In vitro comparison of cold‐stored whole blood and reconstituted whole blood

Cold‐stored whole blood (CSWB) is increasingly used in damage control resuscitation. Haemostatic function of CSWB seems superior to that of reconstituted whole blood, and it is sufficiently preserved for 14–21 days. To provide evidence for a yet insufficiently studied aspect of prehospital CSWB use, we compared in vitro haemostatic properties of CSWB and currently used in‐hospital and prehospital blood component therapies.


INTRODUCTION
The most common preventable cause of death of trauma patients is massive haemorrhage [1]. Most deaths occur before reaching the hospital. A fifth of all trauma deaths could be prevented by better haemostatic control [1]. In developed countries, prehospital blood products are routinely used in damage control resuscitation (DCR). Blood products, typically red blood cells (RBCs) and fresh frozen plasma or lyophilized plasma (LP), may decrease mortality, although data are contradictory [2,3].
Cold-stored whole blood (CSWB) has gained increasing interest in civilian DCR. CSWB, with easier logistics and thus potentially faster transfusions, can provide haemostatically active cold-stored platelets for prehospital transfusions. CSWB, compared to component therapy, contains less anticoagulant and additive solutions and may, therefore, reduce haemodilution in massive transfusion. CSWB may thus be superior to conventional component therapy in DCR. However, mortality and morbidity data are lacking as randomized controlled trials are still underway [4][5][6].
In previous studies [7][8][9][10][11][12][13], haemostatic function of CSWB has been superior to that of reconstituted whole blood (RWB) and sufficiently preserved up to 14-21 days. However, to the best of our knowledge, no studies have compared the haemostatic properties of CSWB to current prehospital and in-hospital component therapies for DCR. Here, prior to introducing CSWB to clinical use, we compared in vitro haemostatic properties of CSWB and the current component therapies used in massive haemorrhage.

MATERIALS AND METHODS
Twenty-four adult male O RhD positive regular donors were recruited through Finnish Red Cross Blood Service (FRCBS) electronic newsletter for blood donors. Standard FRCBS donor eligibility criteria were used, including haemoglobin level above 135 g/L. Use of non-steroidal antiinflammatory drugs and herbal medications was prohibited 2 weeks prior to donation. Blood was collected following FRCBS standard procedures.
The study was approved by the ethics committee for Helsinki and Uusimaa Hospital District (HUS/699/2021). All participants signed an informed consent prior to donation.

Blood product preparation
We compared three different products (Figure 1), with eight donations allocated to each group: CSWB group, RWB mimicking hospital massive transfusion protocol with 1:1:1 blood product ratio (RWB group) and RBCs and LP with 1:1 ratio used in prehospital setting (RBC+LP group). To determine baseline parameters, control samples were drawn from eight random study participants before donation. All donated units were stored at +22 C prior to processing.
In the CSWB group, whole blood was collected with a Terumo Imuflex ® WB-SP collection set (Terumo Europe N.V., Leuven, Belgium) containing 63 mL of citrate phosphate dextrose (CPD). Whole blood was leukoreduced 18-24 h after donation with a platelet-sparing filter.
One CSWB donation coagulated in the collection bag due to erroneous placing of the bag on the mixer scale and was discarded. CSWB units were stored at +5 C after leukoreduction.
In the RWB and RBC+LP groups, whole blood was collected with Fresenius CompoFlow ® Quadruple T&B collection set (Fresenius Kabi AG, Bad Homburg, Germany) containing 63 mL of CPD and 100 mL of SAGM. Whole blood was leukoreduced 18-24 h after donation with a platelet-sparing filter, then separated into RBC, plasma and buffy coat. RBC units were stored at +5 C. Plasma was discarded, as only solvent/detergent-treated pooled plasma is used in Finland.
Buffy coats from four donors were pooled to form one platelet unit, stored in PAS-IIIM at +22 C.
In the RWB group, RBCs were mixed with blood group AB OctaplasLG ® (Octapharma Nordic AB, Vantaa, Finland) and platelets in 1:1:1 ratio. On day 1 (d1), platelets were from the donations for this study. On day 14 (d14), platelets were O RhD positive buffy coat platelets from FRCBS surplus stock at unit expiry (5 days or within 12 h afterwards). In the RBC+LP group, RBCs were mixed with single-donor blood group AB LP (LyoPlas N-w ® , DRK-Blutspendedienst West, Hagen, Germany) in 1:1 ratio. Samples for laboratory analyses were obtained on d1 and d14 after donation. Samples were analysed within 3 h from collection.

