Clinical effectiveness of fresh frozen plasma compared with fibrinogen concentrate: a systematic review

The initial rationale behind this systematic review was formulated by the lead author (SKL), who subsequently recruited the coauthors, arranged for an unrestricted educational grant from CSL Behring (Marburg, Germany), a manufacturer of fibrinogen concentrate, and liaised closely with Fishawack Communications (Oxford, UK), which provided editorial support. Within this structure, we had full control of the ideas, data, reporting and conclusions of the research and were not influenced by any commercial activities. Our study did not require ethical approval.

Studies reporting the administration of FFP or fibrinogen concentrate in a perioperative or massive trauma setting were identified in MEDLINE and EMBASE (1 January 1995 to 31 December 2010). The reference lists of identified studies and relevant reviews were checked for additional citations and to identify key FFP RCTs conducted prior to 1995.

In the electronic database search, we used 'perioperative', 'surgery', 'h(a)emostasis', 'bleeding', 'traumatic injuries', 'blood component transfusion' as both input words and subject headings. The FFP search also included 'fresh frozen plasma', 'FFP', 'plasma' and 'therapeutic plasma', and the fibrinogen concentrate search comprised 'fibrinogen/therapeutic use', 'fibrinogen concentrate', 'fibrinogen concentrates'. Additional limits were added: (1) language: English or German; (2) date range: 1 January 1995 to 31 December 2010; and (3) article type: clinical trial, meta-analysis, RCTrandomized controlled trial, case reports, clinical trial phases I to IV, comparative study, controlled clinical trial and journal article. Titles and abstracts were screened for relevance, and full publications of the potentially relevant studies were assessed against the formal eligibility criteria.

Studies were eligible for inclusion in the systematic review if they reported a predefined outcome of the administration of FFP or fibrinogen concentrate specifically within the perioperative or massive trauma setting. All preparations of FFP ('standard' as well as 'pathogen-reduced') were eligible for inclusion. Abstracts of work presented at congresses were excluded.

Outcomes of interest included blood loss, allogeneic transfusion requirements, survival, hospital and/or ICU length of stay (LOS), plasma fibrinogen levels, thrombotic events, acute lung injury (ALI), transfusion-associated circulatory overload (TACO), infections (bacterial contamination and viral transmission) and multiple organ failure (MOF). A meta-analysis of the effect of FFP on morbidity was recently published by Murad and colleagues [5]. To avoid redundancy, outcomes related to morbidity (thrombotic events, ALI, TACO, infections [bacterial contamination and viral transmission] and MOF) were not extracted and tabulated for FFP studies.

All study types were included (RCTs, non-RCTs, prospective comparator studies, retrospective comparator studies, prospective and retrospective noncomparator studies, and case reports). Study design and quality of evidence were captured at the data extraction stage.

Studies were included regardless of patient age, sample size or duration of patient follow-up. We excluded preclinical studies, studies which included patients with congenital clotting factor deficiencies or other haematological disorders and studies reporting the prophylactic administration of FFP or fibrinogen concentrate in the absence of surgery. Studies reporting mixed populations were included only if > 50% of the cases were surgical and/or trauma patients. Studies reporting the administration of FFP for liver disease or reversal of vitamin K anticoagulation were excluded if there was no surgical intervention. Studies in which control groups received autologous platelet-rich plasma were also excluded. For assessment of the effect of each intervention on survival, studies that reported no deaths in either study arm were excluded from the survival outcome but included in all other applicable outcomes of interest.

Data from eligible studies were extracted according to the population, intervention, comparison and outcome (PICO) method into a standard form and included study size, study type, number of study arms (including the presence of a control group), the number and age (paediatric or adult) of patients in each arm, indication for and nature of the intervention (including the control group where applicable), the presence of chronic liver disease, treatment with anticoagulants, methods for measuring (where applicable) blood loss and fibrinogen levels, and relevant outcome data. Where possible, blood loss and transfusion requirements were recorded as intra-, peri- or postoperative (where perioperative was the intra- and postoperative periods combined), and survival rates were recorded by the time period reported (6 hours, 24 hours, 30 days, in-hospital, and so on). Consequently, in some instances, multiple outcomes from a single study were included.

