A multi-centre study of therapeutic efficacy and safety of platelet components treated with amotosalen and ultraviolet A pathogen inactivation stored for 6 or 7 d prior to transfusion

Authors


Dr Miguel Lozano, Hospital Clínic Universitari, Department of Haemotherapy and Haemostasis, Villarroel 170, 08036 Barcelona, Spain. E-mail: mlozano@clinic.ub.es

Summary

Bacteria in platelet components (PC) may result in transfusion-related sepsis (TRS). Pathogen inactivation of PC with amotosalen (A-PC) can abrogate the risk of TRS and hence facilitate storage to 7 d. A randomized, controlled, double-blinded trial to evaluate the efficacy and safety of A-PC stored for 6–7 d was conducted. Patients were randomized to receive one transfusion of conventional PC (C-PC) or A-PC stored for 6–7 d. The primary endpoint was the 1 h corrected count increment (CCI) with an acceptable inferiority of 30%. Secondary endpoints included 1- and 24-h count increment (CI), 24-h CCI, time to next PC transfusion, red blood cell (RBC) use, bleeding and adverse events. 101 and 100 patients received A-PC or C-PC respectively. The ratio of 1-h CCI (A-PC:C-PC) was 0·87 (95% confidence interval: 0·73, 1·03) demonstrating non-inferiority (P = 0·007), with respective mean 1-h CCIs of 8163 and 9383; mean 1-h CI was not significantly different. Post-transfusion bleeding and RBC use were not significantly different (P = 0·44, P = 0·82 respectively). Median time to the next PC transfusion after study PC was not significantly different between groups: (2·2 vs. 2·3 d, P = 0·72). Storage of A-PCs for 6–7 d had no impact on platelet efficacy.

Since the inception of transfusion therapy, bacterial contamination resulting in transfusion-related sepsis (TRS) has been a major cause of morbidity and mortality among transfusion recipients (Novak, 1939). With the introduction of effective platelet component (PC) transfusion therapy in the early 1960s and the discovery that the optimal conditions for PC storage for 5 or 7 d was at room temperature with agitation in improved plastic containers to retain platelet function (Murphy & Gardner, 1975), bacterial contamination of PC has persisted, and continues to limit the maximal storage time to 5 d (Esber, 1986; Murphy, 2002). Extension of PC storage >5 d offers the potential to increase availability and reduce wastage.

In order to reduce the risk of TRS, different strategies have been implemented including initial blood flow diversion, improved skin disinfection, and bacterial detection. However, despite implementing sensitive bacterial screening of PC, TRS continues to occur in the recipients of platelet transfusions (Eder et al, 2009; Dumont et al, 2010). For example, Eder et al (2009) reported that, even with increased sample volume for bacterial culture (8 ml) with inlet-diversion at collection, there were still at least 1·2 cases of platelet TRS per 100 000 transfusions. Probably the figure might be even higher because a passive report was used and it is known that, very often, septic reaction goes under recognized (Cawley et al, 2011). The PASSPORT (Post Approval Surveillance Study of Platelet Outcomes, Release Tested) study found that PC collected by apheresis which produced negative bacterial cultures when performed 24–36 h after collection, had a positive bacterial culture if re-cultured on day 8 in 66·2 per 100 000 PC (Dumont et al, 2010). As a consequence of this study, the USA Food and Drug Administration (FDA) requested that in spite of performing bacterial screening, 5 d should be the maximum shelf life of PC due to the residual bacterial septic risk. The recently published experience of the Welsh Blood Services also supports the fact that initial screening of PC does not detect all of the contaminated units (Pearce et al, 2011).

Since 1939, strategies have been proposed to decrease the bacterial burden in blood for transfusion, e.g., addition of antibiotics, such as sulfanilamide (Novak, 1939). However, almost 60 years elapsed before a method for inactivating not only bacteria but also other pathogens in labile cellular blood components was developed (Lin et al, 1997).

