Professional Documents
Culture Documents
Nadine Shehata, Alan Tinmouth, Gary Naglie, John Freedman, and Kumanan Wilson
T
ransfusion specialists have debated the signifi-
BACKGROUND: The significance of ABO matching for cance of the ABO blood group in platelet (PLT)
platelet (PLT) transfusion has not been clearly defined. transfusion for more than 50 years.1 Although it
The primary objective of this report is to assess is well established that after transfusion of ABO-
whether ABO-identical PLT transfusion is associated incompatible PLTs, anti-A and anti-B can cause incidents
with improved mortality and/or morbidity for patients of intravascular hemolysis, it is not definitively known
with hematologic/oncologic disorders. whether ABO matching of PLT transfusion improves
STUDY DESIGN AND METHODS: A systematic review other outcomes. Providing ABO-identical PLTs can be pro-
to January 2009 was conducted. Data on mortality, blematic because only limited inventories are available.
morbidity, PLT refractoriness, and PLT increment after Also, there is the wastage of PLTs that are not ABO identi-
transfusion were abstracted. cal, and if patients require HLA-matched PLTs for refrac-
RESULTS: A total of 100 citations were identified. Nine- tory thrombocytopenia, these PLTs may not be ABO
teen studies were included in the systematic review. A identical.2 In addition, periodic shortages of PLT products
total of 1502 patients from three randomized controlled
trials and 16 observational studies were included. Sur-
vival, bleeding events, and transfusion reactions were
only considered as secondary outcomes in the reports From the Department of Medicine, University of Toronto, Divi-
reviewed. The PLT count increment was the primary sion of Hematology, St Michael’s Hospital, Canadian Blood Ser-
outcome of several studies and was consistently higher vices, Toronto, Ontario; the University of Ottawa Centre for
with ABO-identical PLT transfusion. The largest differ- Transfusion Research, Clinical Epidemiology Program, Ottawa
ence in increment between ABO-identical and noniden- Health Research Institute, Department of Medicine, the Ottawa
tical PLT transfusion was 4 ¥ 109/L. No consistent Hospital, University of Ottawa, Canadian Blood Services,
benefit in clinical outcomes was noted. Survival was Ottawa, Ontario; the Departments of Medicine and Health
assessed in three reports with conflicting results. Policy, Management and Evaluation, University of Toronto, Divi-
Although two studies described bleeding as an sion of General Internal Medicine, University Health Network,
outcome, the assessment of hemorrhage was consid- Geriatrics Program, Toronto Rehabilitation Institute, Toronto,
ered inadequate. In six studies, ABO-nonidentical PLT Ontario; and the Departments of Pathology, University of
transfusion was not associated with transfusion reac- Toronto, Li Ka Shing Knowledge Institute, Toronto, Ontario,
tions, and the results from four studies addressing the Canada.
impact of ABO-identical PLT transfusion on PLT and red Address reprint requests to: Nadine Shehata, MD, Queen
blood cell utilization were conflicting. Wing, 2-065c, St Michael’s Hospital, 30 Bond Street, Toronto,
CONCLUSION: ABO-identical PLT transfusion results ON M5B 1W8, Canada; e-mail:shehatan@smh.toronto.on.ca.
in a higher PLT increment. Randomized controlled trials GN is supported by the Mary Trimmer Chair in Geriatric
are required to definitely determine the effect of ABO- Medicine Research, University of Toronto. KW is supported by
identical PLT transfusion on survival, bleeding events, the Canada Research Chair in public health policy.
or transfusion reactions. Each author contributed to the design and conduct of the
work, NS conducted the systematic review and prepared the
manuscript, NS and AT reviewed the citations and validated
data abstraction, and all authors have contributed to the writing
and revision of the manuscript.
