A systematic literature review on the use of platelet transfusions in patients with thrombocytopenia.

Objective: Investigate globally, current treatment patterns, benefit-risk assessments, humanistic, societal and economic burden of platelet transfusion (PT). Methods: Publications from 1998 to June 27, 2018 were identified, based on databases searches including MEDLINE®; Embase and Cochrane Database of Systematic Reviews. Data from studies meeting pre-specified criteria were extracted and validated by independent reviewers. Data were obtained for efficacy and safety from randomized controlled trials (RCTs); data for epidemiology, treatment patterns, effectiveness, safety, humanistic and societal burden from real-world evidence (RWE) studies; and economic data from both. Results: A total of 3425 abstracts, 194 publications (190 studies) were included. PT use varied widely, from 0%-100% of TCP patients; 1.7%-24.5% in large studies (>1000 patients). Most were used prophylactically rather than therapeutically. 5 of 43 RCTs compared prophylactic PT with no intervention, with mixed results. In RWE studies PT generally increased platelet count (PC). This increase varied by patient characteristics and hence did not always translate into a clinically significant reduction in bleeding risk. Safety concerns included infection risk, alloimmunization and refractoriness with associated cost burden. Discussion: In RCTs and RWE studies there was significant heterogeneity in study design and outcome measures. In RWE studies, patients receiving PT may have been at higher risk than those not receiving PT creating potential bias. There were limited data on humanistic and societal burden. Conclusion: Although PTs are used widely for increasing PC in TCP, it is important to understand the limitations of PTs, and to explore the use of alternative treatment options where available.

PT has been used for over 50 years for active bleeding and prophylaxis in high-risk populations (eg, cancer patients or TCP patients undergoing invasive procedures) [9][10][11][12]. Most guidelines recommend a PC threshold of 50 × 10 9 /L to prevent hemorrhage prior to invasive procedures but vary depending on the type of procedure [10,11,13,14]. Although widely used, PTs are associated with a variety of risks including infection (that may result in sepsis) [15][16][17], transfusion reactions and alloimmunization (up to 40%) [18][19][20]. PTs are associated with higher odds ratios of arterial thrombosis and mortality among TTP and HIT patients [21], and with higher rates of transfusion-associated circulatory overload (TACO) [22] and transfusion-related acute lung injury [23]. Any transfusion-related intervention can increase the risk of TACO which has been shown to be under-reported [24]. Thrombocytopenic patients who receive PTs also have a variety of underlying conditions and diseases, making the decision to transfuse difficult in light of potential risks [11,12,[25][26][27].
There are practical, logistical and quality control issues associated with platelet preparation and storage and donor platelet quality varies [28,29]. There is currently no routine testing of platelet quality (such as the percentage of active platelets) [29] and quality is negatively affected by prolonged storage times [30]. PTs must be delivered under strict local/national guidelines that may be difficult to meet [31][32][33][34]. Although 94% of PTs in the USA are collected through apheresis, leukocyte reduction in whole blood collection is often avoided in an effort to reduce costs, despite evidence suggesting that it reduces -related adverse reactions [35]. In addition, the demand for platelet components is rising substantially worldwide, placing pressure on an already scarce resource [36].
There is currently a lack of concrete evidence on the efficacy and effectiveness of PT in patients with TCP (ie, at risk for bleeding), making clinical decisions difficult. To date, minimal research has been conducted to evaluate and understand the current burden and benefit-risk trade-off of PT use in TCP patients. Therefore, this multi-topic, global systematic literature review (SLR) was conducted to investigate current treatment patterns, benefit-risk assessments, as well as the economic, societal and humanistic burden of therapeutic and prophylactic PT in the TCP patient population.

Data sources and searches
A SLR was conducted to identify key literature evaluating the use of PT in the TCP population specific to each of the following domains: (1) Randomized controlled trials (RCTs): namely, efficacy and safety, and (2) Realworld evidence (RWE): namely, epidemiology and treatment patterns, effectiveness and safety, as well as humanistic and societal burden, and (3) Economic burden. Separate and unique searches were performed for each domain. Efficacy and effectiveness were reviewed separately. Efficacy is defined as the evaluation of whether an intervention produces the expected result under ideal circumstances, such as RCTs, whereas effectiveness is a measure of the degree of beneficial effect in 'real-world' clinical settings [37].
Publications indexed from 1998 to June 27, 2018 (May 23, 2018 for the economic burden domain) were identified from the following sources: MEDLINE® (1946 to present); Embase (1974 to present); Cochrane Database of Systematic Reviews; Cochrane Central Register of Controlled Trials; Database of Abstracts of Reviews of Effects; Health Technology Assessment database, UK NHS Economic Evaluation database; and reference lists from relevant systematic reviews. Full search terms and search strategies are provided in the appendix (Appendix Table A1). Additional manual searches were conducted.
This study is reported in accordance with the Cochrane Handbook for Systematic Reviews of Interventions [38] and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Protocols statement [39].

