Thromboelastometry versus standard coagulation tests versus restrictive protocol to guide blood transfusion prior to central venous catheterization in cirrhosis: study protocol for a randomized controlled trial

This is a double-blinded, parallel-group, superiority, single-center, randomized controlled clinical trial.

The participating hospital is a private tertiary 657-bed hospital in São Paulo, Brazil, with an active liver transplant program (mean ± standard deviation (SD) 118 ± 15 liver transplants per year). Patients will be screened and recruited in two mixed medical-surgical intensive care units, with 48 beds in total, and 3673 admissions per year.

All adult patients (aged 18 years or older) admitted to the ICU with chronic liver failure and an indication for bedside central venous line placement for intravenous (I.V.) medication administration (including vasopressors), hemodialysis and/or hemodynamic monitoring will be eligible for inclusion. Chronic liver failure will be defined as liver cirrhosis of any etiology diagnosed via biopsy or clinical evaluation. The exclusion criteria will be: acute liver failure, an indication for a peripherally inserted central venous catheter, current use of orally administered (e.g., inhibitors of factor Xa, thrombin and vitamin K-dependent factors) or parenterally administered anticoagulants (e.g., nonfractionated heparin or low-molecular-weight heparin) in full therapeutic dosage, von Willebrand disease and enrollment in another clinical trial with blood transfusion as intervention.

Before a CVC, every included patient will be submitted to blood sampling for SCTs (PT, INR, aPTT and platelet count), ROTEM® analysis, serum fibrinogen, hemoglobin, hematocrit, ionized calcium and arterial pH. Blood samples of 5 mL will be obtained directly from a peripheral vein by applying a light tourniquet to avoid blood stasis, or from a central venous line after discarding the first 5 mL of blood (in the case of an indication for a second CVC, e.g., hemodialysis catheter), or from an arterial line. Blood samples for SCTs, ROTEM® and fibrinogen will be conditioned in citrate tubes (3.2%; Sarstedt Monovette®, Wedel, Germany) and processed immediately. Blood samples for hemoglobin level, hematocrit and platelet count will be conditioned in 2.6-mL tubes containing EDTA KE (Sarstedt Monovette®, Wedel, Germany).

Thromboelastometry analysis (ROTEM® delta, Pentapharm Co., Munich, Germany) will be performed by pipetting 300 μL citrated whole blood and 20 μL 0.2-M calcium chloride (CaCl₂) with specific activators into a plastic cup [38]. Extrinsically activated (EXTEM), intrinsically activated (INTEM), and fibrinogen polymerization (FIBTEM) assays will be performed. ROTEM® analysis will run for up to 60 min. The parameters coagulation time (CT; s), clot formation time (CFT; s), maximum clot firmness (MCF; mm), alpha angle (α; degrees) and amplitude at 10 min (A10; mm) will be recorded [38].

The EXTEM assay contains a preparation of human recombinant tissue factor, which activates the coagulation cascade and can assess coagulation status via CT_EXTEM, CFT_EXTEM and MCF_EXTEM [27]. The FIBTEM assay uses cytochalasin D which is a potent platelet inhibitor and serves to evaluate fibrinogen function and fibrin polymerization [27]. The A10 is a surrogate marker of MCF and can predict coagulopathy with high sensitivity and specificity (1.0 and 0.7, respectively) with the advantage of a faster evaluation of coagulation disorders [39].

Prothrombin time, INR and aPTT (ACL TOP 700 system, Instrumentation Laboratory, Bedford, MA, USA), serum fibrinogen (Clauss method) and platelet count (Sysmex® XN-Series, Sysmex Corporation, Lincolnshire, IL, USA) will be performed according to the hospital’s standard analytical methods.

All CVCs will be performed with real-time ultrasound (Bard Site ~ Rite® 5 Ultrasound System, Bard Access Systems, Salt Lake City, UT, USA) guidance, using the modified Seldinger (over the guidewire) technique. Only ICU physicians and medical residents trained in ultrasound-guided vascular access will be allowed to perform the procedures. To obtain a valid training status, the ICU physician must have performed at least 50 ultrasound-guided CVCs and at least 12 ultrasound-guided insertions the previous year. The choice of insertion site (jugular, femoral or subclavian) will be at the inserter’s discretion. All procedures will be performed according to best practices guidelines [40] and audited by a staff nurse. A chest X-ray will be ordered after each procedure (except for femoral vein site) to assess catheter positioning and complications. A blinded-to-intervention nurse will inspect the insertion site for local complications within the first 24 h after catheter placement using the HEME tool checklist [41].