A method for inactivating pathogens in PC by utilizing amotosalen and ultraviolet A light (A-UVA, Intercept™ Blood System, Cerus Corp, Amersfoort, the Netherlands) was developed to inactivate a broad spectrum of viruses, bacteria, protozoa, and leucocytes in PCs and plasma (Lin et al, 1997, 2005; Grass et al, 1998; Nussbaumer et al, 2007). PCs treated with A-UVA (A-PC) have demonstrated viability in healthy volunteers (Snyder et al, 2004) and corrected prolonged bleeding times in thrombocytopenic patients (Slichter et al, 2006). Three Phase III trials with repeated transfusion of A-PC in thrombocytopenic patients demonstrated sufficient therapeutic responses that were, in terms of platelet count increments (CI) (van Rhenen et al, 2003; Janetzko et al, 2005) and prevention of bleeding, comparable to conventional PCs stored up to 5 d (van Rhenen et al, 2003; McCullough et al, 2004; Janetzko et al, 2005). This technology has been implemented into routine practice in European blood centres. (Osselaer et al, 2009; Cazenave et al, 2010). However, data on the therapeutic efficacy and safety of A-PC stored for >5 d prior to transfusion are limited.

Under routine operating conditions only a small proportion of PC are likely to be stored >5 d, thus making conduct of clinical trials to evaluate the therapeutic efficacy and safety of PC stored >5 d challenging. To mimic clinical practice haematology patients were randomized to receive one transfusion of either A-PC or conventional PC (C-PC) in a series of transfusions with PC stored ≤5 d. We designed a multicentre prospective, randomized, controlled, double blinded, intent to treat study in which C-PC stored 6 or 7 d were compared with A-PC stored for 6 or 7 d. The primary outcome was the 1 h corrected count increment (1-h CCI), analysed using a non-inferiority hypothesis with a 30% margin. Secondary outcome measures included: 1- and 24-h post-transfusion platelet counts, count increments (CI), platelet transfusion interval, haemostasis, red blood cells (RBC) use, and safety.

Methods

Study design and patients

The purpose of the study was to evaluate the therapeutic efficacy and safety of PC treated with A-UVA (A-PC) and stored for 6 or 7 d in comparison to C-PC stored for 6 or 7 d. The study was designed as a prospective, randomized, double-blind parallel group non-inferiority trial enrolling haematology-oncology patients with thrombocytopenia requiring platelet transfusion support. Patients were recruited from haematology-oncology departments at the following clinical study centres: Hospital Clínic, Barcelona, Spain; Western General Hospital, Edinburgh, Scotland; Centre Hospitalier Universitaire Pontchaillou, Rennes, France and University Hospital, Uppsala, Sweden. The study protocol and consent forms were approved by the local ethics committees at each study centre. Furthermore, the study was conducted in accordance with the applicable European regulations governing clinical investigations of medical devices (Council Directive 93/42/EEC, EN ISO 14155-1:2003, EN ISO 14155-2:2003) and the ICH Tripartite Guideline for Good Clinical Practice legislations (ICH-GCP). The study EudraCT number was 2005-003183-33. Additionally, the study was registered at http://www.clinicaltrials.gov website (ref. NCT00261924).

Patients 16 years of age and older undergoing a haematopoietic stem cell transplant or admitted for the treatment of one of the following conditions: acute or chronic leukaemia, lymphoma, multiple myeloma, or myelodysplasia were eligible to participate in the study if they were thrombocytopenic or expected to develop thrombocytopenia requiring platelet transfusion within 30 d of randomization. Exclusion criteria were: prior history of clinical refractoriness to platelet transfusion; a history or diagnosis of immune-mediated thrombocytopenia, thrombotic thrombocytopenic purpura, or haemolytic uraemic syndrome; treatment with interleukin (IL)-11 (Neumega® Pfizer, Philadelphia, USA) or other investigational platelet growth factors; disseminated intravascular coagulation; clinically or radiologically detectable splenomegaly; and previous participation in the study.

After giving written informed consent, eligible patients were screened and randomized in a 1:1 ratio (using a permuted block design), in one of two study groups, to receive a single study transfusion of either A-PC or C-PC. All other platelet components given to the patient prior to and after study platelet transfusion were conventional products stored for 5 d or less.

Platelet products and study procedures

At each study centre, both study A-PC and C-PC were prepared using pooled whole blood derived buffy-coat or apheresis. All PCs were prepared with leucocyte reduction according to standard operating procedures of the local blood transfusion centre. C-PC were suspended in platelet additive solution (PAS) with plasma (nominal 65:35): T-Sol® (Fenwal, Mont Saint Guibert, Belgium) or SSP (MacoPharma, Mouvaux, France) and A-PC were suspended in InterSol® (Fenwal) with plasma (nominal 65:35) and treated with A-UVA according to the manufacturer’s instructions.

