Optimal timing and dosing of platelet transfusions
Nancy M Heddle, McMaster University, HSC 3N43, 1200 Main Street West, Hamilton ON, L8N 3Z5, Canada
Background Over the past 20 years there have been more than 20 randomized controlled trials (RCTs) that have investigated various aspects of platelet transfusion therapy in haematology/oncology patients. These studies have focused on the best platelet product, the importance of ABO compatibility, pathogen inactivation of platelets, platelet triggers and the optimal platelet dose.
Aims This article summarizes current evidence to support the timing and dosing of platelet transfusions and to explore some ideas of where clinical research in this area may be heading.
Materials and Methods The articles reviewed in this presentation were identified through a search of PubMed using the term, platelet transfusion and setting limits to identify clinical studies, human studies and manuscripts in English.
Results and Discussion Three RCTs have informed practices around platelet transfusion trigger with the largest study by Rebulla et al., being the primary study that has changed practices worldwide, with a move towards a lower prophylactic platelet transfusion trigger of 10 × 109/l. Two groups (Germany and Oxford, UK) are currently investigating whether we can push the boundaries of prophylactic platelet transfusions even further by eliminating this form of therapy. Preliminary results from these studies have been published but we will await the final results to determine whether this research will indeed change practice. Over the past year there has also been two major studies (one by the BEST Collaborative, and the second by the US Transfusion Medicine/Hemostasis Network), that provide new information to guide platelet dosing. The Study by the BEST Collaborative (SToP) compared low dose platelets to standard dose platelets with WHO bleeding greater than or equal to Grade 2 as the primary outcome. The US study (PLADO) compared three doses (low, medium and high) and measured the same outcome (WHO bleeding ≥ Grade 2).
Conclusions Although all of these studies further our knowledge to prescribe platelet transfusions, they also raise some interesting questions about the clinical relevance of the outcomes that we are currently using for these studies. The trend over the past decade has been to use bleeding as the primary outcome; however, bleeding is a complex composite outcome (Grades 2, 3 and 4) comprised of some surrogate components (Grades 2 and 3). It is also an outcome that may be difficult to measure and grade in a consistent and reliable manner. The clinical relevance of this outcome is also complex and may vary depending on the perspective from which it is viewed.
A prescription for a medication or drug typically identifies specific characteristics such as: the type of product (generic or brand); the dose; and frequency and the duration of administration. Blood products such as platelets are also drugs. Should the same prescription characteristics apply to orders for blood products? Although ideally the answer to this question should be yes, platelets (and other blood products) have unique characteristics that make them different from most drugs. Platelets are derived from volunteer blood donations, whereas most drugs are manufactured from chemical compounds; platelets are a limited resource dependent upon availability of donors, whereas manufactured drug production can be ramped up or down depending on demand; the number of platelets per product (dose) is variable depending on the donor’s blood count and the collection process, whereas other drugs can be manufactured such that each capsule or tablet contains a specific concentration of the active compound; the duration of the treatment period with platelets is variable and unknown when treatment is started, whereas other drugs are given for a specified duration of time. These are just some of the challenges associated with the optimal platelet prescription.
In 1962, Gaydos  published on the relationship between platelet count and bleeding. As platelet counts dropped, the frequency of bleeding increased. These findings led to randomized controlled trials (RCTs) investigating the efficacy of prophylactic platelet transfusions using a trigger in the range of 20–30 × 109/l. In the 1970s, there were three RCTs conducted to determine the efficacy of prophylactic platelet transfusions to prevent bleeding complications in patients with chemotherapy induced thrombocytopenia [2–4]. All three studies were conducted in the US and in total they enrolled 108 patients (two studies enrolled adults and 1 enrolled paediatric patients). A meta-analysis looking at the outcomes of all cause mortality, mortality from bleeding, remission rates and total number of red cell transfusions showed no differences between the patients receiving prophylactic platelet transfusions, and those receiving only therapeutic platelet transfusions; however, true differences may not have been detected because of the small sample sizes (low power) in each of the studies . The meta-analysis did show significantly more bleeding events in the therapeutic platelet arm of the study (RR 0·49; 95% CI 0·28, 0·87) and a higher mean number of platelet transfusions required (WMD 15·80; 95% CI 2·45, 29·15) for those receiving prophylactic platelet transfusion therapy. Based on these early studies, patient safety trumped any concerns about blood product demand and availability and prophylactic platelet transfusions became the standard of practice with most centers using a trigger of 20 × 109/l.
