Study patients

In the present study, 118 adults undergoing first-time elective CABG or elective aortic valve replacement between August 2012 and July 2015 were randomly assigned by a computer program into two groups: the RAP group (n = 54) in which the retrograde autologous priming was applied and the non-RAP (n = 64) group in which the same setting was used without the possibility to save priming volume (Fig. 1). All patients were admitted to our unit the day before the planned operation. All patients received aspirin 100 mg/d until the day before the operation. Exclusion criteria were age < 18 years, LVEF ≤ 20%, emergency operations, reoperations, combined procedures, myocardial infarction 24 h before surgery, preoperative cortisone, coumarin, dual platelet inhibitor, or IV heparin therapy, thrombocytopenia (< 100 Gpt/l), liver disease, preoperative dialysis, hematological or oncological systemic disease or systemic infection. As an investigator initiated project, the trial was registered as RAP 18080 at the local research portal of the federal state of Saxony-Anhalt (https://forschung-sachsen-anhalt.de/) on August 18th, 2012.

Randomization

Consenting patients meeting inclusion criteria were randomly assigned at the day of surgery. The randomization was computerized by software. This software generated a binary result by chance. Each was strictly associated with one study group and was linked to the subject’s ID. The ID, the time of randomization, and the assigned study group were stored in a MySQL database.

Anticoagulation management

The Heparin concentration management system has been used in our institution since the year 2010 and we developed a protocol for the rational use of the device. Both study groups were managed according to this protocol [20]. In all patients, tranexamic acid (Cyklokapron®) was used as a standard perioperative bleeding prophylaxis to prevent hyperfibrinolysis. According to our institutional standards all patients received a cumulative dosage of 4 g tranexamic acid.

Anesthesia and cardiopulmonary bypass circuit

Anesthesia was initiated using an intravenous technique with propofol, fentanyl and pancuronium bromide on demand. After initiation anaesthesia was performed as balanced anesthesia, administered as low fresh gas flow with inhalational sevoflurane over the ventilator or the oxygenator of the CPB system; additional fentanyl was administered intravenously. Normothermic extracorporeal circulation was accomplished in all patients with an open non-heparin-coated system including a poly-2-methoxyethyl-acrylate (PMEA) coated hollow-fiber membrane oxygenator (CAPIOX® FX15) with integrated arterial filter, and reservoir (Terumo). The CPB system was primed with 850 ml Jonosteril® and 10.000 IU of porcine heparin. Management of CPB included systemic temperature of 36 °C, α-stat pH management, target mean perfusion pressure between 50 and 70 mmHg, and pump flow rates of ≥ 2.4 l/min/m². Myocardial protection was achieved with intermittent antegrade warm cardioplegia according to Calafiore’s protocol [21]. All patients received an intraoperative infusion of tranexamic acid (bolus of 2 g after the administration of the initial heparin bolus and another 2 g were given into the CPB circuit during reperfusion). In all patients, a cell saver (Medtronic autoLog™ or Sorin Xtra®) was used to collect wound blood prior to CPB, blood from the pleural space and CPB volume, which could not be reinfused before protamine administration. This blood was processed if the collected volume exceeded 600 ml.

Retrograde autologous priming

The only structural difference to the standard CPB circuit is a RAP bag (CE, Terumo, Belgium). Before RAP started, mean arterial pressure was elevated to approximately 80 mmHg using small doses of i.v. noradrenaline. Then priming fluid in the arterial line of the heart-lung-machine was displaced with patient’s blood using arterial pressure, the venous line was drained by slowly rotating the arterial pump. The priming fluid of the CPB circuit was slowly drained into the RAP bag. The RAP bag stayed connected with the venous reservoir for crystalloid fluid replacement during CPB. The retrograde priming procedure took approximately 3 to 5 min.

Transfusion guidelines

Packed red blood cells (PRB) were transfused when hemoglobin concentration (Hb) was < 7.0 g/dl or hematocrit (Hk) < 21% during CPB, and after cardiac surgery if Hb < 8.0 g/dl or Hk < 24% [22–24]. The transfusion of fresh frozen plasma (FFP) and platelet concentrates was based on the clinical decision of the anesthetist. According to the departmental protocol a hemorrhage of over 1000 ml within the first 6 postoperative hours was an indication for surgical re-exploration.