Study selection
Two independent reviewers screened titles and abstracts to identify relevant publications based on pre-defined inclusion and exclusion criteria (see Appendix Table A2), and then full text publications were assessed, with any discrepancies resolved by a third independent reviewer.
The patient population was restricted to adults with TCP receiving PT (e.g. such as, but not limited to, patients with chemotherapy-induced TCP, patients with bone marrow suppression, and patients with or without CLD undergoing elective invasive procedures).
Specific to epidemiology, treatment patterns and economic burden, only studies that pertained to the USA, Japan or European Union Five (France, Germany, Italy, Spain and UK) were included. For the remaining searches, studies were not restricted to a geographic location.

Data extraction and quality assessment
Data from studies meeting pre-specified criteria were extracted using Excel tables by 1 reviewer and validated for accuracy and quality by a second reviewer. Outcomes of interest varied according to the domain evaluated, and are summarized in Appendix Table A2. For RCTs, quality was assessed using the Cochrane Risk of Bias Tool for RCTs [40], and RWE studies were assessed using the Newcastle-Ottawa Scale [41]. A risk of bias assessment in RCTs is provided in Appendix Table A3 and a quality assessment of RWE studies in Appendix Table A4.

Data synthesis and analysis
Detailed evidence tables were created and studies summarized by project reviewers. Due to the heterogeneity of interventions examined, and the range of methods between studies, tables were designed to capture relevant study findings. Study results are presented as a descriptive narrative synthesis.

Results
A total of 3425 abstracts were identified through database searches as well as additional sources (Figure 1). Following omission of duplicates (n = 625) 2800 abstracts were screened and 2456 excluded. A total of 344 full-text publications were assessed and from these 150 were excluded leaving 194 publications, with 190 studies included (see Appendix Table A5).

RWE: epidemiology and treatment patterns
The treatment patterns of PT were identified in 79 publications. The patient populations were TCP overall in 55 publications, TCP and elective invasive procedure in 20 publications and CLD TCP and elective invasive procedure in 4 publications. The epidemiology and treatment patterns reported in 77 of these 79 publications are summarized in Appendix Table A6. The remaining 2 studies looked at either compliance among 113 patients who received PT [42] or the role of the CD40 ligand in adverse reactions to PT [43].
Patient types in which platelet transfusions were used The use of PT was reported in several studies including large cohorts (>1000 patients with TCP). A review of admissions to a single institution over 5 years, found that of 40,693 patients, 9158 (22.5%) patients had TCP (PC <100 × 10 9 /L) [44]. Approximately a quarter (24.5%) of those patients with TCP were transfused platelets. Use of PT in a total of 3743 patients with chronic ITP was analyzed and 1.7% received platelets [45]. In a population of 18204 patients undergoing interventional radiology procedures, 2060 (11.3%) had a PC ≤100 × 10 9 /L prior to their procedure [46]. Approximately a tenth (9.9%) of these patients received pre-procedural platelets, and their median baseline PC was 39 × 10 9 /L compared to 77 × 10 9 /L for those who did not require pre-procedural PT. In a study of 47,159 patients undergoing chemotherapy for solid tumors, over 4800 patients had a PC ≤150 × 10 9 /L [47]. PTs were reported in 2.5% of the 47159 patients treated.
Platelet transfusions by age Use of PT by age was reported in one study treating patients with non-Hodgkin lymphoma [48]. A total of 23 of the 108 patients received platelets during treatment. A quarter (26%) of patients aged 65 years or older required a PT, whilst a smaller proportion (18%) of patients aged less than 65 years required a PT.

