Curley 2018 Platelet Transfusion Threshols in Neonates

The n e w e ng l a n d j o u r na l of m e dic i n e
Original Article
Randomized Trial of Platelet-Transfusion

Thresholds in Neonates
Anna Curley, M.D., Simon J. Stanworth, F.R.C.P., D.Phil., Karen Willoughby, B.Sc.,
Susanna F. Fustolo‑Gunnink, M.D., Vidheya Venkatesh, M.D., Cara Hudson, M.Sc.,
Alison Deary, M.Sc., Renate Hodge, M.Sc., Valerie Hopkins, B.Sc.,
Beatriz Lopez Santamaria, M.Sc., Ana Mora, Ph.D., Charlotte Llewelyn, Ph.D.,
Angela D’Amore, M.D., Rizwan Khan, M.R.C.P.I., Wes Onland, M.D., Ph.D.,
Enrico Lopriore, M.D., Ph.D., Karin Fijnvandraat, M.D., Ph.D.,
Helen New, F.R.C.Path., Ph.D., Paul Clarke, M.D., and Timothy Watts, M.D.,
for the PlaNeT2 MATISSE Collaborators*
A BS T R AC T
BACKGROUND
Platelet transfusions are commonly used to prevent bleeding in preterm infants with The authors’ affiliations are listed in the
thrombocytopenia. Data are lacking to provide guidance regarding thresholds for Appendix. Address reprint requests to
Dr. Curley at the Department of Neona‑
prophylactic platelet transfusions in preterm neonates with severe thrombocytopenia. tology, National Maternity Hospital, Hol‑
les St., Dublin 2, Ireland, or at acurley@
METHODS nmh.ie.
In this multicenter trial, we randomly assigned infants born at less than 34 weeks *A complete list of the PlaNeT2 MATISSE
of gestation in whom severe thrombocytopenia developed to receive a platelet trans- investigators is provided in the Supple‑
fusion at platelet-count thresholds of 50,000 per cubic millimeter (high-threshold mentary Appendix, available at NEJM.org.
group) or 25,000 per cubic millimeter (low-threshold group). Bleeding was docu- Drs. Curley and Stanworth contributed
mented prospectively with the use of a validated bleeding-assessment tool. The pri- equally to this article.
mary outcome was death or new major bleeding within 28 days after randomization. This article was published on November 2,
2018, at NEJM.org.
RESULTS
DOI: 10.1056/NEJMoa1807320
A total of 660 infants (median birth weight, 740 g; and median gestational age, Copyright © 2018 Massachusetts Medical Society.
26.6 weeks) underwent randomization. In the high-threshold group, 90% of the
infants (296 of 328 infants) received at least one platelet transfusion, as compared
with 53% (177 of 331 infants) in the low-threshold group. A new major bleeding
episode or death occurred in 26% of the infants (85 of 324) in the high-threshold
group and in 19% (61 of 329) in the low-threshold group (odds ratio, 1.57; 95%
confidence interval [CI], 1.06 to 2.32; P = 0.02). There was no significant difference
between the groups with respect to rates of serious adverse events (25% in the high-
threshold group and 22% in the low-threshold group; odds ratio, 1.14; 95% CI,
0.78 to 1.67).
CONCLUSIONS
Among preterm infants with severe thrombocytopenia, those randomly assigned
to receive platelet transfusions at a platelet-count threshold of less than 50,000 per
cubic millimeter had a significantly higher rate of death or major bleeding within
28 days after randomization than those in the group that received less than 25,000
per cubic millimeter. (Funded by the National Health Service Blood and Transplant
Research and Development Committee and others; Current Controlled Trials num-
ber, ISRCTN87736839.)
n engl j med nejm.org 1
The New England Journal of Medicine
Downloaded from nejm.org on November 2, 2018. For personal use only. No other uses without permission.
Copyright © 2018 Massachusetts Medical Society. All rights reserved.
P
rophylactic platelet transfusions rhage, immune thrombocytopenia, no adminis-
are commonly administered to preterm in- tration of parenteral vitamin K, or a low proba-
fants to reduce the risk of bleeding.1-5 Poli- bility of survival beyond several hours. Preterm
cies regarding neonatal platelet transfusion vary infants with major bleeding became eligible for
widely among clinicians and institutions, reflect- randomization 72 hours later provided there was
ing the broad nature of consensus-based guideline no further major bleeding.
recommendations.5-8 A randomized trial compar-
ing prophylactic platelet-transfusion thresholds in Intervention
neonates showed no benefit of maintaining a Infants were randomly assigned to receive a plate-
“normal” platelet count (150,000 per cubic milli- let transfusion (at a dose of 15 ml per kilogram of
meter) to prevent intraventricular hemorrhage in body weight) when the platelet count was less
152 preterm infants.9 It excluded infants with se- than 25,000 per cubic millimeter (the low-thresh-
vere thrombocytopenia (defined as a platelet count old group) or less than 50,000 per cubic millimeter
of <50,000 platelets per cubic millimeter). Most (the high-threshold group). One Dutch trial site
current clinical decisions to transfuse platelets administered platelet hyperconcentrates in a dosage
are made for infants with platelet counts below of 10,000 to 20,000 × 106 platelets per kilogram.
this threshold.5,10,11 More restrictive transfusion Platelet products conformed to national standards
thresholds have become common, despite the lack in the United Kingdom, Ireland, and the Nether-
of outcome data.8 In one study, 42% of transfu- lands. The protocol (available with the full text
sions were administered in patients with platelet of this article at NEJM.org) permitted additional
counts of less than 25,000 per cubic millimeter, platelet transfusions for clinically significant bleed-
and 92% were administered in those with counts ing or surgery or invasive procedures. Treating
of less than 50,000 per cubic millimeter.6 clinicians and parents and guardians were aware
Data are lacking from randomized trials to of the treatment assignments, but neonatologists
compare clinically relevant outcomes associated adjudicating the outcomes were unaware of these
with currently used platelet count thresholds in assignments.
preterm infants with thrombocytopenia. We per-
formed a multicenter, randomized trial to assess Outcomes
whether a policy of prophylactic platelet transfu- The primary outcome was a composite of death or
sion in preterm infants, based on a platelet-count major bleeding up to and including day 28. Pre-
threshold of less than 50,000 per cubic millimeter, specified secondary outcomes were the following:
as compared with less than 25,000 per cubic survival up to day 28 after a major bleeding epi-
millimeter, reduced the risk of death and major sode, death up to day 28, the rate and time from
bleeding. randomization to major bleeding up to day 28, at
least one minor bleeding episode up to day 14, at
least one moderate bleeding episode up to day 14,
Me thods
a major bleeding episode after red-cell transfu-
Eligibility Criteria sion, chronic lung disease (dependency on oxy-
Parents and guardians of infants who had a gen or respiratory support at >36 weeks of post-
platelet count of less than 100,000 per cubic milli- menstrual age) up to the end of the trial, stage
meter were identified. Randomization occurred 2 retinopathy of prematurity (unilateral or bilat-
after written informed consent was received if the eral) up to 38 weeks of corrected gestational age,
infant was receiving care in a participating neo- retinopathy of prematurity leading to laser or
natal intensive care unit and the following criteria bevacizumab therapy up to 38 weeks of corrected
were met: a gestational age at birth of less than gestational age, discharge by 38 weeks of corrected
34 weeks, a platelet count of less than 50,000 per gestational age, the number of platelet-transfu-
cubic millimeter, and cranial ultrasonography per- sion episodes per participant up to day 28, receipt
formed within 6 hours before randomization that of at least one platelet transfusion, the median
did not show a major intraventricular hemorrhage. platelet-count nadir before a major bleeding epi-
Exclusion criteria were a major or life-threatening sode, the median platelet count closest to that of
congenital malformation, major bleeding within a major bleeding episode, a new sepsis event up to
the previous 72 hours, fetal intracranial hemor- end of the trial, a new necrotizing enterocolitis
2 n engl j med nejm.org