At each centre, study investigators determined whether samples of all study products (A-PC and C-PC) or only C-PC underwent bacterial culture on day 3 after collection. Unless specifically required by the investigator, A-PC did not require bacterial culture before release. When platelet products were cultured, they were released for transfusion only if the results were negative after 3 or 4 d of culture. Also at each centre, the investigator determined which study products (A-PC and C-PC) were gamma-irradiated, prior to release on day 6 or 7, for prophylaxis against transfusion-associated graft-versus-host disease (TA-GVHD). As A-UVA treatment is Conformité Européene (CE) registered for leucocyte inactivation for prevention of TA-GVHD not all centres used gamma-irradiation to prevent GVHD for A-PCs.

The target threshold platelet content per transfusion dose was 3·0 × 1011 per unit. A sample was withdrawn from all study products at day 5 of storage to determine platelet dose. For single donor apheresis PC, a mean threshold dose of 3·0 × 1011 per unit after pathogen inactivation (PI) treatment required over-collection by approximately 10% to compensate for processing losses. For whole blood buffy-coat derived platelet components, blood centres optimized platelet processing and an additional buffy-coat unit could be added to the pool prior to the PI process to compensate for processing losses. C-PC contained approximately the same platelet concentration and volume as the A-PC.

Study transfusions were ordered by primary care physicians, blinded to treatment assignment or transfusion sequence, and administered according to standard of care clinical practice for each study centre. The study transfusion was postponed if the subject had a body temperature >39°C within 12 h prior to the transfusion, a planned invasive procedure, or active haemorrhage associated with reduction in haemoglobin >1·0 g/l or 10% within 24 h, in the previous 12 h prior to the study transfusion. The suggested platelet count threshold for prophylactic platelet transfusion was 10–20 × 109/l. After transfusion, empty PC bags were returned to the transfusion service and keep at 4°C for at least 24 h in case culture was required. A pre-transfusion platelet count was measured each day prior to transfusion. A nominal 1-h post-transfusion count was measured between 10 min and 4 h after the study platelet transfusion. A nominal 24-h post-transfusion platelet count was measured between 16 and 24 h after transfusion.

Study endpoints

The primary endpoint was the 1-h CCI. Secondary endpoints were 1-h post-transfusion CI, 24-h CI and CCI, 1-h and 24-h post-transfusion platelet count, the number of platelet transfusions given within 24 h after the study transfusion, the number of RBC units transfused within 24 h after the study transfusion, the time interval between the study transfusion and the next platelet transfusion, and clinical haemostatic assessments using the World Health Organization (WHO) Bleeding Criteria (McCullough et al, 2004). Clinical haemostasis was assessed both pre- and post-study platelet transfusion to include the period 6 h before and 6 h after each study transfusion. Haemostatic assessments were performed by trained personnel blinded to treatment assignment to determine the bleeding grade on a 5-point scale of 0 (none) to 4 (maximum) for each of eight potential bleeding sites: mucocutaneous, gastrointestinal, respiratory, genitourinary, body cavity, musculoskeletal, neurologic, and invasive sites (Miller et al, 1981; McCullough et al, 2004). The haemostatic score was defined as the maximum score of sub-scores for the eight organ systems and sites of potential bleeding.

Safety assessments and the safety monitoring committee

Global safety was assessed by monitoring patients for changes in health status using previously described methods (McCullough et al, 2004; Snyder et al, 2005). Patients were monitored for all adverse events occurring during the 4 d after each study platelet transfusion. Serious adverse events (SAE) were recorded from randomization until 15 d after study transfusions. An acute transfusion reaction was defined as an AE or SAE within the 24-h period following a study transfusion that was attributed to the transfusion by the investigator. AE occurring from 24 h to 4 d after study transfusions also were assessed for relationship to study transfusions. Patients were assessed specifically from initiation of the study transfusion to 24 h following the transfusion for events indicative of acute transfusion reactions. Vital signs and transfusion-related laboratory abnormalities, blood culture results and platelet culture results were also recorded. The signs and symptoms of AE classified as acute transfusion reactions, as well as the potential assessments to evaluate suspected transfusion related sepsis, were recorded. The study investigator at each site graded and recorded the overall AE and the causal relationship with the study platelet transfusion.

No interim analyses were conducted in this study. A safety monitoring committee (SMC) reviewed each SAE description when it occurred and performed two interim reviews of safety data during the course of the study: the first after approximately 100 patients and the second after 150 patients were transfused.

Estimation of sample size for conduct of the study and detailed statistical methods for data analysis, including the bootstrap modeling approach, are described in the Data S1.

