A Pilot Trial of Platelets Stored Cold versus at Room Temperature for Complex Cardiothoracic Surgery

The Norwegian Regional Ethics Committee North (Tromsø, Norway) approved the study (identifier 2014/692). The trial was registered on ClinicalTrials.gov on July 8, 2015, with identifier NCT02495506. This two-stage pilot trial was designed to inform future pivotal trials by providing a preliminary test of this hypothesis: Relative to transfusion with room temperature–stored platelets, cold-stored platelets reduce postoperative bleeding with sufficient platelet aggregation in patients who develop indications for platelet transfusion during complex cardiothoracic surgery. Stage I was a two-arm randomized trial that compared transfusion with cold-stored and room temperature–stored platelet concentrates stored for up to 7 days. An external safety monitor was appointed, and interim analysis on clinical effect and safety was performed during the study. Based on acceptable safety and potential for clinical effect after inclusion of 25 patients in the cold-stored arm, an exploratory, single-arm prospective observational study, stage II, extended the duration of platelet cold storage to 8 to 14 days. Stage I was conducted from March 2015 to October 2017, and the observational stage II extension was conducted from October 2017 to August 2018.

Adult patients hospitalized at Haukeland University Hospital, Bergen, Norway, for elective and semiurgent (within 24 h from hospital admission) cardiothoracic surgery were identified and screened for eligibility by their surgeons. The inclusion screening criteria were scheduled elective or semiurgent cardiothoracic surgery with either expected CPB time over 120 min or use of dual platelet inhibition drugs less than 48 h before surgery. Patients with congenital coagulopathies or hemostatic disorders were ineligible. Informed consent was obtained from patients who fulfilled the screening criteria; however, only the subset of consented patients who developed indications for platelet transfusion and had orders for study platelets received by the blood bank were enrolled into the study.

Indications for platelet, erythrocyte, and plasma transfusion, as well as hemostatic adjuncts, was decided according to standard practice at our institution in collaboration between the surgeon and anesthesiologist. The decision to transfuse platelets was based on prolonged CPB time, excessive bleeding needing a balanced transfusion approach, and/or problems achieving surgical hemostasis. Excessive bleeding is present when the patient is hemodynamically unstable because of blood loss and requiring ongoing transfusion of blood products to maintain mean arterial blood pressure above 60 mmHg and end tidal carbon dioxide above 3.5 kPa combined with too much bleeding to close the chest. After chest closure, chest tube output and hemodynamic parameters as described above were used to guide transfusion needs in the postsurgical phase. A transfusion episode was defined as ended if no additional platelet transfusions were given within 30 min after completion of the previous platelet transfusion. As an additional safety precaution, because the clinical performance of platelets stored in the cold beyond 7 days was unknown, a maximum of two units of study platelets per patient was enforced in stage II. If prolonged bleeding with additional need for platelet transfusion occurred in stage II, additional nonstudy platelets were transfused. A bolus of tranexamic acid (10 mg/kg IV) was given to patients during induction of anesthesia before surgery and followed by continuous infusion (10 mg · kg⁻¹ · h⁻¹) until the end of surgery.

For stage I, patients in the cold-stored arm received leukocyte-depleted apheresis platelets in platelet additive solution stored at 2° to 6°C under constant agitation for up to 7 days. Patients in the room temperature–stored arm received leukocyte-depleted apheresis platelets in platelet additive solution stored under constant agitation for up to 7 days in room temperature (20° to 24°C). For stage II, patients received leukocyte-depleted apheresis platelets in platelet additive solution stored at 2° to 6°C for 8 to 14 days without agitation. The platelets were suspended in 63% PAS III (InterSol, Fresenius Kabi, Germany) up to March 11, 2016, and thereafter PAS IIIM (T-PAS+, Terumo BCT, USA) and 37% donor plasma according to standard procedures in our institution. Bacterial testing of the platelet concentrates was performed on day 1 according to procedures in our department and the Norwegian requirements. All study platelet concentrates showed negative results.

The trial was registered on ClinicalTrials.org with an original plan to use evidence of altered platelet function confirmed by point-of-care measurements as the primary outcome. However, during data collection, we replaced aggregometry with chest drain output as the primary outcome to better address the pilot study’s hypothesis based on a reduction of clinically significant bleeding. We evaluated aggregometry as a secondary outcome.