In 1978, Slichter and Harker  measured stool blood loss in non-transfused patients with chronic thrombocytopenia and showed that significant blood loss (>50 ml/day) did not occur until the patients’ platelet counts were less than 5 × 109/l. Based on these data, clinicians began to question the relevance of the 20 trigger and suggested a lower trigger threshold might be feasible. Between 1991 and 2001, six observational studies were performed to assess the safety of a lower prophylactic platelet transfusion trigger: three were retrospective [7–9] and three were prospective cohort studies [10–12]; however, the scientific evidence that changed transfusion practice came from three randomized controlled trials that compared transfusion triggers of 10 × 109 vs. 20 × 109/l [13–15].
The RCTs evaluating transfusion triggers
The three RCT trigger studies were published between 1997 and 2002. The Italian study by Rebulla  was the largest of the three RCTS, randomizing 276 patients, of which 255 were included in the primary analysis. The patient populations were all adults with a diagnosis of acute myeloid leukaemia and receiving their first course of induction chemotherapy. Patients in the low threshold group received prophylactic platelet transfusions when their platelet counts fell below 10 × 109/l (non-febrile) or if febrile when their platelet counts were between 10 and 20 × 109/l. Patients in the high threshold arm received prophylactic platelet transfusions when their counts fell below the threshold of 20 × 109/l. The primary outcome was the frequency of bleeding (≥Grade 2) categorized on an eight-point scale. Secondary outcomes included mortality, remission rates, severity of bleeding, transfusion requirements, and length of hospital stay.
In 1997, Heckman randomized 82 patients (78 included in the primary analysis) with acute leukaemia (initial induction chemotherapy or reinduction following relapse) to a threshold of 10 × 109/l or 20 × 109/l. All platelet products transfused in this study were apheresis platelets. Bleeding was defined as blood loss documented in the clinical notes or observed by an investigator. Other outcomes included survival (at the time of analysis), remission rates, red cell and platelet transfusion requirements, length of hospital stay and adverse events .
The third study reported by Zumberg in 2002 randomized 159 patients (children and adults) receiving hemopoeitic progenitor cell transplants (autologous and allogeneic). All randomized patients were included in the analysis. Transfusions using apheresis platelets were given in the low threshold group when the patient’s platelet count fell below 10 × 109/l (threshold or 10–15 × 109/l was used if a post-transfusion count at 12 hours was <10 × 109/l). In the high threshold arm, prophylactic platelet transfusions were given when the patient’s platelet count fell below 20 × 109/l. Outcomes included mortality, frequency of bleeding events, red cell and platelet transfusion requirements and duration of hospital stay .
A meta-analysis of the three studies concluded that there were no significant differences between transfusion triggers for the outcomes of all cause mortality, remission rates, red cell transfusion requirements, and the need for HLA-matched platelets. The overall Relative Risk for major or more severe bleeding events was 0·99 (95% CI 0·66, 1·48). Patients transfused at the lower threshold (10 × 109/l) used significantly fewer platelet transfusions and experienced fewer reactions .
Based on the results of these studies, a lower prophylactic platelet transfusion threshold was gradually accepted and most recent guidelines now suggest that a trigger of 10 × 109/l can be used in patients without significant risk factors for bleeding.
Which product is best?
There are three methods of platelet preparation: two involve the separation of platelets from whole blood (platelet rich plasma methods and the buffy coat method) and the third involves the collection of platelets by apheresis. There has been on-going controversy as to which product is best; however, a recent systematic review identified an increased corrected count increment (CCI) at 1 and 24 hours as the only outcome favoring the apheresis platelet product . The clinical relevance of this finding is not known.
Platelet dose – the debate
For more than a decade, there has been a debate as to the optimal dose of platelet transfusions. Some transfusion medicine physicians advocate for high-dose platelet products (>6 × 1011 platelets/product), claiming a longer interval between transfusion and fewer prophylactic platelet transfusion requirements. Patients transfused with the high-dose strategy may also have a lower risk of bleeding because of their higher platelet count. Others have suggested that a low-dose strategy (1·5–2·99 × 1011 platelets/product) would be optimal. Their arguments include: evidence that you only need ∼7100 platelets/ul per day to maintain endothelium integrity; fewer total platelets transfused which would benefit availability, resources and cost; the duration of thrombocytopenia may be shortened as fewer circulating platelets could result in higher levels of thrombopoietin to stimulate thrombopoiesis; and, fewer transfusions could be association with a lower risk of transfusion-related immune modulation (TRIM) .