Platelet-Tr ansfusion Thresholds in Neonates
event up to the end of the trial, a serious adverse Oversight

event up to the end of the trial, and platelet trans- This trial was undertaken in accordance with the
fusion–related adverse events up to the end of the Declaration of Helsinki and Good Clinical Prac-
trial. An additional secondary outcome was the tice guidelines. The protocol was approved by inde-
neurodevelopmental outcome at 2 years; these pendent ethics committees in the United Kingdom,
data were not available as of this writing. the Netherlands, and Ireland. An independent data
A bleeding-assessment tool was designed and and safety monitoring committee reviewed the in-
validated for use in this trial.12 Grading of bleed- terim data analysis and monitored patient safety
ing as minor, moderate, major, or severe was as- in 6 monthly data review sessions. The first two
signed centrally with the use of a computer algo- authors take responsibility for the accuracy and
rithm and based on a modified version of the completeness of data and the fidelity of the trial
World Health Organization grading system used in to the protocol. The statistical analysis plan is
other platelet-transfusion trials12-14 (see the Sup- available with the protocol at NEJM.org.
plementary Appendix, available at NEJM.org). The
outcome of “major bleeding” included intracra- Statistical Analysis
nial hemorrhage (leading to neurosurgical inter- Data on the frequency of bleeding outcomes in
vention or radiologic imaging showing midline infants with severe thrombocytopenia were avail-
shift), intraventricular hemorrhage filling 50% or able from the observational Platelets for Neonatal
more of the cerebral ventricle, pulmonary hemor- Transfusion–Study 1 (PlaNeT-1),6 in which 30 of
rhage (fresh bleeding through an endotracheal 169 infants (18%) died or had a major bleeding
tube with increased ventilatory requirements), episode. However, that study included term in-
frank rectal bleeding, and severe bleeding (fatal fants, and therefore overall mortality and the inci-
bleeding, life-threatening bleeding associated with dence of new major bleeding episodes were ex-
shock, or bleeding requiring fluid boluses or red- pected to be higher in our trial. We estimated that
cell transfusion). Our definition of rectal bleeding the assignment of 329 infants to each group
was pragmatic; we defined any amount of fresh would provide 80% power (with the use of a two-
visible blood as rectal bleeding; we also performed sided test, at a 5% significance level), assuming
a prespecified sensitivity analysis excluding rec- an event rate for the primary outcome of 20% in
tal-only bleeding as a component of the primary the low-threshold group, to detect a difference
outcome. of 8 percentage points (assuming that the event
rate in the high-threshold group was 12%). This
Data Collection number of infants was rounded to a total of 660.
A bleeding-assessment form was completed daily At the time of the preplanned interim analysis
for 14 days after randomization. Thereafter, data after 87 infants were enrolled, the event rate in
on infants who were not discharged or transferred the low-threshold group was 18%, suggesting that
were collected weekly. For infants transferred to the sample-size calculation was adequate.
another hospital before day 28, primary outcome All analyses were performed according to the
data up to and including day 28, as a minimum, intention-to-treat principle with adjustments for
were documented. Reporting of safety outcomes trial site as a random effect, gestational age, and
was mandatory only at participating hospitals. intrauterine growth restriction. All tests were two-
sided; a P value of less than 0.05 was considered to
Randomization indicate statistical significance. SAS software, ver-
A Web-based service (www.sealedenvelope.com) sion 9.4 (SAS Institute), was used to conduct the
assigned infants randomly in a 1:1 ratio. Mini- analyses. All odds ratios and hazard ratios are
mization was used for the factors of intrauterine presented as the high-threshold group as com-
growth restriction and gestational age, with a 60% pared with the low-threshold group.
chance of simple random assignment. Intrauter- The primary outcome was analyzed with the
ine growth restriction was defined as birth weight use of a mixed logistic-regression model and was
in less than the 10th percentile in conjunction with compared with the use of cranial imaging per-
an estimated fetal weight crossing percentiles formed within the period from day 23 to day 38
downward during pregnancy, ultrasonographic (5 days before to 10 days after day 28), and details