Therapeutic versus prophylactic platelet transfusions
The use of PT for therapeutic or prophylactic purposes was reported in 3 studies. Over 7400 PTs administered to 503 patients over 6 months were analyzed [49]. Patients receiving prophylactic PT were compared to those receiving therapeutic PT. Nearly three quarters (74%) were given prophylactically for bleeding risk, whilst 18.1% were given therapeutically for active bleeding. A small proportion (2.8%) of transfusions was administered prophylactically for procedures. The reasons for 904 PTs administered to 296 patients with cancer were reported by Habr in 2015 [50]. Patients received 33.2% of transfusions for prophylactic reasons (80% for when PC was below 20 × 10 9 /L), 28.4% for 'securing an invasive procedure' and 38.4% for the treatment of bleeding. In another study, the use of PTs was analyzed in 106 patients who received 140 autologous peripheral blood stem cell transplantations [51]. Patients received a total of 235 transfusions, of which 65.5% were for prophylaxis (including fever of unknown origin and septicaemia) and 34.5% for therapeutic indications. Mean baseline PC in those receiving PT was 34.8 × 10 9 /L compared to 93.0 × 10 9 /L in those who did not receive PT. Giannini et al. studied 121 patients with liver cirrhosis, and the prevalence of TCP (PC <150 × 10 9 /L) and severe TCP (PC <75 × 10 9 /L) was 84% and 51%, respectively [54]. A pre-procedural PT was administered to 7 (14%) of 50 patients with TCP who had an invasive procedure. There was no significant difference in mean PC between those who received PT (45.3 × 10 9 /L) and those who did not (51.9 × 10 9 /L). Of the patients with severe TCP, 32 patients had an invasive procedure. The proportion of bleeding and nonbleeding patients with severe TCP who had an invasive procedure and received prophylactic PT was 40% and 14%, respectively. A total of 363 patients with cirrhosis were included in a study by Napolitano et al. examining PC and bleeding following invasive procedures [55]. Bleeding events were recorded in 8 patients with a pre-procedural PC <150 × 10 9 /L, 5 of whom had received prophylactic PT. Of an additional 10 patients with a pre-procedural PC <150 × 10 9 /L who did not experience post-procedural bleeding, 6 patients had received prophylactic PT. The authors reported that post-infusion PC was barely affected.
In summary, the use of PT in patients with TCP varied widely across studies, from 0% to 100% of patients; in studies with large cohorts (>1000 patients with TCP), PT administration ranged from 1.7% [45] to 24.5% [44] of patients. When indications for use were reported, transfusions were generally prophylactic rather than therapeutic.

RCTs: efficacy and safety
The efficacy and safety of PTs were discussed in 47 publications covering 43 primary RCTs, with 1 publication including 2 RCTs [56] and 5 publications [57-61] including additional analyses of 4 of the RCTs [62][63][64][65]. Prophylactic PT was the intervention in 34 of the 43 primary RCTs, being compared with no treatment in 5 studies, with other treatments in 2 studies, and included in both intervention and comparator arms in the remaining 27 studies. The efficacy and safety results for the 7 RCTs comparing prophylactic PT with best supportive care alone or with other treatments are summarized in Table 1. The remaining 9 of the 43 primary RCTs looked at other drugs to treat TCP, with therapeutic PT included as best supportive care in both arms. There was significant heterogeneity in primary outcome measures, with some reporting the effects on PC and requirement for therapeutic PT, while others focused on adverse events (AEs) including risk of bleeding and transfusion reactions.
Underlying etiology of TCP was CLD in 8, chemotherapy-induced in 6, Dengue fever in 2, hematological malignancies in 2, and mixed in 25 (broadly 'hematological' in 16) RCTs. All were performed in a non-emergency setting, when patients were thrombocytopenic but not actively bleeding, and PT was used prophylactically to raise PC. Most studies allowed for therapeutic use of PT in any of the treatment arms if patients began actively bleeding, and strategies of prophylactic versus therapeutic use were formally compared in 1 study [68].
Platelet transfusions: bleeding events and platelet count response Only 5 RCTs (of the 7 in Table 1) compared prophylactic PT with no intervention/best supportive care. In 2 studies in patients with Dengue fever prophylactic PT did not significantly reduce bleeding rates, with 1 study [66] showing the primary outcome of clinical bleeding occurred in 21% of the PT group vs 26% of the controls (p = 0.16), and the other that prophylactic PT did not prevent severe bleeding or shorten time to bleeding cessation [26]. Lye et al. reported that prophylactic PT was associated with more AEs (13 vs 2 in control group, p = 0.0064), although most were nonsevere and all fully resolved [66]. The effect of prophylactic PT on PC was transient, with no difference in mean daily PC between the 2 groups except on Day 2. Khan-Assir et al. [26] reported that approximately half the patients showed no response to PT when measured by post-transfusion platelet increment (PPI), although overall PPI was higher at 24 and 72 h post-transfusion in the group who had received Those who had prothrombin time 14.7-20.0 s, PC 40-100 × 10 9 /L, and/or active treatment with acetylsalicylic acid were randomized to either fresh frozen plasma (FFP) and/or prophylactic PT or no transfusion. Median blood loss and incidence of intratracheal bleeding were similar between the 2 groups.
Two RCTs (included in Table 1), both in patients with CLD and TCP, compared prophylactic PT to other treatments. Basu et al. randomized patients to either prophylactic PT or a TPO-R agonist (romiplostim or eltrombopag) prior to elective percutaneous liver biopsy [59,64]. Following treatment the pre-procedural PC achieved was significantly lower with prophylactic PT (183.8 × 10 9 /L) than romiplostim (232.0 × 10 9 /L, p < 0.05) but similar to eltrombopag (189.9 × 10 9 /L, p = not significant [NS]). No post-biopsy bleeding or hematoma was observed in either group. Stanca et al. [69] reported that intranasal desmopressin was as effective as FFP and/or prophylactic PT in achieving hemostasis in patients with CLD and TCP undergoing dental extraction.