Results

Age and dose of study products

The majority of PC (86%) were whole blood-derived, prepared by the buffy coat method. Approximately 80% of study PC (77·1% A-PC and 78·3% C-PC) were stored for 7 d prior to transfusion, the remaining products were stored for 6 d. The mean platelet content of the study product at day 5 of storage was equal between groups [4·2 × 1011 ± 0·67, (mean ± SD) P = 0·97)]. Study PC were ABO compatible with the recipient for 99·1% (A-PC) and 98·1% (C-PC) of transfusions (P = 1·0). Study PC were gamma irradiated for 14·3% of the A-PC transfusions and for 73·6% of C-PC transfusions (P < 0·0001).

Study populations for analysis of endpoints

A total of 242 patients were enrolled in four clinical sites. Most patients had haematological malignancies. A total of 211 patients received a study platelet transfusion [full analysis set (FAS) population]. Of the 211 patients who received a study platelet transfusion, 199 patients were eligible for the 1-h CCI per protocol analyses and 186 patients were eligible for the 24-h CCI per protocol analyses. Patient demographics, alloimunization risk factors, and primary diagnoses were balanced between the treatment groups (Tables SI, SII and SIII).

Two control arm patients (C-PC group) with aberrantly high 1-h post-transfusion platelet counts (110 and 128 × 109/l) were excluded from the per protocol group. These post-transfusion platelet count increments were excluded because they were deemed implausible and, upon clinical inquiry, the investigator concluded that these post-transfusion blood samples were drawn through an indwelling catheter and that the catheter’s lumen had not been adequately flushed following the study transfusion. As a consequence, the platelet count was falsely elevated because residual platelet product was sampled from the catheter through which the platelet transfusion had been administered, and the post-transfusion count partially reflected the high residual platelet concentration in the catheter rather than the platelet concentration in the patient’s peripheral blood.

Primary endpoint

The primary hypothesis of this study was whether A-PC, when stored for 6 or 7 d, provided sufficient 1-h post-transfusion CCI compared to C-PC stored for 6 or 7 d. For the A-PC group the mean 1-h CCI was 8163·0 vs. 9383·1 for the C-PC group (Table I). The ratio of the means was 0·87 (95% confidence interval: 0·732–1·034) Thus, the lower bound of the ratio of means was greater than the pre-defined inferiority margin (0·70), and the null hypothesis of inferiority of A-PC compared to C-PC was rejected (P = 0·007).

Table I.   One-hour platelet count and count increment – 1-h per protocol population
 Treatment groupP-value
Amotosalen-PC
N = 101
Conventional-PC
N = 98*
  1. SD, standard deviation.

  2. *Two patients were excluded from the 1-h per protocol population reference group due to invalid platelet counts (>100 × 109/l).

  3. †Wilcoxon rank-sum test.

  4. ‡Derived from bootstrap sampling distribution.

  5. §Derived from bootstrap sampling distribution. P-value reflects rejection of inferiority hypothesis at the 0·025 significance level.

Pre-transfusion platelet count (109/l)
 Mean (±SD)9·8 (4·24)9·6 (5·38)0·277†
 Median9·09·0
 Min, max3·0, 28·02·0, 31·0
1-h post-transfusion platelet count (109/l)
 Mean (±SD)29·2 (13·68)31·1 (15·76)0·360‡
 Median27·029·0
 Min, max7·0, 68·04·0, 91·0
1-h platelet count increment (109/l)
 Mean (±SD)19·4 (13·44)21·6 (14·56)0·268§
 Median17·020·0
 Min, max−3·0, 58·0−3·0, 76·0
1-h platelet corrected count increment
 Mean (±SD)8163·0 (5370·2)9383·1 (5905·5)0·007‡
 Median7500·08859·9
 Min; max−1757·3; 27 255·6−1350·5; 30 445·3

Secondary endpoints

Platelet counts.  Similar to the primary efficacy endpoint, the 1-h CI was not a significantly different between treatment groups (P = 0·268, Table I). Conversely, patients treated with A-PC platelets had significantly lower 24-h CI (P = 0·01) and 24-h CCI (P = 0·003) compared to patients receiving C-PC (Table II).

Table II.   Twenty-four-hour platelet count and count increment – 24-h per protocol population.
 Treatment groupP-value
Amotosalen-PC
N = 92
Conventional-PC
N = 94
  1. SD, standard deviation.

  2. *Wilcoxon rank-sum test.

  3. †Derived from bootstrap sampling distribution.

