Antenatal interventions for fetomaternal alloimmune thrombocytopenia

Abstract

Background

Fetomaternal alloimmune thrombocytopenia results from the formation of antibodies by the mother which are directed against a fetal platelet alloantigen inherited from the father. The resulting fetal thrombocytopenia (reduced platelet numbers) may cause bleeding, particularly into the brain, before or shortly after birth. Antenatal treatment of fetomaternal alloimmune thrombocytopenia includes the administration of intravenous immunoglobulin (IVIG) and/or corticosteroids to the mother to prevent severe fetal thrombocytopenia. IVIG and corticosteroids both have short‐term and possibly long‐term side effects. IVIG is also costly and optimal regimens need to be identified.

Objectives

To determine the optimal antenatal treatment of fetomaternal alloimmune thrombocytopenia to prevent fetal and neonatal haemorrhage and death.

Search methods

We searched the Cochrane Pregnancy and Childbirth Group's Trials Register (28 February 2011) and bibliographies of relevant publications and review articles.

Selection criteria

Randomised controlled studies comparing any intervention with no treatment, or comparing any two interventions.

Data collection and analysis

Two review authors independently assessed eligibility, trial quality and extracted data.

Main results

We included four trials involving 206 people. One trial involving 39 people compared a corticosteroid (prednisone) versus IVIG alone. In this trial, where analysable data were available, there was no statistically significant differences between the treatment arms for predefined outcomes. Three trials involving 167 people compared IVIG plus a corticosteroid (prednisone in two trials and dexamethasone in one trial) versus IVIG alone. In these trials there was no statistically significant difference in the findings between the treatment arms for predefined outcomes (intracranial haemorrhage; platelet count at birth and preterm birth). Lack of complete data sets and important differences in interventions precluded the pooling of data from these trials.

Authors' conclusions

The optimal management of fetomaternal alloimmune thrombocytopenia remains unclear. Lack of complete data sets for two trials and differences in interventions precluded the pooling of data from these trials which may have enabled a more developed analysis of the trial findings. Further trials would be required to determine optimal treatment (the specific medication and its dose and schedule). Such studies should include long‐term follow up of all children and mothers.

Plain language summary

Antenatal interventions for fetomaternal alloimmune thrombocytopenia

The optimal management of pregnant women with a previous child affected by fetomaternal alloimmune thrombocytopenia remains unclear.

Differences between the platelets of an unborn baby and their mother, because of a fetal platelet alloantigen inherited from the father and not shared by the mother, can lead to the development of antibodies in the mother. These attack the baby's platelets and can lead to bleeding complications for the baby. This is termed fetomaternal alloimmune thrombocytopenia, meaning a reduction in the number of platelets in the baby’s blood because of their destruction by the mother's antibodies. The baby is at risk of bleeding in the brain (intracranial haemorrhage) before or shortly after birth and the baby may die or have long‐lasting problems. Fetomaternal alloimmune thrombocytopenia often occurs in the first pregnancy and is usually only diagnosed following the birth of a baby with a low platelet count. Active antenatal management to prevent severe thrombocytopenia is confined to those women who have had a previously affected infant. 
 
 Fetal blood sampling from the umbilical vein is used to monitor the fetal platelet count but carries a risk of fetal loss. Intrauterine platelet transfusions are also associated with a risk of fetal loss and are now only used as treatment for women who do not respond to medical treatment. Medication is intravenous immunoglobulin (IVIG) with or without immunosuppression with corticosteroids. The four small trials (n = 206) provide insufficient evidence to determine the best treatment for fetomaternal alloimmune thrombocytopenia.

Background

Fetomaternal alloimmune thrombocytopenia (FMAIT) (thrombocytopenia meaning a reduction in the number of platelets in the blood which can lead to bleeding into the skin and brain), also commonly known as 'neonatal alloimmune thrombocytopenia' (NAIT), occurs when the mother produces antibodies against a platelet alloantigen that the fetus has inherited from the father (Moulinier 1953; Shulman 1964). The mother's antibodies cross the placenta and destroy the baby's platelets, and this may result in internal bleeding: in severe cases this is in the brain. As a result, babies may die in utero or have long‐lasting disability. Fetomaternal alloimmune thrombocytopenia is usually only diagnosed following the birth of a thrombocytopenic baby, or, less commonly, it may be suspected following the detection of an antenatal intracranial haemorrhage (ICH).

Fetomaternal alloimmune thrombocytopenia (FMAIT): nature and incidence

Platelets are blood cells that play an important role in blood clotting. Platelet antigens are proteins found on the surface of these cells. Platelet‐specific alloantigens or human platelet alloantigens (HPAs) are expressed predominantly on platelets, but can also be found on endothelial cells, smooth muscle cells and fibroblasts (Rozman 2002).

HPAs have been numbered according to the order in which they were described, with the high frequency antigen designated 'a' and the low frequency antigen 'b'. Antibodies against HPAs have been implicated in immune mediated phenomena, such as when there is reduced response to transfused platelets, post‐transfusion purpura (which presents with a rapid decline in the person's platelet count 5 to 12 days after transfusion) and FMAIT.

It has long been believed that, in cases of FMAIT, the source of antigen is fetal platelets, which have been shown to enter the mother's circulation as early as the 16th week of pregnancy (Gruel 1986). It is probable that the immunising antigen can enter the maternal circulation as early as 10 to 12 weeks. The ß3 Integrin, on which the HPA‐1a antigen (the most common cause of FMAIT) is expressed, is known to be expressed on the villous trophoblast and it may be from this source that the mother is immunised (Kumpel 2003). The IgG antibodies produced by the mother are transported across the placenta to the fetal blood system, presumably by the MHC‐class 1‐related FcRn (neonatal Fragment crystalline Receptor) (Simister 1997), where IgGs bind to fetal platelets. The antibody‐coated platelets are removed from the fetal circulation by the reticuloendothelial system, resulting in fetal thrombocytopenia of varying severity. Whether these same anti‐platelet antibodies binding have any effect on fetal platelet production via binding to megakaryocytes remains uncertain.

The first case of FMAIT within a family is usually detected at or shortly after birth. The newborn usually presents with skin bleeding or, in a small percentage of cases, is found to have a low platelet count by chance. However, in severe cases intracranial haemorrhage (ICH) may occur (a risk of 10% to 30%) either in the womb, or during or shortly after birth, which may lead to long lasting disability or even death.The fetus or neonate is usually otherwise well and there is no evidence of thrombocytopenia or an immune‐mediated pathology in the mother, although the presentation of gestational thrombocytopenia may be confusing. More unusual presentations include unexplained hydrocephalus, fetal anaemia and intrauterine death in the second and third trimesters.

Postnatal treatment of severe fetal thrombocytopenia is with antigen‐negative platelets until the platelet count recovers, usually seven to 10 days following birth, although platelet transfusion may sometimes be needed for up to four to six weeks (Rayment 2003).

The incidence of FMAIT in Caucasian populations, based on population surveys, is between 1 in 1000 and 1 in 1500 live births (Mueller Eckhardt 1985; Blanchette 1990; Kaplan 1994; Williamson 1998; Murphy 2000; Kjeldsen‐Kragh 2007). The incidence of severe thrombocytopenia (< 50 x 10*9/l) is 1 in 1695 live births (Turner 2005). However, in the absence of neonatal screening, it is likely that the true incidence is higher. In one prospective study, only 37% of cases of severe fetomaternal alloimmune thrombocytopenia were detected (Turner 2005). One source of confusion is that the severity of thrombocytopenia and especially the incidence of ICH are far different depending upon whether a case is accrued in the neonatal nursery (e.g. following investigation for bleeding) or is identified in an all‐inclusive screening study. FMAIT is the most important cause of severe fetal and neonatal thrombocytopenia, both because of its frequency and the severity of the bleeding associated with it (Burrows 1993). A fetal or neonatal platelet count of < 20 x 10*9/l is usually caused by FMAIT due to anti‐HPA‐1a, as are approximately half of the cases in which the neonatal platelet count is < 50 x 10*9/l (Murphy 2007).

In the UK, The National Blood Service Oxford and Cambridge Platelet Immunology Laboratories have demonstrated maternal antibodies directed against HPA‐1a in 85.1% of cases of FMAIT, HPA‐5b in 10.4% and Gov antigens in 5%. Similar frequencies have been reported elsewhere for Caucasian populations (Mueller Eckhardt 1989; Berry 2000). However, in other countries the gene frequency for the various platelet alloantigens differs (Kim 1995; Tanaka 1996; Chang 1998). Thus, in Japan where the HPA 1a/1b polymorphism does not exist, HPA‐5b incompatibility is the most common cause of FMAIT with HPA 4a and 6W on glycoprotein IIIa also being commonly implicated (Tanaka 1996).

Unlike its red cell equivalent (haemolytic disease of the newborn), FMAIT often occurs in the first pregnancy. However, there is currently no consensus regarding the utility of screening previously unaffected women for anti‐platelet antibodies and thus determining early in gestation whether a pregnancy is affected by FMAIT. Active antenatal management of this disease is therefore confined to those women who have had a previously affected fetus (Murphy 2000).

Following an affected pregnancy, the father can be tested for presence of the relevant HPA: the risk of recurrence in subsequent pregnancies is virtually 100% if the father is homozygous for the responsible HPA and 50% if he is heterozygous. In cases where the father is heterozygous or unavailable, the baby can be tested by amniocentesis for the relevant HPA.

The severity of FMAIT usually increases with each pregnancy. Attempts have been made to predict a fetus at‐risk from severe thrombocytopenia by the use of serial antibody titres in order to determine which at‐risk pregnancies need treatment. However, although an increasing titre largely correlates with increasing thrombocytopenia, occasionally the antibody may be undetectable or of low‐titre in severely affected cases (Rayment 2003; Bertrand 2006; Ghevaert 2007; Killie 2007). As standard laboratory tests are currently unhelpful in predicting the degree of thrombocytopenia in an individual fetus, the clinical history of an affected sibling is currently the best indicator of risk to a current pregnancy (Bussel 1997; Birchall 2003; Radder 2003). A review of the literature found that the recurrence rate of ICH in the subsequent pregnancies of women with a history of FMAIT with ICH was 72% (95% confidence interval (CI) 46 to 98), without the inclusion of fetal deaths, and 79% (95% CI 61 to 97) with their inclusion (Radder 2003). In contrast, the risk of ICH in those with a history of FMAIT but without ICH was estimated to be 7% (95% CI 0.5 to 13).

Antenatal management and outcomes

The goal of antenatal management is to prevent severe thrombocytopenia and thus ICH which may result in death, either in utero or after birth, or long lasting disability. The antenatal treatment of FMAIT has evolved over the past 25 years, largely based on published case series, detailing outcomes with differing regimens, including serial intrauterine platelet transfusion (Murphy 1990), weekly intravenous immunoglobulin (IVIG) and immunosuppression with corticosteroids (Murphy 1994; Bussel 1988; Kaplan 1998).

Diagnostic fetal blood sampling and Intrauterine transfusions

Fetal blood sampling involves the insertion of a needle into the umbilical or intrahepatic vein to sample fetal blood in order to ascertain the platelet count. The procedure is usually complemented by the transfusion of a specially selected very concentrated platelet suspension that is both HPA and blood group‐compatible, to reduce the risk of bleeding (Murphy 1990; Paidas 1995).