Platelet doses
Six RCTs reported on patient response to different platelet doses, with most using bleeding risk as the primary endpoint and threshold for prophylactic PT of PC <10 × 10 9 /L. The studies differed in their definitions of low (1.1-3.1 × 10 11 ), standard/medium (0.5-6 × 10 11 ) and high (1-5 × 10 11 ) dose PT, with significant overlap between groups. Although median number of PT was usually higher with lower doses of platelets, the median number of overall platelets transfused was lower [63,[70][71][72]. Overall, WHO bleeding ≥ grade 2 did not vary significantly between different platelet doses [63,70,71,73], but 1 study [73] was stopped early due to higher rate of grade 4 bleeding in the low dose group (5.2% vs 0% with standard dose). One study showed that relative risk of requiring subsequent therapeutic PT was higher with low platelet doses and transfusion-free interval was shorter [74].

Platelet storage, administration and preparation
The remaining 21 platelet intervention RCTs focused on different methods of platelet storage, administration and preparation, including pathogen inactivation and white blood cell depletion, and the primary endpoints are presented in Appendix ) increased PC and thereby significantly more patients met the primary outcome of reduced need for pre-procedural prophylactic PT and any bleeding rescue therapy, including therapeutic PT, compared to placebo (65%-93% out of a total of 624 patients combined across studies for TPO-R agonists vs 13%-38% out of a total of 475 patients combined across studies for placebo, all p < 0.01). Two further studies in patients with hematological malignancies suggested that TPO-R agonists (eltrombopag and romiplostim) may be an efficacious alternative to prophylactic PT [95,96]. The final study, in chemotherapy-induced TCP patients, showed that the thrombopoietic agent pegylated recombinant human megakaryocyte growth and development factor can improve PC and reduce need for therapeutic PT [97].
In summary, data from RCTs that compared prophylactic PT with either no intervention or best supportive case were mixed regarding the effect of prophylactic PT on increasing PC and reducing bleeding risk.

RWE: effectiveness and safety
The effectiveness and safety of PTs was discussed in 75 publications, of which 49 included populations of TCP overall, 20 of TCP and invasive elective procedure and 5 of CLD TCP and elective invasive procedure. Data for the effect of PTs on PCs, bleeding and other safety-related events are summarized in Appendix Tables A8-A10. One paper only reported on PT use around the time of delivery in pregnant women with TCP [98].

Platelet transfusions and platelet count response
The effect of PTs on PC was reported in 36 publications with readings 10 min to 72 h post-transfusion, when reported. PTs were generally effective, to some degree, with an increase in PC seen in most patients (−4 × 10 9 /L to 262.9 × 10 9 /L). One study, that focused on 27 patients receiving PT in the ICU setting, reported that a single PT resulted in a median PC increase of 14 × 10 9 /L measured at 5.2 h post-PT (based on 57 non-overlapping PTs), however, no PC increase was reported for 13 patients (48.1%) after 17 PTs [99]. In another study, also specific to the ICU setting, based on 5700 PTs, the median PC increase after a single PT was 23 × 10 9 /L measured at a median of 7 h post-PT, however 21.8% of transfusions had an ineffectual PC increase of <5 × 10 9 /L [100]. The independent predictor of an ineffectual response with the greatest odds ratio (1.84 [95% confidence interval: 1.24-2.73], p = 0.0024) was liver disease followed by a number of other factors. Refractoriness, defined as a PC increase <5 × 10 9 /L following PT or 3 consecutive days of PT, was reported in several studies. In a publication examining the safety of endoscopy interventions, 23% of patients were refractory [101]. In a second publication of all hospitalized patients receiving PT over a 6-month period, 22% of patients were refractory [102]. Charbonnier et al. found that only 10 patients out of a total of 1408 with acute myeloid leukemia (AML) were refractory to PT, whilst Wandt et al. stated that platelet refractoriness of clinical significance related to alloimmunization was not reported in their study in AML patients [103,104].