evidence of uteroplacental insufficiency, or both. of any major bleeding were reported up to day 28.
Missing primary outcomes were inferred by inves- (121 in the low-threshold group and 126 in the
tigators who were unaware of the treatment as- high-threshold group) received at least one plate-
signments. For example, if an infant was dis- let transfusion before randomization.
charged before day 28 and was not readmitted
within the 28-day period, we assumed that no Data Completeness and Adherence to
major bleeding had occurred (details are provided Transfusion Protocol
in the statistical analysis plan and the Supple- The primary outcome assessment was completed
mentary Appendix). for 653 infants (99%). Six infants (2 in the low-
Further prespecified sensitivity analyses were threshold group and 4 in the high-threshold
performed, including a per-protocol analysis, an group) were withdrawn because the consenting
analysis that excluded rectal bleeding, and an process was not followed, duplicate randomiza-
analysis that assessed sensitivity to missing pri- tions occurred in error, or randomization that was
mary outcome data. Binary outcomes were ana- performed in error despite major hemorrhage in
lyzed with the use of logistic regression, and count the infant. Of the infants for whom primary out-
variables were analyzed with the use of negative come data were available, 523 (79%) underwent
binomial regression with an offset to account for cranial ultrasonography within 5 days before and
the number of days of follow-up. Several second- 10 days after day 28. Of the 130 infants who did
ary outcomes were analyzed with the use of Cox not undergo cranial ultrasonography in this pe-
proportional-hazards regression to allow for dif- riod, 77 died before day 28, and the primary
fering numbers of days of follow-up. Data on in- outcome was inferred for 53 on the basis of
fants who were transferred to nonparticipating prespecified clinical criteria (see the Supplemen-
hospitals and who had not had an event were tary Appendix). Bleeding-assessment forms were
censored at the time of transfer. The number of completed up to day 14 for 84% of the infants in
platelet transfusions administered to each group the low-threshold group and 81% in the high-
and platelet counts at which transfusions were threshold group. A total of 97% of the transfu-
administered were analyzed to assess adherence sions (93% in the low-threshold group and 99%
to the protocol. in the high-threshold group) were conducted ac-
cording to the protocol. A total of 88% were per-
formed according to the treatment-group threshold
R e sult s
(75% in the low-threshold group and 97% in the
Trial Population high-threshold group). On 124 occasions (30 in
Between June 2011 and August 2017, a total of the low-threshold group and 94 in the high-
3731 infants who were assessed for eligibility, of threshold group), a platelet transfusion was indi-
whom 660 were randomly assigned (331 to the cated according to the protocol but was not ad-
low-threshold group and 329 to the high-thresh- ministered.
old group) (Fig. 1). Of these infants, 563 in the A total of 128 new major bleeding episodes
United Kingdom (85%), 83 in the Netherlands occurred in 80 infants between randomization
(13%), and 14 in Ireland (2%) were enrolled across and day 28. These episodes occurred in 14% of
43 trial sites. We obtained written informed con- infants (45 of 328) in the high-threshold group
sent from the parent or guardian of, on average, and 11% of infants (35 of 330) in the low-thresh-
3 infants (when the platelet count was <100,000 old group. Of these episodes, 7 major bleeding
per cubic millimeter) for every 1 infant in whom episodes that occurred during the trial led to
severe thrombocytopenia developed. Baseline char- death or withdrawal of intensive care (3 in the
acteristics were well matched between the treat- high-threshold group and 4 in the low-threshold
ment groups (Table 1). The overall median birth group).
weight was 740 g (range, 360 to 2490), and the
median gestational age was 26.6 weeks (range, Primary Outcome
22.7 to 33.9). A total of 37% of the infants under- The primary outcome occurred in 26% of the in-
went randomization on or before 5 days of life, fants in the high-threshold group (85 of 324 in-
and 59% underwent randomization by day 10 fants) and 19% of the infants in the low-threshold
(Fig. S1 in the Supplementary Appendix shows group (61 of 329) (Table 2). With adjustment for
randomization according to day of age). A total gestational age and intrauterine growth restriction
of 247 of 634 infants with available data (39%) as covariates and trial site as a random effect, the

