This study is a multicentre, double-blind, randomized therapeutic trial.

The primary objective of this study is to evaluate non-inferiority with regard to prevention and control of haemorrhage:

• of platelet concentrates treated by pathogen reduction(Intercept amotosalen and UVA procedure)
• compared with the usual platelet concentrates (in additive solution intersol), reference arm, and
• compared with platelet concentrates re-suspended in autologous plasma (historic arm) These three products are available and authorised by ANSM (formerly AFSSAPS).

The secondary objectives is to evaluate the transfusion needs, transfusion outcomes and safety and the decreased frequency of grade 2 or higher side effects related to transfusion allergy to platelets.

There is an unresolved difficulty in the evaluation of haemorrhagic symptoms in thrombocytopenia due to the very nature of the scale, which is the international standard at this time (WHO scale). This scale is based on the level of blood loss and is applicable to any haemostasis disorder. We will keep it as the standard but have decided to be particularly rigorous in the data collection and will perform daily haemorrhagic assessment.

Several sequences of missing data can be imputed for one patient. Each sequence of missing data will be replaced if and only if the number of consecutive days missing does not exceed 15%* of the total length of the patient's stay. If one sequence of missing data is longer than the 15%, no replacement will be done for the patient. (Example: If the total stay is 30 days, the maximal length of a missing data sequence accepted is 4 days). The following strategies will be used to replace missing data:

• The first observation is missing: Next observation carried backwards (NOCB) assigns the next known score after the missing value to the missing one.

• The last observation is missing: Last observation carried forward (LOCF) assigns the last known score before the missing value to the missing one. (Suppose that the situation is stable whilst the patient is leaving the hospital.)

• Sequence of one or several missing data with non-missing data before and after the sequence: Last and Next1 assigns the average of the person's last known and next known observation to the missing value. The score is rounded down to the nearest whole number if needed. (Ex mean (1+2) =1)

* 15% rounded up to nearest whole number.

1 : Engels, J.M. 2003. Imputation of Missing Longitudinal Data : a Comparison of Methods. Journal of Clinical Epidemiology 56 (2003) 968-976

Following a quality analysis of EFFIPAP study's recruitment, it was decided by the sponsor to increase the number of patients. Approximatively thirty additional patients will be included in order to replace non analyzable patients for the following reasons : wrongly included, non-transfused patients, consent withdrawal. Those 30 additional patients will allow us to reach our initial target of 810 analyzable patients in order to respond to the main objective of the study

Inclusion Criteria:

• Patients aged 18 years or older
• Patient hospitalised for bone marrow aplasia, with expected stay of over 10 days and in principle requiring platelet transfusion support (at least twice).
• Signed informed consent
• Patients with DIC can be included; they will undergo a separate analysis.
• A negative pregnancy test is necessary before inclusion in all women of childbearing age

Exclusion Criteria:

• Patient included in this trial previously during a prior aplasia episode.
• Patient requiring curative anticoagulant treatment at the time of inclusion (vitamin K antagonists, heparin (LMWH and NFH), anti-IIa and Xa at curative doses for treatment or prophylaxis of arterial or venous thromboembolic disease (TED) or as part of the treatment for cardiac valvulopathy and complications of atrial fibrillation).
• Thrombocytopenia due to increased destruction
• Patient requires washed platelet concentrates (i.e., with residual plasma less than that remaining during the addition of an additive solution) due to previous intolerance to platelets (cf IgA deficiency, history of major allergic reaction)
• Patient requiring products "CMV negative " (previously included in a protocol transfusion CMV negative)
• Patients with platelet transfusion refractoriness during a previous period of cytopenia, including patient with platelet refractoriness related to an anti-HLA alloimmunization (thus, patient already known as requiring compatible platelets HLA)
• Patient who requires compatible HLA platelets due to a refractory state relative to anti-HLA alloimmunization
• Patient presenting a platelet transfusion refractoriness at the time of previous aplasia.
• Protected adults and persons deprived of liberty