Blood sampling and laboratory analyses
Citrated blood was centrifuged at 2500 g for 10 min before coagulation assays. For thrombin generation assay, citrated plasma was additionally centrifuged at 2500 g for 10 min, aliquoted and frozen at À80 C until analysis.
The blood count was analysed using Sysmex XN-9000 ® analyser (Sysmex Corporation, Kobe, Japan). Coagulation tests were done using routine analysers and methods; for fibrinogen, pro- Calibrated automated thrombogram ® (CAT, Diagnostica Stago, Asnieres, France) was performed with 5 pM tissue factor, without thrombomodulin addition. Lag time, time to peak, thrombin peak and endogenous thrombin potential (ETP) were reported.

Statistical analysis
Statistical analyses were performed using IBM SPSS Statistics, version 26. Non-parametric tests were used. As the control samples were not drawn from all donors, controls were treated as an independent group in statistical analyses. To quantify differences between the groups, ratios of group medians were calculated. Differences between the groups were tested with Kruskal-Wallis test. Time-dependent changes within groups were tested with Wilcoxon signed-rank test.
Bonferroni correction was used to control the familywise error rate. A p-value of <0.05 was considered statistically significant.

Haemoglobin and haematocrit levels
Compared to controls, haemoglobin and haematocrit levels on d1 in the CSWB group were similar ( p = 0.094), and in the RWB and RBC +LP groups significantly lower (p < 0.001) ( Figure 2, Table S1). In the CSWB group, haemoglobin and haematocrit remained steady, and no clinically relevant changes occurred during storage. In the RWB and RBC+LP groups, haemoglobin concentrations and haematocrits were similar on d1 and d14 and were significantly higher in the CSWB group than in the RWB group.

Plasma coagulation factor levels
Compared to controls, coagulation factor levels on d1 in the CSWB group were similar ( p = 0.305-0.903), except for lower FVIII activity (p = 0.042) ( Figure 2, Table S2). All coagulation factor levels on d1 were significantly lower in the RWB ( p ≤ 0.010) and RBC+LP (p ≤ 0.001) groups than in the controls.
In the CSWB group, coagulation factor levels decreased during storage, reflected in significantly decreased PT and increased APTT ( Figure 2, Table S1). The storage effect was most prominent in FVIII levels (median 87 IU/dL on d1 and 43 IU/dL on d14, p = 0.018), which, in turn, prolonged APTT ( Figure 2, Table S2).
Fibrinogen level decreased during storage, but remained normal.
Although FXII, FXI, FX, FV and FII levels decreased during storage, they remained within physiological range (Table S2). In the RWB and RBC+LP groups, the variation in coagulation factor levels between d1 and d14 was minimal. In the RBC+LP group, the FVIII level was significantly lower and thus APTT longer on d14 ( Figure 2, Table S2). This was possibly due to variation in the single-donor LP composition.
Both on d1 and d14, coagulation factor levels were generally two-fold higher in the CSWB group than in the RWB and RBC+LP groups ( Figure 2, Table S2). In these groups, coagulation factor levels were approximately 50-70 IU/dL and thus below the physiological range. FVIII levels were significantly higher in the RWB group than in the RBC+LP group both on d1 (p = 0.036) and d14 (p < 0.001).