With reports of a fetal loss rate of around 6% per pregnancy (Overton 2002), serial (weekly) platelet transfusions now take second place for the management of affected fetuses to maternal administration of weekly IVIG, with or without the use of corticosteroids. An important unresolved issue in the management of at‐risk pregnancies is how to safely minimise or eliminate fetal blood sampling (Radder 2003; Berkowitz 2007). Some fetal medicine teams have completely abandoned fetal blood sampling with or without platelet transfusion, whilst some restrict repeated invasive testing to those pregnancies not responding to medical treatment.

IVIG

IVIG is widely used in immune mediated diseases. Its mechanism of action is unclear and may be due to Fragment crystallizable Receptor (FcR) blockade on macrophages (leading to inhibition of uptake of the antibody‐coated platelets by fetal macrophages), enhanced expression of FCy/RllB inhibitory receptors on splenic macrophages (Samuelsson 2001), suppression of maternal antibody production or reduction of placental transfer of the pathological antibodies (Clark 1998). IVIG has been demonstrated to modulate the function of antigen presenting cells (Bayry 2003). There is conflicting evidence on its efficacy in preventing ICH; with some reports documenting good results (Lynch 1992; Bussel 1996a; Bussel 1997) and others reporting failure of IVIG to prevent haemorrhage, particularly in severely affected fetuses (Zimmermann 1992; Kroll 1994; Murphy 1994; Sainio 1999). However, in the latter cases only IVIG 1 g/kg/week was utilised.

IVIG is most often given to the mother on a weekly basis until delivery, but various treatment schedules have been used. The only way to monitor the efficacy of this treatment during pregnancy is by fetal blood sampling, which carries a risk of fetal loss with each procedure, as described above. After birth, the neonatal platelet count and the presence or absence of ICH provide measures of IVIG efficacy. The risks of IVIG include renal dysfunction, transmission of blood‐borne diseases such as hepatitis C and HIV (since IVIG is a pooled blood product) although such risks are low in most countries due to rigorous donor screening and viral inactivation procedure. Severe headaches and febrile reactions which may occur during administration of the infusion have also been described, but inability to tolerate infusions, even 2 g/kg/week is extremely infrequent (Cantu 1995; Berkowitz 2007).

Corticosteroids

Corticosteroids are administered alongside IVIG as a means of supporting the action of IVIG and secondarily reducing headache. Dexamethasone is now avoided as it has been recognised that this drug may cross the fetal blood‐brain barrier and has been associated with oligohydramnios at higher doses (Bussel 1988) and a lack of efficacy at lower doses (Bussel 1996a). At present, prednisone seems to be the corticosteroid of choice for FMAIT (Berkowitz 2006 (stand); Berkowitz 2006 (high)). Although the mothers experience the expected side effects of systemic corticosteroids, clinical experience suggests no abnormalities of children of mothers treated with usual doses of prednisone throughout pregnancy, although premature rupture of amniotic membranes and low birthweight babies have been reported in other settings (Lockshin 1998).

Long‐term outcomes in children treated antenatally for fetomaternal alloimmune thrombocytopenia

The goal of antenatal care is to reduce fetal/neonatal morbidity and mortality but also to allow the child to develop normally in the long term. Follow up of children treated with IVIG or with serial blood transfusions suggests normal neurological and physical development. In particular, those who receive antenatal IVIG appear to have normal immunological development, but for a clinically insignificant elevation of the IgE (Radder 2004). Additionally, children who receive treatment antenatally, with either IVIG and/or corticosteroids, have better long‐term developmental and behavioural outcomes when compared to their older siblings who had not been treated antenatally (Ward 2006).

Why a systematic review?

The majority of the evidence for the antenatal management of FMAIT has come from observational studies, for which there is a substantial body of evidence. Table 1 provides details of the main findings from 27 observational studies. A number of these studies have been referenced in the background for this review (Nicolini 1988; Murphy 1990; Lynch 1992; Zimmermann 1992; Kroll 1994; Murphy 1994; Kaplan 1998; Sainio 1999; Overton 2002). This review will focus on identifying randomised controlled trials of the treatment interventions for FMAIT with the aim of determining the optimal antenatal management.

1. Results from observational studies of antenatal treatment for fetomaternal alloimmune thrombocytopenia with IVIG.

Study	No. fetuses treated	IvIg admin	Steroid admin(daily)	IUT	Second‐line Rx‐nonrespon	No. with ICH	No.plt ct <30x10*9/l	No. deaths	Comments
Daffos 1984	1	Nil	Nil	IUT prior to delivery	Nil	0	0	0
Bussel 1988	7	1 g/kg/week (n = 7)	Dexamethasone 5 mg/day (n = 3), 3 mg/day (n = 2)	Nil	Nil	0	0	0
Mir 1988	1	0.8 mg/kg at 30 weeks, then 0.4 mg/kg/week from 31 weeks	Nil	Nil	Nil	0	0	0
Kaplan 1988	7	0.4 g/kg/5 days weekly from 35 weeks if plt < 20 x 10*9/l (n = 1)		1 IUT 38 weeks. Serial IUT x 3 (n = 1)	Nil	0	0	0
Nicolini 1988	1	Nil	Nil	Weekly IUT from 26 weeks	Nil	0	0	0
Daffos 1988	1	Nil	Prednisolone 10 mg from 23 weeks	Nil	Nil	0	0	0
Murphy 1990	1	Nil	Nil	Weekly IUT from 29 weeks until delivery 34 weeks	Nil	0	0	0
Nicolini 1990	2	Nil	Nil	Weekly IUT from 26 and 27 weeks until delivery	IVIG from week 25, with persistent thrombocytopenia (n = 1)	0	0	0
Kaplan 1991	6	Nil	Prednisone 10 mg from 22 to 26 weeks until delivery at 37 to 38 weeks (n = 4). 25 mg from 21 weeks (n = 2)	Nil	IVIG 1 g/kg/week from 26 weeks until delivery at 38 weeks (n = 1)	0	1	0
Lynch 1992	18	1 g/kg/week from diagnosis of thrombocytopenia at 20 to 22 weeks (one case received 0.5 mg/kg/week)	9 also received dexamethasone 1.5 to 5 mg or prednisolone 10 mg	Nil	Nil	0	2	0
Wenstrom 1992	6	1 g/kg/week	3 received dexamethasone 1.5 mg	Nil	Prednisone 60 mg/day (n = 2)	0	1	0
Marzusch 1992	1	1 g/kg/week from 17 to 34 weeks	Nil	Nil	Failed IUT prior to delivery (n = 1)	0	1	0
Kroll 1994	1	1 g/kg/week from 20 weeks	Nil	Nil	Weekly IUT from time of ICH at 32 weeks until delivery at 35 weeks	1	0	0
Zimmermann 1992	1	Intrauterine transfusion to fetus	Nil	Nil	Single IUT given prior to delivery	0	0	0
Bowman 1992	1	Intrauterine transfusion to fetus	Nil	Nil	IVIG 1 g/kg/week given to mother	0	0	0
Murphy 1994	15	1 g/kg/week beginning at 21 to 34 weeks to 31 to 38 weeks (n = 4) Prednisone 20 mg/day plus IVIG 1 g/kg/week beginning 14 to 20 weeks up to 16 to 33 weeks (n = 3)**	Prednisone 20 mg/day from 16 to 26 weeks to 25 to 36 weeks (n = 2).**	Weekly IUT from 18 to 29 weeks until delivery at 33 to 35 weeks, apart from one case 2 (n = 4) *	Prednisone alone (n = 1) Prednisone + IVIG (n = 2) weekly IUTs were commenced at from 26 to 34 weeks until delivery at 33 to 36 weeks.	1	0	3	*Severely affected babies with initial fetal platelet count < 20 x 10*9/l were treated with fetal platelet transfusions. **Mildly affected fetuses with initial fetal platelet count > 50 x 10*9/l were initially treated either with steroids or IvIgG alone or a combination of both.
Kanhai 1996	19	1 g/kg/week (n = 8) Intrauterine transfusion to fetus (n = 1)		IUT schedule based on previous obstetric history (n = 19)	Prednisone 60 mg/day for 3 weeks (n = 1)	0	0	1 following cordocentesis an ELSCS. 1 fetus: non fatal sagittal sinus thrombosis 1 following cordocentesis an ELSCS. 1 fetus: non fatal sagittal sinus thrombosis
Kaplan 1998	37	From diagnosis of thrombocytopenia (n = 27)1 g/kg/week	Corticosteroids 0.5 mg/kg/day	Nil	Nil	2	Not stated	3
Mackenzie 1999	14	1 g/kg/week (n = 11)	Nil	IUT weekly from 27 weeks until delivery (n = 3)	Nil	0	1 (IUT treatment), 1 (IVIG treatment)	2 related to cordocentesis
Sainio 1999	15	1 g/kg/week from 25 to 32 weeks to 35 to 39 weeks (n = 7) IVIG Plus prednisone (30 mg/day 24 to 29 weeks (n = 1)	Prednisone 20 mg/day from 22 to 35 weeks (n = 1). IVIG ‐1 g/kg for one week only at 35 weeks	Weekly IUT from 25 to 29 weeks to 33 to 36 weeks, apart from 2 cases*	7 cases receiving IVIG +/‐ prednisone required IUT prior to delivery	0	3	0	* 2 pregnancies were managed with a single transfusion: one immediately prior to delivery at 34 weeks, and the other at 28 weeks prior to delivery at 37 weeks.
Dawkins 1999	17	1 g/kg/week (n = 11)	Nil	Nil	Nil	Not stated	2	0
Silver 2000	18	1 g/kg/week (n = 8). Intrauterine transfusion (n = 2)	Nil	Nil	Nil	0	Not stated	0
Radder 2001	56	1 g/kg/week			Weekly IUT	0	25 had platelet count < 50 prior to delivery/IUT	1
Overton 2002	12	1 g/kg/week (n = 3) Intrauterine transfusion in addition (n = 2)	Nil	Weekly IUT from 22 to 24 weeks until delivery	Nil	0	Not stated	2 (1 procedure related)
Lucas 2002	2 (in 1 woman)	0.4 g/kg x 3/week (n = 1) and 0.5 g/kg x 2/week (n = 1)	Nil	Weekly IUT x 5 (n = 1) X1 (n = 1)	Nil	0	0	0
Birchall 2003	56	1 g/kg/week (n = 16)	Prednisone 0.5 mg/kg/day plus IVIG (n = 2 )	Weekly IUT (n = 33)		2 (both before 1st FBS)	Not stated	3 (2 related to FBS)
van den Akker 2006	32	1g/kg from weeks 32 to 38 followed by induction of labour	Nil	Nil	Nil	0	4	0

ELSCS: elective lower section caesarean section 
 FBS: fetal blood sampling 
 IUT: intrauterine transfusion

Objectives

The objective of this review was to determine the optimal antenatal management of FMAIT to prevent fetal and neonatal haemorrhage and death.

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials comparing any intervention with no treatment.

Randomised controlled trials comparing any two interventions, for example a corticosteroid versus IVIG alone or IVIG plus a corticosteroid versus IVIG alone.

Types of participants

Pregnant women with a previous child affected by FMAIT.

Types of interventions

• No treatment.

• IVIG.

• Intrauterine platelet transfusion.

• Corticosteroids.

• Any other treatment aiming to avoid ICH.

• Any other treatment aiming to raise the fetal platelet count.

Types of outcome measures

Primary outcomes

Fetal/neonatal death and survival.

ICH.

Platelet count at birth.

Secondary outcomes

Other bleeding.

Miscarriage. 
 Premature birth.

Long‐term neonatal outcome (disability). 
 Cost of treatment. 
 Acceptability to mothers. 
 Acceptability to carers.