3731 Infants with platelet count

<100,000/mm3 were assessed for eligibility 542 Were excluded
86 Had major or life-threatening
congenital malformation
154 Had major or severe bleeding
within previous 72 hr
30 Had clinically significant fetal
intracranial hemorrhage
10 Had known or family history
of alloimmune thrombocytopenia
121 Were unlikely to survive more than
a few hr
12 Did not receive parenteral
vitamin K
44 Had other reason
85 Had unknown reason
3189 Were eligible
1342 Did not undergo randomization
559 Had recovery of platelet count
above 100,000/mm3 before consent
108 Had parents who were too upset to
discuss research
65 Stayed in hospital too briefly to permit
recruitment
122 Were missed
348 Had other reason
140 Had unknown reason
1847 Had parents who were approached
818 Had parents who did

not provide consent
1029 Had parents who provided consent
660 Had platelet count <50,000/mm3

and underwent randomization
331 Were in low-threshold group 329 Were in high-threshold group
329 Had primary outcome reported 324 Had primary outcome reported
256 Completed trial up to day 28 265 Completed trial up to day 28
73 Did not complete trial up to day 28 59 Did not complete trial up to day 28
62 Were transferred to nonparticipating hospital 49 Were transferred to nonparticipating hospital
before day 28 before day 28
8 Were discharged home before day 28 7 Were discharged home before day 28
3 Had other reason 3 Had other reason
2 Did not have primary outcome reported 5 Did not have primary outcome reported
1 Received palliative care 2 Received palliative care
1 Underwent duplicate randomization in error 2 Had a problem with consent
1 Underwent duplicate randomization in error
331 Were evaluated 328 Were evaluated

2 Were excluded from primary outcome analysis 1 Was excluded from all analyses owing to missing
1 Was receiving palliative care risk-adjustment covariate
1 Underwent duplicate randomization in error 5 Were excluded from primary outcome analysis
2 Were receiving palliative care
2 Had problem with consent
1 Underwent duplicate randomization in error
Figure 1. Eligibility, Randomization, and Follow-up.
Table 1. Characteristics of the Trial Population.*
Low-Threshold Group High-Threshold Group

Variable (N = 331) (N = 329)
Intrauterine growth restriction — no./total no. (%)†‡ 125/331 (38) 120/328 (37)
Antenatal glucocorticoids — no./total no. (%) 289/329 (88) 292/326 (90)
Full course of glucocorticoids, ≥2 doses — no./total no. (%) 191/281 (68) 194/281 (69)
Clinical evidence of chorioamnionitis — no./total no. (%) 26/330 (8) 28/322 (9)
Cesarean delivery — no./total no. (%)† 201/329 (61) 208/328 (63)
Female sex — no./total no. (%)† 140/331 (42) 123/328 (38)
Median weight at birth (IQR) — g† 743 (605–990) 728 (600–940)
Median gestation at birth (IQR) — wk† 26.7 (24.9–28.7) 26.6 (24.9–28.9)
Median corrected gestational age at randomization (IQR) — wk† 28.9 (26.9–31.6) 29.0 (27.2–31.5)
Median weight at randomization (IQR) — g§ 892 (670–1190) 860 (668–1170)
Median postnatal age at randomization (IQR) — days† 7.0 (3.7–18.9) 8.4 (4.0–21.0)
Randomization ≤5 days of age — no./total no. (%) 125/331 (38) 116/328 (35)
Was receiving treatment for necrotizing enterocolitis at random‑ 49/331 (15) 58/328 (18)
ization — no./total no. (%)†¶
Was receiving antibiotic treatment for sepsis at randomization 206/331 (62) 209/328 (64)
— no./total no. (%)†
Major bleeding before randomization — no./total no. (%)† ‖ 62/331 (19) 60/328 (18)
Pulmonary bleeding — no./total no. (%) 31/62 (50) 22/60 (37)
Frank rectal bleeding — no./total no. (%) 8/62 (13) 9/60 (15)
Intraventricular hemorrhage — no./total no. (%)** 40/59 (68) 39/58 (67)
Intracranial hemorrhage — no./total no. (%) 10/62 (16) 7/60 (12)
Other bleeding — no./total no. (%) 7/62 (11) 4/60 (7)
Median platelet count at randomization (IQR) — × 10−3 per 38 (28–44) 38 (29–44)
cubic millimeter†
* There were no significant differences between the groups in this post hoc analysis. IQR denotes interquartile range.
† Data were missing for one infant in the high-threshold group.
‡ Intrauterine growth restriction was defined as a birth weight below the 10th percentile and an estimated fetal weight
crossing percentiles downward during pregnancy with or without ultrasonographic evidence of uteroplacental insuffi‑
ciency.
§ Data were missing for one infant in the low-threshold group.
¶ In infants with necrotizing enterocolitis of stage IIa or higher (abdominal distention, tenderness, absent bowel
sounds, and radiologic finding of pneumatosis intestinalis), necrotizing enterocolitis was defined with the use of the
Modified Bell’s staging criteria.15
‖ Some infants had more than one bleeding episode before randomization.
** Intraventricular hemorrhage was defined as intraventricular hemorrhage filling of at least 50% of a cerebral ventricle.
odds ratio was 1.57 (95% confidence interval [CI], The percentage of infants surviving with
1.06 to 2.32; P = 0.02). bronchopulmonary dysplasia at 36 weeks of cor-
rected age was 63% in the high-threshold group
Secondary Outcomes and 54% in the low-threshold group (odds ratio,
There were no significant differences between the 1.54; 95% CI, 1.03 to 2.30). A post hoc analysis
groups with respect to the rates of minor or worse of the composite outcome of death or broncho-
bleeding. Time to major bleeding or death is pulmonary dysplasia (to allow for deaths before
shown in Figure 2. A total of 48 infants (15%) died a possible diagnosis of bronchopulmonary dys-
in the high-threshold group and 33 (10%) died in plasia) yielded a similar odds ratio. There were
the low-threshold group (odds ratio, 1.56; 95% CI, no significant differences between the groups
0.95 to 2.55). with respect to rates of survival with retinopathy

Table 2. Primary and Secondary Outcomes, According to Treatment Group.
Low-Threshold High-Threshold Odds Ratio or