Data collection

Patient demographics, preoperative characteristics and postoperative outcomes were analyzed for both groups. For all patients, more than 200 variables were entered in a database. All clinical laboratory parameters were assessed by the central department of Laboratory Medicine of the University Hospital.

Study endpoints

The primary endpoint for this study was determined as intraoperative blood transfusion. Other endpoints were in-hospital blood transfusion (intraoperative or postoperative), defined safety criteria like intra- and postoperative complications (renal failure, stroke, prolonged ventilation, reintubation, ICU stay, in-hospital stay, death).

Statistical analyses

Categorical variables were expressed as frequencies and percentages. Metric variables were expressed as mean ± standard deviation (SD). The benefit of the RAP use over the control was tested with the chi-square test and was further expressed as the absolute risk reduction (ARR). For clinical decision making the effect measure number needed to treat (NNT) was calculated [25]. To predict risk factors for intraoperative and in-hospital blood transfusion, simple and multivariable logistic regression analyses were done to estimate the odds ratios (OR) with their 95% confidence intervals (CI). Discrimination was assessed using the area under the receiver operating characteristic curve (AUROC). AUROC analysis was also utilized to calculate cut off values. Sample size calculation was based on a two-sided chi-square-test with significance level alpha = 5%, a power of 90% and expected intraoperative transfusion rates of 5% in the RAP group and 25% in the non-RAP group, respectively, and resulted in n = 65 patients per group. However, the study had to be terminated due to relevant changes in value measurement procedures of the central department of Laboratory Medicine. Especially, the change from troponin I to troponin T measurement as well as the change of the measurement methods for kidney, liver, hematological and hemolysis parameters was critical. Because it was not possible to establish a trouble-free conversion, we felt that the clear analysis of the study was too heavily biased, so we decided to terminate the enrolment. The authors did not look into the data before terminating the enrolment.

Patients

Primary endpoint

The intraoperative transfusion rate was 3.7% (2 of 54 patients) in the RAP group and 17.2% (11 of 64 patients) in the non-RAP group. The primary endpoint defined as rate of intraoperative blood transfusion was significantly reduced in the RAP group (p = 0.04). The absolute risk reduction for RAP managed patients was 13.5 percent points. Non-RAP managed patients had an increased relative risk for an intraoperative red blood cell transfusion of 4.6 (95% CI, 1.07 to 20.03; p = 0.039) compared to RAP managed patients. The number of patients needed to treat to prevent at least one red blood cell transfusion was around 8 (NNT = 7.42). With regard to the fact, that in most of the cases two red blood cell concentrates (costs per RCC: 272 € [11]) are administered, the RAP procedure (costs per RAP bag: 60 €) with 480 € costs for the treatment of eight patients versus 544 € costs for two RCC is cost-effective.

Additional analyses

Secondary endpoints and additional analyses

The postoperative MAP was 79.8 ± 9.7 mmHg in the RAP group and 81.2 ± 9.2 mmHg in the Non-RAP group. The noradrenalin dose up to 16 h postoperative was 2505 ± 3087 μg in RAP patients and 2465 ± 2929 μg in Non-RAP patients. Postoperative adrenalin was used in 7.41% (4/54; 195.3 ± 745.3 μg) of RAP and 12.5% (8/64; 604.4 ± 1989.8 μg) of Non-RAP managed patients.

Study limitations

Our study is subject to the limitations inherent in studies from a single center. Furthermore, the treatment group variable (non-RAP vs. RAP) is associated with large confidence intervals (Tables 3 and 4). This is mainly a consequence of the binary nature of this variable compared to the more informative continuous variables BMI, baseline hematocrit, and blood loss. Additionally, because of the small sample population and numbers of complications, the uses of body mass index, preoperative hematocrit value, blood loss 12 h post-operative and RAP as predictors for in-hospital blood transfusions in cardiac surgery require further evaluation.