Chest drain output was measured from chest closure until 8:00 am the next morning as the best estimate of cumulative postoperative blood loss. The secondary outcomes of platelet function were measured as changes in multiple electrode impedance aggregometry (Multiplate, Roche Diagnostics, Germany) after platelet transfusion, blood cell counts, conventional coagulation tests, blood usage, and hemostatic viscoelastic assays up until 8:00 am the next morning. Adverse events (venous or arterial thromboembolism, transfusion reactions), length of stay in intensive care, and mortality were measured up to 28 days.

Study patients who developed indications for platelet transfusion during complex cardiothoracic surgery were expected to be heterogeneous in terms of their baseline risks of surgical outcomes and complications. In a small pilot study, between-group variations in underlying disease severity or other potential confounders (e.g., intraoperative administration of anticoagulants, other medications, or transfusion with nonassigned platelets) could produce biased treatment effect estimates and misinterpretations, especially for a single primary outcome measured by postoperative chest tube drainage. In addition to the standard intention-to-treat analysis, we conducted a post hoc analysis restricted to the more homogeneous subset of study patients who received the correctly randomized treatment and had surgical bleeding more amenable to treatment with platelet transfusion. Patients were excluded from the post hoc analysis if they (1) were reoperated within 24 h because of excessive bleeding uncontrolled by the cardiothoracic surgical procedures, (2) were anticoagulated with heparin because of the need for venoarterial extracorporeal membrane oxygenation, (3) had iatrogenic spleen bleeding, or 4) received nonassigned platelets (fig. 1). Chest drain output of more than 200 ml/h for more than 3 to 4 h or 500 ml/h for 2 h were the indications for reopening the chest. All patients that were removed from the post hoc analysis because excessive bleeding and reoperation had an identified surgically correctable bleeding.

For stage I, the study weeks were randomized to maintain an inventory for the study without excessive waste. A randomization list for study weeks was generated by use of electronic software. The randomization list was only available to study coworkers at the Department of Immunology and Transfusion Medicine. Surgeons screening the patients for eligibility were blinded to the randomization, as were the clinical personnel responsible for the treatment and follow up of the patients. Clinicians were blinded to patient randomization when deciding whether to transfuse and when ordering platelets. Enrolled stage I study patients were randomized to either the cold-stored or room temperature–stored arms based on the week of their procedure. In the room temperature arm, there were 19 weeks with one patient/week and 3 weeks with two patients/week. In the cold-stored arm, there were 17 weeks with one patient/week and 4 weeks with two patients/week. For stage II, all enrolled patients were assigned to receive platelets stored cold for 8 to 14 days.

The personnel handling platelet concentrates during transfusion could observe both temperature and labeling of the product. Clinicians reported the effect of transfusion and potential adverse events after procedure in the electronic medical records. Study coworkers from the Department of Immunology and Transfusion Medicine who obtained written informed consent, collected data from electronic medical records, and registered information in the study database were not blinded.

Clinical data were extracted by study personnel from electronic medical records. All transfusions were registered with start time and location (operating theater or intensive care unit), end time and location, and component details. Only transfused blood components were reported. Blood samples for laboratory analyses were collected from a radial artery catheter at baseline day of surgery, immediately after weaning CPB, 1 h after arrival at the intensive care unit, immediately before and 20 min after platelet transfusion episodes, and the following morning.

The trial was initially registered on ClinicalTrials.gov with a plan to use noninferiority testing of between-arm differences in platelet function. Because of the lack of evidence needed to set acceptable, nonarbitrary tolerance margins for the noninferiority testing, we conducted standard tests of superiority, commensurate with the early phase of the trial.

A formal sample size calculation was not performed for this pilot study. Instead, the sample size was based on previous publications investigating the clinical effect of cold-stored compared to room temperature–stored platelets (n = 7 to 14,⁹  n = 8,³  and n = 12¹³ ). A previous clinical study in complex cardiothoracic surgery patients investigating post-CPB platelet dysfunction and chest tube output in normovolemic ultrafiltration included 25 patients.¹⁴  We concluded that 20 patients or more per arm in stage I would provide sufficient statistical precision for the preliminary estimation of effect size of our primary outcome to inform future pivotal trials while minimizing risks in a pilot trial. For stage II, a sample size of 10 was chosen.