Studies were categorised by type, then assessed by the effect of the intervention on each applicable outcome criterion. Each intervention (FFP or fibrinogen concentrate) could show a positive, negative or null effect against the comparator group or against absolute values where no comparator group was available. As such, data from the observational and case studies were generally descriptive in nature and were only included as supportive evidence. Studies without comparator groups were not included in any formal calculations of the efficacy of the intervention. Where outcome measures were compared between groups within a study, these comparisons were assessed on the basis of a difference in group averages (mean or median as available). The primary purpose of this review was achieved by describing the numbers and the quality of studies identified in relation to the effect of FFP or fibrinogen concentrate on the stated outcome measures. To avoid redundancy and the inclusion of multiple instances of data from the same study, meta-analyses were not intended to be formally included in the analysis (that is, outcomes from meta-analyses were not counted towards the number of studies reporting the effect of an intervention for each outcome). Instead, meta-analyses were included descriptively as supportive evidence. Meta-analysis was not a stated aim of this review, and analysis of the data extracted from the eligible trials formed the basis of the conclusions reached.

The selection process and study flow are depicted in Figure 1. The majority of identified citations were easily excluded at the level of title or abstract on the basis of no relevance to this review. After screening, there were 91 studies that fulfilled our eligibility criteria, reported outcomes of interest and were formally included in the analysis (Tables 1, 2 and 3). There was also one meta-analysis that was used as supportive evidence. Only a minority of selected studies (n = 18) were high-quality prospective studies with randomisation procedures and a control arm (Table 4). A further 39 nonrandomised and observational studies with a comparator group were included. The majority (62%) of FFP studies involved the comparison of the intervention group against a different dosage (typically by assessing the effect of different FFP:RBC ratios), formulation (various kinds of 'pathogen-reduced' plasmas) or blood product (for example, cryoprecipitate or whole blood). There were only 20 studies in which FFP was compared with no FFP or with a non blood product (for example, colloid or crystalloid). There were five fibrinogen concentrate comparator studies, two comparing fibrinogen concentrate with no fibrinogen concentrate or with a non blood product and, importantly, three studies comparing the effect of fibrinogen concentrate directly with FFP (administered in combination with other allogeneic products).

The average amount of FFP administered in each study varied greatly, ranging from a nominal prophylactic dose to tens of units (typically in the massive transfusion studies). Several studies did not report the FFP dose administered. The dosage of fibrinogen concentrate typically ranged from 2 to 8 g. In studies of massive transfusion, the intervention group was classified as those patients receiving the highest FFP:RBC ratio (typically 1:1), whereas the control groups comprised those patients who received lower FFP:RBC ratios. Control groups from studies in which the effect of FFP:RBC ratios was not assessed typically received various forms of colloid or crystalloid, apart from those studies directly comparing the administration of fibrinogen concentrate with that of FFP (in addition to other allogeneic components).

Studies containing overlapping or duplicate populations were excluded prior to the completion of the data tables. Two articles reported the same data comparing the outcomes of liver transplant patients who were administered either standard FFP or solvent/detergent-treated FFP (SD-FFP) [21, 22]. Data from the more recent publication were used [22]. The same study of infants undergoing craniofacial surgery was published in two articles [23, 24]. Where the same outcome data were reported in both articles, the earlier publication was used [23]. Two articles also reported data from the same trial of patients undergoing cardiovascular (CV) surgery [25, 26]. The article reporting the greater number of outcomes of interest was retained [26].

