EMPIRICUS micafungin versus placebo during nosocomial sepsis in Candida multi-colonized ICU patients with multiple organ failures: study protocol for a randomized controlled trial

The study involves 23 ICUs in French hospitals (mainly university affiliated) and was approved by the local Independent Ethic Committee (Comité de Protection des Personnes CPP Sud Est V) on 7 December 2011 and the French Health Authorities (AFSSAPS) on 2 December 2011. The University Hospital of Grenoble is the sponsor of the trial.

Each patient will have given written informed consent concerning the study design and outcomes. If the information could not be delivered to the patient, it will be delivered to his/her relatives.

The study is divided in two consecutive periods for each patient:

In case of discharge from ICU before having completed the 14-day study treatment period, the patient must receive or have received the study treatment during at least seven days.

All patients who received at least one dose of study treatment will be included in the intent-to-treat analysis.

The EMPIRICUS study primarily aims to evaluate the efficacy of early empirical treatment with micafungin in adult ICU patient with suspected invasive candidiasis in increasing proven-invasive-candidiasis-free-survival at day 28.

Secondary objectives are:

Patients are eligible for inclusion if they fulfill all the inclusion and non inclusion criteria described in Table 1. Patients are screened by the clinical research monitors and investigators in each center on a daily basis according to patient status and known yeast colonization.

If an invasive fungal infection is evidenced by the laboratory tests of baseline samples from a patient who has been included, they are withdrawn from the study and treated according to current IDSA guidelines [31].

Proven Candida IFIs are defined according to the modified criteria of EORTC/MSG Fungal Infections Consensus Group 2008:

The Steering Committee is composed of three intensivists (JFT, EA, MW), two mycologists (MC and JPG) and two pharmacists (ES, VJ). JFT, EA, MW are in charge of questions to the investigators, of checking the clinical consistencies of the outcome variables, and of asking for complementary clinical data for classifying patients and events. They will validate before database lock, blindly to the study group, all the outcome variables. Definite validation will require consensus. MC and JPG are in charge of the relationship with mycologists between centers, processes and analyses. VJ is responsible for pharmacokinetics dosage and models. ES is responsible for adverse event collection and declaration.

The Data Safety Monitoring Board (DSMB), composed of five external experts, regularly monitors the safety of the trial and will inform the Steering Committee in case of events putting into question the clinical and biological safety of the study drugs (occurrence of SAEs, publication of results of a similar trial, and so on). When the inclusion of the 50th and 150th patients is achieved, the DSMB will be provided with two statistical analyses of the SAEs reported with micafungin, liver function tests (bilirubin, ALT, AST, alkaline phosphatases, prothrombin time), venous thromboembolic and cardiovascular conditions, invasive candidiasis, deaths and study withdrawals.

The Adjudication Committee composed of the investigators and a mycologist will also adjudicate all suspected or proven candidiasis cases before the unblinding of study data.

The Adjudication Committee will review all the data from dying patients before day 28 and of patients for whom an antifungal treatment will be started.

Micafungin is provided by Astellas Laboratories to the Albert Michallon Hospital University of Grenoble, research pharmacist. Lc2 (Lentilly, France) is the CRO responsible for the preparation and labeling of batches and for providing each site with the necessary number of therapeutic units throughout the study progress. FOVEA (Rueil-Malmaison, France) is in charge of the monitoring of each patient file and Delta Consultant (Eybens, France), which has created the e-CRF, generates the queries, prepare a statistical analysis plan for safety interim analyses and the final primary analysis (primary end-point and secondary clinical end-points). OUTCOMEREA (Paris, France) has created the randomization list and will perform the final statistical analysis.

Eligible patients are randomly assigned, in a 1:1 ratio, to receive micafungin 100 mg/day (Mycamine®, Astellas Pharma, Levallois Perret, France) or matched placebo (100 ml NaCl 0.9%), administered intravenously as a one-hour perfusion. Randomization is stratified by centers. They also are randomized between two groups: A and B. Randomization is performed via the website https://​empiricus.​calystene.​fr. After having completed the ‘randomization’ web page, the investigator receives the patient’s number. The pharmaceutical order including the patient’s number is transmitted to the pharmacist who prepares treatment (micafungin or placebo) according to the randomization list.

