and SF were not evaluated in the cardiothoracic population. Therefore, the reliability of this algorithm in this setting is uncertain. However, as there is no published algorithm for cardiothoracic patients we modified the originally published algorithm to incorporate specific characteristics of our patient’s cohort e.g. the presence of prosthetic heart valves or other extracorporeal devices.
Due to the retrospective nature of our study we could not ensure that the same blood sample was used for SF and BC.
It has to be mentioned that the SF test is not available in the United States yet.