Received for publication September 26, 2008; revision
received April 14, 2009, and accepted April 17, 2009.
doi: 10.1111/j.1537-2995.2009.02273.x
TRANSFUSION 2009;49:2442-2453.
put additional strain on the blood service and the blood The data were abstracted by one reviewer (NS) and
transfusion specialists when providing ABO-matched verified by a second author (AT). We abstracted study
products. A recent survey of 3152 North American labora- characteristics using standardized data abstraction forms.
tories found that 17% of transfusion medicine laboratories We abstracted information on the following study
do not have a policy regarding the use of ABO- characteristics: the year of study, country where the study
nonidentical PLTs.2 Thirty percent of institutions that had took place, whether the study was single or multicentered,
a policy provided ABO-incompatible PLT or plasma prod- setting, characteristics of participants, inclusion/
ucts if there was an urgent need for PLT products, and 50% exclusion criteria, study design, sampling methods,
indicated that ABO-incompatible PLTs or plasma would sample size, the dose of PLTs transfused, the PLT product
be provided if ABO-compatible products are not avail- transfused (e.g., random donor, single donor), the use of
able.2 Therefore, determining the clinical value of trans- leukoreduction, the duration of PLT storage, the method
fusing ABO-identical PLTs is essential. to determine the PLT increment, the time to determine the
We conducted a systematic review of the literature PLT increment after transfusion, the transfusion criteria,
to determine whether the transfusion of ABO-identical and the definition of PLT refractoriness. We evaluated the
PLTs improves outcomes in patients with hematologic/ quality of the randomized controlled trials by examining
oncologic disorders, including patients undergoing the randomization technique, the mechanism for con-
hematopoietic stem cell and bone marrow transplantation cealment of allocation, whether the study was blinded,
(BMT), compared to ABO-nonidentical (i.e., plasma- or whether the groups were comparable at baseline, the
PLT-incompatible) PLT transfusion.The primary outcomes primary outcomes assessed, the use of an intention-to-
that we considered were mortality and morbidity (i.e., treat analysis, and the extent of losses to follow-up. The
bleeding and transfusion reactions). We considered PLT quality of the observational studies was evaluated by
refractoriness, the posttransfusion increase in PLT count, examining the method of data collection, the sampling
and changes in PLT utilization as secondary outcomes. methods, the outcomes assessed, and the statistical
analysis methods.
MATERIALS AND METHODS We also extracted information on the PLT transfusion
criteria, survival, hemorrhage, transfusion reactions, the
Identification and selection of studies frequency of PLT refractoriness, PLT and red blood cell
We conducted a systematic search of the MEDLINE (1950 (RBC) utilization, and the PLT increments associated with
to Week 2 November 2007) and HEALTHSTAR (1966 to ABO-identical and ABO-nonidentical transfusion. We
November 2007) databases. The search was updated to did not conduct a meta-analysis since the study designs
January 2009. We used the following medical subject and reporting of outcomes were too heterogeneous to
heading or text words to search the databases: “ABO combine.
blood-group system,” “blood grouping and crossmatch-
ing,” “ABO matching,” “ABO identical,” “ABO unmatched,”
“blood group antigens,” “platelet transfusion,” “platelet Definitions for ABO-matched transfusions
refractoriness,” “platelet count,” and “platelet increment.”
Because there were various definitions for ABO matched
The full search strategy is provided in the Appendix. We
transfusions, we defined 1) ABO-identical PLT transfusion
also conducted a bibliographic search to ensure that all
as both the donor and the recipient being the same ABO
relevant publications were included.
type, 2) ABO-compatible PLT transfusion when the donor
We included a study in the systematic review if it met
PLTs do not carry A or B antigens that are incompatible
the following criteria: 1) it was published in English; 2) it
with the recipient’s plasma (i.e., PLTs from an O donor to
was an original report; and 3) one of the primary out-
an A, B, or AB recipient or PLTs from an A or B donor to an
comes was morbidity, mortality, refractoriness, or PLT
AB recipient), 3) ABO-incompatible PLT transfusion (i.e.,
increment. We excluded studies based on the following
all other PLT transfusions), and 4) ABO-nonidentical
criteria: 1) they were case reports, letters, or abstracts or 2)
PLT transfusion to reflect ABO-compatible and ABO-
the objective was not to analyze ABO-incompatible
incompatible PLT transfusion. If there was insufficient
transfusion.
information to classify the PLT transfusion, the classifica-
tion used by the investigators was used.