Three massive transfusion studies reported data from an overlapping population [2, 27, 28]; however, two of these publications compared cohorts of patients pre- and postimplementation of a massive transfusion protocol. The publication that specifically compared different FFP:RBC ratios was used in this review [28]. Overlapping data were reported in another three massive transfusion studies [4, 29, 30]. One reported the results from the first year of a prospective study assessing the effect of the FFP:RBC ratio on survival and compared these with matched historical cases [30], the second reported the results from the entire prospective study only [29] and the entire prospective data set was analysed against matched historical cases in the third [4]. Survival data were taken from the most recent publication in which the entire prospective data set was compared against matched historical cases [4]. Additional outcomes of interest were found in the earliest publication, so they were included where appropriate [30]. There were overlapping populations in another three massive transfusion publications [31‐33]. For the purposes of this review, they were classified as a single study because subsets of patients were analysed in each publication.

There is a relative paucity of high-quality evidence reporting the outcome of administration of FFP in a perioperative or massive trauma setting, despite the long period of its usage. Also, few high-quality trials were identified that reported the outcome of administration of fibrinogen concentrate perioperatively, and none assessed the intervention during massive trauma. Controlled trials with robust blinding and randomisation procedures are the gold standard when assessing the efficacy and safety of interventions. However, the low number of RCTs conducted for each intervention may reflect the difficulties in designing and implementing such trials in bleeding patients who are in potentially life-threatening situations.

Despite there being fewer fibrinogen concentrate comparator studies than FFP comparator studies (5 versus 52, respectively), the nature of the comparison must also be considered. Of the FFP comparator studies analysed, only 38% (20 of 52) compared FFP with no FFP or with a non blood product (for example, colloid or crystalloid). Many FFP studies assessed outcomes against a control group that received a different dosage or formulation of FFP reflecting the widely held assumption that 'standard' FFP is efficacious in these situations. In contrast, 40% (2 of 5) of fibrinogen concentrate studies compared the intervention with either no fibrinogen concentrate or a colloid or crystalloid control, and the other 60% (3 of 5) compared the use of fibrinogen concentrate with FFP. Therefore, the low number of comparator trials for fibrinogen concentrate still provided useful evidence and allowed valuable comparisons to be made between the intervention and a non blood control, as well as a direct comparison with FFP. Whilst observations can be made about the relative efficacies of FFP and fibrinogen concentrate in the perioperative setting, it is notable that, despite a number of noncomparator studies of fibrinogen concentrate in the trauma setting (with one recent study involving 128 patients), there are no comparator studies involving the use of fibrinogen concentrate in patients with haemorrhage due to massive trauma. This lack of evidence highlights a need for more research in this area so that the efficacy of fibrinogen concentrate can be assessed in this clinical setting.

Our systematic literature search identified 70 FFP and 21 fibrinogen concentrate publications reporting the effects of the intervention on a number of outcomes of interest. However, although 74% of identified FFP studies had a comparator group, many involved different doses or formulations of FFP rather than a non FFP arm. Furthermore, the majority of fibrinogen concentrate studies (76%) were of low quality and did not report the effect of the intervention in relation to a comparator group. Figure 2 summarises the results of all studies in our review with a comparator group for each intervention by reported outcome and also as a combined total of all outcomes. Figure 2a reinforces the doubt about the efficacy of FFP, assessed across a range of outcomes, with no single outcome reflecting a benefit for FFP in > 50% of extracted measures. When all outcomes were added together, a benefit of FFP administration was found in only 28% of measures, only slightly more than those reporting a negative effect of FFP (22%). If only those studies comparing FFP with a non FFP group are considered, the picture is even less convincing, because only 7% of reported outcomes supported a beneficial effect of FFP (Figure 2c). In contrast, the evidence for the efficacy of fibrinogen concentrate was far more consistent, with no negative outcomes reported for any measure (Figure 2b). Fibrinogen concentrate was shown to reduce blood loss, reduce allogeneic transfusion requirements, reduce ICU and hospital LOS and increase plasma fibrinogen levels in over two-thirds of reported outcomes. In the five comparator trials, 70% of outcomes showed a benefit of fibrinogen concentrate over the control. Importantly, the control was FFP in three of the studies, thus providing some evidence that fibrinogen concentrate is more efficacious than FFP across a range of clinical outcomes in the perioperative setting.

