b-glucan Antigenemia Anticipates Diagnosis of Blood Culture Negative Intraabdominal Candidiasis

Transcription

1 b-glucan Antigenemia Anticipates Diagnosis of Blood Culture Negative Intraabdominal Candidiasis Frederic Tissot 1, Frederic Lamoth 1, Philippe M. Hauser 2, Christina Orasch 1,3, Ursula Flückiger 3, Martin Siegemund 4, Stefan Zimmerli 5, Thierry Calandra 1, Jacques Bille 2, Philippe Eggimann 6 *, Oscar Marchetti 1 *, and the Fungal Infection Network of Switzerland (FUNGINOS) 1 Infectious Diseases Service, Department of Medicine, 2 Institute of Microbiology, and 6 Adult Intensive Care Service, Lausanne University Hospital, Lausanne, Switzerland; 3 Division of Infectious Diseases and Hospital Epidemiology and 4 Intensive Care Service, Basel University Hospital, Basel, Switzerland; and 5 Institute for Infectious Diseases, University of Bern, Bern, Switzerland Rationale: Life-threatening intraabdominal candidiasis (IAC) occurs in 30 to 40% of high-risk surgical intensive care unit (ICU) patients. Although early IAC diagnosis is crucial, blood cultures are negative, and the role of Candida score/colonization indexes is not established. Objectives: The aim of this prospective Fungal Infection Network of Switzerland (FUNGINOS) cohort study was to assess accuracy of 1,3-b-D-glucan (BG) antigenemia for diagnosis of IAC. Methods: Four hundred thirty-four consecutive adults with abdominal surgery or acute pancreatitis and ICU stay 72 hours or longer were screened: 89 (20.5%) at high risk for IAC were studied (68 recurrent (Received in original form November 19, 2012; accepted in final form June 3, 2013) * These authors contributed equally to this work. The FUNGINOS Foundation received unrestricted grant support from Essex Schering-Plough Switzerland; Gilead Switzerland; Merck, Sharp and Dohme- Chibret Switzerland; Novartis Switzerland; and Pfizer Switzerland. The present project was supported by the following funding sources: Associates of Cape Cod U.S.A.: discounted Fungitell lab kits; Biomérieux France: unrestricted grant support; European Community s Seventh Framework program (FP ) under grant agreement n_health-f allfun; Foundation for the Advancement in Medical Microbiology and Infectious Diseases (FAMMID), Lausanne, Switzerland: unrestricted grant support. None of the funding sources has been involved in study design and conduct; patient recruitment; data collection, analysis, and interpretation; writing of the manuscript; or decision to submit the manuscript for publication. Author Contributions: F.T.: screening and inclusion of eligible patients; data collection, management, analysis, and interpretation; literature search; writing of the manuscript. F.L.: revision of the study protocol; screening and inclusion of eligible patients; data collection, management, and interpretation; literature search; revision of the manuscript. P.M.H.: laboratory analyses and interpretation of results, revision of the manuscript. C.O.: screening and inclusion of eligible patients, data collection and interpretation, revision of the manuscript. U.F.: submission to the Institutional Ethical Board, screening and inclusion of eligible patients, data collection and interpretation, revision of the manuscript. M.S.: screening and inclusion of eligible patients, data collection and interpretation, revision of the manuscript. S.Z.: study design, writing of the study protocol, writing of research grants, data analysis and interpretation, revision of the manuscript. T.C.: study design, writing of research grants, data analysis and interpretation, revision of the manuscript. J.B.: study design; data collection, analysis, and interpretation; writing of the manuscript. P.E.: study design; screening, inclusion, and follow-up of eligible patients and control subjects with bacterial severe sepsis/septic shock; review of individual patients case report form; results interpretation; revision of the manuscript. O.M.: study design, writing of the study protocol and submission to the Institutional Ethical Board, writing of research grants, data analysis and interpretation, literature search, writing and final revision of the manuscript. Correspondence and requests for reprints should be addressed to Oscar Marchetti, M.D., Infectious Diseases Service, Department of Medicine, Lausanne University Hospital, Rue du Bugnon 46, CH-1011 Lausanne, Switzerland. oscar. marchetti@chuv.ch This article has an online supplement, which is accessible from this issue s table of contents at Am J Respir Crit Care Med Vol 188, Iss. 9, pp , Nov 1, 2013 Copyright ª 2013 by the American Thoracic Society Originally Published in Press as DOI: /rccm OC on June 19, 2013 Internet address: AT A GLANCE COMMENTARY Scientific Knowledge on the Subject Invasive candidiasis is a severe complication in the hospital setting. Blood culture negative intraabdominal candidiasis is the most frequent form of this life-threatening infection in nonneutropenic patients. Early antifungal therapy is critical for outcome, but conventional culture-based tools lack diagnostic accuracy and provide results after days of incubation. Prophylactic or empirical antifungal agents are thus largely used. The role of Candida score, colonization index, and corrected colonization index for management of patients with intraabdominal candidiasis remains to be established. Clinicians urgently need alternative noninvasive and rapid diagnostic tools. Recent metaanalyses have shown that b-glucan (BG) antigenemia allows the early detection of invasive candidiasis and aspergillosis in hematological patients with neutropenia. BG has thus been included among the microbiological diagnostic criteria of the updated European Organization for the Research and Treatment of Cancer Mycoses Study Group classification of invasive mycoses in immunocompromised hosts. In nonneutropenic patients, BG has been shown to be accurate for the detection of Candida bloodstream infections. However, the utility of this fungal biomarker in blood culture negative intraabdominal candidiasis is unknown. What This Study Adds to the Field This prospective observational Fungal Infection Network of Switzerland (FUNGINOS) cohort study adds to existing evidence on the diagnostic accuracy of BG antigenemia. Detection of this fungal blood biomarker is accurate for anticipating diagnosis of blood culture negative intraabdominal candidiasis and is superior to Candida score, colonization index, and corrected colonization index. BG values reflect severity of infection, and BG kinetics may predict response to antifungal therapy and clinical outcome. This proof-of-concept observation opens the way to individualized management of surgical intensive care unit patients at high risk for intraabdominal candidiasis. A preemptive approach triggered by a positive BG result may target prompt initiation of antifungal therapy to patients in whom infection is identified at an early stage. This new straightforward strategy is based on the use of a simple blood fungal biomarker in a strictly selected clinical setting where traditional microbiological diagnostics have failed. It is an attractive approach for future clinical investigations aimed at maximizing chances of success while minimizing unnecessary antifungal exposure.