Data extraction and analysis
Two reviewers (NS, AT) independently assessed the
MEDLINE and HEALTHSTAR abstracts for inclusion/ RESULTS
exclusion criteria. When a discrepancy occurred between
the reviewers, a third reviewer (KW) assessed the abstract Results of the literature search
and/or the full publication was retrieved and a decision We retrieved a total of 77 citations from MEDLINE and 23
was arrived at by consensus. from HEALTHSTAR. Twenty-three studies fulfilled the
coagulation disorders
additional studies were identified when the search was
updated.8,9 Twenty studies were included in this system-
atic review.
<20 ¥ 109/mL
Not reported
‡ Patients did not have high fever (>38.3°C), splenomegaly, coagulation abnormalities, or active bleeding. Patients with AB blood group were excluded.
Characteristics of the studies
Randomized controlled trials
PLT product
One study indicated that recipients of PLT transfusions
PRD†
PRD§
PRD
were paired randomly; however, we did not classify this
study as a randomized controlled trial because recipients
Method to determine
divided apheresis units.10 Thus, three of the studies
PLT increment
included in our final analysis were randomized controlled
* CCI = posttransfusion - pretransfusion PLT count (¥109)/L ¥ body surface area (m2)/number of PLTs infused (¥1011).
CCI*
CCI
CCI
trials (Table 1).11-14 Two publications described the same
study, so for the purposes of this systematic review they
were considered as one study.11,12 One of the randomized
controlled trials used a crossover design.14 Two trials were
conducted in the United States and one in the United
ASCT = autologous stem cell transplant; PRD = pooled random-donor PLT product.
Acute leukemia‡
study, leukoreduced PLT products were only used after
febrile reactions.11 The PLT increment was measured using
the corrected count increment (CCI),11,13,14 and PLT trans-
fusion triggers, where reported, were below 20 ¥ 109/L.11,13
The primary outcomes as defined by investigators were
the PLT increment11,14 or efficacy and consequence of
Center status
Single center
Single center
Single center
United States
Lee, 198914
PLTs; one report did not describe the type of PLT product
transfused.16 The patient populations were heteroge-
neous, including individuals with hematologic malignan-
cies,8,9,15,17,18,23,24,26,27 solid tumors,8,9,17,19,23,26 idiopathic
TRANSFUSION 2445
SHEHATA ET AL.
Primary outcome
individuals who were refractory to PLT transfusions;23-25
PLT transfusions
PLT refractoriness
healthy or convalescent medical patients;16 and hemato-
PLT increment
logic patients without fever, sepsis, or splenomegaly.10,24
Two studies excluded patients who did not have post-
transfusion PLT counts.20,26 One study excluded 15% of
PLT transfusions because of incomplete data.9 PLT
refractoriness was described as an outcome in seven
1 PRD/10 kg
studies, with various definitions for refractoriness used
PLT dose
(Table 2).17,18,20,23-26 The criteria for PLT transfusion was
3 ¥ 1011
8 PRD
generally a PLT count of less than 20 ¥ 109/L.8-10,15,18-22,27
The reported primary outcomes by the investigators
were the PLT increment using PLTs from the same apher-
follow-up
esis donor,10 the effect of PLT transfusion on survival,27 the
Loss of
Yes†
NR
NR
effect of ABO and other factors on PLT increment,9,15-18,25
the incidence of hepatic venoocclusive disease associated
with ABO-identical PLT transfusion,19 the effect of factors
Intent to treat
* Sixty percent of women in the ABO-identical group and four or more pregnancies compared to 10% in the incompatible group.
on PLT transfusion refractoriness,20 the impact of leukore-
yes or no
analysis,
TABLE 3. The quality of randomized controlled trials
duced ABO-identical PLT transfusion on morbidity and
No
No
No
† Seventy-five transfusions were excluded from the analysis, 10 patients were excluded from the analysis of refractoriness.
mortality,21 and the effect of ABO compatibility.8,22-24 One
report did not indicate a specific primary outcome.26
Survival, bleeding events, and PLT refractoriness were
yes or no
Blinding,
not reported in any of the prospective studies. The
No
No
No
number of days with hemorrhage was reported in one
NA = not applicable; NR = not reported; PRD = pooled random donor; RCT = randomized controlled trial.