Inclusion in the trial does not preclude concomitant care. The researchers will monitor whether patients will receive any blood transfusion related to the procedure within the first 24 h.

Patients will be randomly assigned to three groups, each including a protocol to guide blood product transfusion (fresh frozen plasma (FFP), platelets and/or cryoprecipitate) prior to CVC, namely: (1) a transfusion protocol based on standard coagulation tests – Coagulogram-based group, (2) a transfusion protocol based on rotational thromboelastometry – Thromboelastometry-based group or (3) a restrictive transfusion protocol – Restrictive group – also based on SCTs, but with wider transfusion trigger values.

The Coagulogram-based group (control group) is based on our institutional protocol and takes into account the SCTs PT/INR, aPTT, platelet count and serum fibrinogen levels to guide transfusion prior to CVC (Fig. 1).

Accordingly, if INR ≤1.5, aPTT ≤50 s, platelet count ≥50,000/μL and serum fibrinogen ≥150 mg/dL, no transfusion is indicated. Otherwise, if INR >1.5 or aPTT >50 s, FFP is transfused at 10 mL per kg of body weight; and/or if platelet count <50,000/μL, 1 unit per 10 kg of body weight of random platelets (up to 10 units) or 1 unit of apheresis platelets is transfused; and/or if serum fibrinogen <150 mg/dL, 1 unit per 10 kg of body weight of cryoprecipitate is transfused (up to 10 units).

The thromboelastometry-based transfusion protocol uses EXTEM and FIBTEM assays from the ROTEM® and was adapted from Görlinger et al. [42] (Fig. 2). No transfusion is necessary when CT_EXTEM is ≤80 s and A10_EXTEM is ≥40 mm. The CT_EXTEM will be used to assess coagulation factor deficiency. For patients in whom CT_EXTEM is >80 s, transfusion of 10 mL per kg of body weight of FFP is performed. If the patient presents an A10_EXTEM <40 mm, we further evaluate the A10_FIBTEM. If A10_FIBTEM is ≥10 mm (indicating adequate fibrinogen function), random platelet units (1 unit per 10 kg of body weight; maximum 10 units) or 1 unit of apheresis platelets is transfused. Otherwise, if A10_FIBTEM is <10 mm (indicating fibrinogen deficiency), cryoprecipitate (1 unit per 10 kg of body weight; maximum 10 units) is transfused.

The restrictive transfusion protocol is also based on SCTs, but it uses wider transfusion triggers and it does not take into account serum fibrinogen and aPTT levels (Fig. 3). If INR is ≤5.0 and platelet count is ≥25,000/μL, no transfusion is required. If INR is >5.0, FFP is transfused at 10 mL per kg of body weight; and/or platelet count is <25,000/μL, random platelet units (1 unit per 10 kg of body weight; maximum 10 units) or 1 unit of apheresis platelets is transfused.

The primary efficacy endpoint is the proportion of patients transfused with any blood product (i.e., FFP, platelets or cryoprecipitate) prior to CVC. The primary safety endpoint is the incidence of major bleeding within the first 24 h after catheter insertion. Major bleeding was defined according to the HEME tool [41]. According to the HEME tool, major bleeding is defined as overt bleeding with any of the following (in the absence of other causes): decrease in hemoglobin of 20 g/L or more, transfusion of 2 or more units of RBC with no increase in hemoglobin level, decrease in systolic BP by 10 mmHg or more while the patient is sitting up, spontaneous decrease in systolic BP of 20 mmHg or more or increase in heart rate by 20 bpm or more; bleeding at any one of the following sites: intracranial, intraspinal, intraocular, pericardial, retroperitoneal or intraarticular; or wound-related bleeding requiring an intervention.