2 Tissot, Lamoth, Hauser, et al.: b-glucan in Intraabdominal Candidiasis 1101 gastrointestinal tract perforation, 21 acute necrotizing pancreatitis). Diagnostic accuracy of serum BG (Fungitell), Candida score, and colonization indexes was compared. Measurements and Main Results: Fifty-eight of 89 (65%) patients were colonized by Candida; 29 of 89 (33%) presented IAC (27 of 29 with negative blood cultures). Nine hundred twenty-one sera were analyzed (9/patient): median BG was 253 pg/ml (46 9,557) in IAC versus 99 pg/ml (8 440) in colonization (P, 0.01). Sensitivity and specificity of two consecutive BG measurements greater than or equal to 80 pg/ml were 65 and 78%, respectively. In recurrent gastrointestinal tract perforation it was 75 and 77% versus 90 and 38% (Candida score > 3), 79 and 34% (colonization index > 0.5), and 54 and 63% (corrected colonization index > 0.4), respectively. BG positivity anticipated IAC diagnosis (5 d) and antifungal therapy (6 d). Severe sepsis/septic shock and death occurred in 10 of 11 (91%) and 4 of 11 (36%) patients with BG 400 pg/ml or more versus 5 of 18 (28%, P ¼ 0.002) and 1 of 18 (6%, P ¼ 0.05) with BG measurement less than 400 pg/ml. b-glucan decreased in IAC responding to therapy and increased in nonresponse. Conclusions: BG antigenemia is superior to Candida score and colonization indexes and anticipates diagnosis of blood culture negative IAC. This proof-of-concept observation in strictly selected highrisk surgical ICU patients deserves investigation of BG-driven preemptive therapy. Keywords: Candida; colonization; ICU; surgery; biomarkers Invasive candidiasis is a frequent infection in critically ill patients (1, 2). Some 30 to 40% of episodes of recurrent gastrointestinal (GI) tract perforation or acute necrotizing pancreatitis are complicated by intraabdominal candidiasis (IAC) (3 10). In this setting, a significant morbidity and a mortality of 52 to 63% have been reported (3, 4, 11). Delay in initiation of antifungal therapy is a major determinant of clinical outcome (4, 12). Clinical signs of IAC are not specific, and early microbiological documentation remains a major challenge. Cultures from nonsterile sites are frequently positive but lack specificity for differentiating infection from colonization (10). Blood cultures are positive in a minority of patients, and results of intraabdominal cultures are available late in the course of infection. Intensive research is dedicated to the development of alternative tools for early diagnosis of IAC (8, 13, 14). Candida score (CS), colonization index (CI), corrected colonization index (CCI), and predictive rules help in discriminating Candida colonization from infection (8, 13, 14). Excellent negative predictive values (NPVs) (.90 95%) and low positive predictive values (PPVs) (,15% for CS and predictive rules) of these parameters have been reported in a mixed medical and surgical intensive care unit (ICU) population, whereas their role in patients with IAC remains to be established (8, 13, 14). New approaches for early diagnosis of IAC are urgently needed to guide antifungal therapy on a preemptive basis. Laboratory tests detecting fungal cell components and/or antibodies directed against these components have been investigated as diagnostic biomarkers of invasive candidiasis (15 18). 1,3-b-D-glucan (BG) is a fungal cell wall antigen circulating in blood of patients with invasive mycoses. Sensitivity of 56 to 93% and specificity of 71 to 100% have been reported for diagnosis of candidemia (19 23). Limited data are available on BG in patients with blood culture negative deep-seated candidiasis, in particular postsurgical IAC, one of the most difficult-to-diagnose forms of this life-threatening infection (8, 22, 24 26). This Fungal Infection Network of Switzerland (FUNGINOS) cohort study aimed at prospectively assessing accuracy of serum BG compared with culture-based documentation, CI, CCI, and CS for early diagnosis of blood culture negative IAC in highrisk surgical ICU patients, and BG kinetics according to severity, response to therapy, and outcome of IAC. Preliminary results of this study have been presented in the form of abstracts (27 29). METHODS Study Design This observational FUNGINOS cohort study was conducted in ICUs of two Swiss University Hospitals (Basel, Lausanne). Consecutive patients Figure 1. Screening for eligibility and inclusion of patients with abdominal surgery or acute pancreatitis and intensive care unit (ICU) stay of 72 hours or more. GI ¼ gastrointestinal; IAC ¼ intraabdominal candidiasis.