Adverse events for the mother.

1. Steroid induced:

• osteoporosis;

• diabetes;

• hypertension;

• weight gain;

• fluid retention.

2. Imunoglobulin induced:

• transfusion transmitted disease;

• renal dysfunction;

• fever;

• immunoglobulin induced headache.

3. Emergency caesarean section.

Search methods for identification of studies

Electronic searches

We searched the Cochrane Pregnancy and Childbirth Group’s Trials Register by contacting the Trials Search Co‐ordinator (28 February 2011).

The Cochrane Pregnancy and Childbirth Group’s Trials Register is maintained by the Trials Search Co‐ordinator and contains trials identified from:

1. quarterly searches of the Cochrane Central Register of Controlled Trials (CENTRAL);

2. weekly searches of MEDLINE;

3. weekly searches of EMBASE;

4. handsearches of 30 journals and the proceedings of major conferences;

5. weekly current awareness alerts for a further 44 journals plus monthly BioMed Central email alerts.

Details of the search strategies for CENTRAL, MEDLINE and EMBASE, the list of handsearched journals and conference proceedings, and the list of journals reviewed via the current awareness service can be found in the ‘Specialized Register’ section within the editorial information about the Cochrane Pregnancy and Childbirth Group.

Trials identified through the searching activities described above are each assigned to a review topic (or topics). The Trials Search Co‐ordinator searches the register for each review using the topic list rather than keywords. 

For the previous update, we also searched EMBASE (1980 to November 2009) using the strategy given in Appendix 1.

Searching other resources

We checked the citation lists of relevant publications and review articles for further relevant trials.

We did not apply any language restrictions.

Data collection and analysis

Selection for inclusion

Two authors (Rachel Rayment (RR) and Susan J Brunskill (SJB)) initially screened all the titles and abstracts of papers, identified by the review search strategy, for relevance. We excluded only studies which were clearly irrelevant at that stage. RR and SJB independently assessed the full‐text versions of all other studies for inclusion/exclusion using the criteria indicated above. We did not mask the authors and institutions of the trials under consideration. We planned to resolve discrepancies by discussion with a third review author, but this proved unnecessary.

Data extraction and management

Aside from details relating to included trial quality, we extracted the following two groups of data.

1. Trial characteristics: place of publication, date of publication, population characteristics, setting, detailed nature of intervention, detailed nature of comparator, detailed nature of outcomes.

2. Results of included studies in respect of each of the main outcomes indicated in the review question. We recorded as post hoc results encountered during the data extraction process on outcomes we had not defined as important a priori, but felt to be wholly relevant. For continuous outcomes, we recorded mean and standard deviations. For dichotomous outcomes, we recorded the numbers of outcomes in treatment and control groups.

Two authors (RR and SJB) undertook data extraction independently. We extracted data onto trial specific data extraction forms. These forms were not piloted, given that the data extracted for the included studies was found to be identical between the two authors. We sought and received clarification with respect to the numbers included in the final analysis from the principal investigator of one trial (Bussel 1996a). We sought and received additional outcome data from the other three trials whereby means and standard deviations were calculated by the authors following receipt of individual patients data. This data has so far not been forthcoming. One review author (SJB) transcribed the extracted data into the systematic review computer software Review Manager (RevMan 2008).

Assessment of risk of bias in included studies

Two review authors (RR and SJB) independently assessed risk of bias for each study using the criteria outlined in The Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2008). We resolved disagreements by discussion.

(1) Sequence generation (checking for possible selection bias)

For each included study the method used to generate the allocation sequence was described in sufficient detail to allow an assessment of whether it should produce comparable groups. We assessed the methods as:

• adequate (any truly random process, e.g. random number table; computer random number generator);

• inadequate (any non random process, e.g. odd or even date of birth; hospital or clinic record number) or

• unclear.

(2) Allocation concealment (checking for possible selection bias)

For each included study the method used to conceal the allocation sequence was described in sufficient detail so as to determine whether intervention allocation could have been foreseen in advance of, or during recruitment, or changed after assignment. We assessed the methods as:

• adequate (e.g. telephone or central randomisation; consecutively numbered sealed opaque envelopes);

• inadequate (open random allocation; unsealed or non‐opaque envelopes, alternation; date of birth);

• unclear.

(3) Blinding (checking for possible performance bias)

For each included study the methods used, if any, to blind study participants and personnel from knowledge of which intervention a participant received were described. We assessed blinding separately for different outcomes or classes of outcomes. We assessed the methods as:

• adequate, inadequate or unclear for participants;

• adequate, inadequate or unclear for personnel;

• adequate, inadequate or unclear for outcome assessors.

(4) Incomplete outcome data (checking for possible attrition bias through withdrawals, dropouts, protocol deviations)

For each included study, and for each outcome or class of outcomes, the completeness of data including attrition and exclusions from the analysis were described. We have stated whether attrition and exclusions were reported, the numbers included in the analysis at each stage (compared with the total randomised participants), reasons for attrition or exclusion where reported, and whether missing data were balanced across groups. The methods were assessed as:

• adequate;

• inadequate;

• unclear.

(5) Selective reporting bias

For each included study how we investigated the possibility of selective outcome reporting bias and what we found was reported. The methods were assessed as:

• adequate (where it is clear that all of the study’s prespecified outcomes and all expected outcomes of interest to the review have been reported);

• inadequate (where not all the study’s prespecified outcomes have been reported; one or more reported primary outcomes were not prespecified; outcomes of interest are reported incompletely and so cannot be used; study fails to include results of a key outcome that would have been expected to have been reported);

• unclear.

(6) Other sources of bias

For each included study any important concerns we have about other possible sources of bias were described. We have reported assessment of whether each study was free of other problems that could put it at risk of bias as:

• yes;

• no;

• unclear.

(7) Overall risk of bias

Explicit judgements have been made about whether studies are at high risk of bias, according to the criteria given in the Handbook (Higgins 2008). With reference to (1) to (6) above, we assessed the likely magnitude and direction of the bias and whether it was considered likely to impact on the findings.

Measures of treatment effect

Where possible, we analysed the extracted data using the most recent version of RevMan (RevMan 2008). We calculated mean differences (MD) for each of the continuous outcomes where data were presented for the outcome within the trial. We calculated risk ratios (RR) for the dichotomous outcomes. We assessed data using a fixed‐effect model method and 95% confidence intervals (CI).

Two trials did not report standard deviations alongside mean values (Berkowitz 2006 (high); Berkowitz 2006 (stand)). Following email contact, the trialists supplied individual patient data relating to the outcome fetal platelet count at birth and the not prespecified outcome fetal blood counts after initiation of treatment. We have included these data in this review following calculation by one author (SJB) of mean and standard deviation values.

One trial (Berkowitz 2007) reported outcome data by absolute numbers and mean values with corresponding range and 95% CI. For continuous outcomes, where the confidence interval has been reported, we have calculated the standard deviation following instructions in the Handbook (Higgins 2008). For continuous outcomes where only the range was given, we have not calculated the standard deviation (SD) for the purposes of this review as the range is a very unstable value and is not able to provide an accurate estimation of the extent of variation around the mean (Higgins 2008). Additionally, the author of this trial provided individual patient data, which enabled calculation of means and standard deviations for the non‐prespecified outcome 'fetal blood count after initiation of treatment'.

We did not pool summary data across studies due to the diversity in treatment groups and in the doses of interventions administered. We have presented the outcome data extracted from the included trials, and discussed the findings within the review. All authors were involved in drawing conclusions and making specific recommendations for future research.

Assessment of heterogeneity

We did not undertake assessment of clinical and statistical heterogeneity given that it was not appropriate to pool summary outcome data across the included studies. 
 
 We did not undertake any subgroup or sensitivity analyses in this review.

Results

Description of studies

We identified 195 citations. We excluded 173 papers during the initial screening and assessed 22 papers for eligibility on the basis of the full text of the abstract or paper using the criteria indicated above. We did not identify any further trials in the search of the citation lists of relevant publications and review articles.

Trials excluded from the review

We excluded 13 studies (in 14 publications) from the review. See the Characteristics of excluded studies table.

Ongoing trials

One trial is ongoing (MacGregor 2005), seeCharacteristics of ongoing studies for further details.

Trials awaiting assessment

One trial was prematurely closed due to lack of patient accrual in 2009 (Oepkes 2005) and has been classified in Studies awaiting classification because no data have been made available to make a further assessment of eligibility in this review. If data become available from this trial we will incorporate them into a future update of this systematic review. For this trial (Oepkes 2005) there are currently two related publications. SeeCharacteristics of studies awaiting classification.

Trials included in the review

Four trials, in three publications (206 participants) met the inclusion criteria for this systematic review (Bussel 1996a; Berkowitz 2006 (high); Berkowitz 2006 (stand); Berkowitz 2007). SeeCharacteristics of included studies table.

We identified no eligible trials that compared any intervention with no treatment. All included trials compared two interventions in the antenatal management of FMAIT: corticosteroid versus IVIG alone (Berkowitz 2006 (stand)) or IVIG plus a corticosteroid versus IVIG alone (Bussel 1996a; Berkowitz 2006 (high); Berkowitz 2007).

Corticosteroid versus IVIG alone

One trial (Berkowitz 2006 (stand)) compared IVIG 1 g per kg per week alone with prednisone 0.5 mg per kg per day in standard risk women. In this trial, standard risk was defined as no ICH in a previous sibling and pre‐treatment fetal platelet count > 20 x 10*9/l. The primary aim of these trials was to "evaluate the safety and efficacy of risk‐based stratified antenatal treatment for women who had previously delivered a fetus affected by alloimmune thrombocytopenia and were pregnant again with an affected fetus". Treatment was initiated after fetal blood sampling at approximately 20 weeks' gestation.

IVIG plus a corticosteroid versus IVIG alone

One trial (Berkowitz 2006 (high)) compared IVIG 1 g/kg per week plus prednisone 1 mg/kg per day with IVIG 1g /kg per week in high‐risk women. High risk was defined by the trial as either a history of peripartum ICH in an affected sibling or an initial fetal platelet count of less than 20 x 10*9/l. The second trial (Berkowitz 2007) compared IVIG 1 g/kg per week plus prednisone 0.5 mg/kg per day with IVIG 2 g/kg per week alone in standard risk pregnancies this time defined as women without an ICH in a previous pregnancy (but who could potentially have an initial fetal platelet count < 10‐20 x 10*9/l. The primary outcome of this trial was the development of a fetal or neonatal ICH. Two trials (Bussel 1996a; Berkowitz 2006 (high)) initiated treatment after fetal blood sampling at approximately 20 weeks' gestation whilst the third trial (Berkowitz 2007) only assessed the platelet count after a (unspecified) number of weeks of treatment. The third trial (Bussel 1996a) compared the addition of low‐dose dexamethasone to IVIG in the antenatal management of women with FMAIT. The primary aim of this study was to determine whether "a lower dose of daily dexamethasone (1.5 mg daily) would augment the positive effect of IVIG on fetal thrombocytopenia without causing significant toxicity".

Risk of bias in included studies

Trials comparing any two interventions

Corticosteroid versus IVIG alone

In the one trial in this comparison (Berkowitz 2006 (stand)), the method of randomisation and the concealment of allocation were 'adequate', being undertaken by "computer generated random number list balanced by computer blocks at the coordinating center and communicated to the participant institutions by telephone". Clinicians, participants and outcome assessors were not blinded to treatment allocation. The trial had no withdrawals. No power calculation was undertaken in this trial as it was reported that "group‐specific sample sizes could not be adequately powered to identify small differences among the groups of each trial because of the rarity of the disorder". The overall quality of this trial was considered adequate.