24-h post-transfusion platelet count (109/l)
 Mean (±SD)20·8 (9·2)24·9 (12·48)0·026*
 Median20·023·5
 Min, Max5·0, 52·01·0, 54·0
24-h platelet count increment (109/l)
 Mean (±SD)11·1 (8·86)15·2 (12·19)0·010†
 Median10·014·0
 Min, Max−3·0, 36·0−13·0, 44·0
24-h platelet corrected count increment
 Mean (±SD)4588·5 (3522·8)6549·3 (5211·1)0·003†
 Median4489·76442·6
 Min; max−1322·5; 15 053·9−5163·9; 24 641·8

The covariates that were significant predictors of both 1-h and 24-h post-study transfusion platelet counts were: pre-transfusion platelet count (P < 0·0001), body surface area <0·0001), and platelet dose (P < 0·0001). An additional covariate, study centre (P = 0·01), was determined to be a significant predictor for the 1-h post-study transfusion platelet count. The effect of study centre was probably a significant covariate because pre-study transfusion platelet counts differed significantly across the study centres. For the 24-h post-study transfusion platelet count, the number of previous transfusions (P = 0·003) in the transfusion cycle was an additional significant covariate. These analyses indicated that study centre practices and factors known to influence response to platelet transfusion had a significant effect on the secondary outcome measures of platelet count and CIs. The effect of treatment group was not significant when these covariates were included in the model.

Additional platelet transfusions.  The number of additional platelet transfusions after the study transfusion was similar between groups. Within 24 h after the start of the study transfusion the proportion of patients in the FAS population who received additional platelet transfusions was 15·2% in the A-PC group and 11·3% in the C-PC group (P = 0·398). The main indications for additional transfusion were a low platelet count (10 in the A-PC group; 5 in C-PC) and a bleeding event (4 in A-PC group; 6 C-PC). Within 15 d after the study transfusion the median number of platelet transfusions administered was 2 for both treatment groups. The median time to the next platelet transfusion after the study transfusion was 2·2 d for the A-PC group and 2·3 d for the reference group (P = 0·717, log-rank test, Fig 1).

Figure 1.

 Time to Next Platelet Transfusion within 15 d of the study transfusion (active surveillance period) – All patients receiving a study transfusion (full analysis set population).

Haemostatic scores.  A majority of the patients in both treatment groups (>75%), had no bleeding observed in the 6 h prior to study transfusion, and the distributions of pre-transfusion haemostatic scores were comparable between the treatment groups (P = 0·801) (Table III). Less than 6% of the patients in both treatment groups had pre-transfusion haemostatic scores of 2 and greater. The post-transfusion haemostatic scores were similar to the pre-transfusion scores in that the majority of patients did not experience bleeding in the monitored organs and sites within 6 h of the study transfusion (84% A-PC, 77% C-PC, P = 0·275). A majority of patients had no change in haemostatic scores before and after the study transfusion (86·7% A-PC, 88·7% C-PC, P = 0·244).

Table III.   Haemostatic scores of all patients receiving the study transfusion (full analysis set population).
 Treatment groupP-value
Amotosalen-PC
N = 105
Conventional-PC
N = 106
  1. *P-values based upon Wilcoxon rank-sum test.

Pre-transfusion haemostatic score
 081 (77·1%)80 (75·5%)0·801*
 118 (17·1%)20 (18·9%)
 24 (3·8%)5 (4·7%)
 31 (1·0%)1 (0·9%)
 41 (1·0%)0
Post-transfusion haemostatic score
 088 (83·8%)82 (77·4%)0·275*
 111 (10·5%)18 (17·0%)
 24 (3·8%)5 (4·7%)
 31 (1·0%)1 (0·9%)
 41 (1·0%)0
Change in haemostatic score (post-transfusion−pre-transfusion)
 −201 (0·9%)0·244*
−111 (10·5%)6 (5·7%)
091 (86·7%)94 (88·7%)
12 (1·9%)4 (3·8%)
21 (1·0%)1 (0·9%)

RBC transfusions

A majority of patients in both treatment groups did not require RBC transfusions within 24 h after the study transfusion (A-PC 76·2%, C-PC 78·3%) (Table SIV). There was no significant difference in the distribution of the number of RBC units transfused between the treatment groups (P = 0·821).