The strongest support for a benefit for FFP derives from studies reporting survival, where 50% suggested that FFP (typically at higher FFP:RBC ratios) reduces mortality; however, FFP was associated with increased mortality in 20% of studies. In general, studies reporting an association of higher doses of FFP with improved survival assessed the effect of FFP:RBC ratios during massive transfusion. This finding is in agreement with the meta-analysis performed by Murad and colleagues [5]. Many studies targeting higher FFP:RBC ratios did so by increasing the amount of FFP administered in the early phase of massive haemorrhage. This temporal aspect of FFP administration was highlighted in a recent publication which found that patients who received an early high FFP:RBC ratio were in less severe shock and less likely to die early from uncontrollable haemorrhage than were those patients in the low FFP:RBC ratio group, who never achieved a high ratio [117]. The survival advantage associated with the higher FFP:RBC ratios currently being lauded in the literature may be due partly to selection, whereby patients in such studies die with a low FFP:RBC ratio, not because of a low ratio. Fibrinogen deficiency manifests early in bleeding patients. It is possible that an improvement in survival rates at higher FFP:RBC ratios was due in part to earlier supplementation of plasma fibrinogen in the resuscitation effort and not to a benefit of FFP per se. The fibrinogen concentrate studies identified were typically small, with a mean of only 10 patients per arm. Consequently, there were almost no deaths reported in either group, making a robust assessment of any survival benefit following the administration of fibrinogen concentrate (early or late) virtually impossible.

In this review, we examined relevant outcomes of interest by analysing the literature regarding one of two interventions: FFP and fibrinogen concentrate. However, haemostatic support during surgery or massive trauma is rarely achieved by the administration of one product alone; therefore, the majority of the studies included in this review involved the administration of other products, particularly RBC, but also PC, cryoprecipitate, prothrombin complex concentrate, tranexamic acid, aprotinin and others. The influence of coadministered products on the outcomes of interest was not studied in this review, though the potential for an impact should be considered when drawing any conclusions regarding the impact of each intervention on these outcomes, particularly in studies where cryoprecipitate was administered, as this would provide a more concentrated dose of fibrinogen than FFP alone.

The benefits of any intervention should outweigh the risks. In this review, we found inconsistent and contradictory evidence concerning the efficacy of FFP. Furthermore, the findings of this review are in line with other published work, where FFP has particularly been associated with an increased risk of mortality when used during nonmassive transfusion [5]. In terms of risks, Murad and colleagues examined the incidence of ALI and MOF in their meta-analysis [5], in which they reported that FFP significantly increased the risk of pulmonary complications (OR = 2.92). Contrary to researchers in other studies [103, 118], however, they reported a significant reduction in the risk of MOF with FFP administration (OR = 0.40). Although the risk of viral and bacterial transmission by FFP exists, it is very low. The introduction of nucleic acid screening for known infectious diseases has led to the transmission of infectious diseases being rare [119]. Other adverse events (AEs) are associated with the use of FFP, such as allergic and haemolytic transfusion reactions, but these are infrequent ≤ 1 event per 100,000 blood components issued [6]). Furthermore, several strategies have been employed successfully to reduce many of the risks associated with 'standard' FFP, such as the introduction of different formulations of plasma (for example, SD-FFP and photochemical treatment FFP, lyophilised and others), the use of leucocyte-depleted plasma and restricting the use of FFP from female donors [9]. Half of all outcomes analysed in this review indicated that FFP had no effect, positive or negative, and it could be argued that administration of FFP is worthwhile on the basis of the possibility that it might be efficacious and at worst will effect no change in clinical parameters. Nonetheless, when the potential risks associated with FFP, however rare, are considered in the context of the efficacy findings in this study, in which fewer than one-third of the reported outcomes favoured FFP over the comparator, the continued use of the product should be questioned in an approach that weighs risk versus benefit.