3 1102 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL TABLE 1. PATIENT DEMOGRAPHICS AND CLINICAL CHARACTERISTICS TABLE 2. RISK FACTORS FOR Candida COLONIZATION AND INFECTION Value (n ¼ 89) Sex, male/female 59 (66)/30 (34) Age, median yr (range) 62 (22 86) Primary diagnosis at ICU admission: Intraabdominal tumor 23 (26) Intestinal ischemic disorder 20 (22) Acute necrotizing pancreatitis 20 (22) GI perforation 10 (11) GI bleeding 5 (6) Ruptured aneurysm of abdominal aorta 4 (4) Others* 7 (8) Inclusion criteria Recurrent GI tract perforation 68 (76) Acute necrotizing pancreatitis 21 (23) Total hospital stay, median (range), d 44.5 (9 176) Hospital stay before inclusion, median (range), d 8 (1 74) Total ICU stay, median (range), d 13 (3 74) ICU stay before inclusion, median (range), d 2 (0 54) Abdominal surgery during study 78 (87) No. of operations, median (range) 3 (0 9) Sites of abdominal surgery during study (>1 site/patient) Stomach 7 (8) Small intestine 42 (47) Colon 42 (47) Biliary tract 18 (20) Pancreas 17 (19) Bacterial infections during study No. of patients 86 (97) No. of episodes x 151 Severity at inclusion SAPS II, median (range) 51 (13 87) APACHE II, median (range) 23 (5 37) Severe sepsis or septic shock 50 (56) Mortality 15 (17) Definition of abbreviations: APACHE ¼ Acute Physiology and Chronic Health Evaluation; GI ¼ gastrointestinal; ICU ¼ intensive care unit; SAPS ¼ Simplified Acute Physiology Score. Data are presented as No. (%) unless otherwise noted. * Cholecystitis (2), rectal prolapse (1), prostatic cancer (1), gastric bypass (1), pancreatic pseudocyst (1). y Causes of recurrent GI tract perforation were: anastomotic leakage (34), ischemic necrosis (14), or recurrent surgery for other local complications (20). z Causes of pancreatitis were: biliary (7), alcoholic (7), trauma (2), and unknown (5). x Documented pathogens: gram-positive bacteria (24%), gram-negative bacteria (39%), anaerobes (7%), mixed infections (30%). with abdominal surgery or acute pancreatitis admitted to the ICU for 72 hours or more were screened (August 1, 2007 to January 31, 2010). Based on previous observations, those with recurrent GI tract perforation (anastomotic leakage, ischemic necrosis, recurrent GI surgery for other local complications) or acute necrotizing pancreatitis (Balthazar grade D-E) not receiving antifungal agents were included and prospectively studied until 2 weeks after ICU discharge (3, 9, 30). Demographic, clinical, and microbiological data were recorded. Institutional Ethical Committees approved the study; written informed consent was obtained from the patients or their legal representatives. The study was designed, analyzed, and reported according to Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) and Standards for the Reporting of Diagnostic Accuracy Studies (STARD) statements for observational diagnostic accuracy studies ( strobe-statement.org, CS and Colonization Indexes Samples from at least three of five nonsterile sites (mouth, urine, stools, skin, and/or respiratory tract) were cultured twice weekly. Candida colonization was defined by recovery of Candida spp. from at least one site and graded as weak, moderate, or heavy by semiquantitative cultures (3). CS, CI, and CCI were calculated at inclusion and twice weekly during the study; they were not used for patients grouping or therapeutic decisions (8, 13). Diagnosis and Management of Candida Infection Value (n ¼ 89) Risk factors for Candida infection at inclusion Central venous catheter 87 (98) Proton-pump inhibitor 86 (97) Urinary catheter 86 (97) Total parenteral nutrition 84 (94) Antibacterial therapy 77 (86) Mechanical ventilation. 24 h 61 (68) Renal replacement therapy: CRRT/intermittent HD 16/3 (21) Immunosuppressive therapy* 6 (7) Candida colonization Colonization at any site: at inclusion/during study 75 (84)/87 (98) Colonization index > 0.5: first wk/during study 51 (57)/71 (80) Corrected colonization index > 0.4: first wk/during 34 (38)/49 (55) study Candida score > 3: first wk/during study 44 (49)/78 (88) Candida infection IAC 29 (33) Pure culture from intraoperative specimens 7/29 (24) Mixed-flora abscess 8/29 (27.5) Mixed-flora peritonitis 14/29 (48.5) Secondary candidemia 2/29 (7) Median days from hospital admission to infection 12 (0 74) (range) Median days from ICU admission to infection (range) 7 (0 70) Candida species (>1 isolate/patient) C. albicans 23/29 (79) C. tropicalis 5/29 (17) C. glabrata 3/29 (10) C. kefyr 1/29 (3) C. lusitaniae 1/29 (3) Other non-albicans Candida species 1/29 (3) Severity of Candida infection No sepsis, sepsis 2/29 (7), 12/29 (41) Severe sepsis, septic shock 4/29 (14), 11/29 (38) Mortality 5/29 (17) Antifungal therapy All included patients (n ¼ 89) No therapy 45 (51) Preemptive therapy for suspected IAC 18 (20) Therapy for documented infection 26 (29) Patients with documented IAC (n ¼ 29) Median days of therapy (range) 16 (4 48) Antifungal agent (>1 agent/patient) None 3/29 (6) Fluconazole 26/29 (89) Caspofungin (before or after fluconazole) 6/29 (12) Definition of abbreviations: CRRT ¼ continuous renal replacement therapy; HD ¼ hemodialysis; IAC ¼ Intraabdominal candidiasis. Data are presented as No. (%) unless otherwise noted. Percentages are referred to the total number of included patients (n ¼ 89), unless otherwise indicated. * Two patients in subgroup II, two in subgroup III, and two in subgroup IV received immunosuppressive therapy: antirejection therapy (ciclosporin and mycophenolate mofetil: n ¼ 1; antithymocytes globulins, azathioprine and tacrolimus: n ¼ 1), corticosteroids > 20 mg prednisone equivalent for > 10 d (n ¼ 3), tumor necrosis factor alpha inhibitor (n ¼ 1). Two sets of blood cultures were drawn according to institutional guidelines at onset of fever/suspected infection and in persistent fever/signs of infection despite antibacterial therapy (automated system Bactec 9240, BACTEC Plus aerobic/f and Lytic anaerobic/f bottles; Becton Dickinson, Sparks, MD). Candidemia was defined as at least one blood culture positive for Candida spp. in the presence of clinical symptoms/ signs of infection. Intraabdominal abscess or peritonitis was classified as IAC according to one of the following culture results from