Group Group Hazard Ratio
Outcome (N = 331) (N = 329) (95% CI)*
Primary outcome
Death or major bleeding episode through trial day 28 — no./total no. (%) 61/329 (19) 85/324 (26) OR, 1.57 (1.06–2.32)†
Secondary outcomes‡
Death through trial day 28 — no./total no. (%) 33/330 (10) 48/326 (15) OR, 1.56 (0.95–2.55)
At least one major bleeding episode through trial day 28 — no./total no. (%) 35/330 (11) 45/328 (14) HR, 1.32 (1.00–1.74)
Survival with bronchopulmonary dysplasia at 36 wk — no./total no. (%)§ 153/281 (54) 169/269 (63) OR, 1.54 (1.03–2.30)
Post hoc outcome of death or bronchopulmonary dysplasia at 36 wk — 200/329 (61) 224/324 (69) OR, 1.56 (1.07–2.27)
no./total no. (%)
Discharge by 38 wk of corrected gestational age — no./total no. (%) 41/328 (12) 29/326 (9) HR, 0.68 (0.46–1.00)
Survival with unilateral or bilateral retinopathy of prematurity of stage 71/297 (24) 82/279 (29) OR, 1.37 (0.91–2.08)
≥2 at 38 wk of corrected gestational age — no./total no. (%)¶
Unilateral or bilateral retinopathy of prematurity of stage ≥2 treated with 29/295 (10) 36/278 (13) OR, 1.38 (0.79–2.42)
laser or bevacizumab therapy — no./total no. (%)
New sepsis event after randomization — no./total no. (%)‖ 175/326 (54) 181/324 (56) HR, 1.10 (0.92–1.33)
New necrotizing enterocolitis event after randomization — no./total no. (%)** 54/326 (17) 42/324 (13) HR, 0.72 (0.37–1.41)
At least one major bleeding episode through trial day 28 — no./total no. (%) 14/330 (4) 11/328 (3) HR, 0.80 (0.40–1.60)
At least one minor or worse bleeding episode through trial day 14 — 232/328 (71) 225/324 (69) HR, 0.96 (0.84–1.09)
no./total no. (%)
At least one moderate or worse bleeding episode up to trial day 14 — 114/328 (35) 111/324 (34) HR, 1.01 (0.86–1.18)
no./total no. (%)
At least one platelet transfusion — no./total no. (%) 177/331 (53) 296/328 (90) HR, 2.75 (2.36–3.21)
No. of platelet transfusions administered in infants who received at least one 2 (1–3) 2 (1–3)
transfusion — median (IQR)
* Odds ratios (ORs) are a relative measure of the odds of the binary outcome occurring by the end of the time period in the high-threshold
group as compared with the low-threshold group. Hazard ratios (HRs) are a relative measure of the hazard (instantaneous event rate) of
the outcome occurring at any time over the time period in the high-threshold group as compared with the low-threshold group. The odds
ratios have been adjusted for trial site as a random effect, intrauterine growth restriction, and gestational age at birth and are based on
marginal effects. The hazard ratios have been adjusted for trial site, intrauterine growth restriction, and gestational age at birth and are
based on marginal Cox models.
† Values for the risk-adjustment covariates were missing for 1 infant in the high-threshold group, and that infant was excluded from the
analysis. Outcome data were missing for 7 infants (2 in the low-threshold group and 5 in the high-threshold group). Outcome data were
inferred for 47 infants (27 in the low-threshold group and 20 in the high-threshold group). P = 0.02 for the comparison of the two groups.
‡ The widths of the confidence intervals for secondary outcomes have not been adjusted for multiplicity of analyses, so the intervals should
not be used to infer definitive treatment effects.
§ Data are shown for infants who were alive at 36 weeks.
¶ Data are shown for infants who were alive 4 weeks after birth.
‖ The hazard ratio was also adjusted for sepsis at randomization.
** The hazard ratio was also adjusted for necrotizing enterocolitis at randomization.
of prematurity or in the rates of other complica- to more than 50,000 per cubic millimeter while
tions of prematurity (Table 2). awaiting planned transfusion. There was a wide
A total of 90% of the infants (296 of 328) in separation between the groups in the numbers of
the high-threshold group and 53% (177 of 331) platelet transfusions on the first trial day (Fig. 3).
in the low-threshold group received at least one
platelet transfusion (hazard ratio, 2.75; 95% CI, Sensitivity Analyses and Subgroup Analyses
2.36 to 3.21) (Table 2). A total of 10% (32 of 328 Results were materially unchanged in a per-proto-
infants) in the high-threshold group did not re- col analysis (odds ratio for the primary outcome,
ceive a transfusion indicated by a low platelet 1.68; 95% CI, 1.11 to 2.55), in an analysis exclud-
count owing to an increase in the platelet count ing rectal-only bleeding (odds ratio, 1.75; 95% CI,
the investigators to be possibly related to the trans-

100 fusion.
90
Patients Who Survived without Major Bleeding (%)
Low-threshold group Discussion