There was consistent evidence that fibrinogen concentrate improved the outcomes studied in this review. More than two-thirds (16 of 23) of all outcomes showed a benefit of fibrinogen concentrate over any comparator, notably 72% (13 of 18) of the outcomes were favourable for fibrinogen concentrate over FFP. Furthermore, no study reported a negative effect versus a comparator on any outcome measure included in this review. In terms of the risks involved with fibrinogen concentrate, there is a low risk of AEs such as allergic reactions, and, in rare cases, administration of fibrinogen concentrate has been associated with thromboembolic events [120, 121]. A number of preclinical studies have provided evidence supporting the safety and tolerability of fibrinogen concentrate, and preclinical models, including one of venous stasis, have shown no evidence of thrombosis formation in treated animals, demonstrating the low thrombogenic potential of the product [120, 122‐125]. In addition, a pharmacosurveillance report and systematic review of thrombotic events in clinical studies (of patients with both congenital and acquired afibrinogenaemia) showed no significant safety concerns associated with fibrinogen concentrate (Haemocomplettan P; CSL Behring, Marburg, Germany) use in perioperative bleeding situations with regard to thrombogenicity [120]. Over a 22-year pharmacosurveillance period, nine thrombotic events possibly related to the administration of fibrinogen concentrate were reported (seven of which were in patients with congenital fibrinogen deficiency) at an incidence of 3.48 per 100,000 treatment episodes. However, further safety studies employing rigorous methods intended to detect conditions such as deep vein thrombosis are still required to confirm the findings from the preclinical and pharmacosurveillance studies.

Among the trials included in this review [26, 54‐63, 71‐74, 104, 112‐116], there was a low incidence of the predefined safety outcomes (thrombotic events, ALI, TACO, infections [bacterial contamination and viral transmission] and MOF). Furthermore, there was a low frequency of AEs of any nature in the 17 fibrinogen concentrate studies in which AEs were reported (Table 10). Of the trials with a comparator group, an AE (postoperative atrial fibrillation) was reported for 1 of 36 (3%) patients receiving fibrinogen concentrate compared with 7 of 37 (19%) of patients in the comparator groups [26, 54, 55, 71]. There were another seven AEs reported in the noncomparator trials and case reports of 240 patients receiving fibrinogen concentrate (3%) [57‐61, 63, 72, 74, 104, 112, 113, 115, 116]. Four of the AEs (all arterial ischaemic events) occurred between 4 and 12 days (median = 7.5 days) after fibrinogen concentrate administration in postoperative surgical patients who had massive perioperative haemorrhage and required more than 12 U of RBC [115]. Of the remaining three AEs, one case of 'jitter and snoring respiration' was reported by nursing staff, though the patient was judged to be alert with normal respiration upon the arrival of the attending physician; one patient complained of attacks of shivering 24 hours after fibrinogen concentrate administration [59]; and one patient who received massive transfusion of a variety of haemostatic products had subsequent acute renal failure [116].

This review presents a consistent picture of the efficacy of fibrinogen concentrate and supports the safety of the product, with a low incidence of thrombotic events reported in the included studies. However, this evidence was derived from a small number of studies with a low number of patients in each arm. Whilst published studies support the efficacy and safety of fibrinogen concentrate during surgical procedures, more trials reporting outcomes from a greater number of patients are needed to reinforce these findings.