Safety

Patients were observed for any adverse events for 24 h after the start of the study platelet transfusion, and all events were recorded, whether or not the events were classified as related to the study transfusion. Approximately 30% of the patients experienced at least one AE <24 h after the study platelet transfusion (31·4% A-PC, 29·2% C-PC) (Table IV). The most common AEs in any treatment group were recorded under the category of ‘other’ (19·0% A-PC, 23·6% C-PC, P = 0·502) (Table SV). These AE comprised a spectrum of signs and symptoms including, fever (14·3% A-PC, 10·4% C-PC, P = 0·410), chills (6·7% A-PC, 2·8% C-PC, P = 0·214), positive blood culture (6·7% A-PC, 1·9% C-PC, P = 0·101), skin rash (2·9% A-PC, 1·9% C-PC, P = 0·683), and nausea/vomiting (1·9% A-PC, 2·8% C-PC, P = 1·00) (Table SV). There were no statistically significant differences between treatment groups in the frequency of any AE deemed by the investigator as related to the study transfusion within the first 24 h or during the 4 d following study transfusion (Table IV).

Table IV.   Summary of adverse events (full analysis set population).
 Number of patients (%)P-value
Amotosalen-PC
N = 105
Conventional-PC
N = 106
Within 24 h of study transfusion
Any adverse event33 (31·4%)31 (29·2%)0·766
Related adverse event10 (9·5%)7 (6·6%)0·461
Within 4 d of study transfusion
Any adverse event88 (83·8%)91 (85·8%)0·705
At least one haemorrhagic adverse event25 (23·8%)31 (29·2%)0·436
Related adverse event10 (9·5%)7 (6·6%)0·461
Severe adverse event8 (7·6%)5 (4·7%)0·408
Within 15 d of study transfusion
Serious adverse event12 (11·4%)10 (9·4%)0·659
Death1 (1·0%)5 (4·7%)0·212

When patients experienced an elevation in temperature following the study transfusion (>1°C increased temperature with rigors or >2°C increased temperature without rigors) within 24 h of the study transfusion, blood and PC cultures were ordered. Positive blood cultures were recorded for nine patients [7 (6·7%) A-PC, 2 (1·9%) C-PC]. Most patients with positive blood cultures had experienced infectious events that preceded the study transfusion. Of the seven A-PC patients with a blood positive culture, five patients had AEs indicating an infection that had started before the study transfusion; two patients with a positive blood culture had no AEs coded to the Medical Dictionary for Regulatory Activities System Organ Class Infections and Infestations (http://www.meddramsso.com) before or after the study transfusion. None of the PC cultured tested positive for bacteria.

The majority of patients in each treatment group experienced at least one AE during the 4 d after the start of the study platelet transfusion (83·8% A-PC, 85·8% C-PC, P = 0·705). These AE, were coded to MedDRA and listed by preferred terms (PT), (Table SV). Two events in the Gastrointestinal System Organ Class were statistically significantly greater in the A-PC treatment group (diarrhoea, P = 0·022; nausea, P = 0·024). No events were statistically significantly greater in the C-PC treatment group (Table SV).

All patients were followed from randomization until 15 d after study transfusion for SAE. Overall, 22 patients (11·4% A-PC, 9·4% reference) experienced at least one SAE. Only one SAE was considered as possibly related to the study transfusion (a patient in the reference group who experienced HLA alloimmunization and haemorrhagic syndrome); all other SAEs were considered unrelated to the study transfusion.

Six study patients died. Five deaths occurred in the C-PC group (4·7%) and one death occurred in the A-PC treatment group (1·0%). Three patients (all C-PC) died prior to completing 15 d of active surveillance period) and three patients (one A-PC group, two C-PC group) died after completing the active surveillance period. Only one death was considered at least possibly related to a study platelet transfusion (C-PC): a female patient experienced severe human leucocyte antigen alloimmunization and haemorrhagic syndrome with the outcome of death (0·9%). All other deaths were considered related to the patient’s underlying disease (Table IV).

Discussion

Extending PC storage beyond 5 d offers the potential to improve PC availability and reduce wastage. However, a major barrier to extending PC storage is the risk of TRS, currently still the most common reported transfusion transmitted infection. While bacterial detection has failed to be sufficiently protective (Murphy et al, 2008; Eder et al, 2009; Dumont et al, 2010), robust inactivation of bacteria will facilitate storage of PC >5 d provided that sufficient therapeutic efficacy is retained. There are data suggesting that A-PC results in acceptable in vitro platelet functional characteristics stored up to 7 d (van Rhenen et al, 2000). Moreover, available paired in vitro studies performed in a perfusion model, mimicking the flow conditions found at sites of vascular damage, suggest that A-PC stored up to 7 d exhibits adhesive and aggregating capacities similar to that of C-PCs stored for the same period of time (Lozano et al, 2007). The perfused surface covered by platelet was higher for A-PC compared to C-PC (27·1% vs. 21·2%) although this difference did not reach statistical difference.(Lozano et al, 2007) Another issue to be considered when extending the storage of A-PC up to 7 d is the potential generation of inflammatory factors. Cognasse et al (2008) investigated the levels of cytokines and chemokines (CD62P, platelet-derived growth factor-AB, IL-8, soluble CD40 ligand, IL-1b and tumour necrosis factor-α) during storage of paired PC, one half treated with amotosalen and the other half left as a control. The results of the study suggested that the levels of the immune modulators measured did not change significantly between days 5 and 7 and that the observed increased during storage was very similar in A-PC compared to C-PC.(Cognasse et al, 2008)