IVIG plus a corticosteroid versus IVIG alone

In the three trials in this comparison, the method of randomisation and the concealment of allocation was 'adequate' being undertaken by a "random number list maintained at a separate site by the study statistician" in one trial (Bussel 1996a) and "computer generated random number list balanced by computer blocks at the coordinating center and communicated to the participant institutions by telephone" in two trials (Berkowitz 2006 (high); Berkowitz 2007). Clinicians, participants and outcome assessors were not blinded to treatment allocation. The trials had no withdrawals. An unreported number of fetuses were lost (due to exsanguination) at an early stage in the study in one trial (Bussel 1996a). It would appear from the trial report that these losses occurred prior to randomisation to treatment group. Subsequent data highlighted that the standard approach to these patients was changed near the end of the study to require platelet transfusions to be administered at the times of fetal blood sampling as a result of the increased rate of exsanguinations (Paidas 1995).

In all trials (Bussel 1996a; Berkowitz 2006 (high); Berkowitz 2006 (stand)) all randomised participants were included in the analysis of outcome assessment. However, in one trial (Bussel 1996a), when outcome assessment was stratified by type of IVIG, three participants were excluded from the analysis, for which an explanation is reported in the trial and detailed in the Characteristics of included studies table in this review. In another trial (Berkowitz 2007) one woman was unable to complete therapy due to a severe rash developing at 30 weeks' gestation. IVIG was discontinued and she received prednisone (1 mg/kg/day alone) for four weeks until the infant was delivered at 34 weeks' gestation. The participant's data were included in the trial and hence in this review.

The treatment groups were comparable at baseline for all analysed variables (platelet count and gestational age at initial fetal blood sample, gestational age at birth, and number of previous siblings with ICH) with the exception of mean platelet count of previous sibling. In one trial the mean platelet count of previous sibling differed markedly between treatment groups (Berkowitz 2006 (high)). 
 
 One trial (Bussel 1996a) documented a power calculation. The power calculation was determined from the sample size and dosage of dexamethasone used in an unreported pilot study by the same authors (in the pilot study 33 women were required to find an 80% chance of determining a statistically significant difference at the 0.05 level in platelet increase, with the addition of dexamethasone). The included trial calculated that to have 90% power with 0.05 level of significance to show a greater platelet increase with the addition of dexamethasone, 54 participants overall would be required. Although 54 participants were enrolled in the trial, this sample size would not be large enough to detect any meaningful difference in treatment effect at a 0.05 level of significance. In one other trial (Berkowitz 2006 (high)), it was reported that "group‐specific sample sizes could not be adequately powered to identify small differences among the groups of each trial because of the rarity of the disorder". The overall quality of these trials was considered adequate.

Effects of interventions

Trials comparing any two interventions

All four included trials presented data for this comparison.

Corticosteroid versus IVIG alone

One trial presented data in this comparison (Berkowitz 2006 (stand)). In this trial the corticosteroid used was prednisone.

Primary outcomes

1. Fetal/neonatal death

The relative risk of death was not statistically significantly different between the two treatment groups (RR 0.95; 95% CI 0.06 to 14.13) (Analysis 1.1). Overall there was one fetal/neonatal death in each treatment arm, both at 32 weeks' gestation. One occurred two weeks following fetal blood sampling where the platelet count was 96 x 10*9/l. The second occurred four weeks after fetal blood sampling where the platelet count was 50 x 10*9/l.

1.1. Analysis.

Comparison 1 Corticosteroid vs IVIG alone, Outcome 1 Fetal/neonatal death.

2. ICH

There were two ICHs in this group, but the trial did not report the treatment arm in which the two ICHs occurred. One ICH occurred in a baby born at 38 weeks' gestation with a birth platelet count of 172 x 10*9/l and another occurred in a baby delivered at 28 weeks' gestation (platelet count 68 x 10*9/l) for persistent bradycardia after fetal blood sampling.

3. Platelet count at birth

The difference in mean platelet count at birth between the two groups was not statistically significant (mean difference (MD) ‐36.30 x 10*9/l , 95%CI ‐85.77 to 13.17) (Analysis 1.2). Data were not available for two neonates in the IVIG alone group.

1.2. Analysis.

Comparison 1 Corticosteroid vs IVIG alone, Outcome 2 Platelet count at birth.

Secondary outcomes

1. Other bleeding

'Other bleeding' was not reported as an outcome in this trial.

2. Miscarriage

Miscarriage was not reported as an outcome in this trial.

3. Premature birth

Mean gestational age at birth was not reported as an outcome in this trial.

4. Long term neonatal outcome (disability)

No significant medical problems were reported in any treated infant during the first year of life. The trial did not give long‐term neonatal outcome as an outcome measure.

5. Cost of treatment

The cost of treatment was not reported as an outcome in this trial.

6. Acceptability to mothers

Acceptability to mothers was not reported as an outcome in this trial.

7. Acceptability to carers

Acceptability to carers was not reported as an outcome in this trial.

8. Adverse events for mother

Adverse events for the mother were not reported as an outcome in this trial.

Non‐prespecified outcomes

1. Satisfactory response to therapy

Satisfactory response to therapy was defined as "fetal platelet count > 25,000/mL³ at the time of the second sampling provided that it had increased by > 10,000 mL³ from the value obtained at the first sampling or fetal platelet count > 40,000mL³ provided that it had not decreased by > 10,000mL³ from the previous value". There was no difference between the two treatments (RR 1.14, 95% CI 0.65 to 1.99).(Analysis 1.3).

1.3. Analysis.

Comparison 1 Corticosteroid vs IVIG alone, Outcome 3 Satisfactory response to therapy.

2. Mean change between pre‐treatment and birth platelet count

The mean difference in fetal platelet counts between pre‐treatment and birth was not statistically different (MD ‐40.84 10*9/l, 95% CI ‐88.25 to 6.57) (Analysis 1.4).

1.4. Analysis.

Comparison 1 Corticosteroid vs IVIG alone, Outcome 4 Mean change between pre‐treatment and birth fetal platelet count.

Mean change was calculated for the 19/19 corticosteroid treated patients and 17/19 IVIG treated patients for whom the platelet count at both time points was reported.

3. Mean change in fetal platelet count between first and second fetal blood sampling

Fetal blood sampling (FBS) was undertaken to document fetal thrombocytopenia. The trial provided data to calculate the mean change in fetal platelet count between the first and second fetal blood sampling. The mean change in fetal platelet count between the first and second fetal blood sampling was not statistically different (MD 15.81 10*9/l, 95% CI ‐23.04 to 54.66) (Analysis 1.5).

1.5. Analysis.

Comparison 1 Corticosteroid vs IVIG alone, Outcome 5 Mean change in fetal platelet count between 1st and 2nd fetal blood sampling.

Mean change was calculated for the 16/19 corticosteroid patients and 16/19 IVIG treated patients for whom the platelet count at both time points was reported.

IVIG plus a corticosteroid versus IVIG alone

Three trials presented data in this comparison (Bussel 1996a; Berkowitz 2007; Berkowitz 2006 (high)).The corticosteroid used was prednisone in two trials (Berkowitz 2006 (high); Berkowitz 2007) and dexamethasone in one trial (Bussel 1996a).

Primary outcomes

1. Fetal/neonatal death

This outcome was not reported in these trials.

2. ICH

ICHs were reported in both treatment arms in Berkowitz 2007, but the relative risk of experiencing an ICH was not statistically significant between the two treatment groups (RR 1.03, 95% CI 0.07 to 15.82) (Analysis 2.1). The ICHs were reported subsequent to delivery in one fetus in each treatment group. In the group treated with IVIG and prednisone, the ICH was reported in a fetus born at 32 weeks' gestation with a birth platelet count of 197 x 10*9/l. In the group treated with IVIG alone, the ICH was reported in a fetus born at 37 weeks' gestation with a birth platelet count of 133 x 10*9/l.

2.1. Analysis.

Comparison 2 IVIG with a corticosteroid vs IVIG alone, Outcome 1 Intracranial haemorrhage.

In another trial (of high risk pregnancies) (Berkowitz 2006 (high)), one ICH was observed in the group treated with IVIG alone. The baby was delivered at 35 weeks' gestation with a birth platelet count of 14 x 10*9/l and an ICH was noted at birth. No ICH was reported in the third trial (Bussel 1996a).

3. Platelet count at birth

All trials contained data that was analysable for this outcome, although we did not pool data due to differences in the dose of IVIG used across the trials (Bussel 1996a; Berkowitz 2006 (high); Berkowitz 2007). In all trials the difference in mean platelet count at birth between the two groups was not statistically significant (MD 14.10 x 10*9/l; 95% CI ‐30.26 to 58.46) (Bussel 1996a), (MD 28.00 10*9/l, 95% CI ‐11.47 to 67.47) (Berkowitz 2006 (high)) and (MD ‐35.40 x 10*9/l, 95% CI ‐258.66 to 187.86) (Berkowitz 2007) (Analysis 2.2).

2.2. Analysis.

Comparison 2 IVIG with a corticosteroid vs IVIG alone, Outcome 2 Platelet count at birth.

Secondary outcomes

1. Other bleeding

None of the trials reported "other bleeding" as an outcome.

2. Miscarriage

None of the trials reported miscarriage as an outcome.

3. Premature birth

Two trials contained data that was analysable for this outcome, although we did not pool data due to differences in the dose of IVIG used across the trials (Bussel 1996a; Berkowitz 2007). The difference in mean gestational age at birth between the two groups was not statistically significant (MD ‐0.20 weeks; 95% CI ‐5.71 to 5.31 (Berkowitz 2007) and MD ‐0.50 weeks; 95% CI ‐2.69 to 1.69 (Bussel 1996a) (Analysis 2.3)).

2.3. Analysis.

Comparison 2 IVIG with a corticosteroid vs IVIG alone, Outcome 3 Premature birth.

4. Long term neonatal outcome (disability)

No significant medical problems were reported in any treated infant during the first year of life. The trials did not give long‐term neonatal outcome as an outcome measure.

5. Cost of treatment

The cost of treatment was not reported as an outcome in these trials.

6. Acceptability to mothers

Acceptability to mothers was not reported as an outcome in these trials.

7. Acceptability to carers

Acceptability to carers was not reported as an outcome in these trials.

8. Adverse events for mother

Adverse events for the mother were not reported as an outcome in these trials.

Non‐prespecified outcomes

1. Satisfactory response to therapy

The number of women with a satisfactory response to therapy was reported in the trial of high‐risk patients (Berkowitz 2006 (high)). Data for this outcome measure were based on the first FBS following the introduction of treatment. There was a higher chance of a a satisfactory response to treatment in the IVIG plus corticosteroid arm (RR 2.54, 95% CI 1.37 to 4.72) (Analysis 2.4).

2.4. Analysis.

Comparison 2 IVIG with a corticosteroid vs IVIG alone, Outcome 4 Satisfactory response to therapy.

2. Mean change in fetal platelet count between pre‐treatment and fetal birth platelet count

Data were available to calculate mean change in fetal platelet count between pre‐treatment and birth fetal platelet count in two trials (Berkowitz 2006 (high); Berkowitz 2007). The mean difference between the treatment arms was not statistically significant in either trial (MD 10.83, 95% CI ‐28.99 to 50.65) (Berkowitz 2006 (high) and MD ‐21.95 10*9/l, 95% CI ‐57.96 to 14.06 (Berkowitz 2007) (Analysis 2.5)).