retrospective report,21 and the frequency of hemolysis in
another observational study.22
Concealment of
allocation
Yes
No
No
Quality of studies
Randomized controlled trials
Of the three randomized controlled trials, two described
the randomization technique,11,13 only one report indi-
Comparable
(Table 3).11,13
Randomization technique
Outcomes
Carr, 199013
Lee, 198914
TRANSFUSION 2447
SHEHATA ET AL.
nor transfusion reactions was a primary endpoint in any sion of RBCs and PLTs, and length of stay. In recipients of
study. A post hoc analysis of one study demonstrated that autologous marrow transplants for lymphoma, there was
in a subgroup of patients with acute leukemia who no difference in the number of days with hemorrhage (0.2
achieved a complete remission and received ABO- days with leukoreduced predominantly ABO identical PLT
identical PLT transfusion, survival was prolonged by 12 transfusion and 0.3 days for nonleukoreduced predomi-
months.12 One study reported that there were no transfu- nantly ABO-unmatched PLT transfusions; p = not signifi-
sion reactions after ABO-incompatible PLT transfusions cant). The same investigators also did not find a difference
and two patients receiving ABO-incompatible PLT trans- in mortality (defined as mean survival months at last
fusion experienced bleeding requiring transfusion sup- follow-up/death) during induction therapy in 115 patients
port.13 In two trials, the frequency of PLT refractoriness in with acute leukemia who received ABO-identical PLT
patients with various hematologic malignancies was transfusion compared to ABO-unmatched transfusion,
reduced by 39 and 61% with ABO-identical PLT transfu- but a difference in survival was reported to be improved
sions,11,13 and a reduction in the number of transfusions for patients receiving washed leukoreduced ABO-
was also demonstrated with ABO-identical PLT transfu- identical PLT transfusion compared to ABO-unmatched
sion.11 ABO-identical PLT transfusions were also associ- transfusion (p = 0.02).27 In pediatric patients undergoing
ated with an increase of 6.6 ¥ 109 to 14.9 ¥ 109/L in the CCI autologous hematopoietic stem cell transplantation,
compared to nonidentical PLTs.11,14 In one of these studies, ABO-identical PLT transfusion was associated with a 17%
a significant PLT increment with ABO-identical PLT trans- reduction in the incidence of hepatic venoocclusive
fusion was only found after the second PLT transfusion disease compared to ABO-compatible transfusion
and not the first transfusion.14 The second study reported (p = 0.02).19 Of the seven studies that included the PLT
as a secondary outcome that the type of ABO matching increment as an outcome,17,18,20,23-26 four reported a signifi-
affected the PLT increment.11 ABO-identical PLT transfu- cantly higher increment with ABO-identical/-compatible
sions were associated with higher CCIs than ABO- PLT transfusions compared with non–ABO-identical
compatible or ABO-incompatible transfusions (7500 vs. transfusions,18,20,23,24 two reported an increase with ABO-
4800 vs. 3000, respectively; p < 0.05).11 The last study did compatible PLT transfusion compared to ABO-
not demonstrate a difference in PLT increment with ABO- incompatible transfusion but did not provide data for the
identical PLT transfusion.13 increment with ABO-nonidentical transfusion,17,18 one
reported an increase with ABO-compatible compared to
Prospective and retrospective studies ABO-incompatible PLT transfusion that was not signifi-
Prospective studies. A 45% difference in PLT incre- cant,25 and one study did not demonstrate an increased
ments with ABO-compatible versus ABO-incompatible PLT increment with ABO-identical PLT transfusion
PLT transfusion was found in one report, but significance (Table 6).26 Only three studies compared outcomes with
was not described.16 Another study demonstrated a ABO-identical versus ABO-compatible PLT transfusion;
19,22,23
significant difference in the PLT increment with ABO- two of the three demonstrated improved outcomes
identical compared to ABO-compatible PLT transfusion.10 with ABO-identical PLT transfusion.19,23
Jimenez and colleagues10 randomly assigned recipients to In 47 patients with lymphoma undergoing autolo-
receive PLTs from a common donor. There was 30% differ- gous BMT, leukoreduced ABO-identical PLT transfusion
ence in PLT recovery between ABO-identical and ABO- was found to be associated with fewer PLT transfusions
compatible PLT transfusion when the data were analyzed (mean, 74 for leukoreduced predominantly ABO-identical