Statistical analyses were performed using IBM SPSS Statistics for Windows, version 24.0 (IBM, USA). For the primary outcome, means with 95% CI, standard deviations with one-sided 95% upper confidence limits, medians, lower and upper quartiles, and 95% CI for the difference in medians are presented. As recommended for pilot studies,¹⁵  the CI for the SD requires only an upper bound not a lower bound. We also report differences in median chest drain output with 95% CI of the difference using the room temperature arm as the comparison. The CIs for differences in medians were calculated using the percentile bootstrap with 99,999 replications, using R, version 4.0.0 (R Core Team 2020, https://www.r-project.org/, accessed June 19, 2020).¹⁶ 

For the secondary outcomes, the results are given as medians (quartiles, minimum, maximum) or count (percentage). Because the study period was limited to 28 days, the length of stay for patients remaining in intensive care at this point was defined as 28 days in the statistical analysis.

In the randomized stage I, primary and secondary outcomes were compared using two-tailed Mann–Whitney U tests. Pre- and posttransfusion measurements were compared using the Wilcoxon signed-rank test. Linear regression analysis of chest drain output with the predictors study arm and time from finished surgery until next morning at 8:00 am was performed to determine the association of time with bleeding volume. It has been shown that multiple electrode impedance aggregometry response depends on platelet count.^17–19  For this reason, we performed post hoc adjustment for platelet count by dividing aggregometry response by platelet count. A P value less than 0.05 was considered significant.

From March 2015 through August 2018, 141 patients were provisionally assigned to one of the three treatment arms according to the study’s randomization protocol at the start of surgery. The exclusions and inclusions are described in detail in figure 1. Only the randomized candidates who developed indications for platelet transfusion and survived long enough for the surgical team to initiate transfusion of the assigned study product were enrolled as study patients in this pilot trial. Enrollment ceased when the target sample size was obtained. 65 consented patients in stage I and 6 consented patients in stage II were ineligible for the study because they did not receive platelet transfusion. The intention-to-treat population included 65 patients receiving platelet transfusion: 25 received room temperature–stored, 25 received cold-stored for up to 7 days, and 15 received cold-stored for 8 to 14 days. Patient characteristics are presented in table 1. The post hoc analysis included 21 patients who received room temperature–stored platelets, 22 patients who received platelets stored cold for up to 7 days and 10 patients who received platelets stored cold for 8 to 14 days (fig. 1).

Most patients received their first platelet transfusion in the operating room during surgical hemostasis, before chest closure (22 of 25 patients in the room temperature arm, 23 of 25 patients in cold-stored up to 7 days arm, and 15 of 15 patients in the cold-stored for 8 to 14 days arm). No platelet transfusions were given during CPB. Six patients in the room temperature arm, two patients in the cold-stored up to 7 days arm, and four patients in the cold-stored for 8 to 14 days arm received more than one transfusion episode.

As shown in figure 2, postoperative blood loss measured as median chest drain output was 720 ml (485 to 1,170, 180 to 2,340) for room temperature, 645 ml (460 to 800, 90 to 1,785) for cold-stored up to 7 days, and 690 ml (500 to 1,880, 100 to 9,210) for 8 to 14 days. In the post hoc analysis, the median chest drain output was 720 ml (490 to 1,020, 180 to 2,340), 590 ml (450 to 720, 90 to 960), and 595 ml (460 to 800, 100 to 1,550) for room temperature, cold storage for up to 7 days and cold storage for 8 to 14 days, respectively. Relative to the room temperature arm, the median chest drain output in the cold-stored up to 7 days arm was not statistically significant different (P_{intention-to-treat} = 0.735, P_{post hoc analysis} = 0.154).

The difference in medians between the room temperature– and cold-stored up to 7 days arm was 75 ml (95% CI, −220, 425). The difference in medians between the room temperature arm and the nonconcurrent cold-stored 8 to 14 days arm was 30 ml (95% CI, −1,040, 355). Means with standard deviations and 95% upper confidence limits are shown in table 2.

The median time from chest closure until postoperative chest drain output registration next morning was 16 (15 to 17, 12 to 19), 16 (16 to 18, 10 to 19), and 16 (13.5 to 17, 11 to 17) hours in the room temperature, cold-stored up to 7 days, and cold-stored for 8 to 14 days arms, respectively. No significant association between the length of this time period and chest drain output was observed (P = 0.458).

Blood usage is reported in figure 3. There were no statistically significant differences in total units of transfused platelets (P = 0.124), plasma (P = 0.079), or red cells (P = 0.543) in the room temperature– and cold-stored up to 7 days arms. In total, 51% of room temperature–stored units transfused were 5 to 7 days old, versus 67% of cold-stored units, whereas 83% of the cold-stored for 8 to 14 days units were transfused beyond storage day 10.