2.5. Analysis.

Comparison 2 IVIG with a corticosteroid vs IVIG alone, Outcome 5 Mean change between pre‐treatment and birth fetal platelet count.

Mean change was calculated for those patients for whom the fetal platelet count at both time points was reported: 17/18 IVIG plus corticosteroid treated patients and 18/20 IVIG (Berkowitz 2006 (high)) and 28/36 IVIG and corticosteroid treated patients and 30/37 IVIG alone (Berkowitz 2007).

3. Mean change in fetal platelet count between first and second fetal blood sampling (FBS)

FBS was undertaken to document fetal thrombocytopenia. Two trials provided data to calculate the mean change in fetal platelet count between the first and second fetal blood sampling (Bussel 1996a; Berkowitz 2006 (high)). The mean difference in fetal platelet count was statistically significant in favour of IVIG plus corticosteroid in one trial (MD 42.0 x 10*9/l, 95% CI 18.91 to 65.09) (Berkowitz 2006 (high)) but not statistically significant between the two treatment groups in the second trial (MD ‐3.50 x 10*9/l, 95% CI ‐24.62 to 17.62) (Bussel 1996a) (Analysis 2.6).

2.6. Analysis.

Comparison 2 IVIG with a corticosteroid vs IVIG alone, Outcome 6 Mean change in fetal platelet count at 2nd from 1st fetal blood sampling.

Mean change was calculated for those patients for whom the fetal platelet count was reported at both time points for all treated patients in one trial (Bussel 1996a) and 18/18 IVIG and corticosteroid treated patients and 18/20 IVIG treated patients in another trial (Berkowitz 2006 (high)).

4. Mean change in fetal platelet count at birth from first fetal blood sampling

One trial provided data to calculate the mean change in fetal platelet count at birth from first fetal blood sampling (Bussel 1996a). The mean change was not statistically significant between the two treatment groups (MD 24.4 x 10*9/l; 95% CI ‐14.17 to 62.97) (Analysis 2.7).

2.7. Analysis.

Comparison 2 IVIG with a corticosteroid vs IVIG alone, Outcome 7 Mean change in fetal platelet count at birth from first fetal blood sampling.

5. Number of ICHs in previous sibling

The number of ICHs in previous (untreated) siblings was measured in one trial (Berkowitz 2006 (high)). The relative risk of a previous sibling having experienced an ICH was not significant between the treatment arms (RR 0.83, 95% CI 0.21 to 3.24) (Analysis 2.8). Interestingly, a high proportion of fetuses (5/7, 71%) whose sibling had sustained a peripartum ICH had initial platelet counts at first blood sampling of < 10 x 10*9/l.

2.8. Analysis.

Comparison 2 IVIG with a corticosteroid vs IVIG alone, Outcome 8 Number of ICHs in previous sibling.

6. Adverse outcomes associated with FBS

One trial provided analysable data for this outcome (Berkowitz 2007). There was no statistical difference in the probability of experiencing an adverse outcome following FBS between these treatment arms (RR 0.97, 95% CI 0.41 to 6.58) (Analysis 2.9). Of the four reported adverse outcomes all involved emergency caesarean section of which three were preterm births within 24 hours of the FBS procedure (two at 33 weeks with birth platelet counts of 104 x 10*9/l and 53 x 10*9/l and one at 32 weeks with a birth platelet count of 52 x 10*9/l).

2.9. Analysis.

Comparison 2 IVIG with a corticosteroid vs IVIG alone, Outcome 9 Adverse outcomes associated with FBS.

7. Adverse outcomes associated with treatment

One trial provided analysable data for this outcome (Berkowitz 2007). There was no statistical difference in the probability of experiencing an adverse outcome associated with treatment between these treatment arms (RR 1.06, 95% CI 0.88 to 1.27) (Analysis 2.10). The full list of adverse outcomes experienced is detailed in Table 2.

2.10. Analysis.

Comparison 2 IVIG with a corticosteroid vs IVIG alone, Outcome 10 Adverse events associated with treatment.

2. Complications of treatment ‐ Berkowitz 2007.

Complication	IVIG alone (n = 22)	IVIG + prednisone (n =28)	Second‐line (n = 13)
Gastrointestinal diabetes mellitus	0	4 (14)	3 (23)
Fluid retention	1 (5)	9 (32)	4 (31)
Mood swings	2 (9)	7 (25)	5 (38)
Moderate to severe fatigue	7 (32)	5 (18)	6 (46)
Joint pain	3 (14)	6 (21)	3 (23)
Moderate to severe headache	9 (41)	10 (36)	6 (46)
Rash	2 (9)	3 (11)	1 (8)
Insomnia/jitteriness	0	2 (7)	4 (31)
TOTAL number of complications	24	46	32
Data on complications experienced were not obtained from 6 women receiving IVIG alone and 4 women receiving IVIG and prednisone

Complications of treatment data presented as number of women (%) of events.

IVIG: intravenous immunoglobulin

Discussion

The antenatal treatment of FMAIT has developed over the past 25 years, largely based on case series, and the optimal management remains unclear. The objective of this review was to determine the optimal antenatal management of FMAIT in the prevention of fetal and neonatal haemorrhage and death based on the evidence from randomised trials.

Four randomised controlled trials (RCTs) met the eligibility criteria for this systematic review (Bussel 1996a; Berkowitz 2006 (high); Berkowitz 2006 (stand); Berkowitz 2007). These studies have all assumed the efficacy of IVIG, based on historical observational data. They have also all assumed that the platelet count is a marker of disease/outcome. Unfortunately, no information is available on the clinical outcome of in‐utero transfusion or even IVIG alone compared to no treatment in the context of a RCT.

1) IVIG alone versus steroids alone

In the Berkowitz 2006 (stand) trial, "standard" treatment with IVIG was compared with long‐term administration of prednisone 0.5 mg per kg per day. No significant difference was demonstrated in efficacy between these treatments for women with a fetus at risk of FMAIT, where the previously affected sibling did not suffer ICH and the birth platelet count was > 20 x 10*9/l. Thus, for standard risk pregnancies either IVIG or prednisone may be suitable, although there is a substantial failure rate with each (Berkowitz 2006 (stand)). At present the only available antenatal method by which to confirm response to treatment is by FBS, which has a significant complication rate. Two trials in this review reported complications of FBS jointly (Berkowitz 2006 (high); Berkowitz 2006 (stand)); there were 11 serious complications out of a total of 175 (6%) FBS procedures. The risks of FBS at different stages of pregnancy must be balanced against the benefit of detecting severe thrombocytopenia which can be treated to avoid perinatal ICH. The most likely serious adverse outcome with FBS and platelet transfusion at 36 weeks' gestation is early onset of labour, whereas FBS before 28 weeks may be complicated by fetal death due to bleeding or cardiac dysrhythmias. Moreover, when exploring the instances of ICH associated with treatment failure, the grade of ICH (grade 1 versus > grade 1) should be considered. Although this trial (Berkowitz 2006 (stand)) did not report maternal complications associated with treatment, they are an increasingly important consideration when determining which treatment modality to use in “standard risk” women, and they should be reported in any new trials.

2) IVIG alone versus IVIG plus steroids

It was reported (Bussel 1996a) that there was an improvement in the fetal platelet count following maternal treatment with both IVIG and IVIG and dexamethasone, and that the platelet count of these treated fetuses at birth was increased in comparison to the platelet count of their respective sibling at birth. However it should be noted that 'second‐line' therapy was non‐randomly administered only to non‐responders, so birth platelet counts were not always valid indicators of the effect of the initial treatment in these cases.

The findings of this trial support findings from other studies suggesting an improved outcome for ICH with increased fetal platelet count (Bussel 1997; Birchall 2003). Indeed, the sibling platelet count and clinical history remain the best surrogate predictive markers for disease at the present time (Radder 2003). Such findings have been suggested in many observational studies, details of which are summarised in additional Table 1.

However, the results from this trial and subsequent conclusions reached by the trials' authors should be read with caution. A number of important clinical outcomes were not reported by the trial authors: fetal death; incidence of ICH; fetal loss as a result of fetal blood sampling; financial cost of treatment; adverse events; and acceptability of the treatment options to the mother. The sample size was small and hence had limited power to detect small but important differences especially in outcomes of survival, fetal loss and occurrence of ICH. For this trial, power calculations to determine study sample size were based on platelet count increments. It may not be possible to design appropriately powered studies for this population that would measure clinically relevant outcomes such as bleeding risk, since ethical constraints would clearly make endpoints of severe bleeding or mortality more difficult to justify. Perhaps alternative outcome measures should be developed based on improved understanding of the underlying mechanism of disease, and/or relying on non‐invasive techniques.

The more recent trial (Berkowitz 2006 (high)) suggests a superior response with combined therapy where there was a history of peripartum ICH in an affected sibling or an initial fetal platelet count of less than 20 x 10*9/l, but their results should also be interpreted with caution due to the low numbers (a problem common to all "rare" disorders) and the differences in sibling platelet count in the two groups. In this trial, one ICH occurred due to failure of treatment in the group in which the intention to treat was with IVIG alone. However, treatment had been empirically intensified by the addition of prednisone from 28 weeks, suggesting that, in this case, neither treatment regimen was adequate. Thirteen out of 20 (65%) women receiving IVIG alone required intensification of treatment (by the addition of prednisone), compared with only two of the 18 (11%) women receiving combination treatment (by increasing the dose of IVIG) suggesting, in this high‐risk group, the latter to be superior. This effect was most marked in cases where the pre‐treatment platelet count was < 10 x 10*9/l (82% for those receiving IVIG alone and 18% of those receiving IVIG plus prednisone required intensification of treatment). Information regarding the number of babies born with a platelet count of less than 50 x 10*9/l in each arm (with or without intensification) was not given.

The design of the third trial (Berkowitz 2007) was based on the findings of the Berkowitz 2006 (high) trial where it was noted that the effect of 1g per kg per week of IVIG alone was inferior to combination therapy with prednisone in women with babies at high risk of severe thrombocytopenia and consequent haemorrhage. This trial recruited standard risk women and excluded those with a history of babies affected by an ICH. Complications of FBS were minimised by performing a single procedure at 32 weeks' gestation, and intensifying treatment for non‐responders. Although, in terms of the primary outcome of ICH, both treatments were effective, it was interesting to note that 5/37 (13.5%) of babies in the group treated with the increased dose IVIG alone (group A) and 4/36 (11.1%) in the group treated with combination therapy (group B) had a birth platelet count < 50 x 10*9/l. i.e. that each treatment had a significant failure rate. Equally, 10/37 (27%) of women in group A and 6/36 (17%) of those in group B needed second‐line therapy. Even with this treatment, one baby in each group had a birth platelet count less than 30 x 10*9/l.

The difficulty highlighted in these studies is that of predicting which cases require intensive treatment and monitoring and those which require treatment with modest doses of IVIG and/or prednisone, the optimal dose of which has yet to be determined. Although sibling history is a valuable tool in assessing risk of ICH, it must be remembered that, in most cases, the sibling was the first affected case, and therefore untreated for the duration of the pregnancy. The sibling history is therefore not predictive of response to treatment. However, it does appear, from the Berkowitz 2006 (high) trial that severe thrombocytopenia at 20 weeks (< 20 x 10*9/l) may be predictive of poor response to therapy, and that these cases should be monitored more intensively with FBS and platelet transfusion, despite the inherent risks. This is in agreement with a European survey of antenatal management of FMAIT (Birchall 2003).