according to whether the recipients received PLTs from a PLT transfusion versus 151 for nonleukoreduced predomi-
common donor.10 One study reported a lower percentage nantly ABO-unmatched PLT transfusions; p = 0.003) and
of PLT recovery with ABO-incompatible transfusions,8 fewer RBC transfusions (mean, 9 vs. 16, respectively;
and another study reported a lower CCI with ABO- p = 0.02).21 In a second study of 81 allogeneic marrow
mismatched transfusions.9 The remaining study did not recipients, there was no significant difference between
find a difference in PLT increment with ABO-identical or ABO-identical versus ABO-compatible PLTs in mean
ABO-compatible PLT transfusion.15 number of PLT transfusions (16.3 ⫾ 20.7 for ABO-identical
Retrospective studies. A total of 1214 individuals vs. 15.1 ⫾ 16.6 with ABO-compatible) and mean number
were included in the 11 retrospective studies (Table 6). In of RBC units transfused (5.3 ⫾ 5.5 for ABO-identical com-
one report of 47 patients, no difference in survival after pared to 4.6 ⫾ 3.7 ABO-compatible PLT transfusion).22 In
3 years was observed with leukoreduced ABO-identical patients with acute leukemia, there was also no difference
PLT transfusion compared to nonleukoreduced ABO- in mean number of PLT transfusions (137 ⫾ 100 for ABO-
mismatched transfusion, but the study was underpow- unmatched compared to 125 ⫾ 124 for ABO-identical)
ered to detect potentially meaningful differences.21 The and mean number of RBC units transfused (17 ⫾ 12 for
adverse effects or morbidities described included days ABO-unmatched compared to 16 ⫾ 8 for ABO-identical).27
with hemorrhage, hepatic venoocclusive disease, transfu- Hemolysis did not appear to occur with ABO-compatible
PLT transfusion.22 There was no difference found in PLT studies were small so they were likely not adequately
refractoriness associated with ABO-identical versus ABO- powered to detect a clinically significant difference in
nonidentical PLT transfusion.18 transfusion reactions.
There were conflicting results for the effect of ABO-
identical transfusion on PLT and RBC transfusion and PLT
DISCUSSION
refractoriness. ABO-identical PLT transfusion was found
In this first systematic review to assess whether ABO to be associated with a reduction on RBC and PLT trans-
matching of PLTs improves morbidity and mortality in fusion in two11,21 of four studies.22,27 The difference in
hematology/oncology patients, we found that ABO- results is likely secondary to confounding variables,
identical PLT transfusion was associated with a higher PLT because the studies were small. Two randomized con-
increment. However, there was inconclusive data avail- trolled trials found a reduction in PLT refractoriness
able to determine whether ABO matching affects mortal- with ABO-identical PLT transfusion11,13 but a retrospective
ity (aside from the rare events of hemolytic transfusion report did not find a difference in the frequency of PLT
reactions), bleeding events, or transfusion reactions as refractoriness with ABO-compatible PLT transfusion. The
none of the studies used these measures as primary out- conflicting results on the effect of PLT refractoriness may
comes and none was powered to detect a difference in have been due to a difference in definitions of PLT refrac-
these outcomes. toriness or a difference in the ABO-matched product used
When addressing the effect of ABO-identical PLT for the analysis, that is, the two randomized controlled
transfusion, it is important to consider a hierarchy of trials11,13 compared ABO-identical PLT transfusion to
transfusion outcomes. Mortality and morbidity, that is, nonidentical transfusion11 and incompatible PLT transfu-
hemorrhage and transfusion reactions, are the most sion,13 whereas the retrospective analysis analyzed the
important clinical outcomes. All other outcomes are sec- effect of ABO identical versus compatible transfusions.18
ondary outcomes, for example, PLT refractoriness, and Thus, ABO-identical PLT transfusion appears to reduce
length of stay or resource utilization outcomes, that is, the frequency of PLT refractoriness compared to ABO-
utilization of RBCs and PLTs. The PLT increment can also nonidentical PLT transfusion, but it has yet to be deter-
be considered a secondary outcome, although it is not a mined whether ABO-identical PLT transfusion reduces
clinical outcome, as a higher PLT recovery may be a sur- the frequency of PLT refractoriness compared to ABO-
rogate marker for the reduced risk of bleeding. compatible PLT transfusion.