The secondary efficacy endpoints will be: the proportion of patients transfused with each blood component and the infused volume of each blood component, hemoglobin and hematocrit levels before and up to 24 h after CVC, costs associated to blood product transfusion and laboratory tests, ICU and hospital length of stay, and ICU, 28-day and hospital mortality. Secondary safety endpoints will include: the incidence of minor bleeding complications (i.e., oozing or hematoma with no criteria for major bleeding) and the proportion of patients transfused due to bleeding within the first 24 h after CVC, and the incidence of acute transfusion-related adverse events (i.e., transfusion-associated cardiac overload, acute hemolytic transfusion reactions, anaphylactic reactions, febrile nonhemolytic reactions and urticarial reactions). The transfusion of bleeding patients after CVC will follow the Hospital Israelita Albert Einstein institutional transfusion protocol. Accordingly, if INR >1.5 or aPTT >32 s, FFP (10 mL per kg) is administered. If serum fibrinogen <150 mg/dL, cryoprecipitate (1 unit per 10 kg) or fibrinogen concentrate (2 to 4 g) is administered. If the platelet count <50 x 10³/mm³, random platelets (1 unit per 10 kg) or apheresis platelets (1 unit) are administered. Correction of hypothermia (axillary temperature ≥35 °C), hypocalcemia (ionized calcium ≥1.14 mmol/l) and acidosis (pH ≥7.31) is always recommended.

The participant timeline and study flowchart are presented in Figs. 4 and 5, respectively. A filled Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) Checklist is available (see Additional file 1).

Patients will be recruited in the two mixed medical-surgical ICUs of Hospital Israelita Albert Einstein in São Paulo, Brazil. On a daily basis, the research team will actively search the records and patient list for patients admitted with a diagnosis of cirrhosis. Additionally, on-call and horizontal care staff will be encouraged to recommend eligible patients to the research team at any time. The current recruitment rate is four patients per month with an expected duration of recruitment of 42 months. Recruitment rate will be monitored trimonthly.

Allocation sequence with random numbers will be computer-generated using random, varied size blocks and stratification by severe renal failure (i.e., hemodialysis program). The randomization sequence will be stored in individually separated and sequentially numbered opaque, sealed envelopes and kept secured in a locker with restricted access (centralized at ICU pharmacy). After the consent term is signed and coagulation tests collected, patients will be randomly assigned to study groups with an allocation ratio of 1:1:1.

Allocation concealment will be preserved by holding randomization sequence disclosure until the consent term is signed and coagulation tests collected. Patients will be enrolled by an on-call recruiter, and assignment to intervention will be provided by an independent researcher, who will be the only one to know which intervention the patient will be assigned to. The research subjects, central venous catheter inserter and outcome assessors will be blinded to the assigned intervention and bleeding will be assessed by the use of the HEME tool [41] immediately after CVC and 24 h after the procedure. The catheter inserter is not directly responsible for care of the patient included in the trial and will not have access to the patient’s chart. In case of major bleeding or serious transfusion-related adverse reaction, unblinding to the attending physician is anticipated.

We calculated the sample size based on the primary efficacy endpoint. We assumed a transfusion rate of 90% in the Coagulogram-based group (i.e., control group) [11] and an absolute risk reduction of 25% for the Thromboelastometry-based and Restrictive groups. Using a study power of 0.80 and alpha of 0.05, we need n = 43 patients per group. The sample size was adjusted for multiple comparisons (additional increase of 20%, n = 8 per group) (Coagulogram-based versus Thromboelastometry-based, Coagulogram-based versus Restrictive, and Thromboelastometry-based versus Restrictive) according to Witte et al. [43]. To account for loss to follow-up, we have added 10% (n = 4) more patients to the adjusted sample size. The final required sample will be 55 patients per group (total estimated sample size of 165 patients).

Trained researchers will be responsible for data collection. The principal investigator (PI) will be responsible for double-checking the collected data and input data into the study database. The quality of data will be guaranteed by locking the variables’ input range and random checking for quality 5% of patient data at each interim analysis and final analysis database.