Fibrinogen concentrate (RiaSTAP, CSL Behring, USA) and prothrombin complex concentrate (Octaplex, Octapharma, Switzerland) were administered after CPB. Overall, 35 patients received fibrinogen: 15 (60%) patients in the room temperature arm, 11 (44%) patients in the cold-stored up to 7 days arm, and 9 (60%) patients in the cold-stored for 8 to 14 days arm. In total, 10 patients received prothrombin complex concentrate, 2 (8%) patients in the room temperature arm, 4 (16%) patients in the cold-stored up to 7 days arm, and 4 (26%) patients in the cold-stored for 8 to 14 days arm (Supplemental Digital Content, https://links.lww.com/ALN/C476). There was no statistically significant difference in chest drain output between patients receiving either fibrinogen concentrates or prothrombin complex concentrate compared with patients who did not receive either of these hemostatic adjuncts within each study group (P_{room temperature} = 0.540, P_{cold-stored 7 days} = 0.340, P_{cold-stored for 8 to 14 days} = 0.272).

Platelet counts and function are presented in figure 4 and in the Supplemental Digital Content (https://links.lww.com/ALN/C476). Because of the low number of patients with more than one platelet transfusion episode, the results of platelet count and function are presented for the first transfusion episode only. Platelet count increased after platelet transfusion for all study arms (P < 0.05). Multiple electrode aggregation response to all agonists increased after platelet transfusion for all study arms (P < 0.05), with the exception of RISTOtest (ristocetin) in the cold-stored for 8 to 14 days arm (P = 0.137l fig. 4).

Median thromboelastography maximum amplitude remained within normal ranges in all study arms. No significant difference were observed between the study arms in stage I. Similar, no differences were observed between the two randomized study arms for rotational thromboelastometry in-tem and ex-tem maximum clot firmness at any time point (Supplemental Digital Content, https://links.lww.com/ALN/C476). Information on coagulation and hemoglobin concentration is presented in the Supplementary Digital Content (https://links.lww.com/ALN/C476).

No transfusion reactions related to platelet transfusions were observed. The complexity of the surgical interventions performed in these high-risk patients resulted in an overall 28-day hospital mortality of three deaths (12%) in the room temperature group, two (8%) in the cold-stored 7 days arm and two (13%) in the cold-stored 8 to 14 days arm. Information on thromboembolic events and other safety parameters are shown in table 3. For the evaluation of length of stay in intensive care, one patient in the room temperature arm, two patients in the cold-stored up to 7 days arm, and two patients in the cold-stored for 8 to 14 days arm remained in intensive care at the conclusion of the 28-day study period.

In complex cardiothoracic surgery with prolonged CPB circulation time, thrombocytopenia and platelet dysfunction cause excessive perioperative bleeding and increased utilization of blood products.^20,21  Achieving postoperative hemostasis often requires platelet transfusion.²²  Optimization of blood components for bleeding patients is therefore needed.⁷ 

This pilot study examined the safety and feasibility of cold-stored platelets stored for up to 14 days in patients undergoing complex cardiothoracic surgery. The preliminary tests of the hypothesis showing that cold-stored platelets are sufficiently functional to participate in hemostasis suggest a feasible alternative product for treatment of postoperative bleeding and provide critical information to proceed with pivotal trials. No statistically significant difference in chest drain output was observed between room temperature–stored platelets and cold-stored platelets stored for up to 7 days. We have reported the effect size of our primary outcome measure chest drain output to be used for sample size calculation in future studies (table 2). We recommend that sample size calculations for future studies should be based on what the investigators regard as a clinically significant blood loss, along with the reported upper confidence limits for the standard deviations.

Impedance aggregometry is an established functional analysis for detecting platelet dysfunction that can affect hemostasis. This includes dysfunction caused by platelet disorders, antiplatelet therapy, bleeding, transfusion, and medical interventions. In our study we quantified the dysfunction caused by CPB and surgery by first performing a preoperative test to establish baseline measurement for the patient and then retesting just before platelet transfusion. To investigate to what degree the dysfunction could be reversed through transfusion of cold-stored compared with room temperature–stored platelets, we performed another test after the platelets were transfused. Our findings of significant increase in aggregation response suggest that platelets that have been stored cold for up to 14 days remain functionally active. These results are in correspondence with previous published in vitro results.^5,23 

Postoperative thromboembolic event rates were similar between cold-stored and room temperature–stored platelets. Postoperative complications in complex cardiothoracic surgery are often unsuspected and multifactorial; evaluating thromboembolic events and their relationship to platelet transfusion is difficult.²⁴  The patient population in this study had high logistic euroSCORE, indicating high risk of mortality. Even though we observed no difference in complications, the number of patients in the study is too low to definitively conclude similarity in risk profile between cold-stored and room temperature–stored platelets.