Survival was demonstrated to be improved in a post There were two prospective8,10 and three retrospec-
hoc analysis of one randomized controlled trial of a sub- tive cohort studies19,22,23 that compared ABO-identical
group of individuals with acute leukemia that received and ABO-compatible PLT transfusion. Two of the five
ABO-identical PLT transfusion, but this result was not studies reported a higher increment in PLTs with ABO-
substantiated in two observational reports by the same identical PLT transfusion.10,23 One study did not define
authors of individuals receiving autologous or peripheral ABO compatibility clearly,9 and the remaining two
stem cell transplantation21 and in patients with acute studies did not report on this outcome.19,22 There was an
leukemia.27 The latter report, however, indicated that increased frequency with hepatic venoocclusive disease
survival was improved for patients receiving washed with ABO-compatible PLT transfusion compared to ABO-
leukoreduced ABO-identical PLTs compared to identical PLT transfusion found in one study19 and the
unmatched PLTs. Two retrospective reports of ABO- significance of this outcome needs to be confirmed
identical PLT transfusion in patients undergoing cardio- because this result has not been reported by other
vascular surgery28,29 and one retrospective report in investigators.
patients receiving ABO-incompatible plasma suggest that It is important to consider why outcomes are not con-
there may be an increased risk of mortality with ABO- sistent. Causality for mortality and morbidity is subject to
nonidentical PLT or large-volume plasma transfusion confounding variables in observational studies and there
(relative risk, 1.15; 95% confidence interval [CI], 1.02- are abundant confounding variables associated with the
1.29).30 Based on these data, it is difficult to conclude outcomes of a PLT transfusion. The effect of PLT transfu-
what the effect of ABO matching is on survival in sion reflects recipient factors (e.g., immunosuppressive
hematology/oncology patients. drugs, infections, sepsis) and donor factors such as the
Of the two studies13,21 that reported bleeding events, variability of expression of AB antigens31-33 as well as PLT
the definition of hemorrhage and bleeding was not clearly factors such as the PLT count and quality of the PLT trans-
defined. Carr and colleagues13 reported that there were no fusion product. There may also be a cumulative effect on
transfusion reactions attributable to ABO-incompatible the PLT increment with ABO-identical PLT transfusion,
PLT transfusions, and Shanwell and colleagues22 indicated which may not be detected if the PLT increment is only
that there was no evidence of hemolysis with ABO- assessed after the first or second transfusion.34 Finally,
compatible PLT transfusions, but the sample sizes of the there is a lack of discrimination between different types of
ABO incompatibility in published reports (i.e., plasma or identical transfusion in hematology/oncology patients.
PLT incompatibility).11 However, the results of this review must be interpreted
The PLT increment may be used as a surrogate in the context of its limitations. The studies that were
outcome for bleeding. However, this assumes that there is included were heterogeneous with respect to the patient
a correlation between the PLT increment and the risk of populations, the PLT products used, the definitions of
bleeding. Although it is intuitive that an increase in PLT incompatibility, and the reported outcomes considered.
number may reduce bleeding events and potentially mor- The sample sizes were small and there were no studies
tality, the PLT increment required to reduce bleeding risk that were powered to detect a difference in mortality or in
has not been determined. Preliminary results from a large adverse events (i.e., bleeding or transfusion reactions).