Clinical studies evaluating platelet dose
There have been seven randomized controlled trials that have addressed the question of optimal platelet dose for prophylactic platelet transfusions [18–24]. The first five studies were published between 1998 and 2005 [18–22]. One of these studies compared standard versus low-dose platelet transfusion strategies , whereas the others aimed to compare standard dose with high dose. Two studies included a third treatment arm with patients receiving a very high-dose product [18,20]. Each of these studies had some methodological limitations that have been summarized in other publications [25,26]. A meta-analysis of these studies concluded that high-dose platelet products were associated with a longer interval between transfusion and a higher post-transfusion platelet count. Although bleeding was assessed to varying degrees in three of the studies, the results were too heterogeneous to determine an overall estimate of the odds ratio; hence, no conclusions regarding this outcome could be made .
Because the question of optimal platelet dose remained an area of debate, two groups designed and conducted multicentre RCTs to try and resolve the platelet dose issue [23,24].
SToP platelet dose study design (Strategies for Transfusion of Platelets)
The SToP study was designed and performed by the BEST Collaborative with participation in six centres in three countries. The study was designed as a non-inferiority RCT to determine if a low-dose prophylactic platelet transfusion strategy in patients with chemotherapy induced thrombocytopenia was not inferior to standard dose platelets as measured by the frequency of WHO bleeding ≥Grade 2. Patients in the low-dose arm of the study received a target dose in the range of 150–299 × 109 platelets per transfusion. Patients in the standard dose treatment arm received platelets with a target dose of 300–600 × 109 platelets/transfusion. Bleeding was documented by performing a daily bleeding assessment (personnel performing this assessment were blinded to the patient treatment allocation), which involved a physical examination of the patient for evidence of bleeding and a series of questions posed to the patient to explore signs of bleeding that may have occurred over the previous 24 hours. The patient’s medical charts were also reviewed to identify any documented episodes of bleeding in the nurses’ and physicians’ notes. The daily bleeding assessments along with relevant laboratory and transfusion data were reviewed independently by two adjudicators and a bleeding grade assigned. Disagreements were resolved through further adjudications and/or consensus.
The sample size required for this non-inferiority study was ∼600 patients; however, the study was stopped by the Data Safety Monitoring Committee (DSMC) after 129 patients had been enrolled because of safety concerns. A prespecified stopping rule allowed the DSMC to stop the study if the difference in the cumulative frequency of bleeding between the two treatment arms exceeded 5% at any time after 50 patients had been enrolled in each treatment arm. The difference reached 5·2% and the study was stopped .
PLADO study design (Platelet dose study)
The Transfusion Medicine/Hemostasis Clinical Trials Network in the US also conducted a multicentre platelet dose study with 17 participating sites throughout the US. There were 1351 patients with hypoproliferative thrombocytopenia, randomized to receive either a low-dose (1·1 × 1011/m2 body surface area), a medium-dose (2·2 × 1011/m2 BSA) or a high-dose (4·4 × 1011/m2 BSA) platelet product. The primary outcome was WHO bleeding ≥Grade 2 bleeding captured through a daily bleeding assessment by study personnel blinded to the treatment allocation that the patient was receiving. The WHO bleeding grade was assigned using an electronic algorithm .
Results from SToP and PLADO
The bleeding outcomes were reported using a similar format for both studies (Table 1); however, secondary outcomes were reported differently allowing for only a few comparisons.