Nevertheless, because there is limited clinical data on the therapeutic efficacy of A-PC stored >5 d, we designed a study to evaluate the response to transfusion of PC prepared with extended storage. The study utilized a non-inferiority design with a transfusion protocol modelled on clinical practice where PC stored >5 d were used to supplement PC stored for 5 d or less. We compared the response to transfusion of a single 6- to 7-d-old A-PC in comparison to C-PC stored for 6–7 d. Both types of products were transfused to haematology patients requiring repeated platelet support within a sequence of conventional PC stored ≤5 d analogous to the anticipated intermittent use of older PC. In order to conduct this study with sufficient statistical power and feasible logistics, we selected a surrogate endpoint (1-h CCI), and utilized secondary supporting endpoints directly indicative for haemostasis including: bleeding, time to the next platelet transfusion, and RBC use. Each of these secondary endpoints is a relevant outcome measure of platelet transfusion efficacy.

The inferiority margin for this trial was based on data from the prior SPRINT (S-59 Platelet Recovery in Thrombocytopenia) trial in which haemostasis was evaluated in parallel with 1-h CCI responses (McCullough et al, 2004). SPRINT was a randomized, double-blinded, controlled, non-inferiority trial in which 645 patients received apheresis PC treated with PI or conventional PC stored for up to 5 d. The primary endpoint of the study was the proportion of patients with Grade 2 bleeding (WHO) during a period of repeated platelet transfusion. The SPRINT study showed that the frequency of Grade 2 bleeding was equivalent between treatment groups (McCullough et al, 2004). The study also showed that, while the mean 1-h CCI was 30% lower in the group receiving A-PC in comparison to C-PC, haemostasis was maintained (McCullough et al, 2004). On the basis of this observation, we postulated that a difference (A-PC vs. C-PC) in mean 1-h CCI ≤30% for 7-d old PC would be consistent with maintaining haemostasis.

The current study demonstrated that transfusion of A-PC stored for 6–7 d resulted in a mean 1-h CCI that was 13% lower than that of the group transfused with C-PC, but was not inferior by the pre-specified margin of 30% inferiority (P = 0·007). Despite a 13% decrease in the mean 1-h CCI there was no effect on haemostatic scores after transfusion, haemorrhagic AE, and the use of RBC transfusions (a surrogate marker of Grade 3 bleeding) within 24 h of the start of the study transfusion was similar between groups (A-PC: 23·8% vs. C-PC: 21·7%, P = 0·821).

The post-transfusion 24-h CCI for A-PC showed a statistically significant decrease of 30% (P = 0·003). However this decrease did not impact the proportion of patients requiring another platelet transfusion within 24 h of the study transfusion (A-PC: 15·2%; C-PC: 11·3%, P = 0·398) and the median time to the next platelet transfusion within 15 d of the study transfusion was also similar (A-PC = 2·2 d: C-PC: 2·3 d, P = 0·717). Thus, while the 24-h CCI was significantly lower for A-PC, this did not impact the need for additional transfusions or the time to the next PC transfusion.

The safety profile was similar between treatment groups for the first 24 h and for the 4 d following study transfusion. The majority of patients in each treatment group experienced at least one AE, which was not unexpected considering the clinical severity of the patients enrolled in the study and the high proportion (70%) undergoing stem cell transplantation (>25% allogeneic). Two events, diarrhoea and nausea, were significantly more frequent in the Test group, but there were no other significant differences within the gastrointestinal System Organ Class within 4 d following study transfusion.