randomized controlled trial comparing different prophy- Although one report did find a significant difference in
lactic PLT doses do not show any differences in bleeding mortality with ABO-identical PLT transfusion,12 the
associated with increased PLT count increments.35 improved mortality was described only in a post hoc
However, one small retrospective analysis showed that analysis. In addition, a large proportion of studies were
PLT refractoriness, as defined by a 1-hour CCI below 7.5 uncontrolled, observational studies that are susceptible to
and/or a 24-hour CCI below 4.5, was associated with confounding in that not all known risk factors for transfu-
increased bleeding complications (odds ratio [OR], 3.4; sion are included in the analysis. Furthermore, few studies
95% CI, 1.1-11) and decreased 100-day survival (98% for assessed the effect of ABO-identical PLT transfusion com-
individuals not experiencing PLT transfusion refractori- pared to ABO-compatible transfusion,10,19,22,23 and three
ness compared to 83% with a low CCI after 24 hours, demonstrated superior outcomes with ABO-identical
p < 0.01).36 The level of thrombocytopenia, however, was PLTs.10,19,23 Finally, the quality of the data in these studies
not associated with an increased bleeding risk.36 These may be impacted by insufficient documentation, which is
results need to be substantiated. Higher PLT count incre- a common limitation of retrospective studies.38
ments may also be used as a surrogate for reduced PLT Although the studies included in this systematic
transfusion requirements. However, a reduction in PLT review had several limitations, we conclude that the pre-
requirements was only found in one21 of the two studies ponderance of evidence suggests that ABO-identical PLT
reporting this outcome. Therefore, the greater PLT incre- transfusion is associated with a higher PLT increment.
ments seen with ABO-compatible or -identical transfu- However, the clinical significance of the greater PLT count
sions may not result in meaningful differences in the increments is not clear. Additionally, this systematic
number of PLT transfusions or the number of PLT units review cannot make any conclusions regarding the effects
transfused. This may be due to the small differences in the of ABO on mortality or morbidity. As there is an increased
increments or due to the small sample sizes of the studies. risk of hemolytic transfusion reactions with ABO-
Since the clinical significance of an increment is not nonidentical PLT transfusion, the potential for increased
known, mortality and serious bleeding events are the pre- adverse events with transfusing ABO-nonidentical PLTs
ferred evaluation outcomes.37 should not be downplayed. Transfusion medicine services
There are other surrogate measures that have been should consider measures to increase ABO-identical PLT
used to assess ABO-identical PLT transfusion. The Hb transfusions and physicians should be aware of potential
concentration has been used as a surrogate marker for adverse outcomes when transfusing ABO-nonidentical
hemolytic transfusion reactions.6 In one small study of 16 PLTs. A large randomized controlled trial is needed to
patients with hematologic and oncologic malignancies determine the degree of benefit associated with ABO-
who received ABO-identical or -nonidentical PLT transfu- identical PLT transfusions compared to ABO-compatible
sion, transfusion of ABO-identical PLT transfusion did not PLT transfusions, because supplying ABO-identical PLTs
result in a significant change in Hb concentration.6 The to all patients has significant resource implications given
significance of the use of the Hb concentration as a surro- that there is currently a limited supply of these products
gate marker for hemolytic transfusion reactions remains and demand may not match the need for these products.
to be determined. A large randomized controlled trial powered to detect a
Although there have been a number of studies difference in morbidity (i.e., transfusion reactions and
addressing the effect of ABO PLT matching, the majority of bleeding events) and/or mortality, and that includes
the studies have been observational and are susceptible to standardized outcomes such as the degree of bleeding
important sources of bias. This systematic review, thus, and PLT refractoriness, would more clearly establish
has several strengths including the use of a predefined the potential benefit of ABO-identical PLT transfusion.
systematic search strategy, a duplicate review of all cita- In addition, future publications should clearly specify
tions, and structured data abstraction. These approaches whether PLT transfusions are ABO-compatible or ABO-
limit bias on the part of the reviewers. In addition, by identical and should try to establish differences in
reviewing all the studies, we were able to determine outcomes between ABO-identical and -compatible
whether there is a tendency to improved effects with ABO- transfusions.