4 Tissot, Lamoth, Hauser, et al.: b-glucan in Intraabdominal Candidiasis 1103 Figure 2. 1,3-b-D-Glucan (BG) values at the time of intraabdominal candidiasis (IAC) diagnosis versus Day 7 after inclusion in patients without IAC. Patient subgroups: noncolonized (Ib, see Methods section), colonized by Candida and not receiving antifungals (II), colonized by Candida and receiving preemptive antifungals for suspected IAC (III), and documented IAC (IV). Labels at the top indicate for each subgroup the proportion of patients with BG above the recommended cut-off (>80 pg/ml). Horizontal bars indicate median values with interquartile ranges. specimens obtained at surgery: (1) monomicrobial growth of Candida spp., (2) growth in any amount of Candida spp. within a mixedflora abscess, (3) moderate or heavy growth of Candida spp. in mixed-flora peritonitis treated with appropriate antibacterial therapy according to susceptibility testing (3, 9). Initiation of antifungal therapy was based on international guidelines: (1) documented Candida infection (see above), (2) preemptive therapy for suspected invasive Candida infection in patients colonized at two or more nonsterile sites and with severe sepsis/septic shock or persistent signs of infection despite more than 48 hours of appropriate antibacterial therapy (31, 32). Antifungal prophylaxis was not recommended, regardless of presence/absence of Candida colonization. BG Measurements Blood samples were drawn at study inclusion and then three times weekly at fixed time points. Sera were stored at 2808C and analyzed in batch with duplicate testing by Fungitell on an ELx808 IU Microplate Reader (Associates of Cape Cod, East Falmouth, MA) as per manufacturer s package insert. Two certified specifically trained laboratory technicians blinded to clinical data performed analyses with BG-free materials under laminar airflow, according to manufacturer s Standard Operating Procedure. According to the Fungitell Training Guide, analyses were repeated when BG duplicates differed by more than 20% (165 of 921, 18%) to minimize the effect of potential environmental contaminations and/or pipetting errors. Samples for BG values above the upper limit of quantification (.523 pg/ml) were diluted (58 of 921, 6%). Mean BG values of duplicates were used for data analysis. BG results were not available in real time for therapeutic decisions. Patients Patients were divided in four subgroups by three investigators blinded to BG results (F.T., P.E., O.M.): noncolonized by Candida (I), colonized by Candida and not receiving antifungals (II), colonized by Candida and receiving preemptive antifungals for suspected IAC (III), documented IAC (IV). As only 2 subjects were neither colonized nor infected (subgroup Ia), 14 noncolonized control subjects with microbiologically documented bacterial peritonitis from a severe sepsis/septic shock ICU cohort were included for comparison of BG values with subgroups II to IV (subgroup Ib). TABLE 3. ACCURACY OF 1,3-b-D-GLUCAN, Candida SCORE, Candida COLONIZATION INDEX, AND CORRECTED Candida COLONIZATION INDEX FOR THE DIAGNOSIS OF INTRAABDOMINAL CANDIDIASIS (N ¼ 29) Sensitivity Specificity PPV NPV Efficiency, % BG > 80 pg/ml 13 At inclusion 0.76 ( ) 0.59 ( ) 0.56 ( ) 0.78 ( ) 66 At infection* 0.83 ( ) 0.40 ( ) 0.49 ( ) 0.77 ( ) 58 BG > 80 pg/ml 23 At inclusion 0.66 ( ) 0.83 ( ) 0.73 ( ) 0.78 ( ) 76 At infection* 0.65 ( ) 0.78 ( ) 0.68 ( ) 0.77 ( ) 73 CS > 3 At inclusion 0.86 ( ) 0.50 ( ) 0.54 ( ) 0.84 ( ) 65 At infection* 0.86 ( ) 0.38 ( ) 0.49 ( ) 0.80 ( ) 58 CI > 0.5 At inclusion 0.26 ( ) 0.76 ( ) 0.35 ( ) 0.67 ( ) 59 At infection* 0.88 ( ) 0.34 ( ) 0.49 ( ) 0.80 ( ) 57 CCI > 0.4 At inclusion 0.14 ( ) 0.77 ( ) 0.23 ( ) 0.65 ( ) 56 At infection* 0.50 ( ) 0.43 ( ) 0.35 ( ) 0.59 ( ) 46 Definition of abbreviations: BG ¼ 1,3-b-D-glucan; CCI ¼ corrected Candida colonization index; CI ¼ Candida colonization index; CS ¼ Candida score; IAC ¼ intraabdominal candidiasis; NPV ¼ negative predictive value; PPV ¼ positive predictive value. Numbers in parentheses represent 95% confidence intervals. * The maximal BG up to time of IAC diagnosis vs. within the first week after inclusion in absence of IAC. y Two consecutive BG values.

5 1104 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL TABLE 4. ACCURACY OF 1,3-b-D-GLUCAN FOR THE DIAGNOSIS OF INTRAABDOMINAL CANDIDIASIS IN PATIENTS WITH RECURRENT GASTROINTESTINAL TRACT PERFORATION VERSUS ACUTE NECROTIZING PANCREATITIS Sensitivity Specificity PPV NPV Efficiency, % Recurrent GI tract perforation (n ¼ 24) BG > 80 pg/ml 13 At inclusion 0.83 ( ) 0.52 ( ) 0.57 ( ) 0.80 ( ) 65 At infection* 0.92 ( ) 0.32 ( ) 0.51 ( ) 0.83 ( ) 58 BG > 80 pg/ml 23 At inclusion 0.75 ( ) 0.84 ( ) 0.78 ( ) 0.81 ( ) 80 At infection* 0.75 ( ) 0.77 ( ) 0.72 ( ) 0.80 ( ) 76 Acute necrotizing pancreatitis (n ¼ 5) BG > 80 pg/ml 13 At inclusion 0.40 ( ) 0.82 ( ) 0.50 ( ) 0.75 ( ) 69 At infection* 0.40 ( ) 0.64 ( ) 0.33 ( ) 0.70 ( ) 56 BG > 80 pg/ml 23 At inclusion 0.20 ( ) 0.82 ( ) 0.33 ( ) 0.69 ( ) 62 At infection* 0.20 ( ) 0.82 ( ) 0.33 ( ) 0.69 ( ) 62 Definition of abbreviations: BG ¼ 1,3-b-D-glucan; GI ¼ gastrointestinal; IAC ¼ intraabdominal candidiasis; NPV ¼ negative predictive value; PPV ¼ positive predictive value. Numbers in parentheses represent 95% confidence intervals. * The maximal BG up to time of IAC diagnosis vs. within the first week after inclusion in absence of IAC. y Two consecutive BG values. Accuracy of BG, CS, CI, and CCI for Diagnosis of IAC BG cut-offs were studied over the analytical range 60 to 500 pg/ml (value recommended by manufacturer: 80 pg/ml). Recommended cut-offs for CS (>3), CI (>0.5), and CCI (>0.4) were used (8, 13). Diagnostic accuracy of a single and of two consecutive BG values above a given cut-off was analyzed at two time points (considering the maximal value measured up to that point): (1) at inclusion, (2) at time of IAC diagnosis (compared with Day 7 after inclusion in patients without IAC, most infections being expected within this time interval). As IAC would not be confirmed or ruled out in subgroup III (colonized patients receiving preemptive antifungals for suspected IAC), this subgroup was excluded from the prespecified performance analysis, which was focused on patients with documented IAC (subgroup IV) versus those colonized not receiving antifungals (subgroup II). Receiver operating characteristic (ROC) curves were obtained for diagnosis of IAC with BG, CS, CI, and CCI. Sensitivity, specificity, PPV, NPV, and efficiency [(true positives 1 true