The use of haemostatic products often occurs in response to a 'trigger', and for FFP, historically, this was based on the percentage of blood volume lost. However, there are many difficulties inherent in estimating the volume of blood lost in a patient with a life-threatening massive haemorrhage, for whom resuscitation efforts may continue for some time. Though primarily used during massive trauma bleeding, a ratio-based approach, often integrated into a massive transfusion protocol which aims to coordinate the activities of the various necessary departments and personnel, is steadily replacing the volume of blood loss as the trigger for FFP administration. This approach appears simple and practical, though concerns remain, not the least of which is that the optimum FFP:RBC ratio has yet to be determined [126], an observation highlighted in recent evidence-based guidelines [127]. In addition to uncertainty over the optimum FFP:RBC ratio, there are a number of logistical considerations, such as potential issues with the availability of thawed FFP, the waste that could be created by immediately thawing large quantities of FFP (because of the short shelf-life of thawed FFP) and the need for ABO-compatible plasma early on. Interestingly, three studies prompt questions regarding the survival benefits reported for increased FFP:RBC ratios in massive trauma patients. Snyder and colleagues [89] suggested that the reduction in mortality reported in many studies of this type is due to survival bias brought about by the exclusion from the analyses of patients who died during the early phase of the study. To test this hypothesis, they performed an analysis adjusted for survival bias and found that an FFP:RBC ratio ≥ 1:2 no longer had any effect on in-hospital mortality rates compared with < 1:2. Furthermore, both Riskin and colleagues [85] and Gunter and colleagues [28] found a significant improvement in survival after implementation of a massive transfusion protocol despite unchanged FFP:RBC ratios. This suggests that simply increasing the amount of FFP administered during a massive transfusion is not the key factor in improving survival and that other factors inherent in a transfusion protocol also have a substantial impact. The survival benefit of FFP may be due more to the focus on 'haemostasis' and the timely delivery of blood products rather than to the increase in FFP, which was reinforced in a recent review of the impact of the introduction of goal-directed haemostatic resuscitation on surgical and trauma patients [128].

When a ratio-based approach is not considered appropriate, such as during routine surgical procedures, a laboratory test-based approach is often employed to guide the administration of FFP on the basis of the prothrombin time and International Normalised Ratio. However, only the initiation phase of haemostasis is monitored by conventional coagulation tests. Screening may not indicate abnormalities, such as the presence of critically low fibrinogen levels, if the coagulopathy occurs in the amplification, propagation or stabilisation phase. The change in plasma fibrinogen levels reported in relevant studies in response to different doses of FFP and fibrinogen concentrate indicate that a good response was achieved in all studies in which fibrinogen concentrate was used, with 2 to 4 g typically raising plasma fibrinogen levels by around 1 g/L [26, 54, 55, 59, 71, 72, 104, 113, 115]. The increases in levels were generally more variable and less predictable for FFP [22, 23, 30, 38, 41, 49, 53, 64, 108‐110]. According to recent measurements, because 1 L of FFP provides an average of 2.0 g of fibrinogen [16] and the same amount of fibrinogen can be found in just 100 ml of fibrinogen concentrate [17], the use of fibrinogen concentrate over FFP for substitution of fibrinogen may be more favourable. This reduced infusion volume may help avoid dilutional coagulopathy and the risk of volume overload associated with FFP use in patients with nonmassive bleeding. Conversely, in situations where a patient has experienced massive blood loss, the greater infusion volume of FFP may better restore some of the lost volume (though FFP should never be used solely as a volume expander [9, 119]).

A goal-directed approach using low fibrinogen levels as a trigger for intervention could improve outcomes. The efficacy of fibrinogen concentrate to improve haemostasis has been demonstrated in a number of in vitro and animal models of haemodilution and severe bleeding [120, 122‐125, 129, 130]. These efficacious findings in preclinical studies are supported by the safety record of fibrinogen concentrate, such as the low thrombogenic potential demonstrated in animal models [120, 124, 125], and that reported in published trials. In our experience, fibrinogen concentrate has some clear advantages over FFP. The preparation method carries a reduced risk of immunological side effects, reduced risk of viral transmission, reduced transfusion volume and a known factor content [20]. In addition to the low risk of the product, its low volume facilitates faster infusion times. Furthermore, rapid administration is possible because there is no need to blood-type, thaw or warm fibrinogen concentrate.