Reconstitution of bags is performed either by the pharmacist (under extractor hood) or by an external nurse, both in unblinded conditions. Micafungin and placebo solutions provided to the site for infusion to the patient are indistinguishable to ensure that the medication blinding is maintained for both patient and investigator throughout the study. Study drug administration starts immediately after the inclusion visit and lasts 14 days.

Time zero is defined by randomization. Pharmaceutical reconstitution and biological sampling are immediately performed and the study treatment is started just after.

A form is filled by the pharmacist for each preparation. The nurse in charge of the patient fills and signs a nurse form with the times of start and end of infusions. Empty vials and packs are checked by the research pharmacist before destruction for the evaluation of compliance for all the patients.

If the existence of an invasive candidiasis at inclusion is evidenced by the analysis of baseline samples (results of the exams performed before or during the first 48 hours of the study treatment), the study treatment will be stopped and the patient will receive antifungal treatment usually prescribed in the investigational site, but blinding is not broken and the patient remains in the intent-to-treat analysis. During the treatment period, if an invasive candidiasis is evidenced, the study treatment is stopped; the antifungal treatment is started according to the decision of the investigators. Investigators can use micafungin, or alternatively, another antifungal drug according to current guidelines [31]. In this latter case, the patient stops taking the study drug but remains in the study and the investigator performs the end-of-treatment assessments immediately and the end-of-study visit 28 days later.

An e-CRF is used for each patient to collect the data. Assessments are made at the screening period, inclusion (D0) and on days 1 to 7, 9, 11, 14, 21, 28 and 90 (phone contact). Assessments include sepsis-related organ failure assessment (SOFA), monitoring (catheter, tubes and drains, mechanical ventilation procedures, inotropes), candidiasis evaluation (signs, imaging, mycological tests), laboratory tests (hematology, biochemistry, urinalysis, pregnancy test at inclusion only), microbiology (blood cultures, mycology). Biomarker dosages are performed blindly to the study group by the mycology and pharmacokinetic investigators and included in separated files. The results are not given to the clinical investigators. The existence of molecular markers of resistance (FKS1 and any other) is investigated in case of positive blood culture. Adverse events, compliance and concomitant treatments are collected at each visit.

In cases of invasive candidiasis, treatment used and complications detected by the attending physicians are collected. At D90, information about patient survival is collected.

Table 2 summarizes all the laboratory tests performed throughout the study.

Microbiological analyses of blood cultures and colonization sites are performed by local laboratories according to standardized procedures for in vitro identification of microorganisms. Briefly, specimens for mycological analyses are cultured on Sabouraud-chloramphenicol or any chromogenic medium, according to the usual procedures of the local mycology laboratory, and incubated at 37°C for at least five days. Species identification is performed using carbohydrate assimilation profile or mass spectrometry where available. The objectives of local mycological analyses are to check for the presence of at least one extra-digestive site of Candida sp. colonization before inclusion (inclusion criteria). Blood cultures and sampling of any normally sterile site are immediately collected in case of a new or severe sepsis or a septic shock. Furthermore, blood cultures are systematically performed on days 3, 7, 14 and 28. In case of positive culture for Candida from a deep site or blood, the specimen is stored in Sabouraud broth at 4°C or in standard freezing medium with glycerol at −80°C before being sent to the Grenoble University Hospital at site closure. The centralized analyses performed on these invasive isolates will include the identification of molecular markers associated with antifungal resistance. Samples for identification of molecular markers of resistance are collected on dry tubes, centrifuged, decanted in one aliquot of about 2 to 2.5 ml of serum and stored in a serum bank at −20°C.

Dosages of biomarkers (blood PCR Candida, (1–3)-β-D-glucan level, mannan antigenemia, anti-mannan antibodies) are performed in the centralized laboratories of Parasitology-Mycology of Rennes and Grenoble University Hospitals. Blood samples for mycology and PK/PD as well as isolated strains of proven or treated Candida infections are transferred from the sites to Grenoble each month. The biobank (serum aliquots and strains) is equally dispatched between Grenoble and Rennes laboratories during the study progress.