The collected data will include age, race, gender, weight, hospital and ICU admission dates, main ICU diagnosis, cirrhosis etiology, transplant list status, comorbidities, Simplified Acute Physiology Score (SAPS 3) [44], Sequential Organ Failure Assessment (SOFA) score [45], Model for End-stage Liver Disease (MELD) score [46], transfusion of blood products in the previous 24 h and 24 h after CVC, procedure time, procedure indication, catheter type, number of attempts, insertion site, inserter, elective/emergency insertion, hemoglobin, hematocrit, PT, aPTT, platelet count, serum fibrinogen, ionized calcium, pH, peripheral temperature, ROTEM® parameters (CT, CFT, α angle, MCF and A10), type and quantity of transfused blood products for the procedure, adverse events related to the procedure and transfusion, when indicated, length of ICU and hospital stay and mortality at day 28.

An independent Data Safety Monitoring Board (DSMB) will conduct two interim analyses after the enrollment of 55 and 110 participants. The Lan-DeMets alpha spending function method will be used to determine statistical differences between groups [47]. After DSMB assessment, a report will be generated recommending whether the trial should continue or stop due to efficacy or harm. The final decision to stop the trial will be held by the PI together with the POCKET Trial Investigators. An independent audit will be conducted every 6 months after recruitment of the first patient. Every serious adverse event will be immediately reported to the Ethics Committee and the DSMB, and minor adverse events will be enlisted and periodically released. Every protocol amendment will be subject to approval from the local Ethics Committee.

All analyses will be performed as intention-to-treat. Categorical variables will be summarized as absolute and relative frequencies. Normality will be tested with the Kolmogorov-Smirnov test. Accordingly, continuous variables with normal distribution will be summarized as mean ± SD, or median and interquartile range (IQR) in case of non-normal distribution.

All binary outcomes will be analyzed using a generalized linear model with binomial distribution and logit link function followed by pairwise comparisons adjusted by Bonferroni correction for multiple comparisons. Pairwise comparisons will be presented as odds ratio (OR) along with 95% confidence interval (95% CI), which will be calculated by logistic regression.

A survival curve will be performed with the Kaplan-Meier estimator and the log-rank test will be used to compare groups. To assess time-to-event (28-day mortality), an unadjusted Cox proportional hazards regression model will be performed. The hazard ratio (HR) with its respective 95% CI will be reported.

Prespecified subgroup analysis for the primary outcome will be presented as a Forest plot, and will include the presence of hypothermia (<35 °C), SAPS 3 score (≥50 points) and MELD score (≥25 points). Interaction between groups and each covariate will be assessed by multivariate logistic regression.

We will make maximum effort to reduce missing data. We will use the multiple imputation method to handle missing data when necessary. The Statistical Package for the Social Sciences version 22.0 (IBM Inc., Chicago, IL, USA) will be used for statistical analyses. GraphPad Prism version 6.07 (GraphPad Software Inc., La Jolla, CA, USA) will be used for graph plotting.

Transfusion and laboratory test costs will be assessed. For the analysis of the costs of laboratory tests, we will include the total cost of the test(s) (i.e., ROTEM® or SCTs), which will guide the blood transfusion. The unit cost information regarding blood products and coagulation tests will be obtained from the blood bank and main laboratory of Hospital Israelita Albert Einstein.

The study was designed in accordance to the Declaration of Helsinki. The Hospital Israelita Albert Einstein Ethics Committee approved the study on 12 August 2014 (reference number: 734824). Recruitment of participants started in September 2014 and is currently ongoing. The study was registered at ClinicalTrials.org (identifier: NCT02311985) on 3 December 2014. The research team will release the study results in scientific media regardless of their final results.

A trained research team member will obtain the informed consent term from the research subject or their next of kin. A deferred informed consent term will be applied in case of emergency CVC when the patient lacks clinical condition and no next of kin is immediately available. In this case, the definitive informed consent must be obtained within 24 h after the procedure. The participants may leave the study at any time with no treatment disadvantage. All study information will be kept secured with limited access. The patient database will be password-protected and patients will be identified by trial ID numbers. A translated copy of the Informed Consent Form is available (see Additional file 2).

This trial is currently ongoing (recruitment phase).