Table 1. Summary of bleeding events in the SToP and PLADO studies
|Grade 1||48 (78·7)||53 (91·4)||–||–||–|
|Grade 2||28 (45·9)||28 (48·3)||58||59||60|
|Grade 3||6 (9·8)||5 (8·6)||9||7||8|
|Grade 4||0 (0·0)||3 (5·2)||3||2||2|
|≥Grade 2||30 (49·2)||30 (51·7)||71||69||70|
|No. of days with ≥Grade 2 bleeding median (IQR)||–||–||1 (0–4)||1 (0–4)||1 (0–4)|
|% (proportion) of days with bleeding ≥Grade 2||8·5 (73/854)||12·1 (111/918)||–||–||–|
In PLADO, the percentage of patients with bleeding in the three treatment arms was almost identical (low dose 71%; medium dose 69%; and high dose 70%). In SToP, the proportion of patients with WHO Bleeding ≥Grade 2 was also similar (low dose 51·7%; and standard dose 49·2%). The bleeding frequencies in the two studies are different by about 20%. This difference is surprising considering that 86·6% of patients in SToP had a diagnosis of acute leukaemia compared to only 45% of patients in PLADO (Table 2). There are several reasons why the frequency of bleeding in the two studies could be variable. The WHO bleeding scale has never been validated; hence, the reproducibility in assigning a bleeding grade may be problematic. Indeed, this was observed during adjudication of the daily bleeding events in SToP, with initial disagreement in Grade of bleeding being observed in 31·2% (359/1150) of cases. The difference may also be because of the method used to assign the grade of bleeding: SToP used adjudication, whereas, PLADO used an electronic algorithm.
Table 2. Summary by diagnostic categories of patients enrolled in the SToP and PLADO studies
|Acute leukaemia||103 (86·6)||573 (45·0)|
|Chronic leukaemia||2 (1·7)||81 (6·4)|
|Lymphoma||7 (5·9)||264 (20·8)|
|Myelodysplasia||3 (2·5)||56 (4·4)|
|Plasma cell dyscrasia||2 (1·7)||0 (0·0)|
|Solid tumour||1 (0·8)||18 (1·4)|
|Myeloma||0 (0·0)||154 (12·1)|
|Other||1 (0·8)||126 (9·9)|
If bleeding is similar regardless of the transfused dose, then what is the benefit to change practice? Both studies found that a low-dose platelet transfusion strategy had shorter intervals between transfusion compared to standard and high dose transfusions (Table 3). Both studies also looked at several of the following outcomes: post-transfusion platelet counts; absolute increments; corrected count increments; number of platelet transfusions; number of days with bleeding ≥Grade 2; total number of platelets transfused (×1011); number of platelet donor exposures; and red cell utilization. Unfortunately the two studies did not report these outcomes consistently; hence, a direct comparison is difficult. We can conclude that the frequency of WHO bleeding ≥Grade 2 is similar regardless of platelet dose transfused. The Grade 4 bleeding events in SToP involved retinal bleeds (2 cases) with no permanent vision impairment and one cerebral bleed. Although these three cases raised concern, they could have occurred all in the low dose arm by chance. Some clinicians would be reassured by a similar frequency of Grade 4 bleeding in all three treatment arms in PLADO. However, both SToP and PLADO showed that a low dose strategy required more platelet transfusion episodes; hence, the costs with a low dose strategy may be increased. PLADO also reported the median total number of platelets transfused and showed significantly fewer platelets being transfused in the low dose treatment arm (Table 3). The assessment of potential benefit of a low or high dose strategy should also include other potential risks, as SToP suggested that a low dose strategy might increase donor exposures (trend but not statistically significant). The investigators from the PLADO study are planning a cost benefit analysis which should address some of these issues.
Table 3. Summary of the secondary outcome results for the SToP and PLADO studies
|No. of platelet transfusion episodes||533||325||2547||1912||1572|
|Post-transfusion plt count × 109/l|
| Median (IQR)|| || ||22 (16–30)||34 (24–48)||50 (33–68)|
|Platelet increment × 109/l|
| Median (IQR)|| || ||10 (5–17)||19 (11–30)||38 (22–54)|
| Median (IQR)|| || ||10 (5–13)||10 (6–16)||11 (6–15)|
|Total number of platelets transfused|
| Median (×1011)|| || ||9·25||11·25||19·63|
|Number of platelet transfusions episodes|
| Median|| || ||5||3||3|
|No. of days with bleeding (≥Grade 2)|
| Median (IQR)|| || ||1 (0–4)||1 (0–4)||1 (0–4)|
|Mean no. of platelet transfusion episodes/thrombocytopenic day (SD)||9·5 (7·8)||5·3 (3·3)|| || || |
|Total number of platelet donor exposures||1524||1354|| || || |
|Interval between platelet transfusion|
| Mean (SD)||1·8 (1·1)||2·8 (1·8)|| || || |
| Median (IQR)|| || ||1·1 (0·7–2·1)||1·9 (0·9–3·1)||2·9 (1·1–4·7)|
Are prophylactic platelet transfusions necessary?
