Biological markers and diagnosis of ventilator-associated pneumonia

Evaluation of a new biomarker from bronchoalveolar fluid, the Clara cell protein 10, adds data to the search for a diagnostic marker for ventilator-associated pneumonia (VAP). For more than 15 years, investigators tried to identify such a marker for predicting or diagnosing VAP. Unfortunately, the results of a number of these studies are disappointing. For optimal management of critically ill, ventilated patients with clinical suspicion of VAP, clinicians need accurate microbiological information to decide to treat in case of confirmed infection and to guide the initial choice of antibiotic therapy with identification of the responsible pathogen(s). Thus, today, the potential advantages of biomarkers are to improve the rapidity and performance of current diagnostic procedures and to reduce antibiotic exposure and selective pressure.

Until now, it has usually been considered that the best way to diagnose and appropriately treat bacterial infection was to identify the bacteria(s) responsible for the infection. Th is statement is accepted without debate for nearly all infections, whether they are severe (such as meningitis) or not (such as urinary tract infection). Th e only question is about the delay for obtaining results with the necessity, in some cases, of initiating urgent empirical antibiotic treatment. In contrast, an endless debate exists about the diagnosis of ventilator-associated pneumonia (VAP). Th is debate is not only about clinical evaluation alone versus clinical suspicion plus microbiological confi rmation, but also, in the case of bacteriologic diagnosis, about the choices between invasive and noninvasive methods, bronchoscopic or non-bronchoscopic sampling techniques, and quantitative or non-quantitative bacterial cultures. Recently, the use of biomarkers has been proposed and tested to replace or complete (or both) such techniques [1][2][3][4]. VAP is diffi cult to diagnose in the critically ill patient because of the presence of underlying cardiopulmonary disorders (for example, atelectasis, acute respiratory distress syndrome, congestive heart failure, pulmonary contusion, and pulmonary hemorrhage), the nonspecifi c radio graphic and clinical signs associated with this infection, and the rapid invasion of the normally sterile lower respiratory tract by microorganisms in all patients with an endotracheal tube, requiring that colonization be diff erentiated from true infection [2]. Actually, there is little information about actual clinical practice; individual studies indicate that practice often diff ers from recommen dations in the literature [5]. In contrast, recom men ded design features of future clinical trials of antimicrobial agents for VAP -elaborated by the Infectious Disease Society of America, the American Th oracic Society, the Society of Critical Care Medicine, and the American College of Chest Physicians -stress the requirement of appropriate microbiological confi rmation of infection by deep lower respiratory tract culture for inclusion in the intention-to-treat population of such trials [6].
Th ese diffi culties for diagnosing VAP and identifying responsible pathogen(s) lead to potential over-or underprescription of antibiotics and misguided treatment. Th e strong associations between delayed diagnosis and subsequent delayed initiation of treatment or between inappropriate initial antibiotic therapy and worse outcome are well documented [7]. On the other hand, a recent observational study conducted in 303 patients at risk for multidrug-resistant pneumonia suggests that guideline-compliant antibiotic treatment was associated with increased mortality in comparison with nonguideline-compliant treatment [8].
Th e major potential advantage of biomarkers is not to diagnose VAP by itself but to potentially improve the rapidity and performance of current diagnostic procedures. Th e rationale is to identify and, if necessary, to dose specifi c markers of alveolar infection (not of bronchial colonization): either endogenous mediators released

Abstract
Evaluation of a new biomarker from bronchoalveolar fl uid, the Clara cell protein 10, adds data to the search for a diagnostic marker for ventilator-associated pneumonia (VAP). For more than 15 years, investigators tried to identify such a marker for predicting or diagnosing VAP. Unfortunately, the results of a number of these studies are disappointing. For optimal management of critically ill, ventilated patients with clinical suspicion of VAP, clinicians need accurate microbiological information to decide to treat in case of confi rmed infection and to guide the initial choice of antibiotic therapy with identifi cation of the responsible pathogen(s). Thus, today, the potential advantages of biomarkers are to improve the rapidity and performance of current diagnostic procedures and to reduce antibiotic exposure and selective pressure.
Th e article by Vanspauwen and colleagues [1] in the previous issue of Critical Care adds new data to the search for a biomarker for VAP. Th e investigators evaluated a low-molecular-weight protein secreted in the alveoli, the Clara cell protein 10 (CC-10), which has a possible anti-infl ammatory role in humans [1]. Bronchoscopy with BAL was performed in 196 patients who were treated with mechanical ventilation for more than 48 hours and who had clinical suspicion of VAP. Th e authors found, in the microbiologically confi rmed VAP group, a median CC-10 concentration that was significantly higher than in the non-VAP group (3,019, range 282 to 65,546 versus 2,054, range 62 to 30,240 ng/mL; P = 0.03), but such CC-10 levels are not discriminative for diagnosing VAP. No diff erence between patients with corticosteroids or antibiotic therapy and those without or between patients with Gram-positive and those with Gram-negative bacteria as responsible pathogens was observed.
In this well-conducted study, some strengths and weaknesses are worth mentioning. First, to evaluate new diagnostic techniques, a reference standard is required. For VAP, because histological confi rmation is not a practical reference, quantitative cultures of BAL fl uid are the best possible alternatives [2]. Second, bronchoscopy and lavage may be contraindicated in some patients and remain operator-dependent and time-consuming; in addition, high variability in the use of bronchoscopic sampling across intensive care units exists [5]. Th ird, an ideal diagnostic test for VAP would involve a highly discriminatory, easy-to-sample blood test. In fact, biomarkers potentially used in diagnosing VAP, including CC-10, are BAL-based techniques, just like quantitative cul tures and direct examination of BAL fl uid. Consequently, such a new technique is probably an extension of BAL fl uid analysis in addition to microbiological techniques rather than a simplifi cation of diagnostic procedures for VAP. Fourth, as usual in this type of study, the precise timing of sampling may infl u ence the diagnosis of VAP. Results of CC-10 dosage in the group of non-VAP patients with a distinction between patients with negative cultures and those with more than 0 and less than 10 4 colony-forming units per milliliter (or less than 2% intracellular organisms) and data concerning the follow-up of such patients (particularly the occur rence of subsequent pneumonia) could give additional information about the ability of CC-10 dosage to predict VAP. Finally, this study did not include sequential CC-10 measurements, although recent data concerning PCT suggest that serial dosages may be used as a marker to monitor response and possibly to termi nate antibiotic therapy [21]. Today, even if interest in surrogate markers (that is, BAL fl uid biomarker dosage) to resolve the question of diagnosing or predicting VAP is not demonstrated, interest in the continuing story of biomarkers and VAP is certain.