Platelet transfusions and bleeding events
Bleeding events (including minor, major and fatal) were reported in 31 publications (Appendix Table A9). Therapeutic PT and prophylactic PT were compared in the study by Charbonnier et al. of patients with AML [103]. Death from hemorrhage was reported in 2.4% and 0.4% of patients receiving therapeutic PT or prophylactic PT, respectively. The clinical impact of PT in patients with TTP was also investigated [105]. Of 54 patients analyzed, platelets were administered to 33 patients. Death due to hemorrhage was reported in 1 of the 33 patients who received PT and in 1 of 21 patients who received no PT. In a study of over 10,000 hospitalizations for TTP, PT was associated with higher odds ratios of thrombosis and death [21]. Death due to bleeding was also reported in 4 other publications [106-109] (Appendix Table A9). In 50 patients with CLD who underwent invasive procedures, peri-procedural bleeding was reported in 10 patients, of which 40% received prophylactic PT [54].
In 874 patients with cirrhosis, 21 patients (2.4%) had major bleeding after invasive procedures [110]. Platelets were administered pre-procedure to 4 patients and 1 of these developed major bleeding. Post-procedure, PTs were given to 5 patients, of which 2 were in the major bleeding group (in 1 of these PT was given with FFP). The effect of PT in 79 invasive procedures in 42 patients with cirrhosis was analyzed [111]. In 61 procedures, patients received platelets pre-procedure with 3 patients experiencing post-procedural bleeding. Of 18 invasive procedures where patients did not receive platelets pre-procedure, no patient experienced post-procedural bleeding.
Platelet transfusions and safety-related events A total of 44 publications reported safety-related events. Eighteen studies included mortality data for patients with PT compared to those without PT ( Table 2). Statistically significant odds or hazard ratios were reported for increased risk of death after PT in 8 studies [44,46,112,[116][117][118]123,125]. However, in 1 study, death was more frequently reported in patients who had not received PT compared to those that had (2.6% and 0%, respectively, p = NS) [113]. Despite a higher frequency of deaths for patients administered platelets, the risk of death with PT was actually lower following regression analysis adjusted for covariates, including nadir PC, red blood cell transfusion and need for hemodialysis [112]. Another study reported death in the same proportion of patients with or without PT (24% for both, p = 0.97) [105].
The transfusion of platelets can result in infection and, in severe cases, sepsis. Patient death due to sepsis was reported in 3 publications [105,127,128]. Several studies also reported infection or sepsis associated with PT. Wandt et al. investigated at two different PC triggers (PC count of <10 × 10 9 /L or <20 × 10 9 /L) for prophylactic PT in patients with AML [104]. Four of the 7 patients with major (WHO grade 3 or 4) bleeding complications (all in group with PC trigger of <20 × 10 9 /L) had associated serious infections and sepsis. The safety post-implantation of totally implantable venous access ports was assessed in 181 patients with TCP (55, 58 and 68 patients with mild (PC: 100-150 × 10 9 /L), moderate (PC: 50-100 × 10 9 /L) and severe (PC: <50 × 10 9 /L) TCP, respectively) [129]. Platelets were only administered to patients with severe TCP. Infection was reported in 4% and 9% of patients with mild or moderate TCP, respectively. In the patients with severe TCP, 10% of patients had an infection. Complication rates following dental extraction in 68 patients with TCP were examined, with 32 patients requiring PT [130]. There were 2 cases of infection (2.9% of study population), and 1 of these had received PT. In a multi-center study of patients admitted to ICUs between 2008 and 2013, the association between PT and hospital-acquired infection was investigated [17]. PT was associated with infection. 7.7% of patients with PT had infections compared with 1.4% without PT (p value < 0.01). Infection was also reported in 2 studies in patients with CLD. In 1 study, no significant odds ratio associated with PT for infection was reported [52], whilst a different study reported a statistically significant odds ratio (2.53 [2.0, 3.2], p < 0.001) [131].
In summary, real-world observational studies demonstrated that PT was generally effective to some degree in increasing PC but did not always translate into a clinically significant increase in PC nor a reduction in bleeding risk. While these studies demonstrated an association between PT and safety events, results with regards to increased mortality rate following PT were varied, with either no difference or an increased mortality rate associated with PT.

RWE: humanistic and societal burden
Two publications discussed the humanistic and societal burden of PT. The first reported completed surveys from 294 patients classified as a population of TCP overall and 73 surgeons and anaesthesiologists. Fewer patients rated transfusion as 'very often' or 'always risky' compared to their physicians (20% and 39%, respectively, p = 0.001) [132]. The second study recruited a population of TCP and elective invasive procedure of 25 patients who were receiving their first transfusion. One third of patients were 'concerned or worried' about receiving the transfusion [133].