Thrombin generation
Compared to controls, CAT parameters on d1 in the CSWB group were similar, except for longer time to peak ( p = 0.010) (Figure 3). In the RWB and RBC+LP groups, CAT parameters on d1 were comparable to controls, except for shorter lag time in the RBC+LP group (p = 0.006).
In the CSWB group, lag time increased during storage ( Figure 3).
Concomitantly, time to peak slightly decreased and peak height slightly increased. However, ETP in the CSWB group was similar to that in controls and did not change during storage, indicating overall normal thrombin generation in CSWB. In the RWB and RBC+LP groups, the variation in CAT parameters between d1 and d14 was minimal ( Figure 3).
F I G U R E 2 Blood count and coagulation assays. Individual measurements with median are shown. Laboratory reference range is shown in grey. Statistical comparison between groups is shown in Tables S1 and S2. CSWB, cold-stored whole blood; RBC+LP, packed red blood cells and lyophilized plasma; RWB, reconstituted whole blood.
Both on d1 and d14, time to peak was significantly longer in the CSWB group than in the RWB ( p = 0.007 and p = 0.004, respectively) and RBC+LP ( p = 0.001 and p = 0.002, respectively) groups ( Figure 3). Likewise, on d14, lag time was significantly longer in the CSWB group than in the RWB (p = 0.038) and RBC+LP ( p < 0.001) groups. However, ETP was similar in all groups on d1 and d14, indicating similar thrombin generation. In the CSWB group, clotting times and EXTEM CFT increased during storage (Figure 4, Table 1). EXTEM A5 decreased significantly, but FIBTEM A5 and EXTEM MCF were comparable on d1 and d14. In the RWB group, EXTEM A5 and EXTEM MCF increased during storage.

Viscoelastic properties
FIBTEM A5 was comparable on d1 and d14. In the RBC+LP group, clotting times increased and clot strength decreased during storage. These effects were likely due to variation in the single-donor LP composition.
Both on d1 and d14, EXTEM CT was longer in the CSWB than in the RWB group, and on d14 also, EXTEM CFT was longer ( Figure 4, Table 1). EXTEM A5 and EXTEM MCF were the lowest in the RBC +LP group, explained by the fact that RBC+LP samples contained only some residual platelets.
F I G U R E 3 Thrombin generation assessed with Calibrated automated thrombogram. Individual measurements with median are shown. Control sample range is shown in grey. AUC, area under the curve; CSWB, cold-stored whole blood; ETP, endogenous thrombin potential; RBC +LP, packed red blood cells and lyophilized plasma; RWB, reconstituted whole blood.

Platelet counts, function and VWF levels
Compared to controls, platelet count on d1 was lower in both the CSWB (p = 0.011) and RWB (p < 0.001) groups (Figure 2), whereas VWF:Ag and VWF:Act were similar ( Table 2, Table S1). Compared to controls, platelet function on PFA and MEA on d1 was reduced both in the CSWB ( p ≤ 0.029) and RWB ( p < 0.001) groups. Interestingly, in controls, collagen/ADP closure time was slightly above laboratory reference.
Platelet count in the CSWB group decreased during storage but remained at an adequate haemostatic level ( Figure 2, Table S1). Both on d1 and d14, platelet counts were slightly, yet not significantly, higher in the CSWB than in the RWB group ( Figure 2, Table S1). In the RWB group, platelet counts achieved adequate haemostatic levels (>100 Â 10E9/L) both on d1 and d14.
The platelet function declined significantly during storage in the CSWB group. In PFA, collagen/ADP closure time prolonged (Table 2).
In MEA, the platelet function in both ADP and TRAP channels declined. In the RWB group, storage had no effect on the platelet function. On d1, the platelet function was comparable in CSWB and RWB groups. On d14, the platelet function was better preserved in the RWB than in the CSWB group.
VWF:Ag levels remained stable during storage in the CSWB group, whereas VWF:Act decreased significantly (Table 2). In the RWB group, both VWF:Ag and VWF:Act remained stable during storage. Both on d1 and d14, VWF:Ag levels were slightly, yet not significantly, higher in the RWB group than in the CSWB group. On d14, VWF:Act was higher in the RWB group.