It is not clear from the report (Berkowitz 2006 (high)) how many women were delivered by elective caesarean section, but one of the author's recommendations was to only permit vaginal delivery if the platelet count was greater than 100 x 10*9/l at 32 weeks. Although there is no evidence from RCTs regarding the mode of delivery in affected pregnancies, an observational study in pregnancies with "standard risk" FMAIT has suggested, albeit in a small number of cases (23), that vaginal delivery is not associated with increased risk of ICH (platelet count range 12 ‐252 x 10*9/l) (van den Akker 2006). An alternative approach in such cases would be a further FBS and platelet transfusion at 36 to 37 weeks, and a planned induction of labour within a few days of transfusion with a "minimally invasive" vaginal delivery (avoiding the use of high forceps or ventouse delivery, and the use of fetal scalp electrodes) and so avoid the increased maternal risks of a caesarean section.

There are some important issues concerning the 'standard' therapy in the four RCTs included in this review. In spite of the drawbacks of observational studies, much of current clinical practice is based on their findings. In the absence of RCTs, the efficacy of IVIG alone and IUT alone in the antenatal management of FMAIT cannot be fully assessed. However, the findings from observational data have provided important data which have influenced clinical practice in this clinical area: that the platelet count declines throughout the course of an affected pregnancy (Bussel 1988; Kaplan 1998) and that there may be an association between ICH and a platelet count of less than 20 x 10*9/l (for the relationship of fetal platelet count with gestational age, see Birchall 2003). Such findings have been suggested in many observational studies (Daffos 1984; Daffos 1988; Mir 1988; Nicolini 1988; Kaplan 1991; Bowman 1992; Marzusch 1992; Wenstrom 1992; Bussel 1997; Mackenzie 1999Lucas 2002), as summarised in Table 1.

IVIG is now commonplace as first‐line treatment in the antenatal management of FMAIT, with intrauterine transfusions of platelets reserved for refractory cases, due to the risk of fetal haemorrhage and/or fetal loss (Daffos 1985; Murphy 1994; Paidas 1995; Silver 2000). A trial comparing IVIG with serial intrauterine platelet transfusions of platelets is unlikely to be carried out, as it is believed that both interventions have found their rightful place. All interventions currently in use have recognised side effects and long‐term implications. For IVIG they include transfusion transmitted disease, headaches and febrile reactions in the mother and bleeding and death for the fetus. Corticosteroids can cause osteoporosis and hypertension in the mother whilst intrauterine platelet transfusion can cause bleeding and fetal loss. It is, therefore, important to have robust data regarding both long‐ and short‐term outcomes for both the mothers and their children. Also, due to the significant costs involved, it is imperative to optimise the cost benefit for interventions considered as management options for fetomaternal alloimmune thrombocytopenia.

From the data published to date, the ideal antenatal treatment of FMAIT is still unclear. However, we have gained some knowledge upon which to base further studies and management strategies in the interim. From these trials it appears that stratification of patients according to the sibling history, and tailoring of treatment accordingly is safe, allowing a less intensive approach to treatment and monitoring where the sibling platelet count was > 20 x 10*9/l and there was no ICH. Where the sibling platelet count was lower, or there was a history of ICH more intensive treatment is still associated with significant failure but the combination of IVIG and prednisone is superior to IVIG alone.

It is important to note that the incidence of ICH, the most important marker of treatment efficacy, can not be reduced to zero in any study where babies are delivered prematurely (which is an independent risk factor for ICH) or by emergency means. It was of interest that three of the four instances of ICH occurred in association with a normal platelet count. In all three cases the baby was premature and/or delivered by emergency caesarean section. Studies powered to demonstrate a reduction in ICH require multi‐centre recruitment since it is estimated that the incidence of ICH in an at‐risk pregnancy where the sibling suffered an ICH is 72% (Radder 2003) and where there was no ICH it is 7%.

Authors' conclusions

Implications for practice.

1. There is no evidence that, when added to IVIG, antenatally administered dexamethasone improves platelet count at birth. Since dexamethasone (albeit at higher doses than that used in Bussel 1996a) has been reported as causing fetal oligohydramnios and this trial also reported significant effects on the mother, dexamethasone is not recommended for the antenatal treatment of FMAIT, especially since alternative corticosteroids are available.

2. IVIG or prednisone can be used as first line treatment for standard‐risk FMAIT, where there was no peripartum haemorrhage in an affected sibling and the pre‐treatment fetal platelet count (if performed) is > 20 x 10*9/l. However, the optimal dose of both prednisone and IVIG has not been established.

3. IVIG in combination with prednisone is more effective in raising the fetal platelet count than IVIG alone in high‐risk pregnancies, where the pre‐treatment fetal platelet count < 20 x 10*9/l or the affected sibling sustained a peripartum ICH. The optimal timing of administration and the dose of prednisone and IVIG is unclear, but studies demonstrating efficacy initiated treatment at 20‐26 weeks.

4. High‐risk pregnancies were defined as those where an ICH occurred peripartum in the sibling. No data are currently available where the sibling ICH occurred in‐utero or neonatally. It is possible that this very high risk group requires dual therapy and we await publication of the trial investigating the management of this group of women (MacGregor 2005).

Implications for research.

In view of the rarity of this condition, international collaborations are needed in order to accrue meaningful data and improve patient management. Future studies will need to include outcomes that can be appropriately assessed and compared across other similar studies. All outcome measures specified in 'Types of outcomes' and identified post‐hoc in 'non‐prespecified outcomes' are important measures for future trials.

Areas of interest for future research include:

1. Dose of IVIG

The ideal study would be one of IVIG compared with placebo since this has not been performed. However, despite their limitations, observational data strongly support the role of IVIG in modulating the course of the disease in at‐risk pregnancies and many would argue that such a trial would be unethical. It would certainly be difficult to recruit patients. Instead, the timing and dosage of treatment needs evaluation in order to avoid the possibility of excessive treatment. The standard of 1 g per kg per week has been derived empirically from anecdotal reports of failure of lower doses (Lynch 1992). The Berkowitz 2007 trial compared IVIG 2 g per kg per week and IVIG 1 g per kg per week PLUS prednisone. Unfortunately, no trial has been performed to compare dose regimens or to determine optimal treatment for different risk groups based on sibling history. Increasing the dose of IVIG in standard risk women from 1 to 2 g per kg per week has substantial implications, both in terms of personal issues of childcare and work for the mother, as well as the high cost per individual treated. The No IntraCranial Haemorrhage (NOIICH) trial aimed to compare different doses of IVIG ‐ low and standard but has been closed due to a lack of patient accrual (Oepkes 2005a).

It is not clear what aspect of IVIG's action is most important in inhibiting disease activity. If it is in the blockade of placental antibody transfer then levels similar to those required in rhesus haemolytic disease of the newborn (20 g/l) may be optimal (Urbaniak 1999). Perhaps this should be confirmed for FMAIT with treatment schedules adjusted for individual participants.

Laboratory‐based studies should focus on the action of IVIG on dendritic cells, T‐cell and maternal antibody production and transfer, to determine relative importance of each function, and the optimal concentrations required for each.

2. Timing of initial treatment

It appears that duration and amount of treatment should be titrated according to risk. The issue of when IVIG should be commenced for different risk groups remains unsolved.

3. Monitoring response by fetal blood sampling

The risks associated with FBS pose many questions.

• Should there be FBS to monitor treatment and if so, how often?

• Should there be a final FBS and platelet transfusion prior to birth?

• Should all high‐risk mothers have FBS at 20 weeks?

• Should FBS be preceded by empirically administered IVIG? If so, what dose should be given and for how long?

• Should low‐risk mothers have sampling at all?

4. Timing and mode of birth

There is no evidence that early birth or caesarean section increases or reduces morbidity/mortality in FMAIT. Outside pregnancies affected by FMAIT, emergency caesarean section is known to be associated with an increased risk of ICH in neonates, compared with that of spontaneous vaginal delivery or elective caesarean section (Ljung 2008). Should the timing of delivery and outcomes of caesarean section efficacy and safety be incorporated into future trials? Should this be stratified by the last fetal platelet count?

5. Laboratory measures to define pregnancies with a high risk of ICH and so identify pregnancies to be treated

The mechanism of disease is poorly understood and limits our management approach in this field.

• What is the role of T‐cells in the pathogenesis of disease? Can their numbers, phenotype or activation state predict clinical disease?

• Could a functional antibody assay be developed which more accurately correlates with disease?

• Are there parameters of the fetus or baby's platelets that might predict the risk of bleeding or the recovery of a normal platelet count?

6. Management of non‐responders

Can non‐responders be predicted by non‐invasive means? Clinical trials should be performed in parallel with laboratory scientific research, with the aim of improving understanding and monitoring of the antibody response in FMAIT.

What is the role of corticosteroids, elevated doses of IVIG and intrauterine platelet transfusion, or a combination of all three? Do alternative immunosuppressive agents have a role to play in these, or even in standard‐risk cases?

7. Antenatal screening

Advances in the laboratory diagnosis and antenatal management of FMAIT have drawn attention to the fact that the first affected fetus/neonate is usually only recognised after bleeding has occurred or severe thrombocytopenia detected by chance. This raises the question of whether routine screening for FMAIT should be considered. However, there are significant shortcomings in the knowledge about FMAIT necessary for the introduction of an antenatal screening programme. These include:

a) the clinical outcome of first affected pregnancies;

b) the identification of laboratory measures predictive of severe disease where antenatal intervention might be justified; and

c) the optimal approach for the antenatal management of pregnant women with HPA antibodies but with no previous history of affected pregnancies.

What's new

Date	Event	Description
10 March 2011	New citation required but conclusions have not changed	A new author helped update the review.
10 March 2011	New search has been performed	Search updated. No new trial reports identified.

History

Protocol first published: Issue 2, 2003
 Review first published: Issue 1, 2005

Date	Event	Description
16 March 2010	New search has been performed	Search updated. Three new studies included (Berkowitz 2006 (high); Berkowitz 2006 (stand); Berkowitz 2007); two studies excluded (Bussel 2005; van den Akker 2006) and one ongoing study identified. One study is awaiting classification (Oepkes 2005). No major changes to results or conclusions.
31 October 2008	Amended	Converted to new review format.