Economic burden
The economic burden of PT was discussed in 26 publications. Most (19 of 26) included a population of TCP overall, and 4 included a population of CLD TCP and elective invasive procedure. Patients in the 3 remaining publications were not classified but involved PT patients in 2 publications. The third was a survey of the National Blood Collection in the USA and included information from hospitals, blood centers and cord blood banks. Costs, including those associated with transfusion-related events, were reported in 23 of the 26 publications. These data are summarized in Table  3. The 3 remaining publications reported on the number of units transfused or transfusion episodes [49,104,154].
The overall costs of PTs were assessed at a tertiary care hospital in the USA [102] where a median hospitalization cost of $27,750 was reported. This varied depending on the service used with internal medicine/other costs lowest at $13,856 and the highest cost associated with BMT ($58,729). Notably, there was a statistically significant difference in mean cost between refractory and non-refractory patients ($103,956 and $37,818, respectively; p < 0.001).
A cost analysis based on data from the Trial of Prophylactic Platelets trial [65,140] suggested that prophylaxis resulted in lower rates of bleeding compared to no prophylaxis. The authors examined the cost of prophylaxis and no prophylaxis policies. The total health care costs per 30 days, per patient were statistically significantly higher in the prophylaxis group compared to the no prophylaxis arm ($16,753 and $14,992, respectively). There was a statistically significant difference in favor of no prophylaxis in the cost of the units transferred and of the investigations and medications between the prophylaxis and no prophylaxis arm. In a separate publication, the health care costs of patients undergoing chemotherapy cycles complicated by TCP was compared to cycles in those same patients not complicated by TCP [142]. The mean cost of providing prophylaxis for a cycle was $792. Furthermore, the mean cost of treatment of bleeding was higher in TCP cycles compared to control cycles ($237 and $14, respectively). The mean cost of treating other complications was also higher in TCP cycles compared to control cycles, resulting in mean total costs of $6866 for TCP cycles and $4875 for control cycles (p < 0.001).
The wholesale acquisition cost of PT was compared to that of the TPO-R agonists romiplostim and eltrombopag in a randomized, double blind clinical pilot trial [64]. Both romiplostim and eltrombopag increased preoperative PC to a similar or greater extent as PT with a cost of less than 50% of PT ($2284 and $2991, based on off-label dosing, respectively compared to $7500 for PT). In another study, the usual standard of PTs was compared to the use of recombinant human interleukin-11 (rhIL-11, oprelvekin) for prophylaxis of severe chemotherapy-induced TCP [141]. The overall cost of the usual standard of PT over the 3-week chemotherapy cycle was $3495 compared to $5328 for rhIL-11 group over the same period. Although, the rhIL-11 group had fewer PTs and therefore avoided potential adverse reactions to transfusion, the cost of the drug was substantial, resulting in a higher overall cost.
An additional cost associated with PT is the treatment of transfusion-related events. The annual cost of such events was reported in 2001 for the USA [135]. Hospitalization due to transfusion-related sepsis cost $6408, whilst treating hepatitis B or C virus transfusion-related events cost between $1228 and $17,412 a year. In a further publication the estimated average treatment cost for acute-transfusion reactions (ATRs) in Germany was reported [136]. Grade 1 ATRs such as chills, fever and urticaria cost on average €104. For Grade 2 ATRs such as urticaria with itching, hypotension or fever >40°C , the average cost was €238. Finally, Grade 3 allergic and bacterial ATRs cost €1200 and €21,984, respectively.
Across publications reporting PT costs in CLD patients, the reported estimated cost for 1-2 PTs ranged from $500 [138] to $1639 [69], while the total estimated costs of PT were reported as $5258-13,117 in 1 publication and $4800-11,000 in another [134,138].
In summary, the available data show that PT represents a substantial cost burden. The economic costs associated with PTs extends beyond the collection and delivery of transfusion units, and there are significant costs associated with transfusion-related events.

Conclusions
PT has been considered the 'gold-standard' treatment for increasing PC in thrombocytopenic patients [155], and is recommended in current guidelines [10,11,13]. However, researching the efficacy and effectiveness . Some variability may be explained by the differences in patient baseline characteristics both between and within studies, including type of procedure, underlying medical condition(s), and patient age. Evidence for the impact of these differences is largely lacking, although one study in patients with non-Hodgkin lymphoma showed that older patients had a higher incidence of PT [48].
In one real-world observational study comparing therapeutic PT versus prophylactic PT in patients with AML [103], death from hemorrhage was reported in more patients receiving therapeutic PT than prophylactic PT. Also, in one RCT comparing therapeutic PT versus prophylactic PT [68], the grade 4 bleeding risk was greater in the therapeutic group; however therapeutic use resulted in a 33% reduction in the mean number of PTs given. Data from trials directly comparing prophylactic PT with no intervention/best supportive care were mixed regarding the effect of prophylactic PT on increasing PC and reducing bleeding risk [26,[65][66][67][68]. One study demonstrated increases in PC with PT but rates of non-response were high [26], and in another study PC increases were only transient [66]. Studies reporting on the response to different platelet doses suggested that lower doses led to an overall reduction in number of platelets transfused [63,70-72], although one study was stopped early due to a higher rate of grade 4 bleeding in the low dose group [73]. PC threshold for intervention varied widely, due to heterogeneity in patient type and procedure, and ranged from <10 to ≤100 × 10 9 /L [46,139]. This is reflected in a recent guideline update from the American Society of Clinical Oncology that recommends PC thresholds ranging from 10 to 50 × 10 9 /L depending on patient characteristics or whether the patient will undergo invasive procedures [10]. These data support further investigation of PT strategies and doses to ensure that the benefits of PTs are maximized whilst minimizing the risks.
Real-world observational studies demonstrated that although PT might be effective to some degree in increasing PC, it did not always translate into a clinically  dollar; WBD = whole blood derived. a PICOS classification based on population, interventions, comparators, outcomes and study design. b A stringent prophylactic-platelet transfusion policy <10 × 10 9 /L for stable patients and <20 × 10 9 /L in the presence of major bleeding or additional risk factors. A trigger of <50 × 10 9 /L was introduced for patients undergoing invasive procedures. c Units of currency not provided in the publication. significant reduction in bleeding risk [54,111], and refractoriness was a significant problem [101,102]. PT was also associated with safety concerns, including an increased infection risk in some studies [17,131]. Studies varied as to whether mortality rate was higher following PT, with 8 showing a significantly increased odds ratio/hazard ratio [44,46,112,[116][117][118]123,125], although in one the rate was lower after regression analysis adjusted for covariates [112], and others showed no difference [105]. It is noteworthy that in individual RWE studies comparing PT versus no PT, patients receiving PT might have been at higher risk than patients who did not receive PT which can create potential bias when comparing results.
Publications that discussed the humanistic burden of PT demonstrated that 20% of patients rated transfusion as 'very often' or 'always risky' [132] while one third of patients were 'concerned or worried' about receiving PT [133]. Limited research on the humanistic and societal burden of PT means that the impact of this procedure on patient experience and quality of life is also largely unknown.
Evidence is lacking for the cost-effectiveness of PT, and most studies examining the economic burden of PT in this SLR did not distinguish between prophylactic and therapeutic use. Available data show that PT represents a substantial cost burden [102], and this includes the higher costs associated with the management of transfusion-related AEs [135,136] and particularly those associated with PT-refractory status [102]. Measures to decrease PT-related costs could include implementing uniform management algorithms, as well as updating guidelines and protocols to include appropriately licensed treatment alternatives such as pharmacotherapy options.
Due to this growing evidence of the limitations of PT, alternative treatment approaches to increase PC are being investigated. Studies with TPO-R agonists demonstrated promising results in reducing patients' need for pre-procedural prophylactic PT and any bleeding rescue therapy, including therapeutic PT, from 65% to 93% vs 13% to 38% for placebo [56,91-94]. Although PTs are used to varying degrees for increasing PC in TCP, it is important to understand the limitations of PTs, and to explore the use of alternative treatment options where available.