When the study was designed, only room temperature–stored platelet concentrates were available for routine use in our hospital. Cold-stored platelets had to be specially prepared for the study. To ensure the ready supply of cold-stored platelets for the study, we had to randomize the availability of either cold-stored or room temperature–stored platelets per week. The randomization schedule enabled us to collect the appropriate type of apheresis platelets in the week(s) before study patients were enrolled. The purpose of randomization is to maximize the chances for balance in the baseline covariates (potentially important confounders) between groups at the time of intervention. If unconcealed, randomization before the decision to transfuse (enrollment) could increase susceptibility to selection bias. To minimize the risk of between-group dissimilarity in the indications for platelet transfusion, all clinical personnel treating our study patients (surgeons, anesthesiologists, nursing staff) were blinded to the randomization both when deciding to transfuse and when ordering platelets. Once study platelets were ordered, no enrolled patients were withdrawn from the study, and no randomized treatment was withheld after either the order or delivery of study product. We believe our randomization scheme and concealment strategies achieved an appropriate balance between scientific rigor and practical logistics.

Most patients received the first transfusion episode during surgical hemostasis when multiple blood components and hemostatic products were transfused simultaneously. This might influence results. The number of patients having previous sternotomies was higher in the cold-stored up to 7 days arm. History of previous sternotomies is associated with increased bleeding and might influence the results; especially chest drain output and blood product use.

The stage II cold-stored arm was different from the stage I cold-stored arm in that the number of platelet units was restricted to two because of safety concerns for their extended storage time. The stage II and stage I inclusion criteria did not differ; however, sometime between stage I and stage II, the surgeons identifying patients for screening and informed consent improved in their selection of the surgical candidates at higher risk for platelet transfusion.

Although the causes of postoperative bleeding measured by chest tube output are multifactorial and difficult to disentangle, we chose to report chest tube output and total blood products transfused as sensitive, if not necessarily specific measures of the contribution of platelet transfusion to postsurgical hemostasis. Postsurgical hemostasis depends on the combination of patients’ intrinsic hemostatic capabilities, preceding and interim medical treatments including transfusion, and the degree of surgical hemostasis. In our study, the majority of blood products was transfused during the surgical hemostatic procedures before chest closure, and study platelet transfusions were initiated before the col lection of postoperative chest tube output. We acknowledge that other measures of hemostasis may be more accurate than chest tube drainage, e.g., hemostasis score,²⁵  and should be considered in future studies.

This pilot trial supports the feasibility of platelets stored cold for up to 14 days and provides critical guidance for future pivotal trials in high-risk cardiothoracic bleeding patients.

The study was performed by the Blood Far Forward research group, a collaboration between military and civilian healthcare services, including the Norwegian Armed Forces Medical Services, Sessvollmoen, Norway, and the Departments of Immunology and Transfusion Medicine, Anesthesia and Intensive Care, and Heart Disease at Haukeland University Hospital, Bergen, Norway. Collaborators at the U.S. Army Institute of Surgical Research contributed to study design, data analysis, data interpretation, and article preparation. The Blood Far Forward Research Group thanks all personnel at the Departments of Immunology and Transfusion Medicine, Anesthesia and Surgical Services, and Heart Disease of Haukeland University Hospital. This study has been our collaborative effort. Finally, the authors thank the members of the Trauma, Hemostasis, and Oxygenation Research Network for inspiring discussions and encouragement.

Supported by the Norwegian Armed Forces Medical Services, Sessvollmoen, Norway, and the Department of Immunology and Transfusion Medicine at Haukeland University Hospital, Bergen, Norway.

Dr. Spinella reports ongoing financial relationships with Secure Transfusion Services (South San Francisco, California), Cerus (Concord, California), Entegrion (Durham, North Carolina), KaloCyte (Baltimore, Maryland), and Haima (Cleveland, Ohio), and a past financial relationship with Hemanext (Lexington, Massachusetts). The opinion or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the views of the U.S. Department of the Army, the U.S. Department of Defense or the Norwegian Armed Forces Medical Services. The remaining authors declare no competing interests.

Full protocol available at: geir@docfish.no. Raw data available at: geir@docfish.no.