negatives)/all results] of recommended cut-offs were calculated and compared. BG diagnostic accuracy was also assessed in a prespecified subgroup analysis comparing patients with recurrent GI tract perforation and those with acute necrotizing pancreatitis. BG Kinetics Time intervals between culture-based diagnosis of IAC and first BG, CS, CI, and CCI values above the cut-off or start of antifungal therapy were compared. BG kinetics were analyzed according to clinical severity, response of IAC to antifungal therapy, and outcome. Statistical Analysis Student t or one-way analysis of variance test and Mann-Whitney U or Kruskall-Wallis test were used for continuous variables. Chi-square or Fisher exact test was used for categorical variables. Two-sided statistical significance was set at P ¼ 0.05 (Stata 11.0; Stata Corporation, College Station, TX/GraphPad Prism 5.0; GraphPad Software, San Diego, CA). RESULTS Study Population Four hundred thirty-four patients with abdominal surgery or acute pancreatitis admitted to the ICU for 72 hours or longer were screened during the 30-month study period. Eighty-nine (20.5%) were included (Figure 1): 68 with recurrent GI tract perforation and 21 with acute necrotizing pancreatitis. Median follow-up was 19 days (range, 3 63 d); all patients were evaluable based on complete clinical and microbiological dataset and collection of blood samples at the scheduled time points. Demographics, clinical characteristics, and surgical conditions are summarized in Table 1. Data on risk factors, Candida colonization and infection, and antifungal therapy are reported in Table 2. Seventy-five (84%) patients were colonized with Candida at inclusion and 87 (98%) during study. Among colonized patients, 58 did not develop IAC; 40 (69%) did not receive antifungals (subgroup II), and 18 (31%) received preemptive antifungal therapy for suspected IAC (subgroup III). IAC was diagnosed in 29 of 89 (33%) patients (subgroup IV): 24 of 68 (35%) with recurrent GI tract perforation and 5 of 21 (24%) with acute necrotizing pancreatitis. Blood cultures were positive in 2 of 29 (7%). Thirteen (45%) patients met IAC criteria at study inclusion; in the remaining 16 (55%), IAC was diagnosed after a median of 5 days (1 33). In five (17%) patients with IAC who died during the study, Candida infection was the primary cause (n ¼ 1) of or contributed to death (n ¼ 4). Table E1 in the online supplement compares clinical characteristics of patient subgroups I to IV. Patients not colonized with IAC differed from those colonized/with IAC for duration of hospital/icu stay and clinical severity. No difference for exposure to potential sources of false-positive BG results was found in patients with and without Candida colonization. BG Results Nine hundred twenty-one serum samples were analyzed (median, 9 samples per patient; range, 2 28). BG was greater than or equal to 80 pg/ml in 45 of 89 patients (50%) at inclusion and 62 of 89 (70%) at Day 7 after inclusion. Colonized patients had significantly higher BG values than those not colonized (Figure 2). In documented IAC, BG was greater than or equal to 80 pg/ml in 22 of 29 (76%) patients at inclusion and 24 of 29 (83%) at time of IAC diagnosis. Median BG (223 pg/ml; range, 12 3,866) was significantly higher than in noncolonized patients or those colonized without IAC. At time of IAC diagnosis, patients with severe sepsis/septic shock had higher BG than those with less severe infections (median, 313 pg/ml; range, 82 9,557; vs. 100 pg/ml; range, 12 1,059; P ¼ 0.007). Severe sepsis/septic shock was observed in 10 of 11 (91%) patients with BG greater than or equal to 400 pg/ml versus 5 of 18 (28%, P ¼ 0.002) patients with BG less than 400 pg/ml: 4 of 11 (36%)

6 Tissot, Lamoth, Hauser, et al.: b-glucan in Intraabdominal Candidiasis 1105 Figure 3. Receiver operating characteristics (ROC) curves for differentiating intraabdominal candidiasis (IAC) from Candida colonization (maximal values at the time of IAC diagnosis versus Day 7 after inclusion in patients without IAC): 1,3-b-D-glucan (BG), two consecutive values greater than or equal to cut-off (A); BG, two consecutive values greater than or equal to cut-off in recurrent gastrointestinal (GI) tract perforation or anastomotic leakage (B); Candida score (CS) (C); colonization index (CI) (D); and corrected colonization index (CCI) (E). patients died versus 1 of 18 (6%, P ¼ 0.05). BG values did not differ in infections due to different Candida species (data not shown). BG was significantly higher in patients with recurrent GI tract perforation than in those with acute necrotizing pancreatitis: median, 114 pg/ml (range, 8 3,866) versus 51 pg/ml (8 3,456) at inclusion (P, 0.001), and 136 pg/ml (13 3,866) versus 71 pg/ml (8 362) at IAC diagnosis (P ¼ 0.006). Accuracy of BG, CS, CI, and CCI for Diagnosis of IAC For this analysis, a median of two BG tests (range, 2 14) were used in patients with documented IAC (subgroup IV) versus two (range, 2 4) in colonized patients not receiving antifungals (subgroup II) (P ¼ 0.54). Table 3 summarizes the diagnostic accuracy of a single and two consecutive BG values greater than or equal to 80 pg/ml and of BG versus CS, CI, and CCI in the prespecified analysis of subgroup IV versus subgroup II. Table 4 shows the diagnostic performance in the subgroup analysis of patients with recurrent perforation versus acute necrotizing pancreatitis. ROC analyses confirmed the best diagnostic accuracy with the recommended 80 pg/ml BG cut-off. The area under the ROC curve for one and two consecutive BG measurements greater than or equal to 80 pg/ml at IAC diagnosis was 0.72 and 0.74, respectively (Figure 3A). BG performed better in patients with recurrent GI perforation than in those with acute necrotizing pancreatitis: the area under the ROC curve for two