Disclosure statement
AN reports research support from Amgen, Bristol-Myers Squibb, GlaxoSmithKline, Octapharma and Rigel and consulting for Amgen, Angle, Argenx, Dova, GlaxoSmithKline, Novartis, ONO Pharmaceutical, Rigel, Shionogi and UCB Biosciences. HL reports grants from Novartis, grants and personal fees from Rigel, personal fees from Dova, grants and personal fees from Argenx, and grants and personal fees from Syntimmune (now Alexion), all outside the submitted work. He also reports previous consulting for Dova on the use of avatrombopag for ITP. MP-R reports personal fees from Shionogi, outside the submitted work. AT reports grants from Shionogi, Kyowa Hakko Kirin, Chugai Pharmaceutical, Zenyaku Kogyo, and the Ministry of Health, Labour and Welfare of Japan, all during the conduct of the study. He also reports grants from the Ministry of Education, Culture, Sports and Technology of Japan, Astellas Pharma, Yamada Bee Farm, and Pfizer, all outside the submitted work. RT reports personal fees of a consultation fee from Shionogi and lecturer fees from Shionogi, Eisai, MSD, Gilead Sciences, Fujifilm Wako and AbbVie, all outside the submitted work. ZY reports being a consultant for Gilead Sciences, Intercept, Bristol-Myers Squibb, Novo Nordisk, Viking, TERNS, Shionogi and AbbVie, all outside the submitted work. RB is an employee of Shionogi. AJ, JL and VT are employees of Creativ-Ceutical.

Funding
Study design and conduct, data collection, management, data analysis and interpretation, and manuscript development was funded by Shionogi.

Availability of data and material
The datasets generated and analyzed during this study are available from the corresponding author on reasonable request.