transplantation in young children. Transplantation 2005; 29. Lin Y, Callum JL, Coovadia A. Reply. ABO-mismatched
80:314-9. platelet transfusions and clinical outcomes after cardiac
20. Klumpp TR, Herman JH, Innis S, Pcarlman E, Culling N, surgery. Transfusion 2002;42:1528-9.
Kotz KW, Slachta C, Goldberg SL, Mangan KF. Factors 30. Shanwell A, Andersson TM, Rostgaard K, Edgren G, Hjal-
associated with response to platelet transfusion following grim H, Norda R, Melbye M, Nyrén O, Reilly M. Post-
hematopoietic stem cell transplantation. Bone Marrow transfusion mortality among recipients of ABO-compatible
Transplant 1996;17:1035-41. but non-identical plasma. Vox Sang 2009;96:316-23.
21. Heal JM, Blumberg N, Kirkley SA, DiPersio JF, Rapoport AP, 31. Ogasawara K, Ueki J, Takenaka M, Furihata K. Study of the
Rowe JM. Leukocyte-reduced transfusions of ABO-identical expression of ABH antigens on platelets. Blood 1993;82:
platelets and clinical outcome in autologous bone marrow 993-9.
transplantation for lymphoma. Bone Marrow Transplant 32. Curtis BR, Edward JT, Hessner MJ, Klein JP, Aster RH.
1994;14:943-8. Blood group A and B antigens are strongly expressed on
22. Shanwell A, Ringdén O, Wiechel B, Rumin S, Akerblom O. platelets of some individuals. Blood 2000;96:1574-81.
A study of the effect of ABO incompatible plasma in plate- 33. Cooling LL, Kelly K, Barton J, Hwang D, Koerner TA, Olson
let concentrates transfused to bone marrow transplant JD. Determinants of ABH expression on human blood
recipients. Vox Sang 1991;60:23-7. platelets. Blood 2005;105:3356-64.
23. Heal JM, Blumberg N, Masel D. An evaluation of cross- 34. Heal JM, Rowe JM, Blumberg N. ABO and platelet transfu-
matching, HLA, and ABO matching for platelet transfu- sion revisited. Ann Hematol 1993;66:309-14.
sions to refractory patients. Blood 1987;70:23-30. 35. Slichter SJ, Kauman RM, Assmann SF, Brecher ME, Gern-
24. Duquesnoy RJ, Anderson AJ, Tomasulo PA, Aster RH. ABO sheimer T, Hillyer CD, McCullough J, Strauss RG, Triulzi D.
compatibility and platelet transfusions of alloimmunized Effects of prophylactic platelet (Plt) dose on transfusion
thrombocytopenic patients. Blood 1979;54:595-9. (Tx) outcomes (PLADO trial) [abstract]. Blood 2008;112:
25. Tosato G, Applebaum FR, Deisserot AR. HLA-matched 285.
platelet transfusion therapy of severe aplastic anemia. 36. Kerkhoffs JL, Eikenboom JC, van de Watering LM, van
Blood 1978;52:846-54. Wordragen-Vlaswinkel RJ, Wijermans PW, Brand A. The
26. Van Eys J, Thomas D, Olivos B. Platelet use in pediatric clinical impact of platelet refractoriness: correlation with
oncology: a review of 393 transfusions. Transfusion 1978; bleeding and survival. Transfusion 2008;48:1959-65.
18:169-73. 37. Norfolk DR, Ancliffe PJ, Contreras M, Hunt BJ, Machin SJ,
27. Blumberg N, Heal JM, Liesveld JL, Phillips GL, Rowe JM. Murphy WG, Williamson LM. Consensus conference on
Platelet transfusion and survival in adults with acute leu- platelet transfusion, Royal College of Physicians of Edin-
kemia. Leukemia 2008;22:631-5. burgh 27-28 November 1997. Br J Hematol 1998;101:609-17.
28. Blumberg N, Heal JM, Hicks GL Jr, Risher WH. Association 38. Audet AM, Goodnough LT, Parvin CA. Evaluating the
of ABO-mismatched platelet transfusions with morbidity appropriateness of red blood cell transfusions: the limita-
and mortality in cardiac surgery. Transfusion 2001;41: tions of retrospective medical record reviews. Int J Qual
790-3. Health Care 1996;8:41-9.