As North America was trying to resolve the platelet dose issue, our colleagues in Germany and the UK raised the question as to whether prophylactic platelet transfusions were even necessary. Wandt et al. , reported the results of a prospective study of 106 patients receiving 140 autologous peripheral blood stem cell transplants where only therapeutic platelet transfusions were given. Bleeding (the primary outcome) was assessed twice a day and WHO bleeding ≥Grade 2 was considered significant. The results from 60 patients in this prospective cohort were also compared to 60 historical control patients who had received both prophylactic and therapeutic platelet transfusions. The only difference in bleeding was Grade 1 where 20% of patients receiving therapeutic transfusions had bleeding compared to 1% in the prophylactic control arm. Based on these results the authors questioned the need for prophylactic platelet transfusions. Both Germany and the UK have implemented RCTs to determine if prophylactic platelet transfusions are necessary. Both studies are parallel design RCTs and are enroling patients with acute leukaemia receiving stem cell transplantation. Germany is enroling just autologous transplant patients; whereas, the UK is including both allogeneic and autologous transplant patients. The interventions (prophylactic vs. therapeutic only) are similar in the two studies. In the UK study WHO bleeding ≥Grade 2 is the primary outcome; whereas, platelet utilization is the primary outcome in the German study. The UK study recruitment is expected to end in 2011. There have been two abstract reports from the German study showing a 25–27% reduction in the number of platelet transfusions [28,29]. The 2009 abstract indicates that the risk of bleeding was 2·3 times higher in the group receiving therapeutic only platelet transfusions (five minor and two major cerebral bleeds in the therapeutic only treatment arm; none in the group receiving prophylactic platelet transfusions). The two major bleeds resulted in death. The authors felt that the two cerebral fatalities could have occurred in the therapeutic arm just by chance, and concluded that their therapeutic only transfusion strategy was a feasible option.
Is bleeding the right outcome?
The landmark trigger study by Rebulla established bleeding as a feasible outcome in platelet transfusion research . This was a major advance, as previous studies had used surrogate outcomes such as absolute and corrected count increments. However, as more studies were performed using bleeding as the outcome, problems and challenges became apparent. The WHO bleeding scale has never been validated using measurement methodology; hence, reproducibility, accuracy and face validity have not been established. The adjudication experience from SToP also raised concerns about how bleeding was measured and graded . There was disagreement between adjudicators when assigning a Grade as well as other difficulties (when does one bleed stop and another start; when is vaginal bleeding abnormal; and difficulty classifying situations where red cells are not required but multiple platelet transfusions are needed). Bleeding is also a complex outcome as it is a composite (three types of bleeding – Grade 2, 3 and 4) and one of the composite components is a surrogate (Grade 2 bleeding). This complexity presents challenges for the use of bleeding as an outcome measure.
In 2007, Nevo et al.  reported on a retrospective analysis of thrombocytopenia and survival in patients who were transfused based on either a trigger of 10 × 109/l or 20 × 109/l. In patients being transfused at the lower trigger, 69·2% had platelet counts <10 × 109/l and 19% had profound thrombocytopenia (>4 platelet counts ≥10 × 109/l). In contrast, only 38·3% of patients transfused at the 20 × 109/l trigger had platelet counts <10 × 109/l and only 7% had profound thrombocytopenia. In the patients with profound thrombocytopenia, there was an increased rate of early mortality (OR 3·18; 95% CI 1·25–8·07) and reduced overall survival (HR 1·95; 96% CI 1·28–2·97). Although a cause effect relationship can not be established from these data, it does raise the question as to whether we should be looking at mortality as an outcome for future platelet transfusion trials.
The platelet prescription is complex as platelets are unlike other prescription drugs. Most centres have adopted the prophylactic platelet transfusion trigger of 10 × 109/l as routine practice. There are data to suggest that a lower platelet dose does not increase the risk of bleeding but more platelet transfusion episodes will be required if a low-dose strategy is adopted. On the other hand, Germany and the UK have suggested that we may not even need to give prophylactic platelet transfusions. If these studies show that prophylactic platelet transfusions can be abandoned, does this mean that we will need to redo the dose studies for therapeutic platelet transfusions? Should we use bleeding as the outcome measure for these studies knowing the challenges and limitations associated with this outcome? Or, should we start thinking about platelet transfusions as a supportive therapy to prevent severe morbidity and mortality and select outcomes accordingly?
The clerical assistance of Heather Patterson during the preparation of this manuscript is acknowledged.