80
This large, multicenter, randomized trial involv-
High-threshold group
70 ing preterm infants with severe thrombocytope-
nia showed that more deaths, major bleeding, or
60
both occurred when a higher prophylactic plate-
50 let-count transfusion threshold of 50,000 per cu-
bic millimeter was used than when a threshold of
40 25,000 per cubic millimeter was used. The re-
cruited cohort comprised high-risk infants with
30
a median gestational age of 26.6 weeks and birth
20 weight of 740 g. They were often critically unwell,
P=0.02 by log-rank test
evidenced by their high background incidence of
10 Marginal Cox proportional-hazards regression, P=0.002 necrotizing enterocolitis and sepsis at recruitment,
(adjusted for trial site, IUGR, and gestational age) overall mortality, and incidence of major bleeding.
0
1 4 7 10 13 16 19 22 25 28 The respective difference in event rate (26% vs.
Trial Day 19%) between the high-threshold and low-thresh-
No. at Risk
old treatment groups implies that reducing the
Low-threshold group 309 297 290 284 279 276 274 271 269 transfusion trigger from 50,000 per cubic milli-
High-threshold group 308 298 282 270 264 256 252 247 243 meter to 25,000 per cubic millimeter may prevent
death or major bleeding in 7 of 100 preterm neo-
Figure 2. Patients Who Survived without Major Bleeding to Trial Day 28,
nates with severe thrombocytopenia.
According to Treatment Group.
Potential limitations of our trial require con-
IUGR denotes intrauterine growth restriction.
sideration. A total of 3.2% (45 of 1393) of plate-
let transfusions were additional transfusions that
were not indicated by the protocol, but overall,
1.14 to 2.67), and in analyses assessing sensitivity adherence to the protocol was good. The results
to missing data (Tables S5 and S6 in the Supple- were similar in direction for the analysis of com-
mentary Appendix). In subgroup analyses, there ponents of the primary outcome separately, and
were no significant interactions between treatment the trial findings remained consistent in a per-
and the presence of intrauterine growth restric- protocol analysis. The rate of rectal bleeding, a
tion, gestational age, or postnatal age at random- component of the primary outcome, may not have
ization (Table S4 in the Supplementary Appendix). been clinically significant; however, the results
were materially unchanged when these bleeding
Serious Adverse Events episodes were excluded in a sensitivity analysis.
Not including major bleeding, 92 serious adverse Only 37% of the infants were recruited on or be-
events were reported in 74 infants in the low- fore 5 days of life, but this is probably because
threshold group and 91 serious adverse events necrotizing enterocolitis and late-onset sepsis,
were reported in 81 infants in the high-threshold both clinically significant causes of severe neonatal
group; these included multiorgan failure, necro- thrombocytopenia, are less common in the first
tizing enterocolitis, renal failure, respiratory fail- postnatal week than at later ages.
ure, and sepsis. There was no significant differ- Our trial highlights the importance of trials of
ence in rates of serious adverse events between the platelet transfusion involving patients with condi-
groups (25% in the high-threshold group and tions other than hematologic cancers.16 It adds to
22% in the low-threshold group; odds ratio, 1.14; previous data indicating a tenuous relationship
95% CI, 0.78 to 1.67) (Table 2, and Table S3 in the between platelet count and bleeding.1,6,13 Although
Supplementary Appendix). One serious adverse retrospective studies have suggested that platelet
event (respiratory deterioration attributed to tran- transfusions may cause harm in neonates inde-
sient acute lung inflammation) was considered by pendently of the disease process, data from

randomized, controlled trials to support this are A

lacking.17 A recent randomized trial of platelet Low-threshold group High-threshold group
transfusion (added to standard care) in adults with
Median Platelet Count (×10 –3 per mm3)

200
intracranial bleeding associated with antiplatelet
agents showed increased rates of death or depen-
dence (defined as a score of 4 to 6 on the modified 150
Rankin scale [a measure of disability in which
scores range from 0 to 6, with higher scores indi-
cating greater disability and a score of 6 indicating 100
death]) with platelet transfusions.18 Associations
between platelet transfusion and increased risks of
50
thrombosis and death have been reported in large
retrospective studies involving adult patients with
platelet-consumptive disorders.19 0
In our trial, the reasons for the between-group 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
differences in mortality and major bleeding are Trial Day
unknown. Platelets have recognized immunolog- No. at Risk
ic and inflammatory effects, outside of effects on Low-threshold group 331 283 261 221 221 196 174 159 140 139 131 127 132 124
High-threshold group 328 271 242 219 196 184 187 158 147 133 144 122 125 130
hemostasis.20,21 The effect of transfusing adult
platelets to a delicately balanced neonatal hemo- B
static system with relatively hypofunctional plate- Low-threshold group High-threshold group
lets is also poorly understood.22 Inflammatory
consequences after transfusion of platelet com- 200
ponents as biologic agents, hemodynamic shifts

No. of Platelet Transfusions
related to platelet-transfusion volume, or both,

150
coupled with fragility of the germinal matrix and
disturbances in organ and brain blood flow could
have contributed to an increased risk of hemor- 100
rhage.23-25 Preterm lungs, with a large capillary bed
and abundant immune cells, may be vulnerable to
proinflammatory injury mediated by platelet trans- 50
fusion–derived bioreactive components.26 Rates

of bronchopulmonary dysplasia appeared to be
0
higher in the group that received transfusions at 1 2 3 4 5 6 7 8 9 10 11 12 13 14
a threshold of 50,000 per cubic millimeter than Trial Day
in the group that received transfusions at a thresh-
old of 25,000 per cubic millimeter, although the Figure 3. Platelet Count and Number of Platelet Transfusions per Day,
lack of adjustment for multiplicity must be con- According to Treatment Group to Trial Day 14.
sidered in interpreting differences between groups Panel A shows the median platelet counts in the infants. I bars denote in‑
in secondary outcomes. Other potential causes terquartile ranges. Panel B shows the number of platelet transfusions per
trial day.
of adverse outcomes include platelet-derived re-
active oxygen species and proangiogenic factors
that may aggravate the aberrant angiogenesis
characterizing bronchopulmonary dysplasia, a threshold of 25,000 per cubic millimeter within
common complication of prematurity.27 Vessel oc- 28 days after randomization.
clusion by platelet microthrombi could also com-
promise collateral blood flow. The views expressed are those of the authors and not neces-
sarily those of the National Health Service (NHS), the NHS
In conclusion, among preterm infants with se- Blood and Transplant, the National Institute for Health Re-
vere thrombocytopenia, the use of a platelet-count search, or the U.K. Department of Health.
threshold of 50,000 per cubic millimeter for pro- Supported by the National Health Service Blood and Trans-
plant Research and Development Committee (Ref No: BS06/1);
phylactic platelet transfusion resulted in a higher Sanquin Research, Amsterdam (grant PPOC-12-012027); Adden-
rate of death or major bleeding than a restrictive brooke’s Charitable Trust; the Neonatal Breath of Life Fund