7 1106 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL Figure 4. Time intervals between reaching values above the cut-off for 1,3-b-D-glucan (BG), Candida score (CS), Candida colonization index (CI), and corrected Candida colonization index (CCI), or start of antifungal therapy (AF) and microbiological diagnosis of intraabdominal candidiasis (IAC) (referred to as Day 0). Vertical bars represent median, interquartile range, minimal, and maximal values. The number of patients reaching values above the cut-off or receiving antifungal therapy over the total number of patients with IAC occurring after inclusion is reported on the left. The median time interval (T) between reaching the cut-off or start of antifungal therapy and microbiological diagnosis of IAC is reported on the right. consecutive BG measurements greater than or equal to 80 pg/ml was 0.79 at inclusion and at IAC diagnosis versus 0.52 and 0.57, respectively (Figure 3B). BG performed better than CS, CI, and CCI at IAC diagnosis (Figures 3C 3E). BG Time Kinetics In 16 patients with IAC occurring after inclusion, BG was greater than or equal to 80 pg/ml at a median of 5 days (range, 223 to 118) before culture-based diagnosis and 6 days (range, 218 to 14) before start of antifungal therapy. Time intervals between first BG, CS, CI, and CCI values above the cut-off and microbiological documentation of IAC are compared in Figure 4. In 4 of 5 (80%) patients with IAC with fatal outcome, progressive BG increase was observed before death (Figure 5A). In 22 patients who responded to antifungal therapy, BG peaked at time of start of antifungal therapy and then declined in 20 of 22 (91%) but remained greater than or equal to 80 pg/ml at end of study in 15 of 20 (75%) (Figure 5B). DISCUSSION This prospective FUNGINOS cohort study on the diagnostic accuracy of BG supports the use of this fungal biomarker for anticipating diagnosis of IAC in high-risk surgical ICU patients, a clinical setting in which blood cultures typically fail to detect invasive infection. In patients with recurrent GI tract perforation, BG greater than or equal to 80 pg/ml discriminated IAC from colonization with 72% PPV and 80% NPV and was superior to CI, CCI, and CS. BG preceded microbiological documentation of IAC with intraabdominal cultures and start of antifungal therapy by a median of 5 and 6 days, respectively, which suggests its potential role for guiding prompt and targeted initiation of antifungal therapy on a preemptive basis. BG kinetics reflected severity of infection, response to antifungal therapy, and clinical outcome. Most studies on BG in ICU patients included heterogeneous surgical and medical populations with a large clinical spectrum of Candida infections (8, 24 26). A retrospective analysis indirectly suggested the diagnostic usefulness of BG in patients colonized by Candida undergoing abdominal surgery (26). Response rate to empirical fluconazole was significantly higher in patients with positive compared with those with negative BG (46.9 vs. 9%). In a prospective study in 57 surgical ICU patients, twice-weekly measurement of BG detected invasive candidiasis before culture-based diagnosis with 87% sensitivity and 73% specificity (25). No information on accuracy of BG for discriminating Candida colonization from infection was reported. A recent study in medical ICU patients with candidemia showed an unprecedented diagnostic yield of BG, which anticipated documentation with blood cultures by 1 to 3 days (23, 33). The present prospective study provides an important piece of lacking information on the diagnostic accuracy of BG in postsurgical IAC, the most frequent blood culture negative form of systemic candidiasis in the ICU. BG greater than or equal to 80 pg/ml performed better in patients with recurrent GI tract perforation than in those with acute necrotizing pancreatitis. This may reflect the different pathogenesis of Candida infection in these two settings: although peritoneal fungal load and potential for BG circulation in blood is high in IAC occurring after GI tract transection, IAC after pancreatitis remains often localized to the retroperitoneal compartment. Although previous studies reported negative BG in patients colonized with Candida, we observed a higher proportion of colonized patients with BG greater than or equal to 80 pg/ml when compared with those who were not colonized (34). Translocation of BG into the bloodstream may explain BG antigenemia in colonized patients without culture-based evidence of IAC. This hypothesis is corroborated by similar exposures of colonized and noncolonized patients to potential sources of false-positive BG results (e.g., surgical gauzes, renal replacement therapy, antibacterial agents). BG discriminated Candida infection from colonization better than CS, CI, and CCI and was positive several days before CI and CCI and at the same time as CS. A prospective validation study of CS in ICU patients identified CS and BG, but not CI, as independent predictors of invasive candidiasis with sensitivities and specificities of 78 and 53% for BG greater than or equal to 75 pg/ml, 78 and 66% for CS greater than or equal to 3, and 72 and 47% for CI greater than or equal to 0.5, respectively. The present study showed higher diagnostic accuracy in IAC, in particular higher PPV of two consecutive BG measurements greater than or equal to 80 pg/ml compared with CS, CI, and CCI. The NPV of CS, CI, and CCI was lower than that previously reported (59 80% vs %) (8, 13). This is explained by the much higher incidence of invasive candidiasis observed in the present cohort (33%) when compared with that in CS validation studies in mixed ICU populations (5%) (8). This high incidence was consistent with the expected 30 to 40% infection rate derived from patient populations with similar

8 Tissot, Lamoth, Hauser, et al.: b-glucan in Intraabdominal Candidiasis 1107 Figure 5. Individual 1,3-b-D-glucan (BG) kinetics in patients with intraabdominal candidiasis (IAC) receiving antifungal therapy (n ¼ 26). Each line indicates individual BG kinetics until end of follow-up or death. D0 refers to the time of start of antifungal therapy. Median BG values at each study time point are reported at the bottom. *Candidemia. Stars indicate patients with severe sepsis/septic shock. Open circles indicate patients with a prolonged IAC course with multiple intraabdominal surgical interventions. (A) Patients not responding to antifungal therapy (n ¼ 4). The four patients, of whom one was candidemic, presented with severe sepsis/septic shock: all died. Whether IAC was the primary cause of death could not be established based on clinical and microbiological criteria. The samples for BG measurement were obtained on the day of death (n ¼ 1), one day before (n ¼ 2), and two days before death (n ¼ 1), respectively. One additional patient with septic shock and post mortem IAC diagnosis did not receive antifungal therapy: BG increased from 1,023 to 1,339 pg/ml at time of death (data not shown). (B) Patients responding to antifungal therapy (n ¼ 22). Ten patients, of whom one was candidemic, presented with severe sepsis/septic shock: all IAC responded to antifungal therapy. Two patients had a prolonged course of IAC requiring multiple intraabdominal surgical interventions. Two additional patients with less severe infections did not receive antifungal therapy, and surgical drainage alone resulted in cure of IAC: peak BG at time of IAC diagnosis was 25 and 334 pg/ml, respectively (data not shown). clinical high-risk profiles (3, 9). Surgical ICU patients at low risk for IAC were ruled out by the restrictive inclusion criteria used in the present cohort, as they would have been excluded by using a test with a high NPV. In this high-incidence population, the PPV has more clinical relevance than the NPV, which constitutes a clear advantage of BG over CS, CI, and CCI in this specific setting. Culture-based monitoring of colonization for CS, CI, and CCI assessment is work-intensive, time-consuming, and expensive, and results are available after days of incubation. In contrast, BG results are known 12 to 24 hours after simple blood sampling. Recent investigations on BG for monitoring response of invasive candidiasis to antifungal therapy have shown that BG clearance is slow (35, 36). Although its usefulness for followup remains unclear because of persistent BG positivity despite resolution of infection, increasing levels suggest failure of therapy. In our study, BG greater than or equal to 400 pg/ml reflected both severity and worse outcome of IAC. Although BG progressively increased in patients who died, in those with uncomplicated surgical course responding to antifungal therapy BG peaked at start of therapy and then declined, while remaining greater than or equal to 80 pg/ml at the end of follow-up in most cases. A major strength of the present observational study was the nonavailability of real-time BG results to clinicians in charge of patients management. This avoided potential biases in the diagnostic classification of IAC and on therapeutic decisions, which were exclusively guided by clinical and microbiological findings, as recommended. Furthermore, this study design provided comprehensive information on BG kinetics in the pre- and postdiagnostic course of IAC and, in particular, the diagnostic performance and its usefulness for predicting severity and outcome. Although the small sample size of our cohort due to the very restrictive inclusion criteria aimed at targeting a specific surgical high-risk subpopulation is a limitation, this is not unusual in diagnostic studies exploring the performance of new fungal biomarkers. The high proportion of mixed-flora peritonitis is another limitation, the pathogenic role of Candida in this frequent polymicrobial form of infection being a matter of debate. Nevertheless, many experts agree that recovery of Candida in intraabdominal cultures from patients with recurrent perforation is clinically significant even in the presence of concomitant bacterial growth (4). Antifungal therapy is thus recommended by Infectious Diseases Society of America (IDSA) and European Society of Clinical Microbiology and Infectious Diseases (ESCMID) guidelines on management of complicated intraabdominal infections and of candidiasis (31, 37, 38). Although clinical and microbiological assessment cannot distinguish the respective pathogenic role of fungi and bacteria in patients with coinfections, our observations suggest that BG may help to estimate the burden of the fungal component and its causal relationship with worse outcome. On the other hand, the present study shows that BG is negative in patients with documented bacterial infections without Candida colonization. A common difficulty in assessing the diagnostic accuracy of new tools is the definition of cases and control subjects, in particular because reference diagnostic standards often have their own intrinsic limitations. Whether IAC was present or absent in colonized patients who received preemptive antifungals for suspected infection is unclear: excluding from the analysis this subgroup may have overestimated diagnostic accuracy of BG as well as that of CS and CI/CCI. One-half of IAC cases have been diagnosed at or shortly after inclusion. Nevertheless, even in these patients positive BG remains a diagnostic clue, because first results are available within 12 to 24 hours, whereas intraabdominal cultures are incubated during days. This supports starting BG monitoring as soon as possible for promptly targeting antifungal therapy in this high-risk surgical ICU setting. This promising individualized approach has been recently

9 1108 AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL investigated in a small randomized pilot study (39). Appropriate laboratory logistics and resources are, however, key conditions for the routine use of this delicate and expensive assay. Although focusing implementation of BG in this easily identified and homogeneous ICU subpopulation is expected to be more cost effective than systematic screening in ICU patients, this prospective observational cohort study did not address clinical efficiency and health-economics issues of BG-based management. In conclusion, this prospective FUNGINOS cohort study shows that BG, a simple blood biomarker, is superior to cultures, CS, CI, and CCI for anticipating diagnosis of blood culture negative postsurgical IAC. This proof-of-concept observation in a clinically strictly predefined high-risk ICU setting, in which traditional diagnostics typically fails, is the first step toward clinical investigations of an individualized BG-driven management of ICU patients at high risk for intraabdominal candidiasis. Author disclosures are available with the text of this article at Acknowledgment: The authors and the FUNGINOS Group thank Mrs. Corine Guyaz, Mrs. Monika Ochsner, Mr. Chris Pluess, and Dr. Chloé Kaech for outstanding assistance in collecting clinical data and blood samples. They also thank Mrs. Sophie Chevalley, Mr. Christian Durussel, Mr. Johnatan Hamilton, and Mr. Benoit Pesse for outstanding technical support in processing blood samples and performing laboratory analyses. References 1. MarchettiO,BilleJ,FluckigerU,EggimannP,RuefC,GarbinoJ,Calandra T, Glauser MP, Tauber