Notes on contributors
Adrian Newland is Professor of Hematology at Barts NHS Trust, UK. He is ex-Chair of the UK National Blood Transfusion Committee and currently chairs the DMC for NHS Blood and Transplant. He has clinical and research interests in TCP and has been Principal Investigator on many treatment studies and has published extensively on the condition.
Roy Bentley MSc, PhD, is an employee of Shionogi with an interest in health economics and outcomes research, patient-reported outcomes and healthcare policy.
Anna Jakubowska received a PhD in Pharmaceutical Sciences from Jagiellonian University and is an employee of Creativ-Ceutical with experience in the preparation of systematic reviews and health economics and outcomes research.
Howard Liebman is the Donald I Feinstein Professor of Hematology and director of the Diagnostic Hematology group of the Jane Anne Nohl Division of Hematology and Center for the Study of Blood Diseases, USA. His research and publications have centered upon disorders of hemostasis and thrombosis, with a particular interest in platelet disorders and immune TCP.
Joanna Lorens is an employee of Creativ-Ceutical with experience in the preparation of systematic reviews.
Markus Peck-Radosavljevic is Professor of Medicine and Chairman at the Department of Gastroenterology & Hepatology, Endocrinology, Rheumatology and Nephrology at Klinikum Klagenfurt in Klagenfurt, Austria. He began his research in liver disease in the field of TCP in portal hypertension in 1994 and has been working on the role of TPO in advanced stage liver disease for many years, including his role as one of the lead-investigators in the L-PLUS-2 trial. He has also been running a translational lab testing novel approaches to the treatment of portal hypertension and hepatocellular carcinoma (HCC) for many years, while working at the Medical University of Vienna. Since January 2016, he has taken over as Chair of Internal Medicine and Gastroenterology in Klagenfurt, which also includes the Endocrinology, the Rheumatology, the Nephrology (including Hemodialysis) and the Emergency Medicine service. He has set up a new Clinical trials study unit there and is running a broad range of clinical trials in various different indications, including HCC, cholestatic liver diseases and non-alcoholic steatohepatitis.
Vanessa Taieb studied at the Ecole Nationale de la Statistique et de l'Analyse de l'Information and is an employee of Creativ-Ceutical with experience in the preparation of systematic reviews and health economics and outcomes research.
Akiyoshi Takami is Professor of Hematology, Department of Internal Medicine, Aichi Medical University, Japan; Executive director for the Japanese Society for Laboratory Hematology; Supervisor for the Japan Society for Hematopoietic Cell Transplantation; and Chairperson for the Subcommittee on the Guideline for the use of PT preparation of the Japan Society of Transfusion Medicine and Therapy.
Ryosuke Tateishi has worked in the HCC field since 1998 and has served on two guideline committees: Japanese Clinical Practice Guideline for liver cancer and Asian Conference on Tumor Ablation Guideline.

Zobair M. Younossi is Chairman of Medicine and Outcomes
Researcher with extensive experience in all types of health services research, including systematic reviews. He is extensively published with a Scopus h-index of 81.                     All patients in the overall study population received PTs. n Value corresponds to percentage of PTs administered to patients in this subgroup out of total PTs administered. o The article stated that 5 patients received 0 PT and that 42 patients did receive PT. However, the article also stated that all patients with PC <20 × 10 9 /L received PT and that 43 patients had PC <20 × 10 9   Organization. a With a per-protocol population with a pre-specified margin of 12.5%, non-inferiority was not achieved for PCT in PAS vs untreated in plasma (4.4%; 95% CI, −4.1% to 12.9%), but was achieved for PCT in PAS vs untreated in PAS (2.6%; 95% CI, −5.9% to 11.1%). b Based on a non-inferiority test with a non-inferiority margin of 0.125 (one-sided 95% CI of difference: −1, 0.07), p value <0.05 indicates that PCT was not inferior to control. c Mean paired difference in CCI was 2400±4301 (one-sided non-inferiority test, upper bound of one-sided 95% CI [of CCI], 4040). Specified non-inferior margin was 2200, indicating that the study failed to show non-inferiority within this specified margin of inferiority. d Non-inferiority criterion was met for the intention-to-treat analysis (indicating that PCT was not inferior to control), but not for the per-protocol analysis. e All measured at 1 h post-transfusion.   Data presented as median (interquartile range). c Complete response was defined as platelet count ≥100 × 10 9 /L. d Refractoriness defined as CCI <4500/µl. e Effective rate is defined as the percentage of cases with increasing platelets (≥20 × 10 9 /L). f Patients classified as refractory if increase in PC <5 × 10 9 /L after transfusion. g Data presented as median (5th and 95th percentile). h Good response defined as 24 h CCI >4500. i Good increment was defined as CCI ≥7500 m 2 /muL at 1 h and/or CCI ≥4500 m 2 /muL at 24 h post-transfusion. j Good response defined as 1 h CCI ≥7.500/muL or 24 h CCI ≥4.500/muL. k Good response defined as increase in platelet count greater than 50 × 10 9 /L at 24 h. l Increment per 10 × Lm 2 .  WHO bleeding Grade 1-4 (89%), Grade 3-4 (8%), death due to bleeding complication (2%) CI = confidence interval; CLD = chronic liver disease; CTCAE = Common Terminology Criteria for Adverse Events; EBL = endoscopic variceal band ligation; FFP = fresh frozen plasma; HIT = heparin-induced thrombocytopenia; HR = hazard ratio; ITP = immune thrombocytopenia; IVIg = intravenous immunoglobulin; OR = odds ratio; PC = platelet count; PICOS = population, interventions, comparators, outcomes, study design; PPT = prophylactic platelet transfusion; PT = platelet transfusion; TCP = thrombocytopenia, TPT = therapeutic platelet transfusion; TTP = thrombotic thrombocytopenia purpura; TXA = tranexamic acid; WHO = World Health Organization. a N of patients unless otherwise stated. b There were 43 catheterization attempts in 41 patients. c A total of 617 procedures were performed in 395 patients. PTs were administered within 24 h preceding the procedure in 329 patient encounters.