9145; and the National Institute for Health Research Clinical of whom died in 2015 and 2016, respectively, for their valued con-
Research Network. tributions and enthusiasm to the success of the trial; the National
Disclosure forms provided by the authors are available with Health Service Blood and Transplant Clinical Trials Unit team, in-
the full text of this article at NEJM.org. cluding Anna Sidders and Heather Smethurst (trial coordinators),
We thank Dr. Neil Murray and all the hospital staff and research Louise Choo, Brennan Kahan, Helen Thomas, Laura Pankhurst,
teams who helped conduct this trial (for a complete list, see the and Agne Zarankaite (trial statisticians), and Colin Morley, Sandy
Supplementary Appendix), all the families who agreed to partici- Calvert, and Anthony Emmerson (trial steering committee) for their
pate; Dr. Shobha Cherian, consultant neonatologist and principal support; and Adrian Newland, Henry Halliday, Paul White, Gavin
investigator at University Hospital of Wales, Cardiff, and Anna Hen- Murphy, Michael Greaves, Keith Wheatley, and Marc Turner (the
drickson, research nurse at St. Mary’s Hospital, Manchester, both data and safety monitoring committee).
Appendix
The authors’ affiliations are as follows: the Department of Neonatology, National Maternity Hospital, Dublin (A.C.), and University
Maternity Hospital Limerick, Limerick (R.K.) — both in Ireland; Department of Transfusion Medicine, National Health Service (NHS)
Blood and Transplant, the Department of Haematology, John Radcliffe Hospital, Oxford University Hospitals NHS Foundation Trust,
the Radcliffe Department of Medicine, University of Oxford, and Oxford Biomedical Research Centre Haematology Theme, Oxford
(S.J.S.), the Department of Obstetrics and Gynaecology, University of Cambridge (K.W.), NHS Blood and Transplant, Clinical Trials Unit
(K.W., C.H., A. Deary, R.H., V.H., A.M., C.L.), and the Neonatal Intensive Care Unit, Cambridge University Hospitals NHS Trust (A.
D’Amore), Cambridge, the Neonatal Intensive Care Unit, Evelina London Children’s Hospital, Guy’s and St. Thomas’ NHS Foundation
Trust (B.L.S., T.W.), NHS Blood and Transplant (H.N.), and the Department of Hematology, Imperial College London (H.N.), London,
and the Neonatal Intensive Care Unit, Norfolk and Norwich University Hospitals NHS Foundation Trust, and Norwich Medical School,
University of East Anglia, Norwich Research Park, Norwich (P.C.) — all in the United Kingdom; the Center for Clinical Transfusion
Research, Sanquin Blood Supply (S.F.F.-G., K.F.), and the Departments of Epidemiology (S.F.F.-G.) and Neonatology (E.L.), Leiden
University Medical Center, Leiden, and the Departments of Pediatric Hematology (S.F.F.-G., K.F.) and Neonatology (W.O.), Emma
Children’s Hospital, Academic Medical Center, Amsterdam — both in the Netherlands; and the Neonatal Intensive Care Unit, Cloudnine
Hospital, Bangalore, India (V.V.).
References
1. Sparger KA, Assmann SF, Granger S, transfusions in thrombocytopenic prema- 17. Kenton AB, Hegemier S, Smith EO, et
et al. Platelet transfusion practices among ture infants. J Pediatr 1993;123:285-91. al. Platelet transfusions in infants with
very-low-birth-weight infants. JAMA Pedi- 10. Lieberman L, Bercovitz RS, Sholapur necrotizing enterocolitis do not lower
atr 2016;170:687-94. NS, Heddle NM, Stanworth SJ, Arnold mortality but may increase morbidity.
2. Ulusoy E, Tüfekçi O, Duman N, Kum- DM. Platelet transfusions for critically ill J Perinatol 2005;25:173-7.
ral A, Irken G, Oren H. Thrombocytope- patients with thrombocytopenia. Blood 18. Baharoglu MI, Cordonnier C, Al-Sha-
nia in neonates: causes and outcomes. 2014;123:1146-1151, 1280. hi Salman R, et al. Platelet transfusion
Ann Hematol 2013;92:961-7. 11. New HV, Berryman J, Bolton-Maggs versus standard care after acute stroke
3. Gunnink SF, Vlug R, Fijnvandraat K, PHB, et al. Guidelines on transfusion for due to spontaneous cerebral haemor-
van der Bom JG, Stanworth SJ, Lopriore E. fetuses, neonates and older children. Br J rhage associated with antiplatelet therapy
Neonatal thrombocytopenia: etiology, Haematol 2016;175:784-828. (PATCH): a randomised, open-label, phase
management and outcome. Expert Rev 12. Venkatesh V, Curley A, Khan R, et al. 3 trial. Lancet 2016;387:2605-13.
Hematol 2014;7:387-95. A novel approach to standardised record- 19. Goel R, Ness PM, Takemoto CM,
4. Sola-Visner M, Bercovitz RS. Neonatal ing of bleeding in a high risk neonatal Krishnamurti L, King KE, Tobian AA.
platelet transfusions and future areas of population. Arch Dis Child Fetal Neonatal Platelet transfusions in platelet consump-
research. Transfus Med Rev 2016;30:183-8. Ed 2013;98:F260-F263. tive disorders are associated with arterial
5. Resch E, Hinkas O, Urlesberger B, 13. Muthukumar P, Venkatesh V, Curley thrombosis and in-hospital mortality.
Resch B. Neonatal thrombocytopenia- A, et al. Severe thrombocytopenia and Blood 2015;125:1470-6.
causes and outcomes following platelet patterns of bleeding in neonates: results 20. McFadyen JD, Kaplan ZS. Platelets are
transfusions. Eur J Pediatr 2018;177:1045- from a prospective observational study not just for clots. Transfus Med Rev 2015;
52. and implications for use of platelet trans- 29:110-9.
6. Stanworth SJ, Clarke P, Watts T, et al. fusions. Transfus Med 2012;22:338-43. 21. Stolla M, Refaai MA, Heal JM, et al.
Prospective, observational study of out- 14. Curley A, Venkatesh V, Stanworth S, et Platelet transfusion — the new immunol-
comes in neonates with severe thrombocy- al. Platelets for neonatal transfusion — ogy of an old therapy. Front Immunol
topenia. Pediatrics 2009;124(5):e826-e834. study 2: a randomised controlled trial to 2015;6:28.
7. Josephson CD, Su LL, Christensen compare two different platelet count 22. Ferrer-Marin F, Stanworth S, Joseph-
RD, et al. Platelet transfusion practices thresholds for prophylactic platelet trans- son C, Sola-Visner M. Distinct differences
among neonatologists in the United fusion to preterm neonates. Neonatol in platelet production and function be-
States and Canada: results of a survey. Pe- 2014;106:102-6. tween neonates and adults: implications
diatrics 2009;123:278-85. 15. Bell MJ, Ternberg JL, Feigin RD, et al. for platelet transfusion practice. Transfu-
8. Cremer M, Sola-Visner M, Roll S, et al. Neonatal necrotizing enterocolitis: thera- sion 2013;53:2813-21.
Platelet transfusions in neonates: practic- peutic decisions based upon clinical stag- 23. Bassan H. Intracranial hemorrhage in
es in the United States vary significantly ing. Ann Surg 1978;187:1-7. the preterm infant: understanding it, pre-
from those in Austria, Germany, and Swit- 16. Stanworth SJ, Estcourt LJ, Powter G, venting it. Clin Perinatol 2009;36:737-62.
zerland. Transfusion 2011;51:2634-41. et al. A no-prophylaxis platelet-transfu- 24. Poryo M, Boeckh JC, Gortner L, et al.
9. Andrew M, Vegh P, Caco C, et al. A sion strategy for hematologic cancers. Ante-, peri- and postnatal factors associ-
randomized, controlled trial of platelet N Engl J Med 2013;368:1771-80. ated with intraventricular hemorrhage in