Appendices

Appendix 1. EMBASE search strategy

1. NAITP.mp. 
 2. fetomaternal alloimmune thrombocytopenia.mp. 
 3. NAIT.mp. 
 4. newborn alloimmune thrombocytopenia/ or neonatal alloimmune thrombocytopenia.mp. 
 5. neonatal hemorrhage.mp. 
 6. fetus disease.mp. or Fetus Disease/ 
 7. Mother Fetus Relationship/ 
 8. or/1‐7 
 9. thrombocyte transfusion.mp. or Thrombocyte Transfusion/ 
 10. antenatal management.mp. 
 11. fetomaternal transfusion.mp. or Fetomaternal Transfusion/ 
 12. fetoplacental unit.mp. or Fetoplacental Unit/ 
 13. prenatal care.mp. or Prenatal Care/ 
 14. placental transfer.mp. or Placental Transfer/ 
 15. or/9‐14 
 16. (feto‐maternal or foeto‐maternal).mp. 
 17. (feto maternal or foeto maternal).mp. 
 18. or/16‐17 
 19. 8 or 18 
 20. 19 and 15 
 21. random$.ti,ab. 
 22. factorial$.ti,ab. 
 23. (crossover$ or cross over$ or cross‐over$).ti,ab. 
 24. placebo$.ti,ab. 
 25. (doubl$ adj blind$).ti,ab. 
 26. (singl$ adj blind$).ti,ab. 
 27. assign$.ti,ab. 
 28. allocat$.ti,ab. 
 29. volunteer$.ti,ab. 
 30. CROSSOVER PROCEDURE.sh. 
 31. DOUBLE‐BLIND PROCEDURE.sh. 
 32. RANDOMIZED CONTROLLED TRIAL.sh. 
 33. SINGLE‐BLIND PROCEDURE.sh. 
 34. or/21‐33 
 35. ANIMAL/ or NONHUMAN/ or ANIMAL EXPERIMENT/ 
 36. HUMAN/ 
 37. 36 and 35 
 38. 35 not 37 
 39. 34 not 38 
 40. 20 and 39

Data and analyses

Comparison 1. Corticosteroid vs IVIG alone.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Fetal/neonatal death	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
2 Platelet count at birth	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
3 Satisfactory response to therapy	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
4 Mean change between pre‐treatment and birth fetal platelet count	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
5 Mean change in fetal platelet count between 1st and 2nd fetal blood sampling	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only

Comparison 2. IVIG with a corticosteroid vs IVIG alone.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Intracranial haemorrhage	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.1 IVIG 1g/kg/week plus prednisone 0.5 mg/kg/day versus IVIG 2g/kg/week	1	73	Risk Ratio (M‐H, Fixed, 95% CI)	1.03 [0.07, 15.82]
2 Platelet count at birth	3		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
2.1 IVIG 1g/kg/week plus dexamethasone 1.5 mg/day versus IVIG 1 g/kg/week	1	54	Mean Difference (IV, Fixed, 95% CI)	14.10 [‐30.26, 58.46]
2.2 1g/kg/week plus prednisone 1.0 mg/kg/day versus IVIG 1g/kg/week	1	36	Mean Difference (IV, Fixed, 95% CI)	28.00 [‐11.47, 67.47]
2.3 IVIG 1g/kg/week plus prednisone 0.5 mg/kg/day versus IVIG 2g/kg/week	1	73	Mean Difference (IV, Fixed, 95% CI)	‐35.40 [‐258.66, 187.86]
3 Premature birth	2		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
3.1 IVIG 1g/kg/week plus dexamethasone 1.5 mg/day versus IVIG 1 g/kg/week	1	54	Mean Difference (IV, Fixed, 95% CI)	‐0.5 [‐2.69, 1.69]
3.2 IVIG 1g/kg/week plus prednisone 0.5 mg/kg/day versus IVIG 2g/kg/week	1	73	Mean Difference (IV, Fixed, 95% CI)	‐0.20 [‐5.71, 5.31]
4 Satisfactory response to therapy	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
4.1 IVIG 1g/kg/week plus prednisone 1 mg/kg/day versus IVIG 1 g/kg/week alone	1	38	Risk Ratio (M‐H, Fixed, 95% CI)	2.54 [1.37, 4.72]
5 Mean change between pre‐treatment and birth fetal platelet count	2		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
5.1 IVIG 1g/kg/week plus prednisone 1 mg/kg/day versus IVIG 1 g/kg/week alone	1	35	Mean Difference (IV, Fixed, 95% CI)	10.83 [‐28.99, 50.65]
5.2 IVIG 1g/kg/week plus prednisone 0.5 mg/kg/day versus IVIG 2g/kg/week	1	58	Mean Difference (IV, Fixed, 95% CI)	‐21.95 [‐57.96, 14.06]
6 Mean change in fetal platelet count at 2nd from 1st fetal blood sampling	2		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
6.1 IVIG 1g/kg/week plus dexamethasone 1.5 mg/day versus IVIG 1 g/kg/week	1	54	Mean Difference (IV, Fixed, 95% CI)	‐3.5 [‐24.62, 17.62]
6.2 IVIG 1g/kg/week plus prednisone 1 mg/kg/day versus IVIG 1 g/kg/week alone	1	36	Mean Difference (IV, Fixed, 95% CI)	42.0 [18.91, 65.09]
7 Mean change in fetal platelet count at birth from first fetal blood sampling	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
7.1 IVIG 1g/kg/week plus dexamethasone 1.5 mg/day versus IVIG 1 g/kg/week	1	54	Mean Difference (IV, Fixed, 95% CI)	24.40 [‐14.17, 62.97]
8 Number of ICHs in previous sibling	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
8.1 IVIG 1g/kg/week plus prednisone 1 mg/kg/day versus IVIG 1 g/kg/week alone	1	40	Risk Ratio (M‐H, Fixed, 95% CI)	0.83 [0.21, 3.24]
9 Adverse outcomes associated with FBS	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
9.1 IVIG 1g/kg/week plus prednisone 0.5 mg/kg/day versus IVIG 2g/kg/week	1	79	Risk Ratio (M‐H, Fixed, 95% CI)	0.98 [0.14, 6.58]
10 Adverse events associated with treatment	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
10.1 IVIG 1g/kg/week plus prednisone 0.5 mg/kg/day versus IVIG 2g/kg/week	1	73	Risk Ratio (M‐H, Fixed, 95% CI)	1.06 [0.88, 1.27]

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Berkowitz 2006 (high).

Methods	Generation of allocation sequences: computer generated random number list balanced by computed blocks. Randomisation was performed at the co‐ordinating centre and communicated to the participating institutions by telephone. Timing of randomisation: not stated. Power calculation undertaken: yes. Blinding: not stated (B). Number of centres: 42 institutions across the United States Timing and duration of study: enrolment May 1994 to April 2001 Follow up: not stated Source of funding: part funded by an unrestricted clinical research grant from Alpha Therapeutics.
Participants	Number of participants: 40 (19/21) pregnancies randomised. 38 (18/20) pregnancies evaluable and analysed; 0 excluded. Where the father was homozygous, the fetus was presumed to be affected; where the father was heterozygous, the fetal genotype was determined by typing of amniocyte DNA. Definition high risk: either a sibling with an ICH that occurred during the peripartum period or an initial fetal platelet count < 20 x 10*9/l. Maternal age: treatment: 27.5 years; control: 30.3 years.
Interventions	High risk: randomisation between IVIG 1 g per kg per week with prednisone 1 mg per kg per day and IVIG 1 g per kg per week. Treatment initiated after fetal blood sampling at 20 weeks' gestation. Trial does not report preparations of IVIG used.
Outcomes	ICH. Fetal platelet count. Neonatal platelet count. Fetal death. Complications of fetal blood sampling.
Notes	Fetal blood sampling undertaken at 3 to 8 week intervals. If fetal blood sampling was not possible treatment was intensified empirically. These women were not included in the analysis of response to initial therapy.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Adequate. Sequence generation was undertaken by computer generated random number list balanced by computed blocks. Randomisation was performed at the co‐ordinating centre and communicated to the participating institutions by telephone.
Allocation concealment (selection bias)	Low risk	Adequate. Allocation concealment was undertaken by computer generated random number list balanced by computed blocks. Randomisation was performed at the co‐ordinating centre and communicated to the participating institutions by telephone.
Blinding (performance bias and detection bias) All outcomes	High risk	Clinicians, patients and outcome assessors were not blinded to treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	40 women were randomised in this trial. 19 women were randomised to receive IVIG plus prednisone. 21 women were randomised to receive IVIG alone. The paper reports that there were 18 (of 19) evaluable women in the IVIG plus prednisone arm and 20 (of 21) evaluable women in the IVIG alone arm. No details are provided as to whether these women were lost due to attrition or exclusion from the trial, the outcome for these 2 women or at what stage they left the trial. It is also unclear whether any of their outcome data were reported in the paper.
Selective reporting (reporting bias)	Low risk	All outcomes mentioned in Methods are reported in Results.
Other bias	Unclear risk	None reported. It would be difficult to rule out.

Berkowitz 2006 (stand).

Methods	Generation of allocation sequences: computer generated random number list balanced by computed blocks. Randomisation was performed at the co‐ordinating centre and communicated to the participating institutions by telephone. Timing of randomisation: not stated. Power calculation undertaken: yes. Blinding: not stated (B). Number of centres: 42 institutions across the United States. Timing and duration of study: enrolment May 1994 to April 2001. Follow up: not stated. Source of funding: part funded by an unrestricted clinical research grant from Alpha Therapeutics.
Participants	Number of participants: 39 (19/20) pregnancies randomised. 38 (19/19) evaluable for analysis. 0 excluded. Where the father was homozygous, the fetus was presumed to be affected; where the father was heterozygous, the fetal genotype was determined by typing of amniocyte DNA. Definition standard risk: no sibling with an ICH and in‐utero platelet count > 20 x 10*9/l but < 100 x 10*9/l. Maternal age: treatment 30.2 years; control 31 years.
Interventions	Standard risk: randomisation between IVIG 1 g per kg per week and prednisone 0.5 mg per kg per day. Treatment initiated after fetal blood sampling as close to 20 weeks' gestation as possible. Trial does not report preparations of IVIG used.
Outcomes	ICH. Fetal platelet count. Neonatal platelet count. Fetal death. Complications of fetal blood sampling.
Notes	Fetal blood sampling undertaken at 3 to 8 week intervals. If fetal blood sampling was not possible treatment was intensified empirically. These women were not included in the analysis of response to initial therapy.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Adequate. Sequence generation was undertaken by computer generated random number list balanced by computed blocks. Randomisation was performed at the co‐ordinating centre and communicated to the participating institutions by telephone.
Allocation concealment (selection bias)	Low risk	Adequate. Allocation concealment was undertaken by computer generated random number list balanced by computed blocks. Randomisation was performed at the co‐ordinating centre and communicated to the participating institutions by telephone.
Blinding (performance bias and detection bias) All outcomes	High risk	Clinicians, patients and outcome assessors were not blinded to treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	39 women were randomised in this trial. 19 women were randomised to receive IVIG. 20 women were randomised to receive prednisone. The paper reports that there were 19 (of 19) evaluable women in the IVIG arm and 19 (of 20) evaluable women in the IVIG alone arm. No details are provided as to whether the 1 woman not evaluable was lost due to attrition or exclusion from the trial, the final outcome for this 1 women or at what stage she left the trial. It is also unclear whether any of her outcome data was reported in the paper.
Selective reporting (reporting bias)	Low risk	All outcomes mentioned in Methods are reported in Results.
Other bias	Unclear risk	None reported. It would be difficult to rule out.

Berkowitz 2007.