DISCUSSION
CSWB is considered a superior prehospital blood product due to the availability of platelets and easier logistics. Most in vitro research has focused on how storage time influences CSWB haemostatic potential.
To the best of our knowledge, only a few such studies compare CSWB and RWB, and none compare CSWB and the current common prehospital practice of RBC+LP. We, therefore, studied the haemostatic potential of CSWB, RWB and RBC+LP with up to 14 days of CSWB and RBC storage time.
F I G U R E 4 Viscoelastic properties assessed with rotational thromboelastometry (ROTEM). Individual measurements with median are shown. Laboratory reference range is shown in grey. Statistical comparison between groups is shown in Table 1. CFT, clot formation time; CSWB, coldstored whole blood; CT, clotting time; RBC+LP, packed red blood cells and lyophilized plasma; RWB, reconstituted whole blood.
T A B L E 1 Rotational thromboelastometry (ROTEM) results.  In line with previous data [14], we demonstrated that platelets are essential for well-maintained in vitro haemostatic function. CSWB platelet count depends primarily on storage time and leukoreduction, and in the latter case, whether a platelet-sparing filter was used [7-10, 12, 14-18].
In our study, median platelet count was slightly higher in CSWB (leukoreduced with a platelet-sparing filter) than in RWB despite the decrease in platelet count in CSWB over time. Due to the single-donor nature of CSWB, the variance in platelet count is wide and CSWB units may display divergent haemostatic function, which may be clinically relevant.
Platelet function was abnormal in both CSWB and RWB groups.
As fibrinogen concentration and FIBTEM A5 remained stable throughout storage in CSWB, the observed increase in EXTEM CFT and decrease in EXTEM A5, both clearly outside the reference range on day 14, are likely due to impaired platelet function in CSWB.
Indeed, in line with previous studies [9,10,16,18], both ADP-and TRAP-induced platelet aggregation in CSWB decreased already during 14-day storage. As platelets in CSWB are exposed to cold storage, this impaired aggregation is seemingly in contradiction with increased platelet activation and enhanced haemostatic capacity of cold-stored platelet concentrates [19,20], but in accordance with the finding of attenuated aggregation responses to ADP and TRAP in CSWB as compared to cold-stored platelet concentrates [10]. This implies that other CSWB components, namely, RBCs and plasma, may influence platelet functionality. We observed a significant decrease in VWF activity, which possibly contributed to decreased CFT and firmness (EXTEM A5) through impaired platelet-to-platelet adhesion. Nevertheless, the platelet function in CSWB seemed haemostatically sufficient, as MCF remained within the physiological range throughout storage. This also supports that, instead of measuring coagulation factor concentrations or platelet function separately, whole blood-based assays such as viscoelastic tests probably give the most accurate estimate of the total CSWB haemostatic capacity.
Although CSWB can provide platelets for prehospital DCR, platelet concentrates are commonly used for in-hospital DCR. We demonstrated that platelet count in RWB, in contrast to CSWB, was more uniform and achieved good haemostatic levels (median 121 Â 10E9/L) even when RWB was prepared with platelets at the end of their shelf life. Interestingly, in RWB with stable fibrinogen content and older platelets, we detected enhanced CFT and MCF, suggesting platelet activation during traditional storage contributed to enhanced haemostasis [21].
An unwanted but unavoidable by-product of blood product storage is dilution due to anticoagulant and additive solutions. We demonstrated that CSWB, with retained physiological haematocrit, is less diluted than in-hospital (RWB) or prehospital (RBC+LP) component therapy. Accordingly, despite individual variation in CSWB units, coagulation factor levels were generally twice as high in CSWB than in RWB or RBC+LP.
However, both in the RWB and RBC+LP groups, fibrinogen concentration, APTT and PT were within the currently recommended levels in traumatic bleeding [22]. FVIII is labile and, affirming previous studies [8,12,14,15,23] blood has lower VWF [24] and FVIII [25] levels and longer PFA closure times [26]. Indeed, this effect on the platelet function is evident in our control group, where ADP response in PFA was attenuated. Taken together, our study suggests that the coagulation properties of CSWB are superior to those of RWB or RBC+LP and supports its use in prehospital care where platelets are otherwise unavailable.
Leukoreduced CSWB retains its haemostatic function over storage time of 14 days and is comparable with that of RWB. After initial DCR in prehospital care, in-hospital RWB seems a viable option when continued blood product use is warranted.