Measurements of PK parameters (distribution volume, clearances, AUC, Cmax, Cmin, AUC/MIC, Cmax/MIC) are performed centrally in the Pharmacology Department of Georges Pompidou University Hospital, Paris Descartes University (Dr V Jullien). Five blood samples for PK analysis are planned for each patient, three after the first study drug perfusion and two after the second, according to the following sampling scheme: at the end of first perfusion; 15 to 30 minutes after; immediately before the second perfusion and between 1 to 3 hours and 8 to 12 hours later.

For PK studies, samples of 5 ml of whole blood are taken, placed in dry heparin tubes and centrifuged within one hour after collection. Following the separation of plasma into two aliquots in polypropylene tubes, aliquots are frozen at −80°C and sent on dry ice to the laboratory once a month. Micafungin plasma concentration will be determined using an ultra-high pressure liquid chromatographic with fluorimetric detection method that was validated according to European guidelines for the validation of bioanalytical methods (Guideline EMEA/CHMP/EWP/192217/2009). Blood samples used for blood cultures are collected with a fungal specific tube if possible, or with a couple of aerobic/anaerobic bottles.

Hypotheses for the sample size calculation were:

Then, in the micafungin group, the actual incidence of IFIs can be estimated at 11.8% (7.1%/0.6), the candidemia related mortality at 1.4% (11.8% x 12%) and overall events between 31.4% and 38.4%. In the placebo group, the candidemia related mortality can be estimated at 4.13% (11.8% × 35%), the number of additional IFIs diagnosed after randomization at 13.7% [29] and overall events between 49.4% and 56.4%. This difference of 18% in the proportions of events between both groups is then an acceptable hypothesis. A two-sided log-rank test with an overall sample size of 235 subjects (of which 118 are in micafungin group and 117 are in placebo group) achieves 80% power at a 0.0500 significance level to detect a difference of 0.18 between 0.37 and 0.55 - the proportions surviving in groups micafungin and placebo, respectively [34]. This corresponds to a hazard ratio of 0.6013. These results are based on the assumption that the hazard rates are proportional. It could be assumed that 5% of the enrolled population will be included but not randomized. The absence of follow-up of patients at day 28 is very improbable (withdrawal of informed consent) and can be assumed to be 5%. A total number of 260 patients (130 in each arm) should then be included.

Biostatistics and epidemiology team of OUTCOMEREA is in charge of statistical analysis of the study data and will use SAS 9x (SAS, Inc., Cary, NC, USA) and R (R foundation for statistical computing, Vienna, Austria) softwares.

Secondary efficacy end-point will be evaluated similarly in the intent-to-treat population. The Cox model will be used if adjustment variables must be taken into account. Proportional assumption will be assessed using graphical methods. All the analyses will be stratified by center. No interim analyses are planned. Missing, unused data or outliers will lead to queries. In the cases where they are confirmed, the data concerning the independent variables will be replaced using multiple imputation methods.

According to French legislation, serious unexpected adverse events will be immediately communicated to the French Health authorities and placed in the Eudravigilance™ database. Safety analyses will be done after 50 and 150 included patients or more upon request of DSMB. Vital signs, body weight and temperature, clinical examinations and laboratory tests will be described. Adverse events will be summarized by system after coding. Incidence of patients with at least one AE general or of special interest, per group, will be calculated and summarized in tables. The incidence of AEs will be described per group in tables, according to severity and imputability to study drug. SAEs will also be summarized in tables and detailed in a separate section and reported to the Regional Pharmacovigilance Center. Changes between baseline and end-of-treatment in vital signs and laboratory tests will be described for each treatment group.

PK analysis will be performed according to a population approach (NONMEN software, ICON, Dublin, Ireland). PK parameters of population (means, inter and intraindividual variability) will be estimated with overall data. The following covariates will be studied: age, weight, concomitant medications, proteinemia, albuminemia, prothrombin time, ALT, AST, nature and volume of intravenous fluids that were used, evolution of weight between admission to ICU and day of PK sampling, evolution of proteinemia between admission to ICU and day of PK sampling. Significance of relationship between covariates and PK parameters will be evaluated with the modification of the objective function and impact on interindividual variability of PK parameters. A decrease of at least 3.84 points (α = 5%) in objective function will be used as significance criteria for the ascendant phase. For the decreasing phase, an increase in objective function of at least 6.63 points (α = 1%) will be required. The final model will be validated using visual predictive checks and normalized prediction errors. PK/PD statistical analyses will be performed by the Pharmacology Department of Dr V Jullien.