Methods	Generation of allocation sequences: computer generated random numbers with permuted blocks to achieve balance in the design. Treatment assignment was undertaken by study biostatisticians away from the individual centres. Investigators rang biostatistics office to receive treatment assignment: confirmation was faxed to the investigator. Timing of randomisation: not stated. Power calculation undertaken: no. Blinding: not stated. Number of centres: 34 institutions across the United States and 1 Canadian centre. Timing and duration of study: enrolment May 2001 to June 2006. Follow up: not stated. Source of funding: not stated. Two authors (RB, MW) receive Clincial Research Support from IgG America Inc, Linthicum, Maryland, USA.
Participants	Number of participants: 73 (37/36) women randomised (40 and 37 pregnancies respectively). All pregnancies evaluable and analysed; 0 excluded. Women were diagnosed as having FMAIT if they were shown to have an incompatibility to a HPA on the father's platelets and to have circulating antibodies to that antigen. If the father was a heterozygote for the specific antigen, the HPA genotype of the fetal platelets was determined by testing cells obtained via amniocentesis during the current pregnancy. Definition standard risk: women with a documented FMAIT carrying a fetus whose platelets contained the offending antigen: previous sibling had not suffered an ICH. Maternal age: treatment: 31.1 years (range 19 to 43); control: 30.6 years (21 to 39).
Interventions	Randomisation between IVIG 1 g per kg per week plus prednisone 0.5 mg per kg per day and IVIG 2 g per kg per week. Treatment initiated at 20 weeks' gestation. A fetal blood sample taken at 32 weeks' gestation: Second‐line therapy initiated following this if the fetal platelet count < 30 x 10*9/l or the procedure could not be performed. Second‐line therapy: treatment group: prednisone 0.5 mg/kg/day added to regimen; control group: IVIG increased to 2 g per kg per week. Second‐line therapy failure defined as a birth platelet count < 30 x 10*9/l. Trial does not report preparations of IVIG used.
Outcomes	Development of a fetal or neonatal ICH. Fetal and birth platelet counts. Gestational age at delivery. Problems associated with fetal blood sampling. Complications of medical therapy as reported by women on a questionnaire administered after delivery.
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Adequate. Sequence generation was undertaken by computer generated random numbers with permuted blocks to achieve balance in the design. Treatment assignment was undertaken by study biostatisticians away from the individual centres. Investigators rang biostatistics office to receive treatment assignment: confirmation was faxed to the investigator.
Allocation concealment (selection bias)	Low risk	Adequate. Allocation concealment was undertaken by computer generated random numbers with permuted blocks to achieve balance in the design. Treatment assignment was undertaken by study biostatisticians away from the individual centres. Investigators rang biostatistics office to receive treatment assignment: confirmation was faxed to the investigator.
Blinding (performance bias and detection bias) All outcomes	High risk	Clinicians, patients and outcome assessors were not blinded to treatment allocation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All women randomised into the trial were included in the analysis of outcomes.
Selective reporting (reporting bias)	Low risk	All outcomes mentioned in Methods are reported in Results.
Other bias	Unclear risk	None reported. It would be difficult to rule out.

Bussel 1996a.

Methods	Generation of allocation sequences: random number list maintained at a separate site by the study statistician. Timing of randomisation: not stated. Power calculation undertaken: yes. Blinding: not stated. Number of centres: 20 (17 in the United States, 3 in the Netherlands). Timing and duration of study: enrolment March 1990 to January 1993. Follow up: until a normal platelet count was reached, and information relating to the first year of life was requested. Source of funding: not stated.
Participants	Number of participants: 54 randomised, all analysed. 0 excluded. Pregnant women with a serological diagnosis of FMAIT and a thrombocytopenic fetus (fetal platelet count of < 100 x 10 9/l). Where the father was heterozygous, serologic testing was used to confirm that the baby was affected by FMAIT. Age: not stated.
Interventions	IVIG 1 g per kg per week within one week of documentation of fetal thrombocytopenia (20 to 24 weeks' gestation). Randomisation to 1.5 mg/day dexamethasone or no dexamethasone. Different IVIG preparations were used throughout the study: Sandoglobulin, Polygam, Veinoglobulin and immunoglobulin prepared by the Dutch Red Cross. 9/54 participants subsequently received second‐line therapy (IVIG 1 g per kg per week with prednisone 60 mg per day) following an assessment of no response to initial treatment.
Outcomes	Platelet count increases observed between 1st and 2nd FBS or 1st sampling to birth. Fetal/neonatal death. Fetal/neonatal survival. ICH. Platelet count at delivery. Abortion. Premature delivery. One year follow up of neonatal outcome.
Notes	During the study, 2 fetuses were lost to analysis. The fetuses exsanguinated during fetal blood sampling and subsequent practice were altered to include the transfusion of concentrated maternal platelets at the time of sampling, when the platelet count was less than 50 x 10*9/l. When outcomes were analysed by preparation of IVIG, the three women treated in the Netherlands were excluded from this analysis.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Adequate. Sequence generation was undertaken by random number list maintained at a separate site by the study statistician.
Allocation concealment (selection bias)	Low risk	Adequate. Allocation concealment was undertaken by random number list maintained at a separate site by the study statistician.
Blinding (performance bias and detection bias) All outcomes	Unclear risk	Blinding of participants, personnel or outcome assessors was not reported.
Incomplete outcome data (attrition bias) All outcomes	Low risk	All women randomised into the trial were included in the analysis of outcomes.
Selective reporting (reporting bias)	Low risk	All outcomes mentioned in Methods are reported in Results.
Other bias	Unclear risk	None reported. It would be difficult to rule out.

DNA:Deoxyribonucleic acid 
 FBS: fetal blood sampling 
 FMAIT: fetomaternal alloimmune thrombocytopenia 
 HPA: Human Platelet Antigen 
 ICH: Intracranial haemorrhage 
 IVIG: Intravenous immunoglobulin

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Birchall 2003	This study is not a RCT. In this paper the authors report the findings of a study which aimed to determine whether the severity of FMAIT in the current pregnancy could be predicted from the history of FMAIT in previous pregnancies, and to assess the effects of different types of antenatal intervention. Fifty‐six fetuses were enrolled in this single arm study.
Bussel 1988	This study is not a RCT. In this paper the authors report the results of antenatal treatment with intravenous gamma globulin, with or without dexamethasone, in seven pregnant women who had previously had infants who had severe alloimmune thrombocytopenia
Bussel 1996b	This study is not a RCT. The authors present the data from their 3 protocols of IVIG treatment in this review.
Bussel 1997	This study is not a RCT. The authors report the data from their study alongside results from 2 additional studies. The objective of this paper was to study the correlates of thrombocytopenia in affected fetuses rather than a consideration of the management of FMAIT.
Bussel 2005	This study is not a full RCT. Personal communication with the author has discovered this information. The study is in part an RCT.
Dawkins 1999	This study is not a RCT.
Gaddipatti 2001	This study is not a RCT. This was a single arm study with the aim being to identify factors that would predict the response of the fetal platelet count to therapy so that use of fetal blood sampling could be minimised.
Kanhai 1996	This study is not a RCT. This was a study to assess the accuracy of a management program in women at risk for FMAIT and to describe peripartum outcomes.
Lynch 1992	This study is not a RCT. In this paper, the authors report their findings of the treatment of 18 women, who had previously delivered infants with severe alloimmune thrombocytopenia, with weekly infusions of intravenous gamma globulin from the diagnosis of fetal thrombocytopenia until birth.
Radder 2000	This study is not a RCT. The paper reports the findings of a retrospective and descriptive study to evaluate a less invasive management strategy for pregnant women with neonatal alloimmune thrombocytopenia without a history of ICH.
Radder 2001	This study is not a RCT. This was a multi‐arm, non randomised study designed to evaluate whether a less invasive treatment strategy results in a higher platelet count of the neonate and prevents intracranial haemorrhage in pregnant women who are at risk for fetal or neonatal alloimmune thrombocytopenia.
Silver 2000	This study is not a RCT. This was a single arm study designed to determine the effectiveness and likely mechanism of action of IVIG and to evaluate the safety of cordocentesis in cases of neonatal alloimmune thrombocytopenia.
van den Akker 2006	This study is not a RCT, rather a prospective data collection which aims to evaluate the safety of vaginal delivery in pregnancies with FMAIT.

FMAIT: fetomaternal alloimmune thrombocytopenia 
 ICH: intracranial haemorrhage 
 IVIG: intravenous immunoglobulin 
 RCT: randomised controlled trial

Characteristics of studies awaiting assessment [ordered by study ID]

Oepkes 2005.

Methods	Prospective, multicentre, parallel design, single blinded, RCT. Planned number of participants: 212.
Participants	1. Pregnant women with a subsequent pregnancy after prior pregnancy complicated by HPA alloimmunisation who have given birth to a child with a platelet count < 150 x 10*9/l in the first week of life. 2. HPA alloimmunisation must have been confirmed by the presence of maternal anti‐HPA antibodies and the offending HPA antigen in the fetus or homozygous partner. 3. The biological fathers are either homozygous positive for the HPA‐type or heterozygous. 4. In the case of a heterozygous father the platelet antigen genotype of the fetus will be tested before 28 weeks by amniocentesis. 5. At inclusion, the pregnancy is an ultrasonographically proven intrauterine singleton pregnancy with a gestational age between 12 and 28 weeks.
Interventions	Treatment group: low dose IVIG (0.5 g per kg per wk); Control group: standard treatment at high dose IvIG (1.0 g per kg per wk).
Outcomes	Primary: Number of neonates with ICH. Secondary: 1. cord blood platelet count at birth; 2. levels of maternal and neonatal anti‐HPA antibodies and IgG; 3. occurrence of other bleeding in the neonate; 4. type of neonatal treatment.
Notes	Trial was prematurely closed in 2009 due to a lack of patient accrual. If data become available it will be incorporated into future updates of the systematic review. ISRCTN29462550

HPA: Human Platelet Antigen 
 ICH: intracranial haemorrhage 
 IgG: immunoglobulin 
 RCT: randomised controlled trial

Characteristics of ongoing studies [ordered by study ID]

MacGregor 2005.

Trial name or title	Antenatal treatment of Alloimmune Thrombocytopenia.
Methods	RCT.
Participants	Women who have previously had a baby with neonatal alloimmune thrombocytopenia, or who have known platelet antibodies.
Interventions	Standard and High risk: IVIG twice weekly versus IVIG once weekly plus prednisolone for remainder of pregnancy. Standard risk begin treatment between 20 to 30 weeks' gestation. High risk begin treatment between 12 to 30 weeks' gestation.
Outcomes	Not stated.
Starting date	Not stated.
Contact information	Scott MacGregor DO. Director Dividion of Maternal Fetal Medicine, Department Obstetrics and Gynaecology, Evanston Nortwestern Healthcare Evanston Hospital Illinois.
Notes	None.

HPA: Human Platelet Antigen 
 ICH: intracranial haemorrhage 
 IgG: immunoglobulin 
 IVIG: intravenous immunoglobulin

Differences between protocol and review

In addition to the outcomes that we prespecified, there were a number of outcomes that the trials reported data for that were of significant interest to this systematic review. These non‐prespecified outcomes are reported at the end of the results section as "Outcomes not prespecified in the protocol". The outcomes are as follows.

• Satisfactory response to therapy.

• Mean change in fetal platelet count between first and second FBS.

• Mean change in fetal platelet count between pre‐treatment and fetal birth platelet count.

• Mean change in fetal platelet count at birth from first fetal blood sampling.

• Number of ICHs in previous sibling.

• Adverse outcomes associated with FBS.

• Adverse outcomes associated with treatment.

Contributions of authors

Rachel Rayment (RR) prepared the first draft of the protocol that was then reviewed and edited by Peter Soothill (PS), David Roberts (DR) James Bussel (JB) and Mike Murphy (MM).

RR and Susan Brunskill (SB) undertook additional searching and the selection of studies, eligibility and quality assessment, data extraction and analysis for the review and subsequent update.

RR, SB, DR, MM, Jim Bussel (JB) and PS contributed to the preparation of the final draft of the review for the original review and RR, SB, MM, JB and PS contributed to the preparation of the final draft for this update.

Sources of support

Internal sources

• National Blood Service, Research and Development, UK.

External sources

• No sources of support supplied

Declarations of interest

James Bussel has received unrestricted support of clinical research, including staff, over the past four years from IgG of America. He is an author on several of the studies assessed for inclusion in this review and the author of one of the included studies. James Bussel was not involved in the assessment of his studies for the review.

New search for studies and content updated (no change to conclusions)