MG, Pittet D, et al. Epidemiology of candidemia in swiss tertiary care hospitals: secular trends, Clin Infect Dis 2004;38: Vincent JL, Bihari DJ, Suter PM, Bruining HA, White J, Nicolas-Chanoin MH, Wolff M, Spencer RC, Hemmer M. The prevalence of nosocomial infection in intensive care units in Europe: results of the European Prevalence of Infection in Intensive Care (EPIC) study. Epic International Advisory Committee. JAMA 1995;274: Calandra T, Bille J, Schneider R, Mosimann F, Francioli P. Clinical significance of candida isolated from peritoneum in surgical patients. Lancet 1989;2: Dupont H, Paugam-Burtz C, Muller-Serieys C, Fierobe L, Chosidow D, Marmuse JP, Mantz J, Desmonts JM. Predictive factors of mortality due to polymicrobial peritonitis with candida isolation in peritoneal fluid in critically ill patients. Arch Surg 2002;137: , discussion Montravers P, Dupont H, Gauzit R, Veber B, Auboyer C, Blin P, Hennequin C, Martin C. Candida as a risk factor for mortality in peritonitis. Crit Care Med 2006;34: Solomkin JS, Flohr AB, Quie PG, Simmons RL. The role of candida in intraperitoneal infections. Surgery 1980;88: Blumberg HM, Jarvis WR, Soucie JM, Edwards JE, Patterson JE, Pfaller MA, Rangel-Frausto MS, Rinaldi MG, Saiman L, Wiblin RT, et al. Risk factors for candidal bloodstream infections in surgical intensive care unit patients: the NEMIS prospective multicenter study. The National Epidemiology of Mycosis survey. Clin Infect Dis 2001;33: Leon C, Ruiz-Santana S, Saavedra P, Galvan B, Blanco A, Castro C, Balasini C, Utande-Vazquez A, Gonzalez de Molina FJ, Blasco- Navalproto MA, et al. Usefulness of the Candida score for discriminating between candida colonization and invasive candidiasis in non-neutropenic critically ill patients: a prospective multicenter study. Crit Care Med 2009;37: Eggimann P, Francioli P, Bille J, Schneider R, Wu MM, Chapuis G, Chiolero R, Pannatier A, Schilling J, Geroulanos S, et al. Fluconazole prophylaxis prevents intra-abdominal candidiasis in high-risk surgical patients. Crit Care Med 1999;27: Calandra T, Marchetti O. Clinical trials of antifungal prophylaxis among patients undergoing surgery. Clin Infect Dis 2004;39:S Hoerauf A, Hammer S, Muller-Myhsok B, Rupprecht H. Intra-abdominal candida infection during acute necrotizing pancreatitis has a high prevalence and is associated with increased mortality. Crit Care Med 1998;26: Garey KW, Rege M, Pai MP, Mingo DE, Suda KJ, Turpin RS, Bearden DT. Time to initiation of fluconazole therapy impacts mortality in patients with candidemia: a multi-institutional study. Clin Infect Dis 2006;43: Pittet D, Monod M, Suter PM, Frenk E, Auckenthaler R. Candida colonization and subsequent infections in critically ill surgical patients. Ann Surg 1994;220: Ostrosky-Zeichner L, Sable C, Sobel J, Alexander BD, Donowitz G, Kan V, Kauffman CA, Kett D, Larsen RA, Morrison V, et al. Multicenter retrospective development and validation of a clinical prediction rule for nosocomial invasive candidiasis in the intensive care setting. Eur J Clin Microbiol Infect Dis 2007;26: Karageorgopoulos DE, Vouloumanou EK, Ntziora F, Michalopoulos A, Rafailidis PI, Falagas ME. b-d-glucan assay for the diagnosis of invasive fungal infections: a meta-analysis. Clin Infect Dis 2011;52: Lamoth F, Cruciani M, Mengoli C, Castagnola E, Lortholary O, Richardson M, Marchetti O; Third European Conference on Infections in Leukemia. b-glucan antigenemia assay for the diagnosis of invasive fungal infections in patients with hematological malignancies: a systematic review and meta-analysis of cohort studies from the Third European Conference on Infections in Leukemia (ECIL-3). Clin Infect Dis 2012;54: Obayashi T, Yoshida M, Mori T, Goto H, Yasuoka A, Iwasaki H, Teshima H, Kohno S, Horiuchi A, Ito A, et al. Plasma (1 3)-beta-d-glucan measurement in diagnosis of invasive deep mycosis and fungal febrile episodes. Lancet 1995;345: Odabasi Z, Mattiuzzi G, Estey E, Kantarjian H, Saeki F, Ridge RJ, Ketchum PA, Finkelman MA, Rex JH, Ostrosky-Zeichner L. Beta- D-glucan as a diagnostic adjunct for invasive fungal infections: validation, cutoff development, and performance in patients with acute myelogenous leukemia and myelodysplastic syndrome. Clin Infect Dis 2004;39: Ostrosky-Zeichner L, Alexander BD, Kett DH, Vazquez J, Pappas PG, Saeki F, Ketchum PA, Wingard J, Schiff R, Tamura H, et al. Multicenter clinical evaluation of the (1 3) beta-d-glucan assay as an aid to diagnosis of fungal infections in humans. Clin Infect Dis 2005;41: Senn L, Robinson JO, Schmidt S, Knaup M, Asahi N, Satomura S, Matsuura S, Duvoisin B, Bille J, Calandra T, et al. 1,3-Beta-D-glucan antigenemia for early diagnosis of invasive fungal infections in neutropenic patients with acute leukemia. Clin Infect Dis 2008;46: Pickering JW, Sant HW, Bowles CA, Roberts WL, Woods GL. Evaluation of a (1-.3)-beta-d-glucan assay for diagnosis of invasive fungal infections. J Clin Microbiol 2005;43: Nguyen MH, Wissel MC, Shields RK, Salomoni MA, Hao B, Press EG, Shields RM, Cheng S, Mitsani D, Vadnerkar A, et al. Performance of Candida real-time polymerase chain reaction, b-d-glucan assay, and blood cultures in the diagnosis of invasive candidiasis. Clin Infect Dis 2012;54: Posteraro B, De Pascale G, Tumbarello M, Torelli R, Pennisi MA, Bello G, Maviglia R, Fadda G, Sanguinetti M, Antonelli M. Early diagnosis of candidemia in intensive care unit patients with sepsis: a prospective comparison of (1 3)-b-D-glucan assay, Candida score, and colonization index. Crit Care 2011;15:R Presterl E, Parschalk B, Bauer E, Lassnigg A, Hajdu S, Graninger W. Invasive fungal infections and (1,3)-beta-D-glucan serum concentrations in long-term intensive care patients. Int J Infect Dis 2009;13: Mohr JF, Sims C, Paetznick V, Rodriguez J, Finkelman MA, Rex JH, Ostrosky-Zeichner L. Prospective survey of (1 3)-beta-D-glucan and its relationship to invasive candidiasis in the surgical intensive care unit setting. J Clin Microbiol 2011;49: Takesue Y, Kakehashi M, Ohge H, Imamura Y, Murakami Y, Sasaki M, Morifuji M, Yokoyama Y, Kouyama M, Yokoyama T, et al. Combined assessment of beta-d-glucan and degree of candida colonization before starting empiric therapy for candidiasis in surgical patients. World J Surg 2004;28: Tissot F, Hauser PM, Eggimann P, Flueckiger U, Siegemund M, Orasch C, Zimmerli S, Calandra T, Bille J, Lamoth F, et al. Monitoring of