very premature infants. Early Hum Dev intraventricular hemorrhage. Transfusion 27. Mestan KK, Gotteiner N, Porta N,
2018;116:1-8. 2014;54:104-8. Grobman W, Su EJ, Ernst LM. Cord blood
25. Christensen RD, Baer VL, Lambert 26. Aghai ZH, Saslow JG, Mody K, et al. biomarkers of placental maternal vascu-
DK, Ilstrup SJ, Eggert LD, Henry E. Asso- IFN-γ and IP-10 in tracheal aspirates from lar underperfusion predict bronchopul-
ciation, among very-low-birthweight neo- premature infants: relationship with bron- monary dysplasia-associated pulmonary
nates, between red blood cell transfu- chopulmonary dysplasia. Pediatr Pulmonol hypertension. J Pediatr 2017;185:33-41.
sions in the week after birth and severe 2013;48:8-13. Copyright © 2018 Massachusetts Medical Society.

Curley 2018 Platelet Transfusion Threshols in Neonates

Uploaded by

Copyright:

Available Formats

Curley 2018 Platelet Transfusion Threshols in Neonates

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Curley 2018 Platelet Transfusion Threshols in Neonates

Uploaded by

Copyright:

Available Formats

The n e w e ng l a n d j o u r na l of m e dic i n e

Randomized Trial of Platelet-Transfusion

The New England Journal of Medicine

event up to the end of the trial, a serious adverse Oversight

The New England Journal of Medicine

3731 Infants with platelet count

818 Had parents who did

1029 Had parents who provided consent

660 Had platelet count <50,000/mm3

331 Were in low-threshold group 329 Were in high-threshold group

331 Were evaluated 328 Were evaluated

Figure 1. Eligibility, Randomization, and Follow-up.

Table 1. Characteristics of the Trial Population.*

Low-Threshold Group High-Threshold Group

The New England Journal of Medicine

Table 2. Primary and Secondary Outcomes, According to Treatment Group.

Low-Threshold High-Threshold Odds Ratio or

the investigators to be possibly related to the trans-

Low-threshold group Discussion

The New England Journal of Medicine

randomized, controlled trials to support this are A

Median Platelet Count (×10 –3 per mm3)

ponents as biologic agents, hemodynamic shifts

related to platelet-transfusion volume, or both,

fusion–derived bioreactive components.26 Rates

n engl j med nejm.org 9

The New England Journal of Medicine

You might also like