A previous pilot study with a randomized controlled crossover design, comparing A-PC to C-PC, both stored for 7 d, failed to reject inferiority within the specified margin (an absolute difference in 1-h CCI of 2100), suggesting that 7-d-old A-PC may be inferior (Simonsen et al, 2006). In that study, the upper bound of the one-sided 95% confidence interval for the mean difference of 1-h CCI was 2400. Paradoxically, the mean 1-h CCI was higher in the A-PC group (8739 ± 3785) than in reference group (7433 ± 5408). However, this difference was not large enough to reject inferiority. The large variance observed in the C-PC group and the small number of transfusions analysed contributed to the failure to reject inferiority in that study. The median 24-h CCI for A-PC (4490) in the current study was within a range observed in other studies with similar patient populations (Bishop et al, 1991).

The findings in the current study are in contrast with a recently published report of a multicentre, open-label, randomized, non-inferiority trial using the 1-h post transfusion CCI for PC stored from 1 to 7 d in plasma (C-PC) compared with A-PC transfused to haematology patients (Kerkhoffs et al, 2010). Secondary endpoints included bleeding and RBC transfusion requirement. They found that compared to the C-PC group, patients receiving A-PC had a 31% lower 1-h CCI (A-PC: 11·4 ± 5·3 vs. C-PC 17·1 ± 7·3) and more frequent bleeding events (grades 1–3 classified by the Common Toxicity Criteria for Adverse Events (CTCAE) version 3 (A-PC: 32% vs. C-PC: 19%). Most of the bleeding events were grade 1, and no difference in the number of RBC units transfused was detected (4 ± 3 units in both study groups).

Several factors might contribute to these conflicting results. In the study by Kerkhoffs et al (2010), the mean platelet content in the A-PC (3·4 ± 0·8) was 12·8% lower than in C-PC (3·9 ± 1, P < 0·001). Although the CCI theoretically corrects for platelet dose, there is evidence that suggests that this correction is not complete and that lower platelet dose might yield lower CCIs (Davis et al, 1999). Also the pre-transfusion platelet count was 11% lower in the A-PC group compared to C-PC plasma (16 ± 11, 18 ± 13, P = 0·04). A remarkable difference between the study reported here and that reported by Kerkhoffs et al (2010) is the fact that in our study the comparison was made with PC resuspended in PAS while in the previous study the conventional PC were suspended in 100% plasma. Kerkhoffs et al (2006) reported the results of a randomized controlled trial where PC stored in PAS II were compared with PC stored in plasma. A 19·4% lower 1-h CCI was observed in PC stored in PAS II compared to PC stored in plasma (11·2 ± 6·4 PAS II, 13·9 ± 7·0 plasma, P = 0·04). There were no significant differences between the groups regarding bleeding complications and transfusion interval (Kerkhoffs et al, 2006). So it might be possible that the difference observed in their later report (Kerkhoffs et al, 2010), may be partly due to the fact that A-PC were re-suspended in PAS III while the conventional PC were in 100% plasma. In fact, there was no significant difference in 1-h CCI between the group receiving PC re-suspended in PAS III and that receiving A-PC (Kerkhoffs et al, 2010).

Another notable difference is the higher incidence of bleeding in the group receiving A-PC compared to C-PC observed by Kerkhoffs et al (2010) (32% A-PC, 19% C-PC, P = 0·034). The current study observed similar haemostatic capacity in both treatment groups. It is pertinent to note that, in the study reported by Kerkhoffs et al (2010), the evaluation of the bleeding was performed by clinicians not blinded to the treatment assignment, and the bleeding scale was different to that used in our study. Another point of note with regard to the study reported by Kerkhoffs et al (2010) is that, considering the severity of the patients enrolled in the study (about 50% of the patients were undergoing stem cell transplantation) grade 2 (CTCAE scale) and higher bleeding was only detected in the 7% of the C-PC group. Several trials that have used bleeding as a primary endpoint evaluated by research personnel blinded to treatment using daily protocol bleeding assessments reported bleeding frequencies among the study population ranging from 57·5%, in the SPRINT trial (McCullough et al, 2004), to 70%, in the PLADO (PLAtelet DOse) trial (Slichter et al, 2010). Differences in design and execution may account for many of the differences observed between the study reported by Kerkhoffs et al (2010) and ours.

In summary, our study indicates that A-PC stored for up to 7 d provided 1-h CCI and CI within therapeutic ranges not significantly inferior to C-PC. A-PCs were effective by multiple clinical indices including RBC transfusions, time to next PC and haemostasis. The safety profile of A-PC was not different from that of C-PC.

Acknowledgements

The authors wish to acknowledge the physicians, nurses and technicians of the hospitals and transfusion centres in Barcelona, Uppsala, Rennes, Edinburgh and Glasgow for their extraordinary work that